eLife assessment

This important study examines the relationship between expiratory airflow and vocal pitch in adult mice during the production of ultrasonic vocalizations and also identifies a molecularly defined population of brainstem neurons that regulates mouse vocal production across development. The evidence supporting the study's conclusions that expiratory airflow shapes vocal pitch and that these brainstem neurons preferentially regulate expiratory airflow is novel and compelling. This work will be of interest to neuroscientists working on mechanisms and brainstem circuits that regulate vocal production and vocal-respiratory coordination.

Reviewer #1 (Public Review):

Summary:

In this important work, the authors propose and test a model for the control of murine ultrasonic vocalizations (USV) in which two independent mechanisms involving changes in laryngeal opening or airflow control vocal tone. They present compelling experimental evidence for this dual control model by demonstrating the ability of freely behaving adult mice to generate vocalizations with various intonations by modulating both the breathing pattern and the laryngeal muscles. They also present novel evidence that these mechanisms are encoded in the brainstem vocalization central neural pattern generator, particularly in the component in the medulla called the intermediate reticular oscillator (iRO). The results presented clearly advance understanding of the developmental nature of the iRO, its ability to intrinsically generate and control many of the dynamic features of USV, including those related to intonation, and its coordination with/control of expiratory airflow patterns. This work will interest neuroscientists investigating the neural generation and control of vocalization, breathing, and more generally, neuromotor control mechanisms.

Strengths:

Important features and novelty of this work include:

(1) The study employs an effective combination of anatomical, molecular, and functional/ behavioral approaches to examine the hypothesis and provide novel data indicating that expiratory airflow variations can change adult murine USV's pitch patterns.

(2) The results significantly extend the authors' previous work that identified the iRO in neonatal mice by now presenting data that functionally demonstrates the existence of the critical Penk+Vglut2+ iRO neurons in adult mice, indicating that the iRO neurons maintain their function in generating vocalization throughout development.

(3) The results convincingly demonstrate that the iRO neurons encode and can generate vocalizations by modulating both breathing and the laryngeal muscles.

(4) The anatomical mapping and tracing results establish an important set of input and output circuit connections to the iRO, including input from the vocalization-promoting subregions of the midbrain periaqueductal gray (PAG), as well as output axonal projections to laryngeal motoneurons, and to the respiratory rhythm generator in the preBötzinger complex.

(5) These studies advance the important concept that the brainstem vocalization pattern generator integrates with the medullary respiratory pattern generator to control expiratory airflow, a key mechanism for producing various USV types characterized by different pitch patterns.

Weaknesses:

A limitation is that the cellular and circuit mechanisms by which the vocalization pattern generator integrates with the respiratory pattern generator to control expiratory airflow has not been fully worked out, requiring future studies.

Reviewer #2 (Public Review):

Summary:

Both human and non-human animals modulate the frequency of their vocalizations to communicate important information about context and internal state. While regulation of the size of the laryngeal opening is a well-established mechanism to regulate vocal pitch, the contribution of expiratory airflow to vocal pitch is less clear. To consider this question, this study first characterizes the relationship between the dominant frequency contours of adult mouse ultrasonic vocalizations (USVs) and expiratory airflow using whole-body plethysmography. The authors also include data from a single mouse that combines EMG recordings from the diaphragm and larynx with plethysmography to provide evidence that the respiratory central pattern generator can be re-engaged to drive "mini-breaths" that occur during the expiratory phase of a vocal breath. Next, the authors build off of their previous work characterizing intermediate reticular oscillator (iRO) neurons in mouse pups to establish the existence of a genetically similar population of neurons in adults and show that artificial activation of iRO neurons elicits USV production in adults. Third, the authors examine the acoustic features of USV elicited by optogenetic activation of iRO and find that a majority of natural USV types (as defined by pitch contour) are elicited by iRO activation and that these artificially elicited USVs are more likely than natural USVs to be marked by positive intonation (positive relationship between USV dominant frequency and expiratory airflow).

Strengths:

Strengths of the study include the novel consideration of expiratory airflow as a mechanism to regulate vocal pitch and the use of intersectional methods to identify and activate the iRO in adult mice. The establishment of iRO neurons as a brainstem population that regulates vocal production across development is an important finding.

Weaknesses:

The conclusion that the respiratory CPG is re-engaged during "mini-breaths" throughout a given vocal breath would be strengthened by including analyses from more than one mouse.

Author response:

The following is the authors’ response to the original reviews.

In the revised manuscript we have included an additional study that significantly contributes to the conclusions and models of the original version. Briefly, Figure 3 now describes our characterization of the diaphragm and laryngeal muscle activities (electromyography, EMG) during endogenous vocalizations. These EMGs also serve as representations of the brainstem breathing central pattern generator (CPG) inspiratory and post-inspiratory generating neurons, respectively. In our original submission, we found that many of the vocalizations had changes in pitch that mirrored the change in expiratory airflow (we termed positive intonation), and we proposed that the coordination of breathing muscles (like the inspiratory muscles) and larynx patterned this. This mechanism is akin to our findings for how neonatal cries are rhythmically timed and produced (Wei et al. 2022). The newly presented EMG data re-inforces this idea. We found that for vocalizations with positive intonation, the inspiratory diaphragm muscle has an ectopic burst(s) of activity during the expiration phase which corresponds to a decrease in airflow and pitch, and this is followed by laryngeal muscle activity and increased pitch. This can be cycled throughout the expiration to produce complex vocalizations with oscillations in pitch. A basal breath is hardwired for the laryngeal muscle activity to follow the diaphragm, so the re-cycling of this pattern nested within an expiration (a ‘mini-breath’ in a ‘breath’) demonstrates that the vocalization patterning system engages the entire breathing CPG. This contrasts with the canonical model that activity of the laryngeal premotor neurons control all aspects of producing / patterning vocalizations. Furthermore, this mechanism is exactly how the iRO produces and patterns neonatal vocalizations (Wei et al. 2022) and motivates the likely use of the iRO in adult vocalizations.

Response to recommendations for the authors:

Reviewer #1:

(1) The authors should note in the Discussion that the cellular and circuit mechanisms by which the vocalization pattern generator integrates with the respiratory pattern generator to control expiratory airflow have not been fully worked out, requiring future studies.

This was noted in the discussion section “The iRO likely patterns intonation for endogenous phonation”.

(2) Please change the labeling of the last supplemental figure to Figure Supplemental 5.

Thank you for identifying this.

Reviewer #2:

Major concerns

(1) While it is true that modulation of activity in RAm modulates the laryngeal opening, this statement is an incomplete summary of prior work. Previous studies (Hartmann et al., 2020; Zhang et al., 1992, 1995) found that activation of RAm elicits not just laryngeal adduction but also the production of vocal sounds, albeit vocal sounds that were spectrally dissimilar from speciestypical vocalizations. Moreover, a recent study/preprint that used an activity-dependent labeling approach in mice to optogenetically activate RAm neurons that were active during USV production found that re-activation of these neurons elicits USVs that are acoustically similar to natural USVs (Park et al., 2023). While the authors might not be required to cite that recent preprint (as it is not yet peer-reviewed), the fact that activation of RAm elicits vocal sounds is clear evidence that its effects go beyond modulating the size of the laryngeal opening, as this alone would not result in sound production (i.e., RAm activation must also recruit expiratory airflow). The authors should include these relevant studies in their Introduction. Moreover, the rationale for the model proposed by the authors (that RAm controls laryngeal opening whereas iRO controls expiratory airflow) is unclear with regard to these prior studies. The authors should include a discussion of how these prior findings are consistent with their model (as presented in the Introduction, as well as in Figure 4 and relevant Discussion) that RAm modulates the size of laryngeal opening but not expiratory airflow.

An introduction and discussion of the Veerakumar et. al. 2023 and Park et. al. 2024 manuscripts describing RAm in mice has now been included.

The iRO serves to coordinate the breath airflow and laryngeal adduction to produce sound and the intonation within it that mirrors the breath airflow. This occurs because the iRO can control the breathing CPG (synaptic input to the preBötC inspiratory pacemaker) and is premotor to multiple laryngeal muscles (Wei et. al. 2022). The modulation of the expiratory airflow is by inducing momentary contraction of the diaphragm (via excitation of the preBötC) which opposes (a.k.a. slows) expiration. This change in flow results in a decrease in pitch (Fig. 3 in the revised manuscript, Wei et. al. 2022).

It is our understanding that the basic model for RAm evoked USVs is that RAm evokes laryngeal adduction (and presumed abdominal expiratory muscle activation) and this activity is momentarily stopped during the breath inspiration by inhibition from the preBötC (Park et. al. 2024). So, in this basic model, any change in pitch and expiratory airflow would be controlled by tuning RAm activity (i.e., extent of laryngeal adduction). In this case, the iRO induced inspiratory muscle activity should not occur during expiration, which is not so (Fig. 3). Note, the activity of abdominal expiratory muscles during endogenous and RAm evoked USVs has not been characterized, so the contribution of active expiration remains uncertain. This is an important next step.

We have now included a discussion of this topic which emphasizes that iRO and RAm likely have reciprocal interactions (supported by the evidence of this anatomical structure). These interactions would explain why excitation of either group can evoke USVs and, perhaps, the extent that either group contributes to a USV explains how the pitch / airflow changes. An important future experiment will be to determine the sufficiency of each site in the absence of the other.

(2) The authors provide evidence that the relationship between expiratory airflow and USV pitch is variable (sometimes positive, sometimes negative, and sometimes not related). While the representative spectrograms clearly show examples of all three relationship types, no statistical analyses are included to evaluate whether the relationship between expiratory airflow and USV pitch is different than what one would expect by chance. For example, if USV pitch were actually unrelated to expiratory airflow, one might nonetheless expect spurious periods of positive and negative relationships. The lack of statistical analyses to explicitly compare the observed data to a null model makes it difficult to fully evaluate to what extent the evidence provided by the authors supports their claims.

We have now included two null distributions and compared our observed correlation values to these. The two distributions were created by taking each USV / airflow pair and randomly shuffling either the normalized USV pitch values (pitch shuffled) or the normalized airflow values (airflow shuffled) to simulate the distribution of data should no relationship exist between the USV pitch and airflow.

(3) The relationship between expiratory airflow and USV pitch comes with two important caveats that should be described in the manuscript. First, even in USV types with an overall positive relationship between expiratory airflow and pitch contour, the relationship appears to be relative rather than absolute. For example, in Fig. 2E, both the second and third portions of the illustrated two-step USV have a positive relationship (pitch goes down as expiratory airflow goes down). Nonetheless, the absolute pitch of the third portion of that USV is higher than the second portion, and yet the absolute expiratory airflow is lower. The authors should include an analysis or description of whether the relationship between expiratory airflow and USV pitch is relative vs.

absolute during periods of 'positive intonation'.

The relationship between pitch and airflow is relative and this in now clarified in the text. To determine this, we visualized the relationship between the two variables by scatterplot for each of the USVs syllables and, as the reviewer notes, a given airflow cannot predict the resulting frequency and vice versa.

(4) A second important caveat of the relationship between expiratory airflow and USV pitch is  that changes in expiratory airflow do not appear to account for the pitch jumps that characterize mouse USVs (this lack of relationship also seems clear from the example shown in Fig. 2E). This caveat should also be stated explicitly.

The pitch jumps do not have a corresponding fluctuation in airflow, and this is now stated in the results and discussion.

(5) The authors report that the mode of relationship between expiratory airflow and USV pitch (positive intonation, negative intonation, or no relationship) can change within a single USV. Have the authors considered/analyzed whether the timing of such changes in the mode of relationship coincides with pitch jumps? Perhaps this isn’t the case, but consideration of the question would be a valuable addition to the manuscript.

We analyzed a subset of USVs with pitch jumps that were defined by a change >10 kHz, at least 5ms long, and had one or two jumps. The intonation relationships between the sub-syllables within a USV type were not stereotyped as evidenced by the same syllable being composed of combinations of both modes.

(6) The authors incorrectly state that PAG neurons important for USV production have been localized to the ventrolateral PAG. Tschida et al., 2019 report that PAG-USV neurons are located predominantly in the lateral PAG and to a lesser extent in the ventrolateral PAG (see Fig. 5A from that paper). The finding that iRO neurons receive input from VGlut2+ ventrolateral PAG neurons represents somewhat weak evidence that these neurons reside downstream of PAG-USV neurons. This claim would be strengthened by the inclusion of FOS staining (following USV production), to assess whether the Vglut+ ventrolateral PAG neurons that provide input to iRO are active in association with USV production.

This comment correctly critiques that our PAG à iRO tracing does not demonstrate that the labeled PAG neurons are sufficient nor necessary for vocalization. Directly demonstrating that activation and inhibition the PAG-iRO labeled neurons ectopically drives or prevents endogenous USVs is an important next step. While FOS implies this connectivity, it does not definitely establish it and so this experiment is impacted by some of the caveats of our tracing (e.g. PAG neurons that drive sniffing might be erroneously attributed to vocalization).

Our reading of the literature could not identify an exact anatomical location within the mouse PAG and this site appears to vary within a study and between independent studies (like within and between Tschida et. al. 2019 and Chen et. al. 2021). The labeling we observed aligns with some examples provided in these manuscripts and with the data reported for the retrograde tracing from RAm (Tschida et al 2019).

(7) In Figure S5A, the authors show that USVs are elicited by optogenetic activation of iRO neurons during periods of expiration. In that spectrogram, it also appears that vocalizations were elicited during inspiration. Are these the broadband vocalizations that the authors refer to in the Results? Regardless, if optogenetic activation of iRO neurons in some cases elicits vocalization both during inspiration and during expiration, this should be described and analyzed in the manuscript.

The sound observed on the spectrogram during inspiration is an artefact of laser evoked head movements that resulted in the fiber cable colliding with the plethysmography chamber. In fact, tapping an empty chamber yields the same broad band spectrogram signal. The evoked USV or harmonic band vocalization is distinct from this artefact and highlighted in pink.

(8) Related to the comment above, the authors mention briefly that iRO activation can elicit broadband vocalizations, but no details are provided. The authors should provide a more detailed account of this finding.

The broadband harmonic vocalizations we sometimes observe upon optogenetic stimulation of AAV-ChR2 expressing iRO neurons are akin to those previously described within the mouse vocal repertoire (see Grimsley et. al .2011). We have added this citation and mentioned this within the text. 

(9) The effects of iRO stimulation differ in a couple of interesting ways from the effects of PAGUSV activation. Optogenetic activation of PAG-USV neurons was not found to entrain respiration or to alter the ongoing respiratory rate and instead resulted in the elicitation of USVs at times when laser stimulation overlapped with expiration. In contrast, iRO stimulation increases and entrains respiratory rate, increases expiratory and inspiratory airflow, and elicits USV production (and also potentially vocalization during inspiration, as queried in the comment above). It would be informative for the authors to add some discussion/interpretation of these differences.

We have added a section of discussion to describe the how these different results may be explained by the iRO being a vocal pattern generator versus the PAG as a ‘gating’ signal to turn on the medullary vocalization patterning system (iRO and RAm). See discussion section ‘The iRO likely patterns intonation for endogenous phonation’.

(10) The analysis shown in Fig. 4D is not sufficient to support the author’s conclusion that all USV types elicited by iRO activation are biased to have more positive relationships between pitch and expiratory airflow. The increase in the relative abundance of down fm USVs in the opto condition could account for the average increase in positive relationship when this relationship is considered across all USV types in a pooled fashion. The authors should consider whether each USV type exhibits a positive bias. Although such a comparison is shown visually in Fig. 4G, no statistics are provided. All 7 USV types elicited by optogenetic activation of iRO should be considered collectively in this analysis (rather than only the 5 types currently plotted in Fig. 4G).

In the original submission the statistical analysis of r values between opto and endogenous conditions was included in the figure legend (‘panels E-G, two-way ANOVA with Sidak’s post-hoc test for two-way comparisons was used; all p-values > 0.05), and this has not changed in the revised manuscript. We have now provided the suggested comparison of opto vs endogenous USVs without down fm (Fig. 5D). This positive shift in r is statistically significant (…).

(11) The evidence that supports the author’s model that iRO preferentially regulates airflow and that RAm preferentially regulates laryngeal adduction is unclear. The current study finds that activation of iRO increases expiratory (and inspiratory) airflow and also elicits USVs, which means that iRO activation must also recruit laryngeal adduction to some extent. As the authors hypothesize, this could be achieved by recruitment of RAm through iRO’s axonal projections to that region.

Note, it is more likely that iRO is directly recruiting laryngeal adduction as they are premotor to multiple laryngeal muscles like the thyroarytenoid and cricothyroid (Wei et. al. 2022). The ‘Discussion’ now includes our ideas for how the iRO and RAm likely interact to produce vocalizations.

In the recent preprint from Fan Wang’s group (Park et al., 2023), those authors report that RAm is required for USV production in adults, and that activation of RAm elicits USVs that appear species-typical in their acoustic features and elicits laryngeal adduction (assessed directly via camera). Because RAm activation elicits USVs, though, it must by definition also recruits expiratory airflow. Can the authors add additional clarification of how the evidence at hand supports this distinction in function for iRO vs RAm?

See response to ‘Major Concern #1”.

Minor concerns 

(1) The authors might consider modifying the manuscript title. At present, it primarily reflects the experiments in Figure 2.

We have provided a title that we feel best reflects the major point of the manuscript. We hope that this simplicity enables it to be recognized by a broad audience of neuroscientists as well as specialists in vocalization and language.

(2) The statement in the abstract that "patterns of pitch are used to create distinct 'words' is somewhat unclear. Distinct words are by and large defined by combinations of distinct phonemes. Are the authors referring to the use of "tonemes" in tonal languages? If so, a bit more explanation could be added to clarify this idea. This minor concern includes both the Abstract, as well as the first paragraph of the Introduction.

We have clarified this line in the abstract to avoid the confusing comparison between mouse vocalizations and human speech. In the introduction we have expanded our explanation to clarify that variations in pitch are a component of spoken language that add additional meaning and depth to the underlying, phonemic structure. 

(3) Multiple terms are used throughout the manuscript to refer to expiratory airflow: breath shape (in the title), breath pattern, deviations in exhalation, power of exhalation, exhalation strength, etc. Some of these terms are vague in meaning, and a consolidation of the language would improve the readability of the abstract and introduction.

We have chosen a smaller selection of descriptive words to use when describing these breath features.

(4) Similarly, "exhalation" and "expiration" are both used, and a consistent use of one term would help readability.

See point 3.

(5) In a couple of places in the manuscript, the authors seem to state that RAm contains both laryngeal premotor neurons as well as laryngeal motor neurons. This is not correct to our knowledge., but if we are mistaken, we would ask that the authors add the relevant references that report this finding.

It is our understanding that the RAm is defined as the anatomical region consistent with the murine rostral and caudal ventral respiratory groups composed of multiple premotor neuron pools to inspiratory, expiratory, laryngeal, and other orofacial muscles. This is supported by neurons within RAm that reflect multiple phases of the inspiratory and expiratory cycle (Subramanian et. al. 2018) and excitation of sub-regions within RAm modulating multiple parts of the breathing control system (Subramanian et. al. 2018 and Subramanian 2009). Rabies tracing of the various premotor neurons which define the anatomical region of RAm in the mouse shows that they surround the motor neurons in the loose region of the nucleus ambiguus (the anatomical location of RAm) for multiple muscles of the upper airway system, such as the thyroarytenoid (Wu et. al. 2017, Dempsey et. al. 2021 and Wei et. al. 2022). Given that the name RAm reflects a broad anatomical location, we have used it to describe both the premotor and motor neurons embedded within it. We have now clarified this in the text.

(6) The statistical analysis applied in Figure 1C is somewhat confusing. The authors show two distributions that appear different but report a p-value of 0.98. Was the analysis performed on the mean value of the distributions for each animal, the median, etc.? If each animal has two values (one for USV+ breaths and one for USV- breaths), why not instead compare those with a paired t-test (or Wilcoxon rank sign)? Additional information is needed to understand how this analysis was performed.

The original manuscript version used a two-way anova to compare the normalized histogram of instantaneous frequency for breaths with (USV+) or without (USV-) for each animal (first factor: USV+/-, second factor: Frequency). The p-value for the first factor (USV) was 0.98 showing no statistically significant effect of USV on the distribution of the histogram.

For simplicity, we have instead performed the analysis as suggested and include a bar graph. This analysis shows that the instantaneous frequency of USV breaths is, in fact, statistically significantly lower than those without USVs. We have updated the figure legend and text to reflect this.

(7) The use of the word "syllable" to describe parts of a USV that are produced on a single breath may be confusing to some scientists working on rodent USVs. The term 'syllable' is typically used to describe the entirety of a USV, and the authors appear to use the term to describe parts of a USV that are separated by pitch jumps. The authors might consider calling these parts of USVs "sub-syllables".

We have clarified these descriptions throughout the text. We now refer to the categories as ‘syllable types’, define ‘syllables’ as ‘a continuous USV event’ with no more than 20ms of silence within and finally ‘sub-syllables’ to refer to components of the syllable separated by jumps in frequency (but not gaps in time).

(8) In Figure S3, final row, the authors show a USV produced on a single breath that contains two components separated by a silent period. This type of bi-syllabic USV may be rare in adults and is similar to what the authors showed in their previous work in pups (multiple USVs produced on a single expiration, separated by mini-inspirations). One might assume that the appearance of such USVs in pups and their later reduction in frequency represents a maturation of vocalrespiratory coordination. Nonetheless, the appearance of bi-syllabic USVs has not been reported in adult mice to our knowledge, and the authors might consider further highlighting this finding.

We were also struck by the similarity of these USVs to our study in neonates and such types of similarities sparked an interest in the role of the iRO in patterning adult USVs. We now include a description of the presence and abundance of bi- and tri-syllablic calls observed in our recordings to highlight this finding.

(9) Figure 4 is referenced at the end of the second Results section, but it would seem that the authors intended to reference Figure 2. 

For simplicity we included some of the referenced data within Fig. S5. We appreciate the recommendation.

(10) In the optogenetic stimulation experiments, the authors should clarify why bilateral stimulation was applied. Was unilateral stimulation ineffective or less effective? The rationale provided for the use of bilateral stimulation (to further localize neural activation) is unclear.

The iRO is bilateral and, we presume, functions similarly. So, we attempted to maximally stimulate the system. We have clarified this in the methods.

(11) Figure Supplemental '6' should be '5'.

Thanks!

(12) Last sentence of the Introduction: "Lasty" should be "lastly".

Thanks!

(13) There are two references for Hage et al., 2009. These should be distinguished as 2009a and 2009b for clarity.

Thanks!

Author response:

The following is the authors’ response to the current reviews.

We thank the reviewers and editor for their careful review of our work. We believe the resulting manuscript is much stronger. We agree with the comments made by Reviewer #2 regarding additional histology and neuronal data analysis, which will be presented in subsequent work.

---

The following is the authors’ response to the original reviews.

Reviewer 1 (Public Weaknesses):

It was not always clear what the lesion size was. This information is important for future applica- tions, for example, in the visual cortex, where neurons are organized in retinotopy patterns.

We thank the reviewer for this feedback. While there is some variation in lesion volume for a given parameter set, we have added more details of the volumes of lesions created in our testing (Fig. 4 and Fig. 5).

It would be helpful if the author could add some discussion about whether and how this method could be used in other types of array/multi-contact electrodes, such as passive neuropixels, S- probes, and so on. In addition, though an op-amp was used in the design, it would still be helpful if the author could provide a recommended range for the impedance of the electrodes.

We thank the reviewer for this suggestion. We have both added a demonstration of use in a differ- ent multielectrode probe type (with a U-probe) in Fig. 8, and we have added a discussion about which types of multielectrode probes would be suitable on Page 15, Line 420.

“We demonstrated that our electrolytic lesioning technique works with a linear multicontact probe by testing with a U-Probe in ex vivo rabbit cortex. There are no particular limitations that would prevent our specific electrolytic lesioning technique and device from working with any passive multielectrode probe. The main requirements for use are that the probe has two electrodes that can directly (via whatever necessary adapters) connect to the lesioning device, such that arbitrary current can be passed into them as the anode and cathode. This would limit use of probes, like Neuropixels, where the on-chip acquisition and digitization circuitry generally precludes direct connection to electrodes [1], [2]. The impedance of the multielectrode probe should not be an issue, due to the use of an op amp. We showed use  with a Utah array (20-800 kΩ) and a U-Probe (1-1.5 MΩ). The specific op amp used here has a voltage range of ± 450 V, which assuming a desired output of 150 µA of current would limit electrode impedance to 6 MΩ. Though a different op amp could easily be used to accommodate a higher electrode impedance, it is unlikely that this would be necessary, since most electrodes have impedances between 100 kΩ to 1 MΩ [3].”

Reviewer 2 (Public Weaknesses):

In many of the figures, it is not clear what is shown and the analysis techniques are not well described.

We thank the reviewer for this feedback. We hope that our edits to both the figures and the text have improved clarity for readers.

The flexibility of lesioning/termination location is limited to the implantation site of the multielec- trode array, and thus less flexible compared to some of the other termination methods outlined in Appendix 2.

We thank the reviewer for this point. You are right that the lesioning location is limited to the multielectrode array’s implantation site, while other methods in Appendix 2 do not require prox- imity of the lesion location and the electrophysiology recording site. However, we believe that the closeness of the lesioning location to the microelectrode array is a strength - guaranteeing record- ings from the perilesional area - even with the small negative of reduced flexibility. Multielectrode arrays can be implanted in many areas of cortex. If one wanted to study distal effects of a lesion, additional electrophysiology probes could be implanted to record from those areas. We have noted this on Page 3, Line 117.

“While the link between the lesion location and the multielectrode location technically con- strains the lesion to an area of cortex in which a multielectrode array could be implanted, we see the connection as a positive, because it ensures recording some neuroelectrophysiology from the perilesional area in which recovery is hypothesized to occur (see Appendix 1Data Availabilityappendix.41).”

Although the extent of the damage created through the Utah array will vary based on anatomical structures, it is unclear what is the range of lesion volumes that can be created with this method, given a parameter set. It was also mentioned that they performed a non-exhaustive parameter search for the applied current amplitude and duration (Table S1/S2) to generate the most suitable lesion size but did not present the resulting lesion sizes from these parameter sets listed. Moreover, there’s a lack of histological data suggesting that the lesion size is precise and repeatable given the same current duration/amplitude, at the same location.

We thank the reviewer for this thoughtful feedback. We have added figures (Figs. 4 and 5), where we show the relationship between estimated lesion volume and the current amplitude and duration parameters. These figures include more data from the tests in Supplementary File 1 and Supplementary File 2. While there is some variation in lesion volume for a given current amplitude and duration, there is still a clear relationship between the parameters and lesion volume.

It is unclear what type of behavioral deficits can result from an electrolytic lesion this size and type (∼3 mm in diameter) in rhesus macaques, as the extent of the neuronal loss within the damaged parenchyma can be different from past lesioning studies.

While we appreciate the reviewer’s interest in the behavioral deficits associated with our lesions in rhesus macaques, reporting these falls beyond the scope of this manuscript. Future work will explore the behavioral deficits associated with these lesions

The lesioning procedure was performed in Monkey F while sedated, but no data was presented for Monkey F in terms of lesioning parameters, lesion size, recorded electrophysiology, histological, or behavioral outcomes. It is also unclear if Monkey F was in a terminal study.

We apologize for not being more explicit about the parameters used for the lesion in Monkey F. We have added this in Results on Page 5, Line 209 and in Methods on Page 19, Line 586.

“After this validation and refinement, one proof-of-concept lesion (150 µA direct current passed through adjacent electrodes for 45 seconds) was performed in an in vivo sedated rhe- sus macaque (Monkey F) in order to validate the safety of the procedure.”

“This lesion was created by applying 150 µA of direct current to two adjacent electrodes in the microelectrode array for 45 seconds.”

We also clarified the parameters used for the other lesions in Monkeys H and U in Results on Page 7, Line 233 and in Methods on Page 19, Line 586.

“In all of the fourteen lesions across two awake-behaving rhesus macaques (150 µA direct current passed through adjacent electrodes for 30 or 45 seconds (30s for Monkey U and 45s for Monkey H, except lesion H200120 which was for 50 seconds)), the current source worked as expected, providing a constant current throughout the duration of the procedure.”

“In these lesions, 150 µA of direct current was applied to two adjacent electrodes in the mi- croelectrode array for 30 or 45 seconds (30s for Monkey U, 45s for Monkey H), except in lesion H200120 where current was applied for 50 seconds.”

Monkey F was euthanized shortly after the lesion, so we now mention this on Page 19, Line 583.

“Based on this, and a lack of physiological signs of pain from the anaesthetized pig studies, a lesion was performed on a sedated rhesus macaque who was subsequently euthanized due to unrelated health complications (Monkey F; 16 year-old adult, male rhesus macaque) in order to further verify safety before use in awake-behaving rhesus.”

Because Monkey F was sedated and then euthanized shortly after, there is no behavioral data. As the lesion in sedated Monkey F was used to validate the safety of the procedure, any further data and analysis fall beyond the scope of this manuscript.

As an inactivation method, the electrophysiology recording in Figure 5 only showed a change in pairwise comparisons of clustered action potential waveforms at each electrode (%match) but not a direct measure of neuronal pre and post-lesioning. More evidence is needed to suggest robust neuronal inactivation or termination in rhesus macaques after electrolytic lesioning. Some exam- ples of this can be showing the number of spike clusters identified each day, as well as analyzing local field potential and multi-unit activity.

The reviewer has pointed out some short comings of the original analysis, which we believe have since been addressed with the revised analysis. LFP and spiking activity are functional measures that are more ambiguous in terms of loss and are also the subject of another manuscript currently under revision.

The advantages over recently developed lesioning techniques are not clear and are not discussed.

We thank the reviewer for noting this. We have added a section, also responding to their later request for us to compare our work to Khateeb et al. 2022, by adding a section to the Discussion on Page 16, Line 434.

“Perhaps the most unique advantage of our technique in comparison with other existing inactivation methods lies in Design Consideration #1: stable electrophysiology pre- and post-inactivation (Appendix 1Data Availabilityappendix.41). While several methods exist that allow for localization and size control of the inactivation (Design Consideration #2) and cross compatibility across regions and species (Design Consideration #3), few have achieved compatibility with stable electrophysiology. For example, some studies record electrophysiology only after the creation of the lesion, preventing comparison with baseline neuronal activity [4]. One recent study, Khateeb, et al., 2022, developed an inactivation method that is effectively combined with stable electrophysiology by creating photothrombotic lesions through a chronic cranial window integrated with an electrocorticography (ECoG) array [5], which may be appropriate for applications where local field potential (LFP) recording is sufficient. This approach has trade-offs with regards to the three design considerations presented in Appendix 1Data Availabilityappendix.41.

While Khateeb, et al., present a toolbox with integrated, stable electrophysiology from an ECoG array pre- and post- inactivation (Design Consideration #1), it demonstrated recordings from an ECoG array with limited spatial resolution. While a higher density ECoG array that would provide higher spatial resolution could be used, increasing the density of opaque electrodes might occlude optical penetration and constrain photothrombotic lesions. Further, ECoG arrays are limited to recording LFP, not electrophysiology at single neuron resolution, potentially missing meaningful changes in the neuronal population activity after lesioning. Khateeb, et al., demonstrated localization and control the size of inactivation (Design Consideration #2). In this manuscript, we have shown that the amount and duration of direct current are significant determinants of lesion size and shape, while with photothrombotic lesions, light intensity and aperture diameter are the significantly relevant parameters. One potential advantage of photothrombotic approaches is the use of optical tools to monitor anatomical and physiological changes after lesioning through the cranial window, though the research utility of this monitoring remains to be demonstrated.

Although the method presented by Khateeb, et al., shows some cross-compatibility (Design Consideration #3), it has greater limitations in comparison with the method presented here. For example, while Khateeb, et al., notes that the approach could be adapted for use in smaller organisms, no modification is needed for use in other species with this work’s approach–so long as a multielectrode probe is implantable. In this manuscript we demon- strate electrolytic lesioning spanning two multielectrode probes across rabbits, pigs, sheep, and rhesus macaques, and our same device could be easily used with other smaller species, like rats, in which multielectrode probes have been successfully implanted [6]. Further, the approach in Khateeb, et al., is limited to superficial brain structures, due to the need for opti- cal accessibility. As noted, fiber optics could allow access to deeper structures, which would bring associated additional tissue damage, but deeper structure lesioning was not demon- strated. In contrast, the approach presented here can be used in any region of cortex in which a multielectrode probe can be implanted, which, depending on the probe used, does not limit it to surface structures. For example, we demonstrated use of our lesioning tech- nique with a linear U-probe (Fig. 8figure.caption.25), which could be used to reach deeper layers of cortex or specific deep cortical structures. In both techniques, the location of the lesion is tied to the location of the electrophysiology (for Khateeb et al., wherever the cra- nial window and ECoG array are; for this technique, wherever the multielectrode probe has been implanted), which ensures that the electrophysiology will include recordings from the perilesional area. Neither work addresses the potential of their technique to induce chronic post-lesion behavioral effects, which is a key goal for future work.”

There is a lack of quantitative histological analysis of the change in neuronal morphology and loss.

We appreciate the reviewer’s desire for a quantitative histological analysis, however this falls out- side of the scope of this manuscript. We are not attempting to make strong claims about the number of neurons lost through lesioning or thoroughly characterize morphological changes in the neurons. The histology is intended to show that lesioning did lead to a loss of neurons, but the precise num- ber of neurons lost is neither in scope nor is likely to be highly conserved across lesions.

There is a lack of histology data across animals and on the reliability of their lesioning techniques across animals and experiments.

We thank the reviewer for this point. As stated above, we have now added Fig. 4 and Fig. 5, which includes volume estimates based on the histology from more of our ex vivo and in vivo testing across animals.

There is a lack of data on changes in cortical layers and structures across the lesioning and non- lesioning electrodes.

We acknowledge that the histology does not have the level of detail that is expected from many modern studies. However, the goal here was dramatically different: we sought to calibrate a novel lesion device, ensure it’s safe use in large mammals (specifically, non-human primates) and pro- vide estimates of the lesion size to compare with the literature. The extent of histology that could be performed and the tools available to us prevent such an in depth analysis. We can say based on shank length of the Utah arrays used and known anatomy that we have affected layer 2/3 and maybe a bit of layer 4.

Reviewer 1 (Recommendations For The Authors):

Figure 5b. It would be helpful if the author could plot the delta match separately for the lesion elec- trodes, near neighbor electrodes, and far neighbors. This would help understand the lesion effect, specifically whether the effect is selective (e.g., more potent for the lesion and adjacent electrodes.)

The fact that neuron loss is not particularly selective can already be seen in the spike waveform plots, arranged spatially on the array. Plenty of clear change is observed far from the lesion elec- trodes (marked with black dots) as well as nearby. We have made mention of this localized non- specificity in the main text and have ensured to remphasize in the figure legened. While a nice suggestion, we currently don’t feel this result rises to the level of a figure given it is not highly specific spatially.

Reviewer 2 (Recommendations For The Authors):

Overall the quality of the paper, the figures and the analysis used could be significantly improved. There is a lack of scientific rigor in the presentation of figures and analysis techniques. It is not clear what the authors are trying to communicate through the figures and their choice of figures to show is confusing (see below).

We thank the reviewer for their pointed critiques and believe we have addressed their concerns with many changes to the text, a revamped waveforms analysis, and both the expansion and addition of results.

The neurophysiology data shown doesn’t suggest neuronal loss, it only shows change which needs strong control data to show it is due to a lesion.

As detailed below, we have presented a revised analysis that provides this control. While the reviewer is right to point out we can distinguish actual neuron loss from neuron silencing, we be- lieve the new analysis rigorously indicates new rates of sample turnover beyond those expected from healthy state.

The histology figure should be replaced with a high-quality representation without folds.

We understand the reviewer’s suggestion. While ideally we would have many histology slices from each lesion, due to cost, we were only able to collect one histology slice per lesion. The folds were introduced by the company that performed the H&E staining, and we unfortunately cannot remove the folds. Therefore, despite the folds, this is the best and only image from this lesion. We hope that the markings on the figure and the comment in the caption is sufficient to explain to readers that the folds are not a result of the lesion but instead a result of the histology process.

The authors suggest that this lesioning method will be compatible with any available multielec- trode probe theoretically. Since all testing was done with a Utah array, it will be helpful to add an explanation about potential constraints that will make a given array compatible with this method.

We thank the reviewer for this suggestion. As stated above, we have both added a demonstration of use in a different multielectrode probe type (with a U-probe) in Fig. 8, and we have added a discussion about which types of multielectrode probes would be suitable on Page 15, Line 420.

The authors should cite and discuss previous studies using electrolytic lesioning in awake-behaving animals to study the causal connection between the brain and behavior. (One example study: Morissette MC, Boye SM. Electrolytic lesions of the habenula attenuate brain stimulation reward. Behavioural brain research. 2008 Feb 11;187(1):17-26.)

We thank the reviewers for this suggestion. We have added a mention of existing electrolytic le- sioning studies on Page 2, Line 88.

“Prior termination studies mostly measure behavioral output, with no simultaneous measures of neuronal activity during the behavior, impairing their ability to provide insight into the causal connection between the brain and behavior [7]–[11], or with no baseline (i.e., pre- lesion) measures of neuronal activity [4].”

The authors should compare their technique with other recent lesioning studies in primates (e.g. Khateeb et al, 2022)

We again thank the reviewer for this point. Specifically not mentioning Khateeb et al. 2022 was a submission error on our part; we cited the paper in Appendix 2 in the version uploaded to the eLife submission portal, but we had uploaded the version prior to citing it to bioRxiv. We have combined addressing this with addressing a previous comment, as mentioned above, with a section in the Discussion on Page 16, Line 434.

In Appendix 2, the authors suggest that a major limitation of optogenetics and chemogenetic in- activation methods is the lack of rhesus-compatible constructs. However, several viral constructs have successful implementation in rhesus monkeys so far (e.g. Galvan A, Stauffer WR, Acker L, El-Shamayleh Y, Inoue KI, Ohayon S, Schmid MC. Nonhuman primate optogenetics: recent advances and future directions. Journal of Neuroscience. 2017 Nov 8;37(45):10894-903; Tremblay et al, Neuron 2020)

We thank the reviewer for pointing us to these papers. We have added a more thorough description of what we meant by lack of rhesus-compatible constructs in that Appendix.

“However, other challenges exist with using optogenetics as an inactivation method in nonhu- man primates, including difficulty reliably affecting behavior [12]. While several constructs for rhesus macaques have been developed [13], [14], reports of successfully inducing be- havioral effects have a small effect size and are less numerous than might be expected [12], and several null results have been published [15]–[17]. Other remaining challenges include the need to develop a head-mounted, battery powered light delivery system for multi-day delivery of light and difficulty integrating illumination with simultaneous chronic neuro- electrophysiology.”

For Figure 5b, only pairwise comparison results from monkey U (L11-14) are shown. It is unclear why such results from monkey H were shown in Figure 5a but not in 5b.

We thank the reviewer for pointing out this unconventional one monkey result. As described in the original submission, we previously omitted Monkey H from the analysis in Figure 5b (now Figure 7) since some of the lesions were closely spaced together, preventing well defined pre- and post- lesion rates of turnover. Never-the-less we have included Monkey H in all the revised analysis and believe even the less cleanly separated data shows useful indications of neuron loss or silencing evoked by the lesion.

Behavioral data (during a motor task) from the awake behaving monkeys (U and H) would greatly strengthen the claim that this lesioning method is capable of creating a behavioral effect and can be adopted to study the relationship between neural function and behavior outcomes.

While we are grateful for the reviewer’s interest in the application of our lesioning technique to studies involving behavior, a behavioral analysis of the effects of our electrolytic lesions falls be- yond the scope of this Tools and Resources manuscript. We would also like to point out that we do not claim that we have achieved a behavioral deficit in this manuscript.

Figure 2 would benefit from an illustration of the Utah array placement and the location of the sites used for lesioning. The authors can either overlay the illustrations on the current ex-vivo and histology images or create a separate schematic to demonstrate that for the readers. Also, Figure 2B needs to be replaced with one without the folds to avoid confusion for the readers.

We have added Figure 2 - figure supplement 1, which shows both the location within the Utah array of the two electrodes used to create the lesions as well as the relative size of the surface area of the lesion and the array. Unfortunately, as the lesion was created under the array, the exact location of the array relative to the lesion is unknown.

As mentioned above, Figure 2B is the only histological image from that lesion. We hope that the markings in the image as well as the caption sufficiently explain that the folds are unrelated to the lesion itself.

Figure 3, the conical region is not well delineated. Data across animals and lesion volume with respect to different parameters should be included.

We have included a supplemental figure, Figure 3 - figure supplement 1, where we have used a dashed white line to clearly indicate the area of damaged parenchyma, in case it was not clear in Figure 3a. We have also added volume estimates from lesions across animals and different param- eters. The ex vivo estimates are shown in Figure 4 and the in vivo estimates are shown in Figure 5.

Figure 4: it is not clear what is being communicated, and where the voltage traces are from.

We thank the reviewer for noting this confusion. We have added some lines in the text to explain what the voltage traces show, both in the caption to Fig. 6 and in the text on Page 7, Line 238.

“Traces only capture the values while the lesioning device was turned on (45 seconds for most lesions and 50 seconds for lesion H200120). A) Voltage traces. Discontinuity at the beginning of the traces indicates transient voltages that were too rapid to be captured by the voltmeter, lasting between 0.13 and 0.33 s. The fluctuating voltages, especially the rapid in- crease in voltage at the beginning of lesioning, emphasize the importance of using a current source to deliver consistent amounts of current into the brain.”

“The voltage across the microelectrode array fluctuated much more than the current did, em- phasizing that we made the correct choice in using a current source to ensure delivery of consistent amounts of current into the brain (Fig. 6figure.caption.19).”

Figure 5: why did the authors choose to use matching units as a measure of the lesion? It is surprising that there are still units on the location that the authors claim to be a lesion. To clarify that it would be helpful to show the location of the lesion in Figure 4a. Also, what can we conclude about the lesion induction when we see units on the lesion electrode? The change in unit match shows that there is a change in the network (although the authors need to show control for that so we know those changes don’t happen due to natural dynamics). It is not clear what is the time duration for pre-pre and post-post (i.e. minutes, seconds, hours). Do these comparisons come from the same time frame or are they coming from two fragments of time for both pre and post- conditions?

Aside from post-mortem histology and tissue assays, there is no good way to confirm neuron loss with chronically implanted electrode arrays in nonhuman primates. Waveforms were chosen as they are the one readily isolated physical measure of the system we are injuring. Although functional measures of activity could indicate neuron loss (topic of following papers), there are many conceivable changes in firing rate patterns that could manifest spuriously as loss, making the estimation of loss even more ambiguous and challenging this way.

We believe the new Figure 7 will make the procedure much more clear, while also providing the control requested by the reviewer, illustrating that new statistical categories of altered waveforms emerge during a lesion, beyond those associated with typical changes in waveform composition within multi-unit recordings seen during recording sample turnover fom healthy animals. We further note that by confining this analysis to four day spans at most, we have limited the impact of daily sample turnover described in the literature (Gallego, 2020).

The time duration for pre-session versus pre-session (pre-post and post-post), is some multiple of the approximate 24 hours between each daily recording session. Therefore, since restricting our- selves to four days separation, between 24 and 96 hours. Spikes are sampled from successful trial periods (so on the order of seconds, compiled into minutes across the whole recording session). Although already described in the main text, these points have been reemphasized in the figure legend.

CNO (line 931) needs to be explained.

We thank the reviewer for this point. We have defined CNO and its relevance in Appendix 2.

“Additionally, chronic inactivation over days may be logistically challenging, as the half life of clozapine N-oxide (CNO, a ligand used to activate DREADD receptors) is on the order of hours.”

eLife assessment

This paper reports a valuable new method for creating localized damage to candidate brain regions for functional and behavioral studies. The authors present solid support for their ability to create long-term local lesions with mm spatial resolution. The paper is likely to be of broad interest to brain researchers working to establish causal links between neural circuits and behavior.

Reviewer #1 (Public Review):

In the paper, the authors illustrated a novel method for Electrolytic Lesioning through a microelectronics array. This novel lesioning technique is able to perform long-term micro-scale local lesions with a fine spatial resolution (mm). In addition, it allows a direct comparison of population neural activity patterns before and after the lesions using electrophysiology. This new technique addresses a recent challenge in the field and provides a precious opportunity to study the natural reorganization/recovery at the neuronal population level after long-term lesions. It will help discover new causal insights investigating the neural circuits controlling behavior.

Comments on revised version:

We appreciate the revisions made by the authors in response to our comments on the previous version of their manuscript. They carefully addressed the majority of the concerns and performed additional experiments. The new figure illustrating the lesion volume as a function of electrolytic lesioning parameters provides a valuable reference for future experiments. In addition, the latest results on different versions of passive multielectrode probes, U-probe, demonstrate that the technique is applicable beyond the specific technical setup they employ. Overall, we believe that the revised manuscript is significantly improved.

Reviewer #2 (Public Review):

This work by Bray et al. presented a customized way to induce small electrolytic lesions in the brain using chronically implanted intracortical multielectrode arrays. This type of lesioning technique has the benefit of high spatial precision and low surgical complexity while allowing simultaneous electrophysiology recording before, during, and after the lesion induction. The authors have validated this lesioning method with a Utah array, both ex vivo and in vivo using pig models and awake-behaving rhesus macaques. Given its precision in controlling the lesion size, location, and compatibility with multiple animal models and cortical areas, the authors believe this method can be used to study cortical circuits in the presence of targeted neuronal inactivation or injury and to establish causal relationships before behavior and cortical activity.

Strengths:

- Overall the techniques, parameters, and data analysis methods are better described in the revised version.

- The authors added the section "Relationship Between Applied Current and Lesion Volume" as well as Figure 4 and 5 to address our comments regarding parameter testing. Multiple combinations of current amplitude and duration were tested and the induced lesion volumes were estimated, providing a better picture of why certain parameters were chosen for in vivo studies.

- The authors added Figure 7 which addressed our comment "more evidence is needed to suggest robust neuronal inactivation or termination in rhesus macaques after electrolytic lesioning." They went into more details to explain the observed changes in pairwise comparisons of spike waveforms (difference in projected radii). Particularly in Fig 7C, they identified a new cluster from the pre-post lesioning group, which effectively represented neuronal loss from the<br /> recorded population.

- The authors discussed their method in the context of other literature and stating its strength and limitation.

Major comments:

-The lack of histology limits the validation of lesion induction, ideally cell loss and neuronal loss in vivo needs to be quantified. In addition based on the lack of access to histology, it is not clear how the lesion volumes are calculated which also impacts the scientific rigor of the work. The authors mention that layers 2/3 and maybe 4 have been impacted. The lack of information on the extent of the lesion severely limits the use of their technique for neuroscience experiments.

-The lack of histology in combination with behavioral measures still limits the impact of the paper in the context of NHP research.

- Figure 5 involves fitting an exponential model to the generated lesion volume given the applied current amplitude and duration. However, the data from ex vivo sheep and pig cortex with the same current amplitude & three durations showed very large variability in lesion volume at Time = 2min (larger than the difference from 2 to ~2.2min). Very limited data points exist for the other two parameter combinations. These may suggest that the exponential fit is not the best model in this scenario.

- Regarding the comment on neuronal inactivation, the authors still did not show any evidence of single unit activity loss or changes in local field potential/multi-unit activity from the region being lesioned.

- Regarding this comment "The lesioning procedure was performed in Monkey F while sedated, but no data was presented for Monkey F in terms of lesioning parameters, lesion size, recorded electrophysiology, histological, or behavioral outcomes. It is also unclear if Monkey F was in a terminal study" the authors explained that "a lesion was performed on a sedated rhesus macaque (monkey F) who was subsequently euthanized due to unrelated health complications, in order to further verify safety before use in awake-behaving rhesus" but still no histology data is shown regarding monkey F to demonstrate this verification. Given that NHPs are highly valuable resources, it's important to make use of all collected data and to show that the induced lesion is comparable to those in the pig cortex.

Reviewer #1 (Public Review):

This paper describes "Ais", a new software tool for machine-learning-based segmentation and particle picking of electron tomograms. The software can visualise tomograms as slices and allows manual annotation for the training of a provided set of various types of neural networks. New networks can be added, provided they adhere to a Python file with an (undescribed) format. Once networks have been trained on manually annotated tomograms, they can be used to segment new tomograms within the same software. The authors also set up an online repository to which users can upload their models, so they might be re-used by others with similar needs. By logically combining the results from different types of segmentations, they further improve the detection of distinct features. The authors demonstrate the usefulness of their software on various data sets. Thus, the software appears to be a valuable tool for the cryo-ET community that will lower the boundaries of using a variety of machine-learning methods to help interpret tomograms.

eLife assessment

This work describes a new software platform for machine-learning-based segmentation of and particle-picking in cryo-electron tomograms. The program and its corresponding online database of trained models will allow experimentalists to conveniently test different models and share their results with others. The paper provides solid evidence that the software will be valuable to the community.

Reviewer #2 (Public Review):

Summary:

Last et al. present Ais, a new deep learning-based software package for the segmentation of cryo-electron tomography data sets. The distinguishing factor of this package is its orientation to the joint use of different models, rather than the implementation of a given approach. Notably, the software is supported by an online repository of segmentation models, open to contributions from the community.

The usefulness of handling different models in one single environment is showcased with a comparative study on how different models perform on a given data set; then with an explanation of how the results of several models can be manually merged by the interactive tools inside Ais.

The manuscripts present two applications of Ais on real data sets; one is oriented to showcase its particle-picking capacities on a study previously completed by the authors; the second one refers to a complex segmentation problem on two different data sets (representing different geometries as bacterial cilia and mitochondria in a mouse neuron), both from public databases.

The software described in the paper is compactly documented on its website, additionally providing links to some YouTube videos (less than an hour in total) where the authors videocapture and comment on major workflows.

In short, the manuscript describes a valuable resource for the community of tomography practitioners.

Strengths:

A public repository of segmentation models; easiness of working with several models and comparing/merging the results.

Weaknesses:

A certain lack of concretion when describing the overall features of the software that differentiate it from others.

Reviewer #3 (Public Review):

Summary:

In this manuscript, Last and colleagues describe Ais, an open-source software package for the semi-automated segmentation of cryo-electron tomography (cryo-ET) maps. Specifically, Ais provides a graphical user interface (GUI) for the manual segmentation and annotation of specific features of interest. These manual annotations are then used as input ground-truth data for training a convolutional neural network (CNN) model, which can then be used for automatic segmentation. Ais provides the option of several CNNs so that users can compare their performance on their structures of interest in order to determine the CNN that best suits their needs. Additionally, pre-trained models can be uploaded and shared to an online database.

Algorithms are also provided to characterize "model interactions" which allows users to define heuristic rules on how the different segmentations interact. For instance, a membrane-adjacent protein can have rules where it must colocalize a certain distance away from a membrane segmentation. Such rules can help reduce false positives; as in the case above, false negatives predicted away from membranes are eliminated.

The authors then show how Ais can be used for particle picking and subsequent subtomogram averaging and for the segmentation of cellular tomograms for visual analysis. For subtomogram averaging, they used a previously published dataset and compared the averages of their automated picking with the published manual picking. Analysis of cellular tomogram segmentation was primarily visual.

Strengths:

CNN-based segmentation of cryo-ET data is a rapidly developing area of research, as it promises substantially faster results than manual segmentation as well as the possibility for higher accuracy. However, this field is still very much in the development and the overall performance of these approaches, even across different algorithms, still leaves much to be desired. In this context, I think Ais is an interesting package, as it aims to provide both new and experienced users with streamlined approaches for manual annotation, access to a number of CNNs, and methods to refine the outputs of CNN models against each other. I think this can be quite useful for users, particularly as these methods develop.

Weaknesses:

Whilst overall I am enthusiastic about this manuscript, I still have a number of comments:

On page 5, paragraph 1, there is a discussion on human judgement of these results. I think a more detailed discussion is required here, as from looking at the figures, I don't know that I agree with the authors' statement that Pix2pix is better. I acknowledge that this is extremely subjective, which is the problem. I think that a manual segmentation should also be shown in a figure so that the reader has a better way to gauge the performance of the automated segmentation.

On page 7, the authors mention terms such as "emit" and "absorb" but never properly define them, such that I feel like I'm guessing at their meaning. Precise definitions of these terms should be provided.

For Figure 3, it's unclear if the parent models shown (particularly the carbon model) are binary or not. The figure looks to be grey values, which would imply that it's the visualization of some prediction score. If so, how is this thresholded? This can also be made clearer in the text.

Figure 3D was produced in ChimeraX using the hide dust function. I think some discussion on the nature of this "dust" is in order, e.g. how much is there and how large does it need to be to be considered dust? Given that these segmentations can be used for particle picking, this seems like it may be a major contributor to false positives.

Page 9 contains the following sentence: "After selecting these values, we then launched a batch particle picking process to determine lists of particle coordinates based on the segmented volumes." Given how important this is, I feel like this requires significant description, e.g. how are densities thresholded, how are centers determined, and what if there are overlapping segmentations?

The FSC shown in Figure S6 for the auto-picked maps is concerning. First, a horizontal line at FSC = 0 should be added. It seems that starting at a frequency of ~0.045, the FSC of the autopicked map increases above zero and stays there. Since this is not present in the FSC of the manually picked averages, this suggests the automatic approach is also finding some sort of consistent features. This needs to be discussed.

Page 11 contains the statement "the segmented volumes found no immediately apparent false positive predictions of these pores". This is quite subjective and I don't know that I agree with this assessment. Unless the authors decide to quantify this through subtomogram classification, I don't think this statement is appropriate.

In the methods, the authors note that particle picking is explained in detail in the online documentation. Given that this is a key feature of this software, such an explanation should be in the manuscript.

eLife assessment

In this study, camera trapping and species distribution models are used to show that human disturbance in mountain forests in the eastern Himalayas pushes medium-sized and large mammal species into narrower habitat space, thus increasing their co-occurrence. While the collected data provide a useful basis for further work, the study presents incomplete evidence to support the claim that increased co-occurrence may indicate positive interactions between species.

Reviewer #1 (Public Review):

Summary:

This study examines the spatial and temporal patterns of occurrence and the interspecific associations within a terrestrial mammalian community along human disturbance gradients. They conclude that human activity leads to a higher incidence of positive associations.

Strengths:

The theoretical framework of the study is brilliantly introduced. Solid data and sound methodology. This study is based on an extensive series of camera trap data. Good review of the literature on this topic.

Weaknesses:

The authors do not delve into the different types of association found in the study. A more ecological perspective explaining why certain species tend to exhibit negative associations and why others show the opposite pattern (and thus, can be used as indicator species) is missing. Also, the authors do not clearly distinguish between significant (true) non-random associations and random associations.

Anthropogenic pressures can shape species associations by increasing spatial and temporal co-occurrence, but above a certain threshold, the positive influence of human activity in terms of species associations could be reverted. This study can stimulate further work in this direction.

Reviewer #2 (Public Review):

Summary:

This study analyses camera trapping information on the occurrence of forest mammals along a gradient of human modification of the environment. The key hypotheses are that human disturbance squeezes wildlife into a smaller area or their activity into only part of the day, leading to increased co-occurrence under modification. The method used is joint species distribution modelling (JSDM).

Strengths:

The data source seems to be very nice, although since very little information is presented, this is hard to be sure of. Also, the JSDM approach is, in principle, a nice way of simultaneously analysing the data.

Weaknesses:

The manuscript suffers from a mismatch of hypotheses and methods at two different levels.

(1) At the lower level, we would need to better understand what the individual species do and "like" (their environmental niche).

(2) The hypothesis clearly asks for an analysis of the statistical interaction between human disturbance and co-occurrence. Yet, the study is not set up in a way to test this directly.

The hypotheses point towards presenting the spatial and the temporal niche, and how it changes, species for species, under human disturbance. To this, one could then add the layer of interspecific associations.

The change in activity and space use could be analysed by looking at the activity times and spatial distribution directly. If biotic interactions change along the disturbance gradient, then observed data are already the outcome of such changed interactions. We thus cannot use the data to infer them! But we can show, for each species, that the habitat preferences change along the disturbance gradient - or not, as the case may be.

The per-species models are simplistic: the predictors are only linear, and there are no statistical interactions. It is unclear how spatial autocorrelations of residuals were treated, although they form the basis for the association analysis. Why are times of day and day of the year not included as predictors IN INTERACTION with niche predictors and human disturbance, since they represent the temporal dimension on which niches are hypothesised to change?

The discussion has little to add to the results. The complexity of the challenge (understanding a community-level response after accounting for species-level responses) is not met, and instead substantial room is given to general statements of how important this line of research is. What is the advance in ecological understanding at the community level?

Author response:

The following is the authors’ response to the original reviews.

Reviewer #1 (Public Review):

Summary:

This study examines the spatial and temporal patterns of occurrence and the interspecific associations within a terrestrial mammalian community along human disturbance gradients. They conclude that human activity leads to a higher incidence of positive associations.

Strengths:

The theoretical framework of the study is brilliantly introduced. Solid data and sound methodology. This study is based on an extensive series of camera trap data. Good review of the literature on this topic.

Weaknesses:

The authors use the terms associations and interactions interchangeably.

This is not the case. In fact, we state specifically that "... interspecific associations should not be directly interpreted as a signal of biotic interactions between pairs of species…" However, co-occurrence can be an important predictor of likely interactions, such as competition and predation. We stand by our original text.

It is not clear what the authors mean by "associations". A brief clarification would be helpful.

Our specific definition of what is meant here by spatial association can be found in the Methods section. To clarify, the calculation of the index of associations is based on the covariance for the two species of the residuals (epsilon) after consideration of all species-specific response to known environmental covariates. These covariances are modelled to allow them to vary with the level of human disturbance, measured as human presence and human modification. After normalization, the final index of association is a correlation value that varies between -1 (complete disassociation) and +1 (complete positive association).

Also, the authors do not delve into the different types of association found in the study. A more ecological perspective explaining why certain species tend to exhibit negative associations and why others show the opposite pattern (and thus, can be used as indicator species) is missing.

Suggesting the ecological underpinnings of the associations observed here would mainly be speculation at this point, but the associations demonstrated in this analysis do suggest promising areas for the more detailed research suggested.

Also, the authors do not distinguish between significant (true) non-random associations and random associations. In my opinion, associations are those in which two species co-occur more or less than expected by chance. This is not well addressed in the present version of the manuscript.

Results were considered to be non-random if correlation coefficients (for spatial association) or overlap (for temporal association) fell outside of 95% Confidence Intervals. This is now stated clearly in the Methods section.  In Figure 3—figure supplement 1-3 and Figure 4—figure supplement 1-3, p<0.01 levels are also presented.

The obtained results support the conclusions of the study.

Anthropogenic pressures can shape species associations by increasing spatial and temporal co-occurrence, but above a certain threshold, the positive influence of human activity in terms of species associations could be reverted. This study can stimulate further work in this direction.

Reviewer #2 (Public Review):

Summary:

This study analyses camera trapping information on the occurrence of forest mammals along a gradient of human modification of the environment. The key hypotheses are that human disturbance squeezes wildlife into a smaller area or their activity into only part of the day, leading to increased co-occurrence under modification. The method used is joint species distribution modelling (JSDM).

Strengths:

The data source seems to be very nice, although since very little information is presented, this is hard to be sure of. Also, the JSDM approach is, in principle, a nice way of simultaneously analysing the data.

Weaknesses:

The manuscript suffers from a mismatch of hypotheses and methods at two different levels.

(1) At the lower level, we first need to understand what the individual species do and "like" (their environmental niche). That information is not presented, and the methods suggest that the representation of each species in the JSDM is likely to be extremely poor.

The response of each species to the environmental covariates provides a window into their environmental niche, encapsulated in the beta coefficients for each environmental covariate. This information is presented in Figure 2.

(2) The hypothesis clearly asks for an analysis of the statistical interaction between human disturbance and co-occurrence. Yet, the model is not set up this way, and the authors thus do a lot of indirect exploration, rather than direct hypothesis testing.

Our JSDM model is set up specifically to examine the effect of human disturbance on co-occurrence, after controlling for shared responses to environmental variables.  It directly tests the first hypothesis, since, if increase in indices of human disturbance had not tended to increase the measured spatial correlations between species as detected by the model, we would have rejected our stated hypothesis that human modification of habitats results in increased positive spatial associations between species.

Even when the focus is not the individual species, but rather their association, we need to formulate what the expectation is. The hypotheses point towards presenting the spatial and the temporal niche, and how it changes, species for species, under human disturbance. To this, one can then add the layer of interspecific associations.

Examining each species one by one and how each one responds to human disturbance would miss the effects of any meaningful interactions between species.  The analysis presented provides a means to highlight associations that would have been overlooked.  Future research could go on to analyze the strongest associations in the community and the strongest effects of human disturbance so as to uncover the underlying interactions that give rise to them and the mechanisms of human impact.  We believe that this will prove to be a much more productive approach than trying to tackle this problem species by species and pair by pair.

The change in activity and space use can be analysed much simpler, by looking at the activity times and spatial distribution directly. It remains unclear what the contribution of the JSDM is, unless it is able to represent this activity and spatial information, and put it in a testable interaction with human disturbance.

The topic is actually rather complicated. If biotic interactions change along the disturbance gradient, then observed data are already the outcome of such changed interactions. We thus cannot use the data to infer them! But we can show, for each species, that the habitat preferences change along the disturbance gradient - or not, as the case may be.

Then, in the next step, one would have to formulate specific hypotheses about which species are likely to change their associations more, and which less (based e.g. on predator-prey or competitive interactions). The data and analyses presented do not answer any of these issues.

We suggest that the so-called “simpler” approach described above is anything but simple, and this is precisely what the Joint Species Distribution Model improves upon.  As pointed out in the Introduction, simply examining spatial overlap is not enough to detect a signal of meaningful biotic interaction, since overlap could be the result of similar responses to environmental variables.  With the JSDM approach, this would not be considered a positive association and would then not imply the possible existence of meaningful interaction.

Another more substantial point is that, according to my understanding of the methods, the per-species models are very inappropriate: the predictors are only linear, and there are no statistical interactions (L374). There is no conceivable species in the world whose niche would be described by such an oversimplified model.

While interaction terms can be included in the JSDM, this would considerably increase the complexity of the models.  In previous work, we have found no strong evidence for the importance of interaction terms and they do not improve the performance of the models.

We have no idea of even the most basic characteristics of the per-species models: prevalences, coefficient estimates, D2 of the model, and analysis of the temporal and spatial autocorrelation of the residuals, although they form the basis for the association analysis!

The coefficient estimates for response to environmental variables used in the JSDM are provided in Figure 2 and Figure 2—source data 1.

Why are times of day and day of the year not included as predictors IN INTERACTION with niche predictors and human disturbance, since they represent the temporal dimension on which niches are hypothesised to change?

Also, all correlations among species should be shown for the raw data and for the model residuals: how much does that actually change and can thus be explained by the niche models?

The discussion has little to add to the results. The complexity of the challenge (understanding a community-level response after accounting for species-level responses) is not met, and instead substantial room is given to general statements of how important this line of research is. I failed to see any advance in ecological understanding at the community level.

We agree that the community-level response to human disturbance is a complex topic, and we believe it is also a very important one.  This research and its support of the spatial compression hypothesis, while not providing definitive answers to detailed mechanisms, opens up new lines of inquiry that makes it an important advance.  For example, the strong effects of human disturbance on certain associations that were detected here could now be examined with the kind of detailed species by species and pair by pair analysis that this reviewer appears to demand.

Reviewer #1 (Recommendations For The Authors):

L27 indicates instead of "idicates".

We thank the reviewer for catching that error.

L64 I would refer to potential interactions or just associations. It is always hard to provide evidence for the existence of true interactions.

We have revised to “potential interactions” to qualify this statement.

L69 Suggestion: distort instead of upset.

We thank the reviewer for catching that error.

L70-71 Here, authors use the term associations. Please, be consistent with the terminology throughout the manuscript.

We thank the reviewer for raising this important point.  The term “co-occurrence” appears to be used inconsistently in the literature, so we have tried to refer to it only when referencing the work of us. For us, co-occurrence means “spatial overlap” without qualification as to whether it is caused by interaction or simply by similar responses to environmental factors (see Blanchet et al. 2020, Argument 1). In our view, interactions refer to biotic effects like predation, competition, commensalism, etc., while associations are the statistical footprint of these processes.   In keeping with this understanding, in Line 73, we changed "association" to the stronger word "interaction," but in Line 76, we keep the words "spatiotemporal association", which is presumed to be the result of those interactions. In Line 91, we have changed “interactions” to “associations,” as we do not believe interactions were demonstrated in that study. 

L76 "Species associations are not necessarily fixed as positive or negative..." This sentence is misleading. I would say that species associations can vary across time and space, for instance along an environmental gradient.

We thank the reviewer for pointing out the potential for confusion.  In Line 79, we have changed as suggested.

L78 "Associations between free-ranging species are especially context-dependent" Loose sentence. Please, explain a bit further.

We have changed the sentence to be more specific; ”Interactions are known to be context-dependent; for example, gradients in stress are associated with variation in the outcomes of pairwise species interactions.”

L83-85 This would be a good place to introduce the 'stress gradient' hypothesis, which has also been applied to faunal communities in a few studies. According to this hypothesis, the incidence of positive associations should increase as environmental conditions harden.

In our review of the literature, we find that the stress gradient hypothesis is somewhat controversial and does not receive strong support in vertebrates.  We have added the phrase “…the controversial stress-gradient hypothesis predicts that positive associations should increase as environmental conditions become more severe…”

L86-88 Well, overall, the number of studies examining spatiotemporal associations in vertebrates is relatively small. That is, bird associations have not received much more attention than those of mammals. I find this introductory/appealing paragraph a bit rough. I think the authors can do better and find a better justification for their work.

We thank the reviewer for the comments.  We have rewritten the paragraph extensively to make it clearer and to provide a stronger justification for the study.

L106 "[...] resulting in increased positive spatial associations between species" I'd say that habitat shrinking would increase the level of species clustering or co-occurrence, but in my opinion, not necessarily the incidence of positive associations. It is not clear to me if the authors use positive associations as a term analogous to co-occurrence.

We thank the reviewer for raising this very important distinction.  Habitat shrinking would increase levels of species co-occurrence, but this is not particularly interested.  We wanted to test whether there were effects on species interactions, as revealed by associations.  We find that the terms association and co-occurrence are used somewhat loosely in the literature and so have made some new effort to clarify and systematize this in the manuscript.  For example, there appear to be a differences in the way “co-occurrence” is used in Boron 2023 and in Blanchet 2020. We do not use the term "positive spatial association" as analogous to "spatial co-occurrence.". Spatial co-occurrence, which for us has the meaning of spatial overlap, could simply be the result of similar reactions to environmental co-variates, not reflecting any biotic interaction. Joint Species Distribution Models enable the partitioning of spatial overlap and segregation into that which can be explained by responses to known environmental factors, and that which cannot be explained and thus might be the result of biotic interactions.  It is only the latter that we are calling spatial association, which can be positive or negative.   These associations may be the statistical footprint of biotic interactions.

Results:

Difference between random and non-random association patterns. It is not clear to me if the reported associations are significant or not. The authors only report the sign of the association (either positive or negative) but do not clarify if these associations indicate that two species coexist more or less than expected by chance. In my opinion, that is the difference between true ecological associations (e.g., via facilitation or competition effects) and random co-existence patterns. This is paramount and should be addressed in a new version of the manuscript.

This information is provided in Figure 3—figure supplement 1,2,3 and Figure 4—figure supplement 1,2,3.  This is referenced in the text as follows, “… correlation coefficients for 18 species pairs were positive and had a 95 % CI that did not overlap zero, and the number increased to 65 in moderate modifications but dropped to 29 at higher modifications" and so on. This criterion for significance (ie., greater than expected by chance) is now stated at the end of the Materials and methods.  In Figure 3—figure supplement 1,2,3 and Figure 4—figure supplement 1,2,3, those correlations that were significant at p<0.01 are also shown.

I am also missing a more ecological explanation for the observed findings. For instance, the top-ranked species in terms of negative associations is the red fox, whereas the muntjac seems to be the species whose presence can be used as an indicator for that of other species. What are the mechanisms underlying these patterns? Do red foxes compete for food with other species? Do the species that show positive associations (red goral, muntjac) have traits or a diet that are more different from those of other species? More discussion on these aspects (role of traits and the trophic niche) would be necessary to better understand the obtained results.

The purpose of this paper was to test the compression hypotheses, and we have tried to keep that as the focus.  However, the analysis does open up interesting lines of inquiry for future research to decipher the details of the interactions between species and the mechanisms by which human disturbance facilitates or disrupts these interactions. The reviewer raises some interesting possibilities, but at this point, any discussion along these lines would be largely speculation and could lengthen the paper without great benefit. 

Reviewer #2 (Recommendations For The Authors):

The manuscript should be accompanied by all data and code of analysis.

All data and RScripts have been made available in Science Data Bank: https://doi.org/10.57760/sciencedb.11804.

The sentence "not much is known" is weak: it suggests the authors did not bother to quantify what IS known, and simply waved any previous knowledge aside. Surely we have some ideas about who preys on whom, and which species have overlapping resource requirements (e.g., due to jaw width). For those, we would expect a particularly strong signal, if the association is indeed indicative of interactions.

We believe that the reviewer is referring to the statement in Line 90-92 about the lack of understanding of the resilience of terrestrial mammal associations to human disturbance.  We have added a reference to one very recent publication that addresses the issue (Boron et al., 2023), but otherwise we stand by our statement. We have, however, added a qualifier to make it clear that we did indeed look for previous knowledge; "However, a review of the literature indicates that ...."

Figures:

Fig. 1. This reviewer considers that this is too trivial and should be deleted.

This is a graphical statement of the hypotheses and may be helpful to some readers.

Fig. 2. Using points with error bars hides any potential information.

Done as suggested.

That only 4 predictors are presented is unacceptably oversimplified.

Only 4 predictors are included because, in previous work, we found that adding additional predictors or interactions did little to improve the model’s performance (Li et al. 2018, 2021 and 2022) and could lead to over-fitting.

Fig. 5. and 6. aggregate extremely strongly over species; it remains unclear which species contribute to the signal, and I guess most do not.

The number of detection events presented in Table 1 should help to clarify the relative contribution of each species to the data presented in Figures 5 and 6.

This reviewer considers that the introduction 'oversells' the paper.

L55: can you give any such "unique ecological information"

L60: Lyons et al. (Kathleen is the first name) has been challenged by Telford et al. (2016 Nature) as methodologically flawed.

The first name has been deleted.  The methodological flaw has to do with interpretation of the fossil record and choice of samples, not with the need to partition shared environmental preferences and interactions.

L61 contradicts line 64: Blanchet et al. (2022, specifying some arguments from Dormann et al. 2018 GEB) correctly point out that logically one cannot infer the existence or strength from co-occurrence data. It is thus wrong to then claim (citing Boron et al.) that such data "convey key information about interactions". The latter statement is incorrect. A tree and a beetle can have extremely high association and nothing to do with each other. Association does not mean anything in itself. When two species are spatially and temporally non-overlapping, they can exhibit perfect "anti-association", yet, by the authors' own definition, cannot interact.

We believe that the reviewer’s concerns arise from a misunderstanding of how we use the term association.  In our usage, an association is not the same as co-occurrence or overlap, which may simply be the result of shared responses to environmental variables.  The co-occurring tree and beetle would not be found to have any association in our analysis, only shared environmental sensitivities.  In contrast, associations can be the statistical footprint of interactions, and would be overlaid onto any overlap due to similar responses to the environment.  In the case of negative associations, such as might be the result of competitive exclusion or avoidance of predators, the two species would share environmental responses but show lower than expected spatial overlap.  Even though they might be only rarely found in the same vicinity, they would indeed be interacting when they were together.

Joint Species Distribution Models "allow the partitioning of the observed correlation into that which can be explained by species responses to environmental factors... and that which remains unexplained after controlling for environmental effects and which may reflect biotic interactions." (Garcia Navas et al. 2021). It is the latter that we are calling “associations.”

L63: Gilbert reference: Good to have a reference for this statement.

This point is important, but the reviewer’s comments below have made it clear that it is even more important to point out that strong interactions should be expected to lead to significant associations.  We have added a statement to clarify this.

L70-72: Incorrect, interactions play a role, not associations (which are merely statistical).

In this, we agree, and we have revised the statement to refer to interactions, not associations. In our view, an interaction is a biological phenomenon, while an association is the resulting statistical signal that we can detect.

L75: Associations tell us nothing, only interactions do. Since these can not be reliably inferred, this statement and this claim are wrong.

We thank the reviewer for raising this point, but we beg to disagree. Strong interactions should be expected to lead to significant associations that can be detected in the data. Associations, which can be measured reliably, are the evidence of potential interactions, and hence associations can tell us a great deal.  We have added a note to this effect after the Gilbert reference above to clarify this point.

However, we do accept that associations must be interpreted with caution. As Blanchet et al. 2020 explain, " …the co-occurrence signals (e.g. a significant positive or negative correlation value) estimated from these models could originate from any abiotic factors that impact species differently. Therefore, this correlation cannot be systematically interpreted as a signal of biotic interactions, as it could instead express potential non-measured environmental drivers (or combinations of them) that influence species distribution and co-distribution.”  Or alternatively an association could be the result of interaction with a 3rd species. 

L87: Regarding your claim, how would you know you DO understand? For that, you need to formulate an expectation before looking at the data and then show you cannot show what you actually measure. (Jaynes called this the "mind-projection fallacy".)

We are not sure if the reviewer is criticizing our paper or the entire field of community ecology.  Perhaps it is the statement that “….resilience of interspecific spatiotemporal associations of terrestrial mammals to human activity remains poorly understood….”  Since we are confident that the reviewer believes that mammals do interact, we guess that it is the term “association” that is questioned.  We have revised this to “…the impacts of human activity on interspecific interactions of terrestrial mammals remains poorly understood…” 

In this particular case, we did formulate an expectation before looking at the data, in the form of the two formal hypotheses that are clearly stated in the Introduction and illustrated in Figure 1. If the hypotheses had not been supported, then we would have accepted that we do not understand. But as the data are consistent with the hypotheses, we submit that we do understand a bit more now.

eLife assessment

This study provides a valuable resource by thoroughly benchmarking multiple sequencing-based tRNA quantification methods. The suggested best practice is supported by solid evidence from in silico experiments in multiple scenarios. The major weakness of the manuscript is the incomplete validation of newly generated experimental datasets.

Reviewer #1 (Public Review):

Summary:

In the manuscript titled "Benchmarking tRNA-Seq quantification approaches by realistic tRNA-Seq data simulation identifies two novel approaches with higher accuracy," Tom Smith and colleagues conducted a comparative evaluation of various sequencing-based tRNA quantification methods. The inherent challenges in accurately quantifying tRNA transcriptional levels, stemming from their short sequences (70-100nt), extensive redundancy (~600 copies in human genomes with numerous isoacceptors and isodecoders), and potential for over 100 post-transcriptional chemical modifications, necessitate sophisticated approaches. Several wet-experimental methods (QuantM-tRNA, mim-tRNA, YAMAT, DM-tRNA, and ALL-tRNA) combined with bioinformatics tools (bowtie2-based, SHRiMP, and mimseq) have been proposed for this purpose. However, their practical strengths and weaknesses have not been comprehensively explored to date. In this study, the authors systematically assessed and compared these methods, considering factors such as incorrect alignments, multiple alignments, misincorporated bases (experimental errors), truncated reads, and correct assignments. Additionally, the authors introduced their own bioinformatic approaches (referred to as Decision and Salmon), which, while not without flaws (as perfection is unattainable), exhibit significant improvements over existing methods.

Strengths:

The manuscript meticulously compares tRNA quantification methods, offering a comprehensive exploration of each method's relative performance using standardized evaluation criteria. Recognizing the absence of "ground-truth" data, the authors generated in silico datasets mirroring common error profiles observed in real tRNA-seq data. Through the utilization of these datasets, the authors gained insights into prevalent sources of tRNA read misalignment and their implications for accurate quantification. Lastly, the authors proposed their downstream analysis pipelines (Salmon and Decision), enhancing the manuscript's utility.

Weaknesses:

As discussed in the manuscript, the error profiles derived from real-world tRNA-seq datasets may still harbor biases, as reads that failed to "align" in the analysis pipelines were not considered. Additionally, the authors did not validate the efficacy of their "best practice" pipelines on new real-world datasets, preferably those generated by the authors themselves. Such validation would not only confirm the improvements but also demonstrate how these pipelines could alter biological interpretations.<br /> Because tRNA-sequencing methods have not been widely used (compared to mRNA-seq), many readers would not be familiar with the characteristics of different methods introduced in this study (QuantM-tRNA, mim-tRNA, YAMAT, DM-tRNA, and ALL-tRNA; bowtie2-based, SHRiMP, and mimseq; what are the main features of "Salmon?"). The manuscript will read better when the basic features of these methods are described in the manuscript, however brief.

Reviewer #2 (Public Review):

Summary:

The authors provided benchmarking study results on tRNA-seq in terms of read alignment and quantification software with optimal parameterization. This result can be a useful guideline for choosing optimal parameters for tRNA-seq read alignment and quantification.

Strengths:

Benchmarking results for read alignment can be a useful guideline to choose optimal parameters and mapping strategy (mapping to amino acid) for various tRNAseq.

Weaknesses:

The topic is highly specific, and the novelty of the analysis might not be widely useful for general readers.

Some details of the sequencing data analysis pipeline are not clear for general readers:

(1) The explanation of the parameter D for bowtie2 sounds ambiguous. "How much effort to expend" needs to be explained in more detail.

(2) Please provide optimal parameters (L and D) for tRNA-seq alignment.

(3) I think the authors chose L=10 and D=100 based on Figure 1A. Which dataset did you choose for this parameterization among ALL-tRNAseq, DM-tRNAseq, mim-tRNAseq, QuantM-tRNA-seq, and YAMAT-seq?

(4) Salmon does not need a read alignment process such as Bowtie2. Hence, it is not clear "Only results from alignment with bowtie2" in Figure legend for Figure 4a.

Author response:

We thank the reviewers for their critical appraisal of our manuscript. We will address the points of confusion and/or lack of clarity in a revised manuscript. We agree with reviewer 1 that applying the best practice pipeline(s) on new experimental data and comparing this approach with current practices would be a useful demonstration of how this alters the biological interpretation. This is something we are in the process of completing but believe this is best addressed in a separate manuscript where we can focus on the associated biological findings, allowing this manuscript to remain focused on the accurate quantification of tRNA-Seq data.

Author response:

The following is the authors’ response to the original reviews.

eLife assessment

This manuscript provides useful information about the lipid metabolite 15d-PGJ2 as a potential regulator of myoblast senescence. The authors provide experimental evidence that 15d-PGJ2 inhibits myoblast proliferation and differentiation by binding and regulating HRas. However, the manuscript is incomplete in its current form, as it lacks robust support from the data regarding the main conclusions related to senescence and technical concerns related to the senescence models used in this study.

We are grateful to the editors and the reviewers for their time and comments in sharpening the science and the writing of the manuscript. We have attached a detailed response to emphasize that the manuscript does include robust evidence regarding the claims, which could have been missed during the review process. We have provided a better context for these points now.

