Did you consider testing purine:pyrimidine ratios as a genome feature (e.g. G:T or A:C)? I didn’t find any evidence supporting the idea that a correlation between a growth condition and this ratio would be present but I’m curious if you ruled that out for a reason or feel like it is captured by G+C content or R>Y transitions.

This is so awesome! This figure makes it easier to identify “novelty” since the species are ordered phylogenetically. For example Aquificota has an ideal growth temperature of 60℃+ while Camplyobacterota has an ideal growth temperature of 15℃ and below. Or, Thermoproteota seems to have a wide range of diversity regarding ideal growth temperatures. You briefly mentioned this in the discussion but, assuming these predictions could be validated in the lab, researchers could then try to identify the gene families that expand/contract and are linked to a trait of interest with a tool like NovelTree. Very cool! https://research.arcadiascience.com/pub/resource-noveltree

Typo? “...'membrane' if the protein’s GRAVY…”

Missing period? “...incorrectly truncated. To determine…”

Although your work focuses on predicting the ideal growth conditions of microbes based on a phylogenetically balanced dataset, it makes me wonder if it is possible to create a model that can point a researcher to a set of microbes that are most likely to thrive in a certain set of conditions. Rather than perform directed evolution, can we identify the microbe that thrives in a certain set of conditions and introduce the genetic information that is required for a function of interest? Or perhaps a different model is unnecessary and one could even parse through the predicted conditions of the 85k+ microbes data you generated? Just a thought!

This is awesome!!! The impact/relevance of your work is incredibly clear, all data/code is on GitHub (with a very robust README) and you candidly express the limitations of the predictive models (e.g. inaccuracy when predicting oxygen tolerance for certain genera or phyla, thus requiring follow-up on the relationship between AAs and metabolic niches or how the lack of precision of the models may not be helpful for cultured microorganisms). I’m looking forward to trying this out myself!

Summary: In this manuscript, Barnum et al. created computational models — favoring simple logistic regression models — that can predict the ideal oxygen tolerance, temperature, salinity and pH conditions of novel taxonomic microbial families (requiring only an unannotated, and potentially incomplete, genome from the user).

The authors leveraged the empirical data of 15.5k+ microbes and curated the dataset to omit microbes that did not have multiple measured values for a growth phenotype, had minor differences between the minimum and maximum value tested for the phenotype (<1.5 pH, 10C, 1.5% NaCl unless salinity was <0.5%), or had fewer than 4 total measurements recorded. Haloarchaea with a salinity optima <3.7% were also excluded and finally, the data set was further balanced to reduce taxonomic bias.

The authors then measured correlations between DNA and protein sequence features and oxygen tolerance, temperature, salinity and pH conditions (expressed as a Spearman’s rank correlation coefficient). No correlations between the tested DNA sequence features and the four physiochemical conditions were identified but numerous correlations between protein sequence features and the physiochemical conditions were identified. For example, a negative correlation between oxygen tolerance and cysteine frequency was revealed (p=-0.49).

Estimators were then evaluated on their ability to accurately predict the four physiochemical conditions based on 9 different sets of features and the authors found that amino acid features alone were sufficient for accurate prediction. Three models were then selected for each condition (optimum, minimum and maximum value predictions). When testing the selected models with family-level holdouts, the predictions were made with lower accuracy albeit their performance was consistent with training and cross-validation data. The models also predicted extreme growth conditions less accurately (e.g. salinity > 15% or pH > 5).

To test the models’ vulnerability to phylogenetic bias, the selected models were compared to models where the prediction was a random value or the average value of the closest relatives. As expected, the chosen models considerably outperformed the models strongly influenced by phylogeny.

To test the models’ vulnerability to genome completeness, protein and genome sequences were subsampled to 10-100% completeness for 20 different species in each condition range and evaluated for prediction accuracy. The selected models showed negligible differences between 10% and 100% genome completeness for oxygen tolerance, temperature and salinity. pH prediction experienced a bigger impact by genome completeness.

