Leituras que destaco sobre ambientes híbridos de aprendizagem: "Mais do que a integração de ambientes físicos e virtuais de aprendizagem, a educação híbrida deve afirmar-se como um conceito de educação total caracterizado pelo uso de soluções combinadas, envolvendo a interação entre diferentes modalidades, abordagens pedagógicas e recursos tecnológicos. [...] Com efeito, em tempos de profundas transformações, de um mundo estruturado de uma forma complexa onde coabitam o analógico e o digital, o real e o virtual, o humano e a máquina, o offline e o online, do reconhecimento de que vivemos numa nova ordem social, cultural, económica, política e até ética e da vertiginosa evolução das tecnologias digitais, deparamo-nos com a necessidade de repensar o paradigma educacional onde a comunicação possa assumir um papel fulcral unindo e aproximando atores humanos e não-humanos. Esta centralidade do processo comunicacional, e não do professor, do aluno ou da tecnologia, remetem para variáveis comunicacionais significativas como a interação, ligação, conexão e participação, essenciais à relação pedagógica. [...] Implementar modelos de educação híbrida, não como um processo de disrupção pura, mas como um processo de inovação sustentada, permitirá avançar para a ideia de uma comunidade educativa unida nos seus propósitos de mudança. [...] Nesta perspetiva, e apesar das múltiplas definições que existem na literatura, o termo Blended Learning é entendido como integrante desta realidade híbrida e como a combinação de diferentes ambientes de aprendizagem, quer na geografia física, quer virtual."

Moreira, J. A. & Horta, M. J. (2020). Educação e Ambientes Híbridos de Aprendizagem: um Processo de Inovação Sustentada. Revista UFG, vol. 20: e66027, 1-29.

挖矿

On

remove colorbar if we dont pass list of years

超心理学

在元层面上，你在研究某个现象并得到一些积极的发现。这并不能告诉你太多，除非你让其他研究人员研究一个你知道不存在的现象——但他们自己相信存在——看看有多少人得到积极的发现。这个数字告诉你，无论现象是否真实，有多少研究会发现积极的结果。除非对真实现象的研究比对安慰剂现象的研究表现得显著更好，否则你没有发现任何东西。

试图建立安慰剂科学将是一个后勤噩梦。你必须找到一个绝对不存在的现象， somehow 说服全世界的科学家社区它存在，并资助他们研究它几十年而不被发现。

TTP is an autoimmune disorder in which the body's immune system creates antibodies that destroy an enzyme (ADAMTS13).

Hi

peu élégant

"nous accordons une importance égale à X et à Y"

supprimer "qu'aux" et remplacer par "et aux"

supprimer "quels sont"

++++

joli

avant tout

qui ?

à l'analyse

choisie

s'attache

oui

résoudre un noeud ?

?

oui

est-ce vraiment inédit ?

amenés ? pas sure

je trouve ça un peu trop général/vague/liste

eLife assessment

This important study reports a novel approach to studying cerebellar function based on the idea of selective recruitment using fMRI. It provides convincing evidence for task-dependent gating of neocortical input to the cerebellum during a motor task and a working memory task. The study will be of interest to a broad cognitive neuroscience audience.

Reviewer #1 (Public Review):

This is an interesting and well-written paper reporting on a novel approach to studying cerebellar function based on the idea of selective recruitment using fMRI. The study is well-designed and executed. Analyses are sound and results are properly discussed. The paper makes a significant contribution to broadening our understanding of the role of cerebellum in human behavior.

In the revision, the authors did an excellent job in addressing my concerns.

Reviewer #2 (Public Review):

Summary:

Shahshahani and colleagues used a combination of statistical modelling and whole-brain fMRI data in an attempt to separate the contributions of cortical and cerebellar regions in different cognitive contexts.

Strengths:

* The manuscript uses a sophisticated integration of statistical methods, cognitive neuroscience and systems neurobiology.<br /> * The authors use multiple statistical approaches to ensure robustness in their conclusions.<br /> * The consideration of the cerebellum as not a purely 'motor' structure is excellent and important.

Weaknesses:

* The assumption that cortical BOLD responses in cognitive tasks should be matched irrespective of cerebellar involvement does not cohere directly with the notion of 'forcing functions' introduced by Houk and Wise, suggesting the need for future work.

Author response:

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

Reviewer #1 (Public Review):

This is an interesting and well-written paper reporting on a novel approach to studying cerebellar function based on the idea of selective recruitment using fMRI. The study is well-designed and executed. Analyses are sound and results are properly discussed. The paper makes a significant contribution to broadening our understanding of the role of the cerebellum in human behavior.

We thank the reviewer for the positive assessment of our paper.

(1) While the authors provide a compelling case for the link between BOLD and the cerebellar cortical input layer, there remains considerable unexplained variance. Perhaps the authors could elaborate a bit more on the assumption that BOLD signals mainly reflect the input side of the cerebellum (see for example King et al., elife. 2023 Apr 21;12:e81511).

Our paper is based on the assumption that the cerebellar BOLD signal reflects solely the input to the cerebellum and does not reflect the changes in firing rates of Purkinje cells. This assumption relies on two lines of arguments: Studies that have directly looked at the mechanism of vasodilation in the cerebellum, and studies that try to infer the contributions of different neurophysiological mechanisms to overall cerebellar metabolism (Attwell and Iadecola, 2002).

Vasodilatory considerations: The mechanisms that causes vasodilation in the cerebellum, and hence BOLD signal increases, has been extensively studied: Electrical stimulation of mossy fibers (Gagliano et al., 2022; Mapelli et al., 2017), as well as parallel fibers (Akgören et al., 1994; Iadecola et al., 1996; Mathiesen et al., 1998; Yang and Iadecola, 1997) lead to robust increases in cerebellar blood flow. In contrast to the neocortex, the regulation of blood flow in the cerebellum depends nearly purely on the vasodilator Nitric Oxide (NO) (Akgören et al., 1994; Yang and Iadecola, 1997) with stellate cells playing a key role in the signaling cascade (Yang et al., 2000).

Electrical (Mathiesen et al., 2000) and pharmacological (Yang and Iadecola, 1998) stimulation of climbing fibers also leads to robust increases in blood flow. Simultaneous parallel and climbing fiber stimulation seems to combine sub-additively to determine the blood flow changes (K. Caesar et al., 2003).

Importantly, even dramatic changes in spiking rate of Purkinje cells do not lead to changes in vasodilation. For starters, parallel fiber stimulation leads to blood flow increases, even though the net effect on Purkinje cell firing is inhibitory (Mathiesen et al., 1998). More importantly, complete inhibition of the Purkinje cell using a GABA agonist does not change baseline cerebellar blood flow (Kirsten Caesar et al., 2003). Conversely, even a 200-300% increase in simple (and complex) spike firing rate through application of a GABA antagonist does not show any measurable consequences for blood flow, even though it clearly increases the metabolic rate of oxygen consumption in the tissue (Thomsen et al., 2009, 2004).

In sum, this extensive set of studies clearly argues that the cerebellar blood flow response is mostly dictated by synaptic input, and that the firing rate of Purkinje cells does not influence vasodilation. Because the BOLD signal is caused by an supply of oxygen over and above the level of oxygen consumption, this would argue that increases in Purkinje cell firing would not lead to BOLD increases. What is less clear is the degree to which changes in BOLD signal during normal activity are determined by changes in mossy fiber or climbing fiber input. Disruption of either pathway leads to 60-70% reductions in the evoked blood flow response during whisker stimulation (Yang et al., 2000; Zhang et al., 2003) – but it remains unclear to what degree this reflects the distribution of contributions in the healthy animal, as these powerful disruptions may have a number of side-effects.

Metabolic considerations: To estimate the relative contributions climbing fiber / mossy fiber input to the variations in BOLD signal under natural conditions, it is useful to consider the contributions of different cerebellar processes to the overall metabolism of the cerebellum. Assuming an average firing rate of 40Hz for mossy fibers, ~3Hz for Granule cells, and 1Hz for climbing fibers, Howarth et al. (Howarth et al., 2012, 2010) estimated that the transmission from mossy fibers to granular cells, dominates the energy budget with 53%. The subsequent stage, encompassing the transfer of information from Granular cells to Purkinje cells, accounts for 32% of energy expenditure. In contrast, integration within Purkinje cells and the spiking (simple and complex) of these cells represents only 15% of the total energy consumption.

More important for the BOLD signal, however, are the activity-induced variations in metabolic consumption: Purkinje cells fire relatively constantly at a very high frequency (~50Hz) both during awake periods and during sleep (Shin et al., 2007). When providing a signal to the neocortex, firing rate decreases, actually lowering the metabolic demand. Climbing fibers normally fire at ~0.5 Hz and even during activity rarely fire much above 2Hz (Streng et al., 2017). In contrast, granule cells show a low firing rates during rest (typically <1hz) and can spike during activity well above 100Hz. Combined with the sheer number of granule cells, these considerations would suggest that the vast majority of the variation in metabolic demand are due to mossy fiber input and granule cell activity.

Overall, we therefore think it is likely that the main determinant of the cerebellar cortical BOLD signal is mossy fiber input and the transmission of information from mossy fibers to granule cells to Purkinje cells. We admit that the degree to which climbing fiber input contribute to BOLD signal changes is much less clear. We can be quite certain, however, that the firing rate of Purkinje cells does not contribute to the cerebellar BOLD signal, as even dramatic changes in the firing rate do not cause any changes in vasodilation.  We have clarified our line of reasoning in the paper, and hope this more extensive response here will give the reader a better overview over the pertaining literature.

(2) The current approach does not appear to take the non-linear relationships between BOLD and neural activity into account.

Thank you for raising this concern. We did not stress this point in the paper, but one big advantage of our selective recruitment approach is that it is – to some degree- robust against non-linearities in the relationship between neural activity and BOLD signal. This is the case, as long as the shape of the non-linearity is similar in the cerebellum and the neocortex. The results of our motor task (Figure 3) provide a clear example of this: The BOLD signal both in the neocortex and cerebellum incases non-linearly as a function of force – the increase from 2.5N to 6N (a 3.5N increase) is larger than the increase from 6N to 10N (a 4N increase). A similar non-linearity can be observed for tapping speed (6, 10 to 18 taps / s). However, within each condition, the relationship between cortical and cerebellar activity is nearly perfectly linear, reflecting the fact that the shape of the non-linearity for the cerebellum and cortex is very similar.

Most importantly, even if the non-linearity across the two structures is different, any non-linear relationship between neural activity and BOLD signal (of vasodilatory nature) should apply to different conditions (here force and speed increases) similarly. Therefore, if two conditions show overlapping activity levels (as observed for force and speed across medium and high levels, Figure 3), a offset between conditions cannot be caused by a non-linearity in the relationship of cortical and cerebellar activity. Because all conditions are subject to the same non-linearity, all points should lie on a single (likely monotonically increasing) non-linear function. Both for the motor and working memory task, the pattern of results clearly violates this assumption.

(3) The authors may want to address a bit more the issue of closed loops as well as the underlying neuroanatomy including the deep cerebellar nuclei and pontine nuclei in the context of their current cerebello-cortical correlational approach. But also the contribution of other brain areas such as the basal ganglia and hippocampus. 

Cortical-cerebellar communication is of course bi-directional. As discussed in King at al., (2023), however, we are restricting our model to the connections from the neocortex to the cerebellum for the following reasons: First, cerebellar BOLD activity likely reflects mostly neocortical input (see our answer to pt. 1), whereas neocortical activity is determined by a much wider array of projections, including striato-thalamo-cortical and cortico-cortical connections. Secondly, the output of the cerebellum cannot be predicted from the BOLD signal of the cerebellar cortex, as it is unlikely that the firing rate of Purkinje cells contribute to cerebellar BOLD signal (see pt. 1). For these reasons we believe that the relationship between neocortical and cerebellar activity patterns is mostly dictated by the connectivity from cortex to cerebellum, and is therefore best modelled as thus. This is now more clearly discussed in a new paragraph (line 318-323) of the revised manuscript.

We are also ignoring other inputs to the cerebellum, including the spinal chord, the basal ganglia (Bhuvanasundaram et al., 2022; Bostan and Strick, 2018) hippocampus (Froula et al., 2023; Watson et al., 2019), and amygdala (Farley et al., 2016; Jung et al., 2022; Terburg et al., 2024). In humans, however, the neocortex remains the primary source of input to pontine nuclei. Consequently, it stands as the main structure shaping activity within the cerebellar cortex. While it is an interesting question to what degree the consideration of subcortical structures can improve the prediction of cerebellar activity patterns, we believe that considering the neocortex provides a good first approximation.

Reviewer #1 (Recommendations):

(4)  A few sentences to clarify the used models as was done in the King et al. (2024) paper may improve readability.

We have now added the sentences in the introduction (line 25ff):

To approach this problem, we have recently developed and tested a range of cortical-cerebellar connectivity models (King et al., 2023), designed to capture fixed, or task-invariant, transmission between neocortex and cerebellum. For each cerebellar voxel, we estimated a regularized multiple regression model to predict its activity level across a range of task conditions (King et al., 2019) from the activity pattern observed in the neocortex for the same conditions. The models were then evaluated in their ability to predict cerebellar activity in novel tasks, again based only on the corresponding neocortical activity pattern. Two key results emerged from this work. First, while rs-FC studies (Buckner et al., 2011; Ji et al., 2019; Marek et al., 2018) have assumed a 1:1 mapping between neocortical and cerebellar networks, models which allowed for convergent input from multiple neocortical regions to a single cerebellar region performed better in predicting cerebellar activity patterns for novel tasks. Second, when given a cortical activation pattern, the best performing model could predict about 50% of the reliable variance in the cerebellar cortex across tasks (King et al., 2023).

(5) To what extent does this paper demonstrate the limitations of BOLD in neuroscientific research? 

The primary objective of this study was to shed light on the problems of interpreting BOLD activation within the cerebellum. The problem that the BOLD signal mostly reflect input to a region is not unique to the cerebellum, but also applies (albeit likely to a lesser degree) to other brain structures. However, the solution we propose here critically hinges on three features of the cerebellar circuitry: a) the mossy fiber input for the cerebellar hemispheres mostly arise from the neocortex, b) the BOLD signal is likely dominated by this mossy fiber input (see pt. 1), and c) there is very little excitatory recurrent activity in the cerebellum, so output activity in the cerebellum does not cause direct activity in other parts of the cerebellum.

These features motivate us to use a directed cortex->cerebellum connectivity model, which does not allow for any direct connectivity within the cerebellum. While the same approach can also be applied to other brain structures, it is less clear that the approach would yield valid results here. For example, due the local excitatory recurrent connectivity within neocortical columns, the activity here will also relate to local processing.

(6) What if the authors reversed their line of reasoning as in that cerebellum activity is matched to map changes in cerebral cortical activity? Perhaps this could provide further evidence for the assumed directional specificity of the task-dependent gating of neocortical inputs. 

Given (a) that the cerebellar BOLD signal tells us very little about cerebellar output signals (b) that there are many other input signals to the neocortex that are more powerful than cerebellar inputs, and c) that there strong cortical-cortical connections, we believe that this model would be hard to interpret (see also our answer to pt. 3).

Therefore, while the inversion of the linear task-invariant mapping between cortical and cerebellar activity is a potentially interesting exercise, it is unclear to us at this point what strong predictions we would be able to test with this approach.

(7) The statement that cerebellar fMRI activity may simply reflect the transmission of neocortical activity through fixed connections can be better explained. Also in the context of using the epiphenomenon (on page 11) in the paper. To what extent is the issue of epiphenomenon not a general problem of fMRI research?

We have rephrased the introduction of this idea (line 17):

This means that increases in the cerebellar BOLD signal could simply reflect the automatic transmission of neocortical activity through fixed anatomical connections. As such, whenever a task activates a neocortical region, the corresponding cerebellar region would also be activated, regardless of whether the cerebellum is directly involved in the task or not.

Epiphemonal activity: This is indeed a general problem in fMRI research (and indeed research that uses neurophysiological recordings, rather than manipulations of activity). Indeed, we have discussed similar issues in the context of motor activity in ipsilateral motor cortex (Diedrichsen et al., 2009). However, given that we only offer a possible approach to address this issue for the cerebellum (see pt. 5), we thought it best to keep the scope of the discussion focused on this structure.

Reviewer #2 (Public Review):

Summary:

Shahshahani and colleagues used a combination of statistical modelling and whole-brain fMRI data in an attempt to separate the contributions of cortical and cerebellar regions in different cognitive contexts.

Strengths:

The manuscript uses a sophisticated integration of statistical methods, cognitive neuroscience, and systems neurobiology.

The authors use multiple statistical approaches to ensure robustness in their conclusions.

The consideration of the cerebellum as not a purely 'motor' structure is excellent and important. <br />

We thank the reviewer for their positive evaluation.

Weaknesses:

(1) Two of the foundation assumptions of the model - that cerebellar BOLD signals reflect granule cells > purkinje neurons and that corticocerebellar connections are relatively invariant - are still open topics of investigation. It might be helpful for the reader if these ideas could be presented in a more nuanced light.

Please see response to the comment 1 of Reviewer 1 for a more extensive and detailed justification of this assumption. We have now also clarified our rationale for this assumption better in the paper on line 10-14. Finally, we now also raise explicitly the possibility that some of the violations of the task-invariant model could be caused by selectively increase of climbing fiber activity in some tasks (line 340).

(2) The assumption that cortical BOLD responses in cognitive tasks should be matched irrespective of cerebellar involvement does not cohere with the idea of 'forcing functions' introduced by Houk and Wise. 

We are assuming that you refer to the idea that cerebellar output is an important determinant of the dynamics (and likely also of the magnitude) of neocortical activity. We agree most certainly here. However, we also believe that in the context of our paper, it is justified to restrict the model to the connectivity between the neocortex and the cerebellum only (see reviewer 1, comment 3).

Furthermore, if increased cerebellar output indeed occurs during the conditions for which we identified unusually high cerebellar activity, it should increase neocortical activity, and bring the relationship of the cerebellar and cortical activity again closer to the predictions of the linear model. Therefore, the identification of functions for which cerebellar regions show selective recruitment is rather conservative.

Reviewer #2 (Recommendations):

(3) One of the assumptions stated in the abstract -- that the inputs to the cerebellum may simply be a somewhat passive relay of the outputs of the cerebral cortex -- has been challenged recently by work from Litwin-Kumar (Muscinelli et al., 2023 Nature Neuroscience), which argues for complex computational relationships between cortical pyramidal neurons, pontine nuclei and granule cells, which in turn would have a non-linear impact on the relationship between cortical and cerebellar BOLD. The modelling is based on empirical recordings from Wagner (2019, Cell) which show that the synaptic connections between the cortex and granule cells change as a function of learning, further raising concerns about the assumption that the signals inherent within these two systems should be identical. Whether these micro-scale features are indicative of the macroscopic patterns observed in BOLD is an interesting question for future research, but I worry that the assumption of direct similarity is perhaps not reflective of the current literature. The authors do speak to these cells in their discussion, but I believe that they could also help to refine the authors' hypotheses in the manuscript writ large.

We absolutely agree with your point. However, we want to make extremely clear here that our hypothesis (that the inputs to the cerebellum are a linear task-invariant function of the outputs of the cerebral cortex) is the Null-hypothesis that we are testing in our paper. In fact, our results show the first empirical evidence that task-dependent gating may indeed occur. In this sense, our paper is consistent with the theoretical suggestion of (Muscinelli et al., 2023).

You may ask whether a linear task-invariant model of cortical-cerebellar connectivity is not a strawman, given that is most likely incorrect. However, as we stress in the discussion (line 298-), a good Null-model is a useful model, even if it is (as all models) ultimately incorrect. Without it, we would not be able to determine which cerebellar activity outstrips the linear prediction. The fact that this Null-model itself can predict nearly 50% of the variance in cerebellar activity patterns across tasks at a group level, means that it is actually a very powerful model, and hence is a much more stringent criterion for evidence for functional involvement than just the presence of activity.

(4) Further to this point, I didn't follow the authors' logic that the majority of the BOLD response in the cerebellum is reflective of granule cells rather than Purkinje cells. I read through each of the papers that were cited in defense of the comment: "The cerebellar BOLD signal is dominated by mossy fiber input with very little contribution from the output of the cerebellar cortex, the activity of Purkinje cells" and found that none of these studies made this same direct conclusion. As such, I suggest that the authors soften this statement, or provide a different set of references that directly confirm this hypothesis. 

Please see response to the comment 1, Reviewer 1. We hope the answer provides a more comprehensive overview over the literature, which DOES show that spiking behavior of Purkinje cells does not influence vasodilation (as opposed to mossy fiber input). We have now clarified our rationale and the exact cited literature on line 9-14 of the paper.

(5) Regarding the statement: "As such, whenever a task activates a neocortical region, we might observe activity in the corresponding cerebellar regions regardless of whether the cerebellum is directly involved in the task or not." -- what if this is a feature, rather than a bug? That is, the organisation of the nervous system has been shaped over phylogeny such that every action, via efference copies of motor outputs, is filtered through the complex architecture of the cerebellum in order to provide a feed-forward signal to the thalamus/cortex (and other connected structures). Houk and Wise made compelling arguments in their 1995 Cerebral Cortex paper arguing that these outputs (among other systems) could act as 'forcing functions' on the kinds of dynamics that arise in the cerebral cortex. I am inclined to agree with their hypothesis, where the implication is that there are no tasks that don't (in some way) depend on cerebellar activity, albeit to a lesser or greater extent, depending on the contexts/requirements of the task. I realise that this is a somewhat philosophical point, but I do think it is important to be clear about the assumptions that form the basis of the reasoning in the paper. 

This is an interesting point. Our way of thinking about cerebellar function does indeed correspond quite well to the idea of forcing functions- the idea that cerebellar output can “steer” cortical dynamics in a particular way. However, based on patient and lesion data, it is also clear that some cortical functions rely much more critically on cerebellar input than others. We hypothesize here that cerebellar activity is higher (as compared to the neocortical activity) when the functions require cerebellar computation.

We also agree with the notion that cerebellar contribution is likely not an all-or-none issue, but rather a matter of gradation (line 324ff).

(6) Regarding the logic of expecting the cortical patterns for speed vs. force to be matched -- surely if the cerebellum was involved more in speed than force production, the feedback from the cerebellum to the cortex (via thalamus) could also contribute to the observed differences? How could the authors control for this possibility? 

Our model currently indeed does not attempt to quantify the contributions of cerebellar output to cortical activity. However, given that cerebellar output is not visible in the BOLD signal of the cerebellum (see reviewer 1, comment 1), we believe that this is a rational approach. As argued in our response to your comment 2, increased cerebellar output in the speed compared to the force condition should bring the activity relationship closer to the linear model prediction. The fact that we find increased cerebellar (as compared to neocortical) activity in the speed conditions, suggests that there is indeed task-dependent gating of cortical projections to the cerebellum.

Akgören N, Fabricius M, Lauritzen M. 1994. Importance of nitric oxide for local increases of blood flow in rat cerebellar cortex during electrical stimulation. Proc Natl Acad Sci U S A 91:5903–5907.

Attwell D, Iadecola C. 2002. The neural basis of functional brain imaging signals. Trends Neurosci 25:621–625.

Bhuvanasundaram R, Krzyspiak J, Khodakhah K. 2022. Subthalamic Nucleus Modulation of the Pontine Nuclei and Its Targeting of the Cerebellar Cortex. J Neurosci 42:5538–5551.

Bostan AC, Strick PL. 2018. The basal ganglia and the cerebellum: nodes in an integrated network. Nat Rev Neurosci 19:338–350.

Buckner RL, Krienen FM, Castellanos A, Diaz JC, Yeo BTT. 2011. The organization of the human cerebellum estimated by intrinsic functional connectivity. J Neurophysiol 106:2322–2345.

Caesar K., Gold L, Lauritzen M. 2003. Context sensitivity of activity-dependent increases in cerebral blood flow. Proc Natl Acad Sci U S A 100:4239–4244.

Caesar K., Thomsen K, Lauritzen M. 2003. Dissociation of spikes, synaptic activity, and activity-dependent increments in rat cerebellar blood flow by tonic synaptic inhibition. Proc Natl Acad Sci U S A 100:16000–16005.

Farley SJ, Radley JJ, Freeman JH. 2016. Amygdala Modulation of Cerebellar Learning. J Neurosci 36:2190–2201.