Public Reviews:

Reviewer #1 (Public Review):

Summary:

The authors show that upon treatment with Doxorubicin (Doxo), there is an increase in senescence and inflammatory markers in the muscles. They also show these genes get upregulated in C2C12 myoblasts when treated with conditioned media or 15d-PGJ2. 15dPGJ2 induces cell death in the myoblasts, decreases proliferation (measured by cell numbers), and decreases differentiation and fusion. 15d-PGJ2 modified Cys184 of HRas, which is required for its activation as indicated by the FRET analysis with RAF RBD. They also showed that 15d-PGJ2 activates ERK signaling, but not Akt signaling, through the electrophilic center. 15d-PGJ2 inhibits Golgi localization of HRAS (only WT, not C181 or C184 mutant). They also showed that expressing the WT HRas followed by 15d-PGJ2 treatment led to a decrease in the levels of MHC mRNA and protein, and this defect is dependent on C184. This is a well-written manuscript with interesting insights into the mechanism of action of 15d-PGJ2. However, some clarification and experiments will help the paper advance the field significantly.

Strengths:

The data clearly shows that 15d-PGJ2 has a negative role in the myoblast cells and that it leads to modification of HRas protein. Moreover, the induction of biosynthetic enzymes in the PGD2 pathway also supports the induction of 15d-PGJ2 in Doxorubicin-treated cells. Both conditioned media experiments and the 15d-PGJ2 experiments show that 15d-PGJ2 could be the active component secreted by the senescent myoblasts.

Weaknesses:

The genes that are upregulated in the muscles upon injection with Doxo are also markers for inflammation. Since Doxo is also known to induce systemic inflammation, it is important to delineate these two effects (inflammatory cells vs senescent cells). The expression of beta Gal and other markers of senescence in the tissue sections will help to delineate these.

As pointed out Doxo induces systemic inflammation along with inducing DNA damage-mediated senescence. Therefore, along with the inflammatory markers of the SASP (CXCL1/2, TNF1α, IL6, PTGS1/2, PTGDS) we also observed an increase in the mRNA levels of canonical markers of DNA damage-mediated senescence. We observed an increase in the mRNA levels of cell cycle and senescence associated proteins p16 and p21 (Fig. 1C). We also observed an increased nuclear accumulation of p21 (Fig. 1A) and increased levels of phosphorylated H2A.X in the nucleus (Fig. 1B).

In Figure 2, where the defect in the differentiation of myoblasts upon treatment with 15d-PGJ2 is shown, most of the cells die within 48 hours at higher concentrations, making it difficult to perform the experiments. This also shows that 15d-PGJ2 was toxic to these cells. Lower concentrations show a decrease in the differentiation based on the lower number of nuclei in fibers and low expression of MyoD, MyoG, and MHC. However, it is unclear if this is due to increased cell death or defective differentiation. It would be a lot more informative if the cell count, cell division, and cell death could be plotted for these concentrations of the drug during the experiment.

We measured the viability of C2C12 cells after 24 hours of treatment with 15d-PGJ2 using the MTT assay and observed that the viability of cells was decreased after treatment with 15d-PGJ2 (10 µM) but not with 15d-PGJ2 (1 µM, 2 µM, 4 µM, or 5 µM) (see Fig. S2A of the updated manuscript). The results and figures of the manuscript have been updated accordingly.

Also, in the myoblast experiments, are the effects of treatment with Dox reversible?

The treatment with Doxorubicin is irreversible as the senescent phenotype was not reversed after withdrawal of Doxorubicin, even after 20 days.

In Figure 3, most of the experiments are done at a high concentration, which induces almost complete cell death within 48 hours.

Figure 3 is an acute experiment for only 1 hour, at which time no cell death was observed. Specifically, we measured the phosphorylation of Erk and Akt proteins after 1 hour of treatment with 15d-PGJ2 (10 µM) during which we did not observe any cell death. 

Even at such a high concentration of 15dPGJ2, the increase in ERK phosphorylation is minimal.

We observe a ~30% increase in the phosphorylation of Erk proteins after treatment with 15d-PGJ2 in 0.2% serum medium compared to treatment with vehicle (DMSO). This is reproducible and significant.

The experiment Figure 4C shows that C181 and C84 mutants of the HRas show higher levels in Golgi compared with WT. However, this could very well be due to the defect in palmitoylation rather than the modification with 15d-PGJ2.

Our data does not suggest higher levels of C184S mutant in the Golgi compared with WT (Fig. S4A). We observed that the ratio of HRas levels in the Golgi to the HRas levels in the plasma membrane were similar in C2C12 cells expressing HRas C184S and HRas WT (Fig. S4A graph columns 1 and 5).

Though the authors allude to the possibility that intracellular redistribution of HRas by 15d-PGJ2 requires C181 palmitoylation, the direct influence of C184 modification on C181 palmitoylation is not shown. To have a meaningful conclusion, the authors need to compare the palmitoylation and modification with 15d-PGJ2.

Palmitoylation of HRas C181S is required for the localization of HRas at the plasma membrane. The inhibition of palmitoylation of C181, either by mutation (C181S) or treatment with protein palmitoyl transferase inhibitor (2-Bromopalmitate), results in the accumulation of HRas at Golgi(Rocks et al., 2005) (Fig. S4A). Modification of HRas at C184 by 15d-PGJ2 (Fig. 3A) could inhibit the palmitoylation of HRas at C181. However, our data does not support this hypothesis as modification of HRas WT by 15d-PGJ2 does not increase the level of HRas at the Golgi, like in the case of inhibition of cysteine palmitoylation due to C181S mutation.

To test if the inhibition of myoblast differentiation depends on HRas, they overexpressed the HRas and mutants in the C2C12 lines. However, this experiment does not take the endogenous HRAs into consideration, especially when interpreting the C184 mutant. An appropriate experiment to test this would be to knock down or knock out HRas (or make knock-in mutations of C184) and show that the effect of 15d-PGJ2 disappears. 

Endogenous HRas (wild type) is present in the C2C12 cells overexpressing the EGFP-tagged HRas constructs. Therefore, we only observe a partial rescue in the differentiation after 15d-PGJ2 treatment in C2C12 cells expressing the C184S mutant (Fig. 4D and E). However, since HRas is expressed under high expression CMV promoter and in the absence of other regulatory elements, the overexpressed constructs do show a dominant effect over the endogenous HRas, showing cysteine mutant dependent inhibition of differentiation of myoblasts after treatment with 15dPGJ2 (Fig. 4D and E).

Moreover, in this specific experiment, it is difficult to interpret without a control with no HRas construct and another without the 15d-PGJ2 treatment.

The mRNA levels of MyoD, MyoG, and MHC in C2C12 cells expressing HRas constructs after treatment with 15d-PGJ2 were normalized to the mRNA levels in C2C12 cells expressing corresponding constructs and were treated with vehicle (DMSO). mRNA levels in C2C12 cells treated with vehicle were not shown as they were normalized to 1. MHC protein levels in C2C12 cells expressing HRas constructs after 15d-PGJ2 treatment were normalized to that in C2C12 cells treated with vehicle (DMSO). Since the hypothesis to study the effect of HRas cysteine mutations on the differentiation of myoblasts after treatment with 15d-PGJ2, C2C12 cells expressing HRas WT serve as adequate control. Fig. 2 shows the effect of 15dPGJ2 on muscle differentiation when HRas was not overexpressed.

Moreover, the overall study does not delineate the toxic effects of 15d-PGJ2 from its effect on the differentiation.

The inhibition of differentiation in C212 cells after treatment with 15d-PGJ2 cannot be attributed to the general toxicity of 15d-PGJ2 in cells. We show that the inhibition of differentiation of myoblasts after 15d-PGJ2 depends on modification of HRas at C184 i.e. failure to modify HRas at C184 (Fig. 3A) and resultant activation (Fig. 3B) by 15d-PGJ2 rescues this inhibition of differentiation of C2C12 cells (Fig. 4D and E), dissecting the inhibition of differentiation of myoblasts by 15d-PGJ2 from general toxic effects of 15d-PGJ2 on cell physiology.

Please note that the effect of 15d-PGJ2 on cell physiology is context-specific. On one hand, 15d-PGJ2 has been shown to exert tumor-suppressor effects by inhibiting the proliferation of ovarian cancer cells and lung adenocarcinoma cells (de Jong et al., 2011; Slanovc et al., 2024), 15d-PGJ2 also exerts pro-carcinogenic effects by induction of epithelial to mesenchymal transition in breast cancer cells MCF7 and inhibition of tumor-suppressor protein p53 in MCF7 and PC-3 cells (Choi et al., 2020; Kim et al., 2010).

Reviewer #2 (Public Review):

Summary:

In this study, Swarang and colleagues identified the lipid metabolite 15d-PGJ2 as a potential component of senescent myoblasts. They proposed that 15d-PGJ2 inhibits myoblast proliferation and differentiation by binding and regulating HRas, suggesting its potential as a target for restoring muscle homeostasis post-chemotherapy.

Strengths:

The regulation of HRas by 15d-PGJ2 is well controlled.

Weaknesses:

The novelty of the study is compromised as the activation of PGD and 15d-PGJ2, as well as the regulation of HRas and cell proliferation, have been previously reported. 

Literature does not support this statement, and it is important to clarify this misimpression for the field as a whole. 

Let us clarify- 

Covalent modification of HRas by 15d-PGJ2 has been reported only twice in the literature(Luis Oliva et al., 2003; Yamamoto et al., 2011) in fibroblasts and neurons respectively. 

Interaction between Hras and 15d-PGJ2 in skeletal muscles has not been shown before, even though both Hras and 15d-PGJ2 are shown to be key regulators of muscle homeostasis. 

Activation of Hras by 15d-PGJ2 was reported first by Luis Oliva et al (Luis Oliva et al., 2003). However, this study does not comment on the functional implications of activation of Hras signaling. 

Recently, our lab contributed to a study where the functional implication of activation of Hras signaling due to covalent modification by 15d-PGJ2 was shown in the maintenance of senescence phenotype (Wiley et al., 2021). 

15d-PGJ2 was shown to inhibit the differentiation of myoblasts by Hunter et al (Hunter et al., 2001). This study hypothesized that the inhibition of myoblast differentiation is via 15d-PGJ2 mediated activation of the PPARγ signaling, the study also showed inhibition of myoblast differentiation independent of PPARγ activity, suggesting the presence of other mechanisms.

This is the first study to show a molecular mechanism where activation of Hras signaling in skeletal myoblasts due to covalent modification by 15d-PGJ2 at C184 of Hras inhibits the differentiation of skeletal myoblasts.

Additionally, there are major technical concerns related to the senescence models, limiting data interpretation regarding the relevance to senescent cells.

Major concerns:

(1) The C2C12 cell line is not an ideal model for senescence study due to its immortalized nature and lack of normal p16 expression. A more suitable myoblasts model is recommended, with a more comprehensive characterization of senescence features.

C2C12 is a good model for DNA damage-based senescence that is used in this manuscript. Several reports in the literature have shown the induction of senescence in C2C12 cells. Moiseeva et al 2023 show induction of senescence in C2C12 cells after etoposide-mediated DNA damage. Moustogiannis et al 2021 show the induction of replicative senescence in C2C12 cells. In this study, we show that C2C12 cells undergo DNA damage-mediated senescence after treatment with Doxo. We measured the induction of senescence in C2C12 cells upon DNA damage using several physiological (Nuclear Size, Cell Size, and SA β-gal) and molecular markers (mRNA levels of p21 and SASP factors (IL6 and TGFβ), protein levels of p21) of senescence (see Fig. 1 of the updated manuscript). The results and the figures in the manuscript have been updated accordingly.

(2) The source of increased PGD or its metabolites in the conditioned medium is unclear. Including other senescence models, such as replicative or oncogeneinduced senescence, would strengthen the study.

Fig. 1E shows time-dependent increase in the expression of PGD2 biosynthetic enzymes in senescent C2C12 cells. Fig. 1F shows an increase in the levels of 15dPGJ2 secreted by senescent C2C12 cells in the conditioned medium. This data shows that senescent C2C12 cells are the source of PGD and its metabolites in the conditioned medium.

Again, C2C12 is not suitable for replicative senescence due to its immortalized status.

We and others have shown that C2C12 cells undergo senescence, and this manuscript only used DNA damage induced senescence.

(3) In the in vivo part, it is unclear whether the increased expression of PTGS1, PTGS2, and PTGDS is due to senescence or other side effects of DOXO.

We concur that this is a limitation of this study and the subsequent work will demonstrate the origin of prostaglandin biosynthesis after treatment with Doxo in vivo.

(4) Figure 2A lacks an important control from non-senescent cells during the measurement of C2C12 differentiation in the presence of a conditioned medium.

Figure 2A tests the effect of prostaglandin PGD2 and its metabolites secreted by the senescent cells on the differentiation of myoblasts. Therefore, we inhibited the synthesis of PGD2 in senescent cells by treatment with AT-56, and then collected the conditioned medium. Conditioned medium collected from senescent C2C12 cells treated with vehicle (DMSO) served as a control for the experiment, whereas differentiation of C2C12 cells without any treatment serves as a positive control.

There is no explanation of how differentiation was quantified or how the fusion index was calculated.

The fusion index was calculated using a published myotube analyzer software (Noë et al., 2022). Appropriate information has been added to the materials and methods section of the manuscript.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

Line 3: Expand SA in "SA β-gal".

The manuscript has been updated accordingly (See line 3).

Line 68: HRas is highly regulated by lipid modifications.

The manuscript has been updated accordingly (See line 67).

Figures

Figure S1A seemed incomplete (maybe some processing issue).

The Figure has been updated in the revised manuscript (See Fig. S1A).

Figure S1B-H are mislabeled.

The figure has been updated in the revised manuscript (See Fig. S1C, D, E, and F).

Figures S1E-H are not mentioned in the manuscript.

The manuscript has been updated accordingly (See line 120).

Many supplementary figures are not cited in the article.

The manuscript has been updated accordingly. (See lines 85, 120, 123, 166, 225, 356, 364, 412, and 413)

Reviewer #2 (Recommendations For The Authors):

(1) Clarify the injection method for Doxorubicin in B6J mice on line 83 (IP or IM).

Mice were injected intraperitoneally with Doxorubicin (as mentioned in the materials and methods, see lines 83 and 794)

(2) Address missing information in figures or figure legends.

There is missing piece in Sup Fig 1A.

The figure has been updated in the revised manuscript (See Fig. S1A).

Correct labels in Sup Fig 1C and 1D.

The figure has been updated in the revised manuscript (See Fig. S1C, D, E, and F).

How would the authors explain the dramatic differences in the morphology of C2C12 cells treated with DOXO between bright field and SA-beta-gal staining images in Sup Fig 1B and 1C.

The SA β-gal image after treatment with Doxo does show a flattened cell morphology. Another field of view from the same experiment has been added in the figure to show the difference in the cell morphology more prominently in the revised manuscript (See Fig. 1H).

Provide explanations for Sup Fig 1E-1G, including the meaning of the y-axis and the blue dots and red lines.

We have provided an explanation for the multiple reaction monitoring mass spectrometry used to measure the concentration of 15d-PGJ2 in the conditioned medium in the revised manuscript (see lines 119-130 and the legends of Fig. S1C, D, and E)

(3) Please review the calculation of qPCR data in Figure 1C for correctness, ensuring reference samples with an average expression level of 1.

The data in Fig. 1C was plotted using 2-ΔCT instead of 2-ΔΔCT to show the variability in the expression of mRNAs isolated from animals treated with Saline.

(4) Please explain the calculation of 15d-PGJ2/cell concentration in Figure 1F and provide raw data for review, considering the substantial changes and small error bars. The method or result section lacks an explanation of how this calculation was performed. Additionally, there is no mention of the cell number count.

All the raw values (concentration of 15d-PGJ2 measured using mass spec and cell numbers counted at the time of collection of conditioned medium) are provided in the supplementary table 1. The standard curve to calculate the concentration of 15dPGJ2 in the conditioned medium is shown in Fig. S1F. The cell number was counted after trypsinization using a hemocytometer on the day of collection of the conditioned medium.

(5) Please clarify how cell number normalization and doubling time calculation were done in Fig 2B. Consider replacing the figure with a growth curve showing confluence on the y-axis for easier interpretation.

Cells were counted every 24 hours and the normalization was done to the number of cells counted on day 0 of the treatment (to consider attaching efficiency and other cell culture parameters). Doubling time was calculated as the reciprocal of the slope of the graph of log2(normalized cell number) vs time.

eLife assessment

This manuscript outlines an interaction between senescence-related 15d-PGJ2 and the proliferation and differentiation of myoblasts, with potential implications for muscle health. This manuscript is useful in understanding the role of lipid metabolite 15d-PGJ2 in myoblast proliferation and differentiation. However, in its current form, the manuscript is incomplete as there are several concerns in the statistical analysis, lack of clarity on the mechanistic details, and concerns about the use of an immortalized C2C12 myoblasts cell line to draw major conclusions related to senescence-associated secreted phenotype.

Reviewer #1 (Public Review):

Summary:

The authors show that upon treatment with Doxorubicin (Doxo), there is an increase in senescence and inflammatory markers in the muscles. They also show these genes get upregulated in C2C12 myoblasts when treated with conditioned media or 15d-PGJ2. 15dPGJ2 induces cell death in the myoblasts, decreases proliferation (measured by cell numbers), and decreases differentiation and fusion. 15d-PGJ2 modified Cys184 of HRas, which is required for its activation as indicated by the FRET analysis with RAF RBD. They also showed that 15d-PGJ2 activates ERK signaling, but not Akt signaling, through the electrophilic center. 15d-PGJ2 inhibits Golgi localization of HRAS (only WT, not C181 or C184 mutant). They also showed that expressing the WT HRas followed by 15d-PGJ2 treatment led to a decrease in the levels of MHC mRNA and protein, and this defect is dependent on C184. This is a well-written manuscript with interesting insights into the mechanism of action of 15d-PGJ2. However, some clarification and experiments will help the paper advance the field significantly.

Strengths:

The data clearly shows that 15d-PGJ2 has a negative role in the myoblast cells and that it leads to modification of HRas protein. Moreover, the induction of biosynthetic enzymes in the PGD2 pathway also supports the induction of 15d-PGJ2 in Doxorubicin-treated cells. Both conditioned media experiments and the 15d-PGJ2 experiments show that 15d-PGJ2 could be the active component secreted by the senescent myoblasts.

Weaknesses:

The genes that are upregulated in the muscles upon injection with Doxo are also markers for inflammation. Since Doxo is also known to induce systemic inflammation, it is important to delineate these two effects (Inflammatory cells vs senescent cells). The expression of beta Gal and other markers of senescence in the tissue sections will help to delineate these.

In Figure 2, where the defect in the differentiation of myoblasts upon treatment with 15d-PGJ2 is shown, most of the cells die within 48 hours at higher concentrations, making it difficult to perform the experiments. This also shows that 15d-PGJ2 was toxic to these cells. Lower concentrations show a decrease in the differentiation based on the lower number of nuclei in fibers and low expression of MyoD, MyoG, and MHC. However, it is unclear if this is due to increased cell death or defective differentiation. It would be a lot more informative if the cell count, cell division, and cell death could be plotted for these concentrations of the drug during the experiment. Also, in the myoblast experiments, are the effects of treatment with Dox reversible?

In Figure 3, most of the experiments are done at a high concentration, which induces almost complete cell death within 48 hours. Even at such a high concentration of 15dPGJ2, the increase in ERK phosphorylation is minimal.

The experiment Figure 4C shows that C181 and C84 mutants of the HRas show higher levels in Golgi compared with WT. However, this could very well be due to the defect in palmitoylation rather than the modification with 15d-PGJ2. Though the authors allude to the possibility that intracellular redistribution of HRas by 15d-PGJ2 requires C181 palmitoylation, the direct influence of C184 modification on C181 palmitoylation is not shown. To have a meaningful conclusion, the authors need to compare the palmitoylation and modification with 15d-PGJ2.

To test if the inhibition of myoblast differentiation depends on HRas, they overexpressed the HRas and mutants in the C2C12 lines. However, this experiment does not take the endogenous HRAs into consideration, especially when interpreting the C184 mutant. An appropriate experiment to test this would be to knock down or knock out HRas (or make knock-in mutations of C184) and show that the effect of 15d-PGJ2 disappears. Moreover, in this specific experiment, it is difficult to interpret without a control with no HRas construct and another without the 15d-PGJ2 treatment.

Moreover, the overall study does not delineate the toxic effects of 15d-PGJ2 from its effect on the differentiation.

Reviewer #2 (Public Review):

Summary:

In this study, Swarang and colleagues identified the lipid metabolite 15d-PGJ2 as a potential component of senescent myoblasts. They proposed that 15d-PGJ2 inhibits myoblast proliferation and differentiation by binding and regulating HRas, suggesting its potential as a target for restoring muscle homeostasis post-chemotherapy.

Strengths:

The regulation of HRas by 15d-PGJ2 is well controlled.

Weaknesses:

(1) I still think the novelty is limited by previous published findings. The authors themselves noted that the accumulation of 15d-PGJ2 in senescent cells has been reported in various cell types, including human fibroblasts, HEPG2 hepatocellular carcinoma cells, and HUVEC endothelial cells (PMCID: PMC8501892). Although the current study observed similar activation of 15d-PGJ2 in myoblasts, it appears to be additive rather than fundamentally novel. The covalent adduct of 15d-PGJ2 with Cys-184 of H-Ras was reported over 20 years ago (PMID: 12684535), and the biochemical principles of this interaction are likely universal across different cell types. The regulation of myogenesis by both HRas and 15d-PGJ2 has also been previously extensively reported (PMID: 2654809, 1714463, 17412879, 20109525, 11477074). The main conceptual novelty may lie in the connection between these points in myoblasts. But as discussed in another comment, the use of C2C12 cells as a model for senescence study is questionable due to the lack of the key regulator p16. The findings in C2C12 cells may not accurately represent physiological-relevant myoblasts. It is recommended that these findings be validated in primary myoblasts to strengthen the study's conclusions.

(2) The C2C12 cell line is not an ideal model for senescence study.<br /> C2C12 cells are a well-established model for studying myogenesis. However, their suitability as a model for senescence studies is questionable. C2C12 cells are immortalized and do not undergo normal senescence like primary cells as C2C12 cells are known to have a deleted p16/p19 locus, a crucial regulator of senescence (PMID: 20682446). The use of C2C12 cells in published studies does not inherently validate them as a suitable senescence model. These studies may have limitations, and the appropriateness of the C2C12 model depends on the specific research goals.<br /> In the study by Moustogiannis et al. (PMID: 33918414), they claimed to have aged C2C12 cells through multiple population doublings. However, the SA-β-gal staining in their data, which is often used to confirm senescence, showed almost fully confluent "aged" C2C12 cells. This confluent state could artificially increase SA-β-gal positivity, suggesting that these cells may not truly represent senescence. Moreover, the "aged" C2C12 cells exhibited normal proliferation, which contradicts the definition of senescence. Similar findings were reported in another study of C2C12 cells subjected to 58 population doublings (PMID: 21826704), where even at this late stage, the cells were still dividing every 2 or 3 days, similar to younger cells at early passages. More importantly, I do know how the p16 was detected in that paper since the locus was already mutated. In terms of p21, there was no difference in the proliferative C2C12 cells at day 0.<br /> In the study by Moiseeva et al. in 2023 (PMID: 36544018), C2C12 cells were used for senescence modeling for siRNA transfection. However, the most significant findings were obtained using primary satellite cells or confirmed with complementary data.<br /> In conclusion, while molecular changes observed in studies using C2C12 cells may be valid, the use of primary myoblasts is highly recommended for senescence studies due to the limitations and questionable senescence characteristics of the C2C12 cell line.

(3) Regarding source of increased PGD in the conditioned medium, I want to emphasize that it's unclear whether the PGD or its metabolites increase in response to DNA damage or the senescence state. Thus, using a different senescent model to exclude the possibility of DNA damage-induced increase will be crucial.

(4) Similarly for the in vivo Doxorubicin (Doxo) injection, both reviewers have raised concerns about the potential side effects of Doxo, including inflammation, DNA damage, and ROS generation. These effects could potentially confound the results of the study. The physiological significance of this study will heavily rely on the in vivo data. However, the in vivo senescence component is confounded by the side effects of Doxo.

(5) Figure 2A lacks an important control from non-senescent cells during the measurement of C2C12 differentiation in the presence of conditioned medium. The author took it for granted that the conditioned medium from senescent cells would inhibit myogenesis, relying on previous publications (PMID: 37468473). However, that study was conducted in the context of myotonic dystrophy type 1. To support the inhibitory effect in the current experimental settings, direct evidence is required. It would be necessary to include another control with conditioned medium from normal, proliferative C2C12 cells.

(6) Statistical analyses problems.<br /> Only t-test was used throughout the study even when there are more than two groups. Please have a statistician to evaluate the replicates and statistical analyses used.<br /> For the 15d-PGJ2/cell concentration measurements in Figure 1F, there were only two replicates, which was provided in the supplementary table after required. Was that experiment repeated with more biological replicates?<br /> For figure 1C, Fig 1F, 1G, 1J, 2C, 2E, 3A, 3E, 3F, 4D, 4E, please include each data points in bar graphs as used in Fig 1D, or at least provide how many biological replicates were used for each experiment?<br /> There is no error bar in a lot of control groups (Fig 2C, 2E, 3EF, 4E, S4B).<br /> For qPCR data in Figure 1C, the author responded in that the data in was plotted using 2-ΔCT instead of 2-ΔΔCT to show the variability in the expression of mRNAs isolated from animals treated with Saline. This statement does not align with the method section. Please revise.

(7) For Figure 1, the title may not be appropriate as there is insufficient data to support the inhibition of myoblast differentiation.

Author response:

The following is the authors’ response to the original reviews.

Public Reviews:

Reviewer #1 (Public Review):

Summary:

In this manuscript, Eaton et al. examine the regulation of transcription directionality using a powerful genomic approach (more about the methodology below). Their data challenge the notion that the polyadenylation signal-reading Cleavage and Polyadenylation (CPA) complex is responsible for controlling promoter directionality by terminating antisense transcription. Namely, depletion of the required CPA factor RBBP6 has little effect on antisense transcription measured by POINT. They find instead that initiation is intrinsically preferential in the sense direction and additionally maintained by the activities of an alternative processing complex called Integrator, together with the kinase CDK9. In the presence of CDK9 activity, depletion of Integrator endoribonuclease INTS11 leads to globally increased transcription in the antisense direction, and minor effects in the sense direction. However, CDK9 inhibition reveals that sense transcription is also sensitive to INS11 depletion. The authors suggest that CDK9 activity is stronger in the sense direction, preventing INTS11-mediated premature termination of sense transcrpts.

Strengths:

The combination of acute depletion of the studied factors using degron approaches (important to limit possible secondary effects), together with novel and very sensitive nascent transcriptomics methods POINT and sPOINT is very powerful. The applied spike-in normalization means the analysis is more rigorous than most. Using this methodology allowed the authors to revisit the interesting question of how promoter/transcription directionality is determined.

The data quality appears very good and the fact that both global analysis as well as numerous gene-specific examples are shown makes it convincing.

The manuscript is well written and hence a pleasure to read.

We appreciate this positive assessment.

Weaknesses:

I am slightly worried about the reproducibility of the data - it is unclear to me from the manuscript if and which experiments were performed in replicate (lack of table with genomic experiments and GEO access, mentioned in more detail in below recommendations to authors), and the methods could be more detailed.

All sequencing data was deposited with GEO. Multiple biological replicates were performed for each sequencing experiment.  Bigwig files are presented as a table in the GEO submissions. This data has now been made public.

A separate discussion section would be useful, particularly since the data provided challenge some concepts in the field. How do the authors interpret U1 data from the Dreyfuss lab in light of their results? How about the known PAS-density directionality bias (more PAS present in antisense direction than in sense) - could the differential PAS density be still relevant to transcription directionality?

As suggested, we have expanded our discussion to relate our findings to existing data. We think the results from the Dreyfuss lab are very important and highlight the role of U1 snRNA in enforcing transcriptional elongation.  It does this in part by shielding PAS sequences.  Recent work from our lab also shows that U1 snRNA opposes the Restrictor complex and PNUTS, which otherwise suppress transcription (Estell et al., Mol Cell 2023).  Most recently, the Adelman lab has demonstrated that U1 snRNA generally enhances transcription elongation (Mimoso and Adelman., Mol Cell 2023).  Our work does not challenge and is not inconsistent with these studies.

The role of U1 in opposing PAS-dependent termination inspired the idea that antisense transcriptional termination may utilise PASs.  This was because such regions are rich in AAUAAA and comparatively poor in U1 binding sites. However, our RBBP6 depletion and POINT-seq data suggest that PAS-dependent termination is uncommon in the antisense direction. As such, other mechanisms suppress antisense transcription and influence promoter directionality. In our paper, we propose a major role for the Integrator complex.

We do not completely rule out antisense PAS activity and discuss the prior work that identified polyadenylated antisense transcripts. Nevertheless, this was detected by oligo-dT primed RT-PCR/Northern blotting, which cannot determine the fraction of non-polyadenylated RNA that could result from PAS-independent termination (e.g. by Integrator).  To do that requires an analysis of total nascent transcription as achieved by our POINT-seq.  Based on these experiments, Integrator depletion has a greater impact on antisense transcription than RBBP6 depletion. 

I find that the provided evidence for promoter directionality to be for the most part due to preferential initiation in the sense direction should be stressed more. This is in my eyes the strongest effect and is somehow brushed under the rug.

We agree that this is an important finding and incorporated it into the title and abstract.  As the reviewer recommends, we now highlight it further in the new discussion.

References 12-17 report an effect of Integrator on 5' of protein-coding genes, while data in Figure 2 appears contradictory. Then, experiments in Figure 4 show a global effect of INST11 depletion on promoter-proximal sense transcription. In my opinion, data from the 2.5h time-point of depletion should be shown alongside 1.5h in Figure 2 so that it is clear that the authors found an effect similar to the above references. I find the current presentation somehow misleading.

We are grateful for this suggestion and present new analyses demonstrating that our experiment in Figure 2 concurs with previous findings (Supplemental Figures 2A and B). Our original heatmap (Figure 2E) shows a very strong and general antisense effect of INTS11 loss. On the same scale, the effects in the sense direction are not as apparent, which is also the case using metaplots.  New supplemental figure 2A now shows sense transcription from this experiment in isolation and on a lower scale, demonstrating that a subset of genes shows promoter-proximal increases in transcription following INTS11 depletion.  This is smaller and less general than the antisense effect but consistent with previous findings.  Indeed, our new analysis in supplemental figure 2B shows that affected protein-coding genes are lowly expressed, in line with Hu et al., Mol Cell 2023. This explains why a sense effect is not as apparent by metaplot, for which highly expressed genes contribute the most signal.

As a result of our analyses, we are confident that the apparently larger effect at the 2.5hr timepoint (Figure 4) that we initially reported is due to experimental variability and not greater effects of extended INTS11 depletion. Overlaying the 1.5h and 2.5h datasets (Supplemental Figure 4B) revealed a similar number of affected protein-coding genes with a strong (83%) overlap between the affected genes.  To support this, we performed qPCR on four affected protein-coding transcripts which revealed no significant difference in the level of INTS11 effect after 2.5h vs 1.5h (Supplemental Figure 4C).

We now present data for merged replicates in Figures 2 and 4 which reveal very similar average profiles for -INTS11 vs +INTS11 at both timepoints. Overall, we believe that we have resolved this discrepancy by showing that it amounts to experimental variability and because the most acutely affected protein-coding genes are lowly expressed. As detailed above, we show this in multiple ways (and validate by qPCR) We have revised the text accordingly and removed our original speculation that differences reflected the timeframe of INTS11 loss.

Conclusion/assessment:

This important work substantially advances our understanding of the mechanisms governing the directionality of human promoters. The evidence supporting the claims of the authors is compelling, with among others the use of advanced nascent transcriptomics including spike-in normalization controls and acute protein depletion using degron approaches.

In my opinion, the authors' conclusions are in general well supported.

Not only the manuscript but also the data generated will be useful to the wide community of researchers studying transcriptional regulation. Also, the POINT-derived novel sPOINT method described here is very valuable and can positively impact work in the field.

We are grateful for the reviewers' positive assessment of our study.

Reviewer #2 (Public Review):

Summary:

Eaton and colleagues use targeted protein degradation coupled with nascent transcription mapping to highlight a role for the integrator component INST11 in terminating antisense transcription. They find that upon inhibition of CDK9, INST11 can terminate both antisense and sense transcription - leading to a model whereby INST11 can terminate antisense transcription and the activity of CDK9 protects sense transcription from INST11-mediated termination. They further develop a new method called sPOINT which selectively amplifies nascent 5' capped RNAs and find that transcription initiation is more efficient in the sense direction than in the antisense direction. This is an excellent paper that uses elegant experimental design and innovative technologies to uncover a novel regulatory step in the control of transcriptional directionality.

Strengths:

One of the major strengths of this work is that the authors endogenously tag two of their proteins of interest - RBBP6 and INST11. This tag allows them to rapidly degrade these proteins - increasing the likelihood that any effects they see are primary effects of protein depletion rather than secondary effects. Another strength of this work is that the authors immunoprecipitate RNAPII and sequence extracted full-length RNA (POINT-seq) allowing them to map nascent transcription. A technical advance from this work is the development of sPOINT which allows the selective amplification of 5' capped RNAs < 150 nucleotides, allowing the direction of transcription initiation to be resolved.

We appreciate this positive assessment.

Weaknesses:

While the authors provide strong evidence that INST11 and CDK9 play important roles in determining promoter directionality, their data suggests that when INST11 is degraded and CDK9 is inhibited there remains a bias in favour of sense transcription (Figures 4B and C). This suggests that there are other unknown factors that promote sense transcription over antisense transcription and future work could look to identify these.

We agree that other (so far, unknown) factors promote sense transcription over antisense, which was demonstrated by our short POINT.  We have provided an expanded discussion on this in the revision. In our opinion, demonstrating that sense transcription is driven by preferential initiation in that direction is a key finding and we agree that the identification of the underlying mechanism constitutes an interesting avenue for future study.

Reviewer #3 (Public Review):

Summary:

Using a protein degradation approach, Eaton et al show that INST11 can terminate the sense and anti-sense transcription but higher activity of CDK9 in the sense direction protects it from INS11-dependent termination. They developed sPOINT-seq that detects nascent 5'-capped RNA. The technique allowed them to reveal robust transcription initiation of sense-RNA as compared to anti-sense.

Strengths:

The strength of the paper is the acute degradation of proteins, eliminating the off-target effects. Further, the paper uses elegant approaches such as POINT and sPOINT-seq to measure nascent RNA and 5'-capped short RNA. Together, the combination of these three allowed the authors to make clean interpretations of data.

We appreciate this positive assessment.

Weaknesses:

While the manuscript is well written, the details on the panel are not sufficient. The methods could be elaborated to aid understanding. Additional discussion on how the authors' findings contradict the existing model of anti-sense transcription termination should be added.

We have added more detail to the figure panels, which we hope will help readers to navigate the paper more easily. Specifically, the assay employed for each experiment is indicated in each figure panel. As requested, we provide a new and separate discussion section in the revision.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

Congratulations on this important piece of work!

Some specific suggestions.

MAJOR

-The data are not available (Accession "GSE243266" is currently private and is scheduled to be released on Sep 01, 2026.) This should be corrected and as a minimum, the raw sequencing files as well as the spike-in scaled bigwig files should be provided in GEO.

We have made the data public. Raw and bigwig files are provided as part of the GEO upload.

MINOR

- It would be useful for readers if you could include catalog numbers of the reagents used in the study.

We have included this information in our revision.

- A table in experimental procedures summarizing the genomic experiments performed in this study as well as published ones reanalyzed here would be helpful.

This is now provided as part of the resources table.

- It would be easier for reviewers to evaluate the manuscript if the figure legends were included together with the figures on one page. This is now allowed by most journals.

We have used this formatting in the revision.

- Providing some captions for the results sections would be helpful.

We have included subheadings as suggested.

Reviewer #2 (Recommendations For The Authors):

Generally, I would suggest writing the experiment-type above panels where it is not immediately obvious what they are so a reader can appreciate the figures without referencing the legend. E.g. write POINT-seq on Figure 1B just to make it obvious to someone looking at the figures what methodology they are looking at. Likewise, you could write RNAPII ChIP-seq for Supplementary Figures 3D and 3E.

We have carried out this recommendation.

Can a y-axis be indicated on POINT-seq genome browser tracks? This could make them easier to interpret.

Y-axis scales are provided as RPKM as stated in the figure legends.

The authors could address/speculate in the text why there is less POINT-seq signal for the antisense transcript in the treatment condition in Figure 1B? Or could consider including a different example locus where this is not the case for clarity.

Acute depletion of poly(A) factors (like RBBP6) results in a strong read-through beyond the poly(A) signal of protein-coding genes as Figure 1 shows.  However, it also causes a reduction in transcription levels, which can be seen in the figure and is correctly noted by the reviewer in this comment.  We see this with other poly(A) factor depletions (e.g. CPSF73 and CPSF30 – Eaton et al., 2020 and Estell et al., 2021) and other labs have observed this too (e.g for CPSF73-dTAG depletion (Cugusi et al., Mol Cell 2022)).  Plausible reasons include a limited pool of free RNAPII due to impaired transcriptional termination or limited nucleotide availability due to their incorporation within long read-through transcripts. For these reasons, we have retained the example in Figure 1B as a typical representation of the effect. Moreover, the heatmap in Figure 1D fairly represents the spectrum of effects following RBBP6 loss – highlighting the strong read-through beyond poly(A) signals and the marginal antisense effects.

"The established effect of INTS11 at snRNAs was detected in our POINT-seq data and demonstrates the efficacy of this approach (Figure 2B)." The authors could explain this point more clearly in the text and describe the data - e.g. As expected, depletion of INTS11 leads to increased POINT-seq signal at the 3' end of snRNAs, consistent with defects in transcriptional termination. This is highlighted by the RNU5A-1 and RNU5B-1 loci (Figure 2B).

We agree and have added more context to clarify this.

I would suggest adjusting the scale of the heatmap in Figure 2E - I think it would be easier to interpret if the value of 0 was white - with >0 a gradient of orange and <0 a gradient of blue (as is done in Figure 1C). I think making this change would make the point as written in the text clearer i.e. "heatmap analysis demonstrates the dominant impact of INTS11 on antisense versus sense transcription at most promoters (Figure 2E)." I'm assuming most of the sense transcription would be white (more clearly unchanging) when the scale is adjusted.

We agree and have done this. The reviewer is correct that most sense transcription is unchanged by INTS11 loss.  However, as we alluded to in the original submission, a subset of transcripts shows a promoter-proximal increase after INTS11 depletion. We have expanded the analyses of this effect (see responses to other comments) but stress that it is neither as general nor as large as the antisense effect.

The authors make the point that there is mildly increased transcription over the 5' end of some genes upon INST11 depletion and show a track (Supplementary Fig 2A). It is not immediately obvious from the presentation of the meta-analysis in Figure 2D how generalisable this statement is. Perhaps the size of the panel or thickness of the lines in Figure 2D could be adjusted so that the peak of the control (in blue) could be seen. Perhaps an arrow indicating the peak could be added? I'm assuming the peak at the TSS is slightly lower in the control compared to INST11 depletion based on the authors' statement.

We have provided multiple new analyses of this data to highlight where there are promoter-proximal effects of INTS11 loss in the sense direction.  Please see our response to the public review of reviewer 1 and new supplemental figures 2A, 2B, 4A and 4B which highlight the sense transcription increased in the absence of INTS11.

The authors label Figure 4 "Promoters lose their directionality when CDK9 is inhibited" - but in INST11 depleted cells treated with CDK9i they find that there still is a bias towards sense transcription. Suggested edit "Some promoter directionality is lost when CDK9 is inhibited" or similar.

We agree and have made this change.

The authors conclude that INTS11-mediated effects are the result of perturbation of the catalytic activities of Integrator, the authors should perform rescue experiments with the catalytically dead E203Q-INTS11 mutant.

This is a very good suggestion and something we had intended to pursue.  However, as we will describe below (and shown in Supplemental Figure 4G), there were confounding issues with this experiment.

The E203Q mutant of INTS11 is widely used in the literature to test for catalytic functions of INTS11.  However, we have found that this mutation impairs the ability of INTS11 to bind other Integrator modules in cells. Based on co-immunoprecipitation of flag-tagged WT and E203Q derivatives, INTS1 (backbone module), 10 (tail module), and 8 (phosphatase module) all show reduced binding to E203Q vs. WT. Because E203Q INTS11 is defective in forming Integrator complexes, rescue experiments might not fully distinguish the effects of INTS11 activity from those caused by defects in complex assembly. While this may at first seem unexpected, in the analogous 3’ end processing complex, catalytic mutants of CPSF73 (which is highly related to INTS11) negatively affect its interaction with other complex members (Kolev and Steitz, EMBO Reports 2005).

We hypothesise that INTS11 activity is most likely involved in attenuating promoter-proximal transcription, but we cannot formally rule out other explanations and discuss this in our revision. Regardless of how INTS11 attenuates transcription, our main conclusion is on its requirement to terminate antisense transcription whether this involves its cleavage activity or not.

The authors suggest that CDK9 modulates INTS11 activity/assembly and suggest this may be related to SPT5. Is there an effect of CDK9 inhibition on the snRNA's highlighted in Figure 2B?

We believe that snRNAs are different from protein-coding genes concerning CDK9 function. Shona Murphy’s lab previously showed that, unlike protein-coding genes, snRNA transcription is insensitive to CDK9 inhibition, and that snRNA processing is impaired by CDK9 inhibition (Medlin et al., EMBO 2003 and EMBO 2005).  We reproduce these findings by metaanalysis of 15 highly expressed and well-separated snRNAs and by qRT-PCR of unprocessed RNU1-1, RNU5A-1 and RNU7-1 snRNA following CDK9 inhibition. We observe snRNA read-through by POINT-seq following INTS11 loss whether CDK9 is inhibited or not (left panel, below). Note the higher TES proximal signal in CDK9i conditions, which likely reflects the accumulation of unprocessed snRNA as validated by qPCR for three example snRNAs (right panel, below).

Author response image 1.

For Figure 4, would similar results be observed using inhibitors targeting other transcriptional CDKs such as CDK7,12/13?

In response to this suggestion, we analysed four selected protein-coding transcripts (the same 4 that we used to validate the CDK9i results) by qRT-PCR in a background of CDK7 inhibition using the THZ2 compound (new Supplemental Figure 4E).  THZ2 suppresses transcription from these genes as expected.  Interestingly, expression is restored by co-depleting Integrator, recapitulating our findings with CDK9 inhibition.  As CDK7 is the CDK-activating kinase for CDK9, its inhibition will also inhibit CDK9 so THZ2 may simply hit this pathway upstream of where CDK9 inhibitors.  Second, CDK7 may independently shield transcription from INTS11.  We allude to both interesting possibilities.

What happens to the phosphorylation state of anti-sense engaged RNAPII when INTS11 is acutely depleted and/or CDK9 is inhibited? This could be measured by including Ser5 and Ser2 antibodies in the sPOINT-seq assay and complemented with Western Blot analysis.

We have performed the western blot for Ser5 and Ser2 phosphorylation as suggested.  Both signals are mildly enhanced by INTS11 loss, which is consistent with generally increased transcription.  Ser2p is strongly reduced by CDK9 inhibition, which is consistent with the loss of nascent transcription in this condition.  Interestingly, both modifications are partly recovered when INTS11 is depleted in conjunction with CDK9 inhibition. This is consistent with the effects that we see on POINT-seq and shows that the recovered transcription is associated with some phosphorylation of RNAPII CTD.  This presumably reflects the action(s) of kinases that can act redundantly with CDK9.

We have not performed POINT-seq with Ser5p and Ser2p antibodies under these various conditions.  Our rationale is that our existing data uses an antibody that captures all RNAPII (regardless of its phosphorylation status), which we feel most comprehensively assays transcription in either direction. Moreover, the lab of Fei Chen (Hu et al., Mol Cell 2023) recently published Ser5p and Ser2p ChIP-seq following INTS11 loss. By ChIP-seq, they observe a bigger increase in antisense RNAPII occupancy vs. sense providing independent and orthogonal support for our POINT-seq data.  Interestingly, this antisense increase is not paralleled by proportional increases in Ser5p or Ser2p signals.  This suggests that the unattenuated antisense transcription resulting from INTS11 loss does not have high Ser5p or Ser2p.  Since CDK7 and 9 are major Ser5 and 2 kinases, this supports our model that their activity is less prevalent for antisense transcription.  We now discuss these data in our revision.   

The HIV reporter RNA experiments should be performed with the CDK9 inhibitor added to the experimental conditions. Presumably CDK9 inhibition would result in no upregulation of the reporter upon addition of TAT and/or dTAG. Perhaps the amount of TAT should be reduced to still have a dynamic window in which changes can be detected. It is possible that reporter activation is simply at a maximum. Can anti-sense transcription be measured from the reporter?

We have performed the requested CDK9 inhibitor experiment to confirm that TAT-activated transcription from the HIV promoter is CDK9-dependent (new supplemental figure 4F).  Consistent with previous literature on HIV transcription, CDK9 inhibition attenuates TAT-activated transcription.  Importantly, and in line with our other experiments, depletion of INTS11 results in significant restoration of transcription from the HIV promoter when CDK9 is inhibited. Thus, TAT-activated transcription is CDK9-dependent and, as for endogenous genes, CDK9 prevents attenuation by INTS11.

While TAT-activated transcription is high, we do not think that the plasmid is saturated. When considering this question, we revisited previous experiments using this system to study RNA processing (Dye et al., Mol Cell 1999, Cell 2001, Mol Cell 2006). In these cases, mutations in splice sites or polyadenylation sites have a strong effect on RNA processing and transcription around HIV reporter plasmids. Effects on transcription and RNA processing are; therefore, apparent in the appropriate context. In contrast, we find that the complete elimination of INTS11 has no impact on RNA output from the HIV reporter. Our original experiment assessing the impact of INTS11 loss in +TAT conditions used total RNA.  One possibility is that this allows non-nascent RNA to accumulate which might confound our interpretation of INTS11 effects on ongoing transcription.  However, the new experiment described in the paragraph above was performed on chromatin-associated (nascent) RNA to rule this out.  This again shows no impact of INTS11 loss on HIV promoter-derived transcription in the presence of TAT.

To our knowledge, antisense transcription is not routinely assayed from plasmids. They generally employ very strong promoters (e.g. CMV, HIV) to drive sense transcription.  Crucially, their circular nature means that RNAPII going around the plasmid could interfere with antisense transcription coming the other way which does not happen in a linear genomic context. This is why we restricted our use of plasmids to looking at the effects of stimulated CDK9 recruitment (via TAT) on transcription rather than promoter directionality.   

The authors should clearly state how many replicates were performed for the genomics experiments. Ideally, a signal should be quantified and compared statistically rather than relying on average profiles only.

We have stated the replicate numbers for sequencing experiments in the relevant figure legends. All sequencing experiments were performed in at least two biological replicates, but often three. In addition, we validated their key conclusions by qPCR or with orthogonal sequencing approaches.

Reviewer #3 (Recommendations For The Authors):

The authors provide strong evidence in support of their claims.

ChIP-seq of pol2S5 and S2 upon INST11 and CDK9 inhibition will strengthen the observation that transcription in the sense direction is more efficient.

We view the analysis of total RNAPII as the most unbiased way of establishing how much RNAPII is going one way or the other. Importantly, ChIP-seq was very recently performed for Ser2p and Ser5p RNAPII derivatives in the lab of Fei Chen (Hu et al., Mol Cell 2023). Their data shows that loss of INTS11 increases the occupancy of total RNAPII in the antisense direction more than in the sense direction, which is consistent with our finding. Interestingly, the increased antisense RNAPII was not paralleled with an increase in Ser2p or Ser5p. This suggests that, following INTS11 loss, the unattenuated antisense transcription is not associated with full/normal Ser2p or Ser5p. These modifications are normally established by CDK7 and 9; therefore, this published ChIP-seq suggests that they are not fully active on antisense transcription when INTS11 is lost. This supports our overall model that CDK9 (and potentially CDK7 as suggested for a small number of genes in new Supplemental Figure 4E) is more active in the sense direction to prevent INTS11-dependent attenuation. We now discuss these data in our revision.

In Supplementary Figure 2, the eRNA expression increases upon INST11 degradation, I wonder if the effects of this will be appreciated on cognate promoters? Can the authors test some enhancer:promoter pairs?

We noticed that some genes (e.g. MYC) that are regulated by enhancers show reduced transcription in the absence of INTS11. Whilst this could suggest a correlation, the transcription of other genes (e.g. ACTB and GAPDH) is also reduced by INTS11 loss although they are not regulated by enhancers.  A detailed and extensive analysis would be required to establish any link between INTS11-regulated enhancer transcription and the transcription of genes from their cognate promoters.  We agree that this would be interesting, but it seems beyond the scope of our short report on promoter directionality.

Line 111, meta plot was done of 1316 genes. Details on this number should be provided. Overall, the details of methods and analysis need improvement. The layout of panels and labelling on graphs can be improved.

We have now explained the 1316 gene set.  In essence, these are the genes separated from an expressed neighbour by at least 10kb.  This distance was selected because depletion of RBBP6 induces extensive read-through transcription beyond the polyadenylation site of protein-coding genes.  To avoid including genes affected by transcriptional read-through from nearby transcription units we selected those with a 10kb gap between them. This was the only selection criteria so is unlikely to induce any unintended biases. Finally, we have added more information to the figure panels and their legends, which we hope will make our manuscript more accessible.

eLife assessment

The important study uses a new experimental method to provide compelling evidence on how sense- and anti-sense transcription is differentially regulated. The method described here can generally be used to study the alterations in transcription. This paper will be of interest to scientists working in the gene regulation community.

Reviewer #1 (Public Review):

Summary:

In this manuscript, Eaton et al. examine the regulation of transcription directionality using a powerful genomic approach (more about the methodology below).<br /> Their data challenge the notion that the polyadenylation signal-reading Cleavage and Polyadenylation (CPA) complex is responsible for controlling promoter directionality by terminating antisense transcription. Namely, depletion of the required CPA factor RBBP6 has little effect on antisense transcription measured by POINT. They find instead that initiation is intrinsically preferential in the sense direction and additionally maintained by the activities of an alternative processing complex called Integrator, together with the kinase CDK9. In the presence of CDK9 activity, depletion of Integrator endoribonuclease INTS11 leads to globally increased transcription in the antisense direction, and minor effects in the sense direction. However, CDK9 inhibition reveals that sense transcription is also sensitive to INS11 depletion. The authors suggest that CDK9 activity is stronger in the sense direction, preventing INTS11-mediated premature termination of sense transcripts.

Strengths:

The combination of acute depletion of the studied factors using degron approaches (important to limit possible secondary effects), together with novel and very sensitive nascent transcriptomics methods POINT and sPOINT is very powerful. The applied spike-in normalization means the analysis is more rigorous than most. Using this methodology allowed the authors to revisit the interesting question of how promoter/transcription directionality is determined.

The data quality appears very good and the fact that both global analysis as well as numerous gene-specific examples are shown makes it convincing.

The manuscript is well written and hence a pleasure to read.

Weaknesses:

The bias in transcriptional initiation directionality remains to be elucidated.

Conclusion/assessment:

This important work substantially advances our understanding of the mechanisms governing the directionality of human promoters. The evidence supporting the claims of the authors is compelling, with a.o. the use of advanced nascent transcriptomics including spike-in normalization controls and acute protein depletion using degron approaches.

In my opinion the authors' conclusions are well supported.

Not only the manuscript but also the data generated will be useful to the wide community of researchers studying transcriptional regulation. Also, the POINT-derived novel sPOINT method described here is very valuable and can positively impact work in the field.

Reviewer #2 (Public Review):

Summary:

Eaton and colleagues use targeted protein degradation coupled with nascent transcription mapping to highlight a role for the integrator component INST11 in terminating antisense transcription. They find that upon inhibition of CDK9, INST11 can terminate both antisense and sense transcription - leading to a model whereby INST11 can terminate antisense transcription and the activity of CDK9 protects sense transcription from INST11-mediated termination. They further develop a new method called sPOINT which selectively amplifies nascent 5' capped RNAs and find that transcription initiation is more efficient in the sense direction than in the antisense direction. This is an excellent paper which uses elegant experimental design and innovative technologies to uncover a novel regulatory step in the control of transcriptional directionality.

Strengths:

One of the major strengths of this work is that the authors endogenously tag two of their proteins of interest - RBBP6 and INST11. This tag allows them to rapidly degrade these proteins - increasing the likelihood that any effects they see are primary effects of protein depletion rather than secondary effects. Another strength of this work is that the authors immunoprecipitate RNAPII and sequence extracted full length RNA (POINT-seq) allowing them to map nascent transcription. A technical advance from this work is the development of sPOINT which allows the selective amplification of 5' capped RNAs < 150 nucleotides, allowing the direction of transcription initiation to be resolved.

Weaknesses:

While the authors provide strong evidence that INST11 and CDK9 play important roles in determining promoter directionality, their data suggests that when INST11 is degraded and CDK9 is inhibited there remains a bias in favour of sense transcription (Figure 4B and C). This suggests that there are other unknown factors that promote sense transcription over antisense transcription and future work could look to identify these.

Reviewer #3 (Public Review):

Summary:

Using protein degradation approach, Eaton et al show that INST11 can terminate the sense and anti-sense transcription but higher activity of CDK9 in sense direction protects it from INS11-dependent termination. They developed sPOINT-seq that detects nascent 5'-capped RNA. The technique allowed them to reveal robust transcription initiation of sense-RNA as compared to anti-sense.

Strengths:

The strength of paper is acute degradation of proteins, eliminating the off-target effects. Further, the paper uses elegant approaches such as POINT and sPOINT-seq to measure nascent RNA and 5'-capped short RNA. Together, the combination of these three allowed the authors to make clean interpretations of data.

Weaknesses:

While manuscript is well written, the details on panel is not sufficient. The methods can be more elaborate for better understanding. Additional discussion on how authors findings contradict the existing model of anti-sense transcription termination should be added.

in the revised manuscript, authors have added details on panels and elaborated method and other sections for better understanding.

eLife assessment

The study presents valuable findings on the molecular mechanisms of glucose-stimulated insulin secretion from pancreatic islets, focusing on the main regulatory elements of the signaling pathway in physiological conditions. While the evidence supporting the conclusions is solid, the study can be strengthened by the use of a beta cell line or knockout mice. The work will be of interest to cell biologists and biochemists working on diabetes.

Reviewer #1 (Public Review):

Summary:

This study investigated the mechanism by which PGE2 inhibits the release of insulin from pancreatic beta cells in response to glucose. The researchers used a combination of cell line experiments and studies in mice with genetic ablation of the Kv2.2 channel. Their findings suggest a novel pathway where PGE2 acts through EP2/EP4 receptors to activate PKA, which directly phosphorylates a specific site (S448) on the Kv2.2 channel, inhibiting its activity and reducing GSIS.

Strengths:

- The study elegantly demonstrates a potential pathway connecting PGE2, EP2/EP4 receptors, PKA, and Kv2.2 channel activity, using embryonic cell line.<br /> - Additional experiments in INS1 and primary mouse beta cells with altered Kv2.2 function partially support the inhibitory role of PGE2 on GSIS through Kv2.2 inhibition.

Weaknesses:

- A critical limitation is the use of HEK293T cells, which are not pancreatic beta cells. Functional aspects can differ significantly between these cell types.<br /> - The study needs to address the apparent contradiction of PKA activating insulin secretion in beta cells, while also inhibiting GSIS through the proposed mechanism.<br /> - A more thorough explanation is needed for the discrepancies observed between the effects of PGE2 versus Kv2.2 knockdown/mutation on the electrical activity of beta cells and GSIS.

Reviewer #2 (Public Review):

The authors identified new target elements for prostaglandin E2 (PGE2) through which insulin release can be regulated in pancreatic beta cells under physiological conditions. In vitro extracellular exposure to PGE2 could directly and dose-dependently inhibit the potassium channel Kv2.2. In vitro pharmacology revealed that this inhibition occurs through the EP2/4 receptors, which activate protein kinase A (PKA). By screening specific sites of the Kv2.2 channel, the target phosphorylation site (S448) for PKA regulation was found. The physiological relevance of the described signaling cascade was investigated and confirmed in vivo, using a Kv2.2 knockdown mouse model.

The strength of this manuscript is the novelty of the (EP2/4-PKA-Kv2.2 channel) molecular pathway described and the comprehensive methodological toolkit the authors have relied upon.

The introduction is detailed and contains all the information necessary to place the claims in context. Although the dataset is comprehensive and a logical lead is consistently built, there is one important point to consider: to clarify that the described signaling pathway is characteristic of normal physiological conditions and thus differs from pathological changes. It would be useful to carry out basic experiments in a diabetes model (regardless of whether this is in mice or rats).

Author response:

We thank the reviewers for their positive evaluation and constructive feedback on our study.

We acknowledge the concern regarding the use of HEK293T cells. In the revised manuscript, we will provide a more detailed explanation of the role of the PKA pathway in the regulation of GSIS by PGE2. To validate this regulation through Kv2.2, we will overexpress the Kv2.2 mutant channel in beta cells and assess its impact. Additionally, we will verify the specificity of the antibodies for EP1-EP4 receptors by knockdown. To confirm the receptors involved in PGE2 function, we will use additional EP receptor blockers or perform receptor knockdown experiments.

We will clarify that the described signaling pathway operates under normal physiological conditions and differs from pathological changes.

We once again thank the reviewers for their positive evaluation and constructive suggestions.

eLife assessment

This work describes a novel affinity interactomics approach that allows investigators to identify networks of protein-protein interactions in cells. The important findings presented here describe the application of this technique to the SH3 domain of the membrane remodeling Bridging Integrator 1 (BIN1), the truncation of which leads to centronuclear myopathy. The authors present solid evidence that BIN1 SH3 engages with an unexpectedly high number of cellular proteins, many of which are linked to skeletal muscle disease, and evidence is presented to suggest that BIN1 may play a role in mitosis creating the potential for new avenues in drug development efforts. Some of the findings, however, remain rather preliminary, lack sufficient replicates and may require additional experiments to definitively support the conclusions.

Reviewer #1 (Public Review):

Summary:

In this paper, Zambo and coworkers use a powerful technique, called native holdup, to measure the affinity of the SH3 domain of BIN1 for cellular partners. Using this assay, they combine data using cellular proteins and proline-containing fragments in these proteins to identify 97 distinct direct binding partners of BIN1. They also compare the binding interactome of the BIN1 SH3 domain to the interactome of several other SH3 domains, showing varying levels of promiscuity among SH3 domains. The authors then use pathway analysis of BIN1 binding partners to show that BIN1 may be involved in mitosis. Finally, the authors examine the impact of clinically relevant mutations of the BIN1 SH3 domain on the cellular interactome. The authors were able to compare the interactome of several different SH3 domains and provide novel insight into the cellular function of BIN1. Generally, the data supports the conclusions, although the reliance on one technique and the low number of replicates in each experiment is a weakness of the study.

Strengths:

The major strength of this paper is the use of holdup and native holdup assays to measure the affinity of SH3 domains to cellular partners. The use of both assays using cell-derived proteins and peptides derived from identified binding partners allows the authors to better identify direct binding partners. This assay has some complexity but does hold the possibility of being used to measure the affinity of the cellular interactome of other proteins and protein domains. Beyond the utility of the technique, this study also provides significant insight into the cellular function of BIN1. The authors have strong evidence that BIN1 might have an undiscovered function in cellular mitosis, which potentially highlights BIN1 as a drug target. Finally, the study provides outstanding data on the cellular binding properties and partners of seven distinct SH3 domains, showing surprising differences in the promiscuity of these proteins.

Weaknesses:

There are several weaknesses of the study. First, the authors rely completely on a single technique to measure the affinity of the cellular interactome. The native holdup is a relatively new technique that is powerful yet relatively unproven. However, it appears to have the capacity to measure the relative affinity of proteins and the authors describe the usefulness of the technique. Second, and most important, the authors use a relatively small number of replicates for the holdup assays. The holdup technique will have biological variation in the cellular lysate or purified protein that could impact the results, so more replicates would enhance the reliability of the results.

Reviewer #2 (Public Review):

Summary:

The authors report here interesting data on the interactions mediated by the SH3 domain of BIN1 that expand our knowledge on the role of the SH3 domain of BIN1 in terms of mediating specific interactions with a potentially high number of proteins and how variants in this region alter or prevent these protein-protein interactions. These data provide useful information that will certainly help to further dissect the networks of proteins that are altered in some human myopathies as well as the mechanisms that govern the correct physiological activity of muscle cells.

Strengths:

The work is mostly based on improved biochemical techniques to measure protein-protein interaction and provide solid evidence that the SH3 domain of BIN1 can establish an unexpectedly high number of interactions with at least a hundred cellular proteins, among which the authors underline the presence of other proteins known to be causative of skeletal muscle diseases and not known to interact with BIN1. This represents an unexpected and interesting finding relevant to better define the network of interactions established among different proteins that, if altered, can lead to muscle disease. An interesting contribution is also the detailed identification of the specific sites, namely the Proline-Rich Motifs (PRMs) that in the interacting proteins mediate binding to the BIN1 SH3 domain.

Weaknesses:

Less convincing, or too preliminary in my opinion, are the data supporting BIN1 co-localization with PRC1. Indeed, the affinity of PRC1 is significantly lower than that of DNM2, an established BIN1 interacting protein. Thus, this does not provide compelling evidence to support PRC1 as a significant interactor of BIN1. Similarly, the localization data appears somewhat preliminary to substantiate a role of BIN1 in mitotic processes. These findings may necessitate additional experimental work to be more convincing.

Author response:

The following is the authors’ response to the previous reviews.

Reviewer #1

We modified the text regarding PRC1 according to the reviewer’s recommendation.

Reviewer #2

Following the reveiwer’s advise, we introduced the holdup assay, as well as the native holdup assay in more details.

This new part now also discusses the question of replicates in more details. We do not agree with the eLife assessment on this matter, but we think that this assessment was made because analyzing holdup data requires a different approach compared to more conventional interactomic approaches and these differences were not introduced in sufficient depth. We hope that the inclusion of more background reasoning, as well as by providing a more detailed comparison of the measured independent BIN1 interactomes, now included on Figure S4, will eliminate all confusion in the reader.

We thank the reviewer for guiding us to a previous work that was done on Grb2. Indeed, the finding of this earlier work aligns perfectly with our finding suggesting general similarities in SH3 domain mediated interactions.

eLife assessment

This useful manuscript extends prior work to identify OVO as a major transcriptional activator of the female germline gene expression program. Using a combination of solid genomic strategies, the authors demonstrate that OVO binds to the promoters of hundreds of genes in the female germline and promotes their expression.

Reviewer #1 (Public Review):

Summary:

In this manuscript, Benner et al. identify OVO as a transcriptional factor instrumental in promoting expression of hundreds of genes essential for female germline identity and early embryo development. Prior data had identified both ovo and otu as genes activated by OVO binding to the promoters. By combining ChIP-seq, RNA-seq and analysis of prior datasets, the authors extend these data to hundreds of genes and therefore propose that OVO is a master transcriptional regulator of oocyte development. They further speculate that OVO may function to promote chromatin accessibility to facilitate germline gene expression. Overall, the data compellingly demonstrate a much broader role for OVO in activation of genes in the female germline than previously recognized. By contrast, the relationship between OVO, chromatin accessibility and the timing of gene expression is only correlative, and more work will be needed to determine the mechanisms by which OVO promotes transcription.

Strengths

Here Benner at al. convincingly show that OVO is a transcriptional activator that promotes expression of hundreds of genes in the female germline. The ChIP-seq and RNA-seq data included in the manuscript are robust and the analysis is compelling.

Importantly, the set of genes identified are essential for maternal processes, including egg production and patterning of the early embryo. Together, these data identify OVO as a major transcriptional activator of the numerous genes expressed in the female germline, deposited into the oocyte and required for early gene expression. This is an important finding as this is an essential process for development and prior to this study the major drivers of this gene expression program were unknown.

Weaknesses

The novelty of the manuscript is somewhat limited as the authors show that, like two prior, well-studied OVO target genes, OVO binds to promoters of germline genes and activates transcription. The fact that OVO performs this function more broadly is not particularly surprising.

A major challenge to understanding the impact of this manuscript is the fact that the experimental system for the RNA-seq, the tagged constructs, and the expression analysis that provides the rationale for the proposed pioneering function of OVO are all included in a separate manuscript.

Author response:

The following is the authors’ response to the original reviews.

Public Reviews:

Reviewer #1 (Public Review):

Summary:

In this manuscript, Benner et al. identify OVO as a transcriptional factor instrumental in promoting the expression of hundreds of genes essential for female germline identity and early embryo development. Prior data had identified both ovo and otu as genes activated by OVO binding to the promoters. By combining ChIP-seq, RNA-seq, and analysis of prior datasets, the authors extend these data to hundreds of genes and therefore propose that OVO is a master transcriptional regulator of oocyte development. They further speculate that OVO may function to promote chromatin accessibility to facilitate germline gene expression. Overall, the data compellingly demonstrate a much broader role for OVO in the activation of genes in the female germline than previously recognized. By contrast, the relationship between OVO, chromatin accessibility, and the timing of gene expression is only correlative, and more work will be needed to determine the mechanisms by which OVO promotes transcription.

We fully agree with this summary.  

Strengths:

Here Benner et al. convincingly show that OVO is a transcriptional activator that promotes expression of hundreds of genes in the female germline. The ChIP-seq and RNA-seq data included in the manuscript are robust and the analysis is compelling.

Importantly, the set of genes identified is essential for maternal processes, including egg production and patterning of the early embryo. Together, these data identify OVO as a major transcriptional activator of the numerous genes expressed in the female germline, deposited into the oocyte and required for early gene expression. This is an important finding as this is an essential process for development and prior to this study, the major drivers of this gene expression program were unknown.

We are delighted that this aspect of the work came across clearly. Understanding the regulation of maternal effect genes has been something of a black-box, despite the importance of this class of genes in the history of developmental genetics. The repertoire of essential oogenesis/embryonic development genes that are bound by and respond to OVO are well characterized in the literature, but nothing is known about how they are transcriptionally regulated. We feel the manuscript will be of great interest to readers working on these genes.

Weaknesses:

The novelty of the manuscript is somewhat limited as the authors show that, like two prior, well-studied OVO target genes, OVO binds to promoters of germline genes and activates transcription. The fact that OVO performs this function more broadly is not particularly surprising.

Clearly, transcription factors regulate more than one or two genes. Never-the-less we were surprised at how many of the aspects of oogenesis per se and maternal effect genes were OVO targets. It was our hypothesis that OVO would have a transcriptional effect genome-wide, however, it was less clear whether OVO would always bind at the core promoter, as is with the case of ovo and otu. Our results strongly support the idea that core promoter proximal binding is essential for OVO function; a conclusion of work done decades ago, which has not been revisited using modern techniques. 

A major challenge to understanding the impact of this manuscript is the fact that the experimental system for the RNA-seq, the tagged constructs, and the expression analysis that provides the rationale for the proposed pioneering function of OVO are all included in a separate manuscript.

This is a case where we ended up with a very, very long manuscript which included a lot of revisiting of legacy data. It was a tough decision on how to break up all the work we had completed on ovo to date. In our opinion, it was too much to put everything into a single manuscript unless we wanted a manuscript length supplement (we were also worried that supplemental data is often overlooked and sometimes poorly reviewed). We therefore decided to split the work into a developmental localization/characterization paper and a functional genomics paper. As it stands both papers are long. Certainly, readers of this manuscript will benefit from reading our previous OVO paper, which we submitted before this one. The earlier manuscript is under revision at another journal and we hope that this improved manuscript will be published and accessible shortly.

Reviewer #2 (Public Review):

Summary:

In this manuscript, Benner et al. interrogate the transcriptional regulator OVO to identify its targets in the Drosophila germline. The authors perform ChIP-seq in the adult ovary and identify established as well as novel OVO binding motifs in potential transcriptional targets of OVO. Through additional bioinformatic analysis of existing ATAC-seq, CAGE-seq, and histone methylation data, the authors confirm previous reports that OVO is enriched at transcription start sites and suggest that OVO does not act as part of the core RNA polymerase complex. Benner et al. then perform bulk RNA-seq in OVO mutant and "wildtype" (GAL4 mediated expression of OVO under the control of the ovo promoter in OVO mutants) ovaries to identify genes that are differentially expressed in the presence of OVO. This analysis supports previous reports that OVO likely acts at transcription start sites as a transcriptional activator. While the authors propose that OVO activates the expression of genes that are important for egg integrity, maturation, and for embryonic development (nanos, gcl, pgc, bicoid), this hypothesis is based on correlation and is not supported by in vivo analysis of the respective OVO binding sites in some of the key genes. A temporal resolution for OVO's role during germline development and egg chamber maturation in the ovary is also missing. Together, this manuscript contains relevant ChIP-seq and RNA-seq datasets of OVO targets in the Drosophila ovary alongside thorough bioinformatic analysis but lacks important in vivo experimental evidence that would validate the high-quality datasets.

We thank reviewer 2 for the appreciation of the genomics data and analysis. Some of the suggested in vivo experiments are clear next steps, which are well underway. These are beyond the scope of the current manuscript. 

Temporal analysis of ovo function in egg chamber development is not easy, as only the weakest ovo alleles have any egg chambers to examine. However, we will also point out the long-known phenotypes of some of those weak alleles in the text (e.g. ventralized chambers in ovoD3/+). We will need better tools for precise rescue/degradation during egg chamber maturation.     

Strengths:

The manuscript contains relevant ChIP-seq and RNA-seq datasets of OVO targets in the Drosophila ovary alongside thorough bioinformatic analysis

Thank you. We went to great lengths to do our highly replicated experiments in multiple ways (e.g. independent pull-down tags) and spent considerable time coming up with an optimized and robust informatic analysis.

Weaknesses:

(1) The authors propose that OVO acts as a positive regulator of essential germline genes, such as those necessary for egg integrity/maturation and embryonic/germline development. Much of this hypothesis is based on GO term analysis (and supported by the authors' ChIP-seq data). However accurate interpretation of GO term enrichment is highly dependent on using the correct background gene set. What control gene set did the authors use to perform GO term analysis (the information was not in the materials and methods)? If a background gene set was not previously specified, it is essential to perform the analysis with the appropriate background gene set. For this analysis, the total set of genes that were identified in the authors' RNA-seq of OVO-positive ovaries would be an ideal control gene set for which to perform GO term analysis. Alternatively, the total set of genes identified in previous scRNA-seq analysis of ovaries (see Rust et al., 2020, Slaidina et al., 2021 among others) would also be an appropriate control gene set for which to perform GO term analysis. If indeed GO term analysis of the genes bound by OVO compared to all genes expressed in the ovary still produces an enrichment of genes essential for embryonic development and egg integrity, then this hypothesis can be considered.

We feel that this work on OVO as a positive regulator of genes like bcd, osk, nos, png, gnu, plu, etc., is closer to a demonstration than a proposition. These are textbook examples of genes required for egg and early embryonic development. Hopefully, this is not lost on the readers by an over-reliance on GO term analysis, which is required but not always useful in genome-wide studies. 

We used GO term enrichment analysis as a tool to help focus the story on some major pathways that OVO is regulating. To the specific criticism of the reference gene-set, GO term enrichment analysis in this work is robust to gene background set. We will update the GO term enrichment analysis text to indicate this fact and add a table using expressed genes in our RNA-seq dataset to the manuscript and clarify gene set robustness in greater detail in the methods of the revision. We will also try to focus the reader’s attention on the actual target genes rather than the GO terms in the revised text.

We have updated the GO term analysis by including all the expressed genes in our RNA-seq datasets as a background control. Figure 6 has been updated to include the significant GO terms. We have outlined changes in the methods section below.

Lines 794-801:

“Gene ontology enrichment analysis was completed with g:Profiler’s g:GOSt software (Raudvere et al. 2019) on the set of genes overlapping OVO ChIP peaks over the TSS and significantly upregulated in the presence of ectopic OVO (525 genes in total). All genes that were considered to be expressed in our RNA-seq datasets were used as a background control (10,801 genes in total). Default parameters were used for the enrichment analysis except for ‘statistical domain scope’ was set to ‘custom’ (our control background genes were uploaded here), ‘significance threshold’ was set to ‘Bonferroni correction’, and only GO biological process terms were searched for enrichment with the gene list. The GO terms listed in Figure 6 represent the 24 smallest GO term sizes according to Table S5.”

(2) The authors provide important bioinformatic analysis of new and existing datasets that suggest OVO binds to specific motifs in the promoter regions of certain germline genes. While the bioinformatic analysis of these data is thorough and appropriate, the authors do not perform any in vivo validation of these datasets to support their hypotheses. The authors should choose a few important potential OVO targets based on their analysis, such as gcl, nanos, or bicoid (as these genes have well-studied phenotypes in embryogenesis), and perform functional analysis of the OVO binding site in their promoter regions. This may include creating CRISPR lines that do not contain the OVO binding site in the target gene promoter, or reporter lines with and without the OVO binding site, to test if OVO binding is essential for the transcription/function of the candidate genes.

Exploring mechanism using in vivo phenotypic assays is awesome, so this is a very good suggestion. But, it is not essential for this work -- as has been pointed out in the reviews, in vivo validation of OVO binding sites has been comprehensively done for two target genes, ovo and otu. The “rules” appear similar for both genes. That said, we are already following up specific OVO target genes and the detailed mechanism of OVO function at the core promoter. We removed some of our preliminary in vivo figures from the already long current manuscript. We continue to work on OVO and expect to include this type of analysis in a new manuscript.

(3) The authors perform de novo motif analysis to identify novel OVO binding motifs in their ChIP-seq dataset. Motif analysis can be significantly strengthened by comparing DNA sequences within peaks, to sequences that are just outside of peak regions, thereby generating motifs that are specific to peak regions compared to other regions of the promoter/genome. For example, taking the 200 nt sequence on either side of an OVO peak could be used as a negative control sequence set. What control sequence set did the authors use as for their de novo motif analysis? More detail on this is necessary in the materials and methods section. Re-analysis with an appropriate negative control sequence set is suggested if not previously performed.

We apologize for being unclear on negative sequence controls in the methods. We used shuffled OVO ChIP-seq peak sequences as the background for the de novo motif analysis, which we will better outline in the methods of the revision. This is a superior background set of sequences as it exactly balances GC content in the query and background sequences. We are not fond of the idea of using adjacent DNA that won’t be controlled for GC content and shadow motifs. Furthermore, the de novo OVO DNA binding motifs are clear, statistically significant variants of the characterized in vitro OVO DNA binding motifs previously identified (Lu et al., 1998; Lee and Garfinkel, 2000; Bielinska et al., 2005), which lends considerable confidence. We also show that the OVO ChIP-seq read density are highly enriched for all our identified motifs, as well as the in vitro motifs. We provide multiple lines of evidence, through multiple methods, that the core OVO DNA binding motif is 5’-TAACNGT-3’. We have high confidence in the motif data.

We have added the below text to the methods section for further clarity on motif analysis parameters.

Lines 808-812

“The default parameters were used for de novo motif enrichment analysis, including the use of shuffled input sequences as a control. After identifying ‘OVO Motif One’, OVO ChIP peaks that contained that sequence were removed and the resulting ChIP peaks were resubmitted for STREME analysis deriving derivative OVO DNA binding motifs like above.”

(4) The authors mention that OVO binding (based on their ChIP-seq data) is highly associated with increased gene expression (lines 433-434). How many of the 3,094 peaks (conservative OVO binding sites), and what percentage of those peaks, are associated with a significant increase in gene expression from the RNA-seq data? How many are associated with a decrease in gene expression? This information should be added to the results section.

Not including the numbers of the overlapping ChIP peaks and expression changes in the text was an oversight on our part. The numbers that relate to this (666 peaks overlapping genes that significantly increased in expression, significant enrichment according to Fishers exact test, 564 peaks overlapping genes that significantly decreased in expression, significant depletion according to Fishers exact test) are found in figure 4C and will be added to the text.

We have modified the results section to include the overlap between the RNA-seq and ChIP-seq data.

Lines 463-468

“We found that 2,298 genes that were expressed in our RNA-seq data overlapped an OVO ChIP peak. 666 genes significantly increased in expression and were bound by OVO, which is a significant enrichment according to a Fisher’s exact test (Figure 4C, cyan dots, p < 0.01, odds ratio = 2.21). While conversely, 564 genes decreased in expression and were bound by OVO, indicating a significant depletion according to a Fisher’s exact test (Figure 4C, blue dots, p < 0.01, odds ratio = 0.85).”

(5) The authors mention that a change in endogenous OVO expression cannot be determined from the RNA-seq data due to the expression of the OVO-B cDNA rescue construct. Can the authors see a change in endogenous OVO expression based on the presence/absence of OVO introns in their RNA-seq dataset? While intronic sequences are relatively rare in RNA-seq, even a 0.1% capture rate of intronic sequence is likely to be enough to determine the change in endogenous OVO expression in the rescue construct compared to the OVO null.

This is a good point. The GAL4 transcript is downstream of ovo expression in the hypomorphic ovoovo-GAL4 allele. We state in the text that there is a nonsignificant increase in GAL4 expression with ectopic rescue OVO, although the trend is positive. We calculated the RPKM of RNA-seq reads mapping to the intron spanning exon 3 and exon 4 in ovo-RA and found that there is also a nonsignificant increase in intronic RPKM with ectopic rescue OVO (we will add to the results in the revision). We would expect OVO to be autoregulatory and potentially increase the expression of GAL4 and/or intronic reads, but the ovoovoGAL4>UASp-OVOB is not directly autoregulatory like the endogenous locus. It is not clear to us how the intervening GAL4 activity would affect OVOB activity in the artificial circuit. Dampening? Feed-forward? Is there an effect on OVOA activity? Regardless, this result does not change our interpretation of the other OVO target genes.

We have added the analysis of intronic ovo RNA-seq to the results as outlined below.

Lines 512-520

“Transcriptionally, ovo RNA-seq reads are likely derived from the UASp-3xFHA-OVO-B cDNA rescue or are indistinguishable between the genomic locus and rescuing cDNA transgene. We found a nonsignificant increase in exon 3 to exon 4 intronic ovo reads with the expression of ectopic rescue OVO (log2 fold change = 0.76, p-adj = 0.26). These intronic reads would be derived from the endogenous ovo locus, but it is difficult to conclusively determine if the endogenous ovo locus would respond transcriptionally to ectopic OVO downstream of UASp (for example, the pathway for ovo is no longer autoregulatory in ovoovo-GAL4/ovoΔBP; UASp-3xFHA-OVO-B germ cells, there is an additional GAL4>UASp activation step). So, we could not confidently assess whether ovo responded transcriptionally to ectopic rescue OVO.”

(6) The authors conclude with a model of how OVO may participate in the activation of transcription in embryonic pole cells. However, the authors did not carry out any experiments with pole cells that would support/test such a model. It may be more useful to end with a model that describes OVO's role in oogenesis, which is the experimental focus of the manuscript.

We did not complete any experiments in embryonic pole cells in this manuscript and base our discussion on the potential dynamics of OVO transcriptional control and our previous work showing maternal and zygotic OVO protein localization in the developing embryonic germline. Obviously, we are highly interested in this question and continue to work on the role of maternal OVO. We agree that we are extended too far and will remove the embryonic germ cell model in the figure. We will instead focus on the possible mechanisms of OVO gene regulation in light of the evidence we have shown in the adult ovary, as suggested.