The selected models were then used to predict the ideal growth conditions of 85k+ bacteria and archaea. As expected, many of the uncultivated species were predicted to grow in more extreme conditions. The ideal growth conditions of 3.3k+ metagenomes were predicted and compared to the growth conditions of the environment from which the samples were derived. Predicted growth conditions mostly aligned with the organism’s habitat but the authors found that predicted individual genomes can deviate from the conditions of the source environment.

Typo? Do you mean "Mice were anesthetized with..." ?

Typo? Do you mean "Osteopontin activates trophoblast invasion in vitro." ?

Typo? Do you mean "...endothelial cells, induced endothelial..." ?

Typo? Do you mean "...change in the sFlt-1/PIGF-2 ratio" ?

Considering the importance of using the sFlt-1/PIGF ratio as a marker for preeclampsia (the only FDA-approved immunoassay to assess preeclampsia risk is based on the sFlt-1/PIGF ratio), do you believe the BAC/APOL-G1 model is a better model for preeclampsia?

Given the results of your work, is there a reason where one would prefer to use the Alb/APOL-G1 model?

Typo? "mice" ?

Typo? Do you mean "...cell types through Cd44..." ?

This sentence is a bit confusing. I’d re-word this to “Dams carrying offspring with BAC/APOL1-G1 and Alb/APOL1-G1 had higher systolic blood pressure (158.4 mm Hg [141.3-163.4], P=0.005 and 159.0 [137.3-164.8], P=0.002) compared to dams carrying BAC/APOL1-G0 and Alb/APOL1-G0 mice (115.9 [114.5-130.8] and 109.5 [96.3-133.3]) (Figure 1B).

You mention this later in the discussion but I would include a sentence conveying why you didn’t test APOL1-G2 variants since your introduction refers to both APOL1 high-risk variants (G1 and G2) and to evidence of preeclampsia development in APOL1-G2 transgenic mice (citation 16). E.g. “We did not study APOL1-G2 variants because we did not observe physical preeclampsia phenotypes in these transgenic lines.”

Suggestion: "...has been reported."

In an attempt to better understand the cell types and molecular mechanisms implicated in preeclampsia, Yoshida et al. analyzed pregnant female mice carrying BAC/APOL-G1 and Alb/APOL-G1 fetuses – novel IVF-derived preeclampsia mouse models – for physical biomarkers of preeclampsia and performed single-nucleus RNA-seq to identify differentially expressed genes and impacted cell-cell interactions.

When testing physical biomarkers of preeclampsia, the authors found that female mice carrying BAC/APOL-G1 and Alb/APOL-G1 fetuses had higher systolic blood pressure and smaller body weight, while just the female mice carrying BAC/APOL-G1 fetuses had a higher sFlt-1/PIGF-2 ratio. Elevated blood pressure and sFlt-1 levels and lower PIGF levels are phenotypes/markers clinically associated with preeclampsia.

Single-nucleus RNA-sequencing of placenta was conducted and differentially expressed genes were identified between APOL1-G1 and APOL1-G0 or APOL1-G1 and wild type mice. Inflammatory pathways and autoimmune disease pathways that were identified when comparing human preeclampsia RNA-seq data and a normal control were also implicated in the BAC/APOL1-G1 vs. BAC/APOL1-G0 DEG analysis results (e.g. pathogen induced cytokine storm signaling pathway and HIFI⍺ pathway).

The authors then performed cell-cell interaction analyses (Seurat) and found 31 shared activated and 4 shared deactivated pathways in BAC/APOL1-G1 placentas compared to BAC/APOL-G0 and wild type placentas. One of the identified shared activated pathways was the osteopontin/Spp1 signaling pathway which was found to be most upregulated in vascular endothelia – the cell type with the highest APOL1 expression.

Lastly, the authors attempted to determine the impact of maternal monocytes and decidual cells by APOL-G1 placentas and found upregulated signals to monocytes associated with the Cd44 receptor and upregulated expression of Ccl2 – supporting their hypothesis that APOL1-G1 induced preeclampsia in female mice activates maternal monocytes.

Suggestion: “...any physiological measure is a better indicator…” or “..suggest that physiological measures are better indicators…”

Figure 1c: It would be interesting to see last birth vs. oopause in four ways: 1. last wild birth vs. captive oopause (as you show here), 2. last wild birth vs. wild oopause, 3. last captive birth vs. wild oopause, and 4. last captive birth vs. captive oopause. This could really emphasize the importance of data curation and how it could lead to misleading conclusions (Winkler & Goncalves).