Froula JM, Hastings SD, Krook-Magnuson E. 2023. The little brain and the seahorse: Cerebellar-hippocampal interactions. Front Syst Neurosci 17:1158492.

Gagliano G, Monteverdi A, Casali S, Laforenza U, Gandini Wheeler-Kingshott CAM, D’Angelo E, Mapelli L. 2022. Non-linear frequency dependence of neurovascular coupling in the cerebellar cortex implies vasodilation-vasoconstriction competition. Cells 11:1047.

Howarth C, Gleeson P, Attwell D. 2012. Updated energy budgets for neural computation in the neocortex and cerebellum. J Cereb Blood Flow Metab 32:1222–1232.

Howarth C, Peppiatt-Wildman CM, Attwell D. 2010. The energy use associated with neural computation in the cerebellum. J Cereb Blood Flow Metab 30:403–414.

Iadecola C, Li J, Xu S, Yang G. 1996. Neural mechanisms of blood flow regulation during synaptic activity in cerebellar cortex. J Neurophysiol 75:940–950.

Ji JL, Spronk M, Kulkarni K, Repovš G, Anticevic A, Cole MW. 2019. Mapping the human brain’s cortical-subcortical functional network organization. Neuroimage 185:35–57.

Jung SJ, Vlasov K, D’Ambra AF, Parigi A, Baya M, Frez EP, Villalobos J, Fernandez-Frentzel M, Anguiano M, Ideguchi Y, Antzoulatos EG, Fioravante D. 2022. Novel Cerebello-Amygdala Connections Provide Missing Link Between Cerebellum and Limbic System. Front Syst Neurosci 16:879634.

King M, Hernandez-Castillo CR, Poldrack RA, Ivry RB, Diedrichsen J. 2019. Functional boundaries in the human cerebellum revealed by a multi-domain task battery. Nat Neurosci 22:1371–1378.

King M, Shahshahani L, Ivry RB, Diedrichsen J. 2023. A task-general connectivity model reveals variation in convergence of cortical inputs to functional regions of the cerebellum. Elife 12:e81511.

Mapelli L, Gagliano G, Soda T, Laforenza U, Moccia F, D’Angelo EU. 2017. Granular layer neurons control cerebellar neurovascular coupling through an NMDA receptor/NO-dependent system. J Neurosci 37:1340–1351.

Marek S, Siegel JS, Gordon EM, Raut RV, Gratton C, Newbold DJ, Ortega M, Laumann TO, Adeyemo B, Miller DB, Zheng A, Lopez KC, Berg JJ, Coalson RS, Nguyen AL, Dierker D, Van AN, Hoyt CR, McDermott KB, Norris SA, Shimony JS, Snyder AZ, Nelson SM, Barch DM, Schlaggar BL, Raichle ME, Petersen SE, Greene DJ, Dosenbach NUF. 2018. Spatial and Temporal Organization of the Individual Human Cerebellum. Neuron 100:977-993.e7.

Mathiesen C, Caesar K, Akgören N, Lauritzen M. 1998. Modification of activity-dependent increases of cerebral blood flow by excitatory synaptic activity and spikes in rat cerebellar cortex. J Physiol 512 ( Pt 2):555–566.

Mathiesen C, Caesar K, Lauritzen M. 2000. Temporal coupling between neuronal activity and blood flow in rat cerebellar cortex as indicated by field potential analysis. J Physiol 523:235–246.

Muscinelli SP, Wagner MJ, Litwin-Kumar A. 2023. Optimal routing to cerebellum-like structures. Nat Neurosci 26:1630–1641.

Shin S-L, Hoebeek FE, Schonewille M, De Zeeuw CI, Aertsen A, De Schutter E. 2007. Regular patterns in cerebellar Purkinje cell simple spike trains. PLoS One 2:e485.

Streng ML, Popa LS, Ebner TJ. 2017. Climbing Fibers Control Purkinje Cell Representations of Behavior. J Neurosci 37:1997.

Terburg D, van Honk J, Schutter DJLG. 2024. Doubling down on dual systems: A cerebellum–amygdala route towards action- and outcome-based social and affective behavior. Cortex 173:175–186.

Thomsen K, Offenhauser N, Lauritzen M. 2004. Principal neuron spiking: neither necessary nor sufficient for cerebral blood flow in rat cerebellum. J Physiol 560:181–189.

Thomsen K, Piilgaard H, Gjedde A, Bonvento G, Lauritzen M. 2009. Principal cell spiking, postsynaptic excitation, and oxygen consumption in the rat cerebellar cortex. J Neurophysiol 102:1503–1512.

Watson TC, Obiang P, Torres-Herraez A, Watilliaux A, Coulon P, Rochefort C, Rondi-Reig L. 2019. Anatomical and physiological foundations of cerebello-hippocampal interaction. Elife 8:e41896.

Yang G, Huard JM, Beitz AJ, Ross ME, Iadecola C. 2000. Stellate neurons mediate functional hyperemia in the cerebellar molecular layer. J Neurosci 20:6968–6973.

Yang G, Iadecola C. 1998. Activation of cerebellar climbing fibers increases cerebellar blood flow: role of glutamate receptors, nitric oxide, and cGMP. Stroke 29:499–507; discussion 507-8.

Yang G, Iadecola C. 1997. Obligatory role of NO in glutamate-dependent hyperemia evoked from cerebellar parallel fibers. Am J Physiol 272:R1155-61.

Zhang Y, Forster C, Milner TA, Iadecola C. 2003. Attenuation of activity-induced increases in cerebellar blood flow by lesion of the inferior olive. Am J Physiol Heart Circ Physiol 285:H1177-82.

require 会生成缓存 对于相同路径的文件不会进行重复读取 会优先采用缓存的内容

如果需要读取到最新的结果，可以采用清除缓存的办法 js delete require.cache[require.resolve('./myJson.js')];

of the two rollings, the most complete one has

eLife assessment

This important study presents an original and promising approach to combine convolutional neural networks of visual processing with evidence accumulation models of decision-making. While the methodological approach itself is strong and technically sophisticated, the evidence supporting the conclusions is currently incomplete. The study will be of interest to researchers working in the fields of machine learning and cognitive modeling.

Reviewer #1 (Public Review):

Summary:

This paper introduces a new approach to modeling human behavioral responses using image-computable models. They create a model (VAM) that is a combination of a standard CNN coupled with a standard evidence accumulation model (EAM). The combined model is then trained directly on image-level data using human behavioral responses. This approach is original and can have wide applicability. However, many of the specific findings reported are less compelling.

Strengths:

(1) The manuscript presents an original approach to fitting an image-computable model to human behavioral data. This type of approach is sorely needed in the field.<br /> (2) The analyses are very technically sophisticated.<br /> (3) The behavioral data are large both in terms of sample size (N=75) and in terms of trials per subject.

Weaknesses:

Major

(1) The manuscript appears to suggest that it is the first to combine CNNs with evidence accumulation models (EAMs). However, this was done in a 2022 preprint (https://www.biorxiv.org/content/10.1101/2022.08.23.505015v1) that introduced a network called RTNet. This preprint is cited here, but never really discussed. Further, the two unique features of the current approach discussed in lines 55-60 are both present to some extent in RTNet. Given the strong conceptual similarity in approach, it seems that a detailed discussion of similarities and differences (of which there are many) should feature in the Introduction.

(2) In the approach here, a given stimulus is always processed in the same way through the core CNN to produce activations v_k. These v_k's are then corrupted by Gaussian noise to produce drift rates d_k, which can differ from trial to trial even for the same stimulus. In other words, the assumption built into VAM appears to be that the drift rate variability stems entirely from post-sensory (decisional) noise. In contrast, the typical interpretation of EAMs is that the variability in drift rates is sensory. This is also the assumption built into RTNet where the core CNN produces noisy evidence. Can the authors comment on the plausibility of VAM's assumption that the noise is post-sensory?

(3) Figure 2 plots how well VAM explains different behavioral features. It would be very useful if the authors could also fit simple EAMs to the data to clarify which of these features are explainable by EAMs only and which are not.

(4) VAM is tested in two different ways behaviorally. First, it is tested to what extent it captures individual differences (Figure 2B-E). Second, it is tested to what extent it captures average subject data (Figure 2F-J). It wasn't clear to me why for some metrics only individual differences are examined and for other metrics only average human data is examined. I think that it will be much more informative if separate figures examine average human data and individual difference data. I think that it's especially important to clarify whether VAM can capture individual differences for the quantities plotted in Figures 2F-J.

(5) The authors look inside VAM and perform many exploratory analyses. I found many of these difficult to follow since there was little guidance about why each analysis was conducted. This also made it difficult to assess the likelihood that any given result is robust and replicable. More importantly, it was unclear which results are hypothesized to depend on the VAM architecture and training, and which results would be expected in performance-optimized CNNs. The authors train and examine performance-optimized CNNs later, but it would be useful to compare those results to the VAM results immediately when each VAM result is first introduced.

(6) The authors don't examine how the task-optimized models would produce RTs. They say in lines 371-2 that they "could not examine the RT congruency effect since the task-optimized models do not generate RTs." CNNs alone don't generate RTs, but RTs can easily be generated from them using the same EAM add-on that is part of VAM. Given that the CNNs are already trained, I can't see a reason why the authors can't train EAMs on top of the already trained CNNs and generate RTs, so these can provide a better comparison to VAM.

(7) The Discussion felt very long and mostly a summary of the Results. I also couldn't shake the feeling that it had many just-so stories related to the variety of findings reported. I think that the section should be condensed and the authors should be clearer about which explanations are speculations and which are air-tight arguments based on the data.

(8) In one of the control analyses, the authors train different VAMs on each RT quantile. I don't understand how it can be claimed that this approach can serve as a model of an individual's sensory processing. Which of the 5 sets of weights (5 VAMs) captures a given subject's visual processing? Are the authors saying that the visual system of a given subject changes based on the expected RT for a stimulus? I feel like I'm missing something about how the authors think about these results.

Reviewer #2 (Public Review):

In an image-computable model of speeded decision-making, the authors introduce and fit a combined CCN-EAM (a 'VAM') to flanker-task-like data. They show that the VAM can fit mean RTs and accuracies as well as the congruency effect that is present in the data, and subsequently analyze the VAM in terms of where in the network congruency effects arise.

Overall, combining DNNs and EAMs appears to be a promising avenue to seriously model the visual system in decision-making tasks compared to the current practice in EAMs. Some variants have been proposed or used before (e.g., doi.org/10.1016/j.neuroimage.2017.12.078 , doi.org/10.1007/s42113-019-00042-1), but always in the context of using task-trained models, rather than models trained on behavioral data. However, I was surprised to read that the authors developed their model in the context of a conflict task, rather than a simpler perceptual decision-making task. Conflict effects in human behavior are particularly complex, and thereby, the authors set a high goal for themselves in terms of the to-be-explained human behavior. Unfortunately, the proposed VAM does not appear to provide a great account of conflict effects that are considered fundamental features of human behavior, like the shape of response time distributions, and specifically, delta plots (doi.org/10.1037/0096-1523.20.4.731). The authors argue that it is beyond the scope of the presented paper to analyze delta plots, but as these are central to studies of human conflict behavior, models that aim to explain conflict behavior will need to be able to fit and explain delta plots.

Theories on conflict often suggest that negative/positive-trending delta plots arise through the relative timing of response activation related to relevant and irrelevant information. Accumulation for relevant and irrelevant information would, as a result, either start at different points in time or the rates vary over time. The current VAM, as a feedforward neural network model, does not appear to be able to capture such effects, and perhaps fundamentally not so: accumulation for each choice option is forced to start at the same time, and rates are a static output of the CNN.

The proposed solution of fitting five separate VAMs (one for each of five RT quantiles) is not satisfactory: it does not explain how delta plots result from the model, for the same reason that fitting five evidence accumulation models (one per RT quantile) does not explain how response time distributions arise. If, for example, one would want to make a prediction about someone's response time and choice based on a given stimulus, one would first have to decide which of the five VAMs to use, which is circular. But more importantly, this way of fitting multiple models does not explain the latent mechanism that underlies the shape of the delta plots.

As such, the extensive analyses on the VAM layers and the resulting conclusions that conflict effects arise due to changing representations across layers (e.g., "the selection of task-relevant information occurs through the orthogonalization of relevant and irrelevant representations") - while inspiring, they remain hard to weigh, as they are contingent on the assumption that the VAM can capture human behavior in the conflict task, which it struggles with. That said, the promise of combining CNNs and EAMs is clearly there. A way forward could be to either adjust the proposed model so that it can explain delta plots, which would potentially require temporal dynamics and time-varying evidence accumulation rates, or perhaps to start simpler and combine CCNs-EAMs that are able to fit more standard perceptual decision-making tasks without conflict effects.

Reviewer #3 (Public Review):

Summary:

In this article, the authors combine a well-established choice-response time (RT) model (the Linear Ballistic Accumulator) with a CNN model of visual processing to model image-based decisions (referred to as the Visual Accumulator Model - VAM). While this is not the first effort to combine these modeling frameworks, it uses this combination of approaches uniquely. Specifically, the authors attempt to better understand the structure of human information representations by fitting this model to behavioral (choice-RT) data from a classic flanker task. This objective is made possible by using a very large (by psychological modeling standards) industry data set to jointly fit both components of this VAM model to individual-level data. Using this approach, they illustrate (among other results) (1) how the interaction between target and flanker representations influence the presence and strength of congruency effects, (2) how the structure of representations changes (distributed versus more localized) with depth in the CNN model component, and (3) how different model training paradigms change the nature of information representations. This work contributes to the ML literature by demonstrating the value of training models with richer behavioral data. It also contributes to cognitive science by demonstrating how ML approaches can be integrated into cognitive modeling. Finally, it contributes to the literature on conflict modeling by illustrating how information representations may lead to some of the classic effects observed in this area of research.

Strengths:

(1) The data set used for this analysis is unique and is made publicly available as part of this article. Specifically, they have access to data for 75 participants with >25,000 trials per participant. This scale of data/individual is unusual and is the foundation on which this research rests.

(2) This is the first time, to my knowledge, that a model combining a CNN with a choice-RT model has been jointly fit to choice-RT data at the level of individual people. This type of model combination has been used before but in a more restricted context. This joint fitting, and in particular, learning a CNN through the choice-RT modeling framework, allows the authors to probe the structure of human information representations learned directly from behavioral data.

(3) The analysis approaches used in this article are state-of-the-art. The training of these models is straightforward given the data available. The interesting part of this article (opinion of course) is the way in which they probe what CNN has learned once trained. I find their analysis of how distractor and target information interfere with each other particularly compelling as well as their demonstration that training on behavioral data changes the structure of information representations when compared to training models on standard task-optimized data.

Weaknesses:

(1) Just as the data in this article is a major strength, it is also a weakness. This type of modeling would be difficult, if not impossible to do with standard laboratory data. I don't know what the data floor would be, but collecting tens of thousands of decisions for a single person is impractical in most contexts. Thus this type of work may live in the realm of industry. I do want to re-iterate that the data for this study was made publicly available though!

2) While this article uses choice-RT data it doesn't fully leverage the richness of the RT data itself. As the authors point out, this modeling framework, the LBA component in particular, does not account for some of the more nuanced but well-established RT effects in this data. This is not a big concern given the already nice contributions of this article and it leads to an opportunity for ongoing investigation.

Reviewer #3 (Public Review):

Summary:

Pyruvate kinase M2 (PKM2) is a rate-limiting enzyme in glycolysis and its translocation to the nucleus in astrocytes in various nervous system pathologies has been associated with a metabolic switch to glycolysis which is a sign of reactive astrogliosis. The authors investigated whether this occurs in experimental autoimmune encephalomyelitis (EAA), an animal model of multiple sclerosis (MS). They show that in EAA, PKM2 is ubiquitinated by TRIM21 and transferred to the nucleus in astrocytes. Inhibition of TRIM21-PKM2 axis efficiently blocks reactive gliosis and partially alleviates symptoms of EAA. Authors conclude that this axis can be a potential new therapeutic target in the treatment of MS.

Strengths:

The study is well-designed, controls are appropriate and a comprehensive battery of experiments has been successfully performed. Results of in vitro assays, single-cell RNA sequencing, immunoprecipitation, RNA interference, molecular docking, and in vivo modeling etc. complement and support each other.

Weaknesses:

Though EAA is a valid model of MS, a proposed new therapeutic strategy based on this study needs to have support from human studies.

Reviewer #1 (Public Review):

Summary:

Yang, Hu et al. examined the molecular mechanisms underlying astrocyte activation and its implications for multiple sclerosis. This study shows that the glycolytic enzyme PKM2 relocates to astrocyte nuclei upon activation in EAE mice. Inhibiting PKM2's nuclear import reduces astrocyte activation, as evidenced by decreased proliferation, glycolysis, and inflammatory cytokine release. Crucially, the study identifies TRIM21 as pivotal in regulating PKM2 nuclear import via ubiquitination. TRIM21 interacts with PKM2, promoting its nuclear translocation and enhancing its activity, affecting multiple signaling pathways. Confirmatory analyses using single-cell RNA sequencing and immunofluorescence demonstrate TRIM21 upregulation in EAE astrocytes. Modulating TRIM21 expression in primary astrocytes impacts PKM2-dependent glycolysis and proliferation. In vivo experiments targeting this mechanism effectively mitigate disease severity, CNS inflammation, and demyelination in EAE.

The authors supported their claims with various experimental approaches, however, some results should be supported with higher-quality images clearly depicting the conclusions and additional quantitative analyses of Western blots.

Strength:

This study presents a comprehensive investigation into the function and molecular mechanism of metabolic reprogramming in the activation of astrocytes, a critical aspect of various neurological diseases, especially multiple sclerosis. The study uses the EAE mouse model, which closely resembles MS. This makes the results relevant and potentially translational. The research clarifies how TRIM21 regulates the nuclear import of PKM2 through ubiquitination by integrating advanced techniques. Targeting this axis may have therapeutic benefits since lentiviral vector-mediated knockdown of TRIM21 in vivo significantly reduces disease severity, CNS inflammation, and demyelination in EAE animals.

Weaknesses:

The authors reported that PKM2 levels are elevated in the nucleus of astrocytes at different EAE phases compared to cytoplasmic localization. However, Figure 1 also shows elevated cytoplasmic expression of PKM2. The authors should clarify the nuclear localization of PKM2 by providing zoomed-in images. An explanation for the increased cytoplasmic PKM2 expression should provided. Similarly, while PKM2 translocation is inhibited by DASA-58, in addition to its nuclear localization, a decrease in the cytoplasmic localization of PKM2 is also observed. This situation brings to mind the possibility of a degradation mechanism being involved when its nuclear translocation of PKM2 is inhibited.

In Figure 3D, the authors claim that PKM2 expression causes nuclear retention of STAT3, p65, and p50, and inhibiting PKM2 localization with DASA-58 suppresses this retention. The western blot results for the MOG-stimulated group show high levels of STAT3, p50, and p65 in nuclear localization. However, in the MOG and DASA-58 treated group, one would expect high levels of p50, p65, and STAT3 proteins in the cytoplasm, while their levels decrease in the nucleus. These western blot results could be expanded. Additionally, intensity quantification for these results would be beneficial to see the statistical difference in their expressions, especially to observe the nuclear localization of PKM2.

The discrepancy between Figure 7A and its explaining text is confusing. The expectation from the knocking down of TRIM21 is the amelioration of activated astrocytes, leading to a decrease in inflammation and the disease state. The presented results support these expectations, while the images showing demyelination in EAE animals are not highly supportive. Clearly labeling demyelinated areas would enhance readers' understanding of the important impact of TRIM21 knockdown on reducing the disease severity.

Reviewer #2 (Public Review):

This study significantly advances our understanding of the metabolic reprogramming underlying astrocyte activation in neurological diseases such as multiple sclerosis. By employing an experimental autoimmune encephalomyelitis (EAE) mouse model, the authors discovered a notable nuclear translocation of PKM2, a key enzyme in glycolysis, within astrocytes.

Preventing this nuclear import via DASA 58 substantially attenuated primary astrocyte activation, characterized by reduced proliferation, glycolysis, and inflammatory cytokine secretion.<br /> Moreover, the authors uncovered a novel regulatory mechanism involving the ubiquitin ligase TRIM21, which mediates PKM2 nuclear import. TRIM21 interaction with PKM2 facilitated its nuclear translocation, enhancing its activity in phosphorylating STAT3, NFκB, and c-myc. Single-cell RNA sequencing and immunofluorescence staining further supported the upregulation of TRIM21 expression in astrocytes during EAE.

Manipulating this pathway, either through TRIM21 overexpression in primary astrocytes or knockdown of TRIM21 in vivo, had profound effects on disease severity, CNS inflammation, and demyelination in EAE mice. This comprehensive study provides invaluable insights into the pathological role of nuclear PKM2 and the ubiquitination-mediated regulatory mechanism driving astrocyte activation.

The author's use of diverse techniques, including single-cell RNA sequencing, immunofluorescence staining, and lentiviral vector knockdown, underscores the robustness of their findings and interpretations. Ultimately, targeting this PKM2-TRIM21 axis emerges as a promising therapeutic strategy for neurological diseases involving astrocyte dysfunction.

While the strengths of this piece of work are undeniable, some concerns could be addressed to refine its impact and clarity further; as outlined in the recommendations for the authors.

Reviewer #4 (Public Review):

Summary:

The authors report the role of the Pyruvate Kinase M2 (PKM2) enzyme nuclear translocation as fundamental in the activation of astrocytes in a model of autoimmune encephalitis (EAE). They show that astrocytes, activated through culturing in EAE splenocytes medium, increase their nuclear PKM2 with consequent activation of NFkB and STAT3 pathways. Prevention of PKM2 nuclear translocation decreases astrocyte counteracts this activation. The authors found that the E3 ubiquitin ligase TRIM21 interacts with PKM2 and promotes its nuclear translocation. In vivo, either silencing of TRIM21 or inhibition of PKM2 nuclear translocation ameliorates the severity of the disease in the EAE model.

Strengths:

This work contributes to the knowledge of the complex action of the PKM2 enzyme in the context of an autoimmune-neurological disease, highlighting its nuclear role and a novel partner, TRIM21, and thus adding a novel rationale for therapeutic targeting.

Weaknesses:

Despite the relevance of the work and its goals, some of the conclusions drawn would require more thorough proof:

I believe that the major weakness is the fact that TRIM21 is known to have per se many roles in autoimmune and immune pathways and some of the effects observed might be due to a PKM2-independent action. Some of the experiments to link the two proteins, besides their interaction, do not completely clarify the issue. On top of that, the in vivo experiments address the role of TRIM21 and the nuclear localisation of PKM2 independently, thus leaving the matter unsolved.

Some experimental settings are not described to a level that is necessary to fully understand the data, especially for a non-expert audience: e.g. the EAE model and MOG treatment; action and reference of the different nuclear import inhibitors; use of splenocyte culture medium and the possible effect of non-EAE splenocytes.

The statement that PKM2 is a substrate of TRIM21 ubiquitin ligase activity is an overinterpretation. There is no evidence that this interaction results in ubiquitin modification of PKM2; the ubiquitination experiment is minimal and is not performed in conditions that would allow us to see ubiquitination of PKM2 (e.g. denaturing conditions, reciprocal pull-down, catalytically inactive TRIM21, etc.).

for - insight - progress trap - problem with problem solving

summary - This piece is difficult for a non-expert to understand

eLife assessment

The manuscript establishes a sophisticated mouse model for acute retinal artery occlusion (RAO) by combining unilateral pterygopalatine ophthalmic artery occlusion (UPOAO) with a silicone wire embolus and carotid artery ligation, generating ischemia-reperfusion injury upon removal of the embolus. This clinically relevant model is useful for studying the cellular and molecular mechanisms of RAO. The data overall are solid, presenting a novel tool for screening pathogenic genes and promoting further therapeutic research in RAO.

Reviewer #1 (Public Review):

Summary:

Wang, Y. et al. used a silicone wire embolus to definitively and acutely clot the pterygopalatine ophthalmic artery in addition to carotid artery ligation to completely block the blood supply to the mouse inner retina, which mimics clinical acute retinal artery occlusion. A detailed characterization of this mouse model determined the time course of inner retina degeneration and associated functional deficits, which closely mimic human patients. Whole retina transcriptome profiling and comparison revealed distinct features associated with ischemia, reperfusion, and different model mechanisms. Interestingly and importantly, this team found a sequential event including reperfusion-induced leukocyte infiltration from blood vessels, residual microglial activation, and neuroinflammation that may lead to neuronal cell death.