We have removed figure 7 and have re-written the last two paragraphs of the discussion as below.

Lines 645-663

“The requirement for OVO at the TSS of target genes has been well characterized at its own locus as well as its downstream target otu. Our OVO ChIP and expression data confirm findings from previous work that OVO is binding to these target promoters, and in the case of otu, strongly responds transcriptionally to the presence of OVO. Although we did not test the requirement for OVO DNA binding motifs at other OVO bound genes in this work, this has been extensively explored before, showing that removal of OVO

DNA binding sites overlapping the TSS results in a strong decrease in reporter expression (Lü et al. 1998; Bielinska et al. 2005; Lü and Oliver 2001). Removal of more distal upstream OVO DNA binding sites also reduces reporter expression to a lesser degree. However, for most cases tested, removal of OVO DNA binding sites while leaving the rest of the enhancer regions intact, never totally abolished reporter expression. These dynamics are highly similar to work that has been completed on the pioneer factor zelda (zld). Adding zld DNA binding motifs to a stochastically expressed transcriptional reporter increases the activity and response of the reporter (Dufourt et al. 2018). Distally located zld DNA binding motifs influenced reporter expression to a lesser degree than proximal sites. A single zld DNA binding site adjacent to the TSS produced the strongest reporter activity. Importantly, just like the activity of OVO transgenic reporters, there is not an absolute requirement for zld DNA binding to activate reporter expression, however, the addition of TSS adjacent zld DNA binding motifs does strongly influence reporter response. We know that zld achieves this reporter response through its pioneering activity (Xu et al. 2014; Harrison et al. 2011), whether OVO achieves this similar effect on gene expression through a shared mechanism, or in cooperation with other transcription factors needs to be further explored.”

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

The Results section could be streamlined by limiting the discussion of analysis to only those details that are unusual or essential for understanding the science. For example, the fact that MACS3 was used to call peaks seems most suitable for the Methods section.

We have removed the below excerpts from the results section to streamline the text.

‘We compared immuno-purified OVO associated DNA with input DNA as a control, for a total of 12 ChIPseq libraries, which we sequenced using the Illumina system. After quality control and alignment to the Drosophila r6.46 genome (Gramates et al. 2022), we used MACS3 (Zhang et al. 2008)’

The Supplemental Tables are referred to out of order. Table S2 is referred to on line 143 while Table S1 is not referred to until the Methods section.

We have reorganized the order of the tables in the manuscript text.

In the analysis of CAGE-seq data, it is unclear whether there is anything distinctive about the ~2000 regions bound by OVO but that is not near TSS in the ovary dataset. Are these TSS that are not active in the ovary or are these non-promoter bound OVO sites? If they are TSS of genes not in the CAGE-seq data set, are these genes expressed in other tissues or just expressed at lower levels in the ovary?

This was a good point that prompted us to take a closer look at the characteristics of OVO binding and its relationships to promoters and other gene elements. 45% of OVO ChIP peaks overlapped the TSS while 55% were either non-overlapping downstream or upstream of the TSS. When plotting OVO ChIP read density, there was still a striking enrichment of OVO binding over the TSS, even though the ChIP peak was not overlapping the TSS (new figure 1K). This is possibly due to weaker direct OVO binding at the TSS that was not considered significant in the peak calling software or were indirect interactions of the distal OVO binding and the TSS. We outline this in the below text added to the results section on the OVO ChIP. To showcase these results, we have included a new panel in figure 1K. We removed the panel showing the enrichment over the cage-seq TSS, but this same data remains in the heatmap shown in figure 1L, so no information is lost. To directly answer the Cage-seq questions considering the OVO bound over the annotated TSS results, we found that 1,047 chip peaks overlapped CAGE-seq TSS, which is only 347 fewer than the annotated TSS overlap (1,394). Of the 1,394 genes that were bound by over the TSS, all of them were considered to be expressed in our RNA-seq dataset, indicating that these might just be more lowly expressed genes that for whatever reason were not considered to be enriched TSSs in the CAGE-seq data. This difference is likely not significant.

Lines 235-251

“Although OVO ChIP peaks overlapping genes showed a strong read density enrichment over the TSS, we found that only 45% (1,394/3,094) of OVO ChIP peaks directly overlapped a TSS. 43% (1,339/3,094) of OVO ChIP peaks were found to overlap the gene body downstream of the TSS (intronic and exonic sequences) and 12% (366/3,094) did not overlap any gene elements, indicating that they were intergenic.

We were interested in the differences between OVO binding directly over the TSS or at more distal upstream and downstream sites. We decided to plot the OVO ChIP read density of these different classes of OVO binding patterns and found that OVO bound over the TSS produced a sharp read density enrichment over the TSS which was consistent with what was found for all OVO bound genes (Figure 1K). OVO binding along the gene body surprisingly also showed a read density enrichment over the TSS, although the magnitude of read density enrichment was notably less than TSS OVO binding. Intergenic OVO binding also showed these same characteristics with a notable upstream read density enrichment possibly indicative of enhancer binding. This indicates that although the significantly called OVO ChIP peaks did not overlap the TSS, there was still a propensity for TSS sequences to be enriched with OVO ChIP over the input control. This could be due to weaker direct in vivo binding of OVO to these TSSs or indirect interactions between the upstream/downstream OVO bound sequences and the TSS, possibly through a looping enhancer-promoter interaction. However, regardless of the location of the OVO ChIP peak, OVO seemed to always be enriched at or in close proximity to TSSs.”

It would be helpful for the authors to provide a bit more detailed analysis of chromatin states of OVObound regions in GSC, 8c NC, and 32c NC (or some more clarity in the current analysis). Are the regions that are bound by OVO accessible in all these cell types or specifically enriched for accessibility in a subset? The authors state that OVO binding is correlated with open chromatin, but whether these are regions that are open in all cell types analyzed or a subset is not clear from the data presented. Promoters are often accessible regardless of cell type, so it is unclear what exactly is to be concluded from this association. Also, is the proximity to open chromatin features for OVO-bound promoters (as shown in Figure 2C) different than non-OVO-bound promoters (the two classes shown Figure 1L, for example)?

We utilized previously published datasets of staged germ cell chromatin status to look at the association of chromatin status and OVO binding. Unfortunately, not all the same germ cell stages were profiled for each chromatin mark from the datasets derived for these two papers. For example, only H3K4me3 data exists for GSCs, and only gsc and 8c data exists for H3K9me3, while the other chromatin marks had more profiles, even including later stages. We focused specifically on gsc and 32c (essentially stage 5 egg chambers) for the other chromatin marks since that is when the ovo hypomorphic egg chambers arrest. A nice control would have been chromatin states in somatic follicle cells of the ovary, since we know germ cell genes such as ovo and otu are not expressed and presumably the chromatin states in somatic cell types would be different than germ cells. However, chromatin states for somatic follicle cells were not published in these two papers and we are not aware of any other existing datasets to compare too. Essentially, we need to determine the changes in chromatin states with and without OVO, which we are currently working on. 

We did further analyze chromatin states and differential OVO binding in respect to gene elements, and found that OVO binding, regardless of the relationship to the gene element, is always open (gsc and 32c ATAC). OVO binding over the gene body shows the same enrichment for open chromatin and transcriptionally active histone marks. We compared the profiles of these chromatin marks and the promoters of OVO bound and not bound genes and consistent with the suggestion that promoters are generally open, we found that this was the case. However, there is an enrichment for open chromatin and transcriptionally active histone marks for OVO bound genes compared to non-OVO bound genes. This could be a consequence of OVO binding or indirect consequence of a downstream OVO target. Regardless, as has been suggested, future experiments directly measuring chromatin status and OVO needs to be performed. The below excerpts have been added to the text to supplement the comments provided above.

Lines 328-343

“The association of OVO binding with active histone marks and open chromatin was striking, but open chromatin is likely a general phenomenon of promoters (Haines and Eisen, 2008). Indeed, when measuring the read density for GSC and 32C ATAC-seq for OVO bound and OVO non-bound promoters, there is an enrichment for open chromatin at the TSS regardless of OVO binding. However, we did notice an increase in enrichment for OVO bound promoters compared to OVO non-bound promoters (Figure S1G), possibly suggesting that OVO bound promoters are more open or have an increase in accessibility when compared to non-OVO bound promoters. This same relationship held true for the transcriptionally active histone mark H3K27ac in GSCs (Figure S1H). Since only 45% of OVO ChIP peaks overlapped TSSs, we plotted the read density of the above chromatin marks over OVO ChIP peak maximums for OVO bound over the TSS, gene body, or intergenic regions (Figure S2A-D). We found that OVO bound regions that were not overlapping the TSS still showed the same propensity for enrichment of open chromatin and active histone marks. Intergenic regions were especially enriched for open chromatin measured through ATAC-seq. Altogether suggesting that OVO binding genome-wide is tightly associated with open chromatin regardless of germ cell stage, and active transcription in GSCs. In other words, chromatin state data suggests OVO is acting positively on its target genes and raises the possibility that OVO-binding and open chromatin are related.”

For clarity, it would help the reader if the authors mentioned the male-specific TATA-associated factors as a rationale for testing the role of OVO binding in core promoter function. This is currently mentioned in the Discussion on lines 575-577, but would help in understanding the motivation behind the detailed analysis of the promoter binding of OVO in the Results and make the negative result more clearly impactful.

We have introduced the male specific tata factors as suggested and have condensed the two intro paragraphs in this section into one, as shown below.

Lines 347-363

“Our data thus far clearly indicates that OVO binding occurs at or very near the core promoter, a region recognized by an enormous collection of factors that associate with RNA polymerase to initiate transcription (Aoyagi and Wassarman 2000; Vo Ngoc, Kassavetis, and Kadonaga 2019). The highly organized polymerase complex has sequence-specific DNA recognition sites with incredibly precise spacing between them, with an overall DNA footprint of a little less than 100bp (Rice, Chamberlin, and Kane 1993; FitzGerald et al. 2006; Ohler et al. 2002). There are upstream binding sites such as TATA, sites at transcription start, such as the initiator (INR), and downstream promoter elements (DPE) (Vo Ngoc, Kassavetis, and Kadonaga 2019). The combinations of these DNA motifs is not random in mammals and Drosophila (FitzGerald et al. 2006), and distinct combinations of different motifs at the TSS of genes expressed in Drosophila are conserved over tens of millions of years of evolution (Chen et al. 2014). The male germline expresses a number of TATA-associated factors that have been implicated in male-specific promoter usage for gene expression (M. Hiller et al. 2004; M. A. Hiller et al. 2001; Lu et al. 2020; V. C. Li et al. 2009). It is possible that OVO is a female germline specific TATA-associated factor, and if so, OVO binding sites at core promoters should share precise spacing with other core promoter elements, suggesting it is likely part of the complex. If not, then OVO is more likely to facilitate binding of the basal transcriptional machinery. Because of the extended footprint of engaged RNA polymerase, OVO and the basal machinery would not be likely to occupy the same region at the same time.”

The description of the system used for the RNA-seq would benefit from additional clarity. It is not clear as written why it is "Lucky" that there is an mRNA isoform with extended exon 2 required for egg chamber development beyond stage 5. How does this requirement compare to the global requirement for OVO, which seems to be required for germ cell development even before stage 5? Understanding this system is essential for interpreting the RNA-seq results. Indeed, the authors have a separate manuscript (currently on bioRxiv) that explains the details of this system. As such, the current description requires that the reader refer to this additional pre-print. Could the authors include a diagram to better illustrate this system? Furthermore, since this RNA-seq is being performed on tissue that includes nurse cells, follicle cells, and germ cells from multiple stages of development, it is important for the authors to clearly state in which cell types OVO is expressed and likely functional. (While this is well beyond this manuscript, this analysis is the type that might benefit from the use of single-cell sequencing as a means to deconvolute the phenotypic effects of OVO loss.)

We have rewritten the text to better describe the system for RNA-seq. We have also included a figure (Figure S1A) showing the alleles used that should help provide clarity for the readers. We agree that moving forward single cell experiments will be critical to have a better understanding of the transcriptional changes and chromatin dynamics with and without OVO. We have included the below changes to the text.

Lines 409-423

“Previous work from our lab has identified a transheterozygous ovo allelic combination (ovoovo-GAL4/ovoΔBP) that greatly reduces OVO activity resulting in sterility, however, female germ cells are able to survive up until at least stage 5 of oogenesis (Benner et al. 2023). ovoovo-GAL4 is a CRISPR/Cas9 derived T2A-GAL43xSTOP insertion upstream of the splice junction of exon 3 in the ovo-RA transcript (Figure S1A).

Importantly, this insertion in the extended exon 3 would disrupt roughly 90% of the ovo-B transcripts. However, since about 10% of ovo-B transcripts utilize an upstream splice junction in exon 3, these transcripts would not be disrupted with the T2A-GAL4-3xSTOP insertion and thus allow for enough OVO activity for germ cell survival (Benner et al. 2023). Since ovoovo-GAL4 expresses GAL4 in place of full length OVO due to the T2A sequences, we can drive expression of a rescuing OVO-B construct downstream of UASp to generate OVO+ female germ cells, which in fact does rescue the arrested germ cell phenotype of ovoovo-GAL4/ovoΔBP ovaries. Therefore, in order to determine genes that are transcriptionally responsive to OVO, we compared the gene expression profiles in sets of ovaries that had the ovo hypomorphic phenotype with a negative control rescue construct (ovoovo-GAL4/ovoΔBP; UASp-GFP)(Figure 4A) versus those that drive expression of the rescue construct expressing OVO-B (ovoovo-GAL4/ovoΔBP; UASp-3xFHAOVO-B)(Figure 4B).”

Lines 427-432

“The adult female ovary contains somatic cells, germline stem cells, and germline derived nurse cells that would be profiled in a bulk ovary tissue RNA-seq experiment. Although OVO is only required and expressed in germline derived cell types, we chose to dissect one day old post-eclosion ovoovoGAL4/ovoΔBP; UASp-3xFHA-OVO-B female ovaries to enrich for early stages of oogenesis and collected only ovarioles containing the germarium through previtellogenic egg chambers.”

On lines 526-532, it is unclear why the genes fs(1)N, fs(1)M3, and closca are particularly sensitive to the ovoD3 allele. What is this allele trans heterozygous with in the assay that allows development through egg laying? Why might these genes be unique in their sensitivity?

These genes are not particularly sensitive, the transheterozygous hypomorphic ovo ovaries are weak enough to reveal the role of OVO for these genes. We rewrote this paragraph to try and provide more clarity to the relationship between OVO+ binding at these vitelline membrane genes and the phenotype of OVOD3 expressing females.

Lines 562-577

“We also found that the genes fs(1)N, fs(1)M3, and closca, were all bound by OVO and responded transcriptionally to the presence of ectopic rescue OVO. These genes are significant because they constitute a set of genes that are expressed in the germline and the encoded proteins are eventually incorporated into the vitelline membrane providing the structural integrity and impermeability of the egg (Mineo, Furriols, and Casanova 2017; Ventura et al. 2010). Loss-of-function of these three genes results in flaccid eggs that are permeable to dye and fail to develop. The loss-of-function phenotype of fs(1)N, fs(1)M3, and closca closely resembles the dominant antimorph ovoD3 phenotype. The ovoD3 allele is the weakest of the original dominant-negative ovo alleles and produces defective eggs allowing us to explore the role of OVO in late stages (Busson et al. 1983; Komitopoulou et al. 1983). ovoD3/ovo+ transheterozygous females express a repressive form of OVO that results in dominant sterility, and importantly, these females lay flaccid eggs with compromised vitelline membranes that are permeable to the dye neutral red (Oliver, Pauli, and Mahowald 1990). Since OVO+ is bound at the TSS of fs(1)N, fs(1)M3, and closca, and these three genes respond transcriptionally to OVO+, then it is plausible that the repressive OVOD3 is negatively regulating these three genes that are required for vitelline membrane formation. This is evidence that OVO is not only involved in regulating the expression of numerous essential maternal pathways for embryonic development, but it is also essential for regulating genes that are required for egg integrity and maturation.”

The Discussion of OVO as a pioneer factor is highly speculative and based only on correlative data. In fact, the expression data in the embryonic germline is not included in this manuscript, but rather in a separate bioRxiv preprint. This makes it challenging to understand, why this is extensively discussed here. However, there are experiments that could begin to test this proposal. OVO could be expressed in an exogenous tissue and test whether it promotes accessibility. Also, mutations could be made (using gene editing) to identify previously known OVO binding sites in the otu and/or other promoters and these could be assayed for accessibility. By selecting promoters of genes that are not essential for germline development, the authors could directly test the role of OVO in promoting chromatin accessibility. Alternatively, are there reasons that the system used for RNA-seq couldn't be similarly used for ATACseq? It is imperfect but could provide insights into chromatin accessibility in the absence of OVO.

We have largely removed the speculation on pioneering activity, reference to embryonic germline OVO dynamics included in the previous work, and Figure 7. These are excellent suggestions for experiments and ones we are currently pursuing. Below is the modified discussion. 

Lines 645-663

“The requirement for OVO at the TSS of target genes has been well characterized at its own locus as well as its downstream target otu. Our OVO ChIP and expression data confirm findings from previous work that OVO is binding to these target promoters, and in the case of otu, strongly responds transcriptionally to the presence of OVO. Although we did not test the requirement for OVO DNA binding motifs at other OVO bound genes in this work, this has been extensively explored before, showing that removal of OVO

DNA binding sites overlapping the TSS results in a strong decrease in reporter expression (Lü et al. 1998; Bielinska et al. 2005; Lü and Oliver 2001). Removal of more distal upstream OVO DNA binding sites also reduces reporter expression to a lesser degree. However, for most cases tested, removal of OVO DNA binding sites while leaving the rest of the enhancer regions intact, never totally abolished reporter expression. These dynamics are highly similar to work that has been completed on the pioneer factor zelda (zld). Adding zld DNA binding motifs to a stochastically expressed transcriptional reporter increases the activity and response of the reporter (Dufourt et al. 2018). Distally located zld DNA binding motifs influenced reporter expression to a lesser degree than proximal sites. A single zld DNA binding site adjacent to the TSS produced the strongest reporter activity. Importantly, just like the activity of OVO transgenic reporters, there is not an absolute requirement for zld DNA binding to activate reporter expression, however, the addition of TSS adjacent zld DNA binding motifs does strongly influence reporter response. We know that zld achieves this reporter response through its pioneering activity (Xu et al. 2014; Harrison et al. 2011), whether OVO achieves this similar effect on gene expression through a shared mechanism, or in cooperation with other transcription factors needs to be further explored.”

The authors suggest that OVO binding is essential for transcriptional activation, but that this may be indirect and that expression of other transcription factors might be necessary for activating gene expression. Did the motif analysis of the OVO-bound regions suggest additional transcription factors that might provide this function?

We did find other motifs significantly enriched in OVO ChIP peaks. We performed XSTREME analysis on the same set of OVO ChIP peaks which allowed us to determine if any of these motifs were significant matches to DNA binding motifs of known transcription factors. Notably, the DNA binding motifs of GAF and CLAMP were enriched in OVO ChIP peaks. GAF is required in germline clones and the potentially for co-regulation of genes is possible. Other enriched motifs did not match any known binding motifs of other transcription factors but we reported some of the most significantly enriched motifs that were alongside of OVO in Figure S1C-F. The below text outlines changes made to the text incorporating these findings.

Lines 170-182

“Along with the OVO DNA binding motif, other motifs were also significantly enriched in OVO ChIP peaks. The motif 5’-GWGMGAGMGAGABRG-3’ (Figure S1C) was found in 18% of OVO ChIP peaks and is a significant match to the DNA binding motifs of the transcription factors GAF (Trl) (Omelina et al. 2011) and CLAMP (Soruco et al. 2013). Trl germline clones are not viable, indicating that GAF activity is required in the germline during oogenesis (Chen et al. 2009). The possibility that OVO binds with and regulates genes alongside of GAF given the enrichment of both transcription factors DNA binding motifs is intriguing. Other significantly enriched motifs 5’-ACACACACACACACA-3’ (29% of peaks, Figure S1D), 5’RCAACAACAACAACA-3’ (26% of peaks, Figure S1E), and 5’-GAAGAAGAAGAAGAR-3’ (17% of peaks,

Figure S1F) were present in OVO ChIP peaks, however, these motifs did not significantly match known

DNA binding motifs of other transcription factors. Determining the factors that bind to these sequences

will certainly help elucidate our understanding of transcriptional control with relationship to OVO in the female germline.”

The figures would benefit from a bit more detail in the legends (see comments below).

Minor comments:

In multiple places throughout the document, the citations are inadvertently italicized (see lines 57-59, 91, and 327 as examples.)

We have changed this in these locations and other instances in the text.

On line 76, when discussing OVO as a transcription factor this is referencing the protein and not the gene. Thus, should be written OVO and not ovo.

We have made the correction ovo to OVO.

On line 349, "core" promoters is likely what is meant rather than "care" promoters.

We have corrected ‘care’ to ‘core’ in the text.

On line 404, the authors state that they wanted to use a "less conservative log2 fold change" but it is not clear what they are comparing to. This is important to understand the motivation.

We are talking about the gene expression comparison between the ectopic ovo rescue and ovo hypomorphic ovaries. “less conservative” was an unfortunate phrasing. We have rewritten the text to state this directly to the reader.

Lines 435-444

“We then performed RNA-seq in quadruplicate and measured the changes in gene expression between ectopic rescue OVO and hypomorphic OVO ovaries. We used a significance level of p-adj < 0.05 and a log2 fold change cutoff of >|0.5| to call differential expression between these two sets of ovaries. We utilized these log2 fold change cutoffs for two reasons. Our control ovary genotype (ovoovo-GAL4/ovoΔBP; UASp-GFP) has hypomorphic OVO activity, hence germ cells can survive but are arrested. With the addition of ectopic rescue OVO in ovoovo-GAL4/ovoΔBP; UASp-3xFHA-OVO-B ovaries, we predicted that genes that were directly regulated by OVO would transcriptionally respond, however, we were unsure as to what degree the response would be in comparison to hypomorphic OVO. We reasoned that if the changes were not significant between genotypes, then minor changes in gene expression would not matter.”

On line 615, it is unclear what is meant by "showing expression with only 10s of bp of sequence in reporters."

This is in reference to some of the previously studied ovo reporter deletion lines, however, we have decided to remove the below text in the revised discussion.

“, despite being remarkably compact. The OVO-dependent ovo core promoter is very compact; showing expression with only 10s of bp of sequence in reporters.” 

It would be useful to cite and discuss Dufourt et al. Nature Communications 2018 (PMID30518940) regarding the role of Zelda in potentiating transcriptional activation when mentioned on line 624.

We have added this and the relationship to previous similar work on OVO in the discussion.

Lines 645-663

“The requirement for OVO at the TSS of target genes has been well characterized at its own locus as well as its downstream target otu. Our OVO ChIP and expression data confirm findings from previous work that OVO is binding to these target promoters, and in the case of otu, strongly responds transcriptionally to the presence of OVO. Although we did not test the requirement for OVO DNA binding motifs at other OVO bound genes in this work, this has been extensively explored before, showing that removal of OVO

DNA binding sites overlapping the TSS results in a strong decrease in reporter expression (Lü et al. 1998; Bielinska et al. 2005; Lü and Oliver 2001). Removal of more distal upstream OVO DNA binding sites also reduces reporter expression to a lesser degree. However, for most cases tested, removal of OVO DNA binding sites while leaving the rest of the enhancer regions intact, never totally abolished reporter expression. These dynamics are highly similar to work that has been completed on the pioneer factor zelda (zld). Adding zld DNA binding motifs to a stochastically expressed transcriptional reporter increases the activity and response of the reporter (Dufourt et al. 2018). Distally located zld DNA binding motifs influenced reporter expression to a lesser degree than proximal sites. A single zld DNA binding site adjacent to the TSS produced the strongest reporter activity. Importantly, just like the activity of OVO transgenic reporters, there is not an absolute requirement for zld DNA binding to activate reporter expression, however, the addition of TSS adjacent zld DNA binding motifs does strongly influence reporter response. We know that zld achieves this reporter response through its pioneering activity (Xu et al. 2014; Harrison et al. 2011), whether OVO achieves this similar effect on gene expression through a shared mechanism, or in cooperation with other transcription factors needs to be further explored.”

On line 1006 (Figure 1 legend), it is unclear what is meant by "The percentage of OVO ChIP peaks each motif was found". Is a word missing?

This was unclear, we have revised the sentence below.

Lines 1035-1036

“The percentage of OVO ChIP peaks containing each motif and their corresponding p-value are indicated to the right.”

In the Figure 1 legend, please include citations for the Garfinkel motif and Oliver motif.

Included, as below.

Lines 1036-1039

“H) OVO ChIP minus input control ChIP-seq read coverage density centered on the location of the four de novo OVO DNA binding motifs and previously defined in vitro OVO DNA binding motifs (Lü et al. 1998, Bielinska et al. 2005, Lee and Garfinkel 2000).”

In Figure 2 legend, it is unclear if B is all instances of a given motif or the DNA motifs that are bound by ChIP. Please clarify.

We meant only the OVO DNA binding motifs that were within significant OVO ChIP peaks. We have revised the legend below.

Lines 1049-1052

“A, B) OVO ChIP minus input control, GSC and 32c ATAC-seq, GSC H3K27ac, H3K4me3, H3K27me3, H3K9me3, 8c NC H3K9me3, 32c NC H3K27ac, and H3K27me3 ChIP-seq read coverage density centered on each OVO peak maximum or OVO DNA binding motif located within a significant OVO ChIP peak.”

The Figure legend for 2D could use more explanation. What do the lines and circles indicate?

These lines and circles indicate the amount of overlapping peaks measured between the two datasets with solid circles. We have included a better description of what these indicate in the figure legend.

Lines 1054-1058

“D) Total number of significant peaks (left) and the total number of overlapping peaks (top) between OVO

ChIP and GSC and 32c ATAC-seq, GSC H3K27ac, H3K4me3, H3K27me3, H3K9me3, 8c NC H3K9me3, 32c NC H3K27ac, and H3K27me3 ChIP-seq. Lines connecting solid dots indicates the amount of overlapping peaks between those two corresponding datasets.”

In Figure 4C, bring the 564 blue dots forward so they are not masked by the yellow dots.

We have brought the colored dots forward in both figure 4C and 4D.

In Figure 4E, what is the order of the heatmaps?

The order is genes with the highest to lowest OVO read density enrichment. We have included this in the figure 4 legend.

Lines 1086-1087

“The order of the heatmap is genes with the highest to lowest amount of OVO ChIP read density.”

In Figure 5, the order of the tracks is not immediately obvious. It appears to be those chromatin features most associated with OVO ChIP and those less correlated. Additional clarity could be provided by showing these tracks (and in Supplemental Figure S2) in different colors with a reference to the figure legend about what the colors might indicate.

We have changed the colors and order of the tracks to be more similar and consistent in both figures.

Lines 1090-1093

“ovo gene level read coverage tracks for OVO ChIP minus input (black), GSC and 32c ATAC-seq (light blue), GSC and 32C H3K27ac (green), H3K4me3 (dark blue), GSC and 32c H3K27me3 (orange), and GSC and 8c H3K9me3 (pink) ChIP-seq, and ovoΔBP/ovoovo-GAL4; UASp-3xFHA-OVO-B minus ovoΔBP/ovoovo-GAL4; UASp-GFP RNA-seq (red).”

In Figure S1 legend, what is the reference to the da-GAL4 X UAS transgene in the title?

This was an error on our part and we have removed it.

Reviewer #2 (Recommendations For The Authors):

Overall, the manuscript would benefit from revisions of the writing style. At times it is difficult to distinguish between hypothesis and results. The use of colloquial phrases/prose was distracting while reading, which the authors may consider revising. Some sentences were confusing or extraneous, and the authors may consider revising those. Occasionally sentences within the results sections seem more appropriate for the materials and methods.

(1) The manuscript is generally clear; however, it is at times difficult to distinguish between hypothesis and results. The use of colloquial phrases/prose was distracting while reading, which the authors may consider revising. Examples include:

a)  Lines 48-49 "While thematic elements of this complex orchestration have been well studied, coordinate regulation of the symphony has not."

We have edited this sentence below.

Lines 48-50

“While the complex interactions between maternally supplied mRNAs and proteins have been well studied, transcriptional regulation driving the expression of these pathways are less well understood.“

b)  Lines 232-233 "In other words, where exactly does transcription start at these genes."

We have removed this sentence.

c)  Line 385, the word "sham" could be changed to "negative control" or "GFP control"

We have rewritten this sentence below.

Lines 419-423

“Therefore, in order to determine genes that are transcriptionally responsive to OVO, we compared the gene expression profiles in sets of ovaries that had the ovo hypomorphic phenotype with a negative control rescue construct (ovoovo-GAL4/ovoΔBP; UASp-GFP)(Figure 4A) versus those that drive expression of the rescue construct expressing OVO-B (ovoovo-GAL4/ovoΔBP; UASp-3xFHA-OVO-B)(Figure 4B)”

d)  Line 490 "For the big picture"

We have removed this and revised with the below sentence.

Lines 530-531

“To do this, we performed Gene Ontology enrichment analysis with gProfiler software (Raudvere et al. 2019).”

(2) Some sentences were confusing or extraneous, and the authors may consider revising them. Examples include:

a)  Lines 195-196 "Therefore, we plotted the significant ChIP (minus input) read density peaks centered on the location of the motif itself."

We have removed the word ‘peaks’ and ‘itself’, as below.

Lines 200-201

“Therefore, we plotted the significant ChIP (minus input) read density centered on the location of the motif.”

b)  Lines 201-203 "... over the location of the motifs, strongly reinforces the idea that our dataset contains regions centered on sequence-specifically bound OVO transcription factor in the ovary."

We have edited this sentence to clarify below.

Lines 204-208

“While it is possible that OVO comes into contact with regions of DNA in three-dimensional nuclear space non-specifically, the presence of OVO motifs within a large percentage of significant ChIP peaks in vivo and enrichment of OVO ChIP read density at the location of the motifs, strongly reinforces the idea that our OVO ChIP dataset contains regions centered on sequences specifically bound by OVO in the ovary.”

c)  Lines 326-328 "The combinations of these elements...tens of millions of years of evolution."

We have revised this sentence below.

Lines 354-357

“The combinations of these DNA motifs is not random in mammals and Drosophila (FitzGerald et al. 2006), and distinct combinations of different motifs at the TSS of genes expressed in Drosophila are conserved over tens of millions of years of evolution (Chen et al. 2014).”

d)  Lines 444-446 "To address this directly, we tested the idea that genes with... and thus downstream of OVO."

We have removed this sentence in its entirety.

e)  Line 579-580 "Where OVO binding in close proximity, in any ...activates transcription"

We have removed this sentence in its entirety.

(3)    Occasionally sentences within the results sections seem more appropriate for the materials and methods. For example, lines 213-218.

(4)    At the end of line 375, do the authors mean "only" instead of "also"?

We have modified this sentence below.

Lines 411-414

“ovoovo-GAL4 is a CRISPR/Cas9 derived T2A-GAL4-3xSTOP insertion upstream of the splice junction of exon 3 in the ovo-RA transcript (Figure S1A). Importantly, this insertion in the extended exon 3 would disrupt roughly 90% of the ovo-B transcripts. However, since about 10% of ovo-B transcripts utilize an upstream splice junction in exon 3, these transcripts would not be disrupted with the T2A-GAL4-3xSTOP insertion and thus allow for enough OVO activity for germ cell survival (Benner et al. 2023).”

(5)    In line 392 the authors say that they dissected ovaries "one day post-eclosion" but the methods section says that ovaries were 3-5 days old. Please clarify.

We meant one day old for the RNAseq experiments. We have changed this in the text.

Lines 679-681

“Twenty, one day old post-eclosion ovoΔBP/ovoovo-GAL4; UASp-GFP and ovoΔBP/ovoovo-GAL4; UASp-3xFHAOVO-B ovaries were dissected and germariums through previtellogenic egg chambers were removed with microdissection scissors and placed in ice cold PBS making up one biological replicate.”

(6)    In line 668 the authors mention CRISPR/Cas9 in the methods, but no such experiment was described.

We have removed this from the Methods header.

Reviewer #1 (Public Review):

(1) Significance of findings and strength of evidence.

(a) The work presented in this manuscript is intended to support the authors' novel idea that HIV DNA integration strongly favors "triple-stranded" R-loops in DNA formed either during transcription of many, but not all, genes or by strand invasion of silent DNA by transcripts made elsewhere, and that HIV infection promotes R-loop formation mediated by incoming virions in the absence of reverse transcription. The authors were able to demonstrate a reverse transcription-independent increase in R-loop formation early during HIV infection, while also demonstrating increased integration into sequences that contain R-loop structures. Furthermore, this manuscript also identifies that R-loops are present in both transcriptionally active and silent regions of the genome and that HIV integrase interacts with R-loops. Although the work presented supports a correlation between R-loop formation and HIV DNA integration, it does not prove the authors' hypothesis that R-loops are directly targeted for integration. Direct experimentation, such as in vitro integration into defined DNA targets, will be required. Further, the authors provide no explanation as to how current sophisticated structural models of concerted retroviral DNA integration into both strands of double-stranded DNA targets can accommodate triple-stranded structures. Finally, there are serious technical concerns with the interpretation of the integration site analyses.

(2) Public review with guidance for readers around how to interpret the work, highlighting important findings but also mentioning caveats.

(a) Introduction: The authors provide an excellent introduction to R-loops but they base the rationale for this study on mis-citation of earlier studies regarding integration in transcriptionally silent regions of the genome. E "most favored locus" cited in the very old reference 6 comprises only 5 events and has not been reproduced in more recent, much larger datasets. For example, see the study of over 300.000 sites in freshly infected PBMC cited in https://doi.org/10.1371/journal.ppat.1009141, which shows a 15-fold preference for integration in expressed genes and no evidence of clustering of sites (as seen in expressed genes) in non-expressed DNA. Further, as far as I can tell, they present no examples in the Results section of R-loops in non-expressed DNA serving as integration targets.

(b) Figure 1: Demonstrates models for HIV infections in both cell lines and primary human CD4+ T cells. R-loop formation was determined through a method called DRIPc-seq which utilizes an antibody specific for DNA-RNA hybrid structures and sequences these regions of the genome using RNaseH treatment to show that when RNA-DNA hybrids are absent then no R-loops are detected. In these models of in vitro and ex vivo infection, the authors show that R-Loop formation increases following HIV infection between 6 hour post-infection and 12 hours post-infection, depending on the cell model. However, these figures lack a mock-infected control for each cell model to assess R-loop formation at the same time points. They would also benefit from a control showing that virus entry is necessary, such as omitting the VSV G protein donor.

Additionally, they use intracellular staining to confirm DRIPc-Seq results, by demonstrating an increase in R-loop formation at 6 hours post-infection in HeLa cells. It would have been more relevant to use primary T cells for this assay, but HeLa cells probably provided easier and clearer imaging.

(c) Figure 2: This figure shows that cells infected with HIV show more R-loops as well as longer sequences containing R-loop structures. Panel B shows that these R-loops were distributed throughout different genomic features, such as both genic and intergenic regions of the genome. However, the data are presented in such a way that it is impossible to determine the proportion of R-loops in each type of genomic feature. The reader has no way to tell, for example, the proportion of R-loops in genic vs intergenic DNA and how this value changes with time. Furthermore, increased R-loop formation due to HIV infection showed poor correlation with gene expression, suggesting that R-loops were not forming due to transcriptional activation, although the difference between 0 and the remaining time points is not apparent, nor is the meaning of the absurd p values.

(d) Figure 3: This figure shows the use of cell lines carrying R-loop inducible (mAIRN) or non-inducible (ECFP) genes to model the association of HIV integration with R-loop structures. The authors demonstrate the functional validation of R-loop induction in the cell line model. Additionally, when R-loops are induced there is a significant increase in HIV integration in the R-loop forming vector sequence when R-loops are induced with doxycycline. This result shows a correlation between expression and integration that is much stronger in the R-loop forming gene than in the unreferenced ECFP gene but does not prove that integration directly targets R-loops. It is possible, for example, that some features of the DNA sequence, such as base composition affect both integration and R-loop formation independently. As described more fully below, there is also a serious concern regarding the method used to quantify the integration frequencies.

(e) Figure 4: This figure shows evidence of increased HIV integration within regions of the genome containing R-loops with an additional preference for integration within the R-loop and a decrease in frequency of integration further from the R-loop. Identifying a preference for R-loops is very intriguing but the authors do also demonstrate that integration does occur when R-loops are not present. Also Panel A, which shows that regions of cell DNA that form R-loops have a higher frequency of Integration sites than those that do not, should also be controlled for the level of gene expression of the two types of region.

(f) Figure 5: In this figure, the authors demonstrate that HIV integrase binds to R-loops through a number of protein assays, but does not show that this binding is associated with enzymatic activity. ESMA of integrase identified increased binding to DNA-RNA over dsDNA. Additionally, precipitation of RNA-DNA hybrids pulled down HIV integrase. A proximity ligation assay detecting R-loops and HIV-integrase showed co-localization within the nucleus of HeLa cells. HeLa cells were probably used due to their efficiency of transduction but are not physiologically relevant cell types.

(g) Discussion: In the discussion, the authors address how their work relates to previous evidence of HIV integration by association of LEDGF/p75 and CPSF6. They also cite that LEDGF/p75 has possible R-loop binding capabilities. They also discuss what possible mechanisms are driving increases in R-loop formation during HIV infection, pointing to possible HIV accessory proteins. They also state that how HIV integrates in transcriptionally silent regions is still unknown but do point out that they were able to show R-loops appear in many different regions of the genome but did not show that R-loops in transcriptional inactive regions are integration targets. More seriously, they failed to make a connection between their work and the current understanding of the biochemical and structural mechanism of the integration reaction.

eLife assessment

Based on largely indirect evidence, this study proposes that genomic integration of HIV targets DNA/RNA hybrids called R-loops. The evidence is indirect because the authors do not use relevant models systems to show integration and because they artificially induce R-loops in the critical experiments. There are two interrelated findings: 1) VSVg-pseudotyped HIV-1 induces R-loops in various cell types, and 2) VSVg-pseudotyped HIV-1 targets R-loops for integration in an artificial Hela cell model in which R-loops are exogenously induced. The induction of R-loops by a pseudotyped HIV-1 is a potentially valuable finding. Critically, however, because of the caveats above, the evidence is inadequate to support the primary claims in the title, abstract, and manuscript. Furthermore, if these claims were true, the authors do not provide context for how they could be reconciled with well-established structural data showing that HIV-1 integrase catalyzes the integration of viral DNA into dsDNA as a substrate.

Reviewer #2 (Public Review):

Retroviral integration in general, and HIV integration in particular, takes place in dsDNA, not in R-loops. Although HIV integration can occur in vitro on naked dsDNA, there is good evidence that, in an infected cell, integration occurs on DNA that is associated with nucleosomes. This review will be presented in two parts. First, a summary will be provided giving some of the reasons to be confident that integration occurs on dsDNA on nucleosomes. The second part will point out some of the obvious problems with the experimental data that are presented in the manuscript.

(1) 2017 Dos Passos Science paper describes the structure of the HIV intasome. The structure makes it clear that the target for integration is dsDNA, not an R-loop, and there are very good reasons to think that structure is physiologically relevant. For example, there is data from the Cherepanov, Engelman, and Lyumkis labs to show that the HIV intasome is quite similar in its overall structure and organization to the structures of the intasomes of other retroviruses. Importantly, these structures explain the way integration creates a small duplication of the host sequences at the integration site. How do the authors propose that an R-loop can replace the dsDNA that was seen in these intasome structures?

(2) As noted above, concerted (two-ended) integration can occur in vitro on a naked dsDNA substrate. However, there is compelling evidence that, in cells, integration preferentially occurs on nucleosomes. Nucleosomes are not found in R loops. In an infected cell, the viral RNA genome of HIV is converted into DNA within the capsid/core which transits the nuclear pore before reverse transcription has been completed. Integration requires the uncoating of the capsid/core, which is linked to the completion of viral DNA synthesis in the nucleus. Two host factors are known to strongly influence integration site selection, CPSF6 and LEDGF. CPSF6 is involved in helping the capsid/core transit the nuclear pore and associate with nuclear speckles. LEDGF is involved in helping the preintegration complex (PIC) find an integration site after it has been released from the capsid/core, most commonly in the bodies of highly expressed genes. In the absence of an interaction of CPSF6 with the core, integration occurs primarily in the lamin-associated domains (LADs). Genes in LADs are usually not expressed or are expressed at low levels. Depending on the cell type, integration in the absence of CPSF6 can be less efficient than normal integration, but that could well be due to a lack of LEDGF (which is associated with expressed genes) in the LADs. In the absence of an interaction of IN with LEDGF (and in cells with low levels of HRP2) integration is less efficient and the obvious preference for integration in highly expressed genes is reduced. Importantly, LEDGF is known to bind histone marks, and will therefore be preferentially associated with nucleosomes, not R-loops. LEDGF fusions, in which the chromatin binding portion of the protein is replaced, can be used to redirect where HIV integrates, and that technique has been used to map the locations of proteins on chromatin. Importantly, LEDGF fusions in which the chromatin binding component of LEDGF is replaced with a module that recognizes specific histone marks direct integration to those marks, confirming integration occurs efficiently on nucleosomes in cells. It is worth noting that it is possible to redirect integration to portions of the host genome that are poorly expressed, which, when taken with the data on integration into LADs (integration in the absence of a CPSF6 interaction) shows that there are circumstances in which there is reasonably efficient integration of HIV DNA in portions of the genome in which there are few if any R-loops.

(3) Given that HIV DNA is known to preferentially integrate into expressed genes and that R-loops must necessarily involve expressed RNA, it is not surprising that there is a correlation between HIV integration and regions of the genome to which R loops have been mapped. However, it is important to remember that correlation does not necessarily imply causation.

If we consider some of the problems in the experiments that are described in the manuscript:

(1) In an infected individual, cells are almost always infected by a single virion and the infecting virion is not accompanied by large numbers of damaged or defective virions. This is a key consideration: the claim that infection by HIV affects R-loop formation in cells was done with a VSVg vector in experiments in which there appears to have been about 6000 virions per cell. Although most of the virions prepared in vitro are defective in some way, that does not mean that a large fraction of the defective virions cannot fuse with cells. In normal in vivo infections, HIV has evolved in ways that avoid signaling infected the cell of its presence. To cite an example, carrying out reverse transcription in the capsid/core prevents the host cell from detecting (free) viral DNA in the cytoplasm. The fact that the large effect on R-loop formation which the authors report still occurs in infections done in the absence of reverse transcription strengthens the probability that the effects are due to the massive amounts of virions present, and perhaps to the presence of VSVg, which is quite toxic. To have physiological relevance, the infections would need to be carried out with virions that contain HIV even under circumstances in which there is at most one virion per cell.

(2) Using the Sso7d version of HIV IN in the in vitro binding assays raises some questions, but that is not the real question/problem. The real problem is that the important question is not what/how HIV IN protein binds to, but where/how an intasome binds. An intasome is formed from a combination of IN bound to the ends of viral DNA. In the absence of viral DNA ends, IN does not have the same structure/organization as it has in an intasome. Moreover, HIV IN (even Sso7d, which was modified to improve its behavior) is notoriously sticky and hard to work with. If viral DNA had been included in the experiment, intasomes would need to be prepared and purified for a proper binding experiment. To make matters worse, there are multiple forms of multimeric HIV IN and it is not clear how many HIV INs are present in the PICs that actually carry out integration in an infected cell.

(3) As an extension of comment 2, the proper association of an HIV intasome/PIC with the host genome requires LEDGF and the appropriate nucleic acid targets need to be chromatinized.

(4) Expressing any form of IN, by itself, in cells to look for what IN associates with is not a valid experiment. A major factor that helps to determine both where integration takes place and the sites chosen for integration is the transport of the viral DNA and IN into the nucleus in the capsid core. However, even if we ignore that important part of the problem, the IN that the authors expressed in HeLa cells won't be bound to the viral DNA ends (see comment 2), even if the fusion protein would be able to form an intasome. As such, the IN that is expressed free in cells will not form a proper intasome/PIC and cannot be expected to bind where/how an intasome/PIC would bind.

(5) As in comment 1, for the PLA experiments presented in Figure 5 to work, the number of virions used per cell (which differs from the MOI measured by the number of cells that express a viral marker) must have a high, which is likely to have affected the cells and the results of the experiment. However, there is the additional question of whether the IN-GFP fusion is functional. The fact that the functional intasome is a complex multimer suggests that this could be a problem. There is an additional problem, even if IN-GFP is fully functional. During a normal infection, the capsid core will have delivered copies of IN (and, in the experiments reported here, the IN-GFP fusion) into the nucleus that is not part of the intasome. These "free" copies of IN (here IN-GFP) are not likely to go to the same sites as an intasome, making this experiment problematic (comment 4).

(6) In the Introduction, the authors state that the site of integration affects the probability that the resulting provirus will be expressed. Although this idea is widely believed in the field, the actual data supporting it are, at best, weak. See, for example, the data from the Bushman lab showing that the distribution of integration sites is the same in cells in which the integrated proviruses are, and are not, expressed. However, given what the authors claim in the introduction, they should be more careful in interpreting enzyme expression levels (luciferase) as a measure of integration efficiency in experiments in which they claim proviruses are integrated in different places.

(7) Using restriction enzymes to create an integration site library introduces biases that derive from the uneven distribution of the recognition sites for the restriction enzymes.

Reviewer #3 (Public Review):

In this manuscript, Park and colleagues describe a series of experiments that investigate the role of R-loops in HIV-1 genome integration. The authors show that during HIV-1 infection, R-loops levels on the host genome accumulate. Using a synthetic R-loop prone gene construct, they show that HIV-1 integration sites target sites with high R-loop levels. They further show that integration sites on the endogenous host genome are correlated with sites prone to R-loops. Using biochemical approaches, as well as in vivo co-IP and proximity ligation experiments, the authors show that HIV-1 integrase physically interacts with R-loop structures.

My primary concern with the paper is with the interpretations the authors make about their genome-wide analyses. I think that including some additional analyses of the genome-wide data, as well as some textual changes can help make these interpretations more congruent with what the data demonstrate. Here are a few specific comments and questions:

(1) I think Figure 1 makes a good case for the conclusion that R-loops are more easily detected HIV-1 infected cells by multiple approaches (all using the S9.6 antibody). The authors show that their signals are RNase H sensitive, which is a critical control. For the DRIPc-Seq, I think including an analysis of biological replicates would greatly strengthen the manuscript. The authors state in the methods that the DRIPc pulldown experiments were done in biological replicates for each condition. Are the increases in DRIPc peaks similar across biological replicates? Are genomic locations of HIV-1-dependent peaks similar across biological replicates? Measuring and reporting the biological variation between replicate experiments is crucial for making conclusions about increases in R-loop peak frequency. This is partially alleviated by the locus-specific data in Figure S3A. However, a better understanding of how the genome-wide data varies across biological replicates will greatly enhance the quality of Figure 1.

(2) I think that the conclusion that R-loops "accumulate" in infected cells is acceptable, given the data presented. However, in line 134 the authors state that "HIV-1 infection induced host genomic R-loop formation". I suggest being very specific about the observation. Accumulation can happen by (a) inducing a higher frequency of the occurrence of individual R-loops and/or (b) stabilizing existing R-loops. I'm not convinced the authors present enough evidence to claim one over the other. It is altogether possible that HIV-1 infection stabilizes R-loops such that they are more persistent (perhaps by interactions with integrase?), and therefore more easily detected. I think rephrasing the conclusions to include this possibility would alleviate my concerns.

(3) A technical problem with using the S9.6 antibody for the detection of R-loops via microscopy is that it cross-reacts with double-stranded RNA. This has been addressed by the work of Chedin and colleagues (as well as others). It is absolutely essential to treat these samples with an RNA:RNA hybrid-specific RNase, which the authors did not include, as far as their methods section states. Therefore, it is difficult to interpret all of the immunofluorescence experiments that depend on S9.6 binding.

(4) Given that there is no clear correlation between expression levels and R-loop peak detection, combined with the data that show increased detection of R-loop frequency in non-genic regions, I think it will be important to show that the R-loop forming regions are indeed transcribed above background levels. This will help alleviate possible concerns that there are technical errors in R-loop peak detection.

(5) In Figures 4C and D the hashed lines are not defined. It is also interesting that the integration sites do not line up with R-loop peaks. This does not necessarily directly refute the conclusions (especially given the scale of the genomic region displayed), but should be addressed in the manuscript. Additionally, it would greatly improve Figure 4 to have some idea about the biological variation across replicates of the data presented 4A.

(6) The authors do not adequately describe the Integrase mutant that they use in their biochemical experiments in Figure 5A. Could this impact the activity of the protein in such a way that interferes with the interpretation of the experiment? The mutant is not used in subsequent experiments for Figure 5 and so even though the data are consistent with each other (and the conclusion that Integrase interacts with R-loops) a more thorough explanation of why that mutant was used and how it impacts the biochemical activity of the protein will help the interpretation of the data presented in Figure 5.

eLife assessment

This valuable work investigates the role of boundary elements in the formation of 3D genome architecture. The authors established a specific model system that allowed them to manipulate boundary elements and examine the resulting genome topology. The work yielded the first demonstration of the existence of stem and circle loops in a genome and confirms a model which had been posited based on extensive prior genetic work, providing insights into how 3D genome topologies affect enhancer-promoter communication. The evidence is solid, although the degree of generalization remains uncertain.

Reviewer #1 (Public Review):

In this study, the authors engineer the endogenous left boundary of the Drosophila eve TAD, replacing the endogenous Nhomie boundary by either a neutral DNA, a wildtype Nhomie boundary, an inverted Nhomie boundary, or a second copy of the Homie boundary. They perform Micro-C on young embryos and conclude that endogenous Nhomie and Homie boundaries flanking eve pair with head-to-tail directionality to form a chromosomal stem loop. Abrogating the Nhomie boundary leads to ectopic activation of genes in the former neighboring TAD by eve embryonic stripe enhancers. Replacing Nhomie by an inverted version or by Homie (which pairs with itself head-to-head) transformed the stem loop into a circle loop. An important finding was that stem and circle loops differentially impact endogenous gene regulation both within the eve TAD and in the TADs bracketing eve. Intriguingly, an eve TAD with a circle loop configuration leads to ectopic activation of flanking genes by eve enhancers - indicating compromised regulatory boundary activity despite the presence of an eve TAD with intact left and right boundaries.

The results obtained are of high-quality and are meticulously discussed. This work advances our fundamental understanding of how 3D genome topologies affect enhancer-promoter communication.

This study raises interesting questions to be addressed in future studies.

First, given the unique specificity with which Nhomie and Homie pair (and exhibit "homing" activity), the generalizability of TAD formation by directional boundary pairing remains unclear. Testing whether boundary pairing is a phenomenon restricted to exceptional loci picked for study, rather than a broader rule of TAD formation, would best be done through the development of untargeted approaches to study boundary pairing.

Second, boundary pairing is one of several mechanisms that may form chromosomal contact domains such as TADs. Other mechanisms include cohesin-mediated chromosomal loop extrusion and the inherent tendency of transcriptionally active and inactive chromatin to segregate (or compartmentalize). The functional interplay between these possible TAD-forming mechanisms remains to be further investigated.

Reviewer #2 (Public Review):

This study reports a set of experiments and subsequent analyses focusing on the role of Drosophila boundary elements in shaping 3D genome structure and regulating gene expression. The authors primarily focus on the region of the fly genome containing the even skipped (eve) gene; eve is expressed in a canonical spatial pattern in fly embryos and its locus is flanked by the well-characterized neighbor of homie (nhomie) and homie boundary elements. The main focus of the investigation is the orientation dependence of these boundary elements, which had been observed previously using reporter assays. In this study, the authors use Crispr/Cas9 editing followed by recombination-mediated cassette exchange to create a series of recombinant fly lines in which the nhomie boundary element is either replaced with exongenous sequence from phage 𝝀, an inversion of nhomie, or a copy of homie that has the same orientation as the endogenous homie sequence. The nhomie sequence is also regenerated in its native orientation to control for effects introduced by the transgenesis process.

The authors then perform high-resolution Micro-C to analyze 3D structure and couple this with fluorescent and colorimetric RNA in situ hybridization experiments to measure the expression of eve and nearby genes during different stages of fly development. The major findings of these experiments are that total loss of boundary sequence (replacement with 𝝀 DNA) results in major 3D structure changes and the most prominent observed gene changes, while inversion of the nhomie boundary or replacement with homie resulted in more modest effects in terms of 3D structure and gene expression changes and a distinct pattern of gene expression change from the 𝝀 DNA replacement. As the samples in which the nhomie boundary is inverted or replaced with homie have similar Micro-C profiles at the eve locus and show similar patterns of a spurious gene activation relative to the control, the observed effects appear to be driven by the relative orientation of the nhomie and homie boundary elements to one another.

Collectively, the findings reported in the manuscript are of broad interest to the 3D genome field. Although extensive work has gone into characterizing the patterns of 3D genome organization in a whole host of species, the underlying mechanisms that structure genomes and their functional consequences are still poorly understood. The perhaps best understood system, mechanistically, is the coordinated action of CTCF with the cohesin complex, which in vertebrates appears to shape 3D contact maps through a loop extrusion-pausing mechanism that relies on orientation-dependent sequence elements found at the boundaries of interacting chromatin loops. Despite having a CTCF paralog and cohesin, the Drosophila genome does not appear to be structured by loop extrusion-pausing. The identification of orientation-dependent elements with pronounced structural effects on genome folding thus may shed light on alternative mechanisms used to regulated genome structure, which in turn may yield insights into the significance of particular folding patterns.

On the whole, this study is comprehensive and represents a useful contribution to the 3D genome field. The transgenic lines and Micro-C datasets generated in the course of the work will be valuable resources for the research community. Moreover, the manuscript, while dense in places, is generally clearly written and comprehensive in its description of the work. However, I have a number of comments and critiques of the manuscript, mainly centering on the framing of the experiments and presentation of the Micro-C results and on the manner in which the data are analyzed and reported.

As this document now reflects my review of a revised version of the initial preprint, I will begin to add the new content at this point. As discussed in detail in the following paragraphs, my initial impression of the manuscript has not changed, so I have accordingly left the above text unaltered.

In my initial review, I provided a number of suggestions to improve the quality of the manuscript. These suggestions, which took the form of six major and three minor points, largely focused on 1) altering the writing in certain places to make the story more broadly accessible to the readership and 2) the inclusion of key, missing methodological detail to increase the rigor and reproducibility of the study. No new experiments were requested, and all of the points could be readily addressed with rather straightforward textual changes.

In their revised manuscript, the authors elected to directly address one of the major points and two of the minor points (major point 4, minor points 1 and 3). The remainder of my suggestions remain entirely unaddressed. A similar level of responsiveness was afforded to the very reasonable critiques of the other Reviewer and the Reviewing Editor. The authors have instead largely chosen to respond to the points raised exclusively in the rebuttal document. This document sprawls across >22 pages, includes numerous in-line figures, and cites dozens of references. The tone of this document, in many places, is at best forceful. In a less generous interpretation, many sections are combative, dismissive, and borderline unprofessional.

It is my opinion that the authors are doing the scientific community a disservice with their response. While it is my understanding that readers will be able see the rebuttal letter, I find that end result far from satisfying. How many readers will take the trouble to access that file, versus the manuscript itself? Skirting the review critiques places an unfair burden on readers, who are expecting peer-reviewed science, to dig into the accessory files to follow the critique and response, rather than seeing in reflected in the final product as they accustomed. Intentionally or not, the tactics the authors have chosen detract from what is otherwise a novel and well-intentioned new publishing model. It is also worth pointing out that peer review is done as an act of service to the scientific community, as the senior authors are doubtless aware. The other reviewer, the Reviewing Editor, and I have all taken time away from advancing our own careers and those of our trainees to offer the thoughtful critiques that were so pointedly dismissed.

In summary, as the vast majority of my critiques remain unaddressed, I have simply reproduced them below.

Major Points:

(1) The authors motivate much of the introduction and results with hypothetical "stem loop" and "circle loop" models of chromosome confirmation, which they argue are reflected in the Micro-C data and help to explain the observed ISH patterns. While such structures may possibly form, the support for these specific models vs. the many alternatives is not in any way justified. For instance, no consideration is given to important biophysical properties such as persistence length, packing/scaling, and conformational entropy. As the biophysical properties of chromatin are a very trafficked topic both in terms of experimentation and computational modeling and generally considered in the analysis of chromosome conformation data, the study would be strengthened by acknowledgement of this body of work and more direct integration of its findings.

(2) Similar to Point 1, while there is a fair amount of discussion of how the observed results are or are not consistent with loop extrusion, there is no discussion of the biophysical forces that are thought to underly compartmentalization such as block-polymer co-segregation and their potential influence. I found this absence surprising, as it is generally accepted that A/B compartmentalization essentially can explain the contact maps observed in Drosophila and other non-vertebrate eukaryotes (Rowley, ..., Corces 2017; PMID 28826674). The manuscript would be strengthened by consideration of this phenomenon.

(3) The contact maps presented in the study represent many cells and distinct cell types. It is clear from single-cell Hi-C and multiplexed FISH experiments that chromosome conformation is highly variable even within populations of the same cell, let alone between cell types, with structures such as TADs being entirely absent at the single cell level and only appearing upon pseudobulking. It is difficult to square these observations with the models of relatively static structures depicted here. The authors should provide commentary on this point.

(4) Related to Point 4, the lack of quantitative details about the Micro-C data make it difficult to evaluate if the changes observed are due to biological or technical factors. It is essential that the authors provide quantitative means of controlling for factors like sampling depth, normalization, and data quality between the samples.

(5) The ISH effects reported are modest, especially in the case of the HCR. The details provided for how the imaging data were acquired and analyzed are minimal, which makes evaluating them challenging. It would strengthen the study to provide much more detail about the acquisition and analysis and to include depiction of intermediates in the analysis process, e.g. the showing segmentation of stripes.

Author response:

The following is the authors’ response to the original reviews.

Public Reviews:

Reviewer #1 (Public Review):

Summary:

In this study, the authors engineer the endogenous left boundary of the Drosophila eve TAD, replacing the endogenous Nhomie boundary by either a neutral DNA, a wildtype Nhomie boundary, an inverted Nhomie boundary, or a second copy of the Homie boundary. They perform Micro-C on young embryos and conclude that endogenous Nhomie and Homie boundaries flanking eve pair with head-to-tail directionality to form a chromosomal stem loop. Abrogating the Nhomie boundary leads to ectopic activation of genes in the former neighboring TAD by eve embryonic stripe enhancers. Replacing Nhomie by an inverted version or by Homie (which pairs with itself head-to-head) transformed the stem loop into a circle loop. An important finding was that stem and circle loops differentially impact endogenous gene regulation both within the eve TAD and in the TADs bracketing eve. Intriguingly, an eve TAD with a circle loop configuration leads to ectopic activation of flanking genes by eve enhancers - indicating compromised regulatory boundary activity despite the presence of an eve TAD with intact left and right boundaries.

Strengths:

Overall, the results obtained are of high-quality and are meticulously discussed. This work advances our fundamental understanding of how 3D genome topologies affect enhancer-promoter communication.

Weaknesses:

Though convincingly demonstrated at eve, the generalizability of TAD formation by directional boundary pairing remains unclear, though the authors propose this mechanism could underly the formation of all TADs in Drosophila and possibly even in mammals. Strong and ample evidence has been obtained to date that cohesin-mediated chromosomal loop extrusion explains the formation of a large fraction of TADs in mammals. 

(1.1) The difficultly with most all of the studies on mammal TADs, cohesin and CTCF roadblocks is that the sequencing depth is not sufficient, and large bin sizes (>1 kb) are needed to visualize chromosome architecture.  The resulting contact profiles show TAD neighborhoods, not actual TADs.

The problem with these studies is illustrated by comparing the contact profiles of mammalian MicroC data sets at different bin sizes in Author response image 1.  In this figure, the darkness of the “pixels” in panels E, F, G and H was enhanced by reducing brightness in photoshop.

Author response image 1.

Mammalian MicroC profiles different bun sizes

Panels A and C show “TADs” using bin sizes typical of most mammalian studies (see Krietenstein et al. (2023) (Krietenstein et al. 2020)).  At this level of resolution, TADs, the “trees” that are the building blocks of chromosomes, are not visible.  Instead, what is seen are TAD neighborhoods or “forests”.  Each neighborhood consists of several dozen individual TADs.  The large bins in these panels also artificially accentuated TAD:TAD interactions, generating a series of “stripes” and “dots” that correspond to TADs bumping into each other and sequences getting crosslinked.  For example, in panel A there is prominent stripe on the edge of a “TAD” (blue arrow).  In panel C, this stripe resolves into a series of dots arranged as parallel, but interrupted “stripes” (green and blue arrows).  At the next level of resolution, it can be seen that the stripe marked by the blue arrow and magenta asterisk is generated by contacts between the left boundary of the TAD indicated by the magenta bar with sequences in a TAD (blue bar) ~180 kb way.  While dots and stripes are prominent features in contact profiles visualized with larger bin sizes (A and C), the actual TADs that are observed with a bin size of 200 bp (examples are underlined by black bars in panel G) are not bordered by stripes, nor are they topped by obvious dots.  The one possible exception is the dot that appears at the top of the volcano triangle underlined with magenta.

The chromosome 1 DNA segment from the MicroC data of Hseih et al. (2023) (Hsieh et al. 2020) shows a putative volcano triangle with a plume (indicated by a V in Author response image 1 panels D, F and H).  Sequences in the V TAD don’t crosslink with their immediate neighbors, and this gives a “plume” above the volcano triangle, as indicate by the light blue asterisk in panels D, F and H.  Interestingly the V TAD does contact two distant TADs, U on the left and W on the right. The U TAD is ~550 kb from V, and the region of contact is indicated by the black arrow.  The W TAD is ~585 kb from V, and the region of contact is indicated by the magenta arrow.  While the plume still seems to be visible with a bin size of 400 bp (light blue asterisk), it is hard to discern when the bin size is 200 bp, as there are not enough reads.

The evidence demonstrating that cohesin is required for TAD formation/maintenance is based on low resolution Hi-C data, and the effects that are observed are on TAD neighborhoods (forests) and not TADs (trees).  In fact, there is published evidence that cohesin is not required in mammals for TAD formation/maintenance.  In an experiment from Goel et al. 2023 the authors depleted the cohesin component Rad21 and then visualized the effects on TAD organization using the high resolution region capture MicroC (RCMC) protocol.  The MicroC contact map in this figure visualizes a ~250 kb DNA segment around the Ppm1pg locus at 250 bp resolution.  On the right side of the diagonal is the untreated control, while the left side shows the MicroC profile of the same region after Rad21 depletion.  The authors indicated that there was a 97% depletion of Rad21 in their experiment.  However, as is evident from a comparison of the experimental and control, loss of Rad21 has no apparent effect on the TAD organization of this mammalian DNA segment.

Several other features are worth noting.  First, unlike the MicroC experiments shown in Author response image 1, there are dots at the apex of the TADs in this chromosomal segment.  In the MicroC protocol, fixed chromatin is digested to mononucleosomes by extensive MNase digestion.  The resulting DNA fragments are then ligated, and dinucleosome-length fragments are isolated and sequenced. 

DNA sequences that are nucleosome free in chromatin (which would be promoters, enhancers, silencers and boundary elements) are typically digested to oligonucleotides in this procedure and won’t be recovered. This means that the dots shown here must correspond to mononucleosome-length elements that are MNase resistant.  This is also true for the dots in the MicroC contact profiles of the Drosophila Abd-B regulatory domain (see Fig. 2B in the paper).  Second, the TADs are connected to each other by 45o stripes (see blue and green arrowheads).  While it is not clear from this experiment whether the stipes are generated by an active mechanism (enzyme) or by some “passive” mechanism (e.g., sliding), the stripes in this chromosomal segment are not generated by cohesin, as they are unperturbed by Rad21 depletion.  Third, there are no volcano triangles with plumes in this chromosomal DNA segment.  Instead, the contact patterns (purple and green asterisks) between neighboring TADs closely resemble those seen for the Abd-B regulatory domains (compare Goel et al. 2023 with Fig. 2B in the paper).  This similarity suggests that the TADs in and around Ppm1g may be circle-loops, not stem-loops.  As volcano triangles with plumes also seem to be rare in the MicroC data sets of Krietenstein et al. (Krietenstein et al. 2020) and Hesih et al. (Hsieh et al. 2020) (with the caveat that these data sets are low resolution: see Author response image 1), it is possible that much of the mammalian genome is assembled into circle-loop TADs, a topology that can’t be generated by the cohesin loop extrusion (bolo tie clip) /CTCF roadblock model.

While Rad21 depletion has no apparent effect on TADs, it does appear to impact TAD neighborhoods.  This is in a supplemental figure in Goel et al. (Goel et al. 2023).  In this figure, TADs in the Ppm1g region of chromosome 5 are visualized with bin sizes of 5 kb and 1 kb.  A 1.2 Mb DNA segment is shown for the 5 kb bin size, while an 800 kb DNA segment is shown for the 1 kb bin size.  As can be seen from comparing the MicroC profiles in Author response image 2 with that in Goel et al. 2023, individual TADs are not visible.  Instead, the individual TADs are binned into large TAD “neighborhoods” that consist of several dozen or more TADs.

Unlike the individual TADs shown in Goel et al. 2023, the TAD neighborhoods in Author response image 2 are sensitive to Rad21 depletion.  The effects of Rad21 depletion can be seen by comparing the relative pixel density inside the blue lines before (above the diagonal) and after (below the diagonal) auxin-induced Rad21 degradation.  The reduction in pixel density is greatest for more distant TAD:TAD contacts (farthest from the diagonal).  By contrast, the TADs themselves are unaffected (Goel et al. 2023), as are contacts between individual TADs and their immediate neighbors.  In addition, contacts between partially overlapping TAD neighborhoods are also lost.  At this point it isn’t clear why contacts between distant TADs in the same neighborhood are lost when Rad21 is depleted; however, a plausible speculation is that it is related to the functioning of cohesin in holding newly replicated DNAs together until mitosis and whatever other role it might have in chromosome condensation.

Author response image 2.

Ppm1g full locus chr5

Moreover, given the unique specificity with which Nhomie and Homie are known to pair (and exhibit "homing" activity), it is conceivable that formation of the eve TAD by boundary pairing represents a phenomenon observed at exceptional loci rather than a universal rule of TAD formation. Indeed, characteristic Micro-C features of the eve TAD are only observed at a restricted number of loci in the fly genome…..

(1.2) The available evidence does not support the claim that nhomie and homie are “exceptional.”  To begin with, nhomie and homie rely on precisely the same set of factors that have been implicated in the functioning of other boundaries in the fly genome.  For example, homie requires (among other factors) the generic boundary protein Su(Hw) for insulation and long-distance interactions (Fujioka et al. 2024).  (This is also true of nhomie: unpublished data.)  The Su(Hw) protein (like other fly polydactyl zinc finger proteins) can engage in distant interactions.  This was first shown by Sigrist and Pirrotta (Sigrist and Pirrotta 1997), who found that the su(Hw) element from the gypsy transposon can mediate long-distance regulatory interactions (PRE dependent silencing) between transgenes inserted at different sites on homologous chromosomes (trans interactions) and at sites on different chromosomes.

The ability to mediate long-distance interactions is not unique to the su(Hw) element, or homie and nhomie.  Muller et al. (Muller et al. 1999) found that the Mcp boundary from the Drosophila BX-C is also able to engage in long-distance regulatory interactions—both PRE-dependent silencing of mini-white and enhancer activation of mini-white and yellow.  The functioning of the Mcp boundary depends upon two other generic insulator proteins, Pita and the fly CTCF homolog (Kyrchanova et al. 2017).  Like Su(Hw) both are polydactyl zinc finger proteins, and they resemble the mammalian CTCF protein in that their N-terminal domain mediates multimerization (Bonchuk et al. 2020; Zolotarev et al. 2016).  Figure 6 from Muller et el. 1999 shows PRE-dependent “pairing sensitive silencing” interactions between transgenes carrying a mini-white reporter, the Mcp and scs’ (Beaf dependent)(Hart et al. 1997) boundary elements, and a PRE closely linked to Mcp.  In this experiment flies homozygous for different transgene inserts were mated and the eye color was examined in their transheterozygous progeny.  As indicated in the figure, the strongest trans-silencing interactions were observed for inserts on the same chromosomal arm; however, transgenes inserted on the left arm of chromosome 3 can interact across the centromere with transgenes inserted on the right arm of chromosome 3. 

Figure 5C (left) from Muller et el. 1999 shows a trans-silencing interaction between w#11.102 at 84D and w#11.16 approximately 5.8 Mb away, at 87D.  Figure 5C (right) shows a trans-silencing interaction across the centromere between w#14.29 on the left arm of chromosome 3 at 78F and w#11.102 on the right arm of chromosome 3 at 84D. The eye color phenotype of mini-white-containing transgenes is usually additive: homozygyous inserts have twice as dark eye color as the corresponding hemizygous inserts.  Likewise, in flies trans-_heterozygous for _mini-white transgenes inserted at different sites, the eye color is equivalent to the sum of the two transgenes.  This is not true when mini-white transgenes are silenced by PREs.  In the combination shown in panel A, the t_rans-_heterozygous fly has a lighter eye color than either of the parents.  In the combination in panel B, the _trans-_heterozygous fly is slightly lighter than either parent.

As evident from the diagram in Figure 6 from Muller et el. 1999, all of the transgenes inserted on the 3rd chromosome that were tested were able to participate in long distance (>Mbs) regulatory interactions.  On the other hand, not all possible pairwise interactions are observed.  This would suggest that potential interactions depend upon the large scale (Mb) 3D folding of the 3rd chromosome.

When the scs boundary (Zw5 dependent) (Gaszner et al. 1999) was added to the transgene to give sMws’, it further enhanced the ability of distant transgenes to find each other and pair.  All eight of the sMws’ inserts that were tested were able to interact with at least one other sMws’ insert on a different chromosome and silence mini-white.  Vazquez et al. () subsequently tagged the sMws’ transgene with LacO sequences (ps0Mws’) and visualized pairing interactions in imaginal discs.  Trans-heterozygous combinations on the same chromosome were found paired in 94-99% of the disc nuclei, while a trans-heterozygous combination on different chromosomes was found paired in 96% of the nuclei (Table 3 from Vazquez et al. 2006).  Vazquez et al. also examined a combination of four transgenes inserted on the same chromosome (two at the same insertion site, and two at different insertion sites).  In this case, all four transgenes were clustered together in 94% of the nuclei (Table 3 from Vazquez et al. 2006).  Their studies also suggest that the distant transgenes remain paired for at least several hours.  A similar experiment was done by Li et al. (Li et al. 2011), except that the transgene contained only a single boundary, Mcp or Fab-7.  While pairing was still observed in trans-heterozygotes, the frequency was reduced without scs and scs’.

It is worth pointing out that there is no plausible mechanism in which cohesin could extrude a loop through hundreds of intervening TADs, across the centromere (ff#13.101_ßà_w#11.102: Figure 6 from Muller et el. 1999; w#14.29_ßà_w#11.02: Figure 6 from Muller et el. 1999 and 5) and come to a halt when it “encounters” Mcp containing transgenes on different homologs.  The same is true for Mcp-dependent pairing interactions in cis (Fig. 7 in Muller et al. (Muller et al. 1999)) or Mcp-dependent pairing interactions between transgenes inserted on different chromosomes (Fig. 8 in Muller et al. (Muller et al. 1999); Line 8 in Table 3 from Vazquez et al. 2006). 

These are not the only boundaries that can engage in long-distance pairing.  Mohana et al. (Mohana et al. 2023) identified nearly 60 meta-loops, many of which appear to be formed by the pairing of TAD boundary elements.  Two examples (at 200 bp resolution from 12-16 hr embryos) are shown in Author response image 3.

Author response image 3.

Metaloops on the 2nd and 3rd chromosomes: circle-loops and multiple stem-loops

One of these meta-loops (panel A) is generated by the pairing of two TAD boundaries on the 2nd chromosome.  The first boundary, blue, (indicated by blue arrow) is located at ~2,006, 500 bp between a small TAD containing the Nplp4 and CG15353 genes and a larger TAD containing 3 genes, CG33543, Obp22a and Npc2a.  Nplp4 encodes a neuropeptide.  The functions of CG15354 and CG33543 are unknown.  Obp22a encodes an odorant binding protein, while Npc2a encodes the Niemann-Pick type C-2a protein which is involved sterol homeostasis.  The other boundary (purple: indicated by purple arrow) is located between two TADs 2.8 Mb away at 4,794,250 bp.  The upstream TAD contains the fipi gene (CG15630) which has neuronal functions in male courtship, while the downstream TAD contains CG3294, which is thought to be a spliceosome component, and schlaff (slf) which encodes a chitin binding protein.  As illustrated in the accompanying diagram, the blue boundary pairs with the purple boundary in a head-to-head orientation, generating a ~2.8 Mb loop with a circle-loop topology.  As a result of this pairing, the multi-gene (CG33543, Obp22a and Npc2a) TAD upstream of the blue boundary interacts with the CG15630 TAD upstream of the purple boundary.  Conversely the small Nplp4:CG15353 TAD downstream of the blue boundary interacts with the CG3294:slf TAD downstream of the purple boundary.  Even if one imagined that the cohesin bolo tie clip was somehow able to extrude 2.8 Mb of chromatin and then know to stop when it encountered the blue and purple boundaries, it would’ve generated a stemloop, not a circle-loop.

The second meta-loop (panel B) is more complicated as it is generated by pairing interactions between four boundary elements.  The blue boundary (blue arrow) located ~4,801,800 bp (3L) separates a large TAD containing the RhoGEF64C gene from a small TAD containing CG7509, which encodes a predicted subunit of an extracellular carboxypeptidase.  As can be seen in the MicroC contact profile and the accompanying diagram, the blue boundary pairs with the purple boundary (purple arrow) which is located at ~7,013, 500 (3L) just upstream of the 2nd internal promoter (indicated by black arrowhead) of the Mp (Multiplexin) gene.  This pairing interaction is head-to-tail and generates a large stem-loop that spans ~2.2 Mb.  The stem-loop brings sequences upstream of the blue boundary and downstream of the purple boundary into contact (the strings below a bolo tie clip), just as was observed in the boundary bypass experiments of Muravyova et al. (Muravyova et al. 2001) and Kyrchanova et al. (Kyrchanova et al. 2008).  The physical interactions result in a box of contacts (right top) between sequences in the large RhoGEF64C TAD and sequences in a large TAD that contains an internal Mp promoter.  The second pairing interaction is between the brown boundary (brown arrow) and the green boundary (green arrow).  The brown boundary is located at ~4 805,600 bp (3L) and separates the TAD containing CG7590 from a large TAD containing CG1808 (predicted to encode an oxidoreductase) and the Dhc64C (Dynein heavy chain 64C) gene.  The green boundary is located at ~6,995,500 bp (3L), and it separates a TAD containing CG32388 and the biniou (bin) transcription factor from a TAD that contains the most distal promoter of the Mp (Multiplexin) gene (blue arrowhead).  As indicated in the diagram, the brown and green boundaries pair with each other head-to-tail, and this generates a small internal loop (and the final configuration would resemble a bolo tie with two tie clips).  This small internal loop brings the CG7590 TAD into contact with the TAD that extends from the distal Mp promoter to the 2nd internal Mp promoter.  The resulting contact profile is a rectangular box with diagonal endpoints corresponding to the paired blue:purple and brown:green boundaries.  The pairing of the brown:green boundaries also brings the TADs immediately downstream of the brown boundary and upstream of the green boundary into contact with each other, and this gives a rectangular box of interactions between the Dhc64C TAD, and sequences in the bin/CG3238 TAD.  This box is located on the lower left side of the contact map.

Since the bin and Mp meta-loops in Author response image 3B are stem-loops, they could have been generated by “sequential” cohesin loop extrusion events.  Besides the fact that cohesin extrusion of 2 Mb of chromatin and breaking through multiple intervening TAD boundaries challenges the imagination, there is no mechanism in the cohesion loop extrusion/CTCF roadblock model to explain why cohesion complex 1 would come to a halt at the purple boundary on one side and the blue boundary on the other, while cohesin complex 2 would instead stop when it hits the brown and green boundaries.  This highlights another problem with the cohesin loop extrusion/CTCF roadblock model, namely that the roadblocks are functionally autonomous: they have an intrinsic ability to block cohesin that is entirely independent of the intrinsic ability of other roadblocks in the neighborhood.  As a result, there is no mechanism for generating specificity in loop formation.  By contrast, boundary pairing interactions are by definition non-autonomous and depend on the ability of individual boundaries to pair with other boundaries: specificity is built into the model. The mechanism for pairing, and accordingly the basis for partner preferences/specificity, are reasonably well understood.  Probably the most common mechanism in flies is based on shared binding sites for architectural proteins that can form dimers or multimers (Bonchuk et al. 2021; Fedotova et al. 2017).  Flies have a large family of polydactyl zinc finger DNA binding proteins, and as noted above, many of these form dimers or multimers and also function as TAD boundary proteins.  This pairing principle was first discovered by Kyrchanova et al. (Kyrchanova et al. 2008).  This paper also showed that orientation-dependent pairing interactions is a common feature of endogenous fly boundaries.  Another mechanism for pairing is specific protein:protein interactions between different DNA binding factors (Blanton et al. 2003).  Yet a third mechanism would be proteins that bridge different DNA binding proteins together.  The boundaries that use these different mechanisms (BX-C boundaries, scs, scs’) depend upon the same sorts of proteins that are used by homie and nhomie.  Likewise, these same set of factors reappear in one combination or another in most other TAD boundaries.  As for the orientation of pairing interactions, this is most likely determined by the order of binding sites for chromosome architectural proteins in the partner boundaries.

…and many TADs lack focal 3D interactions between their boundaries.

(1.3) The idea that flies differ from mammals in that they “lack” focal 3D interactions is simply mistaken.  One of the problems with drawing this distinction is that most all of the “focal 3D interactions” seen mammalian Hi-C experiments are a consequence of binning large DNA segments in low resolution restriction enzyme-dependent experiments.  This is even true in the two “high” resolution MicroC experiments that have been published (Hsieh et al. 2020; Krietenstein et al. 2020).  As illustrated above in Author response image 1, most of the “focal 3D interactions” (the dots at the apex of TAD triangles) seen with large bin sizes (1 kb and greater) disappear when the bin size is 200 bp and TADs rather than TAD neighborhoods are being visualized.