This was incredibly engaging to read!!!!! I really appreciate you writing this rebuttal and keeping the translational/clinical audience in mind. And a HUGE thank you for making this available to anyone via bioRxiv so that the audience can think critically about the differing opinions and methodologies.

Suggestion: “...take issue with the wider use of the term…”

I assume this refers to PrR where the numerator is determined based on the average time that an individual in the population no longer creates offspring, is that correct? If so, there would be the possibility that individuals may still have the ability to reproduce but aren’t creating offspring which would make physiological measures helpful to determine reproductive cessation despite social factors for all species, not just for the Asian elephant no?

I deeply appreciate this caveat and this is a powerful reminder to the non-evolutionary biology audience. I can see how it may be advantageous to study reproductive capabilities and cessation in unnatural contexts to better understand the mechanisms of reproductive aging but I agree with you in the sense that it wouldn’t be accurate to discuss the evolution of a trait in conditions that differ from how the trait evolved/emerged (captivity in this case).

This question is a bit philosophical but I’m curious if you think the conditions of humans most closely resemble captive conditions? In certain societies, modern medicine and infrastructure removes the risk of death from predation, starvation, disease, etc. (longer lifespan) and the U.S. is experiencing a rise in infertility (arguably shorter reproductive lifespan). Would it be accurate to do comparative studies surrounding reproductive health at all given that human reproductive measurements may be more similar to those from captive populations?

Suggestion: “..predominantly captive populations from Winkler & Goncalves…”

Summary: “Menopause has not evolved as a general trait in mammals: A response to ‘Do mammals have menopause?’” Chapman et al. offer a rebuttal to Winkler et al.’s assertion that menopause (or “oopause”) is widespread across mammals. (See below for a summary of “Do mammals have menopause?” by Winkler and Goncalves.) The authors address three main issues with the assertion and methodology: * dismissal of wild data and the use of captive populations — captive populations have artificially long lifespans and artificially early reproductive termination, thus, excluding wild data can lead to misleading conclusions * the use of maximal lifespan — the PrR measure is a better method to compare populations with different lifespans, as maximal lifespan can over- or under-estimate population lifespan * misinterpretations of the data sources — Chapman et. al found numerous instances where Winkler & Goncalves contradict the analysis in the sources they cite or used an oopause age that was younger than that of the cited source

Chapman et al. also note that the term “oopause” is unnecessary, as the definition of menopause, from a cross-species comparative perspective, is “the irreversible loss of the physiological capacity to produce offspring due to intrinsic biological factors” which essentially encompasses the idea of “oopause.”

Overall, the authors stress that to accurately conduct cross-species comparisons centered around menopause, we must focus on the species that experience prolonged life after reproduction in the wild using validated demographic measures such as post-reproductive representation (PrR).

Summary: “Do mammals have menopause?” by Winkler & Goncalves In an attempt to devise a definition of reproductive senescence that is useful for comparative studies, Winkler and Goncalves argue in “Do mammals have menopause?” that “oopause, the permanent age-associated cessation of ovulation across all mammalian species,” is common among mammals. The authors assert that because a majority of mammalian species in captivity in zoos live significantly longer than their wild counterparts, demographic measurements in free-ranging/wild individuals are not reliable indicators of oopause. Thus, their analysis excluded demographic studies that only observed wild populations (primarily order Cetacea from “Analyses of ovarian activity reveal repeated evolution of post-reproductive lifespans in toothed whales'' by Ellis et al.) and concluded that most mammals experience oopause. Despite this conclusion, the authors do acknowledge that physiological data (e.g. follicles or corpora count, frequency/length of menstrual/estrous cycles, changing hormone concentration, microscopic analysis of cell types in vaginal smears, etc.) is lacking and needed for most mammalian orders to better understand oopause and potentially delay reproductive aging.

Typo? “...activation to inhibition…”

Even though the Aβ plaque number and area weren’t considerably different according to Thioflavin S imaging, do you think you could have seen a difference in the soluble Aβ40 or Aβ42 measurements if you used the hippocampal tissue?