Strengths:

Clear demonstration of the surgery procedure with informative illustrations, images, and superb surgical videos.

Two-time points of ischemia and reperfusion were studied with convincing histological and in vivo data to demonstrate the time course of various changes in retinal neuronal cell survivals, ERG functions, and inner/outer retina thickness.

The transcriptome comparison among different retinal artery occlusion models provides informative evidence to differentiate these models.

The potential applications of the in vivo retinal ischemia-reperfusion model and relevant readouts demonstrated by this study will certainly inspire further investigation of the dynamic morphological and functional changes of retinal neurons and glial cell responses during disease progression and before and after treatments.

Weaknesses:

It would be beneficial to the manuscript and the readers if the authors could improve the English of this manuscript by correcting obvious grammar errors, eliminating many of the acronyms that are not commonly used by the field, and providing a reason why this complicated but clever surgery procedure was designed and a summary table with the time course of all the morphological, functional, cellular, and transcriptome changes associated with this model.

Reviewer #2 (Public Review):

Summary:

The authors of this manuscript aim to develop a novel animal model to accurately simulate the retinal ischemic process in retinal artery occlusion (RAO). A unilateral pterygopalatine ophthalmic artery occlusion (UPOAO) mouse model was established using silicone wire embolization combined with carotid artery ligation. This manuscript provided data to show the changes in major classes of retinal neural cells and visual dysfunction following various durations of ischemia (30 minutes and 60 minutes) and reperfusion (3 days and 7 days) after UPOAO. Additionally, transcriptomics was utilized to investigate the transcriptional changes and elucidate changes in the pathophysiological process in the UPOAO model post-ischemia and reperfusion. Furthermore, the authors compared transcriptomic differences between the UPOAO model and other retinal ischemic-reperfusion models, including HIOP and UCCAO, and revealed unique pathological processes.

Strengths:

The UPOAO model represents a novel approach to studying retinal artery occlusion. The study is very comprehensive.

Weaknesses:

Some statements are incorrect and confusing. It would be helpful to review and clarify these to ensure accuracy and improve readability.

Author response:

eLife assessment:

The manuscript establishes a sophisticated mouse model for acute retinal artery occlusion (RAO) by combining unilateral pterygopalatine ophthalmic artery occlusion (UPOAO) with a silicone wire embolus and carotid artery ligation, generating ischemia-reperfusion injury upon removal of the embolus. This clinically relevant model is useful for studying the cellular and molecular mechanisms of RAO. The data overall are solid, presenting a novel tool for screening pathogenic genes and promoting further therapeutic research in RAO.

Thank you for recognizing the sophistication and clinical relevance of our mouse model for acute retinal artery occlusion. We are grateful for your supportive feedback.

Public Reviews:

Reviewer #1:

Summary:

Wang, Y. et al. used a silicone wire embolus to definitively and acutely clot the pterygopalatine ophthalmic artery in addition to carotid artery ligation to completely block the blood supply to the mouse inner retina, which mimics clinical acute retinal artery occlusion. A detailed characterization of this mouse model determined the time course of inner retina degeneration and associated functional deficits, which closely mimic human patients. Whole retina transcriptome profiling and comparison revealed distinct features associated with ischemia, reperfusion, and different model mechanisms. Interestingly and importantly, this team found a sequential event including reperfusion-induced leukocyte infiltration from blood vessels, residual microglial activation, and neuroinflammation that may lead to neuronal cell death.

Strengths:

Clear demonstration of the surgery procedure with informative illustrations, images, and superb surgical videos.

Two-time points of ischemia and reperfusion were studied with convincing histological and in vivo data to demonstrate the time course of various changes in retinal neuronal cell survivals, ERG functions, and inner/outer retina thickness.

The transcriptome comparison among different retinal artery occlusion models provides informative evidence to differentiate these models.

The potential applications of the in vivo retinal ischemia-reperfusion model and relevant readouts demonstrated by this study will certainly inspire further investigation of the dynamic morphological and functional changes of retinal neurons and glial cell responses during disease progression and before and after treatments.

We sincerely appreciate your detailed and positive feedback. These evaluations are invaluable in highlighting the significance and impact of our work. Thank you for your thoughtful and supportive review.

Weaknesses:

It would be beneficial to the manuscript and the readers if the authors could improve the English of this manuscript by correcting obvious grammar errors, eliminating many of the acronyms that are not commonly used by the field, and providing a reason why this complicated but clever surgery procedure was designed and a summary table with the time course of all the morphological, functional, cellular, and transcriptome changes associated with this model.

Thank you for your thorough review of the manuscript. We sincerely apologize for any grammatical errors resulting from our English language proficiency and have taken the necessary steps to polish the article. Additionally, we have heeded your advice and reduced the use of field-specific acronyms to enhance readability for both the manuscript and its readers.

Regarding the rationale behind the design of the UPOAO model, we have provided a description in Introduction section. Our group focuses on the research of pathogenesis and clinical treatment for RAO. The absence of an accurate mouse model simulating the retinal ischemic process has hampered progress in developing neuroprotective agents for RAO. To better simulate the retinal ischemic process and possible ischemia-reperfusion injury following RAO, we developed a novel vascular-associated mouse model called the unilateral pterygopalatine ophthalmic artery occlusion (UPOAO) model. We drew inspiration from the widely employed middle cerebral artery occlusion (MCAO) model, commonly used in cerebral ischemic injury research, which guided the development of the UPOAO model.

We appreciate your valuable suggestion regarding the inclusion of a summary table outlining the time course of morphological, functional, cellular, and transcriptome changes associated with this model. To address this, we intend to include a supplementary table at the end of the article, which will offer a comprehensive overview of the experimental results, thereby aiding in clarity and interpretation.

Once again, we thank you for your insightful comments and suggestions, which have greatly contributed to the improvement of our manuscript.

Reviewer #2:

Summary:

The authors of this manuscript aim to develop a novel animal model to accurately simulate the retinal ischemic process in retinal artery occlusion (RAO). A unilateral pterygopalatine ophthalmic artery occlusion (UPOAO) mouse model was established using silicone wire embolization combined with carotid artery ligation. This manuscript provided data to show the changes in major classes of retinal neural cells and visual dysfunction following various durations of ischemia (30 minutes and 60 minutes) and reperfusion (3 days and 7 days) after UPOAO. Additionally, transcriptomics was utilized to investigate the transcriptional changes and elucidate changes in the pathophysiological process in the UPOAO model post-ischemia and reperfusion. Furthermore, the authors compared transcriptomic differences between the UPOAO model and other retinal ischemic-reperfusion models, including HIOP and UCCAO, and revealed unique pathological processes.

Strengths:

The UPOAO model represents a novel approach to studying retinal artery occlusion. The study is very comprehensive.

We greatly appreciate your positive assessment of our work and are encouraged by your recognition of its significance.

Weaknesses:

Some statements are incorrect and confusing. It would be helpful to review and clarify these to ensure accuracy and improve readability.

We sincerely appreciate your meticulous review of the manuscript. Taking into account your valuable feedback, we will thoroughly address the inaccuracies identified in the revised version. Additionally, we will commit to polishing the article to ensure improved readability. We apologize for any confusion caused by these inaccuracies and genuinely thank you for bringing them to our attention.

eLife assessment

This study describes useful mouse models of knock-ins of human STING1 variants and an assessment of these variants' action in mouse immune cells. While the data included in the manuscript are solid, because of the authors' interpretation and conclusions made, the work is currently considered incomplete.

Reviewer #1 (Public Review):

Summary:

This manuscript by Aybar-Torres et al investigated the effect of common human STING1 variants on STING-mediated T cell phenotypes in mice. The authors previously made knock-in mice expressing human STING1 alleles HAQ or AQ, and here they established a new knock-in line Q293. The authors stimulated cells isolated from these mice with STING agonists and found that all three human mutant alleles resist cell death, leading to the conclusion that R293 residue is essential for STING-mediated cell death (there are several caveats with this conclusion, more below). The authors also bred HAQ and AQ alleles to the mouse Sting1-N153S SAVI mouse and observed varying levels of rescue of disease phenotypes with the AQ allele showing more complete rescue than the HAQ allele. The Q293 allele was not tested in the SAVI model. They conclude that the human common variants such as HAQ and AQ have a dominant negative effect over the gain-of-function SAVI mutants.

Strengths:

The authors and Dr. Jin's group previously made important observations of common human STING1 variants, and these knock-in mouse models are essential for understanding the physiological function of these alleles.

Weaknesses:

However, although some of the observations reported here are interesting, the data collectively does not support a unified model. The authors seem to be drawing two sets of conclusions from in vitro and in vivo experiments, and neither mechanism is clear. Several experiments need better controls, and these knock-in mice need more comprehensive functional characterization.

(1) In Figure 1, the authors are trying to show that STING agonist-induced splenocytes cell death is blocked by HAQ, AQ and Q alleles. The conclusion at line 134 should be splenocytes, not lymphocytes. Most experiments in this figure were done with mixed population that may involve cell-to-cell communication. Although TBK1-dependence is likely, a single inhibitor treatment of a mixed population is not sufficient to reach this conclusion.<br /> (2) Q293 knock-in mouse needs to be characterized and compared to HAQ and AQ. Is this mutant expressed in tissues? Does this mutant still produce IFN and other STING activities? Does the protein expression level altered on Western blot? Is the mutant protein trafficking affected? In the authors' previous publications and some of the Western blot here, expression levels of each of these human STING1 protein in mice are drastically different. HAQ and AQ also have different effects on metabolism (pmid: 36261171), which could complicate interoperation of the T cell phenotypes.<br /> (3) HAQ/WT and AQ/WT splenocytes are protected from STING agonist-induced cell death equally well (Figure 1G). HAQ/SAVI shows less rescue compared to AQ/SAVI. These are interesting observations, but mechanism is unclear and not clearly discussed. E.g., how does AQ protect disease pathology better than HAQ (that contains AQ)? Does Q293 allele also fully rescue SAVI?<br /> (4) Figure 2 feels out of place. First of all, why are the authors using human explant lung tissues? PBMCs should be a better source for lymphocytes. In untreated conditions, both CD4 and B cells show ~30% dying cells, but CD8 cells show 0% dying cells. This calls for technical concerns on the CD8 T cell property or gating strategy because in the mouse experiment (Figure 1A) all primary lymphocytes show ~30% cell death at steady-state. Second, Figure 2C, these type of partial effect needs multiple human donors to confirm. Three, the reconstitution of THP1 cells seems out of place. STING-mediated cell death mechanism in myeloid and lymphoid cells are likely different. If the authors want to demonstrate cell death in myeloid cells using THP1, then these reconstituted cell lines need to be better validated. Expression, IFN signaling, etc. The parental THP1 cells is HAQ/HAQ, how does that compare to the reconstitutions? There are published studies showing THP1-STING-KO cells reconstituted with human variants do not respond to STING agonists as expected. The authors need to be scientifically rigorous on validation and caution on their interpretations.<br /> (5) Figure 2G, H, I are confusing. AQ is more active in producing IFN signaling than HAQ and Q is the least active. How to explain this?<br /> (6) The overall model is unclear. If HAQ, AQ and Q are loss-of-function alleles and Q is the key residue for STING-mediated cell death, then why AQ is the most active in producing IFN signaling and AQ/SAVI rescues disease most completely? If these human variants act as dominant negatives, which would be consistent with the WT/het data, then how do you explain AQ is more dominant negative than HAQ?<br /> (7) As a general note, SAVI disease phenotypes involve multiple cell types. Lymphocyte cell death is only one of them. The authors' characterization of SAVI pathology is limited and did not analyze immunopathology of the lung.<br /> (8) Line 281, the discussion on HIV T cell death mechanism is not relevant and over-stretching. This study did not evaluate viral infection in T cells at all. The original finding of HAQ/HAQ enrichment in HIV/AIDS was 2/11 in LTNP vs 0/11 in control, arguably not the strongest statistics.

Reviewer #2 (Public Review):

Aybar-Torres and colleagues utilize common human STING alleles to dissect the mechanism of SAVI inflammatory disease. The authors demonstrate that these common alleles alleviate SAVI pathology in mice, and perhaps more importantly use the differing functionality of these alleles to provide insight into requirements of SAVI disease induction. Their findings suggest that it is residue A230 and/or Q293 that are required for SAVI induction, while the ability to induce an interferon-dependent inflammatory response is not. This is nicely exemplified by the AQ/SAVI mice that have an intact inflammatory response to STING activation, yet minimal disease progression. As both mutants seem to be resistant STING-dependent cell death, this manuscript also alludes to the importance of STING-dependent cell death, rather than STING-dependent inflammation, in the progression of SAVI pathology. While I have some concerns, I believe this manuscript makes some important connections between STING pathology mouse models and human genetics that would contribute to the field.

Some points to consider:

(1) While the CD4+ T cell counts from HAQ/SAVI and AQ/SAVI mice suggest that these T cells are protected from STING-dependent cell death, an assay that explores this more directly would strengthen the manuscript. This is also supported by Fig 2C, but I believe a strength of this manuscript is the comparison between the two alleles. Therefore, if possible, I would recommend the isolation of T cells from these mice and direct stimulation with diABZI or other STING agonist with a cell death readout.<br /> (2) Related to the above point - further exemplifying that the Q293 locus is essential to disease, even in human cells, would also strengthen the paper. It seems that CD4 T cell loss is a major component of human SAVI. While not completely necessary, repeating the THP1 cell death experiments from Fig 2 with a human T cell line would round out the study nicely.<br /> (3) While I found the myeloid cell counts and BMDM data interesting, I think some more context is needed to fully loop this data into the story. Is myeloid cell expansion exemplified by SAVI patients? Do we know if myeloid cells are the major contributors to the inflammation these patients experience? Why should the SAVI community care about the Q293 locus in myeloid cells?<br /> (4) The functional assays in Figure 4 are exciting and really connect the alleles to disease progression. To strengthen the manuscript and connect all the data, I would recommend additional readouts from these mice that address the inflammatory phenotype shown in vitro in Figure 5. For example, measuring cytokines from these mice via ELISA or perhaps even Western blots looking for NFkB or STING activation would be supportive of the story. This would also allow for some tissue specificity. I believe looking for evidence of inflammation and STING activation in the lungs of these mice, for example, would further connect the data to human SAVI pathology.

Author response:

We deeply appreciate the editors’ and reviewers’ invaluable time and effort. We would also like to extend our gratitude to eLife for its unwavering commitment to a transparent review and publication model. Below, we present our point-by-point responses to the comments.  

Besides the WT allele, equivalent to the mouse TMEM173 gene, the human TMEM173 gene has two common alleles: the HAQ and AQ alleles carried by billions of people. The main conclusions and interpretation, summarized in the Title and Abstract, are (i) Different from the WT TMEM173 allele, the HAQ or AQ alleles are resistant to STING activation-induced cell death; (ii) STING residue 293 is critical for cell death; (iii) HAQ, AQ alleles are dominant to the SAVI allele; iv) One copy of the AQ allele rescues the SAVI disease in mice. We propose that STING research and STING-targeting immunotherapy should consider human TMEM173 heterogeneity. These interpretations and conclusions were based on Data and Logic. We welcome alternative, logical interpretations from our peers and potential collaborations to advance the human TMEM173 research.  

Reviewer #1 (Public Review):

Responses to Comment 1: We greatly appreciate Reviewer 1's insights. We will change the “lymphocytes” to “splenocytes” (line 134) as suggested. We respectfully disagree with Reviewer 1’s comments on TBK1 (lines 129 – 134). First, we used two different TBK1 inhibitors: BX795 and GSK8612. Second, because BX795 also inhibits PDK1, we used a PDK1 inhibitor GSK2334470; Third, both BX795 and GSK8612 completely inhibited diABZI-induced splenocyte cell death (Figure 1B). The logical conclusion is “TBK1 activation is required for STING-mediated mouse spleen cell death ex vivo”. (line 118). 

This manuscript uncovers a significant aspect of the interplay between the common human TMEM173 alleles and the rare SAVI mutation (lines 23-26). Our discovery that the common human TMEM173 alleles are resistant to STING activation-induced cell death is a substantial finding. It further strengthens the argument that the HAQ and AQ alleles are functionally distinct from the WT allele 1-3. We wish to underscore the crucial message of this study-that 'STING research and STING-targeting immunotherapy should consider TMEM173 heterogeneity in humans' (line 37), which has been largely overlooked in current STING clinical trials 4.  

Regarding STING-Cell death, as we stated in the Introduction (lines 62-79). (i) STING-mediated cell death is cell type-dependent 5-7 and type I IFNs-independent 5,7,8. (ii) The in vivo biological significance of STING-mediated cell death is not clear 7,8. (iii) The mechanisms of STING-Cell death remain controversial. Multiple cell death pathways, i.e., apoptosis, necroptosis, pyroptosis, ferroptosis, and PANoptosis, are proposed 7,9,10. SAVI patients (WT/SAVI) and mouse models had CD4 T cellpenia 8,11. SAVI/HAQ, SAVI/AQ restored T cells in mice. Thus, the manuscript provides some answers to the biological significance of STING-cell death. Next, splenocytes from Q293/Q293 mice are resistant to STING cell death. The logical conclusion is that the amino acid 293 is critical for STING cell death. How aa293 mediates this function needs future investigation. Similarly, how TBK1 mediates STING cell death, independent of type I IFNs and NFκB induction, needs future investigation.

Responses to Comment 2: These are all very interesting questions that we will address in future studies. This manuscript, titled “The common TMEM173 HAQ, AQ alleles rescue CD4 T cellpenia, restore T-regs, and prevent SAVI (N153S) inflammatory disease in mice” does not focus on Q293 mice. We have been researching the common human TMEM173 alleles since 2011 from the discovery12 , mouse model1,3, human clinical trial2, and human genetics studies 3. This manuscript is another step towards understanding these common human TMEM173 alleles with the new discovery that HAQ, AQ are resistant to STING cell death. 

Responses to Comment 3: We aim to address these worthy questions in future studies. In this manuscript, Figure 6 shows AQ/SAVI had more T-regs than HAQ/SAVI (lines 246 – 256). In our previous publication on HAQ, AQ knockin mice, we showed that AQ T-regs have more IL-10 and mitochondria activity than HAQ T-regs 3. We propose that increased IL-10+

Tregs in AQ mice may contribute to an improved phenotype in AQ/SAVI compared to

HAQ/SAVI. However, we are not excluding other contributions (e.g. metabolic difference) by the AQ allele. We will explore these possibilities in future research.   

Responses to Comment 4: Figure 2 is necessary because it reveals the difference between mouse and human STING cell death. Figure 2A-2B showed that STING activation killed human CD4 T cells, but not human CD8 T cells or B cells. This observation is different from Figure 1A, where STING activation killed mouse CD4, CD8 T cells, and CD19 B cells, revealing the species-specific STING cell death responses. Regarding human CD8 T cells, as we stated in the Discussion (lines 318-320), human CD8 T cells (PBMC) are not as susceptible as the CD4 T cells to STING-induced cell death 8. We used lung lymphocytes that showed similar observations (Figure 2A). For Figure 2C, we used 2 WT/HAQ and 3 WT/WT individuals (lines 738-739). We generate HAQ, AQ THP-1 cells in STING-KO THP-1 cells (Invivogen,, cat no. thpd-kostg) (lines 740-741). 

A recent study found that STING agonist SHR1032 induces cell death in STING-KO THP-1 cells expressing WT(R232) human STING 10 (line 182) independent of type I IFNs. SHR1032 suppressed THP1-STING-WT(R232) cell growth at GI50: 23 nM while in the parental THP1STING-HAQ cells, the GI50 of SHR1032 was >103 nM 10. Cytarabine was used as an internal control where SHR1032 killed more robustly than cytarabine in the THP1-STING-WT(R232) cells but much less efficiently than cytarabine in the THP-1-STING-HAQ cells 10.   

This manuscript rigorously uses mouse splenocytes, human lung lymphocytes, THP-1 reconstituted with HAQ, AQ, and HAQ/SAVI, AQ/SAVI mice, to demonstrate that the common human HAQ, AQ alleles are resistant to STING cell death in vitro and in vivo.

We agree with reviewer 1 that STING-mediated cell death mechanisms in myeloid and lymphoid cells may be different and likely contribute to the different mechanisms proposed in STING cell death research 7,9,10. Our study focuses on the in vivo mechanism of T cellpenia.  

Responses to Comment 5: We stated in the Introduction that “AQ responds to CDNs and produce type I IFNs in vivo and in vitro 3,13,14 ”(line 94, 95). We reported that the AQ knock in mice responded to STING activation 3. We previously showed that there was a negative natural selection on the AQ allele in individuals outside of Africa 3. 28% of Africans are WT/AQ but only 0.6% East Asians are WT/AQ 3. Future research on the AQ allele will address this interesting question that may shed new mechanistic light on STING action.

Responses to Comment 6: The comment here is similar to comment 3. In this manuscript, Figure 6 shows AQ/SAVI had more T-regs than HAQ/SAVI (lines 246 – 256). In our previous publication on HAQ, AQ knockin mice, we showed that AQ T-regs have more IL-10 and mitochondria activity than HAQ T-regs 3. We propose that increased IL-10+ Tregs in AQ mice may contribute to an improved phenotype in AQ/SAVI compared to HAQ/SAVI. However, we are not excluding other contributions (e.g. metabolic difference) by the AQ allele.

Responses to Comment 7: Both radioresistant parenchymal and/or stromal cells and hematopoietic cells influence SAVI pathology in mice 15,16. Nevertheless, the lack of CD 4 T cells, including the anti-inflammatory T-regs, likely contributes to the inflammation in SAVI mice and patients. We characterized lung function, lung inflammation (Figure 4), lung neutrophils, and inflammatory monocyte infiltration (Figure S4). 

Responses to Comment 8: Several publications have linked STING to HIV pathogenesis 17-22  (line 271). The manuscript studies STING activation-induced cell death. It is not stretching to ask, for example, does preventing STING cell death, without affecting type I IFNs production, restore CD4 T cell counts and improve care for AIDS patients?

Reviewer #2 (Public Review):

Response to Comment 1: Please see the Figure below for cell death by diABZI, DMXAA in Splenocytes from WT/WT, WT/HAQ, HAQ/SAVI, AQ/SAVI mice. The HAQ/SAVI and AQ/SAVI splenocytes showed similar partial resistance to STING activationinduced cell death. 

Responses to Comment 2: We examined HAQ, AQ mouse splenocytes, HAQ human lung lymphocytes, THP-1 reconstituted with HAQ, AQ, and HAQ/SAVI, AQ/SAVI mice, to demonstrate that the common human HAQ, AQ alleles are resistant to STING cell death in vitro and in vivo. Additional human T cell line work does not add too much. 

Responses to Comment 3: This is possibly a misunderstanding. We use BMDM for the purpose of comparing STING signaling (TBK1, IRF3, NFκB, STING activation) by WT/SAVI, HAQ/SAVI, AQ/SAVI. Ideally, we would like to compare STING signaling in CD4 T cells from WT/SAVI to HAQ/SAVI, AQ/SAVI mice. However, WT/SAVI has no CD4 T cells. Here, we are making the assumption that the basic STING signaling (TBK1, IRF3, NFκB, STING activation) is conserved between T cells and macrophages. 

Responses to Comment 4: Reviewer 2 suggests looking for evidence of inflammation and STING activation in the lungs of HAQ/SAVI, AQ/SAVI. We would like to elaborate further. First, anti-inflammatory treatments, e.g. steroids, DMARDs, IVIG, Etanercept, rituximab, Nifedipine, amlodipine, et al., all failed in SAVI patients 11. Second, Figure S4 examined lung neutrophils and inflammatory monocyte infiltration. Interestingly, while AQ/SAVI mice had a better lung function than HAQ/SAVI mice (Figure 4D, 4E vs 4H, 4I), HAQ/SAVI and AQ/SAVI lungs had comparable neutrophils and inflammatory monocyte infiltration. Last, SAVI is classified as type I interferonopathy 11, but the lung diseases of SAVI are mainly independent of type I IFNs 23-26. The AQ allele suppresses SAVI in vivo.  Understanding the mechanisms by which AQ rescues SAVI can generate curative care for SAVI patients.  