As described in point #1.1, in the MicroC protocol, fixed chromatin is first digested to mononucloesomes by extensive MNase digestion, processed/biotinylated, and ligated to give dinucleosome-length fragments, which are then sequenced.  Regions of chromatin that are nucleosome free (promoters, enhancers, silencers, boundary elements) will typically be reduced to oligonucleotides in this procedure and will not be recovered when dinucleosome-length fragments are sequenced.  The loss of sequences from typical paired boundary elements is illustrated by the lar meta-loop shown in Author response image 4 (at 200 bp resolution).  Panels A and B show the contact profiles generated when the blue boundary (which separates two TADs that span  the Lar (Leukocyteantigen-related-like) transcription unit interacts with the purple boundary (which separates two TADs in a gene poor region ~620 kb away).  The blue and purple boundaries pair with each other head-to-head, and this pairing orientation generates yet another circle-loop.  In the circle-loop topology, sequences in the TADs upstream of both boundaries come into contact with each other, and this gives the small dark rectangular box to the upper left of the paired boundaries (Author response image 4A).  (Note that this small box corresponds to the two small TADs upstream of the blue and purple boundaries, respectively. See panel B.)  Sequences in the TADs downstream of the two boundaries also come into contact with each other, and this gives the large box to the lower right of the paired boundaries.  While this meta-loop is clearly generated by pairing interactions between the blue and purple boundaries, the interacting sequences are degraded in the MicroC protocol, and sequences corresponding to the blue and purple boundaries aren’t recovered.  This can be seen in panel B (red arrow and red arrowheads).  When a different Hi-C procedure is used (dHS-C) that captures nucleosome-free regions of chromatin that are physically linked to each other (Author response image 4C & D), the sequences in the interacting blue and purple boundaries are recovered and generate a prominent “dot” at their physical intersection (blue arrow in panel D).

Author response image 4.

Lar metaloop. Panels A & bB: MicroC. Panels C & D: dHS-C

While sequences corresponding to the blue and purple boundaries are lost in the MicroC procedure, there is at least one class of elements that engage in physical pairing interactions whose sequences are (comparatively) resistant to MNase digestion.  This class of elements includes many PREs ((Kyrchanova et al. 2018); unpublished data), the boundary bypass elements in the Abd-B region of BX-C (Kyrchanova et al. 2023; Kyrchanova et al. 2019a; Kyrchanova et al. 2019b; Postika et al. 2018), and “tethering” elements (Batut et al. 2022; Li et al. 2023).  In all of the cases tested, these elements are bound in nuclear extracts by a large (>1000 kD) GAGA factor-containing multiprotein complex called LBC.  LBC also binds to the hsp70 and eve promoters (unpublished data).  Indirect end-labeling experiments (Galloni et al. 1993; Samal et al. 1981; Udvardy and Schedl 1984) indicate that the LBC protects a ~120-180 bp DNA segment from MNase digestion.  It is likely that this is the reason why LBC-bound sequences can be recovered in MicroC experiments as dots when they are physically linked to each other.  One such example (based on the ChIP signatures of the paired elements) is indicated by the green arrow in panel B and D of Author response image 4.  Note that there are no dots corresponding to these two LBC elements within either of the TADs immediately downstream of the blue and purple boundaries.  Instead the sequences corresponding to the two LBC elements are only recovered when the two elements pair with each other over a distance of ~620 kb.  The fact that these two elements pair with each other is consistent with other findings which indicate that, like classical boundaries, LBC elements exhibit partner preferences.  In fact, LBC elements can sometimes function as TAD boundaries.  For example, the Fab-7 boundary has two LBC elements, and full Fab-7 boundary function can be reconstituted with just these two elements (Kyrchanova et al. 2018).

Reviewer #2 (Public Review):

"Chromatin Structure II: Stem-loops and circle-loops" by Ke*, Fujioka*, Schedl, and Jaynes reports a set of experiments and subsequent analyses focusing on the role of Drosophila boundary elements in shaping 3D genome structure and regulating gene expression. The authors primarily focus on the region of the fly genome containing the even skipped (eve) gene; eve is expressed in a canonical spatial pattern in fly embryos and its locus is flanked by the well-characterized neighbor of homie (nhomie) and homie boundary elements. The main focus of investigation is the orientation dependence of these boundary elements, which had been observed previously using reporter assays. In this study, the authors use Crispr/Cas9 editing followed by recombination-mediated cassette exchange to create a series of recombinant fly lines in which the nhomie boundary element is either replaced with exongenous sequence from phage 𝝀, an inversion of nhomie, or a copy of homie that has the same orientation as the endogenous homie sequence. The nhomie sequence is also regenerated in its native orientation to control for effects introduced by the transgenesis process.

The authors then perform high-resolution Micro-C to analyze 3D structure and couple this with fluorescent and colorimetric RNA in situ hybridization experiments to measure the expression of eve and nearby genes during different stages of fly development. The major findings of these experiments are that total loss of boundary sequence (replacement with 𝝀 DNA) results in major 3D structure changes and the most prominent observed gene changes, while inversion of the nhomie boundary or replacement with homie resulted in more modest effects in terms of 3D structure and gene expression changes and a distinct pattern of gene expression change from the 𝝀 DNA replacement. As the samples in which the nhomie boundary is inverted or replaced with homie have similar Micro-C profiles at the eve locus and show similar patterns of a spurious gene activation relative to the control, the observed effects appear to be driven by the relative orientation of the nhomie and homie boundary elements to one another.

Collectively, the findings reported in the manuscript are of broad interest to the 3D genome field. Although extensive work has gone into characterizing the patterns of 3D genome organization in a whole host of species, the underlying mechanisms that structure genomes and their functional consequences are still poorly understood. The perhaps best understood system, mechanistically, is the coordinated action of CTCF with the cohesin complex, which in vertebrates appears to shape 3D contact maps through a loop extrusion-pausing mechanism that relies on orientation-dependent sequence elements found at the boundaries of interacting chromatin loops.

(2.1) The notion that mammalian genome is shaped in 3D by the coordinate action of cohesin and CTCF has achieved the status of dogma in the field of chromosome structure in vertebrates.  However, as we have pointed out in #1.1, the evidence supporting this dogma is far from convincing.  To begin with, it is based on low resolution Hi-C experiments that rely on large bin sizes to visualize so-called “TADs.”  In fact, the notion that cohesin/CTCF are responsible on their own for shaping the mammalian 3D genome appears to be a result of mistaking a series of forests for the actual trees that populate each of the forests.

As illustrated in Author response image 1 above, the “TADs” that are visualized in these low resolution data sets are not TADs at all, but rather TAD neighborhoods consisting of several dozen or more individual TADs.  Moreover, the “interesting” features that are evident at low resolution (>1 kb)—the dots and stripes—largely disappear at resolutions appropriate for visualizing individual TADs (~200 bp).

In Goel et al. 2023, we presented data from one of the key experiments in Goel et al. (Goel et al. 2023).  In this experiment,  the authors used RCMC to generate high resolution (~250 bp) MicroC contact maps before and after Rad21 depletion.  Contrary to dogma, Rad21 depletion has absolutely no effect on TADs in a ~250 kb DNA segment—and these TADs look very much like the TADs we observe in the Drosophila genome, in particular in the Abd-B region of BX-C that is thought to be assembled into a series of circle-loops (see Fig. 2B).

While Goel et al. (Goel et al. 2023) observed no effect of Rad21 depletion on TADs, they found that loss of Rad21 disturbs long-distance (but not short-distance) contacts in large TAD neighborhoods when their RCMC data set is visualized using bin sizes of 5 kb and I kb.  This is shown in Author response image 2.  The significance of this finding is, however, uncertain.  It could mean that the 3D organization of large TAD neighborhoods have a special requirement for cohesin activity.  On the other hand, since cohesin functions to hold sister chromosomes together after replication until they separate during mitosis (and might also participate in mitotic condensation), it is also possible that the loss of long-range contacts in large TAD neighborhoods when Rad21 is depleted is simply a reflection of this particular activity.  Further studies will be required to address these possibilities.

As for CTCF: a careful inspection of the ChIP data in Goel et al. 2023 indicates that CTCF is not found at each and every TAD boundary.  In fact, the notion that CTCF is the be-all and end-all of TAD boundaries in mammals is truly hard to fathom.  For one, the demands for specificity in TAD formation (and in regulatory interactions) are likely much greater than those in flies, and specificity can’t be generated by a single DNA binding protein.  For another, several dozen chromosomal architectural proteins have already been identified in flies.  This means that (unlike what is thought to be true in mammals) it is possible to use a combinatorial mechanism to generate specificity in, for example, the long distance interactions in RFig 6 and 7.  As noted in #2.1 above, many of the known chromosomal architectural proteins in flies are polydactyl zinc finger proteins (just like CTCF).  There are some 200 different polydactyl zinc finger proteins in flies, and the function of only a hand full of these is known at present.  However, it seems likely that a reasonable fraction of this class of DNA binding proteins will ultimately turn out to have an architectural function of some type (Bonchuk et al. 2021; Fedotova et al. 2017).  The number of different polydactyl zinc finger protein genes in mammals is nearly 3 times that of flies.  It is really possible that of these, only CTCF is involved in shaping the 3D structure of the mammalian genome?

Despite having a CTCF paralog and cohesin, the Drosophila genome does not appear to be structure by loop extrusion-pausing. The identification of orientation-dependent elements with pronounced structural effects on genome folding thus may shed light on alternative mechanisms used to regulated genome structure, which in turn may yield insights into the significance of particular folding patterns.

(2.2) Here we would like to draw the reviewer’s and reader’s attention to Author response image 3, which shows that orientation-dependent pairing interactions have a significant impact on physical interactions between different sequences.  We would also refer the reader to two other publications.  One of these is Kyrchanova et al. (Kyrchanova et al. 2008), which was the first to demonstrate that orientation of pairing interactions matters.  The second is Fujioka et al. (Fujioka et al. 2016), which describes experiments indicating that nhomie and homie pair with each other head-to-tail and with themselves head-to-head.

On the whole, this study is comprehensive and represents a useful contribution to the 3D genome field. The transgenic lines and Micro-C datasets generated in the course of the work will be valuable resources for the research community. Moreover, the manuscript, while dense in places, is generally clearly written and comprehensive in its description of the work. However, I have a number of comments and critiques of the manuscript, mainly centering on the framing of the experiments and presentation of the Micro-C results and on manner in which the data are analyzed and reported. They are as follows:

Major Points:

(1) The authors motivate much of the introduction and results with hypothetical "stem loop" and "circle loop" models of chromosome confirmation, which they argue are reflected in the Micro-C data and help to explain the observed ISH patterns. While such structures may possibly form, the support for these specific models vs. the many alternatives is not in any way justified. For instance, no consideration is given to important biophysical properties such as persistence length, packing/scaling, and conformational entropy. As the biophysical properties of chromatin are a very trafficked topic both in terms of experimentation and computational modeling and generally considered in the analysis of chromosome conformation data, the study would be strengthened by acknowledgement of this body of work and more direct integration of its findings.

(2.3) The reviewer is not correct in claiming that “stem-loops” and “circle-loops” are “hypothetical.”  There is ample evidence that both types of loops are present in eukaryotic genomes, and that loop conformation has significant readouts in terms of not only the physical properties of TADs but also their functional properties.  Here we would draw the reviewer’s attention to Author response image 3 and Author response image 4 for examples of loops formed by the orientation-dependent pairing of yet other TAD boundary elements.  As evident from the MicroC data in these figures, circle-loops and stem-loops have readily distinguishable contact patterns.  The experiments in Fujioka et al. (Fujioka et al. 2016) demonstrate that homie and nhomie pair with each other head-to-tail, while they pair with themselves head-to-head.  The accompany paper (Bing et al. 2024) also provides evidence that loop topology is reflected both in the pattern of activation of reporters and in the MicroC contact profiles.  We would also mention again Kyrchanova et al. (Kyrchanova et al. 2008), who were the first to report orientation-dependent pairing of endogenous fly boundaries.

At this juncture it would premature to try to incorporate computational modeling of chromosome conformation in our studies.  The reason is that the experimental foundations that would be essential for building accurate models are lacking.  As should be evident from RFigs. 1-3 above, studies on mammalian chromosomes are simply not of high enough resolution to draw firm conclusions about chromosome conformation: in most studies only the forests are visible.  While the situation is better in flies, there are still too many unknown.  As just one example, it would be important to know the orientation of the boundary pairing interactions that generate each TAD.  While it is possible to infer loop topology from how TADs interact with their neighbors (a plume versus clouds), a conclusive identification of stem- and circle-loops will require a method to unambiguously determine whether a TAD boundary pairs with its neighbor head-to-head or headto-tail.

(2) Similar to Point 1, while there is a fair amount of discussion of how the observed results are or are not consistent with loop extrusion, there is no discussion of the biophysical forces that are thought to underly compartmentalization such as block-polymer co-segregation and their potential influence. I found this absence surprising, as it is generally accepted that A/B compartmentalization essentially can explain the contact maps observed in Drosophila and other non-vertebrate eukaryotes (Rowley, ..., Corces 2017; PMID 28826674). The manuscript would be strengthened by consideration of this phenomenon.

(2.4) Compartments in mammals have typically been identified and characterized using lowresolution data sets, and these studies have relied on visualizing compartments using quite large bin sizes (>>1 kb).  Our experiments have nothing to do with the large-scale compartments seen in these Hi-C experiments.  Instead, we are studying the properties of individual TADs: how TADs are formed, the relationship between TAD topology and boundary:boundary pairing, and the impact of TAD topology on interactions between TADs in the immediate neighborhood.  There is no evidence to date that these large compartments or “block polymer co-segregation” have a) any impact on the properties of individual boundary elements, b) have a role in determining which boundary elements actually come together to form a given TAD, c) impact the orientation of the interactions between boundaries that generate the TAD or d) determine how TADs tend to interact with their immediate neighbors.  

In more recent publications (c.f., Harris et al. 2023) compartments have shrunk in size and instead of being units of several hundred kb, the median length of the “compartmental” unit in mammalian cells is about12 kb. This is not too much different from the size of fly TADs.  However, the available evidence does not support the idea that block polymer co-segregation/co-repulsion drive the TAD:TAD interactions seen in MicroC experiments.  For example, according to this “micro-compartment” model, the specific patterns of interaction between TADs in the CG3294 meta-loop in Author response image 3 would be driven by block polymer co-segregation and co-repulsion. In this model, the TAD upstream of the blue boundary (which contains CG33543, the odorant binding protein gene Obp22a and the Npc2a gene which encodes a protein involved in sterol homeostasis) would share the same chromatin state/biophysical properties as the TAD upstream of the purple boundary, which has the fipi gene. While it is true that CG33543, Obp22a and also the fipi gene are not expressed in embryos, Npc2a is expressed at high levels during embryogenesis, yet it is part of the TAD that interacts with the fipi TAD.  The TAD downstream of the blue boundary contains CG15353 and Nplp4 and it interacts with the TAD downstream of the purple boundary which contains CG3294 and slf.  CG15353 and Nplp4 are not expressed in the embryo and as such should share a compartment with a TAD that is also silent. However, slf is expressed at a high level in 1216 hr embryos, while CG3294 is expressed at a low level.  In neither case would one conclude that the TADs upstream and downstream of the blue and purple boundaries, respectively, interact because of shared chromatin/biophysical states that drive block polymer co-segregation corepulsion. 

One might also consider several gedanken experiments involving the long-range interactions that generate the CG3294 meta-loop in Author response image 3.    According to the micro-compartment model the patchwork pattern of crosslinking evident in the CG3294 meta-loop arises because the interacting  TADs share the same biochemical/biophysical properties, and this drives block polymer cosegregation and co-repulsion.  If this model is correct, then this patchwork pattern of TAD:TAD interactions would remain unchanged if we were to delete the blue or the purple boundary.  However, given what we know about how boundaries can find and pair with distant boundaries (c.f., Figure 6 from Muller et el. 1999 and the discussion in #1.2), the result of these gedanken experiments seem clear: the patchwork pattern shown in Author response image 3A will disappear.  What would happen if we inverted the blue or the purple boundary? Would the TAD containing CG33543, Obp22a and Npc2a still interact with fipi as would be expected from the compartment model?  Or would the pattern of interactions flip so that the CG33543, Obp22a and Npc2a TAD interacts with the TAD containing CG3294 and slf?  Again we can anticipate the results based on previous studies: the interacting TADs will switch when the CG3294 meta-loop is converted into a stem-loop.  If this happened, the only explanation possible in the compartment model is that the chromatin states change when the boundary is inverted so that TAD upstream of blue boundary now shares the same chromatin state as the TAD downstream of the purple boundary, while the TAD downstream of the blue boundary shares same state as the TAD upstream of the purple boundary.  However, there is no evidence that boundary orientation per se can induce a complete switch in “chromatin states” as would be required in the compartment model. 

While we have not done these experimental manipulations with the CG3294 meta-loop, an equivalent experiment was done in Bing et al. (Bing et al. 2024).  However, instead of deleting a boundary element, we inserted a homie boundary element together with two reporters (gfp and LacZ) 142 kb away from the eve TAD.  The result of this gedanken “reverse boundary deletion” experiment is shown in Author response image 5.  Panel A shows the MicroC contact profile in the region spanning the transgene insertion site and the eve TAD in wild type (read “deletion”) NC14 embryos.  Panel B shows the MicroC contact profile from 12-16 hr embryos carrying the homie dual reporter transgene inserted at -142 kb.  Prior to the “deletion”, the homie element in the transgene pairs with nhomie and homie in the eve TAD and this generates a “mini-metaloop.”  In this particular insert, the homie boundary in the transgene (red arrow) is “pointing” in the opposite orientation from the homie boundary in the eve TAD (red arrow).  In this orientation, the pairing of the transgene homie with eve nhomie/homie brings the LacZ reporter into contact with sequences in the eve TAD.  Since a mini-metaloop is formed by homie_à _nhomie/homie pairing, sequences in TADs upstream and downstream of the transgene insert interact with sequences in TADs close to the eve TAD (Author response image 5B).  Taken together these interactions correspond to the interaction patchwork that is typically seen in “compartments” (see boxed region and inset).  If this patchwork is driven as per the model, by block polymer co-segregation and co-repulsion, then it should still be present when the transgene is deleted.  However, panel A shows that the interactions linking the transgene and the sequences in TADs next to the transgene to eve and TADs next to eve disappear when the homie boundary (plus transgene) is “deleted” in wild type flies.

Author response image 5.

Boundary deletion and compartments

A second experiment would be to invert the homie boundary so that instead of pointing away from eve it points towards eve.  Again, if the compartmental patchwork is driven by block polymer co-segregation and co-repulsion, inverting the homie boundary in the transgene should have no effect on the compartmental contact profile.  Inspection of Fig. 7 in Bing et al. (Bing et al. 2024) will show that this prediction doesn’t hold either.  When homie is inverted, sequences in the eve TAD interact with the gfp reporter not the LacZ reporter.  In addition, there are corresponding changes in how sequences in TADs to either side of eve interact with sequences to either side of the transgene insert.  

Yet another “test” of compartments generated by block polymer co-segregation/co-repulsion is provided by the plume above the eve volcano triangle.  According to the compartment model, sequences in TADs flanking the eve locus form the plume above the eve volcano triangle because their chromatin shares properties that drive block polymer co-segregation.  These same properties result in repulsive interactions with chromatin in the eve TAD, and this would explain why the eve TAD doesn’t crosslink with its neighbors.  If the distinctive chromatin properties of eve and the neighboring TADs drive block polymer co-segregation and co-repulsion, then inverting the nhomie boundary or introducing homie in the forward orientation should have absolutely no effect on the physical interactions between chromatin in the eve TAD and chromatin in the neighboring TADs.  However, Figures 4 and 6 in this paper indicate that boundary pairing orientation, not block polymer co-segregation/co-repulsion, is responsible for forming the plume above the eve TAD. Other findings also appear to be inconsistent with the compartment model. (A) The plume topping the eve volcano triangle is present in NC14 embryos when eve is broadly expressed (and potentially active throughout the embryo).  It is also present in 12-16 hr embryos when eve is only expressed in a very small subset of cells and is subject to PcG silencing everywhere else in the embryo.  B) According to the compartment model the precise patchwork pattern of physical interactions should depend upon the transcriptional program/chromatin state that is characteristic of a particular developmental stage or cell type.  As cell fate decisions are just being made during NC14 one might expect that most nuclei will share similar chromatin states throughout much of the genome.  This would not be true for 12-16 hr embryos.  At this stage the compartmental patchwork would be generated by a complex mixture of interactions in cells that have quite different transcriptional programs and chromatin states.  In this case, the patchwork pattern would be expected to become fuzzy as a given chromosomal segment would be in compartment A in one group of cells and in compartment B in another.   Unlike 12-16 hr embryos,  larval wing discs would be much more homogeneous and likely give a distinct and relatively well resolved compartmental pattern. We’ve examined the compartment patchwork of the same chromosomal segments in NC14 embryos, 12-16 hr embryos and larval wing disc cells.  While there are some differences (e.g., changes in some of the BX-C TADs in the wing disc sample) the compartmental patchwork patterns are surprisingly similar in all three cases. Nor is there any “fuzziness” in the compartmental patterns evident in 12-16 hr embryos, despite the fact that there are many different cell types at this stage of development.  C) TAD interactions with their neighbors and compartmental patchworks are substantially suppressed in salivary gland polytene chromosomes.  This would suggest that features of chromosome structure might be the driving force behind many of the “compartmental” interactions as opposed to distinct biochemical/biophysical of properties of small chromosomal segments that drive polymer co- segregation/co-repulsion.  

(3) The contact maps presented in the study represent many cells and distinct cell types. It is clear from single-cell Hi-C and multiplexed FISH experiments that chromosome conformation is highly variable even within populations of the same cell, let alone between cell types, with structures such as TADs being entirely absent at the single cell level and only appearing upon pseudobulking. It is difficult to square these observations with the models of relatively static structures depicted here. The authors should provide commentary on this point.

(2.5) As should be evident from Author response image 1, single-cell Hi-C experiments would not provide useful information about the physical organization of individual TADs, TAD boundaries or how individual TADs interact with their immediate neighbors.  In addition, since they capture only a very small fraction of the possible contacts within and between TADs, we suspect that these single-cell studies aren’t likely to be useful for making solid conclusions about TAD neighborhoods like those shown in Author response image 1 panels A, B, C and D, or Author response image 2.  While it might be possible to discern relatively stable contacts between pairs of insulators in single cells with the right experimental protocol, the stabilities/dynamics of these interactions may be better judged by the length of time that physical interactions are seen to persist in live imaging studies such as Chen et al. (2018), Vazquez et al. (2006) and Li et al. (2011).

The in situ FISH data we’ve seen also seems problematic in that probe hybridization results in a significant decondensation of chromatin.  For two probe sets complementary to adjacent ~1.2 kb DNA sequences, the measured center-to-center distance that we’ve seen was ~110 nM.  This is about 1/3rd the length that is expected for a 1.2 kb naked DNA fragment, and about 1.7 times larger than that expected for a beads-on-a-string nucleosome array (~60 nM).  However, chromatin is thought to be compacted into a 30 nM fiber, which is estimated to reduce the length of DNA by at least another ~6 fold.  If this estimate is correct, FISH hybridization would appear to result in a ~10 fold decompaction of chromatin.  A decompaction of this magnitude would necessarily be followed by a significant distortion in the actual conformation of chromatin loops.

(4) The analysis of the Micro-C data appears to be largely qualitative. Key information about the number of reads sequenced, reaps mapped, and data quality are not presented. No quantitative framework for identifying features such as the "plumes" is described. The study and its findings would be strengthened by a more rigorous analysis of these rich datasets, including the use of systematic thresholds for calling patterns of organization in the data.

Additional information on the number of reads and data quality have been included in the methods section. 

(5) Related to Point 4, the lack of quantitative details about the Micro-C data make it difficult to evaluate if the changes observed are due to biological or technical factors. It is essential that the authors provide quantitative means of controlling for factors like sampling depth, normalization, and data quality between the samples.

In our view the changes in the MicroC contact patterns for the eve locus and its neighbors when the nhomie boundary is manipulated are not only clear cut and unambiguous but are also readily evident in the Figs that are presented in the manuscript.  If the reviewer believes that there aren’t significant differences between the MicroC contact patterns for the four different nhomie replacements, it seems certain that they would also remain unconvinced by a quantitative analysis.

The reviewer also suggests that biological and/or technical differences between the four samples could account for the observed changes in the MicroC patterns for the eve TAD and its neighbors.  If this were the case, then similar changes in MicroC patterns should be observed elsewhere in the genome.  Since much of the genome is analyzed in these MicroC experiments there is an abundance of internal controls for each experimental manipulation of the nhomie boundary.  For two of the nhomie replacements, nhomie reverse and homie forward, the plume above the eve volcano triangle is replaced by clouds surrounding the eve volcano triangle.  If these changes in the eve MicroC contact patterns are due to significant technical (or biological) factors, we should observe precisely the same sorts of changes in TADs elsewhere in the genome that are volcano triangles with plumes.   Author response image 6 shows the MicroC contact pattern for several genes in the Antennapedia complex.  The deformed gene is included in a TAD which, like eve, is a volcano triangle topped by a plume.  A comparison of the deformed MicroC contact patterns for nhomie forward (panel B) with the MicroC patterns for nhomie reverse (panel C) and homie forward (panel D) indicates that while there are clearly technical differences between the samples, these differences do not result in the conversion of the deformed plume into clouds as is observed for the eve TAD.  The MicroC patterns elsewhere in Antennapedia complex are also very similar in all four samples.  Likewise, comparisons of regions elsewhere in the fly genome indicate that the basic contact patterns are similar in all four samples.   So while there are technical differences which are reflected in the relative pixel density in the TAD triangles and the LDC domains, these differences do not result in converting plumes into clouds nor do the alter the basic patterns of TAD triangles and LDC domains.  As for biological differences— the embryos in each sample are at roughly the same developmental stage and were collected and processed using the same procedures. Thus, the biological factors that could reasonably be expected to impact the organization of specific TADs (e.g., cell type specific differences) are not going to impact the patterns we see in our experiments. 

Author response image 6.

(6) The ISH effects reported are modest, especially in the case of the HCR. The details provided for how the imaging data were acquired and analyzed are minimal, which makes evaluating them

challenging. It would strengthen the study to provide much more detail about the acquisition and analysis and to include depiction of intermediates in the analysis process, e.g. the showing segmentation of stripes.

The imaging analysis is presented in Fig. 5 is just standard confocal microscopy.  Individual embryos were visualized and scored.  An embryo in which stripes could be readily detected was scored as ‘positive’ while an embryo in which stripes couldn’t be detected was scored as ‘negative.’   

Recommendations for the authors:

Editor comments:

It was noted that the Jaynes lab previously published extensive genetic evidence to support the stem loop and circle loop models of Homie-Nhomie interactions (Fujioka 2016 Plos Genetics) that were more convincing than the Micro-C data presented here in proof of their prior model. Maybe the authors could more clearly summarize their prior genetic results to further try to convince the reader about the validity of their model.

Reviewer #1 (Recommendations For The Authors):

Below, I list specific comments to further improve the manuscript for publication. Most importantly, I recommend the authors tone down their proposal that boundary pairing is a universal TAD forming mechanism.

(1) The title is cryptic.

(2) The second sentence in the abstract is an overstatement: "In flies, TADs are formed by physical interactions between neighboring boundaries". Hi-C and Micro-C studies have not provided evidence that most TADs in Drosophila show focal interactions between their bracketing boundaries. The authors rely too strongly on prior studies that used artificial reporter transgenes to show that multimerized insulator protein binding sites or some endogenous fly boundaries can mediate boundary bypass, as evidence that endogenous boundaries pair.

Please see responses #1.1 and #1.3 and figures Author response image 1 and Author response image 3.  Note that using dHS-C, most TADs that we’ve looked at so far are topped by a “dot” at their apex.

(3) Line 64: the references do not cite the stated "studies dating back to the '90's'".

The papers cited for that sentence are reviews which discussed the earlier findings.  The relevant publications are cited at the appropriate places in the same paragraph.  

(4) Line 93: "On the other hand, while boundaries have partner preferences, they are also promiscuous in their ability to establish functional interactions with other boundaries." It was unclear what is meant here.

Boundaries that a) share binding sites for proteins that multimerized, b) have binding sites for proteins that interact with each other, or c) have binding sites for proteins that can be bridged by a third protein can potentially pair with each other.  However, while these mechanisms enable promiscuous pairing interactions, they will also generate partner preferences (through a greater number of a, b and/or c).

(5) It could be interesting to discuss the fact that it remains unclear whether Nhomie and Homie pair in cis or in trans, given that homologous chromosomes are paired in Drosophila.

The studies in Fujioka et al. (Fujioka et al. 2016) show that nhomie and homie can pair both in cis and in trans.  Given the results described in #1.2, we imagine that they are paired in both cis and trans in our experiments.

(6) Line 321: Could the authors further explain why they think that "the nhomie reverse circle-loop also differs from the nhomie deletion (λ DNA) in that there is not such an obvious preference for which eve enhancers activate expression"?

The likely explanation is that the topology/folding of the altered TADs impacts the probability of interactions between the various eve enhancers and the promoters of the flanking genes.  

(7) The manuscript would benefit from shortening the long Discussion by avoiding repeating points described previously in the Results.

(8) Line 495: "If, as seems likely, a significant fraction of the TADs genome-wide are circle loops, this would effectively exclude cohesin-based loop extrusion as a general mechanism for TAD formation in flies". The evidence provided in this manuscript appears insufficient to discard ample evidence from multiple laboratories that TADs form by compartmentalization or loop extrusion. Multiple laboratories have, for example, demonstrated that cohesin depletion disrupts a large fraction of mammalian TADs. 

Points made here and in #9 have been responded to in #1.1, #2.1 and #2.4 above.  We would suggest that the evidence for loop extrusion falls short of compelling (as it is based on the analysis of TAD neighborhoods, not TADs—that is forests, not trees) and given the results reported in Goel et al. (in particular Fig. 4 and Sup Fig. 8) is clearly suspect. This is not to mention the fact that cohesin loop-extrusion can’t generate circle-loops TADs, yet circle-loops clearly exist.  Likewise, as discussed in #2.4, it is not clear to us that the shared chromatin states, polymer co-segregation and co-repulsion account for the compartmental patchwork patterns of TAD;TAD interactions. The results from the  experimental manipulations in this paper and the accompanying paper, together with studies by others (e.g., Kyrchanova et al. (Kyrchanova et al. 2008), Mohana et al. (Mohana et al. 2023) would also seem to be at odds with the model for compartments as currently formulated.  

The unique properties of Nhomie and Homie, namely the remarkable specificity with which they physically pair over large distances (Fujioka et al. 2016) may rather suggest that boundary pairing is a phenomenon restricted to special loci. Moreover, it has not yet been demonstrated that Nhomie or Homie are also able to pair with the TAD boundaries on their left or right, respectively.

Points made here were discussed in detail in #1.2.  As described in detail in #1.2, It is not the case that nhomie and homie are in “unique” or “special.”  Other fly boundaries can do the same things.  As for whether nhomie and homie pair with their neighbors:  We haven’t done transgene experiments (e.g., testing by transvection or boundary bypass).  Likewise, in MicroC experiments there are no obvious dots at the apex of the neighboring TADs that would correspond to nhomie pairing with the neighboring boundary to the left and homie pairing with the neighboring boundary to the right. However, this is to be expected. As we discussed in in #1.3 above, only MNase resistant elements will generate dots in standard MicroC experiments.  On the other hand, when boundary:boundary interactions are analyzed by dHS-C (c.f., Author response image 4), there are dots at the apex of both neighboring TADs.  This would be direct evidence that nhomie pairs with the neighboring boundary to the left and homie pairs with the neighboring boundary to the right.

(9) The comment in point 8 also applies to the concluding 2 sentences (lines 519-524) of the Discussion.

See response to 8 above. Otherwise, the concluding sentences are completely accurate. Validation of the cohesin loop extrusion/CTCF roadblock model will required demonstrating a) that all TADs are either stem-loops or unanchored loops and b) that TAD endpoints are always marked by CTCF. 

The likely presence of circle-loops and evidence that TAD boundaries that don’t have CTCF (c.f.,Goel et al. 2023) already suggests that this model can’t (either fully or not all) account for TAD formation in mammals. 

(10) Figs. 3 and 6: It would be helpful to add the WT screenshot in the same figure, for direct comparison.

It is easy enough to scroll between Figs-especially since nhomie forward looks just like WT.

(11) Fig. 6: It would be helpful to show a cartoon view of a circle loop to the right of the Micro-C screenshot, as was done in Fig. 3.

Good idea.   Added to the Fig.

(12) Fig. 5: It would be helpful to standardize the labelling of the different genotypes throughout the figures and panels ("inverted" versus "reverse" versus an arrow indicating the direction).

Fixed.

Reviewer #2 (Recommendations For The Authors):

Minor Points:

(1) The Micro-C data does not appear to be deposited in an appropriate repository. It would be beneficial to the community to make these data available in this way.

This has been done.

(2) Readers not familiar with Drosophila development would benefit from a gentle introduction to the stages analyzed and some brief discussion on how the phenomenon of somatic homolog pairing might influence the study, if at all.

We included a rough description the stages that were analyzed for both the in situs and MicroC. We thought that an actual description of what is going on at each of the stages wasn’t necessary as the process of development is not a focus of this manuscript.  In other studies, we’ve found that there are only minor differences in MicroC patterns between the blastoderm stage and stage 12-16 embryos.  While these minor differences are clearly interesting, we didn’t discuss them in the text.   In all of experiments chromosomes are likely to be paired.  In NC14 embryos (the stage for visualizing eve stripes and the MicroC contact profiles in Fig. 2) replication of euchromatic sequences is thought to be quite rapid.  While homolog pairing is incomplete at this stage, sister chromosomes are paired.  In stage 12-16 embryos, homologs will be paired and if the cells are arrested in G2, then sister chromosome will also be paired.  So in all of experiments, chromosomes (sisters and/or homologs) are paired. However, since we don’t have examples of unpaired chromosomes, our experiments don’t provide any info on how chromosome pairing might impact MicroC/expression patterns.

(3) "P > 0.01" appears several times. I believe the authors mean to report "P < 0.01".

Fixed.  

References for Response

Batut PJ, Bing XY, Sisco Z, Raimundo J, Levo M, Levine MS. 2022. Genome organization controls transcriptional dynamics during development. Science. 375(6580):566-570.

Bing X, Ke W, Fujioka M, Kurbidaeva A, Levitt S, Levine M, Schedl P, Jaynes JB. 2024. Chromosome structure i: Loop extrusion or boundary:Boundary pairing? eLife.

Blanton J, Gaszner M, Schedl P. 2003. Protein:Protein interactions and the pairing of boundary elements in vivo. Genes Dev. 17(5):664-675.

Bonchuk A, Boyko K, Fedotova A, Nikolaeva A, Lushchekina S, Khrustaleva A, Popov V, Georgiev P. 2021. Structural basis of diversity and homodimerization specificity of zincfinger-associated domains in drosophila. Nucleic Acids Res. 49(4):2375-2389.

Bonchuk A, Kamalyan S, Mariasina S, Boyko K, Popov V, Maksimenko O, Georgiev P. 2020. Nterminal domain of the architectural protein ctcf has similar structural organization and ability to self-association in bilaterian organisms. Sci Rep. 10(1):2677.

Chen H, Levo M, Barinov L, Fujioka M, Jaynes JB, Gregor T. 2018. Dynamic interplay between enhancer–promoter topology and gene activity. Nat Genet. 50(9):1296.

Fedotova AA, Bonchuk AN, Mogila VA, Georgiev PG. 2017. C2h2 zinc finger proteins: The largest but poorly explored family of higher eukaryotic transcription factors. Acta Naturae. 9(2):47-58.

Fujioka M, Ke W, Schedl P, Jaynes JB. 2024. The homie insulator has sub-elements with different insulating and long-range pairing properties. bioRxiv. 2024.02.01.578481.

Fujioka M, Mistry H, Schedl P, Jaynes JB. 2016. Determinants of chromosome architecture: Insulator pairing in cis and in trans. PLoS Genet. 12(2):e1005889.

Galloni M, Gyurkovics H, Schedl P, Karch F. 1993. The bluetail transposon: Evidence for independent cis‐regulatory domains and domain boundaries in the bithorax complex. The EMBO Journal. 12(3):1087-1097.

Gaszner M, Vazquez J, Schedl P. 1999. The zw5 protein, a component of the scs chromatin domain boundary, is able to block enhancer-promoter interaction. Genes Dev. 13(16):2098-2107.

Goel VY, Huseyin MK, Hansen AS. 2023. Region capture micro-c reveals coalescence of enhancers and promoters into nested microcompartments. Nat Genet. 55(6):1048-1056.

Harris HL, Gu H, Olshansky M, Wang A, Farabella I, Eliaz Y, Kalluchi A, Krishna A, Jacobs M, Cauer G et al. 2023. Chromatin alternates between a and b compartments at kilobase scale for subgenic organization. Nat Commun. 14(1):3303.

Hart CM, Zhao K, Laemmli UK. 1997. The scs' boundary element: Characterization of boundary element-associated factors. Mol Cell Biol. 17(2):999-1009.

Hsieh TS, Cattoglio C, Slobodyanyuk E, Hansen AS, Rando OJ, Tjian R, Darzacq X. 2020. Resolving the 3d landscape of transcription-linked mammalian chromatin folding. Mol Cell. 78(3):539-553.e538.

Krietenstein N, Abraham S, Venev SV, Abdennur N, Gibcus J, Hsieh TS, Parsi KM, Yang L, Maehr R, Mirny LA et al. 2020. Ultrastructural details of mammalian chromosome architecture. Mol Cell. 78(3):554-565.e557.

Kyrchanova O, Chetverina D, Maksimenko O, Kullyev A, Georgiev P. 2008. Orientation-dependent interaction between drosophila insulators is a property of this class of regulatory elements. Nucleic Acids Res. 36(22):7019-7028.

Kyrchanova O, Ibragimov A, Postika N, Georgiev P, Schedl P. 2023. Boundary bypass activity in the abdominal-b region of the drosophila bithorax complex is position dependent and regulated. Open Biol. 13(8):230035.

Kyrchanova O, Kurbidaeva A, Sabirov M, Postika N, Wolle D, Aoki T, Maksimenko O, Mogila V, Schedl P, Georgiev P. 2018. The bithorax complex iab-7 polycomb response element has a novel role in the functioning of the fab-7 chromatin boundary. PLoS Genet. 14(8):e1007442. Kyrchanova O, Sabirov M, Mogila V, Kurbidaeva A, Postika N, Maksimenko O, Schedl P, Georgiev P. 2019a. Complete reconstitution of bypass and blocking functions in a minimal artificial fab-7 insulator from drosophila bithorax complex. Proceedings of the National Academy of Sciences.201907190.

Kyrchanova O, Wolle D, Sabirov M, Kurbidaeva A, Aoki T, Maksimenko O, Kyrchanova M, Georgiev P, Schedl P. 2019b. Distinct elements confer the blocking and bypass functions of the bithorax fab-8 boundary. Genetics.genetics. 302694.302019.

Kyrchanova O, Zolotarev N, Mogila V, Maksimenko O, Schedl P, Georgiev P. 2017. Architectural protein pita cooperates with dctcf in organization of functional boundaries in bithorax complex. Development. 144(14):2663-2672.

Li H-B, Muller M, Bahechar IA, Kyrchanova O, Ohno K, Georgiev P, Pirrotta V. 2011. Insulators, not polycomb response elements, are required for long-range interactions between polycomb targets in drosophila melanogaster. Mol Cell Biol. 31(4):616-625.

Li X, Tang X, Bing X, Catalano C, Li T, Dolsten G, Wu C, Levine M. 2023. Gaga-associated factor fosters loop formation in the drosophila genome. Mol Cell. 83(9):1519-1526.e1514.

Mohana G, Dorier J, Li X, Mouginot M, Smith RC, Malek H, Leleu M, Rodriguez D, Khadka J, Rosa P et al. 2023. Chromosome-level organization of the regulatory genome in the drosophila nervous system. Cell. 186(18):3826-3844.e3826.

Muller M, Hagstrom K, Gyurkovics H, Pirrotta V, Schedl P. 1999. The mcp element from the drosophila melanogaster bithorax complex mediates long-distance regulatory interactions. Genetics. 153(3):1333-1356.

Muravyova E, Golovnin A, Gracheva E, Parshikov A, Belenkaya T, Pirrotta V, Georgiev P. 2001. Loss of insulator activity by paired su(hw) chromatin insulators. Science. 291(5503):495498.

Postika N, Metzler M, Affolter M, Müller M, Schedl P, Georgiev P, Kyrchanova O. 2018. Boundaries mediate long-distance interactions between enhancers and promoters in the drosophila bithorax complex. PLoS Genet. 14(12):e1007702.

Samal B, Worcel A, Louis C, Schedl P. 1981. Chromatin structure of the histone genes of d. Melanogaster. Cell. 23(2):401-409.

Sigrist CJ, Pirrotta V. 1997. Chromatin insulator elements block the silencing of a target gene by the drosophila polycomb response element (pre) but allow trans interactions between pres on different chromosomes. Genetics. 147(1):209-221.

Udvardy A, Schedl P. 1984. Chromatin organization of the 87a7 heat shock locus of drosophila melanogaster. J Mol Biol. 172(4):385-403.

Vazquez J, Muller M, Pirrotta V, Sedat JW. 2006. The mcp element mediates stable long-range chromosome-chromosome interactions in drosophila. Molecular Biology of the Cell. 17(5):2158-2165.

Zolotarev N, Fedotova A, Kyrchanova O, Bonchuk A, Penin AA, Lando AS, Eliseeva IA, Kulakovskiy IV, Maksimenko O, Georgiev P. 2016. Architectural proteins pita, zw5,and zipic contain homodimerization domain and support specific long-range interactions in drosophila. Nucleic Acids Res. 44(15):7228-7241.

eLife assessment

This study presents important findings on the different polymorphs of alpha-synuclein filaments that form at various pH's during in vitro assembly reactions with purified recombinant protein. Of particular note is the discovery of two new polymorphs (1M and 5A) that form in PBS buffer at pH 7. The strength of the evidence presented is convincing. The work will be of interest to biochemists and biophysicists working on protein aggregation and amyloids.

Reviewer #1 (Public Review):

Summary:

Frey et al. report the structures of aSyn fibrils that were obtained under a variety of conditions. These include generation of aSyn fibrils without seeds, but in different buffers and at different pH values. These also include the generation of aSyn fibrils in the presence of seeding fibrils, again performed in different buffers and at different pH values, while the seeds were generated at different conditions. The authors find that fibril polymorphs primarily correlate with fibril growth buffer conditions, and not such much with the type of seed. However, the presence of a seed is still required, likely because fibrils can also seed along their lateral surfaces, not only at the blunt ends.

Strengths:

The manuscript includes an excellent review of the numerous available structures of aSyn.<br /> The text is interesting to read, figures are clear and not redundant.

Weaknesses:

My earlier comments have all been addressed to my satisfaction.

Reviewer #2 (Public Review):

The authors have engaged constructively with some of the points raised. In particular the addition of more details about the experimental cryo-EM procedures has strengthened the manuscript.

I do worry that the FSC values of model-vs-map appear to be higher than expected from the corresponding FSCs between the half-maps (e.g. see Fig 13). The implication of this observation is that the atomic models may have been overfitted in the maps, which would have led to a deterioration of their geometry. A table with rmsd on bond lengths, angles, etc would probably show this. In addition, to check for overfitting, the atomic model for each data set could be refined in one of the half-maps, and then that same model could be used to calculate 2 FSC model-vs-map curves: one against the half-map it was refined in and one against the other half-map. Deviations between these two curves are an indication of overfitting.

In addition, the sudden drop in the FSC curves in Figure 16 shows that something unexpected has happened to this refinement. Are the authors sure that only the procedures outlined in the Methods were used to create these curves? The unexpected nature of the FSC curve for this type (2A) raises doubts about the correctness of the reconstruction.

Reviewer #3 (Public Review):

Summary

The high heterogeneity nature of α-synuclein (α-syn) fibrils posed significant challenges in structural reconstruction of the ex vivo conformation. A deeper understanding of the factors influencing the formation of various α-syn polymorphs remains elusive. The manuscript by Frey et al. provides a comprehensive exploration of how pH variations (ranging from 5.8 to 7.4) affect the selection of α-syn polymorphs (specifically, Type1, 2 and 3) in vitro by using cryo-electron microscopy (cryo-EM) and helical reconstruction techniques. Crucially, the authors identify two novel polymorphs at pH 7.0 in PBS. These polymorphs bear resemblance to the structure of patient-derived juvenile-onset synucleinopathy (JOS) polymorph and diseased tissue amplified α-syn fibrils. The revised manuscript more strongly supports the notion that seeding is a non-polymorph-specific in the context of secondary nucleation-dominated aggregation, underscoring the irreplaceable role of pH in polymorph formation.

Strengths

This study systematically investigates the effects of environmental conditions and seeding on the structure of α-syn fibrils. It emphasizes the significant influence of environmental factors, especially pH, in determining the selection of α-syn polymorphs. The high-resolution structures obtained through cryo-EM enable a clear characterization of the composition and proportion of each polymorph in the sample. Collectively, this work provides a strong support for the pronounced sensitivity of α-syn fibril structures to the environmental conditions, and systematically categorizes previously reported α-syn fibril structures. Furthermore, the identification of JOS-like polymorph also demonstrates the possibility of in vitro reconstruction of brain-derived α-syn fibril structures.

Weaknesses

There are two minor points I recommend the authors to address:

(1) In the response to Weakness 1, point (3), the authors state that "the Type 5 represented only 10-20% of the fibrils in the sample." However, this information is not labeled in the corresponding Figure 4. I suggest the authors verify and label all relevant percentages in the figures to prevent misunderstandings.

(2) While the authors have detailed the helical reconstruction procedure in the Methods section, it is necessary to indicate the scale bar or box size in the figure legend of the 2D representative classes to ensure clarity and reproducibility.

Comments on the revised manuscript:

The authors have responded adequately to these critiques in the revised version of the manuscript.

Author response:

The following is the authors’ response to the original reviews.

Reviewer #1:

We thank the reviewer for their careful reading of our manuscript and have taken all of their grammatical corrections into account.

Reviewer #2 (Public Review):

Weaknesses: 

The paper contains multiple instances of non-scientific language, as indicated below. It would also benefit from additional details on the cryo-EM structure determination in the Methods and inclusion of commonly accepted requirements for cryo-EM structures, like examples of 2D class averages, raw micrographs, and FSC curves (between half-maps as well as between rigid-body fitted (or refined) atomic models of the different polymorphs and their corresponding maps). In addition, cryo-EM maps for the control experiments F1 and F2 should be presented in Figure 9.

We tried to correct the non-scientific language and have included the suggested data on the Cryo-EM analyses including new Figures 11-17.  We did not collect data on the sample used for the seeds in the cross seeding experiments because we had already confirmed in multiple datasets that the conditions in F1 and F2 reproducibly produce fibrils of Type 1 and Type 3, respectively. We have now analyzed cryo-EM data for 6 more samples at pH 7.0 and found that several kinds of polymorphs (Types 1A, 1M, 2A, 2B and 5) are accessible at this pH, however the Type 3 polymorphs are not formed at pH 7.0 under the conditions that we used for aggregation.

Reviewer #2 (Recommendations For The Authors):

Remove unscientific language: "it seems that there are about as many unique atomicresolution structures of these aggregates as there are publications describing them"   

We have rephrased this sentence.

For same reason, remove "Obviously, " 

Done

What does this mean? “polymorph-unspecific” 

Rephrased as non-polymorph-specific

What does this mean? "shallow amyloid energy hypersurface"  

By “shallow hypersurface” we mean that the minimum of the multi-dimensional function that describes the energy of the amyloid is not so deep that subtle changes to the environment will not favor another fold/energy minimum. We have left the sentence because while it may not be perfect, it is concise and seems to get the point across.

"The results also confirm the possibility of producing disease-relevant structure in vitro." -> This is incorrect as no disease-relevant structure was replicated in this work. Use another word like “suggest”.

We have changed to “suggest” as suggested.

Remove "historically" 

Done

Rephrase “It has long been understood that all amyloids contain a common structural scaffold” 

Changed to “It has long been established that all amyloids contain a common structural scaffold..” 

"Amyloid polymorphs whose differences lie in both their tertiary structure (the arrangement of the beta-strands) and the quaternary structure (protofilamentprotofilament assembly) have been found to display distinct biological activities [8]" -> I don't think this is true, different biological activities of amyloids have never been linked to their distinct structures.  

We have added 5 new references (8-12) to support this sentence.

Reference 10 is a comment on reference 9; it should be removed. Instead, as for alphasynuclein, all papers describing the tau structures should be included.  

We have removed the reference, but feel that the addition of all Tau structure references is not merited in this manuscript since we are not comparing them.

Rephrase: "is not always 100% faithful"

Removed “100%”

What is pseudo-C2 symmetry? Do the authors mean pseudo 2_1 symmetry (ie a 2-start helical symmetry)?

Thank for pointing this out.  We did indeed mean pseudo 21 helical symmetry.  

Re-phrase: "alpha-Syn's chameleon-like behavior" 

We have removed this phrase.

"In the case of alpha-Syn, the secondary nucleation mechanism is based on the interaction of the positively charged N-terminal region of monomeric alpha-Syn and the disordered, negatively charged C-terminal region of the alpha-Syn amyloid fibrils [54]" -> I would say the mechanisms of secondary nucleation are not that well understood yet, so one may want to tune this down a bit. 

We have changed this to “mechanism has been proposed to be”

The paragraphs describing experiments by others are better suited for a Discussion rather than a Results section. Perhaps re-organize this part? 

We have left the text intact as we are using a Results and Discussion format.

A lot of information about Image processing seems to be missing: what steps were performed after initial model generation? 

We have added more details in the methods section on the EM data processing and model analysis.

Figure 1: Where is Type 4 on the pH scale?

We have adjusted the Fig 1 legend to clarify that pH scale is only applicable to the structures presented in this manuscript. 

Figure 2: This might be better incorporated as a subpanel of Figure 1.

We agree that this figure is somewhat of a loner on its own and we only added it in order to avoid confusion with the somewhat inconsistent naming scheme used for the Type 1B structure. However, we prefer to leave it as a separate figure so that it does not get dilute the impact of figure 1.

Figure 3: What is the extra density at the bottom of Type 3B from pH 5.8 samples 1 and 2. pH 5.8 + 50mM NaCl (but not pH 5.8 + 100 mM NaCl)? Could this be an indication of a local minimum and the pH 5.8 + 100 mM NaCl structure is correct? Or is this a real difference between 0/50mM NaCl and 100 mM NaCl? 

We did not see the extra density to which the reviewer is referring, however the images used in this panel are the based on the output of 3D-classification which is more likely to produce more artifacts than a 3D refinement. With this in mind, we did not see any significant differences in the refined structures and therefore only deposited the better quality map and model for each of the polymorph types.

Figure 3: To what extent is Type 3B of pH 6.5 still a mixture of different types? The density looks poor. In general, in the absence of more details about the cryo-EM maps, it is hard to assess the quality of the structures presented.

In order to improve the quality of the images in this panel, a more complete separation of the particles from each polymorph was achieved via the filament subset selection tool in RELION 5. In each case, an unbiased could be created from the 2D classes via the relion_helix_inimodel2D program, further supporting the coexistence of 4 polymorphs in the pH 6.5 sample. The particles were individually refined to produce the respective maps that are now used in this figure.

Many references are incorrect, containing "Preprint at (20xx)" statements.  

This has been corrected.

Reviewer #3 (Public Review):

Weaknesses: 

(1) The authors reveal that both Type 1 monofilament fibril polymorph (reminiscent of JOSlike polymorph) and Type 5 polymorph (akin to tissue-amplified-like polymorph) can both form under the same condition. Additionally, this condition also fosters the formation of flat ribbon-like fibril across different batches. Notably, at pH 5.8, variations in experimental groups yield disparate abundance ratios between polymorph 3B and 3C, indicating a degree of instability in fibrillar formation. The variability would potentially pose challenges for replicability in subsequent research. In light of these situations, I propose the following recommendations: 

(a) An explicit elucidation of the factors contributing to these divergent outcomes under similar experimental conditions is warranted. This should include an exploration of whether variations in purified protein batches are contributing factors to the observed heterogeneity.

We are in complete agreement that understanding the factors that lead to polymorph variability is of utmost importance (and was the impetus for the manuscript itself). However the number of variables to explore is overwhelming and we will continue to investigate this in our future research. Regarding the variability between batches of purified protein, we also think that this could be a factor in the polymorph variability observed for otherwise “identical” aggregation conditions, particularly at pH 7 where the largest variety of polymorphs have been observed. However, even variation between identical replicates (samples created from the same protein solution and simply aggregated simultaneously in separate tubes) can lead to different outcomes (see datasets 15 and 16 in the revised Table 1) suggesting that there are stochastic processes that can determine the outcome of an individual aggregation experiment. While our data still indicates that Type 1,2 and 3 polymorphs are strongly selected by pH, the selection between interface variants 3B vs. 3C and 2A vs. 2B might also be affected by protein purity. Our standard purification protocol produces a single band by coomassie-stained SDS-PAGE however minor truncations and other impurities below a few percent would go undetected and, given the proposed roles of the N and C-termini in secondary nucleation, could have a large effect on polymorph selection and seeding. In line with the reviewer’s comments we now include a batch number for each EM dataset. While no new conclusions can be drawn from the inclusion of this additional data, we feel that it is important to acknowledge the possible role of batch to batch variability. 

(b) To enhance the robustness of the conclusions, additional replicates of the experiments under the same condition should be conducted, ideally a minimum of three times.  

The pH 5.8 conditions that yield Type 3 fibrils has already been repeated several times in the original manuscript. Since the pH 7.4 conditions produce the most common a-Syn polymorph (Type 1A) and were produced twice in this manuscript (once as an unseeded and once as a cross-seeded fibrilization) we decided to focus on the intermediate condition where the most variability had been seen (pH 7.0). The revised table 1 now has 6 new datasets (11-16) representing 6 independent aggregations at pH 7.0 starting from two different protein purification batches. The results is that we now produce the type 2A/B polymorphs in three samples and in two of these samples we once again observed the type 1M polymorph.  The other samples produced Type 1A or non-twisted fibrils.

(c) Further investigation into whether different polymorphs formed under the same buffer condition could lead to distinct toxicological and pathology effects would be a valuable addition to the study.  

The correlation of toxicity with structure would in principle be interesting. However the Type 1 and Type 3 polymorphs formed at pH 5.8 and 7.4 are not likely to be biologically relevant. The pH 7 polymorphs (Type 5 and 1M) would be more interesting because they form under the same conditions and might be related to some disease relevant structures. Still, it is rare that a single polymorph appears at 7.0 (the Type 5 represented only 10-20% of the fibrils in the sample and the Type 1M also had unidentified double-filament fibrils in the sample). We plan to pursue this line of research and hope to include it in a future publication.

(2) The cross-seeding study presented in the manuscript demonstrates the pivotal role of pH conditions in dictating conformation. However, an intriguing aspect that emerges is the potential role of seed concentration in determining the resultant product structure. This raises a critical question: at what specific seed concentration does the determining factor for polymorph selection shift from pH condition to seed concentration? A methodological robust approach to address this should be conducted through a series of experiments across a range of seed concentrations. Such an approach could delineate a clear boundary at which seed concentration begins to predominantly dictate the conformation, as opposed to pH conditions. Incorporating this aspect into the study would not only clarify the interplay between seed concentration and pH conditions, but also add a fascinating dimension to the understanding of polymorph selection mechanisms.

A more complete analysis of the mechanisms of aggregation, including the effect of seed concentration and the resulting polymorph specificity of the process, are all very important for our understanding of the aggregation pathways of alphasynuclein and are currently the topic of ongoing investigations in our lab.

Furthermore, the study prompts additional queries regarding the behavior of cross-seeding production under the same pH conditions when employing seeds of distinct conformation. Evidence from various studies, such as those involving E46K and G51D cross-seeding, suggests that seed structure plays a crucial role in dictating polymorph selection. A key question is whether these products consistently mirror the structure of their respective seeds. 

We thank the reviewer for reminding us to cite these studies as a clear example of polymorph selection by cross-seeding. Unfortunately, it is not 100% clear from the G51D cross seeding manuscript (https://doi.org/10.1038/s41467-021-26433-2) what conditions were used in the cross-seeding since different conditions were used for the seedless wild-type and mutant aggregations… however it appears that the wildtype without seeds was Tris pH 7.5 (although at 37C the pH could have dropped to 7ish) and the cross-seeded wild-type was in Phosphate buffer at pH 7.0. In the E46K cross-seeding manuscript, it appears that pH 7.5 Tris was used for all fibrilizations (https://doi.org/10.1073/pnas.2012435118).  In any event, both results point to the fact that at pH 7.0-7.5 under low-seed conditions (0.5%) the Type 4 polymorph can propagate in a seed specific manner.

(3) In the Results section of "The buffer environment can dictate polymorph during seeded nucleation", the authors reference previous cell biological and biochemical assays to support the polymorph-specific seeding of MSA and PD patients under the same buffer conditions. This discussion is juxtaposed with recent research that compares the in vivo biological activities of hPFF, ampLB as well as LB, particularly in terms of seeding activity and pathology. Notably, this research suggests that ampLB, rather than hPFF, can accurately model the key aspects of Lewy Body Diseases (LBD) (refer to: https://doi.org/10.1038/s41467-023-42705-5). The critical issue here is the need to reconcile the phenomena observed in vitro with those in in-vivo or in-cell models. Given the low seed concentration reported in these studies, it is imperative for the authors to provide a more detailed explanation as to why the possible similar conformation could lead to divergent pathologies, including differences in cell-type preference and seeding capability.  

We thank the reviewer for bring this recent report to our attention. The findings that ampLB and hPFF have different PK digestion patterns and that only the former is able to model key aspects of Lewy Body disease are in support of the seed-specific nature of some types of alpha-synuclein aggregation.  We have added this to the discussion regarding the significant role that seed type and seed conditions likely play in polymorph selection.

(4) In the Method section of "Image processing", the authors describe the helical reconstruction procedure, without mentioning much detail about the 3D reconstruction and refinement process. For the benefit of reproducibility and to facilitate a deeper understanding among readers, the authors should enrich this part to include more comprehensive information, akin to the level of detail found in similar studies (refer to:

https://doi.org/10.1038/nature23002).

As also suggested by reviewer #2, we have now added more comprehensive information on the 3D reconstruction and refinement process.

(5) The abbreviation of amino acids should be unified. In the Results section "On the structural heterogeneity of Type 1 polymorphs", the amino acids are denoted using three-letter abbreviation. Conversely, in the same section under "On the structural heterogeneity of Type 2 and 3 structures", amino acids are abbreviated using the one-letter format. For clarity and consistency, it is essential that a standardized format for amino acid abbreviations be adopted throughout the manuscript.

That makes perfect sense and had been corrected.

Reviewing Editor:

After discussion among the reviewers, it was decided that point 2 in Reviewer #3's Public Review (about the experiments with different concentrations of seeds) would probably lie outside the scope of a reasonable revision for this work. 

We agree as stated above and will continue to work on this important point.

eLife assessment

This study presents a valuable strategy to co-deliver peptides and adjuvants to antigen-presenting cells by engineering the Virus-like particle (VLP). The evidence supporting the claims of the authors is convincing, but the antitumour efficacy is unimpressive and would benefit from more antitumor experiments. The work will be of broad interest to bioengineers and medical biologists focusing on cancer vaccines.

Reviewer #1 (Public Review):

Tang et al present an important manuscript focused on endogenous virus-like particles (eVLP) for cancer vaccination with solid in vivo studies. The author designed eVLP with high protein loading and transfection efficiency by PEG10 self-assembling while packaging neoantigens inside for cancer immunotherapy. The eVLP was further modified with CpG-ODN for enhanced dendritic cell targeting. The final vaccine ePAC was proven to elicit strong immune stimulation with increased killing effect against tumor cells in 2 mouse models. Below are my specific comments:

(1) The figures were well prepared with minor flaws, such as missed scale bars in Figures 4B, 4K, 5B, and 5C. The author should also add labels representing statistical analysis for Figures 3C, 3D, and 3E. In Figure 6G, the authors should label which cell type is the data for.

(2) In Figure 3H, the antigen-presenting cells (APCs) increased significantly, but there was also a non-negligible 10% of APCs found in the control group, indicating some potential unwanted immune response; the authors need to explain this phenomenon or add a cytotoxic test on the normal liver or other cell lines for confirmation.

(3) In Figure 3I, the ePAC seems to have a very similar effect on cytotoxic T-cell tumor killing compared to the peptides + CpG group. If the concentrations were also the same, based on that, questions will arise as to what is the benefit of using the compact vector other than just free peptide and CpG? Please explain and elaborate.

(4) In the animal experiment in Figures 4F to L, the activation effect of APCs was similar between ePAC and CpG-only groups with no significance, but when it comes to the HCC mouse model in Figure 5, the anti-tumor effect was significantly increased between ePAC and CpG-only group. The authors should explain the difference between these two results.

Reviewer #2 (Public Review):

Summary:

The authors provided a novel antigen delivery system that showed remarkable efficacy in transporting antigens to develop cancer therapeutic vaccines.

Strengths:

This manuscript was innovative, meaningful, and had a rich amount of data.

Weaknesses:

There are still some issues that need to be addressed and clarified.

(1) The format of images and data should be unified. Specifically, as follows: a. The presentation of flow cytometry results; b, The color schemes for different groups of column diagrams.

(2) The P-value should be provided in Figures, including Figure 1F, 1H, 3C, 3D, and 3E.

(3) The quality of Figure 1C was too low to support the conclusion. The author should provide higher-quality images with no obvious background fluorescent signal. Meanwhile, the fluorescent image results of "Egfp+VSVg" group were inconsistent with the flow cytometry data. Additionally, the reviewer recommends that the authors use a confocal microscope to repeat this experiment to obtain a more convincing result.

(4) The survival situation of the mouse should be provided in Figure 5, Figure 6, and Figure 7 to support the superior tumor therapy effect of ePAC.

(5) To demonstrate that ePAC could trigger a strong immune response, the positive control group in Figure 4K should be added.

(6) In Figure 6G-I and other figures, the author should indicate the time point of detection. Meanwhile, there was no explanation for the different numbers of mice in Figure 6G-I. If the mouse was absent due to death, it may be necessary to advance the detection time to obtain a more convincing result.

(7) In Figure 6B, the rainbow color bar with an accurate number of maximum and minimum fluorescence intensity should be provided. In addition, the corresponding fluorescence intensity in Figure 6B should be noted.

(8) The quality of images in Figure 1D and Figure S1B could not support the author's conclusion; please provide higher-quality images.

(9) In Figure 2F, the bright field in the overlay photo may disturb the observation. Meanwhile, the scale bar should be provided in enlarged images.

Reviewer #3 (Public Review):

Summary:

The authors harnessed the potential of mammalian endogenous virus-like proteins to encapsulate virus-like particles (VLPs), enabling the precise delivery of tumor neoantigens. Through meticulous optimization of the VLP component ratios, they achieved remarkable stability and efficiency in delivering these crucial payloads. Moreover, the incorporation of CpG-ODN further heightened the targeted delivery efficiency and immunogenicity of the VLPs, solidifying their role as a potent tumor vaccine. In a diverse array of tumor mouse models, this novel tumor vaccine, termed ePAC, exhibited profound efficacy in activating the murine immune system. This activation manifested through the stimulation of dendritic cells in lymph nodes, the generation of effector memory T cells within the spleen, and the infiltration of neoantigen-specific T cells into tumors, resulting in robust anti-tumor responses.

Strengths:

This study delivered tumor neoantigens using VLPs, pioneering a new method for neoantigen delivery. Additionally, the gag protein of VLP is derived from mammalian endogenous virus-like protein, which offers greater safety compared to virus-derived gag proteins, thereby presenting a strong potential for clinical translation. The study also utilized a humanized mouse model to further validate the vaccine's efficacy and safety. Therefore, the anti-tumor vaccine designed in this study possesses both innovation and practicality.

Weaknesses:

(1) CpG-ODN is an FDA-approved adjuvant with various sequence structures. Why was CpG-ODN 1826 directly chosen in this study instead of other types of CpG-ODN? Additionally, how does DEC-205 recognize CpG-ODN 1826, and can DEC-205 recognize other types of CpG-ODN?

(2) Why was it necessary to treat DCs with virus-like particles three times during the in vitro activation of T cells? Can this in vitro activation method effectively obtain neoantigen-responsive T cells?

(3) In the humanized mouse model, the authors used Hepa1-6 cells to construct the tumor model. To achieve the vaccine's anti-tumor function, these Hepa1-6 cells were additionally engineered to express HLA-A0201. However, in the in vitro experiments, the authors used the HepG2 cell line, which naturally expresses HLA-A0201. Why did the authors not continue to use HepG2 cells to construct the tumor model, instead of Hepa1-6 cells?

(4) The advantages of low immunogenicity viruses as vaccines compared with conventional adenovirus and lentivirus, etc. should be discussed.

(5) In Figure 6B, the authors should provide statistical results.

(6.) The entire article demonstrates a clear logical structure and substantial content in its writing. However, there are still some minor errors, such as the misspelling of "Spleenic" in Figure 3B, and the sentence from line 234 should be revised.

(7) The authors demonstrated the efficiency of CpG-ODN membrane modification by varying the concentration of DBCO, ultimately determining the optimal modification scheme for eVLP as 3.5 nmol of DBCO. However, in Figure 2B, the author did not provide the modification efficiency when the DBCO concentration is lower than 3.5 nmol. These results should be provided.

(8) In Figure 3, the authors presented a series of data demonstrating that ePAC can activate mouse DC2.4 cells and BMDCs in vitro. However, in Figure 7, there is no evidence showing whether human DC cells can be activated by ePAC in vitro. This data should be provided.

eLife assessment

This study presents fundamental findings that could redefine the specificity and mechanism of action of the well-studied Ser/Thr kinase IKK2 (a subunit of inhibitor of nuclear factor kappa-B kinase (IkB) that propagates cellular response to inflammation). Solid evidence supports the claim that IKK2 exhibits dual specificity that allows tyrosine autophosphorylation and the authors further show that auto-phosphorylated IKK2 is involved in an unanticipated relay mechanism that transfers phosphate from an IKK2 tyrosine onto the IkBa substrate. These are potentially provocative results but open questions remain due to the nature of the in vitro assays and questions about protein purity and identity. Nevertheless, the findings are a starting point for follow-up studies to confirm the unexpected mechanism and further pursue functional significance.

Reviewer #1 (Public Review):

The model of phosphotransfer from Y169 IKK to S32 IkBa is compelling and an important new contribution to the field. In fact, this model will not be without controversy, and publishing the work will catalyze follow-up studies for this kinase and others as well. As such, I am supportive of this paper, though I do also suggest some shortening and modification.

Generally, the paper is well written, but several figures should be quantified, and experimental reproducibility is not always clear. The first 4 figures are slow-going and could be condensed to show the key points, so that the reader gets to Figures 6 and 7 which contain the "meat" of the paper.

Reviewer #2 (Public Review):

Summary:<br /> The authors investigate the phosphotransfer capacity of Ser/Thr kinase IκB kinase (IKK), a mediator of cellular inflammation signaling. Canonically, IKK activity is promoted by activation loop phosphorylation at Ser177/Ser181. Active IKK can then unleash NF-κB signaling by phosphorylating repressor IκBα at residues Ser32/Ser26. Noting the reports of other IKK phosphorylation sites, the authors explore the extent of autophosphorylation.

Semi-phosphorylated IKK purified from Sf9 cells, exhibits the capacity for further autophosphorylation. Anti-phosphotyrosine immunoblotting indicated unexpected tyrosine phosphorylation. Contaminating kinase activity was tested by generating a kinase-dead K44M variant, supporting the notion that the unexpected phosphorylation was IKK-dependent. In addition, the observed phosphotyrosine signal required phosphorylated IKK activation loop serines.

Two candidate IKK tyrosines were examined as the source of the phosphotyrosine immunoblotting signal. Activation loop residues Tyr169 and Tyr188 were each rendered non-phosphorylatable by mutation to Phe. The Tyr variants decreased both autophosphorylation and phosphotransfer to IκBα. Likewise, Y169F and Y188F IKK2 variants immunoprecipitated from TNFa-stimulated cells also exhibited reduced activity in vitro.

The authors further focus on Tyr169 phosphorylation, proposing a role as a phospho-sink capable of phosphotransfer to IκBα substrate. This model is reminiscent of the bacterial two-component signaling phosphotransfer from phosphohistidine to aspartate. Efforts are made to phosphorylate IKK2 and remove ATP to assess the capacity for phosphotransfer. Phosphorylation of IκBα is observed after ATP removal, although there are ambiguous requirements for ADP.

Strengths:

Ultimately, the authors draw together the lines of evidence for IKK2 phosphotyrosine and ATP-independent phosphotransfer to develop a novel model for IKK2-mediated phosphorylation of IκBα. The model suggests that IKK activation loop Ser phosphorylation primes the kinase for tyrosine autophosphorylation. With the assumption that IKK retains the bound ADP, the phosphotyrosine is conformationally available to relay the phosphate to IκBα substrate. The authors are clearly aware of the high burden of evidence required for this unusual proposed mechanism. Indeed, many possible artifacts (e.g., contaminating kinases or ATP) are anticipated and control experiments are included to address many of these concerns. Taken together, the observations are thought-provoking, and I look forward to seeing this model tested in a cellular system.

Weaknesses:

It seems that the analysis hinges on the fidelity of pan-specific phosphotyrosine antibodies.

The analysis often returns to the notion that tyrosine phosphorylation(s) (and critical active site Lys44) dictate IKK2 substrate specificity, but evidence for this seems diffuse and indirect. This is an especially difficult claim to make with in vitro assays, omitting the context of other cellular specificity determinants (e.g., localization, scaffolding, phosphatases).

Multiple phosphorylated tyrosines in IKK2 were apparently identified by mass spectrometric analyses, but the data and methods are not described. It is common to find non-physiological post-translational modifications in over-expressed proteins from recombinant sources. Are these IKK2 phosphotyrosines evident by MS in IKK2 immunoprecipitated from TNFa-stimulated cells? Identifying IKK2 phosphotyrosine sites from cells would be especially helpful in supporting the proposed model.

Reviewer #3 (Public Review):

Summary:

The authors investigate the kinase activity of IKK2, a crucial regulator of inflammatory cell signaling. They describe a novel tyrosine kinase activity of this well-studied enzyme and a highly unusual phosphotransfer from phosphorylated IKK2 onto substrate proteins in the absence of ATP as a substrate.

Strengths:

The authors provide an extensive biochemical characterization of the processes with recombinant protein, western blot, autoradiography, and protein engineering.

Weaknesses:

The identity and purity of the used proteins is not clear. Since the findings are so unexpected and potentially of wide-reaching interest - this is a weakness. Similar specific detection of phospho-Ser/Thr vs phospho-Tyr relies largely on antibodies which can have varying degrees of specificity.

eLife assessment

This important study investigates the sensitivity to endogenous cosolvents of three families of intrinsically disordered proteins involved with desiccation. The findings, drawn from well-designed experiments and calculations, suggest a functional synergy between sensitivity to small molecule solutes and convergent desiccation protection strategy. While the evidence is found to be convincing, the study's conclusions cannot be generalized due to the small number of proteins investigated. This work will be of interest to biochemists and biophysicists interested in the conformation-function relationship of intrinsically disordered proteins.

Reviewer #1 (Public Review):

Summary:

The individual roles of both cosolvents and intrinsically disordered proteins (IDPs) in desiccation have been well established, but few studies have tried to elucidate how these two factors may contribute synergistically. The authors quantify the synergy for the model and true IDPs involved with desiccation and find that only the true IDPs have strong desiccation tolerance and synergy with cosolvents. Using these as model systems, they quantify the local (secondary structure vis-a-vi CD spectroscopy) and global dimensions (vis-a-vi the Rg of SAXS experiments) and find no obvious changes with the co-solvents. Instead, they focus on the gelation of one of the IDPs and, using theory and experiments, suggest that the co-solvents may enable desiccation tolerance, an interesting hypothesis to guide future in vivo desiccation studies. A few minor points that remain unclear to this reviewer are noted.

Strengths:

This paper is quite extensive and has significant strengths worth highlighting. Notably, the number and type of methods employed to study IDPs are quite unusual, employing CD spectroscopy, SAXS measurements, and DSC. The use of the TFE is an exciting integration of the physical chemistry of cosolvents into the desiccation field is a nice approach and a clever way of addressing the gap of the lack of conformational changes depending on the cosolvents. Furthermore, I think this is a major point and strength of the paper; the underlying synergy of cosolvents and IDPs may lie in the thermodynamics of the dehydration process.

Figure S6A is very useful. I encourage readers who are confused about the DSC analysis, interpretation, and calculation to refer to it.

Weaknesses:

Overall, the paper is sound and employs strong experimental design and analysis. However, I wish to point out a few minor weaknesses.

Perhaps the largest, in terms of reader comprehension, focuses on the transition between the model peptides and real IDPs in Figures 1 and 2. Notably, little is discussed with respect to the structure of the IDPs and what is known. Notably, I was confused to find out when looking at Table 1 that many of the IDPs are predicted to be largely unordered, which seemed to contrast with some of the CD spectroscopy data. I wonder if the disorder plots are misleading for readers. Can the authors comment more on this confusion? What are these IDPs structurally?

Related to the above thoughts, the alpha fold structures for the LEA proteins are predicted (unconfidently) as being alpha-helical in contrast to the CD data. Does this complicate the TFE studies and eliminate the correlation for the LEA proteins? Additionally, the notation that the LEA and BSA proteins do not correlate is unclear to this reviewer, aren't many of the correlations significant, having both a large R^2 and significant p-value?

The calculation of synergy seems too simplistic or even problematic to me. While I am not familiar with the standards in the desiccation field, I think the approach as presented may be problematic due to the potential for higher initial values of protection to have lower synergies (two 50%s for example, could not yield higher than 100%). Instead, I would think one would need to really think of it as an apparent equilibrium constant between functional and non-functional LDH (Kapp = [Func]/[Not Func] and frac = Kapp/(1+Kapp) or Kapp = frac/(1-frac) ) Then after getting the apparent equilibrium constants for the IDP and cosolvent (KappIDP and KappCS), the expected additive effect would be frac = (KappIDP+KappCS)/(1+KappIDP+KappCS). Consequently, the extent of synergy could be instead calculated as KappBOTH-KappIDP-KappCS. Maybe this reviewer is misunderstanding. It is recommended that the authors clarify why the synergy calculation in the manuscript is reasonable.

Related to the above, the authors should discuss the utility of using molar concentration instead of volume fraction or mass concentration. Notably, when trehalose is used in concentration, the volume fraction of trehalose is much smaller compared to the IDPs used in Figure 2 or some in Figure 1. Would switching to a different weighted unit impact the results of the study, or is it robust to such (potentially) arbitrary units?

Reviewer #2 (Public Review):

Summary:

The paper aims to investigate the synergies between desiccation chaperones and small molecule cosolutes, and describe its mechanistic basis. The paper reports that IDP chaperones have stronger synergies with the cosolutes they coexist with, and in one case suggests that this is related to oligomerization propensity of the IDP.

Strengths:

The study uses a lot of orthogonal methods and the experiments are technically well done. They are addressing a new question that has not really been addressed previously.

Weaknesses:

The conclusions are based on a few examples and only partial correlations. While the data support mechanistic conclusions about the individual proteins studied, it is not clear that the conclusions can be generalized to the extent proposed by the authors due to small effect sizes, small numbers of proteins, and only partial correlations.

The authors pose relevant questions and try to answer them through a systematic series of experiments that are all technically well-conducted. The data points are generally interpreted appropriately in isolation, however, I am a little concerned about a tendency to over-generalize their findings. Many of the experiments give negative or non-conclusive results (not a problem in itself), which means that the overall storyline is often based on single examples. For example, the central conclusion that IDPs interact synergistically with their endogenous co-solute (Figure 2E) is largely driven by one outlier from Arabidopsis. The rest are relatively close to the diagonal, and one could equally well suggest that the cosolutes affect the IDPs equally (which is also the conclusion in 1F). Similarly, the mechanistic explanations tend to be based on single examples. This is somewhat unavoidable as biophysical studies cannot be done on thousands of proteins, but the text should be toned down to reflect the strength of the conclusions.

The central hypothesis revolves around the interplay between cosolutes and IDP chaperones comparing chaperones from species with different complements of cosolutes. In Table 1, it is mentioned that Arabidopsis uses both trehalose and sucrose as a cosolute, yet experiments are only done with either of these cosolutes and Arabidopsis is counted in the sucrose column. While it makes sense to compare them separately from a biophysical point of view, the ability to test the co-evolution of these systems is somewhat diminished by this. At least it should be discussed clearly.

It would be helpful if the authors could spell out the theoretical basis of how they quantify synergy. I understand what they are doing - and maybe there are no better ways to do it - but it seems like an approach with limitations. The authors identify one in that the calculation only works far from 100%, but to me, it seems there would be an equally strict requirement to be significantly above 0%. This would suggest that it is used wrongly in Figure 6H, where there is no effect of betaine (at least as far as the color scheme allows one to distinguish the different bars). In this case, the authors cannot really conclude synergy or not, it could be a straight non-synergistic inhibition by betaine.

eLife assessment

The authors study how inflammatory priming and exposure to irradiated Mycobacterium tuberculosis or the bacterial endotoxin LPS impact the metabolism of primary human airway macrophages and monocyte-derived macrophages. The work shows that metabolic plasticity is greater in monocyte-derived macrophages than alveolar macrophages. The experimental methods and evidence are solid, and the results and findings are useful for the field of immunometabolism.

Reviewer #1 (Public Review):

Summary:

The researchers demonstrated that when cytokine priming is combined with exposure to pathogens or pathogen-associated molecular patterns, human alveolar macrophages and monocyte-derived macrophages undergo metabolic adaptations, becoming more glycolytic while reducing oxidative phosphorylation. This metabolic plasticity is greater in monocyte-derived macrophages than in alveolar macrophages.

Strengths:

This study presents evidence of metabolic reprogramming in human macrophages, which significantly contributes to our existing understanding of this field primarily derived from murine models.

Weaknesses:

The study has limited conceptual novelty.

Reviewer #2 (Public Review):

Summary:

The authors aimed to functionally characterize primary human airway macrophages and monocyte-derived macrophages, correlating their glycolytic shift in metabolism. They conducted this macrophage characterization in response to type II interferon and IL-4 priming signals, followed by different stimuli of irradiated Mycobacterium tuberculosis and LPS.

Strengths:

(1) The study employs a thorough measurement of metabolic shift in metabolism by assessing extracellular acidification rate (ECAR) and oxygen consumption rate (OCR) of differentially polarized primary human macrophages using the Seahorse XFe24 Analyzer.<br /> (2) The effect of differential metabolic shift on the expression of different surface markers for macrophage activation is evaluated through immunofluorescence flow cytometry and cytokine measurement via ELISA.<br /> (3) The authors have achieved their aim of preliminarily characterizing the glycolysis-dependent cytokine profile and activation marker expression of IFN-g and IL-4 primed primary human macrophages.<br /> (4) The results of the study support its conclusion of glycolysis-dependent phenotypical differences in cytokine secretion and activation marker expression of AMs and MDMs.

Weaknesses:

(1) The data are presented in duplicates for cross-analyses.<br /> (2) The data presented supports a distinct functional profile of airway macrophages (AMs) compared to monocyte (blood)-derived macrophages (MDMs) in response to the same priming signals. However, the study does not attempt to explore the underlying mechanism for this difference.<br /> (3) The study is descriptive in nature, and the results validate IFN-g-mediated glycolytic reprogramming in primary human macrophages without providing mechanistic insights.