Do you intend on making the raw data publicly available in the future?

Typo? “...between these comparisons…”

Typo? “...i-pathway software…”

Typo? “...40X magnification. ImageJ was…”

This inverse relationship is incredibly fascinating! Are you aware of the specifics of the correlation between cognitive function in ovariectomized AD mice that don’t undergo pharmacological inhibition of systemic gonadotropin levels? (I.e. Did brain LH levels increase as cognitive function symptoms worsen or did cognitive function issues occur once a brain LH threshold was reached?) I could imagine that continuous monitoring of systemic LH levels could potentially serve as a proxy for cognitive function issues in menopausal women if this inverse relationship is taken into account?

The authors aimed to determine whether the central activation of the luteinizing hormone via hCG in ovariectomized AD (APP/PSI) mice impacts cognitive function (Morris water maze test), Aβ pathology (Aβ plaque number and area from Thioflavin S imaging and soluble Aβ levels from cortical tissue), hippocampal dendritic spine density (Golgi staining) and/or signaling changes (RNA-seq to see differential gene expression).

Although the authors did present compelling evidence supporting the potential that LHCGR activation improves spatial memory and increases dendritic spine density in OVX PP/PSI mice, the evidence is based on an underlying assumption that the results are due to the activation of LHCGR rather than the potential impact of hCG being present centrally. It would have been helpful to compare the results of centrally delivering hCG and LH in order to identify similarities which may be due to LHCGR activation versus impacts on cognitive function or spine density that may be due to the central presence of LH or hCG and an unknown alternative interaction.

It would be helpful to include the total number of mice/samples tested.

These examples of how you labeled proteins (circadian, sleep deprivation, sleep or combined) are very helpful! Although the categorizations make it easier to see global expression level changes between juveniles, adolescents and adults (Figure H-J), I imagine it is possible that the expression levels may differ for reasons other than the changed variable (sleep deprivation). How did you take that into account while assigning regulation groups?

Will you make the raw data available in the future?

It would be nice to understand how you identified the repeats. Valine also appears to be conserved in the second position of the MR and I could imagine considering a "KV" repeat rather than just the conserved lysine (K). Was there a reason you didn’t highlight it?

I love this illustration! It does a great job at summarizing the role of each part of the CsoS2 protein. What is the difference between the teal and purple shell components? I assumed the teal component refers to the CsoS1A shell protein. Does the purple component refer to other CsoS2 proteins, the truncated CsoS2A protein or something else?

Missing parenthesis: “ACKR1+/FABP4+ capillaries (FDR<10-7 in both).”

This is an interesting finding! I imagine that it could be helpful to use this as a potential in screen for new psoriasis treatments if human diseased skin is accessible (with the caveat that this finding was based on samples from males of European descent).

It would be nice to see whether the reduction of WNT5A+/IL24+ fibroblasts also occurred for class V-VII corticosteroids given that the the higher potency corticosteroids are advised for short-term use. I wonder if a reduction in the WNT5A+/IL24+ fibroblast cells would occur as quickly or with the same magnitude when using less potent/strong corticosteroid medications. I’d also be interested to see if Vtama (tapinarof) leads to a WNT5A+/IL24+ fibroblast reduction as well.

Typo: "...were CN, 3293 (27.6%) had MCI.." :)

Typo: "...includes participant..."

Did you consider breaking apart the MBI analysis by category (decreased motivation, emotional dysregulation, impulse dyscontrol, social inappropriateness, abnormal perception/thought) versus using a global MBI score to better see where the relationship between MBI, SD and dementia is strongest (if any)? This could help extrapolate the potential common cause/pathology.

I agree with this limitation. The NPI-Q nighttime behavior question excludes the ability to identify other types of sleep-wake disorders and even though the score is based on the answer from an informant, there is often a gap between the subjective perception of sleep and objective data via PSG. The question also doesn’t capture sleep behavior or CBT-I attempts. Although CBT-I is moderately effective for insomnia at best, cases where this has been attempted and proved unsuccessful could help point to neurobiological drivers of the SD-MBI-dementia relationship.