Author response image 1.

(A-B). Flow cytometry of HAQ/SAVI, AQ/SAVI, WT/WT or WT/HAQ splenocytes treated with diABZI (100ng/ml) or DMXAA (20µg/ml) for 24hrs. Cell death was determined by PI staining. Data are representative of three independent experiments. Graphs represent the mean with error bars indication s.e.m. p values are determined by one-way ANOVA Tukey’s multiple comparison test. * p<0.05. n.s: not significant.

References.

(1)             Patel, S. et al. The Common R71H-G230A-R293Q Human TMEM173 Is a Null Allele. J Immunol 198, 776-787 (2017). 

(2)             Sebastian, M. et al. Obesity and STING1 genotype associate with 23-valent pneumococcal vaccination efficacy. JCI Insight 5 (2020). 

(3)             Mansouri, S. et al. MPYS Modulates Fatty Acid Metabolism and Immune Tolerance at Homeostasis Independent of Type I IFNs. J Immunol 209, 2114-2132 (2022). 

(4)             Sivick, K. E. et al. Comment on "The Common R71H-G230A-R293Q Human TMEM173 Is a Null Allele". J Immunol 198, 4183-4185 (2017). 

(5)             Gulen, M. F. et al. Signalling strength determines proapoptotic functions of STING. Nat Commun 8, 427 (2017). 

(6)             Kabelitz, D. et al. Signal strength of STING activation determines cytokine plasticity and cell death in human monocytes. Sci Rep 12, 17827 (2022). 

(7)             Murthy, A. M. V., Robinson, N. & Kumar, S. Crosstalk between cGAS-STING signaling and cell death. Cell Death Differ 27, 2989-3003 (2020). 

(8)             Kuhl, N. et al. STING agonism turns human T cells into interferon-producing cells but impedes their functionality. EMBO Rep 24, e55536 (2023). 

(9)             Li, C., Liu, J., Hou, W., Kang, R. & Tang, D. STING1 Promotes Ferroptosis Through MFN1/2-Dependent Mitochondrial Fusion. Front Cell Dev Biol 9, 698679 (2021). 

(10)         Song, C. et al. SHR1032, a novel STING agonist, stimulates anti-tumor immunity and directly induces AML apoptosis. Sci Rep 12, 8579 (2022). 

(11)         Liu, Y. et al. Activated STING in a vascular and pulmonary syndrome. N Engl J Med 371, 507-518 (2014). 

(12)         Jin, L. et al. Identification and characterization of a loss-of-function human MPYS variant. Genes Immun 12, 263-269 (2011). 

(13)         Yi, G. et al. Single nucleotide polymorphisms of human STING can affect innate immune response to cyclic dinucleotides. PLoS One 8, e77846 (2013). 

(14)         Patel, S. et al. Response to Comment on "The Common R71H-G230A-R293Q Human TMEM173 Is a Null Allele". J Immunol 198, 4185-4188 (2017). 

(15)         Gao, K. M. et al. Endothelial cell expression of a STING gain-of-function mutation initiates pulmonary lymphocytic infiltration. Cell Rep 43, 114114 (2024). 

(16)         Gao, K. M., Motwani, M., Tedder, T., Marshak-Rothstein, A. & Fitzgerald, K. A. Radioresistant cells initiate lymphocyte-dependent lung inflammation and IFNgammadependent mortality in STING gain-of-function mice. Proc Natl Acad Sci U S A 119, e2202327119 (2022). 

(17)         Monroe, K. M. et al. IFI16 DNA sensor is required for death of lymphoid CD4 T cells abortively infected with HIV. Science 343, 428-432 (2014). 

(18)         Doitsh, G. et al. Cell death by pyroptosis drives CD4 T-cell depletion in HIV-1 infection. Nature 505, 509-514 (2014). 

(19)         Jakobsen, M. R., Olagnier, D. & Hiscott, J. Innate immune sensing of HIV-1 infection. Curr Opin HIV AIDS 10, 96-102 (2015). 

(20)         Silvin, A. & Manel, N. Innate immune sensing of HIV infection. Curr Opin Immunol 32, 54-60 (2015). 

(21)         Altfeld, M. & Gale, M., Jr. Innate immunity against HIV-1 infection. Nat Immunol 16, 554-562 (2015). 

(22)         Krapp, C., Jonsson, K. & Jakobsen, M. R. STING dependent sensing - Does HIV actually care? Cytokine Growth Factor Rev 40, 68-76 (2018). 

(23)         Luksch, H. et al. STING-associated lung disease in mice relies on T cells but not type I interferon. J Allergy Clin Immunol 144, 254-266 e258 (2019). 

(24)         Stinson, W. A. et al. The IFN-gamma receptor promotes immune dysregulation and disease in STING gain-of-function mice. JCI Insight 7 (2022). 

(25)         Warner, J. D. et al. STING-associated vasculopathy develops independently of IRF3 in mice. J Exp Med 214, 3279-3292 (2017). 

(26)         Fremond, M. L. et al. Overview of STING-Associated Vasculopathy with Onset in Infancy (SAVI) Among 21 Patients. J Allergy Clin Immunol Pract 9, 803-818 e811 (2021).

In the slides and derivation in class, there was a minus sign here. You can use this convention as well, but then you have to interpret J differently (whether a positive J creates alignment or anti-alignment).

telkens 1,3 kg toevoegen

gemakkelijk

• Sampling all negatives at once drops win rates from 70.1% to 57.3%.
• Continuously updating the reference policy (πref) reduces win rates further to 45.8%, likely due to overfitting.
• Removing replay and inter-policy comparisons decreases performance by 6.5 and 2 points, respectively.

DITTO faces limitations such as biases in GPT evaluations, slower training speed compared to other methods, and unclear learning processes that may lead to forgetting previous knowledge.

New comparisons are made not only between the user examples and the latest model outputs but also between outputs from different stages of the model's training. This helps the model learn progressively and avoid overfitting.

Demonstrative alignment actively reduces the probability of non-expert responses, something that reinforcement learning with human feedback was not designed to do.

Instead of reward-training a model on a large population of examples to product comparisons, comparisons data is created using differences between LLM outputs and a single expert demonstrator with very large reward.

Alignment by demonstration essentially does better than all prevalent tuning techniques.

Since it involve s the LLM comparing demonstrations with its own intermediate outputs.

As opposed to a large number of examples required for tuning

Tuning for a small task can require hundres or thousands of examples

Because before tuning they are designed to handle anything in the world.

This method of aligned produces better results than previous methods even without tuning.

DITTO takes user examples and treats them as better than what the model generates. It uses these comparisons to help the model learn and improve.

not a direct comment but a thought: I find this sentence very interesting! this is the god trick problem with current visualization design, because implicit in visualization, data are 'masked' by the marks and channels that represent them -- the data are implicitly immaterial without being given form through the designer via visualization design. in your example diagram, the visualization emphasizes the process of generating data, visualization, and insight -- so, how may other forms of visualization also demonstrate process? there is work in provenance recording and visualization, but these often start with the data and record manipulation.

it's not immediately apparent that you are on the diagram. also, would adding a legend be antithetical to the rhetorical purpose of the diagram?

in prior chapters we get a lot more information about the visualisation and since this is the first hand-drawn and static visual, it would be nice to know more about the diagram up front. who drew it? what type of visualisation is this? for those not familiar, is this a standard vis?

snel bediend moeten worden.

Beschrijving

a enlever

geïnteresseerd

Heetwatertank

Melkdranken

Tot150 personen per dag te bedienen De "koffiehoeken" van grote bedrijven te verbeteren Een assortiment speciale lattes voor te bereiden Klassieke koffies rap klaar te maken.

te serveren.

telkens 1,3 kg toevoegen,

gemakkelijk

Why not?

SAUCE HOLLONDAISE

geïnteresseerd

Heetwatertank

Melkdranken

80 tot120 personen per dag te bedienen. Kwaliteitsvolle koffie te garanderen. Een ruime keuze te bieden met keuze uit 31 koffiespecialiteiten

ongeëvenaarde

Beschrijving

te maken.

telkens 1 kg koffie toevoegen.

gemakkelijk

eLife assessment

This valuable study reveals how a rhizobial effector protein cleaves and inhibits a key plant receptor for symbiosis signaling, while the host plant counters by phosphorylating the effector. The molecular evidence for the protein-protein interaction and modification is solid, though biological evidence directly linking effector cleavage to rhizobial infection is incomplete. With additional functional data, this work could have implications for understanding intricate plant-microbe dynamics during mutualistic interactions.

Reviewer #1 (Public Review):

Bacterial effectors that interfere with the inner molecular workings of eukaryotic host cells are of great biological significance across disciplines. On the one hand they help us to understand the molecular strategies that bacteria use to manipulate host cells. On the other hand they can be used as research tools to reveal molecular details of the intricate workings of the host machinery that is relevant for the interaction/defence/symbiosis with bacteria. The authors investigate the function and biological impact of a rhizobial effector that interacts with and modifies, and curiously is modified by, legume receptors essential for symbiosis. The molecular analysis revealed a bacterial effector that cleaves a plant symbiosis signaling receptor to inhibit signaling and the host counterplay by phosphorylation via a receptor kinase. These findings have potential implications beyond bacterial interactions with plants.

Bao and colleagues investigated how rhizobial effector proteins can regulate the legume root nodule symbiosis. A rhizobial effector is described to directly modify symbiosis-related signaling proteins, altering the outcome of the symbiosis. Overall, the paper presents findings that will have a wide appeal beyond its primary field.

Out of 15 identified effectors from Sinorhizobium fredii, they focus on the effector NopT, which exhibits proteolytic activity and may therefore cleave specific target proteins of the host plant. They focus on two Nod factor receptors of the legume Lotus japonicus, NFR1 and NFR5, both of which were previously found to be essential for the perception of rhizobial nod factor, and the induction of symbiotic responses such as bacterial infection thread formation in root hairs and root nodule development (Madsen et al., 2003, Nature; Tirichine et al., 2003; Nature). The authors present evidence for an interaction of NopT with NFR1 and NFR5. The paper aims to characterize the biochemical and functional consequences of these interactions and the phenotype that arises when the effector is mutated.

Evidence is presented that in vitro NopT can cleave NFR5 at its juxtamembrane region. NFR5 appears also to be cleaved in vivo. and NFR1 appears to inhibit the proteolytic activity of NopT by phosphorylating NopT. When NFR5 and NFR1 are ectopically over-expressed in leaves of the non-legume Nicotiana benthamiana, they induce cell death (Madsen et al., 2011, Plant Journal). Bao et al., found that this cell death response is inhibited by the coexpression of nopT. Mutation of nopT alters the outcome of rhizobial infection in L. japonicus. These conclusions are well supported by the data.

The authors present evidence supporting the interaction of NopT with NFR1 and NFR5. In particular, there is solid support for cleavage of NFR5 by NopT (Figure 3) and the identification of NopT phosphorylation sites that inhibit its proteolytic activity (Figure 4C). Cleavage of NFR5 upon expression in N. benthamiana (Figure 3A) requires appropriate controls (inactive mutant versions) that have been provided, since Agrobacterium as a closely rhizobia-related bacterium, might increase defense related proteolytic activity in the plant host cells.

Key results from N. benthamiana appear consistent with data from recombinant protein expression in bacteria. For the analysis in the host legume L. japonicus transgenic hairy roots were included. To demonstrate that the cleavage of NFR5 occurs during the interaction in plant cells the authors build largely on western blots. Regardless of whether Nicotiana leaf cells or Lotus root cells are used as the test platform, the Western blots indicate that only a small proportion of NFR5 is cleaved when co-expressed with nopT, and most of the NFR5 persists in its full-length form (Figures 3A-D). It is not quite clear how the authors explain the loss of NFR5 function (loss of cell death, impact on symbiosis), as a vast excess of the tested target remains intact. It is also not clear why a large proportion of NFR5 is unaffected by the proteolytic activity of NopT. This is particularly interesting in Nicotiana in the absence of Nod factor that could trigger NFR1 kinase activity.

It is also difficult to evaluate how the ratios of cleaved and full-length protein change when different versions of NopT are present without a quantification of band strengths normalized to loading controls (Figure 3C, 3D, 3F). The same is true for the blots supporting NFR1 phosphorylation of NopT (Figure 4A).

It is clear that mutation of nopT results in a quantitative infection phenotype. Nodule primordia and infection threads are still formed when L. japonicus plants are inoculated with ∆nopT mutant bacteria, but it is not clear if these primordia are infected or develop into fully functional nodules (Figure 5). A quantification of the ratio of infected and non-infected nodules and primordia would reveal whether NopT is only active at the transition from infection focus to thread or perhaps also later in the bacterial infection process of the developing root nodule.

Reviewer #2 (Public Review):

Summary:

This manuscript presents data demonstrating NopT's interaction with Nod Factor Receptors NFR1 and NFR5 and its impact on cell death inhibition and rhizobial infection. The identification of a truncated NopT variant in certain Sinorhizobium species adds an interesting dimension to the study. These data try to bridge the gaps between classical Nod-factor-dependent nodulation and T3SS NopT effector-dependent nodulation in legume-rhizobium symbiosis. Overall, the research provides interesting insights into the molecular mechanisms underlying symbiotic interactions between rhizobia and legumes.

Strengths:

The manuscript nicely demonstrates NopT's proteolytic cleavage of NFR5, regulated by NFR1 phosphorylation, promoting rhizobial infection in L. japonicus. Intriguingly, authors also identify a truncated NopT variant in certain Sinorhizobium species, maintaining NFR5 cleavage but lacking NFR1 interaction. These findings bridge the T3SS effector with the classical Nod-factor-dependent nodulation pathway, offering novel insights into symbiotic interactions.

Weaknesses:

(1) In the previous study, when transiently expressed NopT alone in Nicotiana tobacco plants, proteolytically active NopT elicited a rapid hypersensitive reaction. However, this phenotype was not observed when expressing the same NopT in Nicotiana benthamiana (Figure 1A). Conversely, cell death and a hypersensitive reaction were observed in Figure S8. This raises questions about the suitability of the exogenous expression system for studying NopT proteolysis specificity.

(2)NFR5 Loss-of-function mutants do not produce nodules in the presence of rhizobia in lotus roots, and overexpression of NFR1 and NFR5 produces spontaneous nodules. In this regard, if the direct proteolysis target of NopT is NFR5, one could expect the NGR234's infection will not be very successful because of the Native NopT's specific proteolysis function of NFR5 and NFR1. Conversely, in Figure 5, authors observed the different results.

(3) In Figure 6E, the model illustrates how NopT digests NFR5 to regulate rhizobia infection. However, it raises the question of whether it is reasonable for NGR234 to produce an effector that restricts its own colonization in host plants.

(4) The failure to generate stable transgenic plants expressing NopT in Lotus japonicus is surprising, considering the manuscript's claim that NopT specifically proteolyzes NFR5, a major player in the response to nodule symbiosis, without being essential for plant development.

Author response:

eLife assessment

This valuable study reveals how a rhizobial effector protein cleaves and inhibits a key plant receptor for symbiosis signaling, while the host plant counters by phosphorylating the effector. The molecular evidence for the protein-protein interaction and modification is solid, though biological evidence directly linking effector cleavage to rhizobial infection is incomplete. With additional functional data, this work could have implications for understanding intricate plant-microbe dynamics during mutualistic interactions.

Thank you for this helpful comment. In the revised manuscript version, we will be more prudent with directly linking cleavage of Nod factor receptors by NopT and rhizobial infection.

We plan to modify the Title, the One-Sentence Summary, Abstract, and Discussion regarding this point.

Public Reviews:

Reviewer #1 (Public Review):

Bacterial effectors that interfere with the inner molecular workings of eukaryotic host cells are of great biological significance across disciplines. On the one hand they help us to understand the molecular strategies that bacteria use to manipulate host cells. On the other hand they can be used as research tools to reveal molecular details of the intricate workings of the host machinery that is relevant for the interaction/defence/symbiosis with bacteria. The authors investigate the function and biological impact of a rhizobial effector that interacts with and modifies, and curiously is modified by, legume receptors essential for symbiosis. The molecular analysis revealed a bacterial effector that cleaves a plant symbiosis signaling receptor to inhibit signaling and the host counterplay by phosphorylation via a receptor kinase. These findings have potential implications beyond bacterial interactions with plants.

Thank you for highlighting the broad significance of rhizobial effectors in understanding legume-rhizobium interactions. We fully agree with your assessment and will emphasize these points in the revised Introduction and Discussion sections of our manuscript. Specifically, we will expand our Discussion regarding the potential impact of the NopT interaction with symbiotic receptor kinases on plant immune signaling and regarding the general significance of our work.

Bao and colleagues investigated how rhizobial effector proteins can regulate the legume root nodule symbiosis. A rhizobial effector is described to directly modify symbiosis-related signaling proteins, altering the outcome of the symbiosis. Overall, the paper presents findings that will have a wide appeal beyond its primary field.

Out of 15 identified effectors from Sinorhizobium fredii, they focus on the effector NopT, which exhibits proteolytic activity and may therefore cleave specific target proteins of the host plant. They focus on two Nod factor receptors of the legume Lotus japonicus, NFR1 and NFR5, both of which were previously found to be essential for the perception of rhizobial nod factor, and the induction of symbiotic responses such as bacterial infection thread formation in root hairs and root nodule development (Madsen et al., 2003, Nature; Tirichine et al., 2003; Nature). The authors present evidence for an interaction of NopT with NFR1 and NFR5. The paper aims to characterize the biochemical and functional consequences of these interactions and the phenotype that arises when the effector is mutated.

Thank you for your positive feedback on our manuscript. In the revised Introduction and Discussion sections, we plan to better emphasize the interdisciplinary significance of our work. We will show how the knowledge gained from our study can contribute to a better understanding of microbial interactions with eukaryotic hosts in general, which may have a stimulating effect on future research in various research areas such as pathogenesis and immunity.

To ensure that the readers can easily follow the rationale behind our experiments, we will improve the Results section and provide more detailed explanations of how NopT among 15 examined effectors was selected. Additionally, we will provide more background information on NopT and the roles of NFR1 and NFR5 in symbiotic signaling in the Introduction section. As suggested, we will include the references Madsen et al. (2003) and Tirichine et al. (2003) as well as additional references on rhizobial NopT proteins into our revised manuscript version.

Evidence is presented that in vitro NopT can cleave NFR5 at its juxtamembrane region. NFR5 appears also to be cleaved in vivo. and NFR1 appears to inhibit the proteolytic activity of NopT by phosphorylating NopT. When NFR5 and NFR1 are ectopically over-expressed in leaves of the non-legume Nicotiana benthamiana, they induce cell death (Madsen et al., 2011, Plant Journal). Bao et al., found that this cell death response is inhibited by the coexpression of nopT. Mutation of nopT alters the outcome of rhizobial infection in L. japonicus. These conclusions are well supported by the data.

We appreciate that you recognize the value of our data.

The authors present evidence supporting the interaction of NopT with NFR1 and NFR5. In particular, there is solid support for cleavage of NFR5 by NopT (Figure 3) and the identification of NopT phosphorylation sites that inhibit its proteolytic activity (Figure 4C). Cleavage of NFR5 upon expression in N. benthamiana (Figure 3A) requires appropriate controls (inactive mutant versions) that have been provided, since Agrobacterium as a closely rhizobia-related bacterium, might increase defense related proteolytic activity in the plant host cells.

Thank you for recognizing the use of an inactive NopT variant in Figure 3A. In fact, increased activity of plant proteases induced by Agrobacterium is an important point that should not be neglected. We plan to mention this aspect in our revised Discussion.

In the context of your comments, we are planning to make the following improvements to the manuscript:

(1) We will add a more detailed description of the experimental conditions under which the cleavage of NFR5 by NopT was observed in vitro and in vivo.

(2) We plan to provide more comprehensive data on the phosphorylation of NopT by NFR1, including phosphorylation assays and mass spectrometry results. These additional data support the proposed mechanism by which NFR1 inhibits the proteolytic activity of NopT.

(3) We will expand the Discussion on the cell death response induced by ectopic expression of NFR1 and NFR5 in Nicotiana benthamiana. We will include more details from Madsen et al. (2011) to contextualize our findings with published literature.

We believe these additions and clarifications will enhance the clarity and impact of our findings.

Key results from N. benthamiana appear consistent with data from recombinant protein expression in bacteria. For the analysis in the host legume L. japonicus transgenic hairy roots were included. To demonstrate that the cleavage of NFR5 occurs during the interaction in plant cells the authors build largely on western blots. Regardless of whether Nicotiana leaf cells or Lotus root cells are used as the test platform, the Western blots indicate that only a small proportion of NFR5 is cleaved when co-expressed with nopT, and most of the NFR5 persists in its full-length form (Figures 3A-D). It is not quite clear how the authors explain the loss of NFR5 function (loss of cell death, impact on symbiosis), as a vast excess of the tested target remains intact. It is also not clear why a large proportion of NFR5 is unaffected by the proteolytic activity of NopT. This is particularly interesting in Nicotiana in the absence of Nod factor that could trigger NFR1 kinase activity.

Thank you for your comments regarding the cleavage of NFR5 and its functional implications. In the revised version, we will change our manuscript taking into account the following considerations:

(1) We acknowledge that the Western blots indicate only a small proportion of NFR5 is cleaved when co-expressed with NopT. It is worth noting in this context that the proteins were expressed at high levels which likely do not reflect the natural situation in L. japonicus. Low production of cleaved NFR5 in our Western blots with transformed N. benthamiana or L. japonicus cells thus may simply reflect an experimental effect due to high NFR5 protein synthesis. We suggest that the presence of high amounts of intact NFR5 does not have a significant functional impact on plant responses (cell death in N. benthamiana, rhizobial infection of L. japonicus) whereas NFR5 cleavage (or formation of NFR5 cleavage products) may be crucial for the observation of the observed phenotypic changes. The fraction of cleaved NFR5, although small, may be sufficient to disrupt crucial signaling pathways, leading to observable phenotypic changes. We will address possible differences between experimental and natural protein levels in our revised Discussion.

(2) We studied in our work three biochemical aspects of NopT: (i) physical binding of NopT to NFR1 and NFR5 (ii) proteolytical cleavage of NFR5 by NopT and (iii) phosphorylation of NopT by NFR1. These three biochemical properties appear to influence each other. Phosphorylation of NopT by NFR1 appears to reduce its proteolytic activity, thereby counteracting NFR5 degradation by NopT (NFR5 homeostasis). Moreover, as NopT is a phosphorylation substrate for NFR1, NopT probably interferes with kinase mediated downstream responses of NFR1. Thus, NFR5 cleavage activity of NopT appears to be only one feature of NopT. We plan to mention these considerations in our revised Discussion.

It is also difficult to evaluate how the ratios of cleaved and full-length protein change when different versions of NopT are present without a quantification of band strengths normalized to loading controls (Figure 3C, 3D, 3F). The same is true for the blots supporting NFR1 phosphorylation of NopT (Figure 4A).

Thank you for pointing out this aspect. Following your recommendation, we will quantify the band intensities for cleaved and full-length NFR5 in the experiments with different versions of NopT. These values will be normalized to loading controls. Similarly, the Western blots supporting NFR1 phosphorylation of NopT will be quantified. The data for normalized band intensities will be included into the revised figures. The quantifications will provide a clearer understanding of how the ratios of cleaved to full-length proteins change with different NopT variants and also will provide information to which extent NopT is phosphorylated by NFR1.

It is clear that mutation of nopT results in a quantitative infection phenotype. Nodule primordia and infection threads are still formed when L. japonicus plants are inoculated with ∆nopT mutant bacteria, but it is not clear if these primordia are infected or develop into fully functional nodules (Figure 5). A quantification of the ratio of infected and non-infected nodules and primordia would reveal whether NopT is only active at the transition from infection focus to thread or perhaps also later in the bacterial infection process of the developing root nodule.

Thank you for pointing this out. In the revised version of our manuscript, we will provide data showing that there are no obvious differences in nodule formation in plants inoculated with ∆nopT and wild-type NGR234, respectively. However, quantification of infection threads containing our GFP-labeled rhizobia in primordia and nodules would be difficult to perform due to strong autofluorescence signals in these tissues. The main goal of our study was to identify and characterize the interaction between NopT and Nod factor receptors. We therefore believe that an in-depth analysis of the bacterial infection process at later symbiotic stages is out of the scope of the present work.