Reviewer #3 (Public Review):

Summary:

In this manuscript, the authors explore the contribution of metabolism to the response of two subpopulations of macrophages to bacterial pathogens commonly encountered in the human lung, as well as the influence of priming signals typically produced at a site of inflammation. The two subpopulations are resident airway macrophages (AM) isolated via bronchoalveolar lavage and monocyte-derived macrophages (MDM) isolated from human blood and differentiated using human serum. The two cell types were primed using IFNγ and Il-4, which are produced at sites of inflammation as part of initiation and resolution of inflammation respectively, followed by stimulation with either irradiated Mycobacterium tuberculosis (Mtb) or LPS to simulate interaction with a bacterial pathogen. The authors use human cells for this work, which makes use of widely reported and thoroughly described priming signals, as well as model antigens. This makes the observations on the functional response of these two subpopulations relevant to human health and disease. To examine the relationship between metabolism and functional response, the authors measure rates of oxidative phosphorylation and glycolysis under baseline conditions, primed using IFNγ or IL-4, and primed and stimulated with Mtb or LPS.

Strengths:

• The data indicate that both populations of macrophages increase metabolic rates when primed, but MDMs decrease their rates of oxidative phosphorylation after IL-4 priming and bacterial exposure while AMs do not.<br /> • It is demonstrated that glycolysis rates are directly linked to the expression of surface molecules involved in T-cell stimulation and while secretion of TNFα in AM is dependent on glycolysis, in MDM this is not the case. IL-1β is regulated by glycolysis only after IFN-γ priming in both MDM and AM populations. It is also demonstrated that Mtb and LPS stimulation produces responses that are not metabolically consistent across the two macrophage populations. The Mtb-induced response in MDMs differed from the LPS response, in that it relies on glycolysis, while this relationship is reversed in AMs. The difference in metabolic contributions to functional outcomes between these two macrophage populations is significant, despite acknowledgement of the reductive nature of the system by the authors.<br /> • The observations that AM and MDM rely on glycolysis for the production of cytokines during a response to bacterial pathogens in the lung, but that only MDM shift to Warburg Metabolism, though this shift is blocked following exposure to IL-4, are supported by the data and a significant contribution the study of the innate immune response.

Weaknesses:

• It is unclear whether changes in glycolysis and oxidative phosphorylation in primed cells are due to priming or subsequent treatments. ECAR and OCR analyses were therefore difficult to interpret.<br /> • The data may not support a claim that AM has greater "functional plasticity" without a direct comparison of antigen presentation. Moreover, MDM secrete more IL-1β than AM. The claim that AM "have increased ability to produce all cytokines assayed in response to Mtb stimulation" does not appear to be supported by the data.<br /> • The claim that AM are better for "innate training" via IFNγ may not be consistent with increased IL-1β and a later claim that MDM have increased production and are "associated with optimal training."<br /> • Statistical analyses may not appropriately support some of the conclusions.<br /> • AM populations would benefit from further definition-presumably this is a heterogenous, mixed population.<br /> • The term "functional plasticity" could also be more stringently defined for the purposes of this study.

Conclusion:

Overall, the authors succeed in their goals of investigating how inflammatory and anti-inflammatory cytokine priming contributes to the metabolic reprogramming of AM and MDM populations. Their conclusions regarding the relationship between cytokine secretion and inflammatory molecule expression in response to bacterial stimuli are supported by the data. The involvement of metabolism in innate immune cell function is relevant when devising treatment strategies that target the innate immune response during infection. The data presented in this paper further our understanding of that relationship and advance the field of innate immune cell biology.

Author response:

The following is the authors’ response to the current reviews.

eLife assessment:

The statistical analyses are incomplete.

I find that the eLife assessment mentions “statistical analyses are incomplete” while the reviewers mention that the data are compelling and results are technically solid. Besides all observations in the manuscript are presented with robust and established norms of statistical analysis. Therefore, I would kindly request to remove the statement as it undermines the article.

Public Reviews:

Reviewer #1 (Public Review):

Strengths:

The use of data from before COVID-19 is both a strength and a weakness. Because COVID had effects on vascular health and had higher death rates for groups with the comorbidities of interest here, it has likely shifted the demographics in ways that would shift the results in unpredictable ways if the analysis were repeated with current data. This can be a strength in providing a reference point for studying those changes as well as allowing researchers to study differences between regions without the complication of different public health responses adding extra variation to the data. On the other hand, it limits the usefulness of the data in research concerned with the current status of the various populations.

We completely agree with the observation, but were restricted as the purpose was to use the most robust and technically qualified data from GBD. The post COVID19 GBD data has not yet been released, but I am sure the observations made in the study can help in guiding the issues in the post COVID era too, because genetics is not going to change in these population groups.

However, we did highlight this aspect of COVID19 even in our original version and also in the revised version.

Reviewer #2 (Public Review):

Weaknesses:

The presentation is not focused. It would be better to include p-values and focus presentation on the main effects from the dataset analysis.

The significant p-values were restricted to public health data only to identify and distinguish differences in incidence, prevalence and mortality and how they differ across world populations. These differences have often been interpreted from socio-economic point of view, while our manuscript presents the reasons for differences for main condition (Stroke) and its comorbid condition among different ethnicities from a genetic perspective. This genetic perspective was further explored to identify unique ethnic specific variants and their patterns of linkage disequilibrium in distinguishing the phenotypic variations. Considering the quantum and diversity of data, both for public health and GWAS data, there can be several directions but for presentation we focused only on the most distinguishing and established phenotypic differences. I am sure this will open up avenues for several future investigations including COVID, as has been highlighted by the reviewers too. All observations were presented with robust and established norms of statistical analysis.

---

The following is the authors’ response to the original reviews.

Thanks for the constructive observations on strengths and weaknesses of our manuscript. Interestingly, some of the weaknesses mentioned here also turns out to be the strength of the article. For example COVID19 has been mentioned by the reviewer as a driver to increase the mortality in some comorbid conditions and stroke. Firstly, I must clarify that, our data is from PreCOVID era and we indeed mention that in COVID era, COVID-19 might differentially impact the risk of stroke. Possibly this differential influence on the comorbidities of stroke, is likely to be influenced by its underlying genetics of stroke and its comorbidities.

I have tried to address the concerns raised by the reviewers, which ideally doesn’t impact the original manuscript. Statistical limitation has been commented pertaining to P-values, which has been clarified here. However, certain minor concerns such as abbreviations have been resolved in the revised manuscript. My response to weakness and reviewer’s comments are mentioned below.

Reviewer #1 (Public Review):

Strengths:

The data provided here will provide a foundation for a lot of future research into the causes of the observed correlations as well as whether the observed differences in comorbidities across regions have clinically relevant effects on risk management.

Weaknesses:

• As with any cross-national analysis of rates, the data is vulnerable to differences in classification and reporting across jurisdictions.

GBD data is the most robust and most comprehensive data resource which has been used and accepted globally in predicting the health metrics statistics.

GBD data indeed considers normalisations, regarding classification and reporting.

To the best of our knowledge this is the best available resource to consider all health metrics analysis.

• Furthermore, given the increased death rate from COVID-19 associated with many of these comorbid conditions and the long-term effects of COVID-19 infection on vascular health, it is expected that many of the correlations observed in this dataset will shift along with the shifting health of the underlying populations.

I must clarify that we have used data prior to COVID-19.

But yes the patterns after COVID19 will shift due to the impact of covid. This makes the study even more relevant as the comorbid conditions of stroke are also the risk drivers for COVID19 and mortality. This shift has been reported by some authors, which has been discussed in the discussion.

Therefore, understanding the genetic factors underlying stroke and its comorbid conditions might help in resolving how COVID19 might differentially impact on health outcome.

We did highlight this aspect of COVID19 even in our original version.

Introduction 1st para:

“It is the accumulated risk of comorbid conditions that enhances the risk of stroke further. Are these comorbid conditions differentially impacted by socio-economic factors and ethnogeographic factors. This was clearly evident in COVID era, when COVID-19 differentially impacted the risk of stroke, possibly due to its differential influence on the comorbidities of stroke.”

Discussion 3rd para:

“Studies reported reduction in life expectancy in 31 of 37 high-income countries, deduced to be due to COVID-191 . However, it would be unfair to ignore the comorbid conditions which could also be the critical determinants for reduced life expectancy in these countries.”

Recommendations For The Authors:

On page 5, the authors make a note about Africa and the Middle East having the highest ASMR for high SBP and comment about the relative populations of these regions. The populations of the regions are irrelevant to the rate.

Since the study is on comorbid factors of stroke and its impact on mortality therefore, relative burden seems critical. This has been further elaborated here to justify the comment, which indeed is an integral part of the original manuscript.

Paragraph referred – Results section 2:

“Ethno-regional differences in mortality and prevalence of stroke and its major comorbid conditions

We observed interesting patterns of ASMRs of stroke, its subtypes and its major comorbidities across different regions over the years as shown in figure 1a, table 1 and supplementary files S2 & S3. When assessed in terms of ranks, high SBP is the most fatal condition followed by IHD in all regions, except Oceania (OCE) where IHD and high SBP swap ranks. Africa (AFR; 206.2/100000, 95%UI 177.4-234.2) and Middle East (MDE; 198.6/100000, 95%UI 162.8-234.4) have the highest ASMR for high SBP, even though they rank as only the third and sixth most populous continents (fig. S2), respectively.”

On page 17, the authors are alarmed by a large ratio between prevalence rates and mortality rates for certain conditions. This is confusing since this indicates that these conditions are not as dangerous as the other conditions.

This has been further elaborated here to justify the comment, which indeed is an integral part of the original manuscript.

Paragraph referred – Discussion para 1:

“While the global stroke prevalence is nearly 15 times its mortality rate, prevalence of comorbid conditions such as high SBP, high BMI, CKD, T2D are alarmingly 150- to 500-fold higher than their mortality rates. These comorbid conditions can drastically affect the outcome of stroke.”

In Figure 4, the colors are not defined.

In Structure plot colours are assigned as per each K, it doesn’t directly refer to any population. But the plot distinguishes the stratification of populations as per K. Ramasamy, R.K., Ramasamy, S., Bindroo, B.B. et al. STRUCTURE PLOT: a program for drawing elegant STRUCTURE bar plots in user friendly interface. SpringerPlus 3, 431 (2014). https://doi.org/10.1186/2193-1801-3-431

Reviewer #2 (Public Review):

Strengths:

The idea is interesting and the data are compelling. The results are technically solid.

The authors identify specific genetic loci that increase the risk of a stroke and how they differ by region.

Weaknesses:

The presentation is not focused. It would be better to include p-values and focus presentation on the main effects of the dataset analysis.

I presume the comment is made with reference to results with significant p-values.

P-values are mentioned in the main text when referring to significant decrease or increase with respect to global rates and time e.g. P-values for comparison of a year 2019, are based on regional rates to global rates of 2019. Supplementary table S2a (mortality) and S3a (prevalence) e.g. P-values for comparison between year is based on 2019 rates to 2009 rates in Supplementary table S2b (mortality) and S3b (prevalence) e.g. P-values for proportional mortality and proportional prevalence in Supplementary table S4 and S5 is also based on global rates.

Recommendations For The Authors:

It would be better to minimize the use of acronyms. Often one has to go back to decipher what the acronym stands for. It is fine to have acronyms in figure legends, if necessary. However, at least for regions, please do not use acronyms.

In the revised version we have tried to minimise the Acronyms.

Removed the acronyms for regions and other places wherever possible however, the diseases acronyms have been maintained as per the GBD terms.

Please focus the presentation on the main results. Currently, the presentation wanders and repeats itself a lot.

Since the manuscript tries to address the global and regional rates of prevalence, mortality and its relationship to genetic correlates, it is difficult not to repeat the same to stress the significant observations coming out of different analysis methods. This might reflect on some amount of repetitiveness but the intention was to stress the significant observations.

I would also recommend acknowledging and discussing socioeconomic factors earlier in the manuscript.

Current mention happens in 3rd para of Discussion

“The changing dynamics of stroke or its comorbid conditions can be attributed to multitude of factors. Often global burden of stroke has been discussed from the point of view of socio-economic parameters. Studies indicate that half of the stroke-related deaths are attributable to poor management of modifiable risk factors 8,9. However, we observe that different socio-economic regions are driven by different risk factors.”

eLife assessment

This paper provides a useful analysis of the variation of the burden of strokes across geographic regions, finding differences in the relationship between strokes and their comorbidities. This dataset and the correlations found within will be a resource for directing the focus of future investigations. The statistical analyses are incomplete.

Reviewer #1 (Public Review):

Summary:

The paper measures the prevalence and mortality of stroke and its comorbidities across geographic regions in order to find differences in risks that may lead to more effective guidance for these subpopulations. It also does a genetic analysis to look for variants that may drive these phenotypic variations.

Strengths:

The data provided here will provide a foundation for a lot of future research into the causes of the observed correlations as well as whether the observed differences in comorbidities across regions have clinically relevant effects on risk management.

The use of data from before COVID-19 is both a strength and a weakness. Because COVID had effects on vascular health and had higher death rates for groups with the comorbidities of interest here, it has likely shifted the demographics in ways that would shift the results in unpredictable ways if the analysis were repeated with current data. This can be a strength in providing a reference point for studying those changes as well as allowing researchers to study differences between regions without the complication of different public health responses adding extra variation to the data. On the other hand, it limits the usefulness of the data in research concerned with the current status of the various populations.

Reviewer #2 (Public Review):

Summary:

The authors have analyzed ethnogeographic differences in the comorbidity factors, such as a diabetes and heart disease, for the incidences of stroke and whether it leads to mortality.

Strengths:

The idea is interesting and data are compelling. The results are technically solid.

The authors identify specific genetic loci that increase the risk of a stroke and how they differ by region.

Weaknesses:

The presentation is not focused. It would be better to include p-values and focus presentation on the main effects from the dataset analysis.

eLife assessment

This fundamental study analyzes the roles of post-translational modifications of tubulin by generating a large panel of tubulin mutants and describing their effects on morphogenesis and function of sensory neurons in C. elegans. The work, which is of interest to all cell biologists, in particular researchers with an interest in the microtubule cytoskeleton and neurobiology, presents conclusions that are supported by solid evidence. Demonstrating that all introduced mutations have the intended consequences and exploring their direct effect on microtubules would further increase the impact of the work.

Reviewer #2 (Public Review):

Summary:

The tubulin subunits that make up microtubules can be posttranslationally modified and these PTMs are proposed to regulate microtubule dynamics and the proteins that can interact with microtubules in many contexts. However, most studies investigating the roles of tubulin PTMs have been conducted in vitro either with purified components or in cultured cells. Lu et al. use CRISPR/Cas9 genome editing to mutate tubulin genes in C. elegans, testing the role of specific tubulin residues on neuronal development. This study is a real tour de force, tackling multiple proposed tubulin modifications and following the resulting phenotypes with respect to neurite outgrowth in vivo. There is a ton of data that experts in the field will likely reference for years to come as this is one of the most comprehensive in vivo analyses of tubulin PTMs in vivo.

This paper will be very important to the field, however, it would be strengthened if: 1) the authors demonstrated that the mutations they introduced had the intended consequences on microtubule PTMs, 2) the authors explored how the various tubulin mutations directly affect microtubules, and 3) the findings are made generally more accessible to non C. elegans neurobiologist.

(1) The authors introduce several mutations to perturb tubulin PTMs, However, it is unclear to what extent the engineered mutations affecting tubulin in the intended way. i.e. are the authors sure that the PTMs they want to perturb are actually present in C. elegans. Many of the antibodies used did not appear to be specific and antibody staining was not always impacted in the mutant cases as expected. For example, is there any evidence that S172 is phosphorylated in C. elegans, e.g. from available phosphor-proteomic data? Given the significant amount of staining left in the S172A mutant, the antibody seems non-specific in this context and therefore not a reliable readout of whether MTs are actually phosphorylated at this residue. As another example, there is no evidence presented that K252 is acetylated in C. elegans. At the very least, the authors should consider demonstrating the conservation of these residues and the surrounding residues with other organisms where studies have demonstrated PTMs exist.

(2) Given that the authors have the mutants in hand, it would be incredibly valuable to assess the impact of these mutations on microtubules directly in all cases. MT phenotypes are inferred from neurite outgrowth phenotypes in several cases, the authors should look directly at microtubules and/or microtubule dynamics via EBP-2 when possible OR show evidence that the only way to derive the neurite phenotypes shown is through the inferred microtubule phenotypes. For example, the effect of the acetylation or detyrosination mutants on MTs was not assessed.

(3) There is a ton of data here that will be important for experts working in this field to dig into, however, for the more general cell biologist, some of the data are quite inaccessible. More cartoons and better labeling will be helpful as will consistent comparisons to control worms in each experiment. A good example of this issue is demonstrated in Figure 2 and Figure 4:

- Fig. 2: Please label images with what is being probed in each panel<br /> - Fig 2G is very hard to interpret-cartoon diagramming what is being observed would be helpful.<br /> - Line 182-185: is this referring to your data or to Wu et al? It is not clear in this paragraph when the authors are describing published work versus their own data presented here.<br /> - Fig 2!-2K is not well described. What experiment is being done here? What is dlk-1 and why did you look at this mutant?<br /> - Figure 4C: this phenotype is hard to interpret. Where is the wt control? Where is the quantification?<br /> - There are no WT comparison images in Figure 4I, making the quantification difficult to interpret

(4) In addition, I am left unconvinced of the negative data demonstrating that MBK does not phosphorylate tubulin. First, the data described in lines 207-211 does not appear to be presented anywhere. Second, RNAi is notoriously finicky in neurons, thus necessitating tissue specific degradation using either the ZF/ZIF-1 or AID/TIR1 systems which both work extremely well in C. elegans. Third, there appears to be increasing S172 phosphorylation in Figure 3 supplement 2 with added MBK-2, but there is no anti-tubulin blot to show equal loading, so this experiment is hard to interpret.

Author response:

The following is the authors’ response to the original reviews.

Public Reviews:

Reviewer #1 (Public Review):

(1) The manuscript by Lu et al aims to study the effects of tubulin post-translational modification in C. elegans touch receptor neurons. Authors use gene editing to engineer various predicted PTM mutations in a-tubulin MEC-12 and b-tubulin MEC-7. Authors generate and analyze an impressive battery of mutants in predicted phosphorylation site and acetylation site of b-tubulin MEC-7, K40 acetylation site in a-tubulin MEC-12, enzymatic site of the a-tubulin acetyltransferase MEC-17, and PTM sites in the MEC-12 and MEC-7 C-tails (glutamylation, detyrosination, delta-tubulin). This represents a lot of work, and will appeal to a readership interested in C. elegans touch receptor neurons. The major concern/criticism of this manuscript is whether the introduced mutation(s) directly affects a specific PTM or whether the mutation affects gene expression, protein expression/stability/localization, etc. As such, this work does convincingly demonstrate, as stated in the title, that "Editing of endogenous tubulins reveals varying effects of tubulin posttranslational modifications on axonal growth and regeneration." 

We thank the reviewer for the constructive comments. With regards to the major concern or criticism, we like to point out that we have previously characterized ~100 missense mutations in mec-7 and mec-12 (Zheng et al., 2017, PMID: 28835377; Lee et al., 2021, PMID: 33378215). So, we are familiar with the phenotypes associated with mutations that affect gene expression or protein stability, which mostly result in a null phenotype. When analyzing the PTM site mutants, we compared their phenotypes with the previously categorized phenotypes of null alleles, neomorphic mutations that increase microtubule stability, and antimorphic mutations that prevent polymerization or disrupt microtubule stability. For example, in the case of mec-7 S172 mutations, we found that S172P mutants had the same phenotype as the mec-7 knockout (mild neurite growth defects), suggesting that S172P likely affects protein folding or stability, resulting in the loss of MEC-7. In contrast, S172A and S172E mutations showed phenotypes similar to neomorphic alleles (the emergence of ectopic ALM posterior neurite) and antimorphic alleles (the severe shortening of all neurites in the TRNs), respectively. These phenotypic differences suggested to us that the effects of S172A and S172E mutations cannot be simply attributed to the loss of protein expression and stability. Similar logic was applied to the studies of other PTM-inactivating or -mimicking mutations.

(2) For example, the authors manipulate the C-terminal tail of MEC-12 and MEC-7, to test the idea that polyglutamylation may be an important PTM. These mutants displayed subtle phenotypes. The authors show that branch point GT335 and polyglutamyation polyE recognizing antibodies stain cultured embryonic touch receptor neurons (TRNs), but did not examine staining in C. elegans TRNs in situ. To my knowledge, these antibodies have not been shown to stain the TRNs in any published papers, raising the question of how these "glutamylation" mutations are affecting mec-12 and -7. The rationale for using cultured embryonic TRNs and the relevance of the data and its interpretation are not clear. 

The GT335 and polyE antibodies were used by previous studies (O’Hagan et al., 2011, PMID: 21982591; and O’Hagan et al., 2017, PMID: 29129530) to detect the polyglutamylation signals in the sensory cilia of C. elegans. We initially tried to stain the whole animals using these antibodies but could not get clear and distinct signals in the TRNs. We reason that the tubulin polyglutamylation signals in the TRNs may be weak, and the in situ staining method which requires the antibodies to penetrate multiple layers of tissues (e.g., cuticles and epidermis) to reach the TRN axons may be not sensitive enough to detect the signal. In fact, the TRN axons are located deeper in the worm body compared to the sensory cilia that are mostly exposed to the environment. Another reason could be that the tissues (mostly epidermis) surrounding the TRN axons also have polyglutamylation staining, which makes it difficult to recognize TRN axons. This is a situation different from the anti-K40 acetylation staining, which only occurs in the TRNs because MEC-12 is the only a-tubulin isotype that carries K40. Due to these technical difficulties, we decided to use the in vitro cultured TRNs for the staining experiment, which allows both easy access of the antibodies (thus higher sensitivity) and the dissociation of the TRNs from other tissues. The fact that we were able to observe reduced staining in the ttll mutants and the tubulin mutants that lost the glutamate residues suggest that these antibodies indeed detected glutamylation signals in the cells.

(3) The final paragraph of the discussion is factually incorrect. The C. elegans homologs of the CCP carboxypeptidases are called CCPP-1 and CCPP-6. There are several publications on their functions in C. elegans.

We thank the reviewer for pointing out the mistake in the text. We intended to say that “there is no C. elegans homolog of the known tubulin carboxypeptidases that catalyze detyrosination”, which is true given that the detyrosinase vasohibins (VASH1/VASH2) homologs cannot be found in C. elegans. We are aware of the publications on CCPP-1 and CCPP-6; CCPP-1 is known to regulate tubulin deglutamylation in the cilia of C. elegans (O’Hagan et al., 2011 and 2017), while CCPP-6 may function in the PLM to regulate axonal regeneration (Ghosh-Roy et al., 2012). In the revised manuscript, we have corrected the error.

Reviewer #2 (Public Review):

Summary:

The tubulin subunits that make up microtubules can be posttranslationally modified and these PTMs are proposed to regulate microtubule dynamics and the proteins that can interact with microtubules in many contexts. However, most studies investigating the roles of tubulin PTMs have been conducted in vitro either with purified components or in cultured cells. Lu et al. use CRISPR/Cas9 genome editing to mutate tubulin genes in C. elegans, testing the role of specific tubulin residues on neuronal development. This study is a real tour de force, tackling multiple proposed tubulin modifications and following the resulting phenotypes with respect to neurite outgrowth in vivo. There is a ton of data that experts in the field will likely reference for years to come as this is one of the most comprehensive in vivo analyses of tubulin PTMs in vivo.

This paper will be very important to the field, however would be strengthened if: 1) the authors demonstrated that the mutations they introduced had the intended consequences on microtubule PTMs, 2) the authors explored how the various tubulin mutations directly affect microtubules, and 3) the findings are made generally more accessible to non C. elegans neurobiologists.

(1) The authors introduce several mutations to perturb tubulin PTMs, However, it is unclear to what extent the engineered mutations affect tubulin in the intended way i.e. are the authors sure that the PTMs they want to perturb are actually present in C. elegans. Many of the antibodies used did not appear to be specific and antibody staining was not always impacted in the mutant cases as expected. For example, is there any evidence that S172 is phosphorylated in C. elegans, e.g. from available phosphor-proteomic data? Given the significant amount of staining left in the S172A mutant, the antibody seems non-specific in this context and therefore not a reliable readout of whether MTs are actually phosphorylated at this residue. As another example, there is no evidence presented that K252 is acetylated in C. elegans. At the very least, the authors should consider demonstrating the conservation of these residues and the surrounding residues with other organisms where studies have demonstrated PTMs exist. 

We thank the reviewer for the comments. To our knowledge, there are very few phosphor-proteome data available for C. elegans. We searched a previously published dataset (Zielinska et al., 2009; PMID: 19530675) and did not find the S172 phosphorylation signal in MEC-7. This is not surprising, given that only six touch receptor neurons expressed MEC-7 and the abundance of MEC-7 in the whole animal lysate may be below the detection limit. However, this phosphorylation site S172 is highly conserved across species and tubulin isotypes (Figure 1-figure supplement 1 in the revised manuscript), suggesting that this site is likely phosphorylated in MEC-7.

In the case of K252, the potential acetylation site and the flanking sequences are extremely conserved across species and isotypes. In fact, the 20 amino acids from 241-260 a.a. are identical among the tubulin genes of C. elegans, fruit flies, Xenopus, and humans (Figure 4-figure supplement 1B). Thus, although K252 acetylation was found in the HeLa cells, this site can possibly be acetylated. 

In the case of K40, we observed sequence divergence at the PTM site and adjacent sequences among the tubulin isotypes in C. elegans. MEC-12 is the only C. elegans a-tubulin isotype that has the K40 residue, and the 40-50 a.a. region of MEC-12 appears to be more conserved than other isotypes when compared to Drosophila, frog, and human a-tubulins (Figure 4-figure supplement 1A).

(2) Given that the authors have the mutants in hand, it would be incredibly valuable to assess the impact of these mutations on microtubules directly in all cases. MT phenotypes are inferred from neurite outgrowth phenotypes in several cases, the authors should look directly at microtubules and/or microtubule dynamics via EBP-2 when possible OR show evidence that the only way to derive the neurite phenotypes shown is through the inferred microtubule phenotypes. For example, the effect of the acetylation or detyrosination mutants on MTs was not assessed. 

We thank the reviewer for the suggestions. In this study, we created >20 tubulin mutants. Due to limited time and resources, we were not able to examine microtubule dynamics in every mutant strain using EBP-2 kymographs. We assessed the effects of the tubulin mutations mostly based on the changes on neurite growth pattern. From our previous experience of analyzing ~100 mec-7 and mec-12 missense mutations (Zheng et al., 2017, MBoC; Lee et al., 2021, MBoC), we found that the changes in microtubule dynamics are correlated with the changes in neuronal morphologies. For example, the growth of ectopic ALM-PN is correlated with fewer EBP-2 comets and potentially reduced microtubule dynamics; this correlation holds true for several mec-7 neomorphic missense alleles we examined before (Lee et al., 2021, MBoC) and the PTM site mutants [e.g., mec-7(S172A) and mec-12(4Es-A)] analyzed in this study. Similarly, the shortening of TRN neurites is correlated with more EBP-2 comets and increased microtubule dynamics. For the mutants that don’t show neurite growth defects, our previous experience is that they are not likely to show altered microtubule dynamics in EBP-2 tracking experiments. So, we did not analyze the acetylation mutants (which had no defects in neurite growth) and the detyrosination mutants (which had weak ALM-PN phenotype). Nevertheless, we agree with the reviewer that we could not rule out the possibility that there may be some slight changes to microtubule dynamics in these mutants.

Using tannic acid staining and electron microscopy (EM), we previously examined the microtubule structure in several tubulin missense mutants (Zheng et al., 2017, MBoC) and found that the loss-of-function and antimorphic mutations significantly reduced the number of microtubules and altered microtubule organizations by reducing protofilament numbers. These structural changes are consistent with highly unstable microtubules and defects in neurite growth. On the other hand, neomorphic mutants had only slight decrease in microtubule abundance, maintained the 15-protofilament structure, and had a more tightly packed microtubule bundles that filled up most of the space in the TRN neurite (Zheng et al., 2017, MBoC). These structural features are consistent with increased microtubule stability and ectopic neurite growth. Although we did not directly examine the microtubule abundance and structure using EM in this study, we would expect similar changes that are correlated with the neurite growth phenotypes in the PTM mutants. We agree with the reviewer, it will be informative to conduct more comprehensive analysis on these mutants using EM and other structural biology methods.

(3) There is a ton of data here that will be important for experts working in this field to dig into, however, for the more general cell biologist, some of the data are quite inaccessible. More cartoons and better labeling will be helpful as will consistent comparisons to control worms in each experiment.

Response: We thank the reviewer for the comment. In the revised manuscript, we added some cartoons to Figure 2G to show the location of the synaptic vesicles. The neurite growth phenotype should be quite straightforward. Nevertheless, we added one more Figure (Figure 8) to summarize all the results in the study with cartoons that depicted the changes to neuronal morphologies.

(4) In addition, I am left unconvinced of the negative data demonstrating that MBK does not phosphorylate tubulin. First, the data described in lines 207-211 does not appear to be presented anywhere. Second, RNAi is notoriously finicky in neurons, thus necessitating tissue-specific degradation using either the ZF/ZIF-1 or AID/TIR1 systems which both work extremely well in C. elegans. Third, there appears to be increasing S172 phosphorylation in Figure 3 Supplement 2 with added MBK-2, but there is no anti-tubulin blot to show equal loading, so this experiment is hard to interpret.

We added the results of mbk-1, mbk-2, and hpk-1 mutants and cell-specific knockdown of MBK-2 into Figure 3-figure supplement 1D. Considering the reviewer’s suggestion, we attempted to use a ZIF-1 system to remove the MBK-2 proteins specifically in the TRNs using a previously published method (PMID: 28619826). We fused endogenous MBK-2 with GFP by gene editing and then expressed an anti-GFP nanobodies fused with ZIF-1 in the TRNs to induce the degradation of MBK-2::GFP. To our surprise, unlike the mbk-2p::GFP transcriptional reporter, the MBK-2::GFP did not show detectable expression in the TRNs, although expression can be seen in early embryos, which is consistent with the “embryonic lethal” phenotype of the mbk-2(-) mutants (Figure 3-figure supplement 2A-B in the revised manuscript). We reason that either endogenous MBK-2 is not expressed in the TRNs or is expressed at a very low level. We then crossed mbk-2::GFP with ItSi953 [mec-18p::vhhGFP4::Zif-1] to trigger the degradation of any potential MBK-2 proteins and did not observe the ectopic growth of ALM-PN (Figure 3- figure supplement 2C). These results suggest that MBK-2 is not likely to regulate tubulin phosphorylation in the TRNs, which is consistent with the results of other genetic mutants and the RNAi experiments.

For Figure 3 Supplement 2 (Figure 3-figuer supplement 3 in revised manuscript), because we added the same amount of purified MEC-12/MEC-7 to all reactions and had established equal loading in Figure 3E, we did not do the anti-tubulin staining in this experiment. Since higher concentration (1742 nM) of MBK-2 did not produce stronger signal than the condition with 1268 nM, we don’t think the 1268 nM band represents true phosphorylation. Moreover, the signal is not significantly stronger than the control without MBK-2 and is much lower than the signal generated by CDK1 in Figure 3E. Based on these results, we concluded that MBK-2 is not likely to phosphorylate MEC-7.  

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

General:

A summary table would help the reader digest the vast amount of phenotypic data.

Cartoons to help a non-C. elegans reader understand the figures. 

We added Figure 8 to summarize and illustrate the effects of the various mutants analyzed in this study.

Specific:

The authors engineered mutations into the predicted phosphorylation site of b-tubulin mec-7. These CRISPR-alleles mutations phenocopied previously identified loss-of-function, gain-of-function, and neomorphic mec-7 alleles identified in genetic screens by the Chalfie lab. Next, the authors sought to identify the responsible kinase, taking a candidate gene approach. The most likely family - minibrain - had no effect when knocked down/out. The authors showed that cdk-1 mutants displayed ectopic ALM-PN outgrowth. Whether cdk-1 specifically acts in the TRNs was not demonstrated, calling into question whether CDK-1 phosphorylates S172 in vivo. In their introduction (lines 45-59), the authors built a case for engineering PTM mutations directly into tubulins, because the PTM enzymes may have multiple substrates. This logic applies to the cdk-1 experiment and its interpretation. 

The reviewer is right. Since CDK1 and minibrain kinase are the only known kinases that catalyze S172 phosphorylation, our results suggest that CDK-1 is more likely to catalyze S172 phosphorylation in the TRNs compared to MBK-1/2. Genetic studies found that cdk-1(-); mec-7(S172A) double mutants did not show stronger phenotype than the two single mutants, suggesting that they function in the same pathway. Nevertheless, we could not rule out the possibility that other kinases may also control S172 phosphorylation, and the effect of CDK-1 is indirect. We mentioned this possibility in the revised manuscript.

For a-tubulin MEC-12, acetyl-mimicking K40Q and unmodifiable K40R mutants failed to stain with the anti-acetyl-a-tubulin (K40) antibody and displayed subtle TRN phenotypes. The enzymatically dead MEC-17 had phenotypes similar to those described by Topalidou (2012), confirming the Chalfie lab finding that MEC-17 has functions in addition and independent of its acetyltransferase activity. The authors moved onto a predicted acetylation site in MEC-7 and observed TRN developmental defects, and acknowledged that this may be due to tubulin instability and not a PTM. This is a concern for all mutants, as there is no way to measure whether the protein is expressed, stable, or localized properly. 

We acknowledge that this is a caveat of mutational studies. An amino acid substitution at the PTM site may have multiple effects, including the change of the PTM state and potential alteration of protein conformation. Without direct evidence for enzymatic modification of the PTM site in the neurons, we could not rule out the possibility the phenotype we observed is not related to PTM and instead is the result of abnormal protein conformation and function caused by the mutation.

Nevertheless, as stated in our above response to the first point in the public review, we can phenotypically differentiate loss-of-function and gain-of-function mutants. If the mutation reduces expression or general protein stability, it is more likely to cause a loss-of-function phenotype. For most PTM site mutants, this is not the case. We observed mostly gain-of-function phenotype, suggesting that the missense mutations did not simply inactivate the tubulin protein and instead affected the functional properties of the protein.

From here, the authors manipulate the C-terminal tail of MEC-12 and MEC-7, testing the idea that polyglutamylation may be an important PTM. These mutants displayed subtle phenotypes. The authors show that branch point GT335 and polyglutamyation polyE recognizing antibodies stain cultured embryonic TRNs, but did not examine staining in TRNs. To my knowledge, these antibodies have not been shown to stain the TRNs in any published papers (see next point). The rationale for using cultured embryonic TRNs is not clear. 

See our response to the second point in the public review.

Lines 548-553 There are several publications on CCPP-1 and CCPP-6 functions in TRNs and ciliated sensory neurons. See

PMID: 20519502

PMID: 21982591

PMID: 21943602

PMID: 23000142

PMID: 29129530

PMID: 33064774

PMID: 36285326

PMID: 37287505 

We thank the reviewer for pointing out these references, some of which were cited in the revised manuscript. We made a mistake in the Discussion by saying that there are no C. elegans homologs of tubulin carboxypeptidases while we intended to state that there is no homolog of tubulin detyrosinase in C. elegans. We are aware of the studies of CCPP-1 and CCPP-6 and have corrected the mistake in revised manuscript (also see our response to the third point in the public review).

Reviewer #2 (Recommendations For The Authors):

Figures: 

As stated in the public review, more cartoons and better labeling will be helpful as will consistent comparisons to control worms in each experiment. A good example of this issue is demonstrated in Figure 2 and Figure 4: 

(1) Figure 2: Please label images with what is being probed in each panel. 

We added labels to the panels.

(2) Figure 2G is very hard to interpret - cartoon diagramming what is being observed would be helpful. 

We added cartoons to help illustrate the images.

(3) Line 182-185: is this referring to your data or to Wu et al? It is not clear in this paragraph when the authors are describing published work versus their own data presented here. 

It is from our data. We have made it clear in the revised manuscript.

(4) Figure 2 - 2K is not well described. What experiment is being done here? What is dlk-1 and why did you look at this mutant? 

Figure 2K showed that both wild-type animals and S172A mutants could reconnect the severed axons after laser axotomy. Previous studies have found that dlk-1(-) mutants were not able to regenerate axons due to altered microtubule dynamics (PMID: 19737525; PMID: 23000142). We used dlk-1(-) mutants as a negative control, because DLK-1 promotes microtubule growth following axotomy, and the DLK-1 pathway is essential for regeneration (PMID: 23000142). We want to highlight the phenotypic difference between dlk-1(-) mutants and the S172E mutants. Although both mutants showed similar regrowth length, dlk-1(-) mutants showed unbranched regrowth probably due to the lack of microtubule polymerization, whereas the S172E mutants showed a mesh-like regrowth pattern likely due to highly dynamic and unstable microtubules. We explained the different phenotypes in the revised manuscript.

(5) Figure 4C: this phenotype is hard to interpret. Where is the wt control? Where is the quantification? 

In the Figure legend, we have referred the readers to Figure 1G for the wild-type image. Quantification is provided in the text (~20% of the animals showed the branching defects).

(6) There are no WT comparison images in Figure 4I, making the quantification difficult to interpret 

In the Figure legend, we have referred the readers to Figure 1A for the wild-type control. Moreover, we included a new Figure 8 to summarize the phenotypes of all mutants.

Experimental:

(1) Is it clear that only MEC-7/MEC-12 are the only a- and b-tubulin present in the TRNs? The presence of other tubulins not mutated would complicate the interpretation of the results. 

According to the mRNA levels, the expression of MEC-7 and MEC-12 are >100 fold higher than other tubulin isotypes. For example, single-cell transcriptomic data (Taylor et al., 2021) showed that mec-7 mRNA is at 135,940 TPM in ALM neurons, whereas two other tubulin isotypes, tbb-1 and tbb-2, have expression value of 54 and 554 TPM, respectively in the ALM. So, even if there are some other tubulin isotypes, their abundance is much lower than mec-7 and mec-12 and are not likely to interfere with the effects of the mec-7 and mec-12 mutants.

(2) The in vitro kinase assays should be quantified. 

We have added the quantification.

(3) The idea that Cdk1 phosphorylates tubulin in interphase is surprising and I am left wondering how the authors propose that Cdk1 is activated in interphase. Is cyclin B (or another cyclin) present in interphase in this cell type? Expression but not activation of Cdk1 is not discussed. 

CDK1 can work with cyclin A and cyclin B. C. elegans has one cyclin A gene (cya-1) and four cyclin B genes (cyb-1, cyb-2.1, cyb-2.2, and cyb-3). According to single-cell transcriptomic data of L4 animals, cya-1 and cyb-1 showed weak expression in many postmitotic neurons (including the ALM neurons), while cyb-2.1, cyb-2.2, and cyb-3 had no expression in neurons. So, it is possible that cya-1/cyclin A and cyb-1/cyclin B has low level of expression in the TRNs. A previous study also found the expression of cell cycle regulators (including cyclins) in postmitotic neurons in mouse brain (Akagawa et al., 2021; PMID: 34746147).

(4) What is the significance of neurite swelling and looping in Figure 4H? The underlying cause of this phenotype is not described. 

The neurite swelling and looping phenotype of mec-17(-) mutants were described by Topalidou et al., (2012; PMID: 22658602) and were caused by the bending of the microtubules. It appears that the loss of the a-tubulin acetyltransferase altered the organization of microtubules in the TRNs. These defects were partially rescued by the enzymatically dead MEC-17, suggesting that MEC-17 may play a non-enzymatic (and likely structural) role in regulating microtubule organization. We added more explanation in the revised manuscript.

(5) It is quite surprising that polyglutamylation is not affected in the quintuple ttll mutant. Since the authors made the sextuple ttll mutant, could they demonstrate whether polyglutamylation is further reduced in this mutant via GT335 staining? 

We did not make the comparison of the quintuple and sextuple ttll mutants because they were crossed with TRN markers with different colors for technical reasons. The quintuple mutants CGZ1475 carried uIs115 [mec-17p::TagRFP] IV, whereas the sextuple mutants CGZ1474 carried zdIs5 [mec-4p::GFP] I. As a result, we need to use different secondary antibodies for the antibody staining, which makes the results not compatible.

Polyglutmaylation signal in the cell body was strongly affected by the ttll mutations. In fact, in the ttll-4(-); ttl-5(-); ttll-12(-) triple mutants, the signal is significantly reduced in the cell body of the TRNs, as well as the cell body of other cells. What’s surprising is that the signal in the axons persisted in the ttll triple and quintuple mutants. As the reviewers suggested, we also stained the sextuple mutants and found similar pattern as the triple and quintuple mutants (new Figure 6-figure supplement 1C in the revised manuscript), although the results are not quantitatively comparable due to the use of secondary antibodies with different fluorophores.

Writing:

(1) The beginning of the results section is quite jarring. The information in lines 96-104 should be in the Introduction. 

Due to the nature of this paper, each section deals with a particular PTM. We think it is helpful to discuss some background information before describing our results on each PTM rather than giving all in the introduction. Nevertheless, we modified the beginning of the results to make it more coherent and more connected with the preceding paragraphs.

(2) Line 122-126: conclusions are not supported by the data: it is suggested from previous experiments, but authors do not look at MTs directly. 

We have rephrased the statement to acknowledge that we made such conclusion based on phenotypic similarity with mutants we previously examined.

(3) I am confused by the usage of both mec-12(4EtoA) and mec-12(4Es-A). Are these the same mutations? If so, there needs to be consistency. If not, each case needs to be defined. 

They are the same. We have corrected the mistake and are now using mec-12(4Es-A) to refer to the mutants.

Line 105: phosphor --> phospho 

Line 187: were --> was 

Line 298: is --> are

The above typos are corrected.

Author response:

Reviewer #1 (Public Review):

Summary and Strengths:

The ability of Wolbachia to be transmitted horizontally during parasitoid wasp infections is supported by phylogenetic data here and elsewhere. Experimental analyses have shown evidence of wasp-to-wasp transmission during coinfection (eg Huigins et al), host to wasp transmission (eg Heath et al), and mechanical ('dirty needle') transmission from host to host (Ahmed et al). To my knowledge this manuscript provides the first experimental evidence of wasp to host transmission. Given the strong phylogenetic pattern of host-parasitoid Wolbachia sharing, this may be of general importance in explaining the distribution of Wolbachia across arthropods. This is of interest as Wolbachia is extremely common in the natural world and influences many aspects of host biology.

Weaknesses:

The first observation of the manuscript is that the Wolbachia strains in hosts are more closely related to those in their parasitoids. This has been reported on multiple occasions before, dating back to the late 1990s. The introduction cites five such papers (the observation is made in other studies too that could be cited) but then dismisses them by stating "However, without quantitative tests, this observation could simply reflect a bias in research focus." As these studies include carefully collected datasets that were analysed appropriately, I felt this claim of novelty was rather strong. It is unclear why downloading every sequence in GenBank avoids any perceived biases, when presumably the authors are reanalysing the data in these papers.

Thank you for bringing this to our attention, and we will make the necessary amendments in our revised manuscript.

I do not doubt the observation that host-parasitoid pairs tend to share related Wolbachia, as it is corroborated by other studies, the effect size is large, and the case study of whitefly is clearcut. It is also novel to do this analysis on such a large dataset. However, the statistical analysis used is incorrect as the observations are pseudo-replicated due to phylogenetic non-independence. When analysing comparative data like this it is essential to correct for the confounding effects of related species tending to be similar due to common ancestry. In this case, it is well-known that this is an issue as it is a repeated observation that related hosts are infected by related Wolbachia. However, the authors treat every pairwise combination of species (nearly a million pairs) as an independent observation. Addressing this issue is made more complex because there are both the host and symbiont trees to consider. The additional analysis in lines 123-124 (including shuffling species pairs) does not explicitly address this issue.

We concur with your observation regarding the non-independence of the data due to phylogenetic relationships. While common phylogenetic correction methods are indeed not directly applicable to wsp distances between species pairs, we are investigating the potential of phylogenetic mixed models to address this issue. We hope to include a revised analysis using this approach in our revised manuscript.

The sharing of Wolbachia between whitefly and their parasitoids is very striking, although this has been reported before (eg the authors recently published a paper entitled "Diversity and Phylogenetic Analyses Reveal Horizontal Transmission of Endosymbionts Between Whiteflies and Their Parasitoids"). In Lines 154-164 it is suggested that from the tree the direction of transfer between host and parasitoid can be inferred from the data. This is not obvious to me given the poor resolution of the tree due to low sequence divergence. There are established statistical approaches to test the direction of trait changes on a tree that could have been used (a common approach is to use the software BEAST).

Thank you for your insightful comments regarding the transfer direction of Wolbachia between whiteflies and their parasitoids. We acknowledge the concern about the resolution of the phylogenetic tree and the inference of the direction of Wolbachia transmission based on the available data. We considered the high infection frequency and obligate nature of Wolbachia in En. formosa, which exhibits a 100% infection rate, as a strong indicator that recent transmission of Wolbachia in this clade likely occurred from En. formosa to B. tabaci. We appreciate your recommendation and will ensure that our conclusions are supported by a more statistically sound approach. As you suggested, we will employ the software BEAST to rigorously test the direction of transmission, and we will revise our statements accordingly.

Reviewer #2 (Public Review):

The paper by Yan et al. aims to provide evidence for horizontal transmission of the intracellular bacterial symbiont Wolbachia from parasitoid wasps to their whitefly hosts. In my opinion, the paper in its current form consists of major flaws.

Weaknesses:

The dogma in the field is that although horizontal transmission events of Wolbachia occur, in most systems they are so rare that the chances of observing them in the lab are very slim.

For the idea of bacteria moving from a parasitoid to its host, the authors have rightfully cited the paper by Hughes, et al. (2001), which presents the main arguments against the possibility of documenting such transmissions. Thus, if the authors want to provide data that contradict the large volume of evidence showing the opposite, they should present a very strong case.

In my opinion, the paper fails to provide such concrete evidence. Moreover, it seems the work presented does not meet the basic scientific standards.

We are grateful for your critical perspective on our work. Nonetheless, we are confident in the credibility of our findings regarding the horizontal transmission of Wolbachia from En. formosa to B. tabaci. Our study has documented this phenomenon through phylogenetic tree analyses, and we have further substantiated our observations with rigorous experiments in both cages and petri dishes. The horizontal transfer of Wolbachia was confirmed via PCR, with the wsp sequences in B. tabaci showing complete concordance with those in En. formosa. Additionally, we utilized FISH, vertical transmission experiments, and phenotypic assays to demonstrate that the transferred Wolbachia could be vertically transmitted and induce significant fitness cost in B. tabaci. All experiments were conducted with strict negative controls and a sufficient number of replicates to ensure reliability, thereby meeting basic scientific standards. The collective evidence we present points to a definitive case of Wolbachia transmission from the parasitoid En. formosa to the whitefly B. tabaci.

My main reservations are:

• I think the distribution pattern of bacteria stained by the probes in the FISH pictures presented in Figure 4 looks very much like Portiera, the primary symbiont found in the bacterium of all whitefly species. In order to make a strong case, the authors need to include Portiera probes along with the Wolbachia ones.

We are very grateful for your critical evaluation regarding the specificity of FISH in our study. We assure the reliability of our FISH results based on several reasons.

1) We implemented rigorous negative controls which exhibited no detectable signal, thereby affirming the specificity of our hybridization. 2) The central region of the whitefly nymphs is a typical oviposition site for En. formosa. Post-parasitism, we observed FISH signals around the introduced parasitoid eggs, distinct from bacteriocyte cells which are rich in endosymbionts including Portiera (FIG 3e-f). This observation supports the high specificity of our FISH method. 3) In the G3 whiteflies, we detected the presence of Wolbachia in bacteriocytes in nymphs and at the posterior end of eggs in adult females (FIG 4). This distribution pattern aligns with previously reported localizations of Wolbachia in B. tabaci (Shi et al., 2016; Skaljac et al., 2013). Furthermore, the distribution of Wolbachia in the whiteflies does indeed exhibit some overlap with that of Portiera (Skaljac et al., 2013; Bing et al., 2014). 4) The primers used in our FISH assays have been widely cited (Heddi et al., 1999) and validated in studies on B. tabaci and other systems (Guo et al., 2018; Hegde et al., 2024; Krafsur et al., 2020; Rasgon et al., 2006; Uribe-Alvarez et al., 2019; Zhao et al., 2013). Taking all these points into consideration, we stand by the reliability of our FISH results.

References:

Bing XL, Xia WQ, Gui JD, Yan GH, Wang XW, Liu SS. 2014. Diversity and evolution of the Wolbachia endosymbionts of Bemisia (Hemiptera: Aleyrodidae) whiteflies. Ecol Evol, 4(13): 2714-37.

Guo, Y, Hoffmann, AA, Xu, XQ, Zhang X, Huang HJ, Ju JF, Gong JT, Hong XY. 2018. Wolbachia-induced apoptosis associated with increased fecundity in Laodelphax striatellus (Hemiptera: Delphacidae). Insect Mol Biol, 27: 796-807.

Heddi A, Grenier AM, Khatchadourian C, Charles H, Nardon P. 1999. Four intracellular genomes direct weevil biology: Nuclear, mitochondrial, principal endosymbiont, and Wolbachia. Proc Natl Acad Sci USA, 96: 6814-6819.

Hegde S, Marriott AE, Pionnier N, Steven A, Bulman C, Gunderson E, et al. 2024. Combinations of the azaquinazoline anti-Wolbachia agent, AWZ1066S, with benzimidazole anthelmintics synergise to mediate sub-seven-day sterilising and curative efficacies in experimental models of filariasis. Front Microbiol, 15: 1346068.

Krafsur AM, Ghosh A, Brelsfoard CL. 2020. Phenotypic response of Wolbachia pipientis in a cell-free medium. Microorganisms, 8: 1060.

Rasgon JL, Gamston, CE, Ren X. 2006. Survival of Wolbachia pipientis in cell-free medium. Appl Environ Microbiol, 72: 6934-6937.

Shi P, He Z, Li S, An X, Lv N, Ghanim M, Cuthbertson AGS, Ren SX, Qiu BL. 2016. Wolbachia has two different localization patterns in whitefly Bemisia tabaci AsiaII7 species. PLoS One, 11: e0162558.

Skaljac M, Zanić K, Hrnčić S, Radonjić S, Perović T, Ghanim M. 2013. Diversity and localization of bacterial symbionts in three whitefly species (Hemiptera: Aleyrodidae) from the east coast of the Adriatic Sea. Bull Entomol Res, 103(1): 48-59.

Uribe-Alvarez C, Chiquete-Félix N, Morales-García L, Bohórquez-Hernández A, Delgado-Buenrostro N L, Vaca L, et al. 2019. Wolbachia pipientis grows in Saccharomyces cerevisiae evoking early death of the host and deregulation of mitochondrial metabolism. MicrobiologyOpen, 8: e00675.

Zhao DX, Zhang XF, Chen DS, Zhang YK, Hong XY, 2013. Wolbachia-host interactions: Host mating patterns affect Wolbachia density dynamics. PLoS One, 8: e66373.

• If I understand the methods correctly, the phylogeny presented in Figure 2a is supposed to be based on a wide search for Wolbachia wsp gene done on the NCBI dataset (p. 348). However, when I checked the origin of some of the sequences used in the tree to show the similarity of Wolbachia between Bemisia tabaci and its parasitoids, I found that most of them were deposited by the authors themselves in the course of the current study (I could not find this mentioned in the text), or originated in a couple of papers that in my opinion should not have been published to begin with.

We appreciate your meticulous examination of the sources for our sequence data. All the sequences included in our phylogenetic analysis were indeed downloaded from the NCBI database as of July 2023. The sequences used to illustrate the similarity of Wolbachia between B. tabaci and its parasitoids include those from our previously published study (Qi et al., 2019), which were sequenced from field samples. Additionally, some sequences were also obtained from other laboratories (Ahmed et al., 2009; Baldo et al., 2006; Van Meer et al., 1999). We acknowledge that in our prior research (Qi et al., 2019), the sequences were directly submitted to NCBI and, regrettably, we did not update the corresponding publication information after the article were published. It is not uncommon for sequences on NCBI, with some never being followed by a published paper (e.g., FJ710487- FJ710511 and JF426137-JF426149), or not having their associated publication details updated post-publication (for instance, sequences MH918776-MH918794 from Qi et al., 2019, and KF017873-KF017878 from Fattah-Hosseini et al., 2018). We recognize that this practice can lead to confusion and apologize for the oversight in our work.

References:

Ahmed MZ, Shatters RG, Ren, SX, Jin GH, Mandour NS, Qiu BL. 2009. Genetic distinctions among the Mediterranean and Chinese populations of Bemisia tabaci Q biotype and their endosymbiont Wolbachia populations. J Appl Entomol, 133: 733-741.

Baldo L, Hotopp JCD, Jolley KA, Bordenstein SR, Biber SA, Choudhury RR, et al. 2006. Multilocus sequence typing system for the endosymbiont Wolbachia pipientis. Appl Environ Microbiol, 72: 7098-110.

Fattah-Hosseini S, Karimi J, Allahyari H. 2014. Molecular characterization of Iranian Encarsia formosa Gahan populations with natural incidence of Wolbachia infection. J Entomol Res Soc, 20: 85–100.

Qi LD, Sun JT, Hong XY, Li YX. 2019. Diversity and phylogenetic analyses reveal horizontal transmission of endosymbionts between whiteflies and their parasitoids. J Econ Entomol, 112(2): 894-905.

Van Meer MM, Witteveldt J, Stouthamer R. 1999. Phylogeny of the arthropod endosymbiont Wolbachia based on the wsp gene. Insect Mol Biol, 8: 399-408.

• The authors fail to discuss or even acknowledge a number of published studies that specifically show no horizontal transmission, such as the one claimed to be detected in the study presented.

Thank you for bringing this to our attention. We will address and discuss the published studies that report no evidence of horizontal transmission, as you've highlighted, in the revised version of our manuscript.

Reviewer #3 (Public Review):

This is a very ordinary research paper. The horizontal of endosymbionts, including Wolbachia, Rickettsia etc. has been reported in detail in the last 10 years, and parasitoid vectored as well as plant vectored horizontal transmission is the mainstream of research. For example, Ahmed et al. 2013 PLoS One, 2015 PLoS Pathogens, Chiel et al. 2014 Enviromental Entomology, Ahmed et al. 2016 BMC Evolution Biology, Qi et al. 2019 JEE, Liu et al. 2023 Frontiers in Cellular and Infection Microbiology, all of these reported the parasitoid vectored horizontal transmission of endosymbiont. While Caspi-Fluger et al. 2012 Proc Roy Soc B, Chrostek et al. 2017 Frontiers in Microbiology, Li et al. 2017 ISME Journal, Li et al. 2017 FEMS, Shi et al. 2024 mBio, all of these reported the plant vectored horizontal transmission of endosymbiont. For the effects of endosymbiont on the biology of the host, Ahmed et al. 2015 PLoS Pathogens explained the effects in detail.

Thank you very much for your insightful comments and for highlighting the relevant literature in the field of horizontal transmission of endosymbionts, including Wolbachia and Rickettsia. After careful consideration of the studies you have mentioned, we believe that our work presents significant novel contributions to the field. 1) Regarding the parasitoid-mediated horizontal transmission of Wolbachia, most of the cited articles, such as Ahmed et al. 2013 in PLoS One and Ahmed et al. 2016 in BMC Evolutionary Biology, propose hypotheses but do not provide definitive evidence. The transmission of Wolbachia within the whitefly cryptic species complex (Ahmed et al. 2013) or between moths and butterflies (Ahmed et al. 2016) could be mediated by parasitoids, plants, or other unknown pathways. 2) Chiel et al. (2014 in Environmental Entomology reported “no evidence for horizontal transmission of Wolbachia between and within trophic levels” in their study system. 3) The literature you mentioned about Rickettsia, rather than Wolbachia, indirectly reflects the relative scarcity of evidence for Wolbachia horizontal transmission. For example, the evidence for plant-mediated transmission of Wolbachia remains isolated, with Li et al. 2017 in The ISME Journal being one of the few reports supporting this mode of transmission. 4) While the effects of endosymbionts on their hosts are not the central focus of our study, the effects of transgenerational Wolbachia on whiteflies are primarily demonstrated to confirm the infection of Wolbachia into whiteflies. Furthermore, the effects we report of Wolbachia on whiteflies are notably different from those reported by Ahmed et al. 2015 in PLoS Pathogens, likely due to different whitefly species and Wolbachia strains. 6) More importantly, our study reveals a mechanism of parasitoid-mediated horizontal transmission of Wolbachia that is distinct from the mechanical transmission suggested by Ahmed et al. 2015 in PLoS Pathogens. Their study implies transmission primarily through host-feeding contamination, without the need for Wolbachia to infect the parasitoid, suggesting host-to-host transmission at the same trophic level. In contrast, our findings demonstrate transmission from parasitoids to hosts through unsuccessful parasitism, which represents cross-trophic level transmission. To our knowledge, this is the first experimental evidence that Wolbachia can be transmitted from parasitoids to hosts. We believe these clarifications and the novel insights provided by our research contribute valuable knowledge to the field.

References:

Ahmed MZ, De Barro PJ, Ren SX, Greeff JM, Qiu BL. 2013. Evidence for horizontal transmission of secondary endosymbionts in the Bemisia tabaci cryptic species complex. PLoS One, 8: e53084.

Ahmed MZ, Li SJ, Xue X, Yin XJ, Ren SX, Jiggins FM, Greeff JM, Qiu BL. 2015. The intracellular bacterium Wolbachia uses parasitoid wasps as phoretic vectors for efficient horizontal transmission. PLoS Pathog, 10: e1004672.

Ahmed MZ, Breinholt JW, Kawahara AY. 2016. Evidence for common horizontal transmission of Wolbachia among butterflies and moths. BMC Evol Biol, 16: 118. doi.org/10.1186/s12862-016-0660-x.

Caspi-Fluger A, Inbar M, Mozes-Daube N, Katzir N, Portnoy V, Belausov E, Hunter MS, Zchori-Fein E. 2012. Horizontal transmission of the insect symbiont Rickettsia is plant-mediated. Proc Biol Sci, 279(1734): 1791-6.

Chiel E, Kelly SE, Harris AM, Gebiola M, Li X, Zchori-Fein E, Hunter MS. 2014. Characteristics, phenotype, and transmission of Wolbachia in the sweet potato whitefly, Bemisia tabaci (Hemiptera: Aleyrodidae), and its parasitoid Eretmocerus sp. nr. emiratus (Hymenoptera: Aphelinidae). Environ Entomol, 43(2): 353-62.

Chrostek E, Pelz-Stelinski K, Hurst GDD, Hughes GL. 2017. Horizontal transmission of intracellular insect symbionts via plants. Front Microbiol, 8: 2237.

Li SJ, Ahmed MZ, Lv N, Shi PQ, Wang XM, Huang JL, Qiu BL. 2017. Plantmediated horizontal transmission of Wolbachia between whiteflies. ISME J, 11: 1019-1028.

Li YH, Ahmed MZ, Li SJ, Lv N, Shi PQ, Chen XS, Qiu BL. 2017. Plant-mediated horizontal transmission of Rickettsia endosymbiont between different whitefly species. FEMS Microbiol Ecol, 93(12). doi: 10.1093/femsec/fix138.

Liu Y, He ZQ, Wen Q, Peng J, Zhou YT, Mandour N, McKenzie CL, Ahmed MZ, Qiu BL. 2023. Parasitoid-mediated horizontal transmission of Rickettsia between whiteflies. Front Cell Infect Microbiol, 12: 1077494. DOI: 10.3389/fcimb.2022.1077494

Qi LD, Sun JT, Hong XY, Li YX. 2019. Diversity and phylogenetic analyses reveal horizontal transmission of endosymbionts between whiteflies and their parasitoids. J Econ Entomol, 112: 894-905.

Shi PQ, Wang L, Chen XY, Wang K, Wu QJ, Turlings TCJ, Zhang PJ, Qiu BL. 2024. Rickettsia transmission from whitefly to plants benefits herbivore insects but is detrimental to fungal and viral pathogens. mBio, 15(3): e0244823.

Weaknesses:

In the current study, the authors downloaded the MLST or wsp genes from a public database and analyzed the data using other methods, and I think the authors may not be familiar with the research progress in the field of insect symbiont transmission, and the current stage of this manuscript lacking sufficient novelty.

We appreciate your critical perspective on our study. However, we respectfully disagree with the viewpoint that our manuscript lacks sufficient novelty.

eLife assessment

Using experiments in the white fly, this manuscript provides evidence that the bacterial symbiont Wolbachia can be transmitted from parasitoid wasps to their insect hosts. Characterizing the transfer of Wolbachia between insect species is a valuable attempt to explain the widespread of this intracellular bacterium. This paper is incomplete as it does not furnish sufficient data to support several of its claims for which additional methods and data are necessary.

Reviewer #1 (Public Review):

Summary and Strengths:

The ability of Wolbachia to be transmitted horizontally during parasitoid wasp infections is supported by phylogenetic data here and elsewhere. Experimental analyses have shown evidence of wasp-to-wasp transmission during coinfection (eg Huigins et al), host to wasp transmission (eg Heath et al), and mechanical ('dirty needle') transmission from host to host (Ahmed et al). To my knowledge this manuscript provides the first experimental evidence of wasp to host transmission. Given the strong phylogenetic pattern of host-parasitoid Wolbachia sharing, this may be of general importance in explaining the distribution of Wolbachia across arthropods. This is of interest as Wolbachia is extremely common in the natural world and influences many aspects of host biology.

Weaknesses:

The first observation of the manuscript is that the Wolbachia strains in hosts are more closely related to those in their parasitoids. This has been reported on multiple occasions before, dating back to the late 1990s. The introduction cites five such papers (the observation is made in other studies too that could be cited) but then dismisses them by stating "However, without quantitative tests, this observation could simply reflect a bias in research focus." As these studies include carefully collected datasets that were analysed appropriately, I felt this claim of novelty was rather strong. It is unclear why downloading every sequence in GenBank avoids any perceived biases, when presumably the authors are reanalysing the data in these papers.

I do not doubt the observation that host-parasitoid pairs tend to share related Wolbachia, as it is corroborated by other studies, the effect size is large, and the case study of whitefly is clearcut. It is also novel to do this analysis on such a large dataset. However, the statistical analysis used is incorrect as the observations are pseudo-replicated due to phylogenetic non-independence. When analysing comparative data like this it is essential to correct for the confounding effects of related species tending to be similar due to common ancestry. In this case, it is well-known that this is an issue as it is a repeated observation that related hosts are infected by related Wolbachia. However, the authors treat every pairwise combination of species (nearly a million pairs) as an independent observation. Addressing this issue is made more complex because there are both the host and symbiont trees to consider. The additional analysis in lines 123-124 (including shuffling species pairs) does not explicitly address this issue.

The sharing of Wolbachia between whitefly and their parasitoids is very striking, although this has been reported before (eg the authors recently published a paper entitled "Diversity and Phylogenetic Analyses Reveal Horizontal Transmission of Endosymbionts Between Whiteflies and Their Parasitoids"). In Lines 154-164 it is suggested that from the tree the direction of transfer between host and parasitoid can be inferred from the data. This is not obvious to me given the poor resolution of the tree due to low sequence divergence. There are established statistical approaches to test the direction of trait changes on a tree that could have been used (a common approach is to use the software BEAST).

Reviewer #2 (Public Review):

The paper by Yan et al. aims to provide evidence for horizontal transmission of the intracellular bacterial symbiont Wolbachia from parasitoid wasps to their whitefly hosts. In my opinion, the paper in its current form consists of major flaws.

Weaknesses:

The dogma in the field is that although horizontal transmission events of Wolbachia occur, in most systems they are so rare that the chances of observing them in the lab are very slim.<br /> For the idea of bacteria moving from a parasitoid to its host, the authors have rightfully cited the paper by Hughes, et al. (2001), which presents the main arguments against the possibility of documenting such transmissions. Thus, if the authors want to provide data that contradict the large volume of evidence showing the opposite, they should present a very strong case.

In my opinion, the paper fails to provide such concrete evidence. Moreover, it seems the work presented does not meet the basic scientific standards.

My main reservations are:

- I think the distribution pattern of bacteria stained by the probes in the FISH pictures presented in Figure 4 looks very much like Portiera, the primary symbiont found in the bacterium of all whitefly species. In order to make a strong case, the authors need to include Portiera probes along with the Wolbachia ones.

- If I understand the methods correctly, the phylogeny presented in Figure 2a is supposed to be based on a wide search for Wolbachia wsp gene done on the NCBI dataset (p. 348). However, when I checked the origin of some of the sequences used in the tree to show the similarity of Wolbachia between Bemisia tabaci and its parasitoids, I found that most of them were deposited by the authors themselves in the course of the current study (I could not find this mentioned in the text), or originated in a couple of papers that in my opinion should not have been published to begin with.

- The authors fail to discuss or even acknowledge a number of published studies that specifically show no horizontal transmission, such as the one claimed to be detected in the study presented.

Reviewer #3 (Public Review):

This is a very ordinary research paper. The horizontal of endosymbionts, including Wolbachia, Rickettsia etc. has been reported in detail in the last 10 years, and parasitoid vectored as well as plant vectored horizontal transmission is the mainstream of research. For example, Ahmed et al. 2013 PLoS One, 2015 PLoS Pathogens, Chiel et al. 2014 Enviromental Entomology, Ahmed et al. 2016 BMC Evolution Biology, Qi et al. 2019 JEE, Liu et al. 2023 Frontiers in Cellular and Infection Microbiology, all of these reported the parasitoid vectored horizontal transmission of endosymbiont. While Caspi-Fluger et al. 2012 Proc Roy Soc B, Chrostek et al. 2017 Frontiers in Microbiology, Li et al. 2017 ISME Journal, Li et al. 2017 FEMS, Shi et al. 2024 mBio, all of these reported the plant vectored horizontal transmission of endosymbiont. For the effects of endosymbiont on the biology of the host, Ahmed et al. 2015 PLoS Pathogens explained the effects in detail.

Weaknesses:

In the current study, the authors downloaded the MLST or wsp genes from a public database and analyzed the data using other methods, and I think the authors may not be familiar with the research progress in the field of insect symbiont transmission, and the current stage of this manuscript lacking sufficient novelty.

eLife assessment

This study presents a valuable syngeneic zebrafish model for studying glioblastoma and will be of interest to neuro-oncologists and cancer biologists. Using a feasible in vivo model to study the tumour microenvironment, cell/cell interaction, and immunity, the data are compelling, and opens up new lines of inquiries for future investigation on the impact of efferocytosis on tumor progression and cell of origin in this model as well as assessments of drug resistance mechanisms, using inhibitors to MAPK , Akt and/or mTOR pathway.

Reviewer #1 (Public Review):

Summary:

The authors have developed a zebrafish model of glioblastoma and characterized this, with a particular focus on the role of recruited myeloid cells in the tumours. Microglia/macrophages in the tumours are proposed to have an inflammatory phenotype and are engaged in phagocytosis. Knockout of Irf7 and Irf8 genes enhanced tumour initiation. Depleting mature myeloid cell types with chlodronate also enhanced tumour initiation. It is proposed that in early stage tumours, microglia/macrophages have tumour suppressive activity.

Strengths:

The authors have generated a novel glioblastoma model in zebrafish. Two key strengths of the zebrafish model are that early stage tumours can be studied and in vivo visualization can be readily performed. The authors show video of microglia/macrophages adopting the ameboid phenotype in tumours (as is observed in human tumours) and engaging in phagocytosis. Video 1 was very impressive in my opinion and shows the model is a very useful tool to study microglia/macrophage:glioblastoma cell interactions. The irf7/irf8 knockdown and the chlodronate experiments are consistent with a role for mature myeloid cells in suppressing tumour initiation, suggesting that the model may also be very valuable in understanding immune surveillance in glioblastoma initiation.

Weaknesses:

EGFRvIII is mainly associated with the classical subtype, so the mesenchymal subtype might be unexpected here. This could be commented on. Some more histologic characterization of the tumours would be helpful. Are they invasive, do larger tumours show necrosis and microvascular proliferation? This would help with understanding the full potential of the new model. Current thinking in established human glioblastoma is that the M1/M2 designations for macrophages are not relevant, with microglia macrophage populations showing a mixture of pre- and anti-inflammatory features. Ideally there would be a much more detailed characterization of the intratumoral microglia/macrophage population here, as single markers can't be relied upon. Phagocytosis could have antitumour effects through removal of live cancer cells, or could be cancer promoting if apoptotic cancer cells are being rapidly cleared with concomitant activation of an immunosuppressive phenotype in the phagocytes (i.e. efferocytosis). It may be possible to distinguish between these two types of phagocytosis experimentally. Do the irf7/8 and chlodronate experiments distinguish between effects on microglia/macrophages and dendritic cells?

Update: The more detailed description of the tumour histology is very interesting and the authors have addressed my previous concerns nicely.

Reviewer #2 (Public Review):

Summary:

Glioblastoma is a common primary brain cancer, that is difficult to treat and has a low survival rate. The lack of genetically tractable and immunocompetent vertebrate animal model has prevented discovery of new therapeutic targets and limited efforts for screening of pharmaceutical agents for the treatment of the disease. Here Weiss et al., express oncogenic variants frequently observed in human glioblastoma within zebrafish lacking the tumor suppressor TP53 to generate a patient-relevant in vivo model. The authors demonstrate that loss of TP53 and overexpression of EGFR, PI3KCA, and mScarlet (p53EPS) in neural progenitors and radial glia leads to visible fluorescent brain lesions in live zebrafish. The authors performed RNA expression analysis that uncovered a molecular signature consistent with human mesenchymal glioblastoma and identified gene expression patterns associated with inflammation. Live imaging revealed high levels of immune cell infiltration and associations between microglia/macrophages and tumor cells. To define functional roles for regulators of inflammation on specific immune-related responses during tumorigenesis, transient CRISPR/Cas9 gene targeting was used to disrupt interferon regulator factor proteins and showed Inflammation-associated irf7 and irf8 are required to inhibit p53EPS tumor formation. Further, experiments to deplete the macrophages using clodronate liposomes suggest that macrophages contribute to the suppression of tumor engraftment following transplantation. The authors' conclusions are supported by the data and the experiments are thoroughly controlled throughout. Taken together, these results provide new insights into the regulation of glioblastoma initiation and growth by the surrounding microenvironment and provide a novel in vivo platform for the discovery of new molecular mechanisms and testing of therapeutics.

Strengths/Weaknesses:

The authors convincingly show that co-injection of activated human EGFRviii, PI3KCAH1047R, and mScarlet into TP53 null zebrafish promotes formation of fluorescent brain lesions and glioblastoma-like tumor formation. The authors include histological characterization of the tumors, as well as quantifications of p-ERK and p-AKT staining to highlight increased activation of the MAPK/AKT signaling pathways in their tumor model.

The authors use a transplantation assay to further test the tumorigenic potential of dissociated cells from glial-derived tumors in the context of specific manipulations of the tumour microenvironment.

The authors nicely show high levels of immune cell infiltration and associations between microglia/macrophages and tumor cells. Quantification of the emergence of macrophages over time in relation to tumor initiation and growth is provided and supports the observations of tumor suppressive activity of the phagocytes. The authors also attempt to delineate if other leukocyte populations are involved and observe tumor formation without significant infiltration of neutrophils.

The authors provide evidence for key genetic regulators of the local microenvironment, showing increased p53EPS tumor initiation following Ifr7 gene knock-down and loss of irf7 expression in the TME.

Author response:

The following is the authors’ response to the original reviews.

Reviewer #1:

“EGFRvIII is mainly associated with the classical subtype, so the mesenchymal subtype might be unexpected here. This could be commented on.” 

We acknowledge that EGFRvIII is most often associated with the classical subtype of glioblastoma and agree that mesenchymal subtype classification may be unexpected given the use of her4.1:EGFRvIII as a driver in our model. We would like to highlight the fact that our brain tumors do also express certain markers associated with the classical subtype including neural precursor and neural stem cell markers like sox2, ascl1b, and gli2 (Supplementary Fig 4, 5; Supplementary Table 1-3). However, our transcriptomic data was not found to significantly enrich for classical subtype gene expression, compared to normal brains. This could be due to a significant contribution of normal brain tissue to our analyses (bulk tumor burdened brains were harvested for RNA sequencing), as well as the significant contribution of mesenchymal subtype signatures and/or inflammatory gene expression in our brain tumor-positive samples. Because signatures associated with inflammation consist of some of the most highly upregulated genes in our samples, this could potentially dilute out and/or lessen alterative subtype and/or signature gene expression. Importantly, it is now widely appreciated that patient tumors simultaneously consist of heterogenous tumor cells reflecting multiple molecular subtypes (Couturier et al., 2020; Darmanis et al., 2017; Neftel et al., 2019), providing glioblastoma with a high level of phenotypic plasticity. We also demonstrate that the contribution of additional drivers not always present with EGFRvIII in patient glioblastoma enhances primary brain tumors in vivo. This result is consistent with more aggressive glioblastomas seen in patients with EGFRvIII variants and TP53 loss-of-function mutations (Ruano et al., 2009). It will therefore be interesting in the future to consider how single or multiple driver mutations contribute to subtype-specific gene expression in our model, as well as histopathology, relative to patients. We have included some of these discussion points to our revised manuscript.     

“Some more histologic characterization of the tumors would be helpful. Are they invasive, do larger tumors show necrosis and microvascular proliferation? This would help with understanding the full potential of the new model.”

We have updated our manuscript to include more histolopathological characterization and images (Supplementary Fig 2).

“Current thinking in established glioblastoma is that the M1/M2 designations for macrophages are not relevant, with microglia macrophage populations showing a mixture of pre- and anti-inflammatory features. Ideally, there would be a much more detailed characterization of the intratumoral microglia/macrophage population here, as single markers can’t be relied upon.”

We performed additional gene set enrichment analyses (GSEA) using our sequencing datasets and compared p53EPS gene expression to M1/M2 macrophage expression signatures and expression signatures from MCSF-stimulated macrophages at early and late (M2 polarized) time-points. From this analysis, we detected enrichment for markers of both pro- and antiinflammatory features, however, with stronger and significant enrichment for gene expression signatures associated with classical pro-inflammatory M1 macrophages. We have included these GSEA plots and gene set enrichment lists as supplementary materials (Supplementary Fig 6, Supplementary Table 6). We also performed GSEA against a broad curated set of immunologic gene sets (C7: immunologic signature gene sets, Molecular Signatures Database, (Liberzon et al., 2011)) and have included the list of signatures and enrichment scores as a supplementary table (Supplementary Table 6). 

“Phagocytosis could have anti-tumor effects through removal of live cancer cells or could be cancer-promoting if apoptotic cells are being rapidly cleared with concomitant activation of an immunosuppressive phenotype in the phagocytes (ie. efferocytosis).” 

We looked at efferocytosis-associated gene expression in our sequencing dataset (124 “efferocytosis” genes, GeneCards), and while we detected upregulation of certain genes associated with efferocytosis in p53EPS brains, we did not detect significant enrichment for the entire gene set. Furthermore, we did not detect up-regulation of key efferocytosis receptors including Axl and Tyro3 (Supplementary Table 1, 2), compared to normal brains. While efferocytosis may contribute to tumor growth and evolution, this GSEA combined with our functional data supporting an inhibitory role for phagocytes in p53EPS tumor initiation and engraftment following transplantation (Fig 4, Fig 5, Supplementary Fig 7), suggests that efferocytosis is not a major driver of tumor formation in our model. However, how efferocytosis affects tumor progression in our model and/or relapse following therapy will be an interesting feature to explore in the future using temporal manipulations of phagocytes and/or treatments with chemical inhibitors.

Author response image 1.

Gene Set Enrichment Analysis (GSEA) for efferocytosis-associated gene expression (124 “efferocytosis” genes in GeneCards) in tp53EPS tumor brains, compared to normal zebrafish brains.

Normalized enrichment score (NES) and p-value are indicated. 

“Do the irf7/8 and chlodronate experiments distinguish between effects on microglia/macrophages and dendritic cells?”

In addition to microglia/macrophages, the IRF8 transcription factor has been shown to control survival and function of dendritic cells (Sichien et al., 2016). Chlodronate treatments are also used to deplete both macrophages and dendritic cells in vivo. Therefore, we cannot distinguish the effects of these manipulations in our experiments and have updated our manuscript throughout to reflect this.     

Reviewer #2:

“The authors state that oncogenic MAPK/AKT pathway activation drives glial-derived tumor formation. It would be important to include a wild-type or uninjected control for the pERK and pAKT staining shown in Fig1 I-K to aid in the interpretation of these results. Likewise, quantification of the pERK and pAKT staining would be useful to demonstrate the increase over WT, and would also serve to facilitate comparison with the similar staining in the KPG model (Supp Fig 2D).”

We have updated Fig 1 and Supplementary Fig 3D (formerly Fig 2D), to include histology from tumor-free uninjected control animals, as well as quantifications of p-ERK and p-AKT staining to highlight increased MAPK/AKT signaling pathway activation in our tumor model.  

“The authors use a transplantation assay to further test the tumorigenic potential of dissociated cells from glial-derived tumors. Listing the percentage of transplants that generate fluorescent tumor would be helpful to fully interpret these data. Additionally, it was not clear based on the description in the results section that the transplantation assay was an “experimental surrogate” to model the relapse potential of the tumor cell. This is first mentioned in the discussion. The authors may consider adding a sentence for clarity earlier in the manuscript as it helps the reader better understand the logic of the assay.” 

We have clarified in the text the percentage of transplants that generated fluorescent tumor (1625%, n=3 independent screens). This is also represented in Fig 5C,D. We also added text when introducing the transplantation assay, explaining that transplantation is frequently used as an experimental surrogate to assess relapse potential, and that our objective was to assess tumor cell propagation in the context of specific manipulations within the TME.  

“The authors nicely show high levels of immune cell infiltration and associations between microglia/macrophages and tumor cells. However, a quantification of the emergence of macrophages over time in relation to tumor initiation and growth would provide significant support to the observations of tumor suppressive activity of the phagocytes. Along these lines, the inclusion of a statement about when leukocytes emerge during normal development would be informative for those not familiar with the zebrafish model.”

In zebrafish, microglia colonize the neural retina by 48 hpf, and the optic tectum by 84 hpf (Herbomel et al., 2001), prior to when we typically observe lesions in our p53EPS brains. To validate the emergence of microglia prior to tumor formation in p53EPS, we have now used live confocal imaging through the brains of uninjected control and p53EPS injected zebrafish at 5, 7 and 9 dpf. As expected, microglia were present throughout the cephalic region and in the brain at 5 dpf (120 hpf). At this stage, p53EPS injected zebrafish brains displayed mosaic cellular expression of her4.1:mScarlet; however, cells were sparse and diffuse, and no large intensely fluorescent tumor-like clusters were detected at this stage (n=12/12 tumor negative). At 7 dpf, microglia were observed in the brains of control and p53EPS zebrafish; however, at this stage we detected clusters of her4.1:mScarlet+ cells (n=5/9), indicative of tumor formation. Lesions were found to be surrounded and/or infiltrated by mpeg:_EGFP+ microglia. Finally, at 9 dpf _her4.1:mScarlet+ expression became highly specific to tumor lesions, and these lesions were associated with _mpeg:_EGFP+ microglia/macrophages (n=8/8 of tumor-positive zebrafish). These descriptions along with representative images has been added to Figure 3.

“From the data provided in Figure 4G and Supp Fig 7b, the authors suggest that “increased p53EPS tumor initiation following Irf gene knock-down is a consequence of irf7 and irf8 loss-of-function in the TME.” Given the importance of the local microenvironment highlighted in this study, spatial information on the form of in situ hybridization to identify the relevant location of the expression change would be important to support this conclusion.”