I would re-phrase this sentence. :)

post-stress?

Typo: "...overlapping outcome, has also been..."

Typo: "...white-dark bars under all figures indicate..."

It would be interesting to think about how we can point to biological/neurological associations with sleep changes without referring to previous work that discovered those associations. I.e Could a tool like NovelTree point to the four genes involved in synaptic plasticity that you’ve tested as potential drivers of the difference in sleep rebound phenotypes or perhaps reveal other unknown gene associations? NovelTree Pipeline: doi.org/10.57844/arcadia-z08x-v798

Even though rebound sleep wasn’t seen during the ZT 0-3 state for species other than D. melanogaster, there does seem to be an increase in sleep during the “natural sleep period” (ZT 12-24 state) for D. erecta, D. yakuba and D. willinstoni. Do you think it is possible that a rebound response could occur much later in the next sleep cycle rather than immediately after sleep disruption ends?

The sexual dimorphism here is fascinating, especially because this is also seen in human sleep-wake disorders! For example, insomnia is more prevalent among women than among men and the discrepancy becomes larger with age. Did you consider comparisons between sex for any of the species other than D. virilis? Perhaps looking at the same species could serve as a useful control for ecological and dietary impacts?

Perhaps this is difficult due to an experimental limitation but it could have been nice to include another Drosophila species with a vegetable ecological niche for your analysis. D. mojavensis could be especially useful since it has a similar “siesta” sleep phenotype as D. virilis and I would be curious to see if it also shows a lack of sexual dimorphism and no sleep rebound differences as well.

Recent efforts have attempted to understand the physical properties and functions of the peripheral mechanoreceptors in the genitalia (Krause corpuscles). This paper may be interesting to you, as it dives deeper into the molecular basis of what you're testing clinically: https://www.biorxiv.org/content/10.1101/2023.06.14.545006v1

What is your definition of non-neurogenic FSD? When sexual dysfunction isn’t due to a lack of desire or fantasies as in HSDD? Even though the lubrication sub-score was used as an exclusion criteria, I’m curious if you believe it’s possible that you still included women with neurogenic FSD in this study? Would it have been useful to use DSM diagnostic criteria for HSDD to eliminate neurogenic FSD participants or is my understanding of neurogenic FSD different from yours in this case?

Do you think the decision to decline to participate for 3 NDAB and 5 SCI participants also could have been due to hesitancy with DGNS?

Do you think it could be helpful to do temporal transcriptomic analysis to reveal the dynamic nature of keratinocytes given that there was a two-month delay to reverse the scratching phenotype after TRPV1 ablation? Perhaps that wouldn't be useful since itch, in this case, is hypothesized to be sustained in the CNS rather than in the periphery?

Did you consider testing kappa opioid receptor agonists such as nalfurafine and difelikefalin since KORs may be associated with itch attenuation while MORs may be associated with itch intensification? This would be especially interesting from a translation perspective since difelikefalin is usually used for chronic kidney disease-associated pruritus but is in the pipeline for AD.

Even though you referenced a few papers that discussed Krause corpuscles in humans, this interpretation is difficult to assert based on the data shown. A direct comparison between the cited work and your work may be helpful in order to include this.

Did you consider incorporating additional vertebrate RNA-seq data into your atlas in order to minimize bias or was this possibility not available? If not, did you consider a re-harmonized atlas after generating the axolotl DRG RNA-seq data?

Summary: The authors constructed cross-species harmonized cell atlases of DRG and TG neuronal and non-neuronal cell types using sc/snRNA-seq data from 19 mammalian studies. (Available at https://harmonized.painseq.com) Cell types were characterized based on mouse DRG marker genes. The final reference only included cells or nuclei with consistent cell type annotations from both the LIGER and Seurat computational pipelines. Generated clusters included 18 neuronal DRG subtypes, 14 neuronal TG subtypes, 7 non-neuronal DRG subtypes and 11 non-neuronal TG subtypes. The authors then sequenced nuclei from human DRG samples from Harvard Medical School, the University of Texas-Dallas and Washington University in St. Louis and compared the annotations generated by the harmonized atlas and the 10X Genomics Cellranger v7 pipelines. Finally, scRNA-seq was performed on axolotl DRGs in order to evaluate conservation of subtypes in vertebrates.