Reviewer #2 (Public Review):

Summary:

This manuscript presents data demonstrating NopT's interaction with Nod Factor Receptors NFR1 and NFR5 and its impact on cell death inhibition and rhizobial infection. The identification of a truncated NopT variant in certain Sinorhizobium species adds an interesting dimension to the study. These data try to bridge the gaps between classical Nod-factor-dependent nodulation and T3SS NopT effector-dependent nodulation in legume-rhizobium symbiosis. Overall, the research provides interesting insights into the molecular mechanisms underlying symbiotic interactions between rhizobia and legumes.

Strengths:

The manuscript nicely demonstrates NopT's proteolytic cleavage of NFR5, regulated by NFR1 phosphorylation, promoting rhizobial infection in L. japonicus. Intriguingly, authors also identify a truncated NopT variant in certain Sinorhizobium species, maintaining NFR5 cleavage but lacking NFR1 interaction. These findings bridge the T3SS effector with the classical Nod-factor-dependent nodulation pathway, offering novel insights into symbiotic interactions.

We appreciate that you recognize the value of our manuscript.

Weaknesses:

(1) In the previous study, when transiently expressed NopT alone in Nicotiana tobacco plants, proteolytically active NopT elicited a rapid hypersensitive reaction. However, this phenotype was not observed when expressing the same NopT in Nicotiana benthamiana (Figure 1A). Conversely, cell death and a hypersensitive reaction were observed in Figure S8. This raises questions about the suitability of the exogenous expression system for studying NopT proteolysis specificity.

We appreciate your attention to these plant-specific differences. In view of your comments, we plan to revise the Discussion and explain the different expression systems used for studying NopT effects in planta. Previous studies showed that NopT expressed in tobacco (N. tabacum) or in specific Arabidopsis thaliana ecotypes (with PBS1/RPS5 genes) causes rapid cell death (Dai et al. 2008; Khan et al. 2022). Our data shown in Fig. S8 confirm these findings. As cell death (effector triggered immunity) is usually associated with induction of protease activities, we considered N. tabacum and A. thaliana plants as not suitable for testing NFR5 cleavage by NopT. In fact, no NopT/NFR5 experiments were performed with these plants in our study. In contrast, the expression of NopT in Nicotiana benthamiana did not lead to cell death in our experiments. Khan et al. 2022 also reported that cell death does not occur in N. benthamiana unless the cells were transformed with PBS1/RPS5 constructs. Thus, N. benthamiana is a suitable expression system to analyze NopT protease activity on co-expressed substrates. Our revision aims to better understand the advantages of the N. benthamiana expression system for studying NopT mediated proteolysis of NFR5.

(2) NFR5 Loss-of-function mutants do not produce nodules in the presence of rhizobia in lotus roots, and overexpression of NFR1 and NFR5 produces spontaneous nodules. In this regard, if the direct proteolysis target of NopT is NFR5, one could expect the NGR234's infection will not be very successful because of the Native NopT's specific proteolysis function of NFR5 and NFR1. Conversely, in Figure 5, authors observed the different results.

Our inoculation experiments clearly show that NopT of NGR234 has a negative effect on formation of infection foci (Fig. 5A) and nodule primordia (Fig. 5E). Our biochemical analysis indicates that NopT targets the NFR1/NFR5 complex, which most likely impairs activation of downstream responses such as NIN gene expression. Accordingly, NIN promoter activity was found to be higher in roots inoculated with the Δ_nopT_ mutant as compared to the NGR234 wild-type (Fig. 5B and 5D). It is therefore plausible that NopT impairs rhizobial infection of L. japonicus due to inhibition of NFR1/NFR5 functions. We agree with this Reviewer that it can be expected that “NGR234's infection will not be very successful”. Fig. 5 confirms that Δ_nopT_ mutant is indeed a better symbiont and we do not think that we obtained “unexpectedly different results”. In the revised version, we will try to formulate our discussion text better in order to avoid any misunderstandings. Furthermore, will write as figure title “NopT dampens rhizobial infection…” instead of “NopT regulates rhizobial infection…”. We are also considering changing the title of our manuscript.  

(3) In Figure 6E, the model illustrates how NopT digests NFR5 to regulate rhizobia infection. However, it raises the question of whether it is reasonable for NGR234 to produce an effector that restricts its own colonization in host plants.

We acknowledge the potential paradox of NGR234 producing an effector that appears to restrict its own colonization in host plants. In fact, depending on the host plant, most rhizobial effectors are “double-edged swords” that play either a positive or negative role in the symbiosis. In response to your comment, we will discuss the possibility that NopT may confer selective advantages in interactions between NGR234 and host plants where NopT plays a positive symbiotic role (Dai et al. 2008; Kambara et al. 2009). Inhibition of NFR1/NFR5 functions by NopT in these host plants could be a feedback response in cells in which symbiotic signaling has already started. It is tempting speculate that the interaction between NopT and Nod factor receptors reduces Nod factor perception and downstream signaling to avoid a possible overreaction of symbiotic signaling, which may result in hypernodulation or formation of empty nodules without bacteria. Furthermore, it is tempting to speculate that NopT targets not only Nod factor receptors but also other host proteins to promote symbiosis, e.g. by suppressing excessive immune responses triggered by hyperinfection of rhizobia. In our revised manuscript, we will highlight the need for further investigations to elucidate the precise mechanisms underlying the observed infection phenotype and the role of NopT in modulating symbiotic signaling pathways.  

(4) The failure to generate stable transgenic plants expressing NopT in Lotus japonicus is surprising, considering the manuscript's claim that NopT specifically proteolyzes NFR5, a major player in the response to nodule symbiosis, without being essential for plant development.

Thank you for your comments. The failure to obtain L. japonicus plants constitutively expressing NopT was indeed surprising and suggests that NopT targets not only NFR5 but also other proteins in L. japonicus. The number of NopT substrates in plants could be greater than assumed. For example, we show in our work that NopT can cleave AtLYK5 and LjLYS11. In our manuscript, we don’t provide protocols and data on our efforts to construct L. japonicus plants stably expressing NopT. Indeed, it cannot be completely ruled out that the observed failure is not due to NopT expression, but rather to other factors that influence the transformation and regeneration of explants into whole plants. Our results should therefore not be over-interpreted. We consider a discussion of our failed transformation experiments to be somewhat preliminary and not central to this manuscript. herefore, we plan to modify our Discussion and delete the sentence reporting that stable transgenic plants expressing NopT have not been successfully generated.

geïnteresseerd

zij

BMW van de koffiemachine.

Zwitsere

Melkdranken

Heetwatertank

• Bedrijven te bedienen met 80 tot 120 medewerkers.
• Een nauwkeurige en uniforme maling te bieden
• Koffiespecialiteiten sne tel bereiden
• Een perfecte hygiëne te handhaven

Beschrijving

te maken.

telkens 1 kg koffie bijvullen

gemakkelijk

Author response:

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

We thank the reviewers for their overall careful evaluation of our work, the constructive criticism, and their many helpful suggestions. We feel that our revision built on the strengths identified by the reviewers, and addressed all the concerns they have raised. Both reviewers recognize that our revisions have improved the paper.  Since the first submission we have:

• Rewritten large parts of the papers to improve clarity and make it more concise where possible

• Simulated an alternative working memory model, as recommended by Reviewer 1

• Included 4 new/revised supplementary figures, following the reviewer’s suggestions for additional analysis.

Below we provide a brief response to the Reviewers’ comments on our manuscript revision.

Reviewer #1: Public Review:

Strengths:

Overall, the work offers a very interesting approach of a topic which is hard to accomplish experimentally --therefore the computational take is entirely justified and extremely useful. The authors carefully designed the computational experiments to shed light into the demyelination effects on working memory from multiple levels of description, increasing the reliability of their conclusions. I think this work provides now convincing evidence and has the potential to be influential in future studies of myelin alterations (and related disorders such as multiple sclerosis).

Weaknesses:

In its current form, the authors have improved the clarity of the results and the model details, and have provided a new set of simulations to complement and reinforce the original ones (including the development of a new spatial working memory model based on silent working memory principles). I do not appreciate any significant weaknesses at this point.

We thank the reviewer for these positive comments on our revision and for the suggestion of adding the silent memory model, as we feel this has strengthened our findings.

Reviewer #2: Public Review:

This paper analyzes the effect of axon de-myelination and re-myelination on action potential speed, and propagation failure. Next, the findings are then incorporated in a standard spiking ring attractor model of working memory.

I think the results are not very surprising or solid and there are issues with method and presentation.

The authors did many simulations with random parameters, then averaged the result, and found for instance that the Conduction Velocity drops in demyelination. It gives the reader little insight into what is really going on. My personal preference is for a well understood simple model rather than a poorly understood complex model. The link between the model outcome of WM and data remains qualitative and is further weakened by the existence of known other age-related effects in PFC circuits.

Comments on revised version:

The paper has improved in the revision, although I still think a reduced model would have been nice.

As noted above, in addition to our spiking bump attractor model, our revision includes a second network-level model:  an activity-silent working memory model for continuous features.  We found qualitatively similar effects as in our bump attractor network model, showing that our main conclusions do not critically depend on the exact working memory mechanism (active vs. activity-silent).  This new model was described in two new supplementary figures and a new paragraph in the Results section.

We did not add a reduced model in our revision to this paper, since neither reviewer explicitly recommended that we add one.  As we noted in our private response to reviewers that accompanied our revision: we share the view that understanding simple models can provide critical insights into brain function (and we believe that many of our papers related to attractor dynamics in working memory and decision-making fall into this category, e.g. Wimmer et al. 2014, Esnaola-Acebes et al. 2022, Ibañez et al 2020). We disagree with the reviewer on an important point: we feel that the model complexity that we have chosen is appropriate and necessary to study the phenomenon at hand. Our modeling efforts are principled, with complexity added as necessary. We started with a biophysical single neuron model with firing dynamics fit to empirical data in pyramidal neurons of rhesus monkey dlPFC (Rumbell et al. 2016) – the same type of neurons and cortical region analyzed in the Peters et al. work on structural changes to myelin seen during aging (e.g., Figure 1).  Because simple models do not accurately capture the CV along thin axons like those in the PFC, we attached a multicompartment axon with detailed myelinated segments, and constructed a cohort of feasible models. We then used this cohort to get quantitative estimates of the effects of variable degrees of demyelination and remyelination. This would not be possible with a simpler model. We then study the consequences of de- and re-myelination in a spiking neural network model. Again, we could not use a simpler model (e.g. a firing rate attractor model) without making gross assumptions about how demyelination affects circuit function. In sum, we believe that our models are relatively simple but comprehensive given the phenomenon that we are studying.

The reviewer is correct in that there exist “known other age-related effects in PFC circuits”. These are reviewed in the introduction and we discuss future extensions of our model that would incorporate those effects as well. It is important to note that this is the first comprehensive study of demyelination effects in aging PFC, demonstrating that myelin changes alone predict working memory changes associated with aging.

While we agree that averaging results about different parameter sets provide a limited understanding of the system, we persist in our belief that such analyses provide an important baseline.  We acknowledge that results vary across our model cohort; this is why we included the heatmaps of our single cell model perturbation results (Figure 3 and Supplementary Figure 3), and simulated network models representing a heterogeneity of neuronal axons with healthy and altered myelin sheaths in different degrees, as likely occurs in the aging brain (Figures 7 and 8).  The model framework we present here is well-suited for more targeted analyses and better insights, including those which we are pursuing currently.

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

We thank the reviewers for their careful evaluation of our work, the constructive criticism, and their many helpful suggestions. We feel that our revision builds on the strengths identified by the reviewers, and addresses all the concerns they have raised. We have:

• Rewritten large parts of the papers to improve clarity and make it more concise where possible

• Simulated an alternative working memory model

• Included 4 new/revised supplementary figures, following the reviewer’s suggestions for additional analysis

Reviewer #1 (Public Review):

Summary:

The authors study the effects of myelin alterations in working memory via the complementary use of two computational approaches: one based on the de- and re-myelination in multicompartmental models of pyramidal neurons, and one based on synaptic changes in a spiking bump attractor model for spatial working memory. The first model provides the most precise angle (biophysically speaking) of the different effects (loss of myelin lamella or segments, remyelination with thinner and shorter nodes, etc), while the second model allows to infer the consequences of myelin alterations in working memory performance, including memory stability, duration, and bump diffusion. The results indicate (i) a slowing down and failure of propagation of spikes with demyelination and partial recovery with remyelination, with detailed predictions on the role of nodes and myelina lamella, and (ii) a decrease in memory duration and an increase in memory drift as a function of the demyelination, in agreement with multiple experimental studies.

Strengths:

Overall, the work offers a very interesting approach of a topic which is hard to accomplish experimentally --therefore the computational take is entirely justified and extremely useful. The authors carefully designed the computational experiments to shed light into the demyelination effects on working memory from multiple levels of description, increasing the reliability of their conclusions. I think this work is solid and has the potential to be influential in future studies of myelin alterations (and related disorders such as multiple sclerosis).

We thank the reviewer for these positive comments on our manuscript.

Weaknesses:

In its current form, the study still presents several issues which prevent it from achieving a higher potential impact. These can be summarized in two main items. First, the manuscript is missing some important details about how demyelination and remyelination are incorporated in both models (and what is the connection between both implementations). For example, it is unclear whether an unperturbed axon and a fully remyelinated axon would be mathematically equivalent in the multicompartment model, or how the changes in the number of nodes, myelin lamella, etc, are implemented in the spiking neural network model.

We thank the reviewer for these suggestions to improve the clarity of our manuscript. A ‘fully remyelinated’ axon is not mathematically equivalent to the unperturbed axon: it has shorter and thinner myelinated segments, and additional nodes in between. This is consistent with empirical observations in rhesus monkey dlPFC, as reviewed in Peters et al. (2009): a 90% increase in paranode profiles, and myelin sheaths that were thinner than expected for the size of the enclosed axon. With no empirical observations of fewer numbers of nodes (but rather, the opposite) or bare sections of axon, we assumed that the remyelination process also creates new nodes (which are identical to existing nodes), as also modeled in Scurfield & Latimer (2018). We have added two new sentences to the results to clarify this fact, before presenting the first set of results for the single cell model: (starting at line 137):

“To simulate demyelination, we removed lamellae from selected myelinated segments; for remyelination we replaced a fraction of myelinated segments by two shorter and thinner segments with a node in between. As such, a ‘fully remyelinated axon’ had all the demyelinated segments subsequently remyelinated, but with fewer lamellae and additional nodes compared to the unperturbed control case, consistent with empirical observations (Peters, 2009).”

We also state the maximal amount of remyelination more explicitly in the Results, starting on lines 164-165: "We next examined the extent to which remyelination with shorter and thinner segments, occurring after demyelination, restored axonal AP propagation (Figure 4).”

Also on line 192-193: “Remyelinating all affected segments with 75% of lamellae (the maximal amount of remyelination) nearly eliminated AP failures (1.8 ± 1.1%).”

Finally, in Methods we also clarified the structure of the added node (starting at line 634): “Remyelination was performed by replacing an affected (previously demyelinated) segment with two shorter segments, each including paranodes, juxtaparanodes, and an internode, and a new node between them that was identical to existing nodes.”

We have also provided further details describing how myelin dystrophy was simulated in the network model in Results (lines 243 - 249) and in Methods (lines 722 - 747). How myelin alterations have been implemented in the network model is one of the questions of the reviewer (Question 5 in Reviewer #1: Recommendations for the Authors_)._ We have addressed this question by describing in detail how we adjusted CV and AP failure rate to the values produced by the multicompartment neuron model. Please see our answer to Question 5 for the details.

Second, it is unclear whether some of the conclusions are strong computational predictions or just a consequence of the model chosen. For example, the lack of effect of decreasing the conduction velocity on working memory performance could be due to the choice of considering a certain type of working memory model (continuous attractor), and therefore be absent under other valid assumptions (i.e. a silent working memory model, which has a higher dependence on temporal synaptic dynamics).

Whether some conclusions are strong predictions or just a consequence of the model chosen is an important concern and indeed a general problem of computational modeling of working memory. For example, Stein et al. (Stein et al. Towards biologically constrained attractor models of schizophrenia, Curr. Opin. Neurobiol. 2021) showed that opposed manipulations of E/I ratio can produce the same behavioral pattern in different alternative, plausible biological network models. As long as we do not fully understand the neural mechanisms underlying working memory, modeling studies of how alterations (e.g. in E/I ratio or in the reliability and timing of axonal transmission, as we did here) affect circuit function need to be interpreted critically and tested against new experimental data.

One way to strengthen model predictions is by showing that different computational models make similar predictions. To do this, we implemented an activity-silent working memory model for continuous features, as suggested by the reviewer, and we found qualitatively similar effects as in our bump attractor network model. Thus, our main conclusions do not critically depend on the exact working memory mechanism (active vs. activity-silent).

In the revised manuscript, we have added two new supplementary figures (Supplementary Figure 8 and 9, see the next page) and a new paragraph in the Results section about activity silent working memory (starting at line 319):

“Alternative working memory mechanisms. Working memory in our neural network is maintained in an attractor state with persistent neural activity (Compte et al., 2000; Hansel and Mato, 2013). Other mechanisms have been proposed, including that working memory maintenance may rely on activity-silent memory traces (Mongillo et al., 2008; Stokes, 2015; Barbosa et al., 2020). In activity-silent models, a slowly decaying transient of synaptic efficacy preserves information without the need for persistent ongoing activity. We implemented an activity-silent model, to our knowledge the first one for continuous spatial locations, and tested how working memory performance is affected by AP failures and propagation delays. We found that AP failures corresponding to demyelination caused working memory errors qualitatively similar to the delay-active network (Supplementary Figure 8). On the other hand, increasing propagation delays did not lead to additional working memory errors, unless we include unrealistically high values (uniform distribution in the range of 0 to 100 ms; Supplementary Figure 9). These results are qualitatively similar to the delay active network model. Thus, our main findings do not critically depend on the exact working memory mechanism (active vs. activity-silent).”

Author response image 1.

Action potential failures impair working memory performance in a network model with activity-silent memory traces. (A) Spiking and synaptic activity in an unperturbed, activity-silent working memory model. Top: Raster plot showing the activity for each excitatory neuron (labeled by its preferred direction) in a single trial with a cue stimulus presented at 180°. We modified our spiking neural network model such that it does not show elevated persistent firing throughout the delay period (see Figure 5B for comparison). In particular, we reduced the external background input to excitatory neurons by a factor of 3.61% and we increased the cue stimulus amplitude by 12.5%. Even though spiking activity decays to baseline (close to 0 Hz), a memory trace is imprinted in enhanced synaptic strength due to short-term synaptic facilitation (Mongillo et al., 2008). Selective spiking activity is recovered by a non-selective constant input applied during 300 ms to all excitatory neurons during the two reactivation periods (marked by yellow and green rectangles in the raster plot). The amplitude of the input was 11 mV during the first and 13 mV during the second reactivation period. Reactivation periods are marked in light gray shading in the remaining panels below and the cue period is indicated by dark gray shading. Firing rates (second row), synaptic facilitation variable u (third row), and synaptic depression variable x (bottom row) for the same trial, averaged for 500 neurons around the neuron with 180° as preferred direction (solid lines) and around the neuron with 0° as preferred direction (dashed lines). Note that reactivation recovers the activity bump (C) but also causes elevated firing and subsequent enhancement of synapses at all positions in the networks. (B) Activity in a network with demyelination of 50% of the myelinated segments by removing 60% of the myelin lamellae. AP failures lead to reduced firing rates in the cue and early delay periods and consequently to weaker synaptic enhancement. (C) Average spike counts of the excitatory neurons during the cue period (black lines), and the two reactivation periods indicated in the raster plots in A and B (yellow and green lines). Solid lines correspond to the control network and dashed lines to the perturbed network. (D) Memory strength as a function of time for the control and perturbed networks. (E-F) Trajectories of the bump center (i.e., remembered cue location) read out from the neural activity across the cue and delay periods using a population vector (see Methods). Cue position was 180° in all trials. The perturbed network (F) shows larger working memory errors towards the end of the delay period compared to the control network (E).

Author response image 2.

Effect of propagation delays on control and perturbed activity-silent network models. (A) Memory strength during the whole simulation time for the young, control networks relying on activity-silent working memory (Supplementary Figure 8) with zero propagation delays (blue line), and with propagation delays from a uniform distribution with a range between 0 and 40 ms (yellow line) and between 0 and 100 ms (orange line). (B) Memory strength for perturbed networks when demyelinating 25% of the myelinated segments by removing 50% of the myelin lamellae, without delays (red line), and with uniformly distributed delays between 0 and 40 ms (light gray line) and between 0 and 100 ms (black line). The cue period is indicated by dark gray shading and reactivation periods are marked in light gray. Memory strength was calculated by averaging across 280 trials for one network. Shaded areas indicate SEM for each case. For the young, control networks (A), working memory was not affected by including delays of up to 40 ms. Unrealistically long delays ranging up to 100 ms did cause an impairment (the longest delays found for the most extreme perturbation condition – demyelination of 75% of the segments by removing 100% of the myelin lamellae – were of 49.9 ms on average). When also incorporating AP failures to the networks (B), we observed a similar trend. For this perturbation condition, delays of up to 40 ms were already much larger than the delays quantified in the single neuron model (for the case of 25% of the segments demyelinated by removing 50% of the myelin lamellae, the average delay in the cohort was 3.75 ms).

With additional simulations to address these issues, I consider that the present study would become a convincing milestone in the computational modeling of myelin-related models, and an important study in the field of working memory.

Again, we would like to thank the reviewer for the positive comments. We have addressed all the main issues raised (see below our response to the “recommendations for the authors”).

Reviewer #2 (Public Review):

This paper analyzes the effect of axon de-myelination and re-myelination on action potential speed, and propagation failure. Next, the findings are then incorporated in a standard spiking ring attractor model of working memory.

I think the results are not very surprising or solid and there are issues with method and presentation.

The authors did many simulations with random parameters, then averaged the result, and found for instance that the Conduction Velocity drops in demyelination. It gives the reader little insight into what is really going on. My personal preference is for a well understood simple model rather than a poorly understood complex model. The link between the model outcome of WM and data remains qualitative, and is further weakened by the existence of known other age-related effects in PFC circuits.

We thank the reviewer for the critical assessment of our work. We share the view that understanding simple models can provide critical insights into brain function (and we believe that many of our papers related to attractor dynamics in working memory and decision making fall into this category, e.g. Wimmer et al. 2014, Esnaola-Acebes et al. 2022, Ibañez et al 2020). However, we respectfully disagree with the reviewer on an important point: the model complexity that we have chosen is appropriate and necessary to study the phenomenon at hand. Our modeling efforts are principled, with complexity added as necessary. We started with a biophysical single neuron model with firing dynamics fit to empirical data in pyramidal neurons of rhesus monkey dlPFC (Rumbell et al. 2016) – the same type of neurons and cortical region analyzed in the Peters et al. work on structural changes to myelin seen during aging (e.g., Figure 1). Because simple models do not accurately capture the CV along thin axons like those in the PFC, we attached a multicompartment axon with detailed myelinated segments, and constructed a cohort of feasible models. We then used this cohort to get quantitative estimates of the effects of variable degrees of demyelination and remyelination. This would not be possible with a simpler model. We then study the consequences of de- and re-myelination in a spiking neural network model. Again, we could not use a simpler model (e.g. a firing rate attractor model) without making gross assumptions about how demyelination affects circuit function. In sum, we believe that our models are relatively simple but comprehensive given the phenomenon that we are studying.

The reviewer is correct in that there exist “known other age-related effects in PFC circuits”. These are reviewed in the introduction and we discuss future extensions of our model that would incorporate those effects as well. It is important to note that this is the first comprehensive study of demyelination effects in aging PFC, demonstrating that myelin changes alone predict working memory changes associated with aging.

The specific issues about modeling choices and interpretation of the results are discussed below.

Both for the de/re myelination the spatial patterns are fully random. Why is this justified?