We performed fluorescent in situ hybridization (using HCR RNA-FISH, Molecular Instruments) on whole mount control and irf7 CRISPR-injected p53EPG animals (her4.1:EGFRvIII +her4.1:PI3KCAH1047R + her4.1:GFP, GFP was used in this case because of probe availability).

Representative confocal projections through tumors, as well as single optical sections are presented and discussed in Figure 4, highlighting the location of irf7 expression change following gene knock-down. We found significant irf7 signal in and surrounding p53EPS tumors at early stages of tumor formation_. This expression was reduced and/or lost following _irf7 CRISPR gene targeting, consistent with RT-PCR data (Supplementary Fig 7).          

“The authors used neutral red staining that labels lysosomal-rich phagocytes to assess enrichment at the early stages of tumor initiation. The images in Figure 3 panel A should be labeled to denote the uninjected controls to aid in the interpretation of the data. In Supplemental Figure 6, the neutral red staining in the irf8 CRISPR-injected larvae looks to be increased, counter to the quantification. Can the authors comment if the image is perhaps not representative?”

We have updated Figure 3 and Supplementary Figure 6 to aid in the interpretation of our results. In Fig 3A, we used tumor-negative controls from our injected cohorts. This was done to control for exogenous transgene presence and/or over-expression prior to (or in the absence of) malignant transformation. In Supplementary Fig 6, our images are representative, but we have now used unprocessed images with arrowheads to highlight neutral-red positive foci for clarity. In our original manuscript the images contained software generated markers, which could have obscured and/or confused the neutral red staining we were trying the highlight.    

Recommendations For the Authors:

Reviewer #1: 

“The PI 3-kinase does a lot more than just activating mTOR and Akt – I would suggest modifying that sentence in the introduction.”

We have adjusted text in the introduction to reflect the broad role for PI3K signaling.

Reviewer #2:

“In Supplemental Fig 1, it would be helpful for the authors to provide a co-stain, such as DAPI to label all nuclei, which would allow the reader to assess the morphology of the cells in the context of the surrounding tissue.”

We have included brightfield images in Supplementary Fig 1, that together with her4.1:mScarlet fluorescence, should help readers assess tumor location and morphology in the context of surrounding tissue. Tumor cell morphology at high-resolution can be visualized in Fig 3, Movie 1 and Movie 2.

“The authors state that oncogenic MAPK/AKT pathway activation drives glial-derived tumor formation. The authors may consider testing if the addition of an inhibitor of MAPK signaling may prevent or decrease the formation of glial-derived tumors in this context to further support their results.” 

To further assess the role for MAPK activation, we decided to test the effect of 50uM AZD6244 MAPK inhibitor following transplantation of dissociated primary p53EPS cells into syngeneic CG1 strain zebrafish embryos, similar to as previously described (Modzelewska et al., 2016). Following 5 days of drug treatments, we did not detect significant differences in tumor engraftment or in tumor size between DMSO control and AZD6244-treated cohorts, suggesting that MAPK inhibition is not sufficient to prevent p53EPS engraftment and growth in our model. In the future, assessments of on-target drug effects, possible resistance mechanisms, and/or testing MAPK inhibitors in combination with other targeted agents including Akt and/or mTOR inhibitors (Edwards et al., 2006; McNeill et al., 2017; Schreck et al., 2020) will enhance our understanding of potential therapeutic strategies.

Author response image 2.

Dorsal views of 8 dpf zebrafish larvae engrafted with her4.1:mScarlet+ p53EPS tumor cells following treatment from 3-8dpf with 0.1% DMSO (control) or 50uM AZD6244. Tumor cell injections were performed at 2 dpf into syngeneic CG1 strain embryos. The percentage of total animals with persisting engraftment following drug treatments, as well as tumor size (microns squared, quantified using Carl Zeiss ZEN software) are shown for control and AZD6244 treated larvae. 

“Have the authors tested if EGFR and PI3KCA driven by other neural promoters produce similar results, or not? This would help support the specificity of her4.1 neural progenitors and glia as the cell of origin in this model.”

At this time, we have not tested other neural promoters. However, previous reports describe a zebrafish zic4-driven glioblastoma model with mesenchymal-like gene expression (Mayrhofer et al., 2017), supporting neural progenitors as a cell of origin. In the future it will be interesting to test sox2, nestin, and gfap promoters to further define and support her4.1-expressing neural progenitors and glia as the cell of origin in our model.

“Other leukocyte populations, such as neutrophils, can also respond to inflammatory cues. Can the authors comment if neutrophils are also observed in the TME?”

We performed initial assessments of neutrophils in the TME using our expression datasets as well as her4.1:EGFRvIII + her4.1:PI3KCAH1047R co-injection into Tg(mpx:EGFP) strain zebrafish. We observed tumor formation without significant infiltration of mpx:EGFP+ neutrophils. Future investigations will be important to assess differences in the contributions of different myeloidderived lineages in the TME of p53EPS, as well as how heterogeneity may be altered depending on different oncogenic drivers and/or stage of tumor progression, as seen in human glioblastoma (Friedmann-Morvinski and Hambardzumyan, 2023). We have added text in the disscussion section of our manuscript to indicate the possibility of neutrophils and/or other immune cell types contributing to p53EPS tumor biology. 

Author response image 3.

Control-injected tumornegative and tumor-positive Tg(mpx:EGFP) zebrafish at 10 dpf. Tg(mpx:EGFP) strain embryos were injected at the one-cell stage with her4.1:EGFRvIII + her4.1:PI3KCAH1047R + her4.1:mScarlet.

“It is not clear if the transcriptomics data has been deposited in a publicly available database, such as the Gene Expression Omnibus (GEO). Sharing of these data would be a benefit to the field and facilitate use in other studies.”

We have uploaded all transcriptomic data to GEO under accession GSE246295.

Author response:

The following is the authors’ response to the original reviews.

Reviewer #1 (Recommendations For The Authors):

(1) Original blots in Figures 2E and 2H should be shown as well as the quantification of miR-182-5p overexpression in HepG2 cells. miR-182-5p expression in T2D patients was 2.3-fold higher than ND patients. The lack of insights into the degree of miR-182-5p overexpression precluded proper interpretation of the data presented.

Thank you very much for these comments. We now include the original uncut blots and relevant bands (new supplementary figure 3A) as well as the quantification of miR-182-5p expression in mimic-treated HepG2 cells in the supplement (new supplementary figure 2).

(2) What are the upstream transcriptional regulators of miR-182-5p?

To the best of our knowledge the upstream transcriptional regulators of miR-182-5p are currently unknown.

(3) What's the purpose of the weight cycling cohort? Figure 3A only showed that miR-182-5p expression was highly correlated to body weight, but the cohort can not explain why the human cohort has different miR-182-5p expression. GTT and ITT data are lacking for this cohort and thus cannot demonstrate a causal link between insulin sensitivity and miR-182-5p. The lack of histological evidence cannot show the relationship between NAFLD and miR-182-5p.

The purpose of the weight cycling cohort was to demonstrate that miR-182-5p is dynamically altered and that it can be reversed to almost control levels by weight loss. Thereby we validate in mice that obesity is associated with miR-182-5p upregulation (HFD group without intervention) and we propose that the adverse effects of increased miR-182-5p in obesity might be reversible by weight loss.  We did not perform ITTs and GTTs in this weigh cycling cohort because the HFD-model in C57BL/6 mice is well established and it can be assumed that glucose- and insulin-tolerance deteriorated during HFD feeding (doi.org/10.1038/oby.2007.608; doi:10.1007/978-1-61779-430-8_27 and improved after weight loss (doi:10.1038/s41598-023-40514-w). To corroborate this assumption, we provide plasma insulin along with as other important metabolic marker of the weight cycling model in supplemental figure 5A.

(4) Loss-of-function of miR-182-5p and/or gain-of-function of Lrp6 in vivo or in vitro would clarify the importance of the miR-182-5p-Lrp6 axis and provide more direct evidence for its potential as a therapeutic target.

We absolutely agree with the reviewer that loss of miR-182 and gain of LRP6 function experiments are missing. However, we provide miR-182 gain of function experiments that impressively show increased liver triglycerides after only seven days of miR-182 overexpression. Because these in vivo data are only short-term, we stated our conclusions carefully and point out that we do not have evidence for a direct involvement of miR-182-5p in insulin signaling. We are now planning follow-up studies in which miR-182-5p will be overexpressed and also antagonized for a longer time. However, for the timeframe of this revision process these extensive studies are not feasible and we ask the reviewer for his/her understanding.

(5) The schematic summary is too complex and includes too many assumptions to faithfully represent the data shown in this study.

We agree, the schematic summary is very complex. Therefore we simplified the upper part (new figure 5) and only focused on the clearly regulated genes and main pathways.

Reviewer #2 (Recommendations For The Authors):

(1) Although lots of microarray analyses were performed in this study, the authors didn't systemically investigate the function of miR-182 in T2DM or NAFLD. The current data provided in this manuscript may only support that miR-182 is involved in the homeostasis of glucose or insulin.

We thank the reviewer for this comment and agree that the nature of or data is mostly correlative. We tried to overcome this by performing mechanistic in vitro data. Because overexpression of miR-182-5p decreases inulin signaling in vitro and induces hyperinsulinemia in vivo we still strongly believe that miR-182-5p is highly relevant for the homeostasis of glucose and insulin.

(2) The authors used miRNA mimics to overexpress miR-182 in mice. How to emphasize the target specificity in the liver? Normally, adeno-associated virus 8 (AAV8) is used to specifically target the liver.

Tail vein injections as used in our experimental set-up are known to deliver compounds directly to the liver via the portal vein. For modulation of microRNAs in the liver it is an established technique to deliver mimics (or inhibitors) via the tail vein (doi:10.1007/978-1-62703-435-7_18; doi: 10.1089/10430349950017734). To account for off-target effects we quantified miR-182-5p and target gene expression in spleen and heart. Although miR-182-5p concentrations in mimic treated mice were strongly increased in these tissues, expression in the liver was still highest (new supplementary figure 6A).

(3) The HE and Oil red staining of the mouse liver should be shown in miR-182-5p overexpressing mice compared with the control mice, which could provide a more intuitive view of the fat content in the mouse liver.

Unfortunately the livers were flash frozen and not optimally prepared for later histological analyses. Nevertheless, we performed H&E stainings in all livers and provide representative HE stainings of two control and two miR-182-mimic treated mice (new supplementary figure 5D). The increase hepatic lipid content is clearly visible in the H&E staining of miR-182-mimic treated mice and supports our previous findings of increased hepatic triglycerides (Figure 4H). Due to the freezing process, livers were damaged and Oil red staining was impossible.

(4) After overexpression of miR-182-5p in mice, the serum insulin levels were increased. Does miR-182-5p affect insulin resistance in mice? The insulin tolerance test (ITT) experiment needs to be performed.

We thank the reviewer for this comment. Indeed, the performance of an ITT would have clarified the effects of miR-182 on insulin tolerance best. Because we did not see differences in the GTT after treating mice acutely with the miR-182 mimic we decided to not perform the ITT in this short-term. The increased fasting serum levels after miR-182-5p mimic treatment (Fig. 4G) suggest that rather insulin sensitivity than insulin secretion is disturbed by miR-182-5p. We are aware, that in future experiments mice should be treated for a longer period with miR-182-5p mimics and that an ITT should be performed in these more chronic studies.

(5) In Figure 2H, the author measured the level of p-Akt/Akt to indicate the effect of miR-182-5p on insulin resistance in HepG2 cells. It is best to provide the western blotting results of p-AKT and t-AKT after HepG2 cells are treated with or without insulin.

We now provide the full blots for all western blotting experiments as new supplemental figure 3B. The HepG2 cells were stimulated with 20 nM insulin 10 min before harvest as described in 2.11 and consequently Akt and p-Akt were quantified. We did not analyze Akt and p-Akt without stimulation because Akt is rarely phosphorylated in the basal non-insulin stimulated state.

(6) This study suggests that miR-182-5p may promote insulin resistance and hyperinsulinemia by downregulating LRP6. Nevertheless, to confirm this conclusion, we suggest you transfect miR-182-5p after downregulating the level of LRP6 with its siRNA for further validation.

Because miR-182-5p targets LRP6 as we have validated by luciferase-assays, LRP6 levels are already low after miR-182-5p overexpression. Thus, the additional downregulation of LRP6 by other means (such as siRNAs) does not make sense in our opinion.

(7) The author described that serum miR-182-5p was neither altered in T2D nor correlated with hepatic miR-182-5p expression, so is it suitable as the biomarker of T2D?

Yes, as the reviewer stated correctly, serum concentrations of miR-182-5p were not related to its liver concentrations or the type 2 diabetic state. We therefore suggest that circulating miR-182-5p levels are not a suitable biomarker for T2D. We clarified this in the discussion.

(8) What are the changes in fasting blood glucose levels in HFD, HC, and YoYo mouse models? Is there a correlation between miR-182-5p level and fasting blood glucose level in T2D patients and mouse models?

Unfortunately, we did not measure the fasting blood glucose levels in this mouse model and therefore cannot answer this question. However, we provide the fasting insulin levels of our mouse models and their positive correlations with miR-182-5p (Fig. 3D and Suppl.Fig. 5D). In T2D humans, hepatic miR-182-5p correlates positively with fasting glucose (Fig. 2B).

(9) The capitalization of the letters in "STrengthening the Reporting of OBservational studies in Epidemiology" should be checked. What does the "Among these is miRNAs miR-182-5p" mean? Please clarify it.

The “STrengthening the Reporting of OBservational studies in Epidemiology “ report form is abbreviated as “STROBE” list. We this capitalized the letters that are used to build the abbreviation.

“Among these is miRNAs miR-182-5p” is a typo for which we apologize. It should mean “Among these conserved miRNAs is miR-182-5p.” We corrected this error.

Reviewer #3 (Recommendations For The Authors):

(1) The functional importance of miR-182 on gene expression is not rigorously tested.

(A) Many of the target genes in Fig. 1C and Fig. 3 are controlled by multiple factors that are known to be increased with obesity (e.g., lipogenic genes are increased by hyperinsulinemia), making it likely that their association with miR-182 is correlative rather than a consequence of miR-182 increases.

We thank the reviewer for this comment and agree that miR-182 is not the only factor regulating the here investigated genes. We rather propose, that miR-182 could be an additional upstream regulator that holds the potential to modify entire pathways of insulin signaling and lipogenesis. However, miR-182 should be not viewed as an on/off-switch as it likely plays a modulating role. Although, our in vivo data stemming from humans and mice are correlative we believe that the in vitro data derived in HepG2 cells clearly show a causal role for miR-182-5ß in decreasing LRP6 and insulin signaling, indicated by lower AKT phosphorylation after miR-182-5p overexpression.

(B) 500-fold overexpression of miR-182 does not significantly change gene expression. The authors need to knockdown miR-182 in mice and then feed them a chow versus high-fat diet. If miR-182 is a significant regulator of these genes, the effects of the diet will be blunted.

We thank the reviewer for the constructive criticism and agree that an optimal experiment would be to antagonize miR-182-5p in mice to rescue glucose and lipid metabolism. There here presented in vivo upregulation of miR-182-5p was a proof-of-concept study to confirm our hypothesis in a reasonable timeframe. We are aware, that follow-up studies are needed, and we are now planning studies in which miR-182-5p will be overexpressed and also antagonized for a longer time. However, for the timeframe of this revision process these extensive studies are not feasible and we ask the reviewer for his/her understanding. 

(2) It has previously been shown that miR-182 is in a polycistrionic microRNA locus that is activated directly by SREBP-2. Is this also true in humans? If so, this would indicate that miR-182 is a marker of SREBP activity. How does the nuclear active form of SREBP1 and SREBP2 change in the human livers and HFD-fed mice?

We thank the reviewer for this very interesting question. Suitable experiments to investigate if miR-182-5p is activated by SREBF would be EMSAs or ChIPs. Unfortunately we have only frozen protein lysate of the human livers left in which such experiments cannot be performed. We agree that this should be prioritizes in the future.

(3) Similarly, to test the role of LRP6 in mediating the effects of miR-182, the authors should compare the effects of miR-182 overexpression in the presence and absence of LRP6.

Because miR-182-5p targets LRP6 as we have validated by luciferase-assays, LRP6 levels are already low after miR-182-5p overexpression. Thus, the additional downregulation of LRP6 by other means (such as siRNAs) does not make sense in our opinion.

(4) The methods are a bit confusing. The authors state that "we applied a logistic regression analysis for the 594 mature miRNAs using the NAFLD activity score (NAS) as a cofactor to exclude any bias by hepatic fat content, lobular inflammation, and fibrosis." However, they later showed that miR-182 levels are correlated with NAS. Please clarify.

We excluded NAFLD explicitly as driving factor for the association to T2D by including a surrogate (the NAFLD activity score) as cofactor. It is well known that NAFLD and T2D are indeed likely associated to each other. Since not all our included individuals with T2D have NAFLD and vice versa, a second correlation with NAS revealed also that a high NAS is associated with higher expression of miR-182.

(5) Does two-fold overexpression of miR-182 (which mimics the effects of HFD) have any effect on chow-fed mice?

This is a very interesting question that we unfortunately cannot answer right now. We are planning further mouse studies in which we will include a chow-fed mice as controls.

eLife assessment

Building on on the observation of an increase in miR-182-5p in diabetic patients, the authors investigated the role of miR-182-5p and its target gene LRP6 in dysregulated glucose tolerance and fatty acid metabolism in obese type 2 diabetics. The use of human livers complemented by supporting data in mice and cells are strengths, but the evidence presented remains incomplete. The findings provide valuable insights into the role of miRNAs in the regulation of liver metabolism and insulin sensitivity in individuals with diabetes and fatty liver disease.

Reviewer #1 (Public Review):

Summary:

This study demonstrated a novel exciting link between conserved miRNA-target axis of miR-182-Lrp6 in liver metabolism which causatively contributes to type 2 diabetes and NAFLD in mice and, potentially, humans.

Strengths:

The direct interaction and inhibition of Lrp6 by miR-182 is convincingly shown. The effects of miR-182-5p on insulin sensitivity are also credible for the in vivo and in vitro gain-of-function experiments.

Weaknesses:

However, the DIO cohorts lack key assays for insulin sensitivity such as ITT or insulin-stimulated pAKT, as well as histological evidence to support their claims and strengthen the link between miR-182-5p and T2D or NAFLD. Besides, the lack of loss-of-function experiments limits its aptitude as potential therapeutic target.

Reviewer #2 (Public Review):

Summary:

In this study, Christin Krause et al mapped the hepatic miRNA-transcriptome of type 2 diabetic obese subjects, identified miR-182-5p and its target genes LRP6 as potential drivers of dysregulated glucose tolerance and fatty acid metabolism in obese T2-diabetics.

Strengths:

This study contains some interesting findings and are valuable for the understanding of key regulatory role of miRNAs in the pathogenesis of T2D.

Weaknesses:

The authors didn't systemically investigate the function of miR-182 in T2DM or NAFLD.

Reviewer #3 (Public Review):

Summary:

In this manuscript, Krause and colleagues identify miR-182 as diabetes-associated microRNA: miR-182 is increased in bariatric surgery patients with versus without T2D; miR-182 was the only microRNA associated with three metabolic traits; miR-182 levels were associated with increased body weight in mice under different dietary manipulations; overexpression in Hep-G2 led to a decrease in LRP6; and overexpression in HFD fed mice led to increased insulin and liver TG. The manuscript provides a potentially useful resource of microRNA expression in human livers, though the functional importance of miR-182 remains unclear.

Strengths:

The use of human tissues and good sample sizes is strong.

Weaknesses:

The study remains primarily correlative; the in vivo overexpression is non-physiological; and the mechanisms by which miR-182 exerts its effects are not rigorously tested.

eLife assessment

This study provides a fundamental advance in palaeontology by reporting the fossils of a new invertebrate, Beretella spinosa, and inferring its relationship with already described species. The analysis placed the newly described species in the earliest branch of moulting invertebrates. The study, supported by convincing fossil observation, hypothesizes that early moulting invertebrate animals were not vermiform.

Reviewer #1 (Public Review):

Summary:

Wang and co-workers characterise the fossil of Beretella spinosa from the early Cambrian, Yanjiahe Formation, South China. Combining morphological analyses with phylogenetic reconstructions, the authors conclude that B. spinosa is closely related to Saccorhytus, an enigmatic fossil recently ascribed to Ecdysozoa, or moulting animals, as an extinct "basal" lineage. Finding additional representatives of the clade Saccorhytida strengthens the idea that there existed a diversity of body plans previously underappreciated in Ecdysozoa, which may have implications for our understanding of the earliest steps in the evolution of this major animal group.

Strengths:

I'm not a paleobiologist; therefore, I cannot give an expert opinion on the descriptions of the fossils. However, the similarities with Saccorhytus seem evident, and the phylogenetic reconstructions are adequate. Evolutionary interpretations are generally justified, and the consolidation of Saccorhytida as the extinct sister lineage to extant Ecdysozoans will have significant implications for our understanding of this major animal clade.

Weaknesses:

While I generally agree with the author's interpretations, the idea of Saccorhytida as a divergent, simplified off-shot is slightly contradictory with a probably non-vermiform ecdysozoan ancestor. The author's analyses do not discard the possibility of a vermiform ecdysozoan ancestor (importantly, Supp Table 4 does not reconstruct that character), and outgroup comparison with Spiralia (and even Deuterostomia for Protostomia as a whole) indicates that a more or less anteroposteriorly elongated (i.e., vermiform) body is likely common and ancestral to all major bilaterian groups, including Ecdysozoa. Indeed, Figure 4 b depicts the potential ancestor as a "worm". The authors argue that the simplification of Saccorhytida from a vermiform ancestor is unlikely "because it would involve considerable anatomical transformations such as the loss of vermiform organization, introvert and pharynx in addition to that of the digestive system". However, their data support the introvert as a specialisation of Scalidophora (Fig. 4a and Supp Table 4), and a pharyngeal structure cannot be ruled out in Saccorhytida. Likewise, loss of an anus is not uncommon in Bilateria. Moreover, this can easily become a semantics discussion (to what extent can an animal be defined as "vermiform"? Where is the limit?). Therefore, I suggest to leave the evolutionary scenario more open. Supporting Saccorhytida as a true group at the early steps of Ecdysozoa evolution is important and demonstrates that animal body plans are more plastic than previously appreciated. However, with the current data, it is unlikely that Saccorhytida represents the ancestral state for Ecdysozoa (as the authors admit), and a vermiform nature is not ruled out (and even likely) in this animal group. Suggesting that the ancestral Ecdysozoan might have been small and meiobenthic is perhaps more interesting and supported by the current data (phylogeny and outgroup comparison with Spiralia).

Reviewer #2 (Public Review):

Summary:

This work provides important anatomical features of a new species from the Lower Cambrian, which helps advance our understanding of the evolutionary origins of animal body plans. The authors interpreted that the new species possessed a bilateral body covered with cuticular polygonal reticulation and a ventral mouth. Based on cladistic analyses using maximum likelihood, Bayesian, and parsimony, the new species was placed, along with Saccorhytus, in a sister-group ("Saccorhytida") of the Ecdysozoa. The phylogenetic position of Saccorhytida suggests a new scenario of the evolutionary origin of the crown ecdysozoan body plan.

Strengths:

Although the new species reported in this paper show strange morphologies, the interpretation of anatomical features was based on detailed observations of multiple fossil specimens, thereby convincing at the moment. Morphological data about fossil taxa in the Ediacaran and Early Cambrian are quite important for our understanding of the evolution of body plans (and origins of phyla) in paleontology and evolutionary developmental biology, and this paper represents a valuable contribution to such research fields.

Weaknesses:

The preservations of the specimens, in particular on the putative ventral side, are not good, and the interpretation of the anatomical features need to be tested with additional specimens in future. The monophyly of Cycloneuralia (Nematoida + Scalidophora) was not necessarily well-supported by cladistic analyses (Supplementary Figures 7-9), and the evolutionary scenario (Fig. 4) also need to be tested in future works. On the other hand, the revised version provides important contributions from currently available data, and the above-mentioned problems should be studied in a separate paper in future.

Author response:

The following is the authors’ response to the original reviews.

Public reviews:

Reviewer 1:

Weaknesses:

While I generally agree with the author's interpretations, the idea of Saccorhytida as a divergent, simplified off-shot is slightly contradictory with a probably non-vermiform ecdysozoan ancestor. The author's analyses do not discard the possibility of a vermiform ecdysozoan ancestor (importantly, Supplementary Table 4 does not reconstruct that character),

Saccorhytids are only known from the early Cambrian and their unique morphology has no equivalent among any extinct or extant ecdysozoan groups. This prompted us to consider them as a possible dead-end evolutionary off-shot. The nature of the last common ancestor of ecdysozoan (i.e. an elongated worm-like or non-vermiform animal with capacities to renew its cuticle by molting) remains hypothetical. At present, palaeontological data do not allow us to resolve this question. The animal in Fig. 4b at the base of the tree is supposed to represent an ancestral soft-bodied form with no cuticle from which ecdysozoan evolved via major innovations (cuticular secretion and ecdysis). Its shape is hypothetical as indicated by a question mark. Our evolutionary model is clearly intended to be tested by further studies and hopefully new fossil discoveries.

…and outgroup comparison with Spiralia (and even Deuterostomia for Protostomia as a whole) indicates that a more or less anteroposteriorly elongated (i.e., vermiform) body is likely common and ancestral to all major bilaterian groups, including Ecdysozoa. Indeed, Figure 4b depicts the potential ancestor as a "worm". The authors argue that the simplification of Saccorhytida from a vermiform ancestor is unlikely "because it would involve considerable anatomical transformations such as the loss of vermiform organization, introvert, and pharynx in addition to that of the digestive system". However, their data support the introvert as a specialisation of Scalidophora (Figure 4a and Supplementary Table 4), and a pharyngeal structure cannot be ruled out in Saccorhytida. Likewise, loss of an anus is not uncommon in Bilateria. Moreover, this can easily become a semantics discussion (to what extent can an animal be defined as "vermiform"? Where is the limit?).

We agree that “worm” and “vermiform” are ill-defined terms. They are widely used in various palaeontological and biological papers to describe elongated tubular animals such as edydsozoans and annelids (see Giribet and Edgecombe 2017; popular textbook written by Nielsen 2012; Schmit-Rhaesa 2013; Brusca et al. 2023; Giribet and Edgecombe 2020). Very few other animals are termed “worms”. Changes have been made in the text to solve this semantic problem, for example in the abstract where we added (i.e elongated and tubular) to better define what we mean by “vermiform”.

Priapulid worms or annelids are examples of extremely elongated, tubular animals. In saccorhytids, the antero-posterior elongation is present (as it is in the vast majority of bilaterians) but extremely reduced, Saccorhytus and Beretella having a sac-like or beret-shape, respectively. That such forms may have derived from elongated, tubular ancestors (e.g. comparable with present-day priapulid worms) would require major anatomical transformations that have no equivalent among modern animals. We agree that further speculation about the nature of these transformations is unnecessary and should be deleted simply because the nature of these ancestors is purely hypothetical. We also agree that the loss of anus and the extreme simplification of the digestive system is common among extant bilaterians. In Figure 4b, the hypothetical pre-ecdysozoan animal is slightly elongated (along its antero-posterior axis) but in no way comparable with a very elongated and cylindrical ecdysozoan worm (e.g. extant or extinct priapulid).

Therefore, I suggest to leave the evolutionary scenario more open. Supporting Saccorhytida as a true group at the early steps of Ecdysozoa evolution is important and demonstrates that animal body plans are more plastic than previously appreciated. However, with the current data, it is unlikely that Saccorhytida represents the ancestral state for Ecdysozoa (as the authors admit), and a vermiform nature is not ruled out (and even likely) in this animal group. Suggesting that the ancestral Ecdysozoan might have been small and meiobenthic is perhaps more interesting and supported by the current data (phylogeny and outgroup comparison with Spiralia).

We agree to leave the evolutionary scenario more open, especially the evolutionary process that gave rise to Saccorhytida. Again, we know nothing about the morphology of the ancestral ecdysozoan (typically the degree of body elongation, whether it had a differentiated introvert or not, whether it had a through gut or not). In Fig.4, the ancestral ecdysozoan is supposed to have evolved from a soft-bodied epibenthic animal through key innovations such as the secretion of a cuticle and ecdysis. It is a hypothesis that needs to be tested by further studies and fossil discoveries. Speculations concerning the process through which saccorhytids may have arisen have been deleted.

Reviewer 2:

Weaknesses:

The preservations of the specimens, in particular on the putative ventral side, are not good, and the interpretation of the anatomical features needs to be tested with additional specimens in the future. The monophyly of Cycloneuralia (Nematoida + Scalidophora) was not necessarily well-supported by cladistic analyses, and the evolutionary scenario (Figure 4) also needs to be tested in future works.

Yes, we agree that the animal described in our manuscrip remains enigmatic (e.g. the natures of its internal organs, its lifestyle, etc..). Whereas the dorsal side of the animal is well documented (consistent pattern of pointed sclerites), uncertainties remain concerning its ventral anatomy (typically the mouth location and shape). Additional better-preserved specimens will hopefully provide the missing information. Concerning Cycloneuralia, their monophyly is generally better supported by analyses based on morphological characters than in molecular phylogenies.

Reviewer 3:

Weaknesses:

I, as a paleontology non-expert, experienced several difficulties in reading the manuscript. This should be taken into consideration when assuming a wide range of readers including non-experts.

We have ensured that the text is comprehensible to biologists. The main results are summarized in relatively simple diagrams (e.g. Fig. 4) that can be understood by non-specialized readers. We are aware that technical descriptive terms may appear obscure to non-specialists. We can hardly avoid them in the descriptive parts. However, our figures (e.g. SEM images and 3D-reconstruction) are clear enough to give the reader a clear idea of the morphology of Beretella.

Recommendations for the authors:

All three reviewers appreciate the discovery and found the merit of publishing this manuscript. They also raised some concerns about the data presentation. The authors are requested to perform no additional analysis but to go through all the reviewer comments and rebut or intake them in revising the manuscript.

Reviewer 1:

- Line 41: comma after "ecdysozans".

OK, done.

- Formatting style: add a space before references.

OK, done.

- Line 169: B. spinosa in italics

OK, done.

- Line 157: could the "relatively large opening" in the flattened ventral side of a mouth (even when altered by the fossilisation process)?

Most bilaterians have a mouth. There is no opening on the relatively well-preserved dorsal side of Beretella, that could be interpreted as a mouth. In contrast the flattened ventral side often show a depressed area that could potentially bear a mouth. This ventral area is often pushed in and poorly preserved. The cuticle of this ventral side might have been relatively thinner, perhaps more flexible than that of the dorsal one (with strong sclerites). These differences might explain why the possible oral area is poorly preserved.

- Line 178: "position of the mouth"

OK, done.

- Line 219: "These sclerites, unknown..."

OK, done.

- Line 282: update reference formatting

OK, done.

- Line 298: remove reference to Supplementary Table 4, as it does not refer to the possible vermiform nature of the last common ecdysozoan ancestor?

OK, done.

- Figure 4a: change "paired legs" for "paired appendages"?

OK, done.

- Supplementary Table 4: For TGE and Introvert, the state 0 (absent) should be in bold and underlined (as it is the most likely state).

OK, done.

Reviewer 2:

Line 25: "from the early Cambrian" should be changed into "from the lower Cambrian"

OK, done.

Line 126: The range of maximum length should be reported in µm (rather than mm) just like those of maximum width and height.

OK, done.

Lines 191-192: Please recheck the figure panels of Saccorhytus (Supplementary Figure 4c) and scalidophoran worm (Supplementary Figure 4d). Perhaps, the former should refer to Figure 4d, and the latter to Figure 4c?

OK, done.

Lines 239 and 241: "1" and "2" appear to stand for citations (the other journal style), but I am not certain what they are.

To avoid confusing, we replace ‘1’ and ‘2’ by ‘i’ and ‘ii’.

Figures 3d and 4a: "Cycloneuralia" should be included in the phylogenetic trees.

OK, done.

Figure 3: The caption for the panel d is redundant. It should be changed into, for example, "Phylogenetic tree obtained from cladistic analyses using maximum likelihood (IQTREE)."

OK, done.

Supplementary Figures 6-9: In the captions, more detailed explanations of the results (for example, "50% majority rule consensus of XXX trees" and "strict consensus of all 4 most-parsimonious trees") should be provided.

OK, done.

Supplementary Figures 8 and 9: The caption explains that Cycloneuralia is resolved as a paraphyletic group, but it is not certain because Nematoida, Scalidophora, and Panarthropoda are resolved in a polytomy.

We changed the sentence into:

“Note that Cycloneuralia does not appear as a monophyletic clade”

Reviewer 3:

Line 25 'tiny' - I suggest giving an absolute measure of the size.

We add ‘maximal length 3 mm’.

Line 29 'both forms' - This is hard to follow by a non-expert. Can this be replaced with 'fossil species'?

OK, done.

Line 32 'dead-end' - Is this word necessary? I suggest to skip this word, as it is obvious that this lineage is extinct.

OK, done.

Lines 80, 94, and 172 'Remarks' - I, as a palaeontology non-expert, cannot get this manuscript structure with a repetition of this same section title.

Our systematic descriptions follow the standard rules in palaeontology.

Line 119 - I could not get what this 'Member 5' that was not introduced earlier means.

In Stratigraphy, ‘member’ is a lithostratigraphic subdivision (a Formation is usually subdivided into several Members).

Lines 104, 105, 417, ... - The name of the organization or database hosting these IDs (CUB.... and ELIXX....) should also be supplied.

OK, done.

Lines 341 and 361 - These two Figures (Figures 1 and 2) have the same caption (with an addition to the one for Figure 1). There should be a distinction based on what is presented in each figure.

We corrected the caption of Figure 2 and wrote the following: ‘Beretella spinosa gen. et sp. nov.’.

Line 362-367 - There is no guide about what the individual figure panels (e.g., Figure 2g, 2h, and 2i) show in detail. This guide should be supplied. This also applies to Figure 3a-c - are they anterolateral (a), dorsal (b), and posterolateral (c) views? It is better to write clearly in this way.

OK, done.

Figure 3d - The color contrast is not sufficient, and this figure does not look reader-friendly. Plus, the division into Cycloneuralia and Panarthropoda is indicated above the tree, but it is not clear what range of lineages these clades include. For example, is Pliciloricidae included in Cycloneuralia? Also, is Collinsium included in Panarthropoda? This figure looks quite unreliable, and it should be easy to fix.

OK, done.

Line 277 legend of Figure 3 - Including the parenthesis only with the program name (IQTREE) is not useful at all. Isn't it enough to describe it in Methods?

OK, done. We remove (IQTREE).

Line 380 legend of Figure 3 - I could not get where 'thicker bars' are.

Known fossil record indicated by thicker vertical bars. We added “vertical”.

Line 453 - Give full names of the methods, maximum parsimony, and maximum-likelihood.

OK, done.

Line 489 - State clearly what 'the recent paper' means.

Replace ‘recent’ by ‘present’.

eLife assessment

The authors report that the neurohormone, bursicon, and its receptor, play a role in regulating aspects of the seasonal polyphenism of the bug, Cacopsylla chinensis. This important study shows that low temperature activates the bursicon signaling pathway during the transition from the summer to the winter form and that it affects cuticle pigment and chitin content, and cuticle thickness. In addition, the authors show that the microRNA miR-6012 targets the bursicon receptor, thereby modulating the function of the bursicon signaling pathway. The study's solid set of experiments and results reveal a role of bursicon signaling in regulating features of polyphenism related to the exoskeleton. Nevertheless, they only incompletely substantiate the authors' claims about the regulation of polyphenism itself.

Joint Public Review:

Summary:

Bursicon is a key hormone regulating cuticle tanning in insects. While the molecular mechanisms of its function are rather well studied--especially in the model insect Drosophila melanogaster, its effects and functions in different tissues are less well understood. Here, the authors show that bursicon and its receptor play a role in regulating aspects of the seasonal polyphenism of Cacopsylla chinensis. They found that low temperature treatment activated the bursicon signaling pathway during the transition from summer form to winter form and affect cuticle pigment and chitin content, and cuticle thickness. In addition, the authors show that miR-6012 targets the bursicon receptor, CcBurs-R, thereby modulating the function of bursicon signaling pathway in the seasonal polyphenism of C. chinensis. This discovery expands our knowledge of the roles of neuropeptide bursicon action in arthropod biology.

However, the study falls short of its claim that it reveals the molecular mechanisms of a seasonal polyphenism. While cuticle tanning is an important part of the pear psyllid polyphenism, it is not the equivalent of it. First, there are other traits that distinguish between the two morphs, such as ovarian diapause (Oldfield, 1970), and the role of bursicon signaling in regulating these aspects of polyphenism were not measured. Thus, the phenotype in pear psyllids, whereby knockdown bursicon reduces cuticle tanning seems to simply demonstrate the phenotypes of Drosophila mutants for bursicon receptor (Loveall and Deitcher, 2010, BMC Dev Biol) in another species (Fig. 2I, 4H). Second, the study fails to address the threshold nature of cuticular tanning in this species, although it is the threshold response (specifically, to temperature and photoperiod) that distinguishes this trait as a part of a polyphenism. Whereas miR-6012 was found to regulate bursicon expression, there no evidence is provided that this microRNA either responds to or initiates a threshold response to temperature. In principle, miR-6012 could regulate bursicon whether or not it is part of a polyphenism. Thus, the impact of this work would be significantly increased if it could distinguish between seasonal changes of the cuticle and a bona fide reflection of polyphenism.

Strengths:

This study convincingly identifies homologs of the genes encoding the bursicon subunits and its receptor, showing an alignment with those of another psyllid as well as more distant species. It also demonstrates that the stage- and tissue-specific levels of bursicon follow the expected patterns, as informed by other insect models, thus validating the identity of these genes in this species. They provide strong evidence that the expression of bursicon and its receptor depend on temperature, thereby showing that this trait is regulated through both parts of the signaling mechanism.

Several parallel measurements of the phenotype were performed to show the effects of this hormone, its receptor, and an upstream regulator (miR-6012), on cuticle deposition and pigmentation (if not polyphenism per se, as claimed). Specifically, chitin staining and TEM of the cuticle qualitatively show difference between controls and knockdowns, and this is supported by some statistical tests of quantitative measurements (although see comments below). Thus, this study provides strong evidence that bursicon and its receptor play an important role in cuticle deposition and pigmentation in this psyllid.

The study identified four miRNAs which might affect bursicon due to sequence motifs. By manipulating levels of synthetic miRNA agonists, the study successfully identified one of them (miR-6012) to cause a cuticle phenotype. Moreover, this miRNA was localized (by FISH) to the cuticle, body-wide. To our knowledge, this is the first demonstrated function for this miRNA, and this study provides a good example of using a gene of known function as an entry point to discovering others influencing a trait. Thus, this finding reveals another level of regulation of cuticle formation in insects.

Weaknesses:

(1) The introduction to this manuscript does not accurately reflect progress in the field of mechanisms underlying polyphenism (e.g., line 60). There are several models for polyphenism that have been used to uncover molecular mechanisms in at least some detail, and this includes seasonal polyphenisms in Hemiptera. Therefore, the justification for this study cannot be predicated on a lack of knowledge, nor is the present study original or unique in this line of research (e.g., as reviewed by Zhang et al. 2019; DOI: 10.1146/annurev-ento-011118-112448). The authors are apparently aware of this, because they even provide other examples (lines 104-108); thus the introduction seems misleading as framed.

(2) The data in Figure 2H show "percent of transition." However, the images in 2I show insects with tanned cuticle (control) vs. those without (knockdown). Yet, based on the description of the Methods provided, there appears to be no distinction between "percent of transition" and "percent with tanning defects". This an important distinction to make if the authors are going to interpret cuticle defects as a defect in the polyphenism. Furthermore, there is no mention of intermediate phenotypes. The data in 2H are binned as either present or absent, and these are the phenotypes shown in 2I. Was the phenotype really an all-or-nothing response? Instead of binning, which masks any quantitative differences in the tanning phenotypes, the authors should objectively quantify the degree of tanning and plot that. This would show if and to what degree intermediate tanning phenotypes occurred, which would test how bursicon affects the threshold response. This comment also applies to the data in Figures 4G and 6G. Since cuticle tanning is present in more insect than just those with seasonal polyphenism, showing how this responds as a threshold is needed to make claims about polyphenism.

(3) This study also does not test the threshold response of cuticle phenotypes to levels of bursicon, its receptor, or miR-6012. Hormone thresholds are the most widespread and, in most systems where polyphenism has been studied, the defining characteristic of a polyphenism (e.g., Nijhout, 2003, Evol Dev). Quantitative (not binned) measurements of a polyphenism marker (e.g., chitin) should be demonstrated to result as a threshold titer (or in the case of the receptor, expression level) to distinguish defects in polyphenism from those of its component trait.

(4) Cuticle issue:<br /> (a) Unlike Fig. 6D and F, Figs. 2D and F do not correspond to each other. Especially the lack and reduction of chitin in ds-a+b! By fluorescence microscopy there is hardly any signal, whereas by TEM there is a decent cuticle. Additionally, the dsGFP control cuticle in 2D is cut obliquely with a thick and a thin chitin layer. This is misleading.<br /> (b) In Figs. 2F and 3F, the endocuticle appears to be missing, a portion of the procuticle that is produced post-molting. As tanning is also occurring post-molting, there seems to be a general problem with cuticle differentiation at this time point. This may be a timing issue. Please clarify.<br /> (c) To provide background information, it would be useful analyze cuticle formation in the summer and winter morphs of controls separately by light and electron microscopy. More baseline data on these two morphs is needed.<br /> (d) For the TEM study, it is not clear whether the same part of the insect's thorax is being sectioned each time, or if that matters. There is not an obvious difference in the number of cuticular layers, but only the relative widths of those layers, so it is difficult to know how comparable those images are. This raises two questions that the authors should clarify. First, is it possible that certain parts of the thoracic cuticle, such as those closer to the intersegmental membrane, are naturally thinner than other parts of the body? Second, is the tanning phenotype based on the thickness or on the number of chitin layers, or both? The data shown later in Figure 4I, J convincingly shows that the biosynthesis pathway for chitin is repressed, but any clarification of what this might mean for deposition of chitin would help to understand the phenotypes reported. Also, more details on how the data in Fig. 2G were collected would be helpful. This also goes for the data in Fig. 4 (bursicon receptor knockdowns).

(5) Tissue issue:<br /> The timed experiments shown in all figures were done in whole animals. However, we know from Drosophila that Bursicon activity is complex in different tissues. There is, thus, the possibility, that the effects detected on different days in whole animals are misleading because different tissues--especially the brain and the epidermis, may respond differentially to the challenge and mask each other's responses. The animal is small, so the extraction from single tissue may be difficult. However, this important issue needs to be addressed.

(6) No specific information is provided regarding the procedure followed for the rescue experiments with burs-α and burs-β (How were they done? Which concentrations were applied? What were the effects?). These important details should appear in the Materials and Methods and the Results sections.

(7) Pigmentation<br /> (a) The protocol used to assess pigmentation needs to be validated. In particular, the following details are needed: Were all pigments extracted? Were pigments modified during extraction? Were the values measured consistent with values obtained, for instance, by light microscopy (which should be done)?<br /> (b) In addition, pigmentation occurs post-molting; thus, the results could reflect indirect actions of bursicon signaling on pigmentation. The levels of expression of downstream pigmentation genes (ebony, lactase, etc) should be measured and compared in molting summer vs. winter morphs.

(8) L236: "while the heterodimer protein of CcBurs α+β could fully rescue the effect of CcBurs-R knockdown on the transition percent (Figure 4G 4H)". This result seems contradictory. If CcBurs-R is the receptor of bursicon, the heterodimer protein of CcBurs α+β should not be able to rescue the effect of CcBurs-R knockdown insects. How can a neuropeptide protein rescue the effect when its receptor is not there! If these results are valid, then the CcBurs-R would not be the (sole) receptor for CcBurs α+β heterodimer. This is a critical issue for this manuscript and needs to be addressed (also in L337 in Discussion).

(9) Fig. 5D needs improvement (the magnification is poor) and further explanation and discussion. mi6012 and CcBurs-R seem to be expressed in complementary tissues--do we see internal tissues also (see problem under point 2)? Again, the magnification is not high enough to understand and appreciate the relationships discussed.

(10) The schematic in Fig. 7 is a useful summary, but there is a part of the logic that is unsupported by the data, specifically in terms of environmental influence on cuticle formation (i.e., plasticity). What is the evidence that lower temperatures influence expression of miR-6012? The study measures its expression over life stages, whether with an agonist or not, over a single temperature. Measuring levels of expression under summer form-inducing temperature is necessary to test the dependence of miR-6012 expression on temperature. Otherwise, this result cannot be interpreted as polyphenism control, but rather the control of a specific trait.

eLife assessment

This paper addresses a question regarding the low overlap between genetic variants linked to human complex diseases and variants linked to differences in gene expression. Some of the analyses supporting the main claims are convincing, and the key conclusions are valuable and of interest to readers in the fields of human genetics and functional genomics. However, chromatin accessibility QTL (caQTL) also carry the limitation of not identifying the genes that directly mediate the influence on disease phenotypes.

Reviewer #1 (Public Review):

Most human traits and common diseases are polygenic, influenced by numerous genetic variants across the genome. These variants are typically non-coding and likely function through gene regulatory mechanisms. To identify their target genes, one strategy is to examine if these variants are also found among genetic variants with detectable effects on gene expression levels, known as eQTLs. Surprisingly, this strategy has had limited success, and most disease variants are not identified as eQTLs, a puzzling observation recently referred to as "missing regulation".

In this work, Jeong and Bulyk aimed to better understand the reasons behind the gap between disease-associated variants and eQTLs. They focused on immune-related diseases and used lymphoblastoid cell lines (LCLs) as a surrogate for the cell types mediating the genetic effects. Their main hypothesis is that some variants without eQTL evidence might be identifiable by studying other molecular intermediates along the path from genotype to phenotype. They specifically focused on variants that affect chromatin accessibility, known as caQTLs, as a potential marker of regulatory activity.

The authors present data analyses supporting this hypothesis: several disease-associated variants are explained by caQTLs but not eQTLs. They further show that although caQTLs and eQTLs likely have largely overlapping underlying genetic variants, some variants are discovered only through one of these mapping strategies. Notably, they demonstrate that eQTL mapping is underpowered for gene-distal variants with small effects on gene expression, whereas caQTL mapping is not dependent on the distance to genes. Additionally, for some disease variants with caQTLs but no corresponding eQTLs in LCLs, they identify eQTLs in other cell types.

Altogether, Jeong and Bulyk convincingly demonstrate that for immune-related diseases, discovering the missing disease-eQTLs requires both larger eQTL studies and a broader range of cell types in expression assays. It remains to be seen what fractions of the missing disease-eQTLs will be discovered with either strategy and whether these results can be extended to other diseases or traits.

It should be noted that the problem of "missing regulation" has been investigated and discussed in several recent papers, notably Umans et al., Trends in Genetics 2021; Connally et al., eLife 2022; Mostafavi et al., Nat. Genet. 2023. The results reported by Jeong and Bulyk are not unexpected in light of this previous work (all of which they cite), but they add valuable empirical evidence that mostly aligns with the model and discussions presented in Mostafavi et al.

Reviewer #2 (Public Review):

Summary:

eQTLs have emerged as a method for interpreting GWAS signals. However, some GWAS signals are difficult to explain with eQTLs. In this paper, the authors demonstrated that caQTLs can explain these signals. This suggests that for GWAS signals to actually lead to disease phenotypes, they must be accessible in the chromatin. This implies that for GWAS signals to translate into disease phenotypes, they need to be accessible within the chromatin.

However, fundamentally, caQTLs, like GWAS, have the limitation of not being able to determine which genes mediate the influence on disease phenotypes. This limitation is consistent with the constraints observed in this study.

(1) For reproducibility, details are necessary in the method section.

- Details about adding YRI samples in ATAC-seq: For example, how many samples are there, and what is used among public data? There is LCL-derived iPSC and differentiated iPSC (cardiomyocytes) data , not LCL itself. How does this differ from LCL, and what is the rationale for including this data despite the differences?

- caQTL is described as having better power than eQTL despite having fewer samples. How does the number of ATAC peaks used in caQTL compare to the number of gene expressions used in eQTL?

- Details about RNA expression data: In the method section, it states that raw data (ERP001942) was accessed, and in data availability, processed data (E-GEUV-1) was used. These need to be consistent.

How many samples were used (the text states 373, but how was it reduced from the original 465, and the total genotype is said to be 493 samples while ATAC has n=100; what are the 20 others?), and it mentions European samples, but does this exclude YRI?

(2) Experimental results determining which TFs might bind to the representative signals of caQTL are required.

(3) It is stated that caQTL is less tissue-specific compared to eQTL; would caQTL performed with ATAC-seq results from different cell types, yield similar results?

eLife assessment

This valuable work presents elegant experimental data from the Drosophila embryo supporting the notion that interactions among specific loci, called boundary elements, contribute to topologically associated domain (TAD) formation and gene regulation. The evidence supporting boundary:boundary pairing as a determinant of 3D structures is compelling; however, an inability to deplete loop extruders formally leaves open a possible contribution of loop extrusion. This study will be of interest to the nuclear structure community, particularly those using Drosophila as a model.

Reviewer #1 (Public Review):

Summary:

The authors addressed how long-range interactions between boundary elements are established and influence their function in enhancer specificity. Briefly, the authors placed two different reporters separated by a boundary element. They inserted this construct ectopically ~140 kb away from an endogenous locus that contains the same boundary element. The authors used expression patterns driven by nearby enhancers as an output to determine which enhancers the reporters interact with. They complemented this analysis with 3D DNA contact mapping. The authors found that the orientation of the boundary element determined which enhancers each reporter interacted with. They proposed that the 3D interaction topology, whether being circular or stem configuration, distinguished whether the interaction was cohesin mediated or through an independent mechanism termed pairing.

Strengths:

The transgene expression assays are built upon prior knowledge of the enhancer activities. The 3D DNA contacts confirm that transgene expression correlates with the contacts. Using 4 different orientations covers all combinations of the reporter genes and the boundary placement.

Weaknesses:

The interpretation of the data as a refusal of loop extrusion playing a role in TAD formation is not warranted, as the authors did not deplete the loop extruders to show that what they measure is independent. As the authors show, the single long DNA loop mediated by cohesin loop extrusion connecting the ectopic and endogenous boundary is clearly inconsistent with the results, therefore the main conclusion of the paper that the 3D topology of the boundary elements a consequence of pairing is strong. However, the loop extrusion and pairing are not mutually exclusive models for the formation of TADs. Loop-extruding cohesin complexes need not make a 140 kb loop, multiple smaller loops could bring together the two boundary elements, which are then held together by pairing proteins that can make circular topologies.

Reviewer #2 (Public Review):

In Bing et al, the authors analyze micro-C data from NC14 fly embryos, focusing on the eve locus, to assess different models of chromatin looping. They conclude that fly TADs are less consistent with conventional cohesin-based loop extrusion models and instead rely more heavily on boundary-boundary pairings in an orientation-dependent manner.

Overall, I found the manuscript to be interesting and thought-provoking. However, this paper reads much more like a perspective than a research article. Considering the journal is aimed at the general audience, I strongly suggest the authors spend some time editing their introduction to the most salient points as well as organizing their results section in a more conventional way with conclusion-based titles. It was very difficult to follow the authors' logic throughout the manuscript as written. It was also not clear as written which experiments were performed as part of this study and which were reanalyzed but published elsewhere. This should be made clearer throughout.

It has been shown several times that Drosophila Hi-C maps do not contain all of the features (frequent corner peaks, stripes, etc.) observed when compared to mammalian cells. Considering these features are thought to be products of extrusion events, it is not an entirely new concept that Drosophila domains form via mechanisms other than extrusion. That being said, the authors' analyses do not distinguish between the formation and the maintenance of domains. It is not clear to this reviewer why a single mechanism should explain the formation of the complex structures observed in static Hi-C heatmaps from a population of cells at a single developmental time point. For example, how can the authors rule out that extrusion initially provides the necessary proximity and possibly the cis preference of contacts required for boundary-boundary pairing whereas the latter may more reflect the structures observed at maintenance? Future work aimed at analyzing micro-C data in cohesin-depleted cells might shed additional light on this.

Additional mechanisms at play include compartment-level interactions driven by chromatin states. Indeed, in mammalian cells, these interactions often manifest as a "plume" on Hi-C maps similar to what the authors attribute to boundary interactions in this manuscript. How do the chromatin states in the neighboring domains of the eve locus impact the model if at all?

How does intrachromosomal homolog pairing impact the models proposed in this manuscript (Abed et al. 2019; Erceg et al., 2019). Several papers recently have shown that somatic homolog pairing is not uniform and shows significant variation across the genome with evidence for both tight pairing regions and loose pairing regions. Might loose pairing interactions have the capacity to alter the cis configuration of the eve locus?

In summary, the transgenic experiments are extensive and elegant and fully support the authors' models. However, in my opinion, they do not completely rule out additional models at play, including extrusion-based mechanisms. Indeed, my major issue is the limited conceptual advance in this manuscript. The authors essentially repeat many of their previous work and analyses. The authors make no attempt to dissect the mechanism of this process by modifying extrusion components directly. Some discussion of Rollins et al., 1999 on the discovery of Nipped-B and its role in enhancer-promoter communication should also be made to reconcile their conclusions in the proposed absence of extrusion events.

Reviewer #3 (Public Review):

Bing et al. attempt to address fundamental mechanisms of TAD formation in Drosophila by analyzing gene expression and 3D conformation within the vicinity of the eve TAD after insertion of a transgene harboring a Homie insulator sequence 142 kb away in different orientations. These transgenes along with spatial gene expression analysis were previously published in Fujioka et al. 2016, and the underlying interpretations regarding resulting DNA configuration in this genomic region were also previously published. This manuscript repeats the expression analysis using smFISH probes in order to achieve more quantitative analysis, but the main results are the same as previously published. The only new data are the Micro-C and an additional modeling/analysis of what they refer to as the 'Z3' orientation of the transgenes. The rest of the manuscript merely synthesizes further interpretation with the goal of addressing whether loop extrusion may be occurring or if boundary:boundary pairing without loop extrusion is responsible for TAD formation. The authors conclude that their results are more consistent with boundary:boundary pairing and not loop extrusion; however, most of this imaging data seems to support both loop extrusion and the boundary:boundary models. This manuscript lacks support, especially new data, for its conclusions. Furthermore, there are many parts of the manuscript that are difficult to follow. There are some minor errors in the labelling of the figures that if fixed would help elevate understanding. Lastly, there are several major points that if elaborated on, would potentially be helpful for the clarity of the manuscript.

Major Points:

(1) The authors suggest and attempt to visualize in the supplemental figures, that loop extrusion mechanisms would appear during crosslinking and show as vertical stripes in the micro-C data. In order to see stripes, a majority of the nuclei would need to undergo loop extrusion at the same rate, starting from exactly the same spots, and the loops would also have to be released and restarted at the same rate. If these patterns truly result from loop extrusion, the authors should provide experimental evidence from another organism undergoing loop extrusion.<br /> (2) On lines 311-314, the authors discuss that stem-loops generated by cohesin extrusion would possibly be expected to have more next-next-door neighbor contacts than next-door neighbor contacts and site their models in Figure 1. Based on the boundary:boundary pairing models in the same figure would the stem-loops created by head-to-tail pairing also have the same phenotype? Making possible enrichment of next-next-door neighbor contacts possible in both situations? The concepts in the text are not clear, and the diagrams are not well-labeled relative to the two models.<br /> (3) The authors appear to cite Chen et al., 2018 as a reference for the location of these transgenes being 700nM away in a majority of the nuclei. However, the exact transgenes in this manuscript do not appear to have been measured for distance. The authors could do this experiment and include expression measurements.<br /> (4) The authors discuss the possible importance of CTCF orientation in forming the roadblock to cohesin extrusion and discuss that Homie orientation in the transgene may impact Homie function as an effective roadblock. However, the Homie region inserted in the transgene does not contain the CTCF motif. Can the authors elaborate on why they feel the orientation of Homie is important in its ability to function as a roadblock if the CTCF motif is not present? Trans-acting factors responsible for Homie function have not been identified and this point is not discussed in the manuscript.<br /> (5) The imaging results seem to be consistent with both boundary:boundary interaction and loop extrusion stem looping.<br /> (6) The authors suggest that the eveMa TAD could only be formed by extrusion after the breakthrough of Nhomie and several other roadblocks. Additionally, the overall long-range interactions with Nhomie appear to be less than the interactions with endogenous Homie (Figures 7, 8, and supplemental 5). Is it possible that in some cases boundary:boundary pairing is occurring between only the transgenic Homie and endogenous Homie and not including Nhomie?<br /> (7) In Figure 4E, the GFP hebe expression shown in the LhomieG Z5 transgenic embryo does not appear in the same locations as the LlambdaG Z5 control. Is this actually hebe expression or just a background signal?<br /> (8) Figure 6- The LhomieG Z3 late-stage embryo appears to be showing the ventral orientation of the embryo rather than the lateral side of the embryo as was shown in the previous figure. Is this for a reason? Additionally, there are no statistics shown for the Z3 transgenic images. Were these images analyzed in the same way as the Z5 line images?<br /> (9) Do the Micro-C data align with the developmental time points used in the smFISH probe assays?

Author response:

The following is the authors’ response to the original reviews.

Reviewer #1 (Public Review):

Summary:

The authors addressed how long-range interactions between boundary elements are established and influence their function in enhancer specificity. Briefly, the authors placed two different reporters separated by a boundary element. They inserted this construct ectopically ~140 kb away from an endogenous locus that contains the same boundary element. The authors used expression patterns driven by nearby enhancers as an output to determine which enhancers the reporters interact with. They complemented this analysis with 3D DNA contact mapping. The authors found that the orientation of the boundary element determined which enhancers each reporter interacted with. They proposed that the 3D interaction topology, whether being circular or stem configuration, distinguished whether the interaction was cohesin mediated or through an independent mechanism termed pairing.

Strengths:

The transgene expression assays are built upon prior knowledge of the enhancer activities. The 3D DNA contacts confirm that transgene expression correlates with the contacts. Using 4 different orientations covers all combinations of the reporter genes and the boundary placement.

Weaknesses:

The interpretation of the data as a refusal of loop extrusion playing a role in TAD formation is not warranted, as the authors did not deplete the loop extruders to show that what they measure is independent.

(1.1) To begin with, our findings do not exclude the possibility that cohesin loop extrusion has some sort of role in the formation or maintenance of TADs in flies or other aspects of chromosome structure.  On the other hand, it clearly is not determinative in defining the end-points of TADs or in generating the resulting topology (stem-loop or circle-loop).  Our main point, which we feel we have established unequivocally, is that it can’t explain many essential features of TADs or chromosome loops (see below) in Drosophila.  This reviewer agrees with this point in their next paragraph (below).  We also think that the loop extrusion model’s general acceptance as THE driving force behind TAD formation in mammals is unwarranted and not fully consistent with the available data, as explained below.

As to the reviewer’s specific point regarding depletion of loop extruders, we first note that completely eliminating factors encoding cohesin subunits in fly embryos isn’t readily feasible.  As cohesin is essential starting at the beginning of embryonic development, and is maternally deposited, knockdowns/depletions would likely be incomplete and there would always be some remaining activity.  As long as there is some residual activity—and no disruption in TAD formation is observed—this experimental test would be a failure.  In addition, any defects that are observed might arise not from a failure in TAD formation via loop extrusion but rather because the rapid mitotic cycles would be disrupted.  A far better approach would be to deplete/knockdown cohesin subunits in tissue culture cells, as there is no requirement for the cells to undergo embryonic development.  Moreover, since cell division is relatively slow, the depletion would likely eliminate much if not all of the activity before a checkpoint is reached.

While a drastic depletion of cohesin is not feasible in our model organism, we would draw the reviewer’s attention to an experiment of this type which has already been done in mammalian tissue culture cells by Goel et al. (Goel et al. 2023).  Unlike most Hi-C studies in mammals, the authors used region capture MicroC (RCMC).  In contrast to published genome-wide mammalian MicroC experiments (c.f., (Hsieh et al. 2020; Krietenstein et al. 2020)) which require large bin sizes to visualize mammalian “TADs,” the resolution of the experiments in Goel et al. (Goel et al. 2023) is similar to the resolution in our MicroC experiments (200-400 bp).  A MicroC contact map from Goel et al. shows the Pdm1g locus on chromosome 5 before and after Rad21 depletion.  The contact map visualizes a 250 kb DNA segment, which is only slightly larger than the ~230 kb DNA segment in Fig. 2C in our paper.

In this experiment, there was a 97% reduction in the amount of Rad21.  However, as can be seen by comparing the contact profiles above and below the diagonal, there is little or no difference in TAD organization after cohesin depletion when individual TADs are visualized with a bin size of 250 bp.  These results would indicate that mammalian TADs do not require cohesin.

Note also that the weak 45o stripes connecting different TADs (c.f. blue/green arrowheads) are still present after Rad21 depletion.  In the most popular version of the loop extrusion model, cohesin loads at a site(s) somewhere in the TAD-to-be, and then extrudes both strands until it bumps into CTCF roadblocks.  As illustrated in Figure Sup 2, this mechanism generates a vertical stripe originating at the cohesin loading site and extending until cohesin bumps into the left or right roadblock, at which point the stripe transitions into 45o stripe that ends when cohesin bumps into the other roadblock.  While 45o stripes are visible, there is no hint of a vertical stripe.  This suggests that the mechanism for generating stripes, if it is an active mechanism (rather than passive diffusion) may be quite different.  The 45o stripes must be generated by a factor(s) that is anchored to one (blue arrowhead) or both (green arrowhead) boundaries.  In addition, this factor, whatever it is, is not cohesin.  The reason for this is that the 45o stripes are present both before and after Rad21 depletion.  Moreover, if one were to imagine that the stripes represent a process involved in TAD formation, this process does not require cohesin (see Goel et al 2023).

It is worth noting another observation that is inconsistent with the cohesin loop extrusion/CTCF roadblock model for TAD formation/maintenance.  CTCF is not found at all of the TAD boundaries in this 250 kb DNA region.  This would suggest that there are other DNA binding proteins that have chromosomal architectural functions besides CTCF.  In flies, many of the chromosomal architectural proteins are, like CTCF, polydactyl zinc finger (PZF) proteins (Bonchuk et al. 2021; Bonchuk et al. 2022; Fedotova et al. 2017).  These include Su(Hw), CTCF, Pita, Zipic and CLAMP.  The PZF family in flies is quite large.  There are ~250 different PZF genes, and since only a handful of these have been characterized, it seems likely that additional members of this family will have architectural functions.  Thus far, only one boundary protein, CTCF, has received attention in studies on mammalian chromosome architecture.  As the mammalian genome is much larger and more complicated than the fly genome, it is difficult to believe that CTCF is the sole chromosomal architectural protein in mammals.  In this respect, it is worth noting that there are ~800 members of the PZF family in mammalian genomes (Fedotova et al. 2017).

Goel et al. (Goel et al. 2023) did observe alterations in the contact profiles after Rad21 depletion when they visualized the Ppm1g region at much lower resolution (bin sizes of 5 kb and 1 kb). The 5 kb bin size visualizes a region of ~1.2 Mb, while the 1 kb bin size visualizes a region that spans ~800 kb.  These large triangular units do not correspond to the individual TADs seen when Goel et al. visualized the Ppm1g locus at 250 bp resolution. 

Nor do they correspond to TADs in Fig. 2 of our paper.  Instead they represent TAD neighborhoods which, likely consist of 20-30 or more individual TADs.  Consequently the alterations in contact patterns seen after Rad21 depletion are occurring at the level of TAD neighborhoods.  This can be seen by comparing pixel density inside the blue lines before (above the diagonal) and after Rad21 depletion (below the diagonal) (Goel et al 2023).  The more distant contacts between individual TADs within this neighborhood are preferentially reduced by Rad21 depletion (the region below and to the left of the double arrowhead).  By contrast, the TADs themselves are unaffected, as are contacts between individual TADs and their immediate neighbors (see purple and light green asterisk).  The other interesting feature is the loss of contacts between what appears to be partially overlapping neighborhoods.  This loss of neighborhood-toneighborhood contacts can be seen in the region located between the green and blue lines.  The neighborhood that appears to partially overlap the Ppm1g neighborhood is outlined in purple.

It worth noting that, with the exception of the high resolution experiments in Goel et al., all of the other studies on cohesin (and CTCF) have examined the effects on contact maps within (and between) large neighborhoods (bin sizes >1 kb).  In most cases, these large neighborhoods are likely to be composed of many individual TADs like those seen in Goel et al. and in Fig. 2 of our paper.  We also observe larger neighborhoods in the fly genome, though they do not appear to be as large as those in mammals.  Our experiments do not address what role cohesin might have in facilitating contacts between more distant TADs located within the same neighborhoods, or between TADs in different neighborhoods, or whether loop extrusion is involved.

We would also note that the Drosophila DNA segment in Fig. 2C contains 35 different genes, while the mammalian DNA segment shown in Fig. 1 has only 9.  Thus, in this part of the fly genome, Pol II genes are more densely packed than in the mammalian DNA segment.  Much of the fly genome is also densely packed, and the size of individual TADs will likely be smaller, on average, than in mammals.  Nevertheless, the MicroC profiles are not all that different.  As is also common in flies, each TAD in the Ppm1g region only encompasses one or two genes.  Note also that there are no volcano triangles with plumes as would be predicted for TADs that have a stem-loop topology.

In fact, as shown in Author response image 1, the high-resolution contact profile for the Ppm1g region shows a strong resemblance to that observed for the fly Abd-B regulatory domains.  These regulatory domains are part of larger neighborhood that encompasses the abd-A and Abd-B genes and their regulatory domains.

Author response image 1.

Abd-B regulatory domains

As the authors show, the single long DNA loop mediated by cohesin loop extrusion connecting the ectopic and endogenous boundary is clearly inconsistent with the results, therefore the main conclusion of the paper that the 3D topology of the boundary elements a consequence of pairing is strong. However, the loop extrusion and pairing are not mutually exclusive models for the formation of TADs. Loop-extruding cohesin complexes need not make a 140 kb loop, multiple smaller loops could bring together the two boundary elements, which are then held together by pairing proteins that can make circular topologies.

(1.2) In the pairing model, distant boundaries bump into each other (by random walks or partially constrained walks), and if they are “compatible” they pair with each other, typically in an orientation-dependent manner.  As an alternative, the reviewer argues that cohesin need not make one large 140 kb loop.  Instead it could generate a series of smaller loops (presumably corresponding to the intervening TADs).  These smaller loops would bring homie in the transgene in close proximity to the eve locus so that it could interact with the endogenous homie and nhomie elements in the appropriate orientation, and in this way only one of the reporters would be ultimately activated.

There are two problems with the idea that cohesin-dependent loop extrusion brings transgene homie into contact with homie/nhomie in the eve locus by generating a series of small loops (TADs).  The first is the very large distances over which specific boundary:boundary pairing interactions can occur.  The second is that boundary:boundary pairing interactions can take place not only in cis, but also in trans.

We illustrate these points with several examples. 

Fujioka et al. 2016, Fig 7 shows an experiment in which attP sites located ~2 Mb apart were used to insert two different transgenes, one containing a lacZ reporter and the other containing the eve anal plate enhancer (AP) (Fujioka et al. 2016).  If the lacZ reporter and the AP transgenes also contain homie, the AP enhancer can activate lacZ expression (panel A,).  On the other hand, if one of the transgenes has lambda DNA instead of homie, no regulatory interactions are observed (panel A,).  In addition, as is the case in our experiments using the -142 kb platform, orientation matters.  In the combination on the top left, the homie boundary is pointing away from both the lacZ reporter and the AP enhancer.  Since homie pairs with itself head-tohead, pairing brings the AP enhancer into contact with the lacZ reporter.  A different result is obtained for the transgene pair in panel A on the top right.  In this combination, homie is pointing away from the lacZ reporter, while it is pointing towards the AP enhancer.  As a consequence, the reporter and enhancer are located on opposite sides of the paired homie boundaries, and in this configuration they are unable to interact with each other.

On the top left of panel B, the homie element in the AP enhancer transgene was replaced by a nhomie boundary oriented so that it is pointing towards the enhancer.  Pairing of homie and nhomie head-to-tail brings the AP enhancer in the nhomie transgene into contact with the lacZ reporter in the homie transgene, and it activates reporter expression.  Finally, like homie, nhomie pairs with itself head-to-head, and when the nhomie boundaries are pointing towards both the AP reporter and the lacZ reporter, reporter expression is turned on.

Long distance boundary-dependent pairing interactions by the bithorax complex Mcp boundary have also been reported in several papers.  Fig. 6 from Muller et al. (Muller et al. 1999) shows the pattern of regulatory interactions (in this case PRE-dependent “pairing-sensitive silencing”) between transgenes that have a mini-white reporter, the Mcp and scs’ boundaries and a PRE that is located close to Mcp.  In this experiment flies carrying transgenes inserted at the indicated sites on the left and right arms of the 3rd chromosome were mated in pairwise combinations, and their trans-heterozygous progeny examined for pairing-sensitive silencing of the mini-white reporter.

Two examples of long-distance pairing-sensitive silencing mediated by Mcp/scs’ are shown in Fig. 5b from Muller et al. 1999.  The transgene inserts in panel A are w#12.43 and ff#10.5.  w#12.43 is inserted close to the telomere of 3R at 99B.  ff10.5 is inserted closer to the middle of 3R at 91A.  The estimate distance between them is 11.3 Mb.  The transgene inserts in panel B are ff#10.5 and ff#11.102.  ff#11.102 is inserted at 84D, and the distance between them is 11 Mb.  Normally, the eye color phenotype of the mini-white reporter is additive: homozygyous inserts have twice as dark eye color as hemizygous inserts, while in trans-_heterozygous flies the eye color would be the sum of the two different transgenes.  However, when a PRE is present and the transgene can pair, silencing is observed.  In panel A, the t_rans-_heterozygous combination has a lighter eye color than either of the parents.  In panel B, the _trans-_heterozygous combination is darker than one of the parents (_ff#10.5) but much lighter than the other (ff#11.102).

All ten of the transgenes tested were able to engage in long distance (>Mbs) trans_regulatory interactions; however, likely because of how the chromosome folds on the Mb scale (e.g., the location of meta-loops: see #2.1 and Author response image 3) not all of the possible pairwise silencing interactions are observed.  The silencing interactions shown in Muller et.al. are between transgenes inserted on different homologs.  _Mcp/scs'-dependent silencing interactions can also occur in cis. Moreover, just like the homie and nhomie experiments described above, Muller et.al. (Muller et al. 1999) found that Mcp could mediate long-distance activation of mini-white and yellow by their respective enhancers.

The pairing-sensitive activity of the PRE associated with the Mcp boundary is further enhanced when the mini-white transgene has the scs boundary in addition to Mcp and scs’.  In the experiment shown in Fig. 8 from Muller et al. 1999, the pairing-sensitive silencing interactions of the Mcp/scs’/scs transgene are between transgenes inserted on different chromosomes.  Panel A shows pairing-sensitive silencing between w#15.60, which is on the X chromosome, and w#15.102, which is on the 2nd chromosome.  Panel B shows pairing-sensitive silencing between the 2nd chromosome insert w#15.60 and a transgene, w#15.48, which is inserted on the 3rd chromosome.

The long-distance trans and cis interactions described here are not unique to homie, nhomie, Mcp, scs’, or scs.  Precisely analogous results have been reported by Sigrist and Pirrotta (Sigrist and Pirrotta 1997) for the gypsy boundary when the bxd PRE was included in the mini-white transgene.  Also like the Mcp-containing transgenes in Muller et al. (Muller et al. 1999), Sigrist and Pirrotta observed pairing-sensitive silencing between gypsy bxd_PRE _mini-white transgenes inserted on different chromosomes.  Similar long-distance (Mb) interactions have been reported for Fab-7 (Bantignies et al. 2003; Li et al. 2011).  In addition, there are examples of “naturally occurring” long-distance regulatory and/or physical interactions.  One would be the regulatory/physical interactions between the p53 enhancer upstream of reaper and Xrp1 which was described by Link et al. (Link et al. 2013).  Another would be the nearly 60 meta-loops identified by Mohana et al. (Mohana et al. 2023).

Like homie at -142 kb, the regulatory interactions (pairing-sensitive silencing and enhancer activation of reporters) reported in Muller et al. (Muller et al. 1999) involve direct physical interactions between the transgenes.  Vazquez et al. (Vazquez et al. 2006) used the lacI/lacO system to visualize contacts between distant scs/Mcp/scs’-containing transgenes in imaginal discs.  As indicated in Vasquez et al. 2006, Table 3 lines #4-7,  when both transgenes have Mcp and were inserted on the same chromosome, they colocalized in trans-_heterozygotes (single dot) in 94% to 97% of the disc nuclei in the four pairwise combinations they tested.  When the transgenes both lacked _Mcp (Vasquez et al. 2006, Table 3 #1), co-localization was observed in 4% of the nuclei.  When scs/Mcp/scs’-containing transgenes on the 2nd and 3rd chromosome were combined (Vasquez et al. 2006, Table 3 #8), colocalization was observed in 96% of the nuclei.  They also showed that four different scs/Mcp/scs’ transgenes (two at the same insertion site but on different homologs, and two at different sites on different homologs) co-localized in 94% of the eye imaginal disc nuclei (Vasquez et al. 2006, Table 3 #9).  These pairing interactions were also found to be stable over several hours.  Similar co-localization experiments together with 3C were reported by Li et al. (Li et al. 2011).

The de novo establishment of trans interactions between compatible boundary elements has been studied by Lim et al. (Lim et al. 2018).  These authors visualized transvection (enhancer activation of a MS2 loop reporter in trans) mediated by the gypsy insulator, homie and Fab-8  in NC14 embryos.  When both transgenes shared the same boundary element, transvection/physical pairing was observed in a small subset of embryos.  The interactions took place after a delay and increased in frequency as the embryo progressed into NC14.  As expected, transvection was specific: it was not observed when the transgenes had different boundaries.  For homie it was also orientation-dependent.  It was observed when homie was orientated in the same direction in both transgenes, but not when homie was orientated in opposite directions in the two transgenes.

While one could imagine that loop extrusion-dependent compaction of the chromatin located between eve and the transgene at -142 kb into a series of small loops (the intervening TADs) might be able to bring homie in the transgene close to homie/nhomie in the eve locus, there is no cohesinbased loop extrusion scenario that would bring transgenes inserted at sites 6 Mb, 11 Mb, on different sides of the centromere, or at opposite ends of the 3rd chromosome together so that the distant boundaries recognize their partners and physically pair with each other.  Nor is there a plausible cohesin-based loop extrusion mechanism that could account for the fact that most of the documented long-distance interactions involve transgenes inserted on different homologs.  This is not to mention the fact that long-distance interactions are also observed between boundarycontaining transgenes inserted on different chromosomes.

In fact, given these results, one would logically come to precisely the opposite conclusion.  If boundary elements inserted Mbs apart, on different homologs and on different chromosomes can find each other and physically pair, it would be reasonable to think that the same mechanism (likely random collisions) is entirely sufficient when they are only 142 kb apart.

Yet another reason to doubt the involvement or need for cohesin-dependent loop extrusion in bringing the transgene homie in contact with the eve locus comes from the studies of Goel et al. (Goel et al. 2023).  They show that cohesin has no role in the formation of TADs in mammalian tissue culture cells.  So if TADs in mammals aren’t dependent on cohesin, there would not be a good reason to think at this point that the loops (TADs) that are located between eve and the transgene are generated by, or even strongly dependent on, cohesin-dependent loop extrusion.

It is also important to note that even if loop-extrusion were to contribute to chromatin compaction in this context and make the looping interactions that lead to orientation-specific pairing more efficient, the role of loop extrusion in this model is not determinative of the outcome, it is merely a general compaction mechanism.  This is a far cry from the popular concept of loop extrusion as being THE driving force determining chromosome topology at the TAD level.

Reviewer #2 (Public Review):

In Bing et al, the authors analyze micro-C data from NC14 fly embryos, focusing on the eve locus, to assess different models of chromatin looping. They conclude that fly TADs are less consistent with conventional cohesin-based loop extrusion models and instead rely more heavily on boundaryboundary pairings in an orientation-dependent manner.

Overall, I found the manuscript to be interesting and thought-provoking. However, this paper reads much more like a perspective than a research article. Considering eLIFE is aimed at the general audience, I strongly suggest the authors spend some time editing their introduction to the most salient points as well as organizing their results section in a more conventional way with conclusion-based titles. It was very difficult to follow the authors' logic throughout the manuscript as written. It was also not clear as written which experiments were performed as part of this study and which were reanalyzed but published elsewhere. This should be made clearer throughout.

It has been shown several times that Drosophila Hi-C maps do not contain all of the features (frequent corner peaks, stripes, etc.) observed when compared to mammalian cells. Considering these features are thought to be products of extrusion events, it is not an entirely new concept that Drosophila domains form via mechanisms other than extrusion.

(2.1) While there are differences between the Hi-C contact profiles in flies and mammals, these differences likely reflect in large part the bin sizes used to visualize contact profiles.  With the exception of Goel et al. (Goel et al. 2023), most of the mammalian Hi-C studies have been low resolution restriction enzyme-based experiments, and required bin sizes of >1 kb or greater to visualize what are labeled as  “TADs.”  In fact, as shown by experiments in Goel et al., these are not actually TADs, but rather a conglomeration of multiple TADs into a series of TAD neighborhoods.  The same is true for the MicroC experiments of Krietenstein et al. and Hsieh et al. on human and mouse tissue culture cells (Hsieh et al. 2020; Krietenstein et al. 2020).  This is shown in Author response image 2.  In this image, we have compared the MicroC profiles generated from human and mouse tissue culture cells with fly MicroC profiles at different levels of resolution.

For panels A-D, the genomic DNA segments shown are approximately 2.8 Mb, 760 kb, 340 kb, and 190 kb.  For panels E-H, the genomic DNA segments shown are approximately 4.7 Mb, 870 kb, 340 kb and 225 kb.  For panels I-L, the genomic DNA segments shown are approximately 3 Mb, 550 kb, 290 kb and 175 kb.

As reported for restriction enzyme-based Hi-C experiments, a series of stripes and dots are evident in mammalian MicroC profiles.  In the data from Krietenstein et al., two large TAD “neighborhoods” are evident with a bin size of 5 kb, and these are bracketed by 45o stripes (A: black arrows).  At 1 kb (panel B), the 45o stripe bordering the neighborhood on the left no longer defines the edge of the neighborhood (blue arrow: panel B), and both stripes become discontinuous (fuzzy dots).  At 500 (panel C) and 200 bp (panel D) bin sizes, the stripes largely disappear (black arrows) even though they were the most prominent feature in the TAD landscape with large bin sizes.  At 200 bp, the actual TADs (as opposed to the forest) are visible, but weakly populated.  There are no stripes, and only one of the TADs has an obvious “dot” (green asterisk: panel C).

Author response image 2.

Mammalian MicroC profiles different bin sizes.