Small Typo: “identify anchors between Nefh…”

I would be more interested in knowing what range of similarity is expected for evolutionarily distant species and what outliers exist. E.g. Are there similarities higher than what would be expected due to evolutionary distance?

It would be interesting to see which neuronal cell types were identified when the three datasets were individually annotated versus when they were anchored to your atlas. I would be interested to see if anchoring to the atlas resolved the rare neuronal subtypes like Mrgpra3+Mrgprb4, Mrgpra3+Trpv1, Calca+Oprk1, and Calca+Dcn.

Thanks for sharing these results! It's awesome to see how clinical reporting of differences in pain between males and females can help drive an understanding of the molecular mechanisms associated with those differences. The implication of this finding is especially interesting when we consider how prolactin levels in females increase during pregnancy and breastfeeding!

Summary: Evidence of differences in RNA transcripts between male and female human DRG neurons and of an increase in receptivity of trigeminal nociceptors in the presence of prolactin led this team to provide functional evidence to support or refute the hypothesis that prolactin may differentially sensitize DRG neurons. Using PrlrCre/+;Ai6 transgenic mice and patch clamp recording of cultured mouse DRG cells, an increase in firing frequency was found in female DRG neurons when exogenous mouse prolactin was introduced. Male DRG neurons did not display the same relative increase in firing. Further, immunostaining revealed higher expression levels of prolactin receptors in female human DRG tissue samples in comparison to male human DRG tissue samples. Cultured and PRLR-stained human neurons displayed differential expression of prolactin receptors between male and female samples. Finally, patch clamp recording was used on human DRG cultures in order to compare sex-based action potential differences in the presence of prolactin. As in mice, a notable increase in firing frequency was found in female human DRG neurons in the presence of prolactin but not for male human DRG neurons.

small typo: I believe you meant image (B)?

small typo: "...Fig. S1A) but..."

Have you considered exploring the potential relationship between prolactin and the top 25 pain-associated genes in the female cohort from the “RNA profiling of human dorsal root ganglia reveals sex differences in mechanisms promoting neuropathic pain” study? Even though prolactin receptor transcripts weren’t reported, is there a possible relationship between the genes that were and the impact of prolactin?

Do you have additional experiments in mind that could further explore the role of prolactin in human nociceptors despite the limited availability of human cells? It would be interesting to compare the spatial organization of prolactin receptor expressing neurons in mouse and human DRG tissues considering the probability that the organization of the DRG is likely different?

Your data seems to suggest the opposite, the p-value for the female human vehicle versus prolactin DRG rheobase is 0.0105 while the male and female rodent p-values are both well-above 0.05.

There does appear to be a considerable difference in aggregate appearance in 2Q left versus 2Q middle and 2Q right (with and without 10% PEG) no?

Did you mean "through" here? :)

I would be curious to see if there were any spatial distribution differences between the truncations before photobleaching. Did you notice any differences that “agree” or “disagree” with how the 18-hour incubated IDR + CC fragment showed exceptional recovery? How did spatial distribution of the newly formed full McdB condensates compare to the 18-hour incubated IDR + CC fragment?

I really enjoyed reading about your approach to thoroughly understand the McdB-driven phase separation and oligomerization of S. elongatus carboxysomes! This was obviously a really challenging protein to work with but you did an incredible job! I noticed in Figure 3A that there is a difference in how spatially distributed the McdB proteins are before photobleaching. There seems to be a higher concentration of McdB proteins on the "outskirts" of the newly formed condensate aggregates while the "mature" aggregates have a more even distribution. Any idea as to why this is the case and/or if this could impact recovery differences after photobleaching?

It would be interesting to see if there are differences in response rates and results if you separated the collected specimens by sex. Although I'm not sure if this applies in the context of M. semilimbatus, in some cases there have been instances of sex-related differences in attention and memory.

Considering that spiders are also extremely sensitive to vibration, have you considered whether different environmental conditions could be the cause of the spider's reaction rather than due to the sensory input? Perhaps the environment was controlled for this but not stated in the paper?