We agree that myelin dystrophy during aging could be non-random, that is, localized to certain regions of an axon. Our collaborators (Drs Jennifer Luebke, Maya Medalla, and Patrick Hof) are currently addressing this question using 3D electron microscopy and immunohistochemistry on axons of individual neurons and their associated myelin, but results are not available yet. Early on in this study we examined how the location of myelin alterations affected AP propagation. Focusing demyelination along a section of axon led to more AP slowing and failure than when spatially randomized. Likewise, remyelination of such spatially localized dystrophy led to greater recovery, as there were fewer transitions between long and short internodes (Supplemental Figure 4). Since otherwise the effects in the localized cases were largely similar to those in the spatially random case (see Author response image 3 below), for brevity in this paper we assumed myelin alterations were randomly distributed. Our next paper, extending this study to collateralized axons and which was presented as a poster at the 2023 Society for Neuroscience meeting, will include an examination of localized myelin dystrophy.

Author response image 3.

Effect of localized myelin alterations on CV change. Myelin alterations were either focused on the third of myelinated segments closest to the initial segment (‘proximally clustered’), the third of myelinated segments furthest from the initial segment (‘distally clustered’), or distributed according to a uniform distribution as in the current study. For demyelination, all lamellae were removed from 25% of myelinated segments (showing mean +/- SEM of all 50 cohort models, 30 randomized trials each). For remyelination, affected segments were replaced by two shorter segments with 75% of the original lamellae thickness and a node in between.

We have added two sentences in Methods to justify this assumption more clearly (line 510): “Evidence suggests that aging affects oligodendrocytes in several ways, including the ability for oligodendrocyte precursor cells to mature (Dimovasili et al., 2022). Knowing that individual oligodendrocytes myelinate axons of many different neurons, but without data quantifying how oligodendrocyte dystrophy affects myelination in individual axons, we assumed that myelin alterations were randomly distributed.”

We have also added a sentence in the Discussion alluding to our upcoming study (line 434): “Our model can also be extended to explore interactions between spatially localized myelin perturbations (such as those seen in multiple sclerosis) and axon collateralization (Sengupta et al., 2023), which would affect the distance-dependence of AP failures.”

Similarly, to model the myelin parameters were drawn from uniform distributions, Table 1 (I guess). Again, why is this reasonable?

The reviewer is correct that our initial Latin hypercube sample generated a uniform distribution. However, parameters of the random sample of models selected as biologically feasible were not uniformly distributed. We have added a new figure (Supplementary Figure 1A) to illustrate the parameter distributions, and have added two sentences in Methods (starting on line 596):

“Of the 1600 simulated models, 138 met these criteria; for the present study, we randomly selected 50 models to comprise the young, control model cohort. Along most dimensions, the chosen cohort was approximately normally distributed (Supplementary Figure 1). The g-ratio (ratio of axon to fiber diameter) among models in the cohort was 0.71 ± 0.02, with total axon lengths of 1.2 ± 0.1 cm.”

Author response image 4.

Distribution of parameters and conduction velocities in the single neuron model cohort. (A) Histograms of axon morphology parameters of models selected for the single neuron cohort. Top: axon diameter: middle, length of unperturbed myelin segments; bottom: total myelin thickness in unperturbed segments, computed as the product of lamella thickness and number of lamellae. (B) Histograms of the CV for the 50 axons of the unperturbed model cohort (top), and representative demyelination and remyelination perturbations: mild demyelination (removing 25% of lamellae from 25% of the myelinated segments, second row); severe demyelination (removing all lamellae from 75% of the myelinated segments, third row); and complete (100%) remyelination (where the demyelinated segments from the third row were remyelinated by two shorter segments with 75% of lamellae). CVs averaged over 30 trials in each case. (C) Changes in CV (measured in %) in response to demyelination and remyelination versus the magnitude of current clamp step (+180, +280, or +380 pA). Shown are mean +/- SEM for demyelinating 50% of myelinated segments (removing all lamellae), and subsequent remyelination of those segments by shorter segments with 75% of lamellae.

The focus of most analysis is on the conduction velocity but in the end, this has no effect on WM, so the discussion of CV remains sterile.

CV delays likely do affect brain functions that rely on neuronal oscillations and synchrony, as mentioned in the Discussion. As such, we feel that our single neuron model results on CV delays as well as AP failures are valuable for the scientific community. Yet, given the results of our network models here, the reviewer has a valid point. We have clarified in the introduction that AP failures but not CV delays affected the network output (line 115):

“Higher degrees of demyelination led to slower propagation and eventual failure of APs along the axons of the multicompartment models. In the network models, an increase in AP failure rate resulted in progressive working memory impairment, whereas slower conduction velocities, in the range observed in the multicompartment models, had a negligible effect.”

We have also revised the single neuron section of the Results throughout, to better highlight the effects of myelin dystrophy on AP failures. Revisions to address this in the demyelination section start on line 148:

“AP propagation was progressively impaired as demyelination increased (Figure 3): CV became slower, eventually leading to AP failure. Removing 25% of lamellae had a negligible effect on CV, regardless of how many segments were affected. However, when all lamellae were removed, CV slowed drastically – by 38 ± 10% even when just 25% of the segments were demyelinated in this way, and 35 ± 13% of APs failed. When 75% of segments lost all their lamellae, CV slowed by 72 ± 8% and 45 ± 13% of APs failed.”

Similiarly, we have added several sentences about AP failures that remain after remyelination of the single neuron model (starting on line 190):

“Results for the percentage of AP failures (Figure 4C,F) were consistent with those for CV recovery. Remyelinating all previously demyelinated segments, even adding just 10% of lamellae, brought AP failure rates down to 14.6 ± 5.1%. Remyelinating all affected segments with 75% of lamellae (the maximal amount of remyelination) nearly eliminated AP failures (1.8 ± 1.1%). Incomplete remyelination, where some segments were still demyelinated, still had relatively high AP failure rates. For example, when one eighth of segments were remyelinated with the maximal amount of lamellae and one eighth were left bare, 25.7 ± 11.5% of APs failed across the cohort (Figure 4C, red dashed line and arrow). AP failure rates were slightly lower when starting with partial demyelination: 10.6 ± 7.6% of APs failed in the analogous paradigm (Figure 4F, red dashed line and arrow). In short: combinations of demyelinated and remyelinated segments often led to sizable CV delays and AP failures.”

The more important effect of de/re myelination is on failure. However, the failure is, AFAIK, just characterized by a constant current injection of 380pA. From Fig 2 it seems however that the first spike is particularly susceptible to failure. In other words, it has not been justified that it is fine to use the failure rates from this artificial protocol in the I&F model. I would expect the temporal current trace to affect whether the propagation fails or not.

In general, we did not find the first spike to be more susceptible to failure than latter spikes; the trace in Figure 2 is a representative snapshot intended to illustrate CV slowdown, AP failure, and recovery. Regarding the constant current injection: while the reviewer is correct that neurons do not receive such inputs in vivo, the applied current injections were designed to match in vitro current clamp protocols for these rhesus monkey neurons. While our future studies will include responses to more realistic synaptic inputs, we focused on somatic current injections here. We have added a new panel (C) to Supplementary Figure 1 (see previous response above) showing that the current step magnitude had little effect on the CV change after myelin perturbations; there was little effect on AP failure rates too. We now also state this finding more explicitly in Methods (starting on line 561):

“As done during in vitro electrophysiological experiments (Chang et al., 2005; Ibanez et al., 2020) and past modeling studies (Coskren et al., 2015; Rumbell et al., 2016), we first applied a holding current to stabilize the somatic membrane potential at -70 mV, then injected a current step into the somatic compartment for 2 seconds. …The CV changes in response to myelin alterations were relatively insensitive to variations in the magnitude of suprathreshold somatic current steps (Supplementary Figure 1C), and whether the current was constant or included Gaussian noise. Therefore, here we quantified CV changes and AP failures from responses to constant +380 pA current steps only.”

I don't know if there are many axon-collaterals in the WM circuits and or distance dependence in the connectivity, but if so, then the current implementation of failure would be questionable.

We agree that axon collaterals may affect our results; our unpublished morphological analyses of individual neuron axons indicate that there is a high degree of local axon collateralization in Layer 3 pyramidal neurons in LPFC. In this first study from our group on myelin perturbations, we chose to focus here on unbranched axons. There was some distance dependence of AP failure along the length of the axon. For example, in our most extreme demyelination case (75% of segments losing all their lamellae), about 14% of the axons showed more AP failure at their distal ends relative to the middle (mean difference 6.33%). We are examining this distance dependence more broadly in our next study, now cited in the Discussion (line 434): “Our model can also be extended to explore interactions between spatially localized myelin perturbations (such as those seen in multiple sclerosis) and axon collateralization (Sengupta et al., 2023), which would affect the distance-dependence of AP failures.”

I would also advise against thresholding at 75% failure in Fig3C. Why don't the authors not simply plot the failure rate?

We thank the reviewer for this suggestion, and have made this change. As suggested by the reviewer, we now show the AP failure rate in Figure 3 and Figure 4. The trends shown are nearly identical to those from the high failure trials.

Regarding the presentation, there are a number of dead-end results that are not used further on. The paper is rather extensive, and it would be clearer if written up in half the space. In addition, much information is really supplementary. The issue of the CV I already mentioned, also the Lasso regression for instance remains unused.

We understand the reviewer’s perspective, and we do value brevity when possible. During the revision process we examined the paper carefully, and made things more concise when it was feasible. As mentioned above, reporting CV results is important, though these revisions increased emphasis on results for AP failures in our revision. We combined the two Supplementary Figures about remyelination in the single neuron model into one (Supplementary Figure 3). We also moved the Lasso figure and associated methods to the Supplementary Material (Supplementary Figure 2), and have separated the Lasso results for demyelination and remyelination into their respective paragraphs (lines 154-160 and lines 200-204 respectively). While we do not use the Lasso explicitly later in Results, we cite them in the Discussion when comparing our findings to previous work (starting on line 417):

“Since our single neuron cohort sampled a wide range of parameter space, we used Lasso regression to identify which of the complex, interacting parameters contributed most to CV delays (which preceded AP failures). Parameters including axon diameter, node length, length of myelinated segments, and nodal ion channel densities predicted how our models responded to demyelination and remyelination; these findings are consistent with past modeling studies over more limited parameter ranges (e.g., Goldman and Albus, 1968; Moore et al., 1978; Babbs and Shi, 2013; Young et al., 2013; Schmidt and Knösche, 2019).”

We hope that our revision has struck an appropriate balance between clear and concise writing, and addressing concerns from both reviewers. We greatly value the time you have given to help us to improve our manuscript.

Response to Recommendations for the Authors:

Reviewer #1 (Recommendations for the Authors):

As I mentioned above, I consider that this study is well designed and it offers very interesting results. I have detailed below some of the issues that should be addressed to improve its potential impact in the field:

(1) Across the manuscript, it is not entirely clear how the results of the multicompartmental model compare to existing modeling results on demyelination and CV changes (such as in the papers cited by the authors). Is this section confirming previous results with a new (more accurate) computational model, or are there any new insights previously unreported? A new paragraph in the Discussion putting these results in context would be very useful for the reader.

We thank the reviewer for this suggestion. We have added two new subheadings to organize the Discussion better, and have expanded the single neuron section to three paragraphs. We feel this now clarifies how our model fits in with previous work while stating its novelty more explicitly. Starting on line 391:

“Myelin changes affect AP propagation in a cohort of model neurons

The novelty of our neuron model lies in its systematic exploration of a combination of different myelin perturbation types known to occur in myelin dystrophies, across a wide range of biologically feasible models. Our single neuron model assumed that age-related myelin dystrophies (e.g., Figure 1) alter the insulative properties of lamellae analogously to demyelination, and examined interactions between demyelination and remyelination. Past studies of myelin dystrophy examined how either demyelination or remyelination of all segments affected AP propagation for a few representative axon morphologies. For example, Scurfield and Latimer (2018) explored how remyelination affected CV delays, finding that axons with more transitions between long and short myelinated segments had slower CV (Supplementary Figure 4), and was first to explore how remyelination interacts with tight junctions. However, their study did not couple remyelination and demyelination together or examine AP failures. Other basic findings from our single neuron cohort are consistent with past modeling studies, including that demyelination caused CV slowing and eventual AP failures (Stephanova et al., 2005; Stephanova and Daskalova, 2008; Naud and Longtin, 2019), and, separately, that remyelination with shorter and thinner myelinated segments led to CV slowing (Lasiene et al., 2008; Powers et al., 2012; Scurfield and Latimer, 2018). However, by assuming that some previously demyelinated segments were remyelinated while others were not, we found that models could have much higher AP failure rates than previously reported. Such a scenario, in which individual axons have some segments that are normal, some demyelinated, and some remyelinated, is likely to occur. We also found a few neurons in our cohort showing a CV increase after remyelination, which has not generally been reported before and is likely due to an interplay between ion channels in the new nodes and altered electrotonic lengths in the perturbed myelinated segments (e.g., Waxman, 1978; Naud and Longtin, 2019).

Since our single neuron cohort sampled a wide range of parameter space, we used Lasso regression to identify which of the complex, interacting parameters contributed most to CV delays (which preceded AP failures). Parameters including axon diameter, node length, length of myelinated segments, and nodal ion channel densities predicted how our models responded to demyelination and remyelination; these findings are consistent with past modeling studies over more limited parameter ranges (e.g., Goldman and Albus, 1968; Moore et al., 1978; Babbs and Shi, 2013; Young et al., 2013; Schmidt and Knösche, 2019). Better empirical measurements of these parameters in monkey dlPFC, for example from 3-dimensional electron microscopy studies or single neuron axon studies combined with markers for myelin, would help predict the extent to which myelin dystrophy and remyelination along individual axons with aging affect AP propagation.

Another important feature of our multicompartment model is that it was constrained by morphologic and physiological data in rhesus monkey dlPFC —an extremely valuable dataset from an animal model with many similarities to humans (Upright and Baxter, 2021; Tarantal et al., 2022). While beyond the scope of the current study, this computational infrastructure –with a detailed axon, initial segment, soma, and apical and basal dendrites– enables simultaneous investigations of signal propagation through the dendritic arbor and axon. Our model can also be extended to explore interactions between spatially localized myelin perturbations (such as those seen in multiple sclerosis) and axon collateralization (Sengupta et al., 2023), which would affect the distance-dependence of AP failures. Integrating such results from single neuron models into network models of working memory, as we have done here, is a powerful way to connect empirical data across multiple scales.”

(2) Although the authors provide a well-designed study for the multi-compartmental model, it would be useful to add more details about how an unperturbed model and a completely remyelinated model differ in practice, perhaps right before the first results on the single cell model are presented. Are the new myelin sheaths covering the same % of axon as in the original case? Are there the same number of nodes? It is hard to distinguish which of these results are due to a compensation by the new myelin sheaths and which ones are just the model coming back to its original (and mathematically equivalent) starting point.

A ‘fully remyelinated’ axon is not mathematically equivalent to the unperturbed axon. Newly remyelinated segments had at most 75% of the original number of myelin wraps, with a new node in between, consistent with empirical observations in rhesus monkey dlPFC. Our manuscript changes in response to this recommendation are described in detail above in our response to the public review of the same reviewer.

(3) The authors observe a directed component in the bias that is known to be caused by heterogeneities in network connectivity, as stated in the text. It occurs to me that similar effects could be also caused by an heterogeneous demyelination in parts of the network. Inducing these biases could be another potential effect of demyelination in practice, and could be easily revealed by the author's current model (and displayed in a supplementary figure).

As suggested by the reviewer, we have tested heterogeneous demyelination in parts of the network and the results confirm the reviewer’s intuition. We have included these new results as new Supplementary Figure 7 (see below) and we have added the following sentences in the Legend of Figure 5, line 1265: “When demyelination is restricted to a part of the network, diffusion only increases in the perturbed zone (Supplementary Figure 7).” and in the Discussion (line 457): “In addition to age-related changes in memory duration and precision, our network model predicts an age-related increase in systematic errors (bias) due to an increased drift of the activity bump (Supplementary Figure 11). Moreover, if demyelination is spatially localized in a part of the network, the model predicts a repulsive bias away from the memories encoded in the affected zone (Supplementary Figure 7).”

Author response image 5.

Effect of spatially heterogeneous demyelination of the model neurons according to their preferred angle. We also tested working memory performance in the network when demyelination affects only parts of the network. The figure shows the decoded bump center position during the cue and delay period for the eight possible cue directions when a fraction of neurons was perturbed and the rest of the neurons in the circuit were unaltered (Figure 5B). We perturbed 10% of the neurons around the neuron with preferred direction 90° (left panel), 25% of the neurons around -90° (middle panel), and 50% of the neurons around 180° (right panel). Bump traces for cues that lie inside the perturbed portion of the circuit are shown in blue. Network perturbation in the three cases consisted in demyelinating 25% of the segments along the axons of model neurons, by removing 70% of the myelin lamellae. In each case, 280 trials were simulated for one network. These simulations show an increased drift and diffusion inside the perturbed zone, consistent with the increased drift and diffusion when perturbing the entire network (Figure 6B and Supplementary Figure 11). In particular, spatially heterogeneous demyelination in our network leads to a bias away from the affected zone and to increased trial-to-trial variability. Note that this is a model prediction, but we are not aware of empirical data showing heterogeneous demyelination with aging. Further, note that while our network model has a topological ring structure, neurons in PFC are not anatomically arranged depending on their preferred features. Thus, spatially heterogeneous demyelination would likely affect neurons with different feature preferences (i.e., neurons throughout our ring model).

(4) The bump attractor model of WM relies on a continuous attractor dynamics to encode the information stored in memory --a fixed point dynamics that can only vary via the slow noise-driven drift. This means, as the authors mention, that changes in CV won't affect the performance of WM in their model. This seems to be a limitation of the model, or at least an effect which is highly dependent on the modeler's choice, rather than an accurate prediction. While testing the effects of oscillations (as the authors argue in the Discussion) might be out of the scope of this work, there are other WM models which are more sensitive to temporal differences in activity. The authors should test whether the same (lack of) effects are also found in other WM models. A silent WM model seems to be the ideal candidate for this, as the authors already have the key dynamics of that model incorporated in their computational framework (namely, short-term synaptic facilitation in excitatory synapses).

We fully agree that considering the effects of demyelination in networks with alternative mechanisms would strengthen our manuscript. As suggested by the reviewer, we have simulated demyelination effects (AP failures and changes in CV) in an activity silent working memory model. The results are described in detail above in our response to the public review of the same reviewer.

We also would like to mention that we have now also tested larger conduction delays in the bump attractor model, revealing additional working memory errors. This is shown in the revised version of Supplementary Figure 6 (see below). However, those delays are unrealistically large and thus the main effect in both the bump attractor and the activity-silent model is due to AP failures.

Author response image 6.

Effect of propagation delays on control and perturbed networks. (A) Memory strength (left panels) and diffusion (right panels) for the young, control networks with zero propagation delays (blue solid line), as in Figure 5, and with propagation delays from a uniform distribution with a range between 0 and 100 ms (yellow dashed line). (B) Memory strength and diffusion for perturbed networks when demyelinating 50% of the segments along the axons of model neurons, by removing 60% of the myelin lamellae without delays (red solid line), and with delays from a uniform distribution with a range between 0 and 40 ms (gray dashed line) and between 0 and 85 ms (black dash-dotted line). The measures of working memory performance were calculated by averaging across 20 networks and 280 trials for each network. Shaded areas indicate SEM for each case. For the young, control networks, there was no difference with and without propagation delays, even though the delays used in the network simulations were much larger than the delays quantified in the single neuron model (the longest delays found for the most extreme perturbation condition –demyelination of 75% of the segments by removing 100% of the myelin lamellae– were of 49.9 ms on average; A). Working memory performance was also unaffected in the perturbed network with AP failures for delays ranging between 0 and 40 ms, also larger than the ones quantified in the single neuron model (for the case of 50% of the segments demyelinated by removing 60% of the myelin lamellae, the average delay in the cohort was 4.6 ms and the maximum delay was 15.7 ms; B). However, including extremely long delays of up to 85 ms did further impair memory compared to the impairment level introduced by AP failures alone (B).

(5) Impact of demyelination and remyelination on working memory: Could the authors explain here how these biologically detailed alterations are implemented in the bump attractor model? Is the CV and AP failure rate adjusted to the values produced by the multicompartment neuron model with these myelin alterations?

Yes, the reviewer is right, the CV and AP failure rate have been adjusted to the values produced by the multicompartment neuron model. To clarify this in the manuscript, we have restated the text as follows:

Lines 243 - 249 (Results):

To investigate how myelin alterations affect working memory maintenance, we explored in the network model the same demyelination and remyelination conditions as we did in the single neuron model. Because our network model consists of point neurons (i.e., without detailed axons), we incorporated CV slowing as an effective increase in synaptic transmission delays (see Methods). To simulate AP failures, we adjusted the AP failure rate to the values given by the single neuron model, by creating a probabilistic model of spike transmission from the excitatory presynaptic neurons to both the excitatory and inhibitory postsynaptic neurons (see Methods).

Lines 722 - 747 (Methods):

Modeling action potential propagation failures in the network. The network model is composed of point neurons without an explicit model of the axon. To effectively model the action potential failures at the distal end of the axons quantified with the single neuron model under the different demyelination and remyelination conditions, the AP failure rate was adjusted to the values produced by the single neuron model. To do this, we perturbed the 10 control networks by designing a probabilistic model of spike transmission from the excitatory presynaptic neurons to both the excitatory and inhibitory postsynaptic neurons. From the single neuron model, for each demyelination/remyelination condition, we quantified the probability of AP failure for each of the neurons in the control cohort, as well as the percentage of those neurons that shared the same probabilities of failure. That is, the percentage of neurons that had probability of failure = 0, probability of failure = 1 or any other probability. Then, we computed the probability of transmission, , and we specified for the corresponding percentages of excitatory neurons in the networks. Thus, in the network model, we took into account the heterogeneity observed in the single neuron model under each demyelination/remyelination condition.

Modeling conduction velocity slowing in the network. To explore the effect of CV slowing along the axons of model neurons, we simulated 20 young, control networks and 20 perturbed networks with AP failure rates adjusted for the case of single model neurons with 50% of the segments demyelinated along the axons by removing 60% of the myelin lamellae (we ran 280 trials for each network). Then, we added random delays uniformly distributed with a minimum value of 0 ms in both cases, a maximum value of 100 ms in the control networks, and a maximum values of 40 ms and 85 ms in the perturbed networks, in both the AMPA and NMDA excitatory connections to both E and I neurons (Supplementary Figure 6). These large values were chosen because we wanted to illustrate the potential effect of CV slowing in our network and smaller, more realistic, values did not have any effect.

(6) "We also sought to reveal the effect on working memory performance of more biologically realistic network models with AP transmission probabilities matched to both axons with intact and with altered myelin sheaths, as likely occurs in the aging brain (Figure 1). Thus, we ran network model simulations combining AP failure probabilities corresponding to groups of neurons containing intact axons and axons presenting different degrees of demyelination." I fail to see the difference with respect to the results in previous sections. Is it that now we have subnetworks in which axons are intact and subnetworks with significant AP failures, while before there was no topological separation between both cases? Please clarify.

In Figures 5 and 6 the AP failure rate of the neural population in the network simulations was matched to the AP failure rate of the cohort of single model neurons for each demyelination/remyelination condition. Since not all model neurons have equal features, a given condition produces different levels of impairment in its neuron. Thus, we quantified the probability of AP failure for each neuron in the control cohort, as well as the percentage of those neurons that shared the same probabilities of failure. Then, we computed the probability of AP transmission for the corresponding percentages of excitatory neurons in the networks. Thus, in the network model, we took into account the heterogeneity observed in the single neuron model under each demyelination/remyelination condition.

However, In Figures 7 and 8, we consider additional heterogeneity due to a different degree of demylination/remyelination of different neurons. Here, excitatory neurons in the network model are not perturbed according to a single demyelination/remyelination condition. Instead, we allowed that different percentages of excitatory neurons had AP failure rates corresponding to different demyelination/remyelination conditions: some were unperturbed, while others had different degrees of demyelination (Figure 7) and different degrees of remyelination (Figure 8). We have modified the text for clarification in several places.