Large TAD neighborhoods bordered by stripes are also evident in the Hsieh et al. data set in Author response image 2 panels E and F (black arrows in E and F and green arrow in F).  At 400 bp resolution (panel G), the narrow stripe in panel F (black arrows) becomes much broader, indicating that it is likely generated by interactions across one or two small TADs that can be discerned at 200 bp resolution.  The same is true for the broad stripe indicated by the green arrows in panels F, G and H.  This stripe arises from contacts between the TADs indicated by the red bar in panels G and H and the TADs to the other side of the volcano triangle with a plume (blue arrow in panel H).  As in flies, we would expect that this volcano triangle topped by a plume corresponds to a stem-loop.  However, the resolution is poor at 200 bp, and the profiles of the neighboring TADs are not very distinct.

For the fly data set, stripes can be discerned when analyzed at 800 bp resolution (see arrows in Author response image 3);  however, these stripes are flanked by regions of lower contact, and represent TAD-TAD interactions.  At 400 bp, smaller neighborhoods can be discerned, and these neighborhoods exhibit a complex pattern of interaction with adjacent neighborhoods.  With bin sizes of 200 bp, individual TADs are observed, as are TAD-TAD interactions like those seen near eve.  Some of the TADs have dots at their apex, while others do not—much like what is seen in the mammalian MicroC studies.

Author response image 3.

Mammalian MicroC profiles different bin sizes.

Stripes: As illustrated in Author response image 2 A-D and E-H, the continuous stripes seen in low resolution mammalian studies (>1 kb bins) would appear to arise from binning artefacts.  At high resolution where single TADs are visible, the stripes seem to be generated by TAD-TAD interactions, and not by some type of “extrusion” mechanism.  This is most clearly seen for the volcano with plume TAD in Author response inage 2 G and H.  While stripes in Author response image 2 disappear at high resolution, this is not always true.  There are stripes that appear to be “real” in Geol et al. 2023 for the TADs in the Ppm1g region, and in Author response image 1 for the Abd-B regulatory domain TADs.  Since the stripes in the Ppm1g region are unaffected by Rad21 depletion, some other mechanism must be involved (c.f. (Shidlovskii et al. 2021)).

Dots: The high resolution images of mammalian MicroC experiments in Author response image 2D and H show that, like Drosophila (Author response image 3L), mammalian TADs don’t always have a “dot” at the apex of the triangle.  This is not surprising.  In the MicroC procedure, fixed chromatin is digested to mononucleosomes with MNase.  Since most TAD boundaries in flies, and presumably also in mammals, are relatively large (150-400 bp) nuclease hypersensitive regions, extensive MNase digestion will typically reduce the boundary element sequences to oligonucleotides.

In flies, the only known sequences (at least to date) that end up giving dots (like those seen in Author response image 1) are bound by a large (>1,000 kd) GAF-containing multiprotein complex called LBC.  In the Abd-B region of BX-C, LBC binds to two ~180 bp sequences in Fab-7 (dHS1 and HS3: (Kyrchanova et al. 2018; Wolle et al. 2015), and to the centromere proximal (CP) side of Fab-8.  The LBC elements in Fab-7 (dHS1) and Fab-8 (CP) have both blocking and boundary bypass activity (Kyrchanova et al. 2023; Kyrchanova et al. 2019a; Kyrchanova et al. 2019b; Postika et al. 2018).  Elsewhere, LBC binds to the bx and bxd PREs in the Ubx regulatory domains, to two PREs upstream of engrailed, to the hsp70 promoter, the histone H3-H4 promoters, and the eve promoter (unpublished data).  Based on ChIP signatures, it likely binds to most PREs/tethering elements in the fly genome (Batut et al. 2022; Li et al. 2023).  Indirect end-labeling experiments (Galloni et al. 1993; Samal et al. 1981; Udvardy and Schedl 1984) indicate that LBC protects an ~150-180 bp DNA segment from MNase digestion, which would explain why LBC-bound sequences are able to generate dots in MicroC experiments.  Also unlike typical boundary elements, the pairing interactions of the LBC elements we’ve tested appear to be orientation-independent (unpublished data).

The difference in MNase sensitivity between typical TAD boundaries and LBC-bound elements is illustrated in the MicroC of the Leukocyte-antigen-related-like (Lar) meta-loop in Author response image 4 panels A and B.  Direct physical pairing of two TAD boundaries (blue and purple) brings two TADs encompassing the 125 kb lar gene into contact with two TADs in a gene poor region 620 kb away.  This interaction generates two regions of greatly enhanced contact: the two boxes on either side of the paired boundaries (panel A).  Note that like transgene homie pairing with the eve boundaries, the boundary pairing interaction that forms the lar meta-loop is orientation-dependent.  In this case the TAD boundary in the Lar locus pairs with the TAD boundary in the gene poor region head-to-head (arrow tip to arrow tip), generating a circle-loop.  This circle-loop configuration brings the TAD upstream of the blue boundary into contact with the TAD upstream of the purple boundary.  Likewise, the TAD downstream of the blue boundary is brought into contact with the TAD downstream of the purple boundary.

In the MicroC procedure, the sequences that correspond to the paired boundaries are not recovered (red arrow in Author response image 4 panel B).  This is why there are vertical and horizontal blank stripes (red arrowheads) emanating from the missing point of contact.  Using a different HiC procedure (dHS-C) that allows us to recover sequences from typical boundary elements (Author response image 4 panels C and D), there is a strong “dot” at the point of contact which corresponds to the pairing of the blue and purple boundaries.

There is a second dot (green arrow) within the box that represents physical contacts between sequences in the TADs downstream of the blue and purple boundaries.  This dot is resistant to MNase digestion and is visible both in the MicroC and dHS-C profiles.  Based on the ChIP signature of the corresponding elements in the two TADs downstream of the blue and purple boundaries, this dot represents paired LBC elements.

Author response image 4.

Lar metaloop. Panels A & bB: MicroC. Panels C & D: dHS-C

That being said, the authors' analyses do not distinguish between the formation and the maintenance of domains. It is not clear to this reviewer why a single mechanism should explain the formation of the complex structures observed in static Hi-C heatmaps from a population of cells at a single developmental time point. For example, how can the authors rule out that extrusion initially provides the necessary proximity and possibly the cis preference of contacts required for boundaryboundary pairing whereas the latter may more reflect the structures observed at maintenance?

(2.2) The MicroC profiles shown in Fig. 2 of our paper were generated from nuclear cycle (NC) 14 embryos.  NC14 is the last nuclear cycle before cellularization (Foe 1989).  After the nuclei exit mitosis, S-phase begins, and because satellite sequences are late replicating in this nuclear cycle, S phase lasts 50 min instead of only 4-6 min during earlier cycles (Shermoen et al. 2010).  So unlike MicroC studies in mammals, our analysis of chromatin architecture in NC14 embryos likely offers the best opportunity to detect any intermediates that are generated during TAD formation.  In particular, we should be able to observe evidence of cohesin linking the sequences from the two extruding strands together (the stripes) as it generates TADs de novo.  However, there are no vertical stripes in the eve TAD as would be expected if cohesin entered at a few specific sites somewhere within the TAD and extruded loops in opposite directions synchronously, nor are their stripes at 45o as would be expected if it started at nhomie or homie (see Figure Supplemental 1).  We also do not detect cohesin-generated stripes in any of the TADs in between eve and the attP site at -142 kb. Note that in some models, cohesin is thought to be continuously extruding loops. After hitting the CTCF roadblocks, cohesin either falls off after a short period and starts again or it breaks through one or more TAD boundaries generating the LDC domains. In this dynamic model, stripes of crosslinked DNA generated by the passing cohesin complex should be observed throughout the cell cycle.  They are not. 

As for formation versus maintenance, and the possible involvement of cohesin loop extrusion in the former, but not the latter:  This question was indirectly addressed in point #1.2 above.  In this point we described multiple examples of specific boundary:boundary pairing interactions that take place over Mbs, in cis and in trans and even between different chromosomes.  These long-distance interactions don’t preexist;  instead they must be established de novo and then maintained.  This process was actually visualized in the studies of Lim et al. (Lim et al. 2018) on the establishment of trans boundary pairing interactions in NC14 embryos.  There is no conceivable mechanism by which cohesin-based loop extrusion could establish the long or short distance trans interactions that have been documented in many studies on fly boundary elements.  Also as noted above, its seems unlikely that it is necessary for long-range interactions in cis.  

A more plausible scenario is that cohesin entrapment helps to stabilize these long-distance interactions after they are formed.  If this were true, then one could argue that cohesin might also function to maintain TADs after boundaries have physically paired with their neighbors in cis.  However, the Rad21 depletion experiments of Goel et al. (Goel et al. 2023) would rule out an essential role for cohesin in maintaining TADs after boundary:boundary pairing.  In short, while we cannot formally rule out that loop extrusion might help bring sequences closer together to increase their chance of pairing, neither the specificity of that pairing, nor its orientation can be explained by loop extrusion.  Furthermore, since pairing in trans cannot be facilitated by loop extrusion, invoking it as potentially important for boundary-boundary pairing in cis can only be described as a potential mechanism in search of a function, without clear evidence in its favor.

On the other hand, the apparent loss of contacts between TADs within large multi-TAD neighborhoods (Geol et al. 2023) would suggest that there is some sort of decompaction of neighborhoods after Rad21 depletion.  It is possible that this might stress interactions that span multiple TADs as is the case for homie at -142, or for the other examples described in #1.2 above.  This kind of involvement of cohesin might or might not be associated with a loop extrusion mechanism.

Future work aimed at analyzing micro-C data in cohesin-depleted cells might shed additional light on this.

(2.3) This experiment has been done by Goel et al. (Goel et al. 2023) in mammalian tissue culture cells.  They found that TADs, as well as local TAD neighborhoods, are not disrupted/altered by Rad21 depletion (see Geol at al. 2023 and our response to point #1.1 of reviewer #1).

Additional mechanisms at play include compartment-level interactions driven by chromatin states. Indeed, in mammalian cells, these interactions often manifest as a "plume" on Hi-C maps similar to what the authors attribute to boundary interactions in this manuscript. How do the chromatin states in the neighboring domains of the eve locus impact the model if at all?

(2.4) Chromatin states have been implicated in driving compartment level interactions. 

Compartments as initially described were large, often Mb sized, chromosomal segments that “share” similar chromatin marks/states, and are thought to merge via co-polymer segregation.  They were visualized using large multi-kb bin sizes.  In the studies reported here, we use bin sizes of 200 bp to examine a DNA segment of less than 200 kb which is subdivided into a dozen or so small TADs.  Several of the TADs contain more than one transcription unit, and they are expressed in quite different patterns, and thus might be expected to have different “chromatin states” at different points in development and in different cells in the organism. However, as can be seen by comparing the MicroC patterns in our paper that are shown in Fig. 2 with Fig. 7, Figure Supplemental 5 and Figure Supplemental 6, the TAD organization in NC14 and 12-16 hr embryos is for the most part quite similar.  There is no indication that these small TADs are participating in liquid phase compartmentalization that depends upon shared chromatin/transcriptional states in NC14 and then again in 12-16 hr embryos. 

In NC14 embryos, eve is expressed in 7 stripes, while it is potentially active throughout much of the embryo.  In fact, the initial pattern in early cycles is quite broad and is then refined during NC14.  In 12-16 hr embryos, the eve gene is silenced by the PcG system in all but a few cells in the embryo.  However, here again the basic structure of the TAD, including the volcano plume, looks quite similar at these different developmental stages.  

As for the suggestion that the plume topping the eve volcano triangle is generated because the TADs flanking the eve TAD share chromatin states and coalesce via some sort of phase separation:

This model has been tested directly in Ke et al. (Ke et al. 2024).  In Ke et al., we deleted the nhomie boundary and replaced it with either nhomie in the reverse orientation or homie in the forward orientation.  According to the compartment model, changing the orientation of the boundaries so that the topology of the eve TAD changes from a stem-loop to a circle-loop should have absolutely no effect on the plume topping the eve volcano triangle.  The TADs flanking the eve TAD would still be expected to share the same chromatin states and would still be able to coalesce via phase transition.  However, this is not what is observed.  The plume disappears and is replaced by “clouds” on both sides of the eve TAD. The clouds arise because the eve TAD bumps into the neighboring TADs when the topology is a circle-loop.  

We would also note that “compartment-level” interactions would not explain the findings presented in Muller at al. 1999, in Table 1 or in Author response image 4.  It is clear that the long distant (Mb) interactions observed for Mcp, gypsy, Fab-7, homie, nhomie and the blue and purple boundaries in Author response image 4 arise by the physical pairing of TAD boundary elements.  This fact is demonstrated directly by the MicroC experiments in Fig. 7 and Fig Supplemental 4 and 5, and by the MicroC and dHS-C experiments in Author response image 4.  There is no evidence for any type of “compartment/phase separation” driving these specific boundary pairing interactions.

In fact, given the involvement of TAD boundaries in meta-loop formation, one might begin to wonder whether some of the “compartment level interactions” are generated by the specific pairing of TAD boundary elements rather than by “shared chromatin” states.  For example, the head-tohead pairing of the blue and purple boundaries generates a Lar meta-loop that has a circle-loop topology.  As a consequence, sequences upstream of the blue and purple boundary come into contact, generating the small dark rectangular box on the upper left side of the contact map.  Sequences downstream of the blue and purple boundary also come into contact, and this generates the larger rectangular box in the lower right side of the contact map.  A new figure, Fig. 9, shows that the interaction pattern flips (lower left and top right) when the meta-loop has a stem-loop topology.  If these meta-loops are visualized using larger bin sizes, the classic “compartment” patchwork pattern of interactions emerges.  Would the precise patchwork pattern of “compartmental” interactions involving the four distant TADs that are linked in the two meta-loops shown in Fig. 9 persist as is if we deleted one of the TAD boundaries that forms the meta-loop?  Would the precise patchwork pattern persist if we inverted one of the meta-loop boundaries so that we converted the topology of the loop from a circle-loop to a stem-loop or vice versa?  We haven’t used MicroC to compare the compartment organization after deleting or inverting a meta-loop TAD boundary; however, a comparison of the MicroC pattern in WT in Fig. 1C with that for the homie transgenes in Fig. 7 and Figs. Supplemental 5, 6 and 7 indicates a) that novel patterns of TAD:TAD interactions are generated by this homie dependent mini-meta-loop and b) that the patterns of TAD:TAD interactions depend upon loop topology. Were these novel TAD:TAD interactions generated instead by compartment level interactions/shared chromatin states, they should be evident in WT as well (Fig. 1).  They are not.

How does intrachromosomal homolog pairing impact the models proposed in this manuscript (Abed et al. 2019; Erceg et al., 2019). Several papers recently have shown that somatic homolog pairing is not uniform and shows significant variation across the genome with evidence for both tight pairing regions and loose pairing regions. Might loose pairing interactions have the capacity to alter the cis configuration of the eve locus?

(2.5) At this point it is not entirely clear how homolog pairing impacts the cis configuration/MicroC contact maps.  We expect that homolog pairing is incomplete in the NC14 embryos we analyzed;  however, since replication of eve and the local neighborhood is likely complete, sister chromosomes should be paired.  So we are likely visualizing the 3D organization of paired TADs.

In summary, the transgenic experiments are extensive and elegant and fully support the authors' models. However, in my opinion, they do not completely rule out additional models at play, including extrusion-based mechanisms. Indeed, my major issue is the limited conceptual advance in this manuscript. The authors essentially repeat many of their previous work and analyses.

(2.6) In our view, the current paper makes a number of significant contributions that go well beyond those described in our 2016 publication.  These are summarized below.

A) While our 2016 paper used transgenes inserted in the -142 kb attP site to study pairing interactions of homie and nhomie, we didn’t either consider or discuss how our findings might bear on the loop extrusion model.  However, since the loop extrusion model is currently accepted as established fact by many labs working on chromosome structure, it is critically important to devise experimental approaches which test the predictions of this particular model.  One approach would be to deplete cohesin components; however, as discussed in #1.1, our experimental system is not ideal for this type of approach.  On the other hand, there are other ways to test the extrusion model.  Given the mechanism proposed for TAD formation—extruding a loop until cohesin bumps into CTCF/boundary road blocks—it follows that only two types of loop topologies are possible: stemloop and unanchored loop.  The loop extrusion model, as currently conceived, can’t account for the two cases in this study in which the reporter on the wrong side of the homie boundary from the eve locus is activated by the eve enhancers.  In contrast, our findings are completely consistent with orientation-specific boundary:boundary pairing.

B) In the loop extrusion model, cohesin embraces both of the extruded chromatin fibers, transiently bringing them into close proximity.  As far as we know, there have been no (high resolution) experiments that have actually detected these extruding cohesin complexes during TAD formation.  In order to have a chance of observing the expected signatures of extruding cohesin complexes, one would need a system in which TADs are being formed.  As described in the text, this is why we used MicroC to analyze TADs in NC14 embryos.  We do not detect the signature stripes that would be predicted (see Figure Supp 2) by the current version of the loop extrusion model.

C) Reporter expression in the different -142 kb transgenes provides only an indirect test of the loop extrusion and boundary:boundary pairing models for TAD formation.  The reporter expression results need to be confirmed by directly analyzing the pattern of physical interactions in each instance.  While we were able to detect contacts between the transgenes and eve in our 2016 paper, the 3C experiments provided no information beyond that.  By contrast, the MicroC experiments in the current paper give high resolution maps of the physical contacts between the transgene and the eve TAD.  The physical contacts track completely with reporter activity.  Moreover, just as is the case for reporter activity, the observed physical interactions are inconsistent with the loop extrusion model.

D) Genetic studies in Muller et al. (Muller et al. 1999) and imaging in Vazquez et al. (Vazquez et al. 2006) suggested that more than two boundaries can participate in pairing interactions.  Consistent with these earlier observations, viewpoint analysis indicates the transgene homie interacts with both eve boundaries.  While this could be explained by transgene homie alternating between nhomie and homie in the eve locus, this would require the remodeling of the eve TAD each time the pairing interaction switched between the three boundary elements.  Moreover, two out of the three possible pairing combinations would disrupt the eve TAD, generating an unanchored loop (c.f., the lambda DNA TAD in Ke et al., (Ke et al. 2024)).  However, the MicroC profile of the eve TAD is unaffected by transgenes carrying the homie boundary.  This would suggest that like Mcp, the pairing interactions of homie and nhomie might not be exclusively pairwise.  In this context is interesting to compare the contact profiles of the lar meta-loop shown in Author response image 4 with the different 142 kb homie inserts.  Unlike the homie element at -142 kb, there is clearly only a single point of contact between the blue and purple boundaries.

E) Chen et al. (Chen et al. 2018) used live imaging to link physical interactions between a homie containing transgene inserted at -142 kb and the eve locus to reporter activation by the eve enhancers.  They found that the reporter was activated by the eve enhancers only when it was in “close proximity” to the eve gene.  “Close proximity” in this case was 331 nM.  This distance is equivalent to ~1.1 kb of linear duplex B form DNA, or ~30 nucleosome core particles lined up in a row.  It would not be possible to ligate two DNAs wrapped around nucleosome core particles that are located 330 nM apart in a fixed matrix.  Since our MicroC experiments were done on embryos in which the gene is silent in the vast majority of cells, it is possible that the homie transgene only comes into close enough proximity for transgene nucleosome: eve nucleosome ligation events when the eve gene is off.  Alternatively, and clearly more likely, distance measurements using imaging procedures that require dozens of fluorescent probes may artificially inflate the distance between sequences that are actually close enough for enzymatic ligation.

F) The findings reported in Goel et al. (Goel et al. 2023) indicate that mammalian TADs don’t require cohesin activity; however, the authors do not provide an alternative mechanism for TAD formation/stability.  Here we have suggested a plausible mechanism.

The authors make no attempt to dissect the mechanism of this process by modifying extrusion components directly.

(2.7) See point #1.1

Some discussion of Rollins et al. on the discovery of Nipped-B and its role in enhancer-promoter communication should also be made to reconcile their conclusions in the proposed absence of extrusion events.

(2.8) The reason why reducing nipped-B activity enhances the phenotypic effects of gypsy-induced mutations is not known at this point; however, the findings reported in Rollins et al. (Rollins et al. 1999) would appear to argue against an extrusion mechanism for TAD formation.

Given what we know about enhancer blocking and TADs, there are two plausible mechanisms for how the Su(Hw) element in the gypsy transposon blocks enhancer-promoter interactions in the gypsy-induced mutants studied by Rollins et al.  First, the Su(Hw) element could generate two new TADs through pairing interactions with boundaries in the immediate neighborhood.  This would place the enhancers in one TAD and the target gene in another TAD.  Alternatively, the studies of Sigrist and Pirrotta (Sigrist and Pirrotta 1997) as well as several publications from Victor Corces’ lab raise the possibility that the Su(Hw) element in gypsy-induced mutations is pairing with gypsy transposons inserted elsewhere in the genome.  This would also isolate enhancers from their target genes.  In either case, the loss of nipped-B activity increases the mutagenic effects of Su(Hw) element presumably by strengthening its boundary function.  If this is due to a failure to load cohesin on to chromatin, this would suggest that cohesin normally functions to weaken the boundary activity of the Su(Hw) element, i.e., disrupting the ability of Su(Hw) elements to interact with either other boundaries in the neighborhood or with themselves.  Were this a general activity of cohesin (to weaken boundary activity), one would imagine that cohesin normally functions to disrupt TADs rather than generate/stabilize TADs.

An alternative model is that Nipped-B (and thus cohesion) functions to stabilize enhancerpromoter interactions within TADs.  In this case, loss of Nipped-B would result in a destabilization of the weak enhancer:promoter interactions that can still be formed when gypsy is located between the enhancer and promoter.  In this model the loss of these weak interactions in nipped-b mutants would appear to increase the “blocking” activity of the gypsy element.  However, this alternative model would also provide no support for the notion that Nipped-B and cohesin function to promote TAD formation.

Reviewer #3 (Public Review):

Bing et al. attempt to address fundamental mechanisms of TAD formation in Drosophila by analyzing gene expression and 3D conformation within the vicinity of the eve TAD after insertion of a transgene harboring a Homie insulator sequence 142 kb away in different orientations. These transgenes along with spatial gene expression analysis were previously published in Fujioka et al. 2016, and the underlying interpretations regarding resulting DNA configuration in this genomic region were also previously published. This manuscript repeats the expression analysis using smFISH probes in order to achieve more quantitative analysis, but the main results are the same as previously published. The only new data are the Micro-C and an additional modeling/analysis of what they refer to as the 'Z3' orientation of the transgenes. The rest of the manuscript merely synthesizes further interpretation with the goal of addressing whether loop extrusion may be occurring or if boundary:boundary pairing without loop extrusion is responsible for TAD formation. The authors conclude that their results are more consistent with boundary:boundary pairing and not loop extrusion; however, most of this imaging data seems to support both loop extrusion and the boundary:boundary models. This manuscript lacks support, especially new data, for its conclusions.

(3.1) The new results/contributions of our paper are described in #2.6 above. 

Although there are (two) homie transgene configurations that give expression patterns that would be consistent with the loop extrusion model, that is not quite the same as strong evidence supporting loop extrusion.  On the contrary, key aspects of the expression data are entirely inconsistent with loop extrusion, and they thus rule out the possibility that loop extrusion is sufficient to explain the results.  Moreover, the conclusions drawn from the expression patterns of the four transgenes are back up by the MicroC contact profiles—profiles that are also not consistent with the loop extrusion model.  Further, as documented above, loop extrusion is not only unable to explain the findings reported in this manuscript, but also the results from a large collection of published studies on fly boundaries.  Since all of these boundaries function in TAD formation, there is little reason to think that loop extrusion makes a significant contribution at the TAD level in flies.   Given the results reported by Goel et al. (Goel et al. 2023), one might also have doubts about the role of loop extrusion in the formation/maintenance of mammalian TADs. 

To further document these points, we’ve included a new figure (Fig. 9) that shows two meta-loops.  Like the loops seen for homie-containing transgenes inserted at -142 kb, meta-loops are formed by the pairing of distant fly boundaries.  As only two boundaries are involved, the resulting loop topologies are simpler than those generated when transgene homie pairs with nhomie and homie in the eve locus.  The meta-loop in panel B is a stem-loop.  While a loop with this topology could be formed by loop extrusion, cohesion would have to break through dozens of intervening TAD boundaries and then somehow know to come to a halt at the blue boundary on the left and the purple boundary on the right.  However, none of the mechanistic studies on either cohesin or the mammalian CTCF roadblocks have uncovered activities of either the cohesin complex or the CTCF roadblocks that could explain how cohesin would be able to extrude hundreds of kb and ignore dozens of intervening roadblocks, and then stop only when it encounters the two boundaries that form the beat-IV meta-loop.  The meta-loop in panel A is even more problematic in that it is a circle-loop--a topology that can’t be generated by cohesin extruding a loop until comes into contact with CTCF roadblocks on the extruded strands.

Furthermore, there are many parts of the manuscript that are difficult to follow. There are some minor errors in the labelling of the figures that if fixed would help elevate understanding. Lastly, there are several major points that if elaborated on, would potentially be helpful for the clarity of the manuscript.

Major Points:

(1) The authors suggest and attempt to visualize in the supplemental figures, that loop extrusion mechanisms would appear during crosslinking and show as vertical stripes in the micro-C data. In order to see stripes, a majority of the nuclei would need to undergo loop extrusion at the same rate, starting from exactly the same spots, and the loops would also have to be released and restarted at the same rate. If these patterns truly result from loop extrusion, the authors should provide experimental evidence from another organism undergoing loop extrusion.

(3.2) We don’t know of any reports that actually document cohesion extrusion events that are forming TADs (TADs as defined in our paper, in the RCMC experiments of Goel et al. (Goel et al. 2023), in response #1.1, or in the high-resolution images from the MicroC data of Krietenstein et al (Krietenstein et al. 2020) and Hseih et al. (Hsieh et al. 2020). However, an extruding cohesin complex would be expected to generate stripes because it transiently brings together the two chromatin strands as illustrated by the broken zipper in Figure Supplemental 2 of our paper.  While stripes generated by cohesin forming a TAD have not to our knowledge ever been observed, Fig. 4 in Goel et al. (Goel et al. 2023)) shows 45o stripes outlining TADs and connecting neighboring TADs.  These stripes are visible with or without Rad21.

In some versions of the loop extrusion model, cohesin extrudes a loop until it comes to a halt at both boundaries, where it then remains holding the loop together.  In this model, the extrusion event would occur only once per cell cycle.  This is reason we selected NC14 embryos as this point in development should provide by far the best opportunity to visualize cohesin-dependent TAD formation.  However, the expected stripes generated by cohesin embrace of both strands of the extruding loop were not evident.  Other newer versions of the loop extrusion model are much more dynamic—cohesin extrudes the loop, coming to a halt at the two boundaries, but either doesn’t remain stably bound or breaks through one or both boundaries. In the former case, the TAD needs to be reestablished by another extrusion event, while in the latter case LDC domains are generated.  In this dynamic model, we should also be able to observe vertical and 45o stripes (or stripes leaning to one side or another of the loading site if the extrusion rates aren’t equal on both fibers) in NC14 embryos corresponding to the formation of TADs and LDC domains.  However, we don’t.

(2) On lines 311-314, the authors discuss that stem-loops generated by cohesin extrusion would possibly be expected to have more next-next-door neighbor contacts than next-door neighbor contacts and site their models in Figure 1. Based on the boundary:boundary pairing models in the same figure would the stem-loops created by head-to-tail pairing also have the same phenotype? Making possible enrichment of next-next-door neighbor contacts possible in both situations? The concepts in the text are not clear, and the diagrams are not well-labeled relative to the two models.

(3.3) Yes, we expect that stem-loops formed by cohesin extrusion or head-to-tail pairing would behave in a similar manner.  They could be stem-loops separated by unanchored loops as shown in Fig. 1B and E.  Alternatively, adjacent loops could be anchored to each other (by cohesin/CTCF road blocks or by pairing interactions) as indicated in Fig. 1C and F.  In stem-loops generated either by cohesin extrusion or by head-to-tail pairing, next-next door neighbors should interact with each other, generating a plume above the volcano triangle.  In the case of circle-loops, the volcano triangle should be flanked by clouds that are generated when the TAD bumps into both next-door neighbors.  In the accompanying paper, we test this idea by deleting the nhomie boundary and then a) inserting nhomie back in the reverse orientation, or b) by inserting homie in the forward orientation.  The MicroC patterns fit with the predictions that were made in this paper.

(3) The authors appear to cite Chen et al., 2018 as a reference for the location of these transgenes being 700nM away in a majority of the nuclei. However, the exact transgenes in this manuscript do not appear to have been measured for distance. The authors could do this experiment and include expression measurements.

(3.4) The transgenes used in Chen et al. are modified versions of a transgene used in Fujioka et al. (2016) inserted into the same attP site.  When we visualize reporter transcription in NC14 embryos driven by the eve enhancers using smFISH, HCR-FISH or DIG, only a subset of the nuclei at this stage are active.  The number of active nuclei we detect is similar to that observed in the live imaging experiments of Chen et al.  The reason we cited Chen et al. (Chen et al. 2018) was that they found that proximity was a critical factor in determining whether the reporter was activated or not in a given nucleus.  The actual distance they measured wasn’t important.  Moreover, as we discussed in response #2.6 above, there are good reasons to think that the “precise” distances measured in live imaging experiments like those used in Chen et al. are incorrect.  However, their statements are certainly correct if one considers that a distance of ~700 nM or so is “more distant” relative to a distance of ~300 nM or so, which is “closer.”

(4) The authors discuss the possible importance of CTCF orientation in forming the roadblock to cohesin extrusion and discuss that Homie orientation in the transgene may impact Homie function as an effective roadblock. However, the Homie region inserted in the transgene does not contain the CTCF motif. Can the authors elaborate on why they feel the orientation of Homie is important in its ability to function as a roadblock if the CTCF motif is not present? Trans-acting factors responsible for Homie function have not been identified and this point is not discussed in the manuscript.

We discussed the “importance” of CTCF orientation in forming roadblocks because one popular version of the cohesin loop extrusion/CTCF roadblock model postulates that CTCF must be oriented so that the N-terminus of the protein is facing towards the oncoming cohesin complex, otherwise it won’t be able to halt extrusion on that strand.  When homie in the transgene is pointing towards the eve locus, the reporter on the other side (farther from eve) is activated by the eve enhancers.  One possible way to explain this finding (if one believes the loop extrusion model) is that when homie is inverted, it can’t stop the oncoming cohesin complex, and it runs past the homie boundary until it comes to a stop at a properly oriented boundary farther away.  In this case, the newly formed loop would extend from the boundary that stopped cohesin to the homie boundary in the eve locus, and would include not only the distal reporter, but also the proximal reporter.  If both reporters are in the same loop with the eve enhancers (which they would have to be given the mechanism of TAD formation by loop extrusion), both reporters should be activated.  They are not.

For the boundary pairing model, the reporter that will be activated will depend upon the orientation of the pairing interaction—which can be either head-to-head or head-to-tail (or both: see discussion of LBC elements in #2.1).  For an easy visualization of how the orientation of pairing interactions is connected to the patterns of interactions between sequences neighboring the boundary, please look at Fig. 9.  This figure shows two different meta-loops.  In panel A, head-tohead pairing of the blue and purple boundaries brings together, on the one hand, sequences upstream of the blue and purple boundary, and on the other hand, sequences downstream of the blue and purple boundaries.  In the circle loop configuration, the resulting rectangular boxes of enhanced contact are located in the upper left and lower right of the contact map.  In panel B, the head-to-tail pairing of the blue and purple boundary changes how sequences upstream and downstream of the blue and purple boundaries interact with each other.  Sequences upstream of the blue boundary interact with sequences downstream of the purple boundary, and this gives the rectangular box of enhanced interactions on the top right.  Sequences downstream of the blue boundary interact with sequences upstream of the purple boundary, and this gives the rectangular box of enhanced contact on the lower left.

CTCF: Our analysis of the homie boundary suggests that CTCF contributes little to its activity.  It has an Su(Hw) recognition sequence and a CP190 “associated” sequence.  Mutations in both compromise boundary activity (blocking and -142 kb pairing).  Gel shift experiments and ChIP data indicate there are half a dozen or more additional proteins that associate with the 300 bp homie fragment used in our experiments.

Orientation of CTCF or other protein binding sites:  The available evidence suggests that orientation of the individual binding sites is not important (Kyrchanova et al. 2016; Lim et al. 2018)).  Instead, it is likely that the order of binding sites affects function.

(5) The imaging results seem to be consistent with both boundary:boundary interaction and loop extrusion stem looping.

It is not clear whether the reviewer is referring to the different patterns of reporter expression— which clearly don’t fit with the loop extrusion model in the key cases that distinguish the two models—or the live imaging experiments in Chen et al. (Chen et al. 2018).

(6) The authors suggest that the eveMa TAD could only be formed by extrusion after the breakthrough of Nhomie and several other roadblocks. Additionally, the overall long-range interactions with Nhomie appear to be less than the interactions with endogenous Homie (Figures 7, 8, and supplemental 5). Is it possible that in some cases boundary:boundary pairing is occurring between only the transgenic Homie and endogenous Homie and not including Nhomie?

Yes, it is possible.  On the other hand, the data that are currently available supports the idea that transgene homie usually interacts with endogenous homie and nhomie at the same time.  This is discussed in #2.6D above.  The viewpoints indicate that crosslinking occurs more frequently to homie than to nhomie.  This could indicate that when there are only pairwise interactions, these tend to be between homie and homie.  Alternatively, this could also be explained by a difference in relative crosslinking efficiency.

(7) In Figure 4E, the GFP hebe expression shown in the LhomieG Z5 transgenic embryo does not appear in the same locations as the LlambdaG Z5 control. Is this actually hebe expression or just a background signal?

The late-stage embryos shown in E are oriented differently.  For GlambdaL, the embryo is oriented so that hebe-like reporter expression on the ventral midline is readily evident.  However, this orientation is not suitable for visualizing eve enhancer-dependent expression of the reporters in muscle progenitor cells.  For this reason, the 12-16 hr GeimohL embryo in E is turned so that the ventral midline isn’t readily visible in most of the embryo.  As is the case in NC14 embyros, the eve enhancers drive lacZ but not gfp expression in the muscle progenitor cells.

(8) Figure 6- The LhomieG Z3 (LeimohG) late-stage embryo appears to be showing the ventral orientation of the embryo rather than the lateral side of the embryo as was shown in the previous figure. Is this for a reason? Additionally, there are no statistics shown for the Z3 transgenic images.

Were these images analyzed in the same way as the Z5 line images?

The LeimohG embryo was turned so that the hebe enhancer-dependent expression of lacZ is visible.  While the eve enhancer-dependent expression of lacZ in the muscle progenitor cells isn’t visible with this orientation, eve enhancer-dependent expression in the anal plate is.

(9) Do the Micro-C data align with the developmental time points used in the smFISH probe assays?

The MicroC data aligns with the smFISH images of older embryos: 12-14 hour embryos or stages 14-16.  

Recommendations for the authors:   

Reviewer #1 (Recommendations For The Authors):

This was a difficult paper to review. It took me several hours to understand the terminology and back and forth between different figures to put it together. It might be useful to put the loop models next to the MicroC results and have a cartoon way of incorporating which enhancers are turning on which reporters.

I also found the supercoiled TAD models in Figure 1 not useful. These plectoneme-type of structures likely do not exist, based on the single-cell chromosome tracing studies, and the HiC structures not showing perpendicular to diagonal interactions between the arms of the plectonemes.

We wanted to represent the TAD as a coiled 30nM fiber, as they are not likely to resemble the large loops like those shown in Fig. 1 A, D, and G.

There are no stripes emerging from homies, which is consistent with the pairing model, but there seem to be stripes from the eve promoter. I think these structures may be a result of both the underlying loop extruders + pairing elements.

There are internal structures in the eve TAD that link the upstream region of the eve promoter to the eve PRE and sequences in nhomie.  All three of these sequences are bound by LBC.  Each of the regulatory domains in BX-C also have LBC elements and, as shown in Author response image 1, you can see stripes connecting some of these LBC elements to each other.  Since the stripes that Goel et al. (Goel et al. 2023) observed in their RCMC analysis of Ppm1g didn’t require cohesin, how these stripes are generated (active: e.g, a chromatin remodeler or passive: e.g., the LBC complex has non-specific DNA binding activity that can be readily crosslinked as the chromatin fiber slides past) isn’t clear.

The authors say there are no TADs that have "volcano plumes" but the leftmost TAD TA appears to have one. What are the criteria for calling the plumes? I am also not clear why there is a stripe off the eve volcano. It looks like homie is making a "stripe" loop extrusion type of interaction with the next TAD up. Is this maybe cohesin sliding off the left boundary?

The reviewer is correct, the left-most TAD TA appears to have a plume.  We mentioned TA seems to have a plume in the original text, but it was inadvertently edited out.

Two different types of TADßàTAD interactions are observed.  In the case of eve, the TADs to either side of eve interact more frequently with each other than they do with eve.  This generates a “plume” above the eve volcano triangle.  The TADs that comprise the Abd-B regulatory domains (see Author response image 1) are surrounded by clouds of diminishing intensity.  Clouds at the first level represent interactions with both next-door neighbors; clouds at the second level represent interactions with both next-next-door neighbors; clouds at the third level represent interactions with next-next-next door neighbors.  The Abd-B TADs are close to the same size, so that interactions with neighbors are relatively simple.  However, this is not always the case.  When there are smaller TADs near larger TADs the pattern of interaction can be quite complicated.  An example is indicated by the red bar in Author response image 2

The authors state "In the loop-extrusion model, a cohesin complex initiating loop extrusion in the eve TAD must break through the nhomie roadblock at the upstream end of the eve TAD. It must then make its way past the boundaries that separate eve from the attP site in the hebe gene, and come to a halt at the homie boundary associated with the lacZ reporter." Having multiple loops formed by cohesin would also bring in the 142kb apart reporter and homie. Does cohesin make 140 kb long loops in flies?

A mechanism in which cohesin brings the reporter close to the eve TAD by generating many smaller loops (which would be the intervening TADs) was discussed in #1.2.

Figure 5 title mistakes the transgene used?

Fixed.

In figure 6, the orientation of the embryos does not look the same for the late-stage panels. So it was difficult to tell if the eve enhancer was turning the reporter on.

Here we were focusing mainly on the AP enhancer activation of the reporter, as this is most easily visualized.  It should be clear from the images that the appropriate reporter is activated by the AP enhancer for each of the transgene inserts.

It is not clear to me why the GFP makes upstream interactions (from the 4C viewpoint) in GhomileLZ5 but not in LhomieGZ5? Corresponding interactions for Fig Supp 5 & 6 are not the same. That is, LacZ in the same place and with the same homie orientation does not show a similar upstream enrichment as the GFP reporter does.

We are uncertain as to whether we understand this question/comment.  In GhomieLZ5 (now GhomieL, the lacZ reporter is on the eve side of the homie boundary while gfp is on the hebe enhancer side of the homie boundary.  Since homie is pointing away from gfp, pairing interactions with homie and nhomie in the eve locus bring the eve enhancers in close proximity with the gfp reporter.  This is what is seen in Fig. 7 panel D—lower trace.  In LhomieGZ5 (now GeimohL) the lacZ reporter is again on the eve side of the homie boundary while gfp is on the hebe enhancer side of the homie boundary.  However, in this case homie is inverted so that it is points away from lacZ (towards gfp).  In this orientation, pairing brings the lacZ reporter into contact with the eve enhancers.  This is what is seen in the upper trace in Fig. 7 panel D.

The orientation of the transgene is switch in Fig. Supp 5 and 6.  For these “Z3) transgenes (now called LeimohG and LhomieG the gfp reporter is on the eve side of homie while the lacZ reporter is on the hebe enhancer side of homie.  The interactions between the reporters and eve are determined by the orientation of homie in the transgene.  When homie is pointing away from gfp (as in LeimohG), gfp is activated and that is reflected in the trace in Supp Fig. 5. When homie is pointing away from lacZ, lacZ is activated and this is reflected (though not as cleanly as in other cases) in the trace in Supp Fig. 6.  

I did not see a data availability statement. Is the data publicly available? The authors also should consider providing the sequences of the insertions, or provide the edited genomes, in case other researchers would like to analyze the data.

Data have been deposited.

Reviewer #3 (Recommendations For The Authors):

Minor Points:

(1) There is an inconsistency in the way that some of the citations are formatted. Some citations have 'et al' italicized while others do not. It seems to be the same ones throughout the manuscript. Some examples: Chetverina et al 2017, Chetverina et al 2014, Cavalheiro et al 2021, Kyrchanova et al 2008a, Muravyova et al 2001.

Fixed

(2) Pita is listed twice in line 48.

Fixed

(3) Line 49, mod(mdg4)67.2 is written just as mod(mdg4). The isoform should be indicated.

This refers to all Mod isoforms.

(4) Homie and Nhomie are italicized throughout the manuscript and do not need to be.

This is the convention used previously.  

(5) The supplemental figure captions 1 and 2 in the main document are ordered differently than in the supplemental figures file. This caused it to look like the figures are being incorrectly cited in lines 212-214 and 231-232.

Fixed

(6) Is the correct figure being cited in line 388-389? The line cites Figure 6E when mentioning LlambdaG Z5; however, LlambdaG Z5 is not shown in Figure 6.

Fixed

(7) Section heading 'LhomieG Z5 and GhomieL Z5' could be renamed for clarity. GhomieL Z5 results are not mentioned until the next section, named 'GhomieL Z5'.

Fixed

(8) Can the authors provide better labeling for control hebe expression? This would help to determine what is hebe expression and what is background noise in some of the embryos in Figures 4-6.

Author response image 5 shows expression of the lacZ reporter in GeimohL and GlambdaL.  For the GlambdaL transgene, the hebe enhancers drive lacZ expression in 1216 hr embryos.  Note that lacZ expression is restricted to a small set of quite distinctive cells along the ventral midline.  lacZ is also expressed on the ventral side of the GeimohL embryo (top panel).  However, their locations are quite different from those of the lacZ positive cells in the GlambdaL transgene embryo.  These cells are displaced from the midline, and are arranged as pairs of cells in each hemisegment, locations that correspond to eve-expressing cells in the ventral nerve cord.  The eve enhancers also drive lacZ expression elsewhere in the GeimohL embryo, including the anal plate and dorsal muscle progenitor cells (seen most clearly in the lower left panel).

Author response image 5.

lacZ expression in Giemohl and Glambdal embryos

(9) The Figure 5 title is labeled with the wrong transgene.

Fixed

(10) Heat map scales are missing for Figures 7, supplemental 5, and supplemental 6.

Fixed

(11) Did the authors check if there was a significant difference in the expression of GFP and lacZ from lambda control lines to the Homie transgenic lines?

Yes.  Statistical analysis added in Table Supplemental #1

(12) The Figure 7 title references that these are Z3 orientations, however, it is Z5 orientations being shown.

Fixed

(13) The virtual 4C data should include an axis along the bottom of the graphs for better clarity. An axis is missing in all 4C figures.

References:

Bantignies F, Grimaud C, Lavrov S, Gabut M, Cavalli G. 2003. Inheritance of polycomb-dependent chromosomal interactions in drosophila. Genes Dev. 17(19):2406-2420.

Batut PJ, Bing XY, Sisco Z, Raimundo J, Levo M, Levine MS. 2022. Genome organization controls transcriptional dynamics during development. Science. 375(6580):566-570.

Bonchuk A, Boyko K, Fedotova A, Nikolaeva A, Lushchekina S, Khrustaleva A, Popov V, Georgiev P. 2021. Structural basis of diversity and homodimerization specificity of zinc-fingerassociated domains in drosophila. Nucleic Acids Res. 49(4):2375-2389.

Bonchuk AN, Boyko KM, Nikolaeva AY, Burtseva AD, Popov VO, Georgiev PG. 2022. Structural insights into highly similar spatial organization of zinc-finger associated domains with a very low sequence similarity. Structure. 30(7):1004-1015.e1004.

Chen H, Levo M, Barinov L, Fujioka M, Jaynes JB, Gregor T. 2018. Dynamic interplay between enhancer–promoter topology and gene activity. Nat Genet. 50(9):1296.

Fedotova AA, Bonchuk AN, Mogila VA, Georgiev PG. 2017. C2h2 zinc finger proteins: The largest but poorly explored family of higher eukaryotic transcription factors. Acta Naturae. 9(2):4758.

Foe VE. 1989. Mitotic domains reveal early commitment of cells in drosophila embryos. Development. 107(1):1-22.

Fujioka M, Mistry H, Schedl P, Jaynes JB. 2016. Determinants of chromosome architecture: Insulator pairing in cis and in trans. PLoS Genet. 12(2):e1005889.

Galloni M, Gyurkovics H, Schedl P, Karch F. 1993. The bluetail transposon: Evidence for independent cis-regulatory domains and domain boundaries in the bithorax complex. The EMBO Journal. 12(3):1087-1097.

Goel VY, Huseyin MK, Hansen AS. 2023. Region capture micro-c reveals coalescence of enhancers and promoters into nested microcompartments. Nat Genet. 55(6):1048-1056.

Hsieh TS, Cattoglio C, Slobodyanyuk E, Hansen AS, Rando OJ, Tjian R, Darzacq X. 2020. Resolving the 3d landscape of transcription-linked mammalian chromatin folding. Mol Cell. 78(3):539553.e538.

Ke W, Fujioka M, Schedl P, Jaynes JB. 2024. Chromosome structure ii: Stem-loops and circle-loops. eLife.

Krietenstein N, Abraham S, Venev SV, Abdennur N, Gibcus J, Hsieh TS, Parsi KM, Yang L, Maehr R, Mirny LA et al. 2020. Ultrastructural details of mammalian chromosome architecture. Mol Cell. 78(3):554-565.e557.

Kyrchanova O, Ibragimov A, Postika N, Georgiev P, Schedl P. 2023. Boundary bypass activity in the abdominal-b region of the drosophila bithorax complex is position dependent and regulated. Open Biol. 13(8):230035.

Kyrchanova O, Kurbidaeva A, Sabirov M, Postika N, Wolle D, Aoki T, Maksimenko O, Mogila V, Schedl P, Georgiev P. 2018. The bithorax complex iab-7 polycomb response element has a novel role in the functioning of the fab-7 chromatin boundary. PLoS Genet. 14(8):e1007442.

Kyrchanova O, Mogila V, Wolle D, Deshpande G, Parshikov A, Cleard F, Karch F, Schedl P, Georgiev P. 2016. Functional dissection of the blocking and bypass activities of the fab-8 boundary in the drosophila bithorax complex. PLoS Genet. 12(7):e1006188.

Kyrchanova O, Sabirov M, Mogila V, Kurbidaeva A, Postika N, Maksimenko O, Schedl P, Georgiev P.

2019a. Complete reconstitution of bypass and blocking functions in a minimal artificial fab7 insulator from drosophila bithorax complex. Proceedings of the National Academy of Sciences.201907190.

Kyrchanova O, Wolle D, Sabirov M, Kurbidaeva A, Aoki T, Maksimenko O, Kyrchanova M, Georgiev P, Schedl P. 2019b. Distinct elements confer the blocking and bypass functions of the bithorax fab-8 boundary. Genetics.genetics. 302694.302019.

Li H-B, Muller M, Bahechar IA, Kyrchanova O, Ohno K, Georgiev P, Pirrotta V. 2011. Insulators, not polycomb response elements, are required for long-range interactions between polycomb targets in drosophila melanogaster. Mol Cell Biol. 31(4):616-625.

Li X, Tang X, Bing X, Catalano C, Li T, Dolsten G, Wu C, Levine M. 2023. Gaga-associated factor fosters loop formation in the drosophila genome. Mol Cell. 83(9):1519-1526.e1514.

Lim B, Heist T, Levine M, Fukaya T. 2018. Visualization of transvection in living drosophila embryos. Mol Cell. 70(2):287-296. e286.

Link N, Kurtz P, O'Neal M, Garcia-Hughes G, Abrams JM. 2013. A p53 enhancer region regulates target genes through chromatin conformations in cis and in trans. Genes Dev. 27(22):24332438.

Mohana G, Dorier J, Li X, Mouginot M, Smith RC, Malek H, Leleu M, Rodriguez D, Khadka J, Rosa P et al. 2023. Chromosome-level organization of the regulatory genome in the drosophila nervous system. Cell. 186(18):3826-3844.e3826.

Muller M, Hagstrom K, Gyurkovics H, Pirrotta V, Schedl P. 1999. The mcp element from the drosophila melanogaster bithorax complex mediates long-distance regulatory interactions. Genetics. 153(3):1333-1356.

Postika N, Metzler M, Affolter M, Müller M, Schedl P, Georgiev P, Kyrchanova O. 2018. Boundaries mediate long-distance interactions between enhancers and promoters in the drosophila bithorax complex. PLoS Genet. 14(12):e1007702.

Rollins RA, Morcillo P, Dorsett D. 1999. Nipped-b, a drosophila homologue of chromosomal adherins, participates in activation by remote enhancers in the cut and ultrabithorax genes. Genetics. 152(2):577-593.

Samal B, Worcel A, Louis C, Schedl P. 1981. Chromatin structure of the histone genes of d. Melanogaster. Cell. 23(2):401-409.

Shermoen AW, McCleland ML, O'Farrell PH. 2010. Developmental control of late replication and s phase length. Curr Biol. 20(23):2067-2077.

Shidlovskii YV, Bylino OV, Shaposhnikov AV, Kachaev ZM, Lebedeva LA, Kolesnik VV, Amendola D, De Simone G, Formicola N, Schedl P et al. 2021. Subunits of the pbap chromatin remodeler are capable of mediating enhancer-driven transcription in drosophila. Int J Mol Sci. 22(6).

Sigrist CJ, Pirrotta V. 1997. Chromatin insulator elements block the silencing of a target gene by the drosophila polycomb response element (pre) but allow trans interactions between pres on different chromosomes. Genetics. 147(1):209-221.

Udvardy A, Schedl P. 1984. Chromatin organization of the 87a7 heat shock locus of drosophila melanogaster. J Mol Biol. 172(4):385-403.

Vazquez J, Muller M, Pirrotta V, Sedat JW. 2006. The mcp element mediates stable long-range chromosome-chromosome interactions in drosophila. Molecular Biology of the Cell. 17(5):2158-2165.

Wolle D, Cleard F, Aoki T, Deshpande G, Schedl P, Karch F. 2015. Functional requirements for fab-7 boundary activity in the bithorax complex. Mol Cell Biol. 35(21):3739-3752.

Author Response:

We thank the reviewers for careful reading, acknowledging the strength of our manuscript, and pointing out its weakness, which we will address in the revised version as described below.

(1) We will supplement our analysis with finer statistical testing and analysis, such as cross-validation and a more detailed analysis of the relation between the inferred model and the intrinsic timescales of the system. For the effect of the drug TIMP-1 on the animal, we will first explore the possibility of assessing the results using a multifactor ANOVA test, with the caveat that the distribution of interactions is not Gaussian. We will further test the effect of different group size on the significance of our results by considering subgroups of animals in the drug group, and compare the statistics between the (subsampled) drug group and the controlled group.

(2) Our manuscript is similar with that of Shemesh et al. in that we both analyze socially interacting mice by constructing maximum entropy models (MEM) of the co-localization patterns of mice. The difference is in the setup and the number of mice (4 mice in Shemesh et al, 10-15 in our work), as we outlined in the manuscript. To further supplement our current argument of the difference of our results in the Discussion section, we will learn a MEM model up to triplet interactions for our Eco-HAB mice data, and compare to our current MEM model up to pairwise interactions using test-set validation or the Bayesian information criterion (BIC).

eLife assessment

This useful work investigates the social interactions of mice living together in a system of multiple connected cages. The approach is interesting as it uses some of the tools developed in physics to investigate animal behaviour. However, , some of the analyses require further scrutiny, leaving the evidence supporting the main claim currently incomplete.

Public Review:

Summary:

In this manuscript, Chen et al. investigate the statistical structure of social interactions among mice living together in the ECO-Hab. They use maximum entropy models (MEM) from statistical physics that include individual preferences and pair-wise interactions among mice to describe their collective behavior. They also use this model to track the evolution of these preferences and interactions across time and in one group of mice injected with TIMP-1, an enzyme regulating synaptic plasticity. The main result is that they can explain group behavior (the probability of being together in one compartment) by a MEM that only includes pair-wise interactions. Moreover, the impact of TIMP-1 is to increase the variance of the couplings J_ij, the preference for the compartment containing food, as well as the dissatisfaction triplet index (DTI).

Strengths:

The ECO-Hab is a really nice system to ask questions about the sociability of mice and to tease apart sociability from individual preference. Moreover, combining the ECO-Hab with the use of MEM is a powerful and elegant approach that can help statistically characterize complex interactions between groups of mice -- an important question that requires fine quantitative analysis.

Weaknesses:

However, there is a risk in interpreting these models. In my view, several of the comparisons established in the current study would require finer and more in-depth analysis to be able to establish firmer conclusions (see below). Also, the current study, which closely resembles previous work by Shemesh et al., finds a different result but does not provide the same quantitative model comparison included there, nor a conclusive explanation of why their results are different. In total, I felt that some of the results required more solid statistical testing and that some of the conclusions of the paper were not entirely justified. In particular, the results from TIMP-1 require proper interaction tests (group x drug) which I couldn't find. This is particularly important when the control group has a smaller N than the drug groups.

Author response:

The following is the authors’ response to the current reviews.

Reviewer #1 (Recommendations For The Authors):

I still find it really impressive that the Purkinje cell stimulation so closely mimics the pathogenic phenotypes - in my opinion, the strongest part of the paper. I would like just a little clarification on some of my previous questions.

Major points:

(1) Can the authors clarify where the new units came from? Are these units that were recorded before the initial submission and excluded, but are now included? If so, why were they excluded before? Or are these units that were recorded since the original submission?

The number of units increased in Figure 1 for three reasons: 1) We have now plotted the classifier results in Figure 1 instead of the validation results, which have been moved to Figure 1 Supplement 3. 2) In response to reviewer comments, we no longer include units that had >60 s of recording in both our model creation and validation. We had previously used 30 s for creating the model and a different 30 s for validating the model, if an additional 30 s were available. 3) We changed our model creation and validation strategy based on previous reviewer comments. The new units in Figures 2-4 were taken from our pool of previously collected but unanalyzed data (we collect neural data on a rolling basis and thus these data were not initially available). We were fortunate to have these data to analyze in order to address the concerns about the number of cells included in the manuscript. The number of units increased in Figure 5 because new units were recorded in response to reviewer comments.

(2) Why did some of the neuron counts go down? For example, in Pdx1Cre;Vglut2fl/fl mice, the fraction of units with the control signature went from 11/21 to 7/23. Is this because the classifier changed between the original submission and the revision?

Yes, the proportion of cells matching each classification changed due to the different parameters and thresholds used in the updated classifier model.

Minor points:

In the Discussion: "We find some overlap and shared spike features between the different disease phenotypes and show that healthy cerebellar neurons can adapt multiple disease-associated spike train signatures." I think "adapt" should be "adopt"

In the Discussion: "compare" is misspelled as "compared"

Thank you for bringing these typos to our attention. We will upload a new version of the text with the typos corrected.

---

The following is the authors’ response to the original reviews.

We would like to thank the Reviewers for providing excellent and constructive suggestions that have enabled us to strengthen our overall presentation of our data. We have addressed each of the comments by altering the text, providing additional data, and revising the figures, as requested.

Below are our explanations for how we have altered the manuscript in this revised version.

Recommendations for the authors:

I think you will have seen from the comments that there was great enthusiasm for the importance of this study. There were also shared concerns about how the classifier may be inadequate in its current format, as well as specific suggestions to consider to improve. I hope that you will consider a revision to really amplify the impact of the importance of this study.

Reviewer #1 (Recommendations For The Authors):

Distinct motor phenotypes are reflected in different neuronal firing patterns at different loci in motor circuits. However, it is difficult to determine if these altered firing patterns: 1) reflect the underlying neuropathology or phenotype, 2) whether these changes are intrinsic to the local cell population or caused by larger network changes, and 3) whether abnormal firing patterns cause or reflect abnormal movement patterns. This manuscript attempts to address these questions by recording neural firing patterns in deep cerebellar nucleus neurons in several models of cerebellar dysfunction with distinct phenotypes. They develop a classifier based on parameters of single unit spike trains that seems to do an inconsistent job of predicting phenotype (though it does fairly well for tremor). The major limitation of the recording/classifier experiments is the low number of single units recorded in each model, greatly limiting statistical power. However, the authors go on to show that specific patterns of Purkinje cell stimulation cause consistent changes in interposed nucleus activity that map remarkably well onto behavioral phenotypes. Overall, I did not find the recording/classifier results to be very convincing, while the stimulation results strongly indicate that interposed nucleus firing patterns are sufficient to drive distinct behavioral phenotypes.

We thank the reviewer for their comments. We describe below how we have addressed the major concerns.

Major concerns:

(1) I don't think it's legitimate to use two 30-second samples from the same recording to train and validate the classifier. I would expect recordings from the same mouse, let alone the same unit, to be highly correlated with each other and therefore overestimate the accuracy of the classifier. How many of the recordings in the training and validation sets were the same unit recorded at two different times?

We previously published a paper wherein we measured the correlation (or variability) between units recorded from the same mouse versus units recorded from different mice (see: Van der Heijden et al., 2022 – iScience, PMID: 36388953). In this paper we did not find that nuclei neuron recordings from the same mouse were more correlated or similar to each other than recordings from different mice. 

Upon this reviewer comment, however, we did observe strong correlations between the two 30-second samples from the same recording units. We therefore decided to no longer validate our classifier based on a training and validation sets that had overlapping units. Instead, we generated 12 training sets and 12 non-overlapping validation sets based on our entire database. We then trained 12 classifier models and ranked these based on their classification ability on the validation sets (Figure 1 – supplemental Figure 3). We found that the top two performing classifier models were the same, and used this model for the remainder of the paper. 

(2) The n's are not convincing for the spike signature analyses in different phenotypic models. For example, the claim is that Pdx1Cre;Vglut2fl/fl mice have more "control" neurons than ouabain infusion mice (more severe phenotype). However, the numbers are 11/21 and 7/20, respectively. The next claim is that 9/21 dystonic neurons are less than 11/20 dystonic neurons. A z-test for proportions gives a p-value of 0.26 for the first comparison and a pvalue of 0.44 for the second. I do not think any conclusions can be drawn based on these data.

We included more cells in our analyses and found that the z-test for n the proportion of cells with the “control” and “dystonia” signature is indeed statistically significant. 

(3) Since the spiking pattern does not appear to predict an ataxic phenotype and the n's are too small to draw a conclusion for the dystonic mice, I think the title is very misleading - it does not appear to be true that "Neural spiking patterns predict behavioral phenotypes...", at least in these models.

We have changed the title to: “Cerebellar nuclei cells produce distinct pathogenic spike signatures in mouse models of ataxia, dystonia, and tremor.” We feel that this new title captures the idea that we find differences between spike signatures associated with ataxia, dystonia, and tremor and that these signatures induce pathological movements.

(4) I don't think it can be concluded from the optogenetic experiments that the spike train signatures do not depend on "developmental changes, ...the effect of transgene expression, ... or drug effects outside the cerebellum." The optogenetic experiments demonstrate that modulating Purkinje cell activity is sufficient to cause changes in DCN firing patterns and phenotypes (i.e., proof-of-principle). However, they do not prove that this is why DCN firing is abnormal in each model individually.

Thank you for highlighting this section of the text. We agree that the optogenetic experiments cannot explain why the DCN is firing abnormally in each model. We have edited this section of the text to prevent this conclusion from being drawn by the readers.

Minor points:

(1) It would be nice to see neural recordings in the interposed nucleus during Purkinje terminal stimulation to verify that the firing patterns observed during direct Purkinje neuron illumination are reproduced with terminal activation. This should be the case, but I'm not 100% certain it is.

We have edited the text to clarify that representative traces and analysis of interposed nucleus neurons in response to Purkinje terminal stimulation are the data in Figure 5.

(2) How does the classifier validation (Fig. 1E) compare to chance? If I understand correctly, 24/30 neurons recorded in control mice are predicted to have come from control mice (for example). This seems fairly high, but it is hard to know how impressive this is. One approach would be to repeat the analysis many (1000s) of times with each recording randomly assigned to one of the four groups and see what the distribution of "correct" predictions is for each category, which can be compared against the actual outcome.

We have now also included the proportion of spike signatures in the entire population of neurons and show that the spike signatures are enriched in each of the four groups (control, ataxia, dystonia, tremor) relative to the presence of these signatures in the population (Figure 1E). 

(3) I don't think this is absolutely necessary, but do the authors have ideas about how their identified firing patterns might lead to each of these phenotypes? Are there testable hypotheses for how different phenotypes caused by their stimulation paradigms arise at a network level?

We have added some ideas about how these spike signatures might lead to their associated phenotypes to the discussion.

Reviewer #2 (Recommendations For The Authors):

(1) As mentioned earlier, my main concern pertains to the overall architecture and training of the classifier. Based on my reading of the methods and the documentation for the classifier model, I believe that the classifier boundaries may be biased by the unequal distribution of neurons across cerebellar disease groups (e.g., n=29 neurons in control versus n=19 in ataxics). As the classifier is trained to minimize the classification error across the entire sample, the actual thresholds on the parameters of interest may be influenced by the overrepresentation of neurons from control mice. To address this issue, one possible solution would be to reweight each group so that the overall weight across classes is equal. However, I suggest a better strategy might be to revise the classifier architecture altogether (as detailed below).

We have retrained the classifier model based on equal numbers of ataxic, dystonic, and tremor cells (n=20) but we intentionally included more control cells (n=25). We included more control cells because we assume this is the baseline status for all cerebellar neurons and wanted to avoid assigning disease signatures to healthy neurons too easily. 

(2) As the authors make abundantly clear, one mouse model of disease could potentially exhibit multiple phenotypes (e.g., a mouse with both ataxia and tremor). To address this complexity, it might be more valuable to predict the probability of a certain CN recording producing specific behavioral phenotypes. In this revised approach, the output of the classifier wouldn't be a single classification (e.g., "this is an ataxic mouse") but rather the probability of a certain neural recording corresponding to ataxia-like symptoms (e.g., "the classifier suggests that this mouse has a 76% likelihood of exhibiting ataxic symptoms given this CN recording"). This modification wouldn't require additional data collection, and the exemplar disease models could still be used to train such a revised network/classifier, with each mouse model corresponding to 0% probability of observing all other behavioral phenotypes except for the specific output corresponding to the disease state (e.g., L7CreVgat-fl/fl would be 0% for all categories except ataxia, which would be trained to produce a score of 100%). This approach could enhance the validation results across other mouse models by allowing flexibility in a particular spike train parameter to produce a diverse set of phenotypes.

This is a great comment. Unfortunately, our current dataset is constrained to fully address this comment for the following reasons:

- We have a limited number of neurons on which we can train our classifier neurons. Further dividing up the groups of neurons or complicating the model limited the power of our analyses and resulted in overfitting of the model on too few neurons.

- The recording durations (30 seconds) used to train our model are likely too short to find multiple disease signatures within a single recording. We feel that the complex phenotypes are likely resulting from cells within one mouse exhibiting a mix of disease signatures (as in the Car8wdl/wdl mice).

We think this question would be great for a follow-up study that uses a large number of recordings from single mice to fully predict the mouse phenotype based on the population spike signatures. 

To limit confusion about our classifier model, we have also altered the language of our manuscript and refer to the cells exhibiting a spike signature instead of predicting a phenotype. 

However, the paper falls short in terms of the classifier model itself. The current implementation of this classifier appears to be rather weak. For instance, the crossvalidated performance on the same disease line mouse model for tremor is only 56%. While I understand that the classifier aims to simplify a high-dimensional dataset into a more manageable decision tree, its rather poor performance undermines the authors' main objectives. In a similar vein, although focusing on three primary features of spiking statistics identified by the decision tree model (CV, CV2, and median ISI) is useful for understanding the primary differences between the firing statistics of different mouse models, it results in an overly simplistic view of this complex data. The classifier and its reliance on the reduced feature set are the weakest points of the paper and could benefit from further analysis and a different classification architecture. Nevertheless, it is commendable that the authors have collected high-quality data to validate their classifier. Particularly impressive is their inclusion of data from multiple mouse models of ataxia, dystonia, and tremor, enabling a true test of the classifier's generalizability.

We intentionally simplified our parameter space from a high-dimensional dataset into a more manageable decision tree. We did this for the following reasons:

- The parameters, even though all measuring different features, are highly correlated (see Figure 1 – supplemental Figure 2). Further, we were training our dataset on a limited number of recordings. We found that including all parameters (for example using a linear model) caused overfitting of the data and poor model performance.

- Describing the spike signatures using a lower number of parameters allowed us to design optogenetic parameters that would mimic this parameter space. This would be infinitely more complex with a bigger parameter space. 

We agree with the reviewer that inclusion of multiple mouse models in addition to the optogenetics experiments provide the classifier’s generalizability. 

Minor Comments:

(1) The blown-up CN voltage traces in Figures 5C and Supplementary Figure 2B appear more like bar plots than voltage traces on my machine.

Thank you for bringing this to our attention. We have improved the rendering of the traces.

(2) The logic in lines 224-228 is somewhat confusing. The spike train signatures are undoubtedly affected by all the factors mentioned by the authors. What, I believe, the authors intend to convey is that because changes in CN firing rates can be driven by multiple factors, it is the CN firing properties themselves that likely drive disease-specific phenotypes.

We agree that our discussion of the CN firing needs clarification. We have made the appropriate edits in the text.

Reviewer #3 (Recommendations For The Authors):

It's quite astounding that this can be done from single spike trains from what are almost certainly mixed populations of neurons. Could you add something to the discussion about this? Some questions that could be addressed would be would multiple simultaneous recordings additionally help classify these diseases, or would non-simultaneous recordings from the same animal be useful? Also more discussion about which cells you are likely recording from would be useful.

Thank you for this suggestion. We have added discussion about multiple recordings, simultaneous vs non-simultaneous recordings, and our thoughts on the cell population recorded in this work.

Data in figure 2 is difficult to understand - it appears that the majority of dysregulated cells in 2 ataxic models are classified as dystonia cells, not ataxic cells. This appears surprising as it seems to be at odds with earlier data from Fig 1. In my opinion, it is not discussed adequately in the Results or Discussion section.

We have added further discussion of the ataxia models represented in Figures 1 and 2.

Minor comment:

The colours of the subdivisions of the bars in 2C and 3C, and the rest of the paper appear to be related to the groups in the middle (under "predicted"), but the colours are much paler in the figure than in the legend, although the colours in the bars and the legends match in the first figure (1E). Does this signify something?

These figures were remade with the same colors across the board.

eLife assessment

This important and novel study addresses the challenge of antimicrobial resistance by targeting plasmid proteins that interfere with plasmid transfer as a strategy to limit the spread of antibiotic-resistance genes. The evidence presented and the integration of two approaches to tackle antimicrobial resistance is convincing. This work will interest those working on plasmid transfer and antimicrobial resistance.

Reviewer #1 (Public Review):

The study by Prieto et al. faces the increasingly serious problem of bacterial resistance to antimicrobial agents. This work has an important element of novelty proposing a new approach to control antibiotic resistance spread by plasmids. Instead of targeting the resistance determinant, plasmid-borne proteins are used as antigens to be bound by specific nanobodies (Nbs). Once bound plasmid transfer was inhibited and Salmonella infection blocked. This in-depth study is quite detailed and complex, with many experiments (9 figures with multiple panels), rigorously carried out. Results fully support the authors' conclusions. Specifically, the authors investigated the role of two large molecular weight proteins (RSP and RSP2) encoded by the IncHI1 derivative-plasmid R27 of Salmonella. These proteins have bacterial Ig-like (Big) domains and are expressed on the cell surface, creating the opportunity for them to serve as immunostimulatory antigens. Using a mouse infection model, the authors showed that RSP proteins can properly function as antigens, in Salmonella strains harboring the IncHI1 plasmid. The authors clearly showed increased levels of specific IgG and IgA antibodies against these RSP proteins proteins in different tissues of immunized animals. In addition, non-immunized mice exhibited Salmonella colonization in the spleen and much more severe disease than immunized ones.

However, the strength of this work is the selection and production of nanobodies (Nbs) that specifically interact with the extracellular domain of RSP proteins. The procedure to obtain Nbs is lengthy and complicated and includes the immunization of dromedaries with purified RPS and the construction of a VHH (H-chain antibody variable region) library in E. coli. As RSP is expressed on the surface of E. coli, specific Nbs were able to agglutinate Salmonella strains harboring the p27 plasmid encoding the RSP proteins.

The authors demonstrated that Nbs-RSP reduced the conjugation frequency of p27 thus limiting the diffusion of the amp resistance harbored by the plasmid. This represents an innovative and promising strategy to fight antibiotic resistance, as it is not blocked by the mechanism that determines, in the specific case, the amp resistance of p27 but it targets an antigen associated with HincHI- derivative plasmids. Thus, RPS vaccination could be effective not only against Salmonella but also against other enteric bacteria. A possible criticism could be that Nbs against RSP proteins reduce the severity of the disease but do not completely prevent the infection by Salmonella.

Reviewer #2 (Public Review):

Summary:

This manuscript aims to tackle the antimicrobial resistance through the development of vaccines. Specifically, the authors test the potential of the RSP protein as a vaccine candidate. The RSP protein contains bacterial Ig-like domains that are typically carried in IncHl1 plasmids like R27. The extracellular location of the RSP protein and its role in the conjugation process makes it a good candidate for a vaccine. The authors then use Salmonella carrying an IncHl plasmid to test the efficacy of the RSP protein as a vaccine antigen in providing protection against infection of antibiotic-resistant bacteria carrying the IncHl plasmid. The authors found no differences in total IgG or IgA levels, nor in pro-inflammatory cytokines between immunized and non-immunized mice. They however found differences in specific IgG and IgA, attenuated disease symptoms, and restricted systemic infection.

The manuscript also evaluates the potential use of nanobodies specifically targeting the RSP protein by expressing it in E. coli and evaluating their interference in the conjugation of IncHl plasmids. The authors found that E. coli strains expressing RSP-specific nanobodies bind to Salmonella cells carrying the R27 plasmid thereby reducing the conjugation efficacy of Salmonella.

Strengths:

- The main strength of this manuscript is that it targets the mechanism of transmission of resistance genes carried by any bacterial species, thus making it broad.

- The experimental setup is sound and with proper replication.

Weaknesses:

- The two main experiments, evaluating the potential of the RSP protein and the effects of nanobodies on conjugation, seem as parts of two different and unrelated strategies.

- The survival rates shown in Figure 1A and Figure 3A for Salmonella pHCM1 and non-immunized mice challenged with Salmonella, respectively, are substantially different. In the same figures, the challenge of immunized mice and Salmonella pHCM1 and mice challenged with Salmonella pHCM1 with and without ampicillin are virtually the same. While this is not the only measure of the effect of immunization, the inconsistencies in the resulting survival curves should be addressed by the authors more thoroughly as they can confound the effects found in other parameters, including total and specific IgG and IgA, and pro-inflammatory cytokines.

- Overall the results are inconsistent and provide only partial evidence of the effectiveness of the RSP protein as a vaccine target.

- The conjugative experiments use very long conjugation times, making it harder to asses if the resulting transconjugants are the direct result of conjugation or just the growth of transconjugants obtained at earlier points in time. While this could be assessed from the obtained results, it is not a direct or precise measure.

- While the potential outcomes of these experiments could be applied to any bacterial species carrying this type of plasmids, it is unclear why the authors use Salmonella strains to evaluate it. The introduction does a great job of explaining the importance of these plasmids but falls short in introducing their relevance in Salmonella.

Author response:

The following is the authors’ response to the original reviews.

Public Reviews:

Reviewer #1 (Public Review):

The study by Prieto et al. faces the increasingly serious problem of bacterial resistance to antimicrobial agents. This work has an important element of novelty proposing a new approach to control antibiotic resistance spread by plasmids. Instead of targeting the resistance determinant, plasmid-borne proteins are used as antigens to be bound by specific nanobodies (Nbs). Once bound plasmid transfer was inhibited and Salmonella infection blocked. This in-depth study is quite detailed and complex, with many experiments (9 figures with multiple panels), rigorously carried out. Results fully support the authors' conclusions. Specifically, the authors investigated the role of two large molecular weight proteins (RSP and RSP2) encoded by the IncHI1 derivative-plasmid R27 of Salmonella. These proteins have bacterial Ig-like (Big) domains and are expressed on the cell surface, creating the opportunity for them to serve as immunostimulatory antigens. Using a mouse infection model, the authors showed that RSP proteins can properly function as antigens, in Salmonella strains harboring the IncHI1 plasmid. The authors clearly showed increased levels of specific IgG and IgA antibodies against these RSP proteins proteins in different tissues of immunized animals. In addition, non-immunized mice exhibited Salmonella colonization in the spleen and much more severe disease than immunized ones. 

However, the strength of this work is the selection and production of nanobodies (Nbs) that specifically interact with the extracellular domain of RSP proteins. The procedure to obtain Nbs is lengthy and complicated and includes the immunization of dromedaries with purified RPS and the construction of a VHH (H-chain antibody variable region) library in E. coli. As RSP is expressed on the surface of E. coli, specific Nbs were able to agglutinate Salmonella strains harboring the p27 plasmid encoding the RSP proteins. 

The authors demonstrated that Nbs-RSP reduced the conjugation frequency of p27 thus limiting the diffusion of the amp resistance harbored by the plasmid. This represents an innovative and promising strategy to fight antibiotic resistance, as it is not blocked by the mechanism that determines, in the specific case, the amp resistance of p27 but it targets an antigen associated with HincHI- derivative plasmids. Thus, RPS vaccination could be effective not only against Salmonella but also against other enteric bacteria. A possible criticism could be that Nbs against RSP proteins reduce the severity of the disease but do not completely prevent the infection by Salmonella.

It is true that vaccina2on of mice with purified RSP protein did not provide complete protec2on against infec2on with a Salmonella strain harboring an IncHI plasmid. As this finding is based on an animal model, further inves2ga2on is required to evaluate its clinical efficacy. In any case, even par2al protec2on provided by nanobodies or by a vaccine could poten2ally improve survival rates among cri2cally ill pa2ents infected with a pathogenic bacterium harboring an IncHI plasmid. An addi2onal beneficial aspect of our approach is that it will reduce dissemina2on of IncHI plasmids among pathogenic bacteria, which would reduce the presence of an2bio2c resistance plasmids in the environment and in the bacteria infec2ng pa2ents. 

Reviewer #2 (Public Review):

Summary:

This manuscript aims to tackle the antimicrobial resistance through the development of vaccines. Specifically, the authors test the potential of the RSP protein as a vaccine candidate. The RSP protein contains bacterial Ig-like domains that are typically carried in IncHl1 plasmids like R27. The extracellular location of the RSP protein and its role in the conjugation process makes it a good candidate for a vaccine. The authors then use Salmonella carrying an IncHl plasmid to test the efficacy of the RSP protein as a vaccine antigen in providing protection against infection of antibioticresistant bacteria carrying the IncHl plasmid. The authors found no differences in total IgG or IgA levels, nor in pro-inflammatory cytokines between immunized and non-immunized mice. They however found differences in specific IgG and IgA, attenuated disease symptoms, and restricted systemic infection.

The manuscript also evaluates the potential use of nanobodies specifically targeting the RSP protein by expressing it in E. coli and evaluating their interference in the conjugation of IncHl plasmids. The authors found that E. coli strains expressing RSPspecific nanobodies bind to Salmonella cells carrying the R27 plasmid thereby reducing the conjugation efficacy of Salmonella. 

Strengths:

The main strength of this manuscript is that it targets the mechanism of transmission of resistance genes carried by any bacterial species, thus making it broad.

The experimental setup is sound and with proper replication.

Weaknesses:

The two main experiments, evaluating the potential of the RSP protein and the effects of nanobodies on conjugation, seem as parts of two different and unrelated strategies.

In preparing our manuscript, we were aware that we included two different strategies to combat an2microbial resistance. However, we deemed it valuable to include both in the paper. The development of new vaccines and the inhibi2on of the transfer of an2bio2c resistance determinants are currently considered relevant approaches to combat an2microbial resistance. Our inten2on in the ar2cle is to integrate these two strategies. 

The survival rates shown in Figure 1A and Figure 3A for Salmonella pHCM1 and non-immunized mice challenged with Salmonella, respectively, are substantially different. In the same figures, the challenge of immunized mice and Salmonella pHCM1 and mice challenged with Salmonella pHCM1 with and without ampicillin are virtually the same. While this is not the only measure of the effect of immunization, the inconsistencies in the resulting survival curves should be addressed by the authors more thoroughly as they can confound the effects found in other parameters, including total and specific IgG and IgA, and pro-inflammatory cytokines.

Overall the results are inconsistent and provide only partial evidence of the effectiveness of the RSP protein as a vaccine target.

To address the concerns regarding the disparities in survival rates depicted in Figures 1A and 3A, it is important to refer to several factors that contribute to these variations. Firstly, it should be noted that the data depicted in these figures stem from distinct experimental sets conducted at different times employing different batches of mice. Despite the use of the same strain and supplier, individual animals and their batches can exhibit variability in susceptibility to infection due to inherent biological differences.

Unlike in vitro cell culture experiments, which can achieve high replicability due to the homogeneity of cell lines, in vivo animal studies often exhibit greater variability. This variability is influenced not only by genetic variations within animal populations, even if originating from the same supplier, but also by environmental factors within the animal facility. These factors include temperature variations, the concentration y of non-pathogenic microorganisms in the facility, which can modify the immune responses, or the density of animals in the environment, consequently affecting human traffic and generating potential disturbances. 

When designing experiments with animals, it is desirable for the results to be consistent across different animal batches. If one bacterial strain exhibits higher mortality rates than another across multiple experimental series, this pattern should be reproducible despite the inherent variability in in vivo studies. It is more important to demonstrate consistency in trends than to focus on absolute figures when validating experimental results. 

It is also important to clarify that when we refer to survival rates, it doesn’ t necessarily mean that the animals were found deceased. The animal procedures were approved by the Ethics Committee of Animal Experimentation of the Universitat de Barcelona, which include an animal monitoring protocol. Our protocol requires close daily monitoring of several health and behavioral parameters, each evaluated according to specific criteria. When an animal reaches a predetermined score threshold indicating severe distress or suffering, euthanasia is administered to alleviate further suffering. At this point, biological samples are collected for subsequent analysis.

The conjugative experiments use very long conjugation times, making it harder to assess if the resulting transconjugants are the direct result of conjugation or just the growth of transconjugants obtained at earlier points in time. While this could be assessed from the obtained results, it is not a direct or precise measure.

In the conjuga2on experiments we u2lized a reduced number of donor cells expressing the RSP protein and of recipient cells, as well as long conjuga2on 2mes, to reflect more accurately a situa2on that may occur naturally in the environment. Short conjuga2on 2mes are efficient in controlled laboratory condi2ons using high densi2es of donor and recipient cells, but these condi2ons are not commonly found in the environment. For the interference of the conjuga2ve transfer of the IncHI plasmid we used an E. coli strain displaying the nanobody binding RSP to simulate a process that could be also scaled-up in a natural environment (i.e., a probio2c strain in a livestock farm) and that could be cost effec2ve. See discussion sec2on, lanes 326-328.   

While the potential outcomes of these experiments could be applied to any bacterial species carrying this type of plasmids, it is unclear why the authors use Salmonella strains to evaluate it. The introduction does a great job of explaining the importance of these plasmids but falls short in introducing their relevance in Salmonella.