First, when we describe the impact of demyelination on working memory, we already mention that (line 271): “In each of the 10 networks, we set the AP failure rate of the excitatory neurons according to the distribution of failure probabilities of the neurons in the single neuron cohort for the given demyelination or remyelination condition. Thus, we took into account the heterogeneity of demyelination and remyelination effects from our single neuron cohort (Figure 3A; Supplementary Figure 3). Note that this heterogeneity originates from differences in axon properties, but probabilities of failure for all neurons in the network correspond to the same degree of demyelination (Figure 6). We will also consider networks that contain different combinations of axons with either intact or perturbed myelin (Figure 7 and Figure 8).”

Second, we have combined the text describing Figures 7 and 8 under a single section title, which reads “Simulated heterogenous myelin alterations match empirical data” (line 334) and start this section with (line 337): “Up to this point we have studied network models with AP failure probabilities corresponding to a single degree of myelin alterations (i.e., with all excitatory neurons in the network having AP failure rates matched to those of the single neuron cohort for one particular demyelination or remyelination condition). Next, we sought to reveal the effect on working memory performance of more biologically realistic network models, where excitatory neurons in the networks were perturbed according to a combination of different demyelination or remyelination conditions. That is, we simulated networks with excitatory neurons having AP failure probabilities matched to both neuronal axons with intact and with altered myelin sheaths in different degrees, as likely occurs in the aging brain (Figure 1).”

(7) "Unexpectedly, our model indicates that compared to the performance of networks composed of neurons possessing axons with intact myelin sheaths, both demyelination and remyelination leads to an impaired performance." This conclusion is quite interesting, but I lack intuition from the paper as of why it is happening. In fact, the authors say in the Discussion that "complete remyelination of all the previously demyelinated segments with sufficient myelin, with fewer transitions between long and short segments, recovered working memory function." Would we then see a minimum and then an increase in memory duration in Figure 9B if we extended the X-axis until we hit 100% of new myelin sheaths?

This is a very important question that we have carefully addressed in Results and Discussion. We distinguish between two remyelination cases in the models. Complete remyelination: when all (100%) the previously demyelinated segments have been subsequently remyelinated, and incomplete remyelination: when less than 100% (25%, 50% or 75%) of the demyelinated segments have been remyelinated. Figure 6 (middle and right columns) shows the two cases (black lines for any percentage of lamellae added vs. colored lines): for 100% of the segments remyelinated, the network performance is nearly or completely (when enough lamellae are added) recovered to the young network performance. In fact, with the single neuron model we observe that (lines 192 - 193 in Results): “Remyelinating all affected segments with 75% of lamellae (the maximal amount of remyelination) nearly eliminated AP failures (1.8 ± 1.1%)”. However, incomplete remyelination recovers the performance compared to demyelination (middle and right columns in Figure 6 vs left column), but this performance is worse than the performance of the young networks. The single neuron model shows that (lines 194 - 197 in Results): “Incomplete remyelination, where some segments were still demyelinated, still had relatively high AP failure rates. For example, when one eighth of segments were remyelinated with the maximal amount of lamellae and one eighth were left bare, 25.7 ± 11.5% of APs failed across the cohort (Figure 4C, red dashed line and arrow).”

In Figure 9B (now Figure 8B), we combine intact axons with axons that are only partially remyelinated (i.e., incomplete remyelination). Extending the X-axis in Figure 8B until 100% of new myelin sheaths would not imply a minimum and a subsequent increase, but a continuous impairment: the more axons we perturb (remyelinate) the higher is the impairment compared to the young cases where all the axons are intact.

The sentence "Unexpectedly, our model indicates that compared to the performance of networks composed of neurons possessing axons with intact myelin sheaths, both demyelination and remyelination leads to an impaired performance.", now reads as (lines 379 380 in Results): “Therefore, both demyelination and incomplete remyelination lead to impaired performance in our networks, compared to networks with intact myelin sheaths”. We have also rewritten the corresponding section in Discussion (lines 486 - 489) as follows: “Therefore, it is reasonable to assume that ineffective remyelination may lead to working memory impairment. In fact, complete remyelination of all previously demyelinated segments with sufficient myelin, with fewer transitions between long and short segments, led to full recovery of working memory function.”

(8) [minor] "Our recent network model found that age-related changes in firing rates and synapse numbers in individual neurons can lead to working memory impairment (Ibañez et al., 2020), but did not consider myelin dystrophy." Could you be more precise about which age-related changes were studied in Ibanez et al. 2020? From the paper it seems like it was mostly cellular excitability and synaptic density, so this should be added here for more context.

To clarify this, we have added the following sentences in the Introduccion (line 105):

“Our recent network model revealed that the empirically observed age-related increase in AP firing rates in prefrontal pyramidal neurons (modeled through an increased slope of the f-I curve) and loss of up to 30% of both excitatory and inhibitory synapses (modeled as a decrease in connectivity strength) can lead to working memory impairment (Ibañez et al., 2020), but this model did not incorporate the known changes to myelin structure that occur during normal

aging.”

(9) [minor] "Recurrent excitatory synapses are facilitating, which promotes robust and reliable persistent activity despite spatial heterogeneities in the connectivity or in the intrinsic properties of the neurons." It would be great to add a reference here to justify the inclusion of this type of plasticity in the excitatory circuit (for example Wang, Markram et al. Nat Neuro 2006).

We have added the references suggested by the reviewer and a further one in the Results (line 216):

“Recurrent excitatory synapses are facilitating, as has been empirically observed in PFC (Hempel et al., 2000; Wang et al., 2006), which promotes robust and reliable persistent activity despite spatial heterogeneities in the connectivity or in the intrinsic properties of the neurons.”

References:

Hempel, C. M., Hartman, K. H., Wang, X. J., Turrigiano, G. G., and Nelson, S. B. (2000). Multiple forms of short-term plasticity at excitatory synapses in rat medial prefrontal cortex. J. Neurophysiol. 83, 3031–3041. doi: 10.1152/jn.2000.83.5.3031

Wang, Y., Markram, H., Goodman, P. H., Berger, T. K., Ma, J., and Goldman- Rakic, P. S.(2006). Heterogeneity in the pyramidal network of the medial prefrontal cortex. Nat.Neurosci. 9, 534–542. doi: 10.1038/nn1670

eLife assessment

This manuscript reports a valuable computational study of the effects of axon de-myelination and re-myelination on action potential speed and propagation failure. The manuscript presents solid evidence for the effects of de- and re-myelination in different models of working memory, with potential implications in disorders such as multiple sclerosis. The exposition of the manuscript is targeted for researchers interested in biophysical models of cognitive deficits.

Reviewer #1 (Public Review):

Summary:

The authors study the effects of myelin alterations in working memory via the complementary use of two computational approaches: one based on the de- and re-myelination in multicompartmental models of pyramidal neurons, and one based on synaptic changes in a spiking bump attractor model for spatial working memory. The first model provides the most precise angle (biophysically speaking) of the different effects (loss of myelin lamella or segments, remyelination with thinner and shorter nodes, etc), while the second model allows to infer the consequences of myelin alterations in working memory performance, including memory stability, duration, and bump diffusion, while also exploring the case of myeling alterations in a novel silent working memory model. The results indicate (i) a slowing down and failure of propagation of spikes with demyelination and partial recovery with remyelination, with detailed predictions on the role of nodes and myelina lamella, and (ii) a decrease in memory duration and an increase in memory drift as a function of the demyelination, in agreement with multiple experimental studies.

Strengths:

Overall, the work offers a very interesting approach of a topic which is hard to accomplish experimentally --therefore the computational take is entirely justified and extremely useful. The authors carefully designed the computational experiments to shed light into the demyelination effects on working memory from multiple levels of description, increasing the reliability of their conclusions. I think this work provides now convincing evidence and has the potential to be influential in future studies of myelin alterations (and related disorders such as multiple sclerosis).

Weaknesses:

In its current form, the authors have improved the clarity of the results and the model details, and have provided a new set of simulations to complement and reinforce the original ones (including the development of a new spatial working memory model based on silent working memory principles). I do not appreciate any significant weaknesses at this point.

Reviewer #2 (Public Review):

This paper analyzes the effect of axon de-myelination and re-myelination on action potential speed, and propagation failure. Next, the findings are then incorporated in a standard spiking ring attractor model of working memory.

I think the results are not very surprising or solid and there are issues with method and presentation.<br /> The authors did many simulations with random parameters, then averaged the result, and found for instance that the Conduction Velocity drops in demyelination. It gives the reader little insight into what is really going on. My personal preference is for a well understood simple model rather than a poorly understood complex model. The link between the model outcome of WM and data remains qualitative and is further weakened by the existence of known other age-related effects in PFC circuits.

Comments on revised version:

The paper has improved in the revision, although I still think a reduced model would have been nice.

eLife assessment

The authors develop a self-returning self-avoiding polymer model of chromosome organization and show that their framework can recapitulate at the same time local density and large-scale contact structural properties observed experimentally by various technologies. The presented theoretical framework and the results are valuable for the community of modelers working on 3D genomics. The work provides solid evidence that such a framework can be used, is reliable in describing chromatin organization at multiple scales, and could represent an interesting alternative to standard molecular dynamics simulations of chromatin polymer models.

Reviewer #1 (Public Review):

Carignano et al propose an extension of the self-returning random walk (SRRW) model for chromatin to include excluded volume aspects and use it to investigate generic local and global properties of the chromosome 3D organization inside eukaryotic nuclei. In particular, they focus on chromatin volumic density, contact probability, and domain size and suggest that their framework can recapitulate several experimental observations and predict the effect of some perturbations.

Strengths:

• The developed methodology is convincing and may offer an alternative - less computationally demanding - framework to investigate the single-cell and population structural properties of 3D genome organization at multiple scales.

• Compared to the previous SRRW model, it allows for investigation of the role of excluded volume locally.

• They perform some experiments to compare with model predictions and show consistency between the two.

Weaknesses:

• The model is a homopolymer model and currently cannot fully account for specific mechanisms that may shape the heterogeneous, complex organization of chromosomes (TAD at specific positions, A/B compartmentalization, promoter-enhancer loops, etc.).

• By construction of their framework, the effect of excluded volume is only local and larger-scale properties for which excluded volume could be a main actor (formation of chromosome territories [Rosa & Everaers, PLoS CB 2009], bottle-brush effects due to loop extrusion [Polovnikov et al, PRX 2023], etc.) cannot be captured.

• Apart from being a computationally interesting approach to generating realistic 3D chromosome organization, the method offers fewer possibilities than standard polymer models (eg, MD simulations) of chromatin (no dynamics, no specific mechanisms, etc.) with likely the same predictive power under the same hypotheses. In particular, authors often claim the superiority of their approach to describing the local chromatin compaction compared to previous polymer models without showing it or citing any relevant references that would show it.

• Comparisons with experiments are solid but are not quantified.

Impact:

Building on the presented framework in the future to incorporate TAD and compartments may offer an interesting model to study the single-cell heterogeneity of chromatin organization. But currently, in this reviewer's opinion, standard polymer modeling frameworks may offer more possibilities.

Reviewer #2 (Public Review):

Summary:

The authors introduce a simple Self Returning Excluded Volume (SR-EV) model to investigate the 3D organization of chromatin. This is a random walk with a probability to self-return accounting for the excluded volume effects. The authors use this method to study the statistical properties of chromatin organization in 3D. They compute contact probabilities, 3D distances, and packing properties of chromatin and compare them with a set of experimental data.

Strengths:

(1) Typically, to generate a polymer with excluded volume interactions, one needs to run long simulations with computationally expensive repulsive potentials like the Weeks-Chanlder-Anderson potential. However, here, instead of performing long simulations, the authors have devised a method where they can grow polymer, enabling quick generation of configurations.

(2) Authors show that the chromatin configurations generated from their models do satisfy many of the experimentally known statistical properties of chromatin. Contact probability scalings and packing properties are comparable with Chromatin Scanning Transmission Electron Microscopy (ChromSTEM)  experimental data from some of the cell types.

Weaknesses:

This can only generate broad statistical distributions. This method cannot generate sequence-dependent effects, specific TAD structures, or compartments without a prior model for the folding parameter alpha. It cannot generate a 3D distance between specific sets of genes. This is an interesting soft-matter physics study. However, the output is only as good as the alpha value one provides as input.

a enlever

geïnteresseerd

Heetwatertank

Melkdranken

. Deze brengt de frisheid van cold brew naar de professionele wereld.

Melkdranken

te maken.

telkens 500 gr koffie

gemakkelijk

a enlever

Melkdranken

Heetwatertank

melkdranken

ervaring van superieure kwaliteit is.

de kwaliteitsvolle

te maken.

telkens 500 gr koffie

gemakkelijk tot

PERIOD

a enlever

geïnteresseerd

Melkdranken

Heetwatertank

aangezien deze elegantie, prestaties en gebruiksgemak combineert.

Deze Zwitserse merk combineert een elegant design met hoogwaardige prestaties. Dankzij het kleurenscherm is de Jura X6 Dark Inox een zeer gebruiksvriendelijke machnie die uw medewerkers dagelijks blij zal maken.

bonenmachine van Jura, ideaal

Beschrijving

te maken.

telkens 500 gr koffie

gemakkelijk

via plug&play geleverd

herstellingservice

fighting is healthy, able to share opinions

couples have a hard tike to construct their feelings into words

创始人

他父亲说

调节心态

这些数字都确认过吗？

确认数字

koffiemachines

--> sur le point d'interrogation juste derrière il faut changer "gerepareerd" par hersteld "reparatie" par herstelling.

a enlever

meerdere mogelijkheden !

geïnteresseerd

Melkdranken

Heetwatertank

.Deze is zeer intuïtief en bovenop zeer design !

Beschrijving

telkens 500 gr koffie toevoegen, hiermee kun je ongeveer 62 kopjes serveren !

gemakkelijk

eLife assessment

The study by Kim et al. is a valuable contribution to the topic of obtaining good channel conductance parameters from electrophysiological recordings. While promising in its ability to rapidly construct newly fitted models using generative adversarial networks, the approach is incompletely described and the generated models often substantially deviate from the dynamics observed empirically. The comparison with existing multi-objective optimization methods is also incomplete.

Reviewer #1 (Public Review):

The manuscript describes a GAN-based approach that generates parameters for HH-like channels for multiple C. Elengans neurons. The network is trained on generated data to produce parameter sets that, on the one hand, reproduce voltage responses and IV curves, and on the other hand, are indistinguishable from the ground truth parameters, as tested by the discriminator. It is then shown that these generated parameter sets lead to reasonable reproductions of the recorded responses (but see the section "weaknesses" below for some reservations).

Strengths:

In itself, I find the methodology of high interest, particularly in that it can generate parameter sets to construct models of new recordings at a very low computational cost.

Weaknesses:

Nevertheless, I believe there are some weaknesses in the evaluation of the models that should be addressed before the quality of the methodology can be fully assessed. Firstly, at the methodological level, the authors should provide more clarity on the inverse gradient operation they use, as opposed to just simulating the models, as such an inversion depends not only on the parameters but also on the state of the model. How the state is obtained remains unclear here. Secondly, in the evaluation of their models, the authors could provided more information such as IV curves, as whether these would be accurate is difficult to visually infer from their figures. Thirdly, the authors do not address the question of whether all obtained parameter sets are stable when simulated over longer times, while their figures do include hints that this might not be the case for at least some of their models (e.g. voltage traces that do not converge back to the equilibrium after the stimulus, but rather seem to diverge).

Reviewer #2 (Public Review):

Summary:

Generating biophysically detailed computational models that capture the characteristic physiological properties of biological neurons for diverse cell types is an important and difficult problem in computational neuroscience. One major challenge lies in determining the large number of parameters of such models, which are notoriously difficult to fit into experimental data. Thereby, the computational and energy costs can be significant. The study 'ElectroPhysiomeGAN: Generation of Biophysical Neuron Model Parameters from Recorded Electrophysiological Responses' by Kim et al. describes a computationally efficient approach for predicting model parameters of Hodgkin-Huxley neuron models using Generative Adversarial Networks (GANs) trained on simulation data. The method is applied to generate models for 9 non-spiking neurons in C. elegans based on electrophysiological recordings. While the generated models capture the responses of these neurons to some degree, they generally show significant deviations from the empirically observed responses in important features. While interesting, in its current form, the method has not been demonstrated to generate models that faithfully capture empirically observed responses.

Strengths:

The authors work on an important and difficult problem. A noteworthy strength of their approach is that once trained, the GANs can generate models from new empirical data with very little computational effort. The generated models reproduce the average voltage during current injections reasonably well.

Weaknesses:

Major 1: While the models generated with EP-GAN reproduce the average voltage during current injections reasonably well, the dynamics of the response are not well captured. For example, for the neuron labeled RIM (Figure 2), the most depolarized voltage traces show an initial 'overshoot' of depolarization, i.e. they depolarize strongly within the first few hundred milliseconds but then fall back to a less depolarized membrane potential. In contrast, the empirical recording shows no such overshoot. Similarly, for the neuron labeled AFD, all empirically recorded traces slowly ramp up over time. In contrast, the simulated traces are mostly flat. Furthermore, all empirical traces return to the pre-stimulus membrane potential, but many of the simulated voltage traces remain significantly depolarized, far outside of the ranges of empirically observed membrane potentials. While these deviations may appear small in the Root mean Square Error (RMSE), the only metric used in the study to assess the quality of the models, they likely indicate a large mismatch between the model and the electrophysiological properties of the biological neuron.

Major 2: Other metrics than the RMSE should be incorporated to validate simulated responses against electrophysiological data. A common approach is to extract multiple biologically meaningful features from the voltage traces before, during and after the stimulus, and compare the simulated responses to the experimentally observed distribution of these features. Typically, a model is only accepted if all features fall within the empirically observed ranges (see e.g. https://doi.org/10.1371/journal.pcbi.1002107). However, based on the deviations in resting membrane potential and the return to the resting membrane potential alone, most if not all the models shown in this study would not be accepted.

Major 3: Abstract and introduction imply that the 'ElectroPhysiome' refers to models that incorporate both the connectome and individual neuron physiology. However, the work presented in this study does not make use of any connectomics data. To make the claim that ElectroPhysiomeGAN can jointly capture both 'network interaction and cellular dynamics', the generated models would need to be evaluated for network inputs, for example by exposing them to naturalistic stimuli of synaptic inputs. It seems likely that dynamics that are currently poorly captured, like slow ramps, or the ability of the neuron to return to its resting membrane potential, will critically affect network computations.

how?

eLife assessment

This valuable study by Wu and Zhou combines neurophysiological recordings and computational modelling to address an interesting question regarding the sequence of events from sensing to action. Neurophysiological evidence remains incomplete: explicit mapping of saccade-related activity in the same neurons and a better understanding of the influence of the spatial configuration of stimulus and targets would be required to pinpoint whether such activity might contribute, even partially, to the observed results and interpretations. These results are of interest for neuroscientists investigating decision-making.

Reviewer #1 (Public Review):

Summary:

This valuable study by Wu and Zhou combined neurophysiological recordings and computational modelling to investigate the neural mechanisms that underpin the interaction between sensory evaluation and action selection. The neurophysiological results suggest non-linear modulation of decision-related LIP activity by action selection, but some further analysis would be helpful in order to understand whether these results can be generalised to LIP circuitry or might be dependent on specific spatial task configurations. The authors present solid computational evidence that this might be due to projections from choice target representations. These results are of interest for neuroscientists investigating decision-making.

Strengths:

Wu and Zhou combine awake behaving neurophysiology for a sophisticated, flexible visual-motion discrimination task and a recurrent network model to disentangle the contribution of sensory evaluation and action selection to LIP firing patterns. The correct saccade response direction for preferred motion direction choices is randomly interleaved between contralateral and ipsilateral response targets, which allows the dissociation of perceptual choice from saccade direction.<br /> The neurophysiological recordings from area LIP indicate non-linear interaction between motion categorisation decisions and saccade choice direction.

The careful investigation of a recurrent network model suggests that feedback from choice target representations to an earlier sensory evaluation stage might be the source for this non-linear modulation and that it is an important circuit component for behavioural performance.

The paper presents a possible solution to a central controversy about the role of LIP in perceptual decision-making, but see below.

Weaknesses:

The paper presents a possible solution to a central controversy about the role of LIP in perceptual decision-making. However, the authors could be more clear and upfront about their interpretational framework and potential alternative interpretations.<br /> Centrally, the authors' model and experimental data appears to test only that LIP carries out sensory evaluation in its RFs. The model explicitly parks the representation of choice targets outside the "LIP" module receiving sensory input. The feedback from this separate target representation provides then the non-linear modulation that matches the neurophysiology. However, they ignore the neurophysiological results that LIP neurons can also represent motor planning to a saccade target.<br /> The neurophysiological results with a modulation of the direction tuning by choice direction (contralateral vs ipsilateral) are intriguing. However, the evaluation of the neurophysiological results are difficult, because some of the necessary information is missing to exclude alternative explanations. It would be good to see the actual distributions and sizes of the RF, which were determined based on visual responses not with a delayed saccade task. There might be for example a simple spatial configuration, for example, RF and preferred choice target in the same (contralateral) hemifield, for which there is an increase in firing. It is a shame that we do not see what these neurons would do if only a choice target would be put in the RF, as has been done in so many previous LIP experiments. The authors exclude also some spatial task configurations (vertical direction decisions), which makes it difficult to judge whether these data and models can be generalised. The whole section is difficult to follow, partly also because it appears to mix reporting results with interpretation (e.g. "feedback").

The model and its investigation is very interesting and thorough, but given the neurophysiological literature on LIP, it is not clear that the target module would need to be in a separate brain area, but could be local circuitry within LIP between different neuron types.

Reviewer #2 (Public Review):

Summary:

In this manuscript, the authors recorded activity in the posterior parietal cortex (PPC) of monkeys performing a perceptual decision-making task. The monkeys were first shown two choice dots of two different colors. Then, they saw a random dot motion stimulus. They had to learn to categorize the direction of motion as referring to either the right or left dot. However, the rule was based on the color of the dot and not its location. So, the red dot could either be to the right or left, but the rule itself remained the same. It is known from past work that PPC neurons would code the learned categorization. Here, the authors showed that the categorization signal depended on whether the executed saccade was in the same hemifield as the recorded PPC neuron or in the opposite one. That is, if a neuron categorized the two motion directions such that it responded stronger for one than the other, then this differential motion direction coding effect was amplified if the subsequent choice saccade was in the same hemifield. The authors then built a computational RNN to replicate the results and make further tests by simulated "lesions".

Strengths:

Linking the results to RNN simulations and simulated lesions.

Weaknesses:

Potential interpretational issues due to a lack of evidence on what happens at the time of the saccades.

https://www.washingtonpost.com/climate-environment/2024/06/10/anthropocene-epoch-human-climate-impacts/

eLife assessment

This paper makes a valuable contribution to our understanding of the tradeoffs in eye design - specifically between improvements in optics and in photoreceptor performance. The authors successfully build a formal theory that enables comparisons across a wide range of species and eye types. The conclusion from the modeling is that resources are split relatively evenly between optics and photoreceptors, and hence that both must be considered in eye design. Evidence for this conclusion is solid, and could be strengthened with a more complete comparison with the experiment.

Reviewer #1 (Public Review):

Summary:

Two important factors in visual performance are the resolving power of the lens and the signal-to-noise ratio of the photoreceptors. These both compete for space: a larger lens has improved resolving power over a smaller one, and longer photoreceptors capture more photons and hence generate responses with lower noise. The current paper explores the tradeoff of these two factors, asking how space should be allocated to maximize eye performance (measured as encoded information).

Strengths:

The topic of the paper is interesting and not well studied. The approach is clearly described and seems appropriate (with a few exceptions - see weaknesses below). In most cases, the parameter space of the models are well explored and tradeoffs are clear.

Weaknesses:

- Light level<br /> The calculations in the paper assume high light levels (which reduces the number of parameters that need to be considered). The impact of this assumption is not clear. A concern is that the optimization may be quite different at lower light levels. Such a dependence on light level could explain why the model predictions and experiment are not in particularly good agreement. The paper would benefit from exploring this issue.

- Discontinuities<br /> The discontinuities and non-monotonicity of the optimal parameters plotted in Figure 4 are concerning. Are these a numerical artifact? Some discussion of their origin would be quite helpful.