The prevalence of IncHI plasmids in Salmonella was indicated in the introduc2on sec2on, lanes 65-67. Nevertheless, we understand the reviewer’s cri2cisms and have modified both these sentences in the introduc2on sec2on and also added comments in the results sec2on (lanes 118-128).

Recommendations for the authors:

Reviewer #2 (Recommendations For The Authors):

I understand working with mice can be challenging in terms of repeating experiments to further support the study's claims. For this reason, I think the authors need to discuss more thoroughly the following things:

Can the authors comment on why the presence of Ampicillin leads to a lower upregulation of proinflammatory cytokines in the spleen despite harboring resistance against ampicillin?

At the intestinal level, physiological inflammatory responses play a crucial role in enabling the host to identify foreign and commensal bacterial antigens and initiate a highly regulated and "controlled" immune response (Fiocchi, 2008. Inflamm Bowel Dis. 2008, 14 Suppl 2:S77-8). The administration of antibiotics such as ampicillin, reduces the load of intestinal resident microbiota, thereby lowering the extent of intestinal immune activation. This decline in immune activation extends to systemic levels, potentially accounting for the reduced expression of proinflammatory cytokines observed in the spleen.

There are inconsistent results in the survival rates in Figures 1A and 3A, please discuss how this could alter the observed differences in total and specific IgG and IgA, and pro-inflammatory cytokines.

To address the reviewer concerns regarding the discrepancies in survival rates shown in Figures 1A and 3A, and how these differences might influence the observed variations in total and specific IgG and IgA, as well as pro-inflammatory cytokines, it is important to clarify the terminology used in our study. In our context, "survival" does not solely refer to mortality per se, but encompasses the endpoints defined by our animal welfare protocols, which are rigorously supervised by the Animal Experimentation Ethics Committee of the University of Barcelona. Our protocol mandates close daily monitoring of several health and behavioral parameters, each scored according to specific criteria. When an animal reaches a predefined score threshold indicating severe distress or suffering, euthanasia is conducted to prevent further distress, at which point we collect biological samples for analysis.

In contrast to in vitro cell culture experiments, which often achieve high replicability thanks to the homogeneity of cell lines, in vivo animal studies frequently display greater variability. This variability stems not only from genetic differences within animal populations, even if originating from the same supplier, but also from environmental factors within the animal facility. These factors encompass variations in temperature, the presence of non-pathogenic microorganisms in the facility (capable of altering immune responses) and the density of animals, which can impact human traffic and potentially lead to disturbances. 

The experiments depicted in Figs. 1A and 3A were separated in time, and hence may be influenced by environmental factors within the animal facility. Nevertheless, in the comparative analysis performed between immunized and non-immunized animals, experiments were performed simultaneously and hence under similar environmental conditions in the animal facility. For several parameters (i.e., immunoglobulins and proinflammatory cytokines) statistically significant differences were observed. 

Regarding the conjugation assays, it is not entirely clear to me why the conjugation times are so long. It would be beneficial to have more data about the conjugation efficacy between the donor and recipient without any E. coli expressing the nanobodies at different time intervals. This would help to differentiate between transconjugants and transconjugants obtained from early conjugation events.

This comment is par2ally answered in a previous response, regarding the numbers of donor and recipient cells and dura2on of conjuga2on.  We note here that in fig. 9, the requested experiment with donor and recipient cells without E. coli interferent cells is already present, corresponding to the label “none”. To avoid confusion, we have modified the legend in fig. 9.

eLife assessment

How the triplicate interaction between chemokines with both GAGs and G protein-coupled receptors (GPCR) works and how gradients are created and potentially maintained in vivo are poorly understood. The authors provide solid evidence to show phase separation can drive chemotactic gradient formation. The paper is a useful advance in the field of chemokine biology.

Joint Public Review:

Chemokines are known to create chemotactic gradients and it is generally recognized that in order to create these gradients they need to bind to glycosaminoglycans (GAGs) on cells and in tissues. However, how the triplicate interaction between chemokines with both GAGs and G protein-coupled receptors (GPCR) works and how gradients are created and potentially maintained in vivo is poorly understood. In their manuscript, Yu et al investigated and showed in detail the ability of soluble and cell-bound GAGs to create gradients of the chemokine CCL5. They show in vitro in a modified leukocyte migration assay that soluble GAGs and GAGs on the tumor cell line THP-1 affect leukocyte migration. This useful work contributes to our in-depth understanding of the role of GAGs in chemokine gradient creation which is important for site-directed leukocyte and potentially tumor cell migration and as such is of potential interest for scientists studying immune responses in infection, inflammation, autoimmunity and tumor biology. In their reply to the comments of both reviewers they indicate that liquid-liquid phase separation (LLPS) was not detected at lower CCL5 concentrations. This is important information since, together with the tendency of CCL5 to form oligomers, it may indicate that oligomerization is crucial for LLPS. This info should at least be added to the discussion of the manuscript.

Author response:

The following is the authors’ response to the previous reviews.

Public Reviews:

Reviewer #2 (Public Review):

Although the study by Xiaolin Yu et al is largely limited to in vitro data, the results of this study convincingly improve our current understanding of leukocyte migration.

(1) The conclusions of the paper are mostly supported by the data and in the revised manuscript clarification is provided concerning the exact CCL5 forms (without or with a fluorescent label or His-tag) and amounts/concentrations that were used in the individual experiments. This is important since it is known that modification of CCL5 at the N-terminus affects the interactions of CCL5 with the GPCRs CCR1, CCR3 and CCR5 and random labeling using monosuccinimidyl esters (as done by the authors with Cy-3) is targeting lysines. The revised manuscript more clearly indicates for each individual experiment which form is used. However, a discussion on the potential effects of the modifications on CCL5 in the results and discussion sections is still missing.

Many thanks for the reviewer's suggestion. We fully agree it is important to clarify the potential issue of Cy3 labeling, and believe it is more suitable in the Materials and Methods section (line 312-314).

(2) In general, authors used high concentrations of CCL5 in their experiments. In their reply to the comments they indicate that at lower CCL5 concentrations no LLPS is detected. This is important information since it may indicate the need for chemokine oligomerization for LLPS. This info should be added to the manuscript and comparison with for instance the obligate monomer CCL7 and another chemokine such as CXCL4 that easily forms oligomers may clarify whether LLPS is controlled by oligomerization.

We are pleased by the help of the reviewers and accordingly inserted a brief discussion as suggested (line 240-246).

(3) Statistical analyses have been improved in the revised manuscript.

Thanks to the reviewer for his/her comment.

Author response:

The following is the authors’ response to the original reviews.

eLife assessment

This valuable study uses a novel experimental design to elegantly demonstrate how we exploit stimulus structure to overcome working memory capacity limits. While the behavioural evidence is convincing, the neural evidence is incomplete, as it only provides partial support for the proposed information compression mechanism. This study will be of interest to cognitive neuroscientists studying structure learning and memory.

Public Reviews:

Reviewer #1 (Public Review):

Summary:

Huang and Luo investigated whether regularities between stimulus features can be exploited to facilitate the encoding of each set of stimuli in visual working memory, improving performance. They recorded both behavioural and neural (EEG) data from human participants during a sequential delayed response task involving three items with two properties: location and colour. In the key condition ('aligned trajectory'), the distance between locations of successively presented stimuli was identical to their 'distance' in colour space, permitting a compression strategy of encoding only the location and colour of the first stimulus and the relative distance of the second and third stimulus (as opposed to remembering 3 locations and 3 colours, this would only require remembering 1 location, 1 colour, and 2 distances). Participants recalled the location and colour of each item after a delay.

Consistent with the compression account, participants' location and colour recall errors were correlated and were overall lower compared to a non-compressible condition ('misaligned trajectory'). Multivariate analysis of the neural data permitted decoding of the locations and colours during encoding. Crucially, the relative distance could also be decoded - a necessary ingredient for the compression strategy.

Strengths:

The main strength of this study is a novel experimental design that elegantly demonstrates how we exploit stimulus structure to overcome working memory capacity limits. The behavioural results are robust and support the main hypothesis of compressed encoding across a number of analyses. The simple and well-controlled design is suited to neuroimaging studies and paves the way for investigating the neural basis of how environmental structure is detected and represented in memory. Prior studies on this topic have primarily studied behaviour only (e.g., Brady & Tenenbaum, 2013).

Thanks for the positive comments and excellent summary.

Weaknesses:

The main weakness of the study is that the EEG results do not make a clear case for compression or demonstrate its neural basis. If the main aim of this strategy is to improve memory maintenance, it seems that it should be employed during the encoding phase. From then on, the neural representation in memory should be in the compressed format. The only positive evidence for this occurs in the late encoding phase (the re-activation of decoding of the distance between items 1 and 2, Fig. 5A), but the link to behaviour seems fairly weak (p=0.068).

Thanks for raising this important concern. The reviewer is correct that in principle subjects should employ the compression strategy during the encoding phase when sequence stimuli are presented, yet our results show that the 1-2 trajectory could only be decoded during the late encoding phase.

Meanwhile, subjects could not get enough information to form the compressed strategy for the location and color sequences until the appearance of the 3rd item. Specifically, based on the first two items, the 1st and 2nd item, they only learn whether the 1st-2nd trajectories are congruent between location and color features. However, they could not predict whether it would also apply to the incoming 2nd-3rd trajectory. This is exactly what we found in neural decoding results. The 1st-2nd trajectory could be decoded after the 2nd item presentation, and the 2nd-3rd trajectory appears after the 3rd item onset. Most critically, the 1st-2nd trajectory is reactivated after the 3rd item but only for alignment condition, implicating formation of the full-sequence compression strategy wherein the previously formed 1st-2nd trajectory is reactivated to be connected to the 2nd-3rd trajectory.

Regarding the difference between higher- and lower-correlation groups, previously we used the time window based on the overall 2nd-3rd neural reactivations, which might not be sensitive to reactivation strength. We now re-chose the time window based on the higher-correlation group (bootstrap test, p = 0.037, two sides).

Results have been updated (Figure 5; Results, Page 16). Interpretations about the formation of compression strategy during encoding phase have been added to Results (Page 15-16) and Discussion (Page 18).

Stronger evidence would be showing decoding of the compressed code during memory maintenance or recall, but this is not presented. On the contrary, during location recall (after the majority of memory maintenance is already over), colour decoding re-emerges, but in the un-compressed item-by-item code (Fig. 4B). The authors suggest that compression is consolidated at this point, but its utility at this late stage is not obvious.

Thank you for the important question we apologize for omitting previously - neural evidence for the compressive account.

The reason we did not perform neural decoding during maintenance is that previous EEG/MEG studies including our own failed to reveal robust and sustained time-resolved memory decoding during this period. This is posited to arise from “activity-silent” WM states, wherein memories are not necessarily retained in sustained firing but silently stored within connection weights of WM networks (Stokes, Trends Cogn. Sci., 2015; Rose, Curr Dir Psychol Sci, 2020). Our previous work showed that by transiently perturbing the 'activity-silent' WM using a retrocue or neutral impulse, memories could be reactivated and robustly decoded from neural activities (Huang et al., eLife, 2021). However, due to the lack of transient events during retention in the current design, we do not expect robust decoding results during maintenance. As shown below (AB), this is indeed what we have observed, i.e., no robust neural decoding of trajectories during retention.

We further used alpha-band (8-11 Hz) neural activities, which have been shown to carry WM information (de Vries et al., Trends Cogn. Sci, 2020; Foster et al., Curr. Biol, 2016; Fukuda et al., J. Neurophysiol, 2016; Sutterer et al., PLOS Biol., 2019) to perform decoding analysis of compression trajectories during maintenance. As shown below, the alpha-band decoding results are indeed stronger than raw activities. Importantly, as shown below (CD), the aligned condition indeed showed significant and long-lasting decoding of compression trajectories (1st-2nd, 2nd-3rd) during retention, while the misaligned condition only showed decoding at the beginning (GH), which might be due to the non-specific offset response of the 3rd item. The results, although not as clear as those during encoding and recalling periods, support the reviewer’s hypothesis that the compressive strategy, if exploited, would be demonstrated during both encoding and maintenance periods. New results and related discussion have been added (Page 16, Supplementary Figure 4).

With regards to the observed item-by-item color replay during location recall, the reviewer was concerned that this was not consistent with the compressive account, given the lack of trajectory decoding.

First, item sequences stored in compressive formats need to be converted to sequences during serial recall. In other words, even though color and location sequences are retained in a compressive format (i.e., common 1st-2nd, 2nd-3rd trajectories) throughout the encoding and retention phases, they should be transferred to two sequences as outputs. This is exactly why we performed decoding analysis on individual color and location items rather than trajectories.

Second and most importantly, we observed serial replay of color sequences when recalling locations. In our view, these results constitute strong evidence for common structure, since the spontaneous color replay during location recall for aligned condition highlights the close bound between color and location sequences stored in WM. In fact, item-by-item serial replay has been well acknowledged as a critical neural index of cognitive maps, not only for spatial navigation but also for higher-order tasks (e.g., Liu et al., Cell, 2019; Liu et al., Science, 2021). Therefore, spontaneous color sequence replay during location sequence recall supports their shared underlying cognitive map.

Finally, spontaneous serial replay is also correlated with the reactivation of compressive trajectories during encoding (Supplementary Figure 3). This further indicates that serial replay during recalling is associated with memory reorganization formed during encoding.

Taken together, we posit that memories need to be converted to sequences as outputs, which leads to serial reactivations during recalling. Importantly, the observed spontaneous replay of color sequences for the aligned condition provides strong evidence supporting the associations between color and location sequences in WM.

We have now added relevant interpretations and discussions (Page 11&13).

Reviewer #2 (Public Review):

Summary:

In this study, the authors wanted to test if using a shared relational structure by a sequence of colors in locations can be leveraged to reorganize and compress information.

Strength:

They applied machine learning to EEG data to decode the neural mechanism of reinstatement of visual stimuli at recall. They were able to show that when the location of colors is congruent with the semantically expected location (for example, green is closer to blue-green than purple) the related color information is reinstated at the probed location. This reinstatement was not present when the location and color were not semantically congruent (meaning that x displacement in color ring location did not displace colors in the color space to the same extent) and semantic knowledge of color relationship could not be used for reducing the working memory load or to benefit encoding and retrieval in short term memory.

Weakness:

The experiment and results did not address any reorganization of information or neural mechanism of working memory (that would be during the gap between encoding and retrieval).

We apologize for not presenting clear neural evidence for memory reorganization, particularly neural decoding during WM maintenance and retrieval, in the previous version. As below, we explain why the findings provide converging neural evidence for WM reorganization based on a shared cognitive map.

First, during the encoding phase when location and color sequences are serially presented, our results reveal reactivation of the 1st-2nd trajectories upon the onset of the 3rd item when location and color sequences are aligned with each other. The reactivation of 1st-2nd trajectory right after the emergence of 2nd-3rd trajectory for aligned but not for misaligned sequences strongly supports WM reorganization, since only stimulus sequences that could be compressed based on shared trajectories (aligned condition) show the co-occurrence of 1st-2nd and 2nd-3rd trajectories. Moreover, the relevance of 1st-2nd reactivation to behavioral measurements of color-location reorganization (i.e., behavioral trajectory correlation, Figure 5D) further indicates its link to WM reorganization.

Second, the reason we originally did not perform neural decoding during maintenance is that previous EEG/MEG studies including our own failed to reveal robust and sustained time-resolved memory decoding during this period. This is posited to arise from “activity-silent” WM states, wherein memories are not necessarily retained in sustained firing but silently stored within connection weights of WM networks (Stokes, Trends Cogn. Sci., 2015; Wolff et al., Nat. Neurosci, 2017; Rose et al., Curr Dir Psychol Sci, 2020). Our previous work showed that by transiently perturbing the 'activity-silent' WM using a retrocue or neutral impulse, memories could be reactivated and robustly decoded from neural activities (Huang et al., eLife, 2021). However, due to the lack of transient events during retention in the current design, we do not expect robust decoding results during maintenance. As shown in Supplementary Figure 4(AB), this is indeed what we have observed, i.e., no robust neural decoding of trajectories during retention.

We then used alpha-band (8-11 Hz) neural activities, which have been found to carry WM information (de Vries et al., Trends Cogn. Sci, 2020; Foster et al., Curr. Biol, 2016; Fukuda et al., J. Neurophysiol, 2016; Sutterer et al., PLOS Biol., 2019) to perform decoding analysis of compression trajectories during maintenance. As shown below, the alpha-band decoding results are indeed stronger than raw activities. Importantly, as shown in Supplementary Figure 4(CD), the aligned condition indeed showed significant and long-lasting decoding of compression trajectories (1st-2nd, 2nd-3rd) during retention, while the misaligned condition only showed decoding at the beginning (GH), which might be due to the non-specific offset response of the 3rd item. The results, although not as clear as those during encoding and recalling periods, thus also support WM reorganization.

Finally, during the recalling period, we observed automatic serial replay of color sequences when recalling locations. In our view, these results constitute strong evidence for common structure, since the spontaneous color replay during location recall for aligned condition highlights the close bound between color and location sequences stored in WM. In fact, item-by-item serial replay has been well acknowledged as a critical neural index of cognitive maps, not only for spatial navigation but also for higher-order tasks (e.g., Liu et al., Cell, 2019; Liu et al., Science, 2021). Therefore, spontaneous replay of color sequence during location recall supports their shared underlying cognitive map. Moreover, the spontaneous serial replay is correlated with the reactivation of compressive trajectories during encoding (Supplementary Figure 3). This further indicates that serial replay during recalling is associated with memory reorganization formed during encoding.

Taken together, we have added updated results about the maintenance period (Page 16, Supplementary Figure 4) and included clarifications and interpretations about why the findings during the encoding and retrieval periods support the WM reorganization view (Page 15-16).

There was also a lack of evidence to rule out that the current observation can be addressed by schematic abstraction instead of the utilization of a cognitive map.

The likely impact of the initial submission of the study would be in the utility of the methods that would be helpful for studying a sequence of stimuli at recall. The paper was discussed in a narrow and focused context, referring to limited studies on cognitive maps and replay. The bigger picture and long history of studying encoding and retrieval of schema-congruent and schema-incongruent events is not discussed.

We agree with the reviewer that cognitive map referred here could be understood as schematic abstraction. Cognitive map refers to the internal representation of spatial relations in a specific environment (Tolman 1948). Schematic abstraction denotes a more broad range of circumstances, whereby the gist or structure of multiple environments or episodes can be integrated (Bartlett, 1932; Farzanfar et al., Nat. Rev. Neurosci, 2023).

In other words, schema refers to highly abstract framework of prior knowledge that captures common patterns across related experiences, which does not necessarily occur in a spatial framework as cognitive maps do. Meanwhile, in the current design, we specifically manipulate the consistency of spatial trajectory distance between color and location sequences. Therefore, we would argue that cognitive map is a more conservative and appropriate term to frame our findings.

Relevant discussions have been added (Page 3&19).

We apologize for the lack of more generalized discussion and have added schema-related literatures. Thanks for the suggestion.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

(1) Do time-frequency-domain data (e.g., alpha-band power) in the delay provide evidence for delay-period decoding of trajectory lengths? This might strengthen the case for compression.

Thanks for the suggestion. We now performed decoding analysis of the delay period based on alpha-band power. As shown in supplementary figure 4, both the 1st-2nd and 2nd-3rd trajectories could be decoded for the aligned condition.

Added in supplementary figure 4 and Page 16.  

(2) Do participants erroneously apply the compression strategy in the misaligned condition? This would not show up in the trajectory error correlation analysis, but might be visible when examining correlations between raw trajectory lengths.

Thanks for raising this interesting suggestion. To test the hypothesis, we chose a typical misaligned condition where 1st-2nd trajectory distances are same between location and color sequences, while the 2nd-3rd trajectory distances are different between the two features.

In this case, participants might exploit the compression strategy for the first two items and erroneously apply the strategy to the 3rd item. If so, we would expect better memory performance for the first two items but worse memory for the 3rd item, compared to the rest of misaligned trials. As shown below, the 1st-2nd aligned trials showed marginally significant higher performance than misaligned trials for the first two items (t(32) = 1.907, p = 0.066, Cohen’s d = 0.332) . Unfortunately, we did not find significant worse performance for the 3rd item between the two conditions (t(32) = -0.4847, p = 0.631, Cohen’s d = -0.084). We observed significant interactions between the last two items and the alignment effect (t(32) = 2.082, p = 0.045, Cohen’s d = 0.362), indicating a trend of applying wrong compression strategy to the 3nd item.

Author response image 1.

(3a) Some more detail on some of the methods might help readers. For instance, did trajectories always move in a clockwise direction? Could the direction reverse on the third item? If not, did this induce a response bias? Could such a bias possibly account for the trajectory error correlations

Sorry for the unclear statement. For individual trial, both the color and location features of the three items are randomly selected from nine possible values without any constraint about the directions. That is to say, the trajectories can move in a clockwise or anticlockwise direction, and the direction can also reverse on the third item in some trials. Thus, we think the current design can actually help us to reduce the influence of response bias. Taking a step back, if trajectory error correlations are due to response bias, we should expect consistent significant correlation for all conditions, instead of only observing significant correlation for 1st-2nd and 2nd-3rd trajectories but not for 1st-3rd trajectory and only in aligned trajectory condition but not in misaligned condition. Therefore, we think the trajectory error correlations cannot be simply explained by response bias.

Details have been added (Page 23).

(3b) Is the colour wheel always oriented the same way for a participant? If so, given there are only nine colors, it seems possible that colors are mapped to locations and remembered in a location code instead. This does not seem to be a problem in principle for the behavioural findings, but might change the interpretation of what is being decoded from the EEG. If this is a possibility then this might be acknowledged.

The color wheel is always oriented the same way for each participant. We agree with the reviewer that it is possible that participants tend to map colors to locations and remembered in a location code. We don’t have sufficient evidence to rule out this possibility. One possible way could be running another experiment with varied color wheel during response period. Meanwhile, we would like to point out that the underlying logic of the current design is based on the facts that thinking spatially is intuitive and spatial metaphors like “location” and “distance” is commonly used to describe world, e.g., the well-known mental number line (Dehaene et al., JEP: General, 1993). Therefore, we expected participants to associate or integrate location and color maps based on trajectory distance.

The reviewer is correct that the color decoding would reflect spatial location rather than the genuine color feature. This is actually the point of the experimental design, whereby two irrelevant features could be possibly combined within a common cognitive map. Without the realignment of the two feature maps defined in space, subjects could not at all form the strategy to compress the two sequences. In other words, decoding of color sequences could be understood as neural representation of a series of corresponding locations along the ring that are independent of the physical locations of the items.

Interpretations and clarifications have been added (Page 23&26).

(4) Does the discretisation of the stimulus distribution (to only 9 possible locations) make the compression strategy easier to use? If the features had been continuously distributed across the location/colour circle, would participants still pick up on and use the shared trajectory structure?

Thanks for the question. Without further data, it’s hard to say whether the discretization of the stimulus distribution would make the compression strategy easier to use or not, compared to continuous distribution. Both outcomes seem possible. On the one hand, discrete stimulus distribution would result in discrete trajectory distribution, which helps participants to realize the common trajectory strategy. On the other hand, discrete stimulus distribution would result in category or label representation, which may weaken the effectiveness of structure compression strategy. We postulate that our findings could be generalized to continuous trajectories in a cognitive map within certain resolution.

(5a) Minor point: I disagree that avoiding the same points for location and colour for a given item allows them to be independently decoded. I would argue the contrary - this kind of constraint should create a small anti-correlation that in principle could lead to spurious decoding of one variable (although this seems unlikely here).

We appreciate the concern. As mentioned above, with discrete stimulus distribution (9 possible values for both color and location domains), it is quite possible that a fraction of trials would share same values in location and color. Therefore, the neural decoding for one domain might be confounded by another domain. To dissociate their neural representations, we imposed constraints that color and location could not occupy the same value for a given item.

We agree that this kind of constraint might create a small anti-correlation, even though it is not observed here. Future studies using continuous stimulus distribution would reduce the correlation or anti-correlation between stimuli.

(5b) Very minor point: 1,000 permutations for significance testing seems on the low side. Since some of the p-values are close to 0.05 it may be worth running more permutations.

Thanks for this suggestion. We got similar results using 1000 or 10000 permutations.

(6) Missing reference: H. H. Li et al., 2021 (line 213) seems not to be on the list of references.

Sorry for the mistake. Added.

Reviewer #2 (Recommendations For The Authors):

The study aimed to discuss the working memory mechanism, instead, it seems to be focused on the encoding and recall strategies after a short while, I recommend updating the manuscript to refer to the relevant cognitive mechanism.

There was a strong voice on the effect of using the cognitive map in working memory, without any tests on if indeed a cognitive map was used (for example the novel link between stimuli and how a cognitive map can be used to infer shortcuts). Was the participant required to have any mental map beyond the schema of the shown color ring?

In the current experiment, to discuss if the effect is driven by utilizing a cognitive map or schematic abstraction of color-relatedness, further analysis is required to possibly assess the effects of schema on neural activity and behavior. Namely,<br /> (1) Was there any reinstatement of schematically congruent (expected) colors that were probed by location 1, at locations 2 and 3 in the MAT condition?

Thanks for pointing out this possibility. However, we don’t think there will be stable color expectations given location information under the MAT condition. First, as the trajectory distance varied on a trial-by-trial basis, no prior common trajectory knowledge could be used to make inference about the current stimuli in individual trial. Second, the starting points for color and location (1st item) were randomly and independently selected, such that color sequence could not be predicted based on the location sequence for both aligned and misaligned conditions.

(2) Given that response time can be a behavioral marker of schematic conflict, was the response time faster for congruent than incongruent conditions?

Thanks for this question. Unfortunately, due to the experimental design, the response time could not be used as a behavioral marker to infer mental conflicts, since participants were not required to respond as fast as possible. Instead, they took their own pace to reproduce sequences without time limit. They could even take a short break before submitting their response to initiate the next trial.

(3) In case you cannot rule out that utilizing schema is the cognitive mechanism that supports working memory performance (the behavior), please add the classical literature (on the memory of schematically congruent and incongruent events) to the discussion.

Thanks for this suggestion and we have added relevant literatures now (Page 3&19).

(4) On page 6, 'common structure in the cognitive map' is the schema, isn't it?

Correct. Based on our understanding, ‘common structure in the cognitive map’ is a spatial schema.

(5) In Figure 2 EFG, would you please use a mixed effect model or show evidence that all participants demonstrated a correlation between the location trajectory error and color trajectory error?

Thanks for the suggestion. We have added the mixed effect model results, which are consistent with Figure 2EFG (AT: 1st-2nd trajectory, β = 0.071, t = 4.215, p < 0.001; 2nd-3rd trajectory, β = 0.077, t = 3.570, p < 0.001; 1st-3rd trajectory, β = 0.019, t = 1.118, p = 0.264; MAT: 1st-2nd trajectory, β = 0.031, t = 1.572, p = 0.116; 2nd-3rd trajectory, β = 0.002, t = 0.128 , p = 0.898; 1st-3rd trajectory, β = -0.017, t = -1.024, p = 0.306).

In general, doesn't such correlation just show that good participants/trials were good (some did well in the study and some did poorly throughout?)

We don’t think the trajectory error correlation results just reveal that some participants did well and some participants did poorly. If that is the case, we shouldn’t observe significant correlation in Figure 2D, where we first run correlation for each participant and then test correlation significance at group level. Indeed, trajectory error correlation between color and location domains characterizes the consistent changes between the two domains.

It is worth to note that the correlation was estimated with signed trajectory errors in color and location domains, which meant that we indeed cared about whether the errors in the two domains were consistently varied in the same direction, i.e., whether longer trajectory memory compared to the actual trajectory in location domain would predict longer trajectory memory in color domain.

Moreover, as shown in Figure 2EFG, by dividing trials into 4 bins according to the location trajectory error for each participant and pooling the data across participants, we observed 4 clusters along x-axis (location trajectory error). This suggests that participants’ memory performance is rather consistent instead of being extremely good or bad. Besides, if trajectory error correlation is due to different overall memory performance between participants, we should observe significant trajectory error correlations both in AT and MAT conditions, instead of only under AT condition and for 1st-2nd and 2nd-3rd trajectories but not for 1st-3rd trajectory.

In Figure 2 G, is the marginal error just too big to be sensitive? I am not sure what we are learning here, please clarify.

Sorry for the confusion. To examine this possibility, we excluded errors which are beyond 2.5 * σ, and still observed non-significant 1st-3rd trajectory error correlation between color and location domains (r = 0.119, p = 0.167).

The 1st-3rd trajectory showed nonsignificant behavioral correlation and neural representation, which suggests that the current sequential memory task would encourage participants to organize all information by relying more on the adjacent items and their distance. Thus, we think the 1st-3rd trajectory would serve as a control trajectory, which helps us not only exclude other possible explanation (e.g., systematic response bias), but also validate current findings both in behavioral and neural level.

Results and statements (Page 10-11) added now.

Author response image 2.

(6) Regarding the first lines on page 11, did you do qualitative research to know if less information was encoded in congruent conditions?

The current experimental design is inspired by the mental compression of spatial sequence studies from Dehaene’s lab (Amalric er al., 2017; Roumi et al., 2021), in which they propose that human brain compresses spatial sequence using an abstract language and formalize minimal description length of a sequence as the “language-of-thought complexity.” Based on this evidence, we think less information is required to describe congruent condition compared to incongruent condition. This idea is supported by better memory performance for congruent condition. Unfortunately, we couldn’t manage to quantify how less information was encoded in congruent condition.

eLife assessment

This valuable study uses a novel experimental design to elegantly demonstrate how we exploit stimulus structure to overcome working memory capacity limits. The presented behavioural and neural evidence are solid and in line with the proposed information compression mechanism. This study will be of interest to cognitive neuroscientists studying structure learning and memory.

Reviewer #1 (Public Review):

Summary:

Huang and Luo investigated whether regularities between stimulus features can be exploited to facilitate the encoding of each set of stimuli in visual working memory, improving performance. They recorded both behavioural and neural (EEG) data from human participants during a sequential delayed response task involving three items with two properties: location and colour. In the key condition ('aligned trajectory'), the distance between locations of successively presented stimuli was identical to their 'distance' in colour space, permitting a compression strategy of encoding only the location and colour of the first stimulus and the relative distance of the second and third stimulus (as opposed to remembering 3 locations and 3 colours, this would only require remembering 1 location, 1 colour, and 2 distances). Participants recalled the location and colour of each item after a delay.

Consistent with the compression account, participants' location and colour recall errors were correlated and overall lower compared to a non-compressible condition ('misaligned trajectory'). Multivariate analysis of the neural data permitted decoding of the locations and colours during encoding. Crucially, the relative distance could also be decoded - a necessary ingredient for the compression strategy.

Strengths:

The main strength of this study is a novel experimental design that elegantly demonstrates how we exploit stimulus structure to overcome working memory capacity limits. The behavioural results are robust and support the main hypothesis of compressed encoding across a number of analyses. The simple and well-controlled design is suited to neuroimaging studies and paves the way for investigating the neural basis of how environmental structure is detected and represented in memory. Prior studies on this topic have primarily studied behaviour only (e.g., Brady & Tenenbaum, 2013).

Weaknesses:

The main weakness of the study is that the EEG results could make a clearer case for compression. There is some evidence that distance decoding is present in alpha-band activity in the maintenance delay, but the strongest evidence for this occurs only briefly in the late encoding phase (the re-activation of decoding of the distance between items 1 and 2, Fig. 5A). The link to behaviour (Fig. 5D) seems fairly weak and based on a potentially circular analysis. During location recall, colour decoding re-emerges and is reactivated in sequence, but this finding is consistent both with compression-based and conventional rehearsal mechanisms. Nevertheless, the balance of evidence appears to favour the compression account.

Impact:

This important study elegantly demonstrates that the use of shared structure can improve capacity-limited visual working memory. The paradigm and approach explicitly link this field to recent findings on the role of replay in structure learning and will therefore be of interest to neuroscientists studying both topics.

eLife assessment

This study focuses on the regulation of GLP-1 in enteroendocrine L cells and how this may be stimulated by the mechanogated ion channel Piezo1. The work is innovative and the hypothesis that is being tested may have important mechanistic and translational implications. The data remains incomplete at present and needs a substantial amount of supporting evidence and corrections to be a stronger manuscript and publication.

Reviewer #1 (Public Review):

Summary:

In this manuscript, the authors intended to prove that gut GLP-1 expression and secretion can be regulated by Piezo1, and hence by mechanistic/stretching regulation. For this purpose, they have assessed Piezo1 expression in STC-1 cell line (a mouse GLP-1 producing cell line) and mouse gut, showing the correlation between Piezo1 level and Gcg levels (Figure S1). They then aimed to generate gut L cell-specific Piezo1 KO mice, and claimed the mice show impaired glucose tolerance and GLP-1 production, which can be mitigated by Ex-4 treatment (Figures 1-2). Pharmacological agents (Yoda1 and GsMTx4) and mechanic activation (intestinal bead implantation) were then utilized to prove the existence of ileal Piezo1-regulated GLP-1 synthesis (Figure 3). This was followed by testing such mechanism in a limited amount of primary L cells and mainly in the STC-1 cell line (Figures 4-7).

While the novelty of the study is somehow appreciable, the bio-medical significance is not well demonstrated in the manuscript. The authors stated (in lines between lines 78-83) a number of potential side effects of GLP-1 analogs, how can the mechanistic study of GLP-1 production on its own be essential for the development of new drug targets for the treatment of diabetes. Furthermore, the study does not provide a clear mechanistic insight on how the claimed CaMKKbeta/CaMKIV-mTORC1 signaling pathway upregulated both GLP-1 production and secretion. This reviewer also has concerns about the experimental design and data presented in the current manuscript, including the issue of how proglucagon expression can be assessed by Western blotting.

Strengths:

The novelty of the concept.

Weaknesses:

Experimental design and key experiment information.

Reviewer #2 (Public Review):

Summary:

The study by Huang and colleagues focuses on GLP-1 producing entero-endocrine (EEC) L-cells and their regulation of GLP-1 production by a mechano-gated ion channel Piezo1. The study describes Piezo1 expression by L-cells and uses an exciting intersectional mouse model (villin to target epithelium and Gcg to target GLP-1-producing cells and others like glucagon-producing pancreatic endocrine cells), which allows L-cell specific Piezo1 knockout. Using this model, they find an impairment of glucose tolerance, increased body weight, reduced GLP-1 content, and changes to the CaMKKbeta-CaMKIV-mTORC1 signaling pathway using a normal diet and then high-fat diet. Piezo1 chemical agonist and intestinal bead implantation reversed these changes and improved the disrupted phenotype. Using primary sorted L-cells and cell model STC-1, they found that stretch and Piezo1 activation increased GLP-1 and altered the molecular changes described above.

Strengths:

This is an interesting study testing a novel hypothesis that may have important mechanistic and translational implications. The authors generated an important intersectional genetics mouse model that allowed them to target Piezo1 L-cells specifically, and the surprising result of impaired metabolism is intriguing.

Weaknesses:

However, there are several critical limitations that require resolution before making the conclusions that the authors make.

(1) A potential explanation for the data, and one that is consistent with existing literature [see for example, PMC5334365, PMC4593481], is that epithelial Piezo1, which is broadly expressed by the GI epithelium, impacts epithelial cell density and survival, and as such, if Piezo1 is involved in L-cell physiology, it may be through regulation of cell density. Thus, it is critical to determine L-cell densities and epithelial integrity in controls and Piezo1 knockouts systematically across the length of the gut, since the authors do not make it clear which gut region contributes to the phenotype they see. Current immunohistochemistry data are not convincing.

(2) Calcium signaling in L-cells is implicated in their typical role of being gut chemo-sensors, and Piezo1 is a calcium channel, so it is not clear whether any calcium-related signaling mechanism would phenocopy these results.

(3) Intestinal bead implantation, while intriguing, does not have clear mechanisms - and is likely to provide a point of intestinal obstruction and dysmotility.

(4) Previous studies, some that are very important, but not cited, contradict the presented results (e.g., epithelial Piezo1 role in insulin secretion) and require reconciliation.

Overall, this study makes an interesting observation but the data are not currently strong enough to support the conclusions.

Author response:

Reviewer #1 (Public Review):

Summary:

In this manuscript, the authors intended to prove that gut GLP-1 expression and secretion can be regulated by Piezo1, and hence by mechanistic/stretching regulation. For this purpose, they have assessed Piezo1 expression in STC-1 cell line (a mouse GLP-1 producing cell line) and mouse gut, showing the correlation between Piezo1 level and Gcg levels (Figure S1). They then aimed to generate gut L cell-specific Piezo1 KO mice, and claimed the mice show impaired glucose tolerance and GLP-1 production, which can be mitigated by Ex-4 treatment (Figures 1-2). Pharmacological agents (Yoda1 and GsMTx4) and mechanic activation (intestinal bead implantation) were then utilized to prove the existence of ileal Piezo1-regulated GLP-1 synthesis (Figure 3). This was followed by testing such mechanism in a limited amount of primary L cells and mainly in the STC-1 cell line (Figures 4-7).

While the novelty of the study is somehow appreciable, the bio-medical significance is not well demonstrated in the manuscript. The authors stated (in lines between lines 78-83) a number of potential side effects of GLP-1 analogs, how can the mechanistic study of GLP-1 production on its own be essential for the development of new drug targets for the treatment of diabetes. Furthermore, the study does not provide a clear mechanistic insight on how the claimed CaMKKbeta/CaMKIV-mTORC1 signaling pathway upregulated both GLP-1 production and secretion. This reviewer also has concerns about the experimental design and data presented in the current manuscript, including the issue of how proglucagon expression can be assessed by Western blotting.

Strengths:

The novelty of the concept.

Weaknesses:

Experimental design and key experiment information.

Current GLP-1-based therapies for diabetes use GLP-1 agonists/analogs. Although generally safe, there are some side effect or risks of GLP-1 agonists/analogs. We agree to the reviewer that a mechanistic study on the regulation of GLP-1 production will not directly lead to development of new drug targets for the treatment of diabetes. However, understanding the mechanism of GLP-1 production may shed light onto alternative treatment strategies for diabetes that targeting the production of GLP-1. In our previous studies, we have elucidated the role of mTOR/S6K pathway in regulating GLP-1 production in L cells. Using STC-1 cell line and different mouse models, including Neurog3-Tsc1−/− mice, rapamycin or L-lucine treatment to stimulate mTOR activity, we have demonstrated that mTOR stimulates proglucagon gene expression and thus GLP-1 production (Diabetologia 2015;58(8):1887-97； Mol Cell Endocrinol. 2015 Nov 15:416:9-18.). Based on our previous studies, we found that Piezo1 regulated mTOR/S6K pathway and thus proglucagon expression and GLP-1 production through Ca2+/CaMKKbeta/CaMKIV in our present study. Although we could not exclude involvement of other signaling pathways downstream of Piezo1 in regulating the cleavage of proglucagon, granule maturation and the final release of GLP-1, our present study provided evidence to support the involvement of the Ca2+/CaMKKbeta/CaMKIV/mTOR pathway in mediating the role Piezo1 in proglucagon expression and GLP-1 production. The reviewer also expressed concerns on the use of western blot to detect proglucagon expression. In fact, western blot is often used in detection of proglucagon. Here are some examples from other researchers: Diabetes. 2013 Mar;62(3):789-800. Gastroenterology. 2011 May;140(5):1564-74. 2004 Jul 23;279(30):31068-75. The proglucagon antibody we used in our study was purchased from abcam (Cat#ab23468), which can detect proglucagon of 21 kDa.

Reviewer #2 (Public Review):

Summary:

The study by Huang and colleagues focuses on GLP-1 producing entero-endocrine (EEC) L-cells and their regulation of GLP-1 production by a mechano-gated ion channel Piezo1. The study describes Piezo1 expression by L-cells and uses an exciting intersectional mouse model (villin to target epithelium and Gcg to target GLP-1-producing cells and others like glucagon-producing pancreatic endocrine cells), which allows L-cell specific Piezo1 knockout. Using this model, they find an impairment of glucose tolerance, increased body weight, reduced GLP-1 content, and changes to the CaMKKbeta-CaMKIV-mTORC1 signaling pathway using a normal diet and then high-fat diet. Piezo1 chemical agonist and intestinal bead implantation reversed these changes and improved the disrupted phenotype. Using primary sorted L-cells and cell model STC-1, they found that stretch and Piezo1 activation increased GLP-1 and altered the molecular changes described above.

Strengths:

This is an interesting study testing a novel hypothesis that may have important mechanistic and translational implications. The authors generated an important intersectional genetics mouse model that allowed them to target Piezo1 L-cells specifically, and the surprising result of impaired metabolism is intriguing.

Weaknesses:

However, there are several critical limitations that require resolution before making the conclusions that the authors make.

(1) A potential explanation for the data, and one that is consistent with existing literature [see for example, PMC5334365, PMC4593481], is that epithelial Piezo1, which is broadly expressed by the GI epithelium, impacts epithelial cell density and survival, and as such, if Piezo1 is involved in L-cell physiology, it may be through regulation of cell density. Thus, it is critical to determine L-cell densities and epithelial integrity in controls and Piezo1 knockouts systematically across the length of the gut, since the authors do not make it clear which gut region contributes to the phenotype they see. Current immunohistochemistry data are not convincing.

We appreciate the reviewer’s comment. We agree that Piezo1 may affect L-cell density and epithelial integrity. We will do quantification of L-cell density and test the epithelial integrity by examining the expression of tight junction proteins (ZO-1 and Occludin) and determine the transepithelial resistance in different regions of the gut

(2) Calcium signaling in L-cells is implicated in their typical role of being gut chemo-sensors, and Piezo1 is a calcium channel, so it is not clear whether any calcium-related signaling mechanism would phenocopy these results.

We will examine whether other calcium-related signaling mechanism also contribute the phenotype seen in the IntL-Piezo1-/- mice.

(3) Intestinal bead implantation, while intriguing, does not have clear mechanisms - and is likely to provide a point of intestinal obstruction and dysmotility.

To ascertain if intestinal bead implantation led to intestinal obstruction and dysmotility, we conducted a bowel transit time test. The results revealed no difference in bowel transit time between the sham-operated mice and those implanted with beads.

(4) Previous studies, some that are very important, but not cited, contradict the presented results (e.g., epithelial Piezo1 role in insulin secretion) and require reconciliation.

Overall, this study makes an interesting observation but the data are not currently strong enough to support the conclusions.

We will cite more previous studies on GLP-1 production and discuss the discrepancy between our study and others’ studies. The lack of changes in blood glucose seen in Villin-Piezo1-/- mice reported by Sugisawa et. al. is not surprising (Cell. 2020 Aug 6;182(3):609-624.e21.). Actually, in another recent study from our group, we found similar results when the Villin-Piezo1-/- mice Piezo1fl/fl control mice were fed with normal chow diet. Since Villin-1 is expressed in all the epithelial cells of the gut, including enterocytes and various types of endocrine cells, the effect of L-cell Piezo1 loss may be masked by other cell types under normal condition. However, impair glucose tolerance was seen in Villin-Piezo1-/- mice compared to the Piezo1fl/fl control mice after high fat diet for 8 weeks. We further found that Piezo1 in enterocytes exerted a negative effect on the glucose and lipid absorption. Loss of Piezo1 in enterocytes led to over-absorption of nutrients under high-fat diet （Tian Tao, Qing Shu, Yawen Zhao, Wenying Guo, Jinting Wang, Yuhao Shi, Shiqi Jia, Hening Zhai, Hui Chen, Cunchuan Wang*, Geyang Xu*, Mechanical regulation of lipid and sugar absorption by Piezo1 in enterocytes, Acta Pharmaceutica Sinica B , Accepted, 2024，https://doi.org/10.1016/j.apsb.2024.04.016）.

Author response:

The following is the authors’ response to the original reviews.

Your editorial guidance, reviews, and suggestions have led us to make substantial changes to our manuscript. While we detail point-by-point responses in typical fashion below, I wanted to outline, at a high level, what we’ve done.

(1) Methods. Your suggestions led us to rethink our presentation of our methods, which are now described more cohesively in a new methods section in the main text.

(2) Model Validation & Robustness. Reviewers suggested various validations and checks to ensure that our findings were not, for instance, the consequence of a particular choice of parameter. These can be found in the supplementary materials.

(3) Data Cleaning & Inclusion/Exclusion. Finally, based on feedback, our new methods section fully describes the process by which we cleaned our original data, and on what grounds we included/excluded individual faculty records from analysis.

eLife assessment

Efforts to increase the representation of women in academia have focussed on efforts to recruit more women and to reduce the attrition of women. This study - which is based on analyses of data on more than 250,000 tenured and tenure-track faculty from the period 2011-2020, and the predictions of counterfactual models - shows that hiring more women has a bigger impact than reducing attrition. The study is an important contribution to work on gender representation in academia, and while the evidence in support of the findings is solid, the description of the methods used is in need of improvement.

Reviewer #1 (Public Review):

Summary and strengths

This is an interesting paper that concludes that hiring more women will do more to improve the gender balance of (US) academia than improving the attrition rates of women (which are usually higher than men's). Other groups have reported similar findings but this study uses a larger than usual dataset that spans many fields and institutions, so it is a good contribution to the field.

We thank the reviewer for their positive assessment of the contributions of our work.

Weaknesses

The paper uses a mixture of mathematical models (basically Leslie matrices, though that term isn't mentioned here) parameterised using statistical models fitted to data. However, the description of the methods needs to be improved significantly. The author should consider citing Matrix Population Models by Caswell (Second Edition; 2006; OUP) as a general introduction to these methods, and consider citing some or all of the following as examples of similar studies performed with these models:

Shaw and Stanton. 2012. Proc Roy Soc B 279:3736-3741

Brower and James. 2020. PLOS One 15:e0226392

James and Brower. 2022. Royal Society Open Science 9:220785 Lawrence and Chen. 2015.

[http://128.97.186.17/index.php/pwp/article/view/PWP-CCPR-2015-008]

Danell and Hjerm. 2013. Scientometrics 94:999-1006

We have expanded the description of methods in a new methods section of the paper which we hope will address the reviewer’s concerns.

We agree that our model of faculty hiring and attrition resembles Leslie matrices. In results section B, we now mention Leslie matrices and cite Matrix Population Models by Caswell, noting a few key differences between Leslie matrices and the model of hiring and attrition presented in this work. Most notably, in the hiring and attrition model presented, the number of new hires is not based on per-capita fertility constants. Instead, population sizes are predetermined fixed values for each year, precluding exponential population growth or decay towards 0 that is commonly observed in the asymptotic behavior of linear Leslie Matrix models.

We have additionally revised the main text to cite the listed examples of similar studies (we had already cited James and Brower, 2022). We thank the reviewer for bringing these relevant works to our attention.

The analysis also runs the risk of conflating the fraction of women in a field with gender diversity! In female-dominated fields (e.g. Nursing, Education) increasing the proportion of women in the field will lead to reduced gender diversity. This does not seem to be accounted for in the analysis. It would also be helpful to state the number of men and women in each of the 111 fields in the study.

We have carefully examined the manuscript and revised the text to correctly differentiate between gender diversity and women’s representation.

We have additionally added a table to the supplemental materials (Tab. S3) that reports the estimated number of men and women in each of the 111 fields.

Reviewer #2 (Public Review):

Summary:

This important study by LaBerge and co-authors seeks to understand the causal drivers of faculty gender demographics by quantifying the relative importance of faculty hiring and attrition across fields. They leverage historical data to describe past trends and develop models that project future scenarios that test the efficacy of targeted interventions. Overall, I found this study to be a compelling and important analysis of gendered hiring and attrition in US institutions, and one that has wide-reaching policy implications for the academy. The authors have also suggested a number of fruitful future avenues for research that will allow for additional clarity in understanding the gendered, racial, and socioeconomic disparities present in US hiring and attrition, and potential strategies for mitigating or eliminating these disparities.

We thank the reviewer for their positive assessment of the contributions of our work.

Strengths:

In this study, LaBerge et al use data from over 268,000 tenured and tenure-track faculty from over 100 fields at more than 12,000 PhD-granting institutions in the US. The period they examine covers 2011-2020. Their analysis provides a large-scale overview of demographics across fields, a unique strength that allows the authors to find statistically significant effects for gendered attrition and hiring across broad areas (STEM, non-STEM, and topical domains).

LaBerge et al. find gendered disparities in attrition-using both empirical data and their counterfactual model-that account for the loss of 1378 women faculty across all fields between 2011 and 2020. It is true that "this number is both a small portion of academia... and a staggering number of individual careers," as ." - as this loss of women faculty is comparable to losing more than 70 entire departments. I appreciate the authors' discussion about these losses-they note that each of these is likely unnecessary, as women often report feeling that they were pushed out of academic jobs.

LaBerge et al. also find-by developing a number of model scenarios testing the impacts of hiring, attrition, or both-that hiring has a greater impact on women's representation in the majority of academic fields in spite of higher attrition rates for women faculty relative to men at every career stage. Unlike many other studies of historical trends in gender diversity, which have often been limited to institution-specific analyses, they provide an analysis that spans over 100 fields and includes nearly all US PhD-granting institutions. They are able to project the impacts of strategies focusing on hiring or retention using models that project the impact of altering attrition risk or hiring success for women. With this approach, they show that even relatively modest annual changes in hiring accumulate over time to help improve the diversity of a given field. They also demonstrate that, across the model scenarios they employ, changes to hiring drive the largest improvement in the long-term gender diversity of a field.

Future work will hopefully - as the authors point out - include intersectional analyses to determine whether a disproportionate share of lost gender diversity is due to the loss of women of color from the professoriate. I appreciate the author's discussion of the racial demographics of women in the professoriate, and their note that "the majority of women faculty in the US are white" and thus that the patterns observed in this study are predominately driven by this demographic. I also highly appreciate their final note that "equal representation is not equivalent to equal or fair treatment," and that diversifying hiring without mitigating the underlying cause of inequity will continue to contribute to higher losses of women faculty.

Weaknesses

First, and perhaps most importantly, it would be beneficial to include a distinct methods section. While the authors have woven the methods into the results section, I found that I needed to dig to find the answers to my questions about methods. I would also have appreciated additional information within the main text on the source of the data, specifics about its collection, inclusion and exclusion criteria for the present study, and other information on how the final dataset was produced. This - and additional information as the authors and editor see fit - would be helpful to readers hoping to understand some of the nuance behind the collection, curation, and analysis of this important dataset.

We have expanded upon the description of methods in a new methods section of the paper.

We have also added a detailed description of the data cleaning steps taken to produce the dataset used in these analyses, including the inclusion/exclusion criteria applied. This detailed description is at the beginning of the methods section. This addition has substantially enhanced the transparency of our data cleaning methods, so we thank the reviewer for this suggestion.

I would also encourage the authors to include a note about binary gender classifications in the discussion section. In particular, I encourage them to include an explicit acknowledgement that the trends assessed in the present study are focused solely on two binary genders - and do not include an analysis of nonbinary, genderqueer, or other "third gender" individuals. While this is likely because of the limitations of the dataset utilized, the focus of this study on binary genders means that it does not reflect the true diversity of gender identities represented within the professoriate.

In a similar vein, additional context on how gender was assigned on the basis of names should be added to the methods section.

We use a free, open-source, and open-data python package called nomquamgender (Van Buskirk et al, 2023) to estimate the strengths of (culturally constructed) name-gender associations. For sufficiently strong associations with a binary gender, we apply those labels to the names in our data. We have updated the main text to make this approach more apparent.

We have also added language to the main text which explicitly acknowledges that our approach only assigns binary (woman/man) labels to faculty. We point out that this is a compromise due to the technical limitations of name-based gender methodologies and is not intended to reinforce a gender binary.

I do think that some care might be warranted regarding the statement that "eliminating gendered attrition leads to only modest changes in field-level diversity" (Page 6). while I do not think that this is untrue, I do think that the model scenarios where hiring is "radical" and attrition is unchanged from present (equal representation of women and men among hires (ER) + observed attrition (OA)) shows that a sole focus on hiring dampens the gains that can otherwise be addressed via even modest interventions (see, e.g., gender-neutral attrition (GNA) + increasing representation of women among hires (IR)). I am curious as to why the authors did not include an additional scenario where hiring rates are equal and attrition is equalized (i.e., GNA + ER). The importance of including this additional model is highlighted in the discussion, where, on Page 7, the authors write: "In our forecasting analysis, we find that eliminating the gendered attrition gap, in isolation, would not substantially increase representation of women faculty in academia. Rather, progress towards gender parity depends far more heavily on increasing women's representation among new faculty hires, with the greatest change occurring if hiring is close to gender parity." I believe that this statement would be greatly strengthened if the authors can also include a comparison to a scenario where both hiring and attrition are addressed with "radical" interventions.

Our rationale for omitting the GNA + ER scenario in the presented analysis is that we can reason about the outcomes of this scenario without the need for computation; if a field has equal inputs of women and men faculty (on average) and equal retention rates between women and men (on average), then, no matter the field’s initial age and gender distribution of faculty, the expected value for the percentage of women faculty after all of the prior faculty have retired (which may take 40+ years) is exactly 50%. We have updated the main text to discuss this point.

Reviewer #3 (Public Review):

This manuscript investigates the roles of faculty hiring and attrition in influencing gender representation in US academia. It uses a comprehensive dataset covering tenured and tenure-track faculty across various fields from 2011 to 2020. The study employs a counterfactual model to assess the impact of hypothetical gender-neutral attrition and projects future gender representation under different policy scenarios. The analysis reveals that hiring has a more significant impact on women's representation than attrition in most fields and highlights the need for sustained changes in hiring practices to achieve gender parity.

Strengths:

Overall, the manuscript offers significant contributions to understanding gender diversity in academia through its rigorous data analysis and innovative methodology.

The methodology is robust, employing extensive data covering a wide range of academic fields and institutions.

Weaknesses:

The primary weakness of the study lies in its focus on US academia, which may limit the generalizability of its findings to other cultural and academic contexts.

We agree that the U.S. focus of this study limits the generalizability of our findings. The findings that we present in this work will only generalize to other populations–whether it be to an alternate industry, e.g., tech workers, or to faculty in different countries–to the extent that these other populations share similar hiring patterns, retention patterns, and current demographic representation. We have added a discussion of this limitation to the manuscript.

Additionally, the counterfactual model's reliance on specific assumptions about gender-neutral attrition could affect the accuracy of its projections.

Our projection analysis is intended to illustrate the potential gender representation outcomes of several possible counterfactual scenarios, with each projection being conditioned on transparent and simple assumptions. In this way, the projection analysis is not intended to predict or forecast the future.

To resolve this point for our readers, we now introduce our projections in the context of the related terms of prediction and forecast, noting that they have distinct meanings as terms of art: On one hand, prediction and forecasting involve anticipating a specific outcome based on available information and analysis, and typically rely on patterns, trends, or historical data to make educated guesses about what will happen. Projections are based on assumptions and are often presented in a panel of possible future scenarios. While predictions and forecasts aim for precision, projections (which we make in our analysis) are more generalized and may involve a range of potential outcomes.

Additionally, the study assumes that whoever disappeared from the dataset is attrition in academia. While in reality, those attritions could be researchers who moved to another country or another institution that is not included in the AARC (Academic Analytics Research Centre) dataset.

In our revision, we have elevated this important point, and clarified it in the context of the various ways in which we count hires and attritions. We now explicitly state that “We define faculty hiring and faculty attrition to include all cases in which faculty join or leave a field or domain within our dataset.” Then, we enumerate the number of situations that could be counted as hires and attritions, including the reviewer’s example of faculty who move to another country.

Reviewer #1 (Recommendations For The Authors):

Section B: The authors use an age structured Leslie matrix model (see Caswell for a good reference to these) to test the effect of making the attrition rates or hiring rates equal for men and women. My main concern here is the fitting techniques for the parameters. These are described (a little too!) briefly in section S1B. Some specific questions that are left hanging include:

A 5th order polynomial is an interesting choice. Some statistical evidence as to why it was the best fit would be useful. What other candidate models were compared? What was the "best fit" judgement made with: AIC, r^2? What are the estimates for how good this fit is? How many data points were fitted to? Was it the best fit choice for all of the 111 fields for men and women?

We use a logistic regression model for each field to infer faculty attrition probabilities across career ages and time, and we include the career age predictor up to its fifth power to capture the career-age correlations observed in Spoon et. al., Science Advances, 2023. For ease of reference, we reproduce the attrition risk curves in Fig S4.

We note that faculty attrition rates start low and then reach a peak around 5-7 years after earning PhD, and then decline until around 15-20 years post-PhD, after which, attrition rates increase as faculty approach retirement.

This function shape starts low and ends high, and includes at least one local minimum, which indicates that career age should be odd-ordered in the model and at least order-3, but only including career age up to its 3rd order term tended to miss some of the overserved career-age/attrition correlations. We evaluated the fit using 5-fold cross validation with a Brier score loss metric, and among options of polynomials of degree 1, 3, 5, or 7, we found that 5th order performed well overall on average over all fields (even if it was not the best for every field), without overfitting in fields with fewer data. Example fits, reminiscent of the figure from Spoon et al, are now provided in Figs S4 and S5.

While the model fit with fifth order terms may not be the best fit for all 111 fields (e.g., 7th order fits better in some cases), we wanted to avoid field-specific curves that might be overfitted to the field-specific data, especially due to low sample size (and thus larger fluctuations) on the high career age side of the function. Our main text and supplement now includes justifications for our choice to include career age up to its fifth order terms.

You used the 5th order logistic regression (bottom of page 11) to model attrition at different ages. The data in [24] shows that attrition increases sharply, then drops then increases again with career age. A fifth order polynomial on its own could plausibly do this but I associate logistic regression models like this as being monotonically increasing (or decreasing!), again more details as to how this worked would be useful.

Our first submission did not explain this point well, but we hope that Supplementary Figures S4 and S5 provide clarity. In short, we agree of course that typical logistic regression assumes a linear relationship between the predictor variables and the log odds of the outcome variable. This means that the relationship between the predictor variables and the probability of the outcome variable follows a sigmoidal (S-shaped) curve. However, the relationship between the predictor variables and the outcome variable may not be linear.

To capture more complex relationships, like the increasing, decreasing and then increasing attrition rates as a function of career age, higher-order terms can be added to the logistic regression model. These higher-order terms allow the model to capture nonlinear relationships between the predictor variables and the outcome variable — namely the non-monotonic relationship between rates of attrition and career age — while staying within a logistic regression framework.

"The career age of new hires follows the average career age distribution of hires" did you use the empirical distribution here or did you fit a standard statistical distribution e.g. Gamma?

We used the empirical distribution. This information has been added to the updated methods section in the main text.

How did you account for institution (presumably available)? Your own work has shown that institution types plays a role which could be contributing to these results.

See below.

What other confounding variables could be at play here, what is available as part of the data and what happens if you do/don't account for them?

A number of variables included in our data have been shown to correlate with faculty attrition, including PhD prestige, current institution prestige, PhD country, and whether or not an individual is a “self-hire,” i.e., trained and hired at the same institution (Wapman et. al., Nature, 2022). Additional factors that faculty self-report as reasons for leaving academia include issues of work-life balance, workplace climate, and professional reasons, and in some cases to varying degrees between men and women faculty (Spoon et. al., Sci. Adv., 2023).

Our counterfactual analysis aims to address a specific question: how would women’s representation among faculty be different today if men and women were subjected to the same attrition patterns over the past decade? To answer this question, it is important to account for faculty career age, which we accept as a variable that will always correlate strongly with faculty attrition rates, as long as the tenure filter remains in place and faculty continue to naturally progress towards retirement age. On the other hand, it is less clear why PhD country, self-hire status, or any of the other mentioned variables should necessarily correlate with attrition rates and with gendered differences in attrition rates more specifically. While some or all of these variables may underlie the causal roots of gendered attrition rates, our analysis does not seek to answer causal questions about why faculty leave their jobs (e.g., by testing the impact of accounting for these variables in simulations per the reviewers suggestion). This is because we do not believe the data used in this analysis is sufficient to answer such questions, lacking comprehensive data on faculty stress (Spoon et. al., Sci. Adv., 2023), parenthood status, etc.

What career age range did the model use?

The career age range observed in model outcomes are a function of the empirically derived attrition rates for faculty across academic fields. The highest career age observed in the AARC data was 80, and the faculty career ages that result from our model simulations and projections do not exceed 80.

We have also added the distribution of faculty across career ages for the projection scenario model outputs in the supplemental materials Fig. S3 (see response to your later comment regarding career age for further details). Looking at these distributions, it is observed that very few faculty have career age > 60, both in observation and in our simulations.

What was the initial condition for the model?

Empirical 2011 Faculty rosters are used as the initial conditions for the counterfactual analysis, and 2020 faculty rosters are these as the initial conditions for the projections analysis. This information has been added to the descriptions of methods in the main text.

Starting the model in 2011 how well does it fit the available data up to 2020?

Thank you for this suggestion. We ran this analysis for each field starting in 2011, and found that model outcomes were statistically indistinguishable from the observed 2020 faculty gender compositions for all 111 academic fields. This finding is not surprising, because the model is fit to the observed data, but it serves to validate the methods that we used to extract the model's parameters. We have added these results to the supplement (Fig. S2).

What are the sensitivity analysis results for the model? If you have made different fitting decisions how much would the results change? All this applied to both the hiring and attrition parameters estimates.

We model attrition and hiring using logistic regression, with career age included as an exogenous variable up to its fifth power. A natural question follows: what if we used a model with career age only to its first or third power? Or to higher powers? We performed this sensitivity analysis, and added three new figures to the supplement to present these findings:

First, we show the observed attrition probabilities at each career age, and four model fits to attrition data (Supplementary Figs S4 and S5). The first model includes career age only to its first power, and this model clearly does not capture the full career age / attrition correlation structure. The second model includes career age to its third power, which does a better job of fitting to the observed patterns. The third model includes career age up to its fifth power, which appears to very modestly improve upon the former model. The fourth model includes career age up to its seventh power, and the patterns captured by this model are largely the same as the 5th-power model up to career age 50, beyond which there are some notable differences in the inferred attrition probabilities. These differences would have relatively little impact on model outcomes because the vast majority of faculty have a career age below 50.

Second, we show the observed probability that hires are women, conditional on the career age of the hire. Once again, we fit four models to the data, and find that career age should be included at least up to its fifth order in order to capture the correlation structures between career age and the gender of new hires. However, limited differences result from including career age up to the 7th degree in the model (relative to the 5th degree).

As a final sensitivity analysis, we reproduce Fig. 2, but rather than including career age as an exogenous variable up to its fifth power in our models for hiring and attrition, we include career age up to its third power. Findings under this parameterization are qualitatively very similar to those presented in Fig. 2, indicating that the results are robust to modest changes to model parameterization (shown in supplement Fig. S6).

Far more detail in this and some interim results from each stage of the analysis would make the paper far more convincing. It currently has an air of "black box" too much of the analysis which would easily allow an unconvinced reader to discard the results.

We have added more detailed descriptions of the methods to the main text. We hope that the changes made will address these concerns.

Section C: You use the Leslie model to predict the future population. As the model is linear the population will either grow exponentially (most likely) or dwindle to zero. You mention you dealt with this by scaling the average value of H to keep the population at 2020 levels? This would change the ratio of hiring to attrition. How did this affect the timescale of the results. If a field had very minimal attrition (and hence grew massively over the time period of the dataset) the hiring rate would have to be very small too so there would be very little change in the gender balance. Did you consider running the model to steady state instead?

We chose the 40 year window (2020-2060) for this projection analysis because 40 years is roughly the timespan of a full-length faculty career. In other words, it will take around 40 years for most of the pre-existing faculty from 2020 to retire, such that the new, simulated faculty will have almost entirely replaced all former faculty by 2060.

For three out of five of our projection scenarios (OA, GNA, OA+ER), the point at which observed faculty are replaced by simulated faculty represents steady state. One way to check this intuition is to observe the asymptotic behavior of the trajectories in Fig. 3B; the slopes for these 3 scenarios nearly level out within 40 years.

The other two scenarios (OA + IR, GNA+IR) represent situations where women’s representation among new hires is increasing each year. These scenarios will not reach steady state until women represent 100% of faculty. Accordingly, the steady state outcomes for these scenarios would yield uninteresting results; instead, we argue that it is the relative timescales that are interesting.

What did you do to check that your predictions at least felt realistic under the fitted parameters? (see above for presenting the goodness of fit over the 10 years of the data).

We ran the analysis suggested in a prior comment (Starting the model in 2011 how well does it fit the available data up to 2020?) and found that model outcomes were statistically indistinguishable from the observed 2020 faculty gender compositions for all 111 academic fields, plus the “All STEM” and “All non-STEM” aggregations.

You only present the final proportion of women for each scenario. As mentioned earlier, models of this type have a tendency to lead to strange population distributions with wild age predictions and huge (or zero populations). Presenting more results here would assuage any worries the reader had about these problems. What is the predicted age distribution of men and women in the long term scenarios? Would a different method of keeping the total population in check have yielded different results? Interim results, especially from a model as complex as this one, rather than just presenting a final single number answer are a convincing validation that your model is a good one! Again, presenting this result will go a long way to convincing readers that your results are sound and rigorous.

Thank you for this suggestion. We now include a figure that presents faculty age distributions for each projection scenario at 2060 against the observed faculty age distribution in 2020 (pictured below, and as Fig. S3 in the supplementary materials). We find that the projected age distributions are very similar to the observed distributions for natural sciences (shown) and for the additional academic domains. We hope this additional validation will inspire confidence in our model of faculty hiring and attrition for the reviewer, and for future readers.

In Fig S3, line widths for the simulated scenarios span the central 95% of simulations.

Other people have reached almost identical conclusions (albeit it with smaller data sets) that hiring is more important than attrition. It would be good to compare your conclusions with their work in the Discussion.

We have revised the main text to cite the listed examples of similar studies. We thank the reviewer for bringing these relevant works to our attention.

General comments:

What thoughts have you given to non-binary individuals?

Be careful how you use the term "gender diversity"! In many countries "Gender diverse" is a term used in data collection for non-binary individuals, i.e. Male, female, gender diverse. The phrase "hiring more gender diverse faculty" can be read in different ways! If you are only considering men and women then gender balance may be a better framework to use.

We have added language to the main text which explicitly acknowledges that our analysis focuses on men and women due to limitations in our name-based gender tool, which only assigns binary (woman/man) labels to faculty. We point out that this is a compromise due to the technical limitations of name-based gender methodologies and is not intended to reinforce a gender binary.

We have also taken additional care with referring to “gender diversity,” per reviewer 1’s point in their public review.

Reviewer #2 (Recommendations For The Authors):

Data availability: I did not see an indication that the dataset used here is publicly available, either in its raw format or as a summary dataset. Perhaps this is due to the sensitive nature of the data, but regardless of the underlying reason, the authors should include a note on data availability in the paper.

The dataset used for these analyses were obtained under a data use agreement with the Academic Analytics Research Center (AARC). While these data are not publicly available, researchers may apply for data access here: https://aarcresearch.com/access-our-data.

We also added a table to the supplemental materials (Tab. S3) that reports the estimated number of men and women in each of the 111 fields.

Additionally, a variety of summary statistics based on this dataset are available online, here: https://github.com/LarremoreLab/us-faculty-hiring-networks/tree/main

Gender classification: Was an existing package used to classify gender from names in the dataset, or did the authors develop custom code to do so? Either way, this code should be cited. I would also be curious to know what the error rate of these classifications are, and suggest that additional information on potential biases that might result from automated classifications be included in the discussion, under the section describing data limitations. The reliability of name-based gender classification is particularly of interest, as external gender classifications such as those applied on the basis of an individual's name - may not reflect the gender with which an individual self-identifies. In other words, while for many people their names may reflect their true genders, for others those names may only reflect their gender assigned at birth and not their self-perceived or lived gender identity. Nonbinary faculty are in particular invisibilized here (and through any analysis that assigns binary gender on the basis of name). While these considerations do not detract from the main focus of the study - which was to utilize an existing dataset classified only on the basis of binary gender to assess trends for women faculty-these limitations should be addressed as they provide additional context for the interpretation of the results and suggest avenues for future research.

We use a free, open-source, and open-data python package called nomquamgender (Van Buskirk et al, 2023) to estimate the strengths of (culturally constructed) name-gender associations. For sufficiently strong associations with a binary gender, we apply those labels to the names in our data. We have updated the main text to make this approach more apparent.

We have also added language to the main text which explicitly acknowledges that our approach only assigns binary (woman/man) labels to faculty. We point out that this is a compromise due to the technical limitations of name-based gender methodologies and is not intended to reinforce a gender binary.

As we mentioned in response to the public review, we use a free and open source python package called nomquamgender to estimate the strengths of name-gender associations, and we apply gender labels to the names with sufficiently strong associations with a binary gender. This package is based on a paper by Van Buskirk et. al. 2023, “An open-source cultural consensus approach to name-based gender classification,” which documents error rates and potential biases.

We have also added language to the main text which explicitly acknowledges that our approach only assigns binary (woman/man) labels to faculty. We point out that this is a compromise due to the technical limitations of name-based gender methodologies and is not intended to reinforce a gender binary.

Page 1: The sentence beginning "A trend towards greater women's representation could be caused..." is missing a conjunction. It should likely read: "A trend towards greater women's representation could be caused entirely by attrition, e.g., if relatively more men than women leave a field, OR entirely by hiring..."

We have edited the paragraph to remove the sentence in question.

Pages 1-2: The sentence beginning "Although both types of strategy..." and ending with "may ultimately achieve gender parity" is a bit of a run-on; perhaps it would be best to split this into multiple sentences for ease of reading.

We have revised this run-on sentence.

Page 2: See comments in the public review about a methods section, the addition of which may help to improve clarity for the readers. Within the existing descriptions of what I consider to be methods (i.e., the first three paragraphs currently under "results"), some minor corrections could be added here. First, consider citing the source of the dataset in the line where it is first described (in the sentence "For these analyses, we exploit a census-level dataset of employment and education records for tenured and tenure-track faculty in 12,112 PhD-granting departments in the United States from 2011-2020.") It also may be helpful to include context here (or above, in the discussion about institutional analyses) about how "departments" can be interpreted. For example, how many institutions are represented across these departments? More information on how the authors eliminated the gendered aspect of patterns in their counterfactual model would be helpful as well; this is currently hinted at on page 4, but could instead be included in the methods section with a call-out to the relevant supplemental information section (S2B).

We have added a citation to Academic Analytics Research Center’s (AARC) list of available data elements to the data’s introduction sentence. We hope this will allow readers to familiarize themselves with the data used in our analysis.

Faculty department membership was determined by AARC based on online faculty rosters. 392 institutions are represented across the 12,112 departments present in our dataset. We have updated the main text to include this information.

Finally, we have added a methods section to the main text, which includes information on how the gendered aspect of attrition patterns were eliminated in the counterfactual model.

Page 2: Perhaps some indication of how many transitions from an out-of-sample institution might be helpful to readers hoping to understand "edge cases."

In our analysis, we consider all transitions from out-of-sample institutions to in-sample institutions as hires, and all transitions away from in-sample institutions–whether it be to an out of sample institution, or out of academia entirely–as attritions. We choose to restrict our analysis of hiring and attrition to PhD granting institutions in the U.S. in this way because our data do not support an analysis of other, out-of-sample institutions.

I also would have liked additional information on how many faculty switched institutions but remained "in-sample and in the same field" - and the gender breakdowns of these institutional changes, as this might be an interesting future direction for studies of gender parity. (For example, readers may be spurred to ask: if the majority of those who move institutions are women, what are the implications for tenure and promotion for these individuals?)

While these mid-career moves are not counted as attritions in the present analysis, a study of faculty who switch institutions but remain (in-sample) as faculty could shed light on issues of gendered faculty retention at the level of institutions. We share the reviewer’s interest in a more in depth study of mid-career moves and how these moves impact faculty careers, and we now discuss the potential value of such a study towards the end of the paper. In fact, this subject is the topic of a current investigation by the authors!

Page 3: I was confused by the statement that "of the three types of stable points, only the first point represents an equitable steady-state, in which men and women faculty have equal average career lengths and are hired in unchanging proportions." Here, for example, computer science appears to be close to the origin on Figure 1, suggesting that hiring has occurred in "unchanging proportions" over the study interval. However, upon analysis of Table S2, it appears that changes in hiring in Computer Science (+2.26 pp) are relatively large over the study interval compared to other fields. Perhaps I am reading too literally into the phrase that "men and women faculty are hired in unchanging proportions" - but I (and likely others) would benefit from additional clarity here.

We had created an arrow along with the computer science label in Fig. 1, but it was difficult to see, which is likely the source of this confusion. This was our fault, and we have moved the “Comp. Sci.” label and its corresponding arrow to be more visible in Figure 1.

Changes in women’s representation in Computer Science due to hiring over 2011 - 2020 was +2.26 pp as the reviewer points out, but, consulting Fig. 1 and the corresponding table in the supplement, we observe that this is a relatively small amount of change compared to most fields.

Page 3: If possible it may be helpful to cite a study (or multiple) that shows that "changes in women's representation across academic fields have been mostly positive." What does "positive" mean here, particularly when the changes the authors observe are modest? Perhaps by "positive" you mean "perceived as positive"?

We used the term positive in the mathematical sense, to mean greater than zero. We have reworded the sentence to read “women's representation across academic fields has been mostly increasing…” We hope this change clarifies our meaning to future readers.

Page 3: The sentence that ends with "even though men are more likely to be at or near retirement age than women faculty due to historical demographic trends" may benefit from a citation (of either Figure S3 or another source).

We now cite the corresponding figure in this sentence.

Page 4: The two sentences that begin with "The empirical probability that a person leaves their academic career" would benefit from an added citation.

We have added a citation to the sentences.

Figure 3: Which 10 academic domains are represented in Panel 3B? The colors in appear to correspond to the legend in Panel 3A, but no indication of which fields are represented is provided. If possible, please do so - it would be interesting and informative to be able to make these comparisons.

This was not clear in the initial version of Fig. 3B, so we now label each domain. For reference, the domains represented in 3B are (from top to bottom):

● Health

● Education

● Journalism, Media, Communication

● Humanities

● Social Sciences

● Public Administration and Policy

● Medicine

● Business

● Natural Sciences

● Mathematics and Computing

● Engineering

Page 6: Consider citing relevant figure(s) earlier up in paragraph 2 of the discussion. For example, the first sentence could refer to Figure 1 (rather than waiting until the bottom of the paragraph to cite it).

Thank you for this suggestion, we now cite Fig. 1 earlier in this discussion paragraph.

Page 10: A minor comment on the fraction of women faculty in any given year-the authors assume that the proportion of women in a field can be calculated from knowing the number of women in a field and the number of men. This is, again, true if assuming binary genders but not true if additional gender diversity is included. It is likely that the number of nonbinary faculty is quite low, and as such would not cause a large change in the overall proportions calculated here, but additional context within the first paragraph of S1 might be helpful for readers.

We have added additional context in the first paragraph of S1, explaining that an additional term could be added to the equation to account for nonbinary faculty representation if our data included nonbinary gender annotations. Thank you for making this point.

Page 10: Please include a range of values for the residual terms of the decomposition of hiring and attrition in the sentence that reads "In Figure S1 we show that the residual terms are small, and thus the decomposition is a good approximation of the total change in women's representation."

These residual terms range from -0.51pp to 1.14pp (median = 0.2pp). We have added this information to the sentence in question.

Page 12: It may be helpful to readers to include a description of the information contained in Table S2 in the supplemental text under section S3.

We refer to table S2 twice in the main text (once in the observational findings, and once for the counterfactual analysis), and the contents of table S2 are described thoroughly in the table caption.

Reviewer #3 (Recommendations For The Authors):

(1) There is a potential limitation in the generalizability of the findings, as the study focuses exclusively on US academia. Including international perspectives could have provided a more global understanding of the issues at hand.

The U.S. focus of this study limits the generalizability of our findings, as non-U.S. other faculty may exhibit differences in hiring patterns, retention patterns, and current demographic representations. We have added a discussion of this limitation to the manuscript. Unfortunately, our data do not support international analyses of hiring and attrition.

(2) I am not sure that everyone who disappeared from the AARC dataset could be count as "attrition" from academia. Indeed, some who disappeared might have completely left academia once they disappeared from the AARC dataset. Yet, there's also the possibility that some professors left for academic positions in countries outside of the US, or US institutions that are not included in the AARC dataset. These individuals didn't leave academia. Furthermore, it is also possible that these scholars who moved to an institution outside of US or not indexed by AARC are gender specific. Therefore, analyses that this study conducts should find a way to test whether the assumption that anyone who disappeared from AARC is indeed valid. If not, how will this potentially challenge the current conclusions?

The reviewer makes an important point: faculty who move to faculty positions in other countries and faculty who move to non-PhD granting institutions, or to institutions that are otherwise not included in the AARC data are all counted as attritions in our analysis. We intentionally define hiring and attrition broadly to include all cases in which faculty join or leave a field or domain within our dataset.

The types of transitions that faculty make out of the tenure track system at PhD granting institutions in the U.S. may correlate with faculty attributes, like gender. For example, women or men may be more likely to transition to tenure track positions at non-U.S. institutions. Nevertheless, these types of career transition represent an attrition for the system of study, and a hire for another system. Following this same logic, faculty who transition from one field to another field in our analysis are treated as an attrition from the first field and a hire into the new field.

By focusing on “all-cause” attrition in this way, we are able to make robust insights for the specific systems we consider (e.g.,, STEM and non-STEM faculty at U.S. PhD granting institutions), without being roadblocked by the task of annotating faculty departures and arbitrating which should constitute “valid” attritions.

(3) It would be very interesting to know how much of the attribution was due to tenure failure. Previous studies have suggested that women are less likely to be granted tenure, which makes me wonder about the role that tenure plays in the gendered patterns of attrition in academia.

We note that faculty attrition rates start low and then reach a peak around 5-7 years after earning PhD, and then decline until around 15-20 years post-PhD, after which, attrition rates increase as faculty approach retirement. The first local maximum appears to coincide roughly with the tenure clock timing, but we can only speculate that these attritions are tenure related. Our dataset is unfortunately not equipped to determine the causal mechanisms driving attrition.

We reproduce the attrition risk curve in the supplementary materials, Fig. S4:

(4) The dataset used doesn't fully capture the complexities of academic environments, particularly smaller or less research-intensive institutions (regional universities, historically black colleges and universities, and minority-serving institutions). This could be potentially added to the manuscript for discussions.

We have added this point to the description of this study’s limitations in the discussion.

eLife assessment

Efforts to increase the representation of women in academia have focussed on efforts to recruit more women and to reduce the attrition of women. This study - which is based on analyses of data on more than 250,000 tenured and tenure-track faculty from the period 2011-2020, and the predictions of counterfactual models - shows that hiring more women has a bigger impact than reducing attrition. The study is an important contribution to work on gender representation in academia, and the evidence in support of the findings is convincing.

Reviewer #1 (Public Review):

Summary<br /> This is an interesting paper that concludes that hiring more women will do more to improve the gender balance of (US) academia than improving the attrition rates of women (which are usually higher than men's). Other groups have reported similar findings, i.e. that improving hiring rates does more for women's representation than reducing attrition, but this study uses a larger than usual dataset that spans many fields and institutions so it is a good contribution to the field.

The paper is much improved and far more convincing as a result of the revisions made by the authors.

Strengths<br /> A large data set with many individuals, many institutions and fields of research.<br /> A good sensitivity analysis to test for potential model weaknesses.

Weaknesses<br /> Only a single country with a very specific culture and academic system.<br /> Complex model fitting with many steps and possible places for model bias.