- Discrepancies between predictions and experiment<br /> As the authors clearly describe, experimental measurements of eye parameters differ systematically from those predicted. This makes it difficult to know what to take away from the paper. The qualitative arguments about how resources should be allocated are pretty general, and the full model seems a complex way to arrive at those arguments. Could this reflect a failure of one of the assumptions that the model rests on - e.g. high light levels, or that the cost of space for photoreceptors and optics is similar? Given these discrepancies between model and experiment, it is also hard to evaluate conclusions about the competition between optics and photoreceptors (e.g. at the end of the abstract) and about the importance for evolution (end of introduction).

Reviewer #2 (Public Review):

Summary:

In short, the paper presents a theoretical framework that predicts how resources should be optimally distributed between receptors and optics in eyes.

Strengths:

The authors build on the principle of resource allocation within an organism and develop a formal theory for optimal distribution of resources within an eye between the receptor array and the optics. Because the two parts of eyes, receptor arrays and optics, share the same role of providing visual information to the animal it is possible to isolate these from resource allocation in the rest of the animal. This allows for a novel and powerful way of exploring the principles that govern eye design. By clever and thoughtful assumptions/constraints, the authors have built a formal theory of resource allocation between the receptor array and the optics for two major types of compound eye as well as for camera-type eyes. The theory is formalized with variables that are well characterized in a number of different animal eyes, resulting in testable predictions.

The authors use the theory to explain a number of design features that depend on different optimal distribution of resources between the receptor array and the optics in different types of eyes. As an example, they successfully explain why eye regions with different spatial resolution should be built in different ways. They also explain differences between different types of eyes, such as long photoreceptors in apposition compound eyes and much shorter receptors in camera type eyes. The predictive power in the theory is impressive.

To keep the number of parameters at a minimum, the theory was developed for two types of compound eye (neural superposition, and apposition) and for camera-type eyes. It is possible to extend the theory to other types of eyes, although it would likely require more variables and assumptions/constraints to the theory. It is thus good to introduce the conceptual ideas without overdoing the applications of the theory.

The paper extends a previous theory, developed by the senior author, that develops performance surfaces for optimal cost/benefit design of eyes. By combining this with resource allocation between receptors and optics, the theoretical understanding of eye design takes a major leap and provides entirely new sets of predictions and explanations for why eyes are built the way they are.

The paper is well written and even though the theory development in the Results may be difficult to take in for many biologists, the Discussion very nicely lists all the major predictions under separate headings, and here the text is more tuned for readers that are not entirely comfortable with the formalism of the Results section. I must point out though that the Results section is kept exemplary concise. The figures are excellent and help explain concepts that otherwise may go above the head of many biologists.

Reviewer #3 (Public Review):

Summary:

This is a proposal for a new theory for the geometry of insect eyes. The novel cost-benefit function combines the cost of the optical portion with the photoreceptor portion of the eye. These quantities are put on the same footing using a specific (normalized) volume measure, plus an energy factor for the photoreceptor compartment. An optimal information transmission rate then specifies each parameter and resource allocation ratio for a variable total cost. The elegant treatment allows for comparison across a wide range of species and eye types. Simple eyes are found to be several times more efficient across a range of eye parameters than neural superposition eyes. Some trends in eye parameters can be explained by optimal allocation of resources between the optics and photoreceptors compartments of the eye.

Strengths:

Data from a variety of species roughly align with rough trends in the cost analysis, e.g. as a function of expanding the length of the photoreceptor compartment.

New data could be added to the framework once collected, and many species can be compared.

Eyes of different shapes are compared.

Weaknesses:

Detailed quantitative conclusions are not possible given the approximations and simplifying assumptions in the models and poor accounting for trends in the data across eye types.

eLife assessment

This important study provides solid evidence that both psychiatric dimensions (e.g. anhedonia, apathy, or depression) and chronotype (i.e., being a morning or evening person) influence effort-based decision-making. Notably, the current study does not elucidate whether there may be interactive effects of chronotype and psychiatric dimensions on decision-making. This work is of importance to researchers and clinicians alike, who may make inferences about behaviour and cognition without taking into account whether the individual may be tested or observed out-of-sync with their phenotype.

Reviewer #1 (Public Review):

Summary:

This study uses an online cognitive task to assess how reward and effort are integrated in a motivated decision-making task. In particular the authors were looking to explore how neuropsychiatric symptoms, in particular apathy and anhedonia, and circadian rhythms affect behavior in this task. Amongst many results, they found that choice bias (the degree to which integrated reward and effort affects decisions) is reduced in individuals with greater neuropsychiatric symptoms, and late chronotypes (being an 'evening person').

Strengths:

The authors recruited participants to perform the cognitive task both in and out of sync with their chronotypes, allowing for the important insight that individuals with late chronotypes show a more reduced choice bias when tested in the morning.<br /> Overall, this is a well-designed and controlled online experimental study. The modelling approach is robust, with care being taken to both perform and explain to the readers the various tests used to ensure the models allow the authors to sufficiently test their hypotheses.

Weaknesses:

This study was not designed to test the interactions of neuropsychiatric symptoms and chronotypes on decision making, and thus can only make preliminary suggestions regarding how symptoms, chronotypes and time-of-assessment interact.

Reviewer #2 (Public Review):

Summary:

The study combines computational modeling of choice behavior with an economic, effort-based decision-making task to assess how willingness to exert physical effort for a reward varies as a function of individual differences in apathy and anhedonia, or depression, as well as chronotype. They find an overall reduction in effort selection that scales with apathy and anhedonia and depression. They also find that later chronotypes are less likely to choose effort than earlier chronotypes and, interestingly, an interaction whereby later chronotypes are especially unwilling to exert effort in the morning versus the evening.

Strengths:

This study uses state-of-the-art tools for model fitting and validation and regression methods which rule out multicollinearity among symptom measures and Bayesian methods which estimate effects and uncertainty about those estimates. The replication of results across two different kinds of samples is another strength. Finally, the study provides new information about the effects not only of chronotype but also chronotype by timepoint interactions which are previously unknown in the subfield of effort-based decision-making.

Weaknesses:

The study has few weaknesses. One potential concern is that the range of models which were tested was narrow, and other models might have been considered. For example, the Authors might have also tried to fit models with an overall inverse temperature parameter to capture decision noise. One reason for doing so is that some variance in the bias parameter might be attributed to noise, which was not modeled here. Another concern is that the manuscripts discuss effort-based choice as a transdiagnostic feature - and there is evidence in other studies that effort deficits are a transdiagnostic feature of multiple disorders. However, because the present study does not investigate multiple diagnostic categories, it doesn't provide evidence for transdiagnosticity, per se.

Reviewer #3 (Public Review):

Summary:

In this manuscript, Mehrhof and Nord study a large dataset of participants collected online (n=958 after exclusions) who performed a simple effort-based choice task. They report that the level of effort and reward influence choices in a way that is expected from prior work. They then relate choice preferences to neuropsychiatric syndromes and, in a smaller sample (n<200), to people's circadian preferences, i.e., whether they are a morning-preferring or evening-preferring chronotype. They find relationships between the choice bias (a model parameter capturing the likelihood to accept effort-reward challenges, like an intercept) and anhedonia and apathy, as well as chronotype. People with higher anhedonia and apathy and an evening chronotype are less likely to accept challenges (more negative choice bias). People with an evening chronotype are also more reward sensitive and more likely to accept challenges in the evening, compared to the morning.

Strengths:

This is an interesting and well-written manuscript which replicates some known results and introduces a new consideration related to potential chronotype relationships which have not been explored before. It uses a large sample size and includes analyses related to transdiagnostic as well as diagnostic criteria. I have some suggestions for improvements.

Weaknesses:

(1) The novel findings in this manuscript are those pertaining to transdiagnostic and circadian phenotypes. The authors report two separate but "overlapping" effects: individuals high on anhedonia/apathy are less willing to accept offers in the task, and similarly, individuals tested off their chronotype are less willing to accept offers in the task. The authors claim that the latter has implications for studying the former. In other words, because individuals high on anhedonia/apathy predominantly have a late chronotype (but might be tested early in the day), they might accept less offers, which could spuriously look like a link between anhedonia/apathy and choices but might in fact be an effect of the interaction between chronotype and time-of-testing. The authors therefore argue that chronotype needs to be accounted for when studying links between depression and effort tasks.<br /> The authors argue that, if X is associated with Y and Z is associated with Y, X and Z might confound each other. That is possible, but not necessarily true. It would need to be tested explicitly by having X (anhedonia/apathy) and Z (chronotype) in the same regression model. Does the effect of anhedonia/apathy on choices disappear when accounting for chronotype (and time-of-testing)? Similarly, when adding the interaction between anhedonia/apathy, chronotype, and time-of-testing, within the subsample of people tested off their chronotype, is there a residual effect of anhedonia/apathy on choices or not?<br /> If the effect of anhedonia/apathy disappeared (or got weaker) while accounting for chronotype, this result would suggest that chronotype mediates the effect of anhedonia/apathy on effort choices. However, I am not sure it renders the direct effect of anhedonia/apathy on choices entirely spurious. Late chronotype might be a feature (induced by other symptoms) of depression (such as fatigue and insomnia), and the association between anhedonia/apathy and effort choices might be a true and meaningful one. For example, if the effect of anhedonia/apathy on effort choices was mediated by altered connectivity of the dorsal ACC, we would not say that ACC connectivity renders the link between depression and effort choices "spurious", but we would speak of a mechanism that explains this effect. The authors should discuss in a more nuanced way what a significant mediation by the chronotype/time-of-testing congruency means for interpreting effects of depression in computational psychiatry.

(2) It seems that all key results relate to the choice bias in the model (as opposed to reward or effort sensitivity). It would therefore be helpful to understand what fundamental process the choice bias is really capturing in this task. This is not discussed, and the direction of effects is not discussed either, but potentially quite important. It seems that the choice bias captures how many effortful reward challenges are accepted overall which maybe captures general motivation or task engagement. Maybe it is then quite expected that this could be linked with questionnaires measuring general motivation/pleasure/task engagement. Formally, the choice bias is the constant term or intercept in the model for p(accept), but the authors never comment on what its sign means. If I'm not mistaken, people with higher anhedonia but also higher apathy are less likely to accept challenges and thus engage in the task (more negative choice bias). I could not find any discussion or even mention of what these results mean. This similarly pertains to the results on chronotype. In general, "choice bias" may not be the most intuitive term and the authors may want to consider renaming it. Also, given the sign of what the choice bias means could be flipped with a simple sign flip in the model equation (i.e., equating to accepting more vs accepting less offers), it would be helpful to show some basic plots to illustrate the identified differences (e.g., plotting the % accepted for people in the upper and lower tertile for the SHAPS score etc).

(3) None of the key effects relate to effort or reward sensitivity which is somewhat surprising given the previous literature and also means that it is hard to know if choice bias results would be equally found in tasks without any effort component. (The only analysis related to effort sensitivity is exploratory and in a subsample of N=56 per group looking at people meeting criteria for MDD vs matched controls.) Were stimuli constructed such that effort and reward sensitivity could be separated (i.e., are uncorrelated/orthogonal)? Maybe it would be worth looking at the % accepted in the largest or two largest effort value bins in an exploratory analysis. It seems the lowest and 2nd lowest effort level generally lead to accepting the challenge pretty much all the time, so including those effort levels might not be sensitive to individual difference analyses?

(4) The abstract and discussion seem overstated (implications for the school system and statements on circadian rhythms which were not measured here). They should be toned down to reflect conclusions supported by the data.

eLife assessment

This important work advances our understanding of microglial aging trajectory and heterogeneity. The authors provide an in-depth characterization of microglia in aging and aim to identify molecular checkpoints, that while solid are also deemed incomplete to support all the authors' claims. The study should be of interest to neuroimmunologists and biologists interested in aging.

Reviewer #1 (Public Review):

Summary:

This manuscript by Shea and Villeda furnishes the field with a valuable scRNAseq data set detailing microglial aging in the mouse hippocampus. They provide clear evidence that changes in microglial attributes begin in mid-life, well before time points when mice are traditionally considered to be "aging." It also adds to a growing body of data in the field demonstrating that there is substantial heterogeneity in microglial responses to aging. Using in vitro experiments and transgenic manipulations in mice, the authors show that transforming growth factor beta (TGFb1)-based signaling can potently impact microglial state, consistent with previous findings in the field. They also demonstrate that manipulation of microglial TGFb1-based signaling can impact hippocampus-dependent behaviors.

Limitations of the study lie primarily in reaching too far with interpretations of the data. The authors argue that changes in microglial transcriptome during midlife represent a type of "checkpoint," after which microglial aging can progress along distinct trajectories depending on the status of TGFb1 signaling. They also posit that a specific intermediate "stress response" state in midlife is mechanistically linked to a translational burst that drives the subsequent progression of microglia to an "inflammatory state." Unequivocal data to support these causal links is lacking, however. similarly, key additional experiments would be needed to demonstrate that TGFb1 signaling and microglial progression through these identified intermediate states are causally linked to cognitive decline.

Guidance for readers along with study strengths and caveats:

The present manuscript provides valuable strengthening and expansion to a growing body of data showing prominent changes in the microglial state during aging. Microarray(1), bulkRNAseq(2-5), scRNAseq(6,7), snRNAseq(8,9), and spatial transcriptomic(10) approaches have been leveraged to map changes in microglial transcriptome during aging in rodents, non-human primates, and humans. A number of these studies include the hippocampus (1,8,9,11) and have highlighted variation across brain regions in microglial transcriptomic changes during aging (1,11). They have also revealed differences across sex (7) as well as increased cell-to-cell heterogeneity (6-10), consistent with the idea that individual microglia can follow distinct aging trajectories. Several of these studies revealed that changes in microglial attributes begin in middle age (1,7,11), supporting similar observations from studies that did not use omics (12-14). The present manuscript utilizes scRNAseq of hippocampal microglia at adulthood (6mo), middle age (12mo), late middle age (18mo) and aging (24mo) to show that aging-induced changes in microglia begin in middle age and that microglia exhibit ample phenotypic heterogeneity during the progression of aging.

To gain further insight into the dynamics of microglial aging in the hippocampus, the authors used a bioinformatics method known as "pseudotime" or "trajectory inference" to understand how cells may progress through different functional states, as defined by cellular transcriptome (15,16). These bioinformatics approaches can reveal key patterns in scRNAseq / snRNAseq datasets and, in the present study, the authors conclude that a "stress response" module characterized by expression of TGFb1 represents a key "checkpoint" in microglial aging in midlife, after which the cells can move along distinct transcriptional trajectories as aging progresses. This is an intriguing possibility. However, pseudotime analyses need to be validated via additional bioinformatics as well as follow-up experiments. Indeed, Heumos et al, in their Nature Genetics "Expert Guidelines" Review, emphasize that "inferred trajectories might not necessarily have biological meaning." They recommend that "when the expected topology is unknown, trajectories and downstream hypotheses should be confirmed by multiple trajectory inference methods using different underlying assumptions."(15) Numerous algorithms are available for trajectory inference (e.g. Monocle, PAGA, Sligshot, RaceID/StemID, among many others) and their performance and suitability depends on the individual dataset and nature of the trajectories that are to be inferred. It is recommended to use dynGuidelines(16) for the selection of optimal pseudotime analysis methods. In the present manuscript, the authors do not provide any justification for their use of Monocle 3 over other trajectory inference approaches, nor do they employ a secondary trajectory inference method to confirm observations made with Monocle 3. Finally, follow-up validation experiments that the authors carry out have their own limitations and caveats (see below). Hence, while the microglial aging trajectories identified by this study are intriguing, they remain hypothetical trajectories that need to be proven with additional follow-up experiments.

To follow up on the idea that TGFb1 signaling in microglia plays a key role in determining microglial aging trajectories, the authors use RNAscope to show that TGFb1 levels in microglia peak in middle age. They also treat primary LPS-activated microglia with TGFb1 and show that this restores expression of microglial homeostatic gene expression and dampens expression of stress response and, potentially, inflammatory genes. Finally, they utilize transgenic approaches to delete TGFb1 from microglia around 8-10mo of age and scRNAseq to show that homeostatic signatures are lost and inflammatory signatures are gained. Hence, findings in this study support the idea that TGFb1 can strongly regulate microglial phenotype. Loss of TGFb1 signaling to microglia in adulthood has already been shown to cause decreased microglial morphological complexity and upregulation of genes typically associated with microglial responses to CNS insults(17-19). TGFb1 signaling to microglia has also been implicated in microglial responses to disease and manipulations to increase this signaling can improve disease progression in some cases(19). In this light, the findings in the present study are largely confirmatory of previous findings in the literature. They also fall short of unequivocally demonstrating that TGFb1 signaling acts as a "checkpoint" for determining subsequent microglial aging trajectory. To show this clearly, one would need to perturb TGFb1 signaling around 12mo of age and carry out sequencing (bulkRNAseq or scRNAseq) of microglia at 18mo and 24mo. Such experiments could directly demonstrate whether the whole microglial population has been diverted to the TGFb1-low aging trajectory (that progresses through a translational burst state to an inflammation state as proposed). Future development of tools to tag TGFb1 high or low microglia could also enable fate tracing type experiments to directly show whether the TGFb1 state in middle age predicts cell state at later phases of aging.

The present study would also like to draw links between features of microglial aging in the hippocampus and a decline in hippocampal-dependent cognition during aging. To this end, they carry out behavioral testing in 8-10mo old mice that have undergone microglial-specific TGFb1 deletion and find deficits in novel object recognition and contextual fear conditioning. While this provides compelling evidence that TGFb1 signaling in microglia can impact hippocampus-dependent cognition in midlife, it does not demonstrate that this signaling accelerates or modulates cognitive decline (see below). Age-associated cognitive decline refers to cognitive deficits that emerge as a result of the normative brain aging process(20-21). For a cognitive deficit to be considered age-associated cognitive decline, it must be shown that the cognitive operation under study was intact at some point earlier in the adult lifespan. This requires longitudinal study designs that determine whether a manipulation impacts the relationship between brain status and cognition as animals age (22-24). Alternatively, cross-sectional studies with adequate sample sizes can be used to sample the variability in cognitive outcomes at different points of the adult lifespan(22-24) and show that this is altered by a particular manipulation. For this specific study, one would ideally demonstrate that hippocampal-based learning/memory was intact at some point in the lifespan of mice with microglial TGFb1 KO but that this manipulation accelerated or exacerbated the emergence of deficits in hippocampal-dependent learning/memory during aging. In the absence of these types of data, the authors should tone down their claims that they have identified a cellular and molecular mechanism that contributes to cognitive decline.

A final point of clarification for the reader pertains to the mining of previously generated data sets within this study. The language in the results section, methods, and figure legends causes confusion about which experiments were actually carried out in this study versus previous studies. Some of the language makes it sound as though parabiosis experiments and experiments using mouse models of Alzheimer's Disease were carried out in this study. However, parabiosis and AD mouse model experiments were executed in previous studies (25,26), and in the present study, RNAseq datasets were accessed for targeted data mining. It is fantastic to see further mining of datasets that already exist in the field. However, descriptions in the results and methods sections need to make it crystal clear that this is what was done.

References:

(1) Grabert, K. et al. Microglial brain region-dependent diversity and selective regional sensitivities to aging. Nat. Neurosci. (2016). doi:10.1038/nn.4222<br /> (2) Hickman, S. E. et al. The microglial sensome revealed by direct RNA sequencing. Nat. Neurosci. (2013). doi:10.1038/nn.3554<br /> (3) Deczkowska, A. et al. Mef2C restrains microglial inflammatory response and is lost in brain ageing in an IFN-I-dependent manner. Nat. Commun. (2017). doi:10.1038/s41467-017-00769-0<br /> (4) O'Neil, S. M., Witcher, K. G., McKim, D. B. & Godbout, J. P. Forced turnover of aged microglia induces an intermediate phenotype but does not rebalance CNS environmental cues driving priming to immune challenge. Acta Neuropathol. Commun. (2018). doi:10.1186/s40478-018-0636-8<br /> (5) Olah, M. et al. A transcriptomic atlas of aged human microglia. Nat. Commun. (2018). doi:10.1038/s41467-018-02926-5<br /> (6) Hammond, T. R. et al. Single-Cell RNA Sequencing of Microglia throughout the Mouse Lifespan and in the Injured Brain Reveals Complex Cell-State Changes. Immunity 50, 253-271 (2019).<br /> (7) Li, X. et al. Transcriptional and epigenetic decoding of the microglial aging process. Nat. aging 3, 1288-1311 (2023).<br /> (8) Zhang, H. et al. Single-nucleus transcriptomic landscape of primate hippocampal aging. Protein Cell 12, 695-716 (2021).<br /> (9) Su, Y. et al. A single-cell transcriptome atlas of glial diversity in the human hippocampus across the postnatal lifespan. Cell Stem Cell 29, 1594-1610.e8 (2022).<br /> (10) Allen, W. E., Blosser, T. R., Sullivan, Z. A., Dulac, C. & Zhuang, X. Molecular and spatial signatures of mouse brain aging at single-cell resolution. Cell 186, 194-208.e18 (2023).<br /> (11) Soreq, L. et al. Major Shifts in Glial Regional Identity Are a Transcriptional Hallmark of Human Brain Aging. Cell Rep. 18, 557-570 (2017).<br /> (12) Hefendehl, J. K. et al. Homeostatic and injury-induced microglia behavior in the aging brain. Aging Cell (2014). doi:10.1111/acel.12149<br /> (13) Nikodemova, M. et al. Microglial numbers attain adult levels after undergoing a rapid decrease in cell number in the third postnatal week. J. Neuroimmunol. 0, 280-288 (2015).<br /> (14) Moca, E. N. et al. Microglia Drive Pockets of Neuroinflammation in Middle Age. J. Neurosci. 42, 3896-3918 (2022).<br /> (15) Heumos, L. et al. Best practices for single-cell analysis across modalities. Nat. Rev. Genet. 24, 550-572 (2023).<br /> (16) Saelens, W., Cannoodt, R., Todorov, H. & Saeys, Y. A comparison of single-cell trajectory inference methods: towards more accurate and robust tools. (2018). doi:10.1101/276907<br /> (17) Zöller, T. et al. Silencing of TGFβ signalling in microglia results in impaired homeostasis. Nat. Commun. 9, (2018).<br /> (18) Bedolla, A. et al. Microglia-derived TGF-β1 ligand maintains microglia homeostasis via autocrine mechanism and is critical for normal cognitive function in adult mouse brain. bioRxiv Prepr. Serv. Biol. (2023). doi:10.1101/2023.07.05.547814<br /> (19) Spittau, B., Dokalis, N. & Prinz, M. The Role of TGFβ Signaling in Microglia Maturation and Activation. Trends Immunol. 41, 836-848 (2020).<br /> (20) L. Nyberg, M. Lövdén, K. Riklund, U. Lindenberger, L. Bäckman, Memory aging and brain maintenance. Trends Cogn. Sci. 16, 292-305 (2012).<br /> (21) L. Luo, F. I. M. Craik, Aging and memory: A cognitive approach. Can. J. Psychiatry 53, 346-353 (2008).<br /> (22) Y. Stern, M. Albert, C. Barnes, R. Cabeza, A. Pascual-Leone, P. Rapp.<br /> A framework for concepts of reserve and resilience in aging. Neurobiol. Aging, 124 (2022), pp. 100-103, 10.1016/j.neurobiolaging.2022.10.015<br /> (23) Y. Stern, C.A. Barnes, C. Grady, R.N. Jones, N. Raz. Brain reserve, cognitive reserve, compensation, and maintenance: operationalization, validity, and mechanisms of cognitive resilience. Neurobiol. Aging, 83 (2019), pp. 124-129, 10.1016/j.neurobiolaging.2019.03.022<br /> (24) R. Cabeza, M. Albert, S. Belleville, F.I.M. Craik, A. Duarte, C.L. Grady, U. Lindenberger, L. Nyberg, D.C. Park, P.A. Reuter-Lorenz, M.D. Rugg, J. Steffener, M.N. Rajah. Maintenance, reserve and compensation: the cognitive neuroscience of healthy ageing. Nat. Rev. Neurosci., 19 (11) (2018), Article 11, 10.1038/s41583-018-0068-2<br /> (25) Palovics, R. et al molecular hallmarks of heterochronic parabiosis at single-cell resolution. Nature 603, 309-314 (2022)<br /> (26) Sala Frigerio, C. et al. The major risk factors for Alzheimer's Disease: age, sex, and genes modulate the microglial response to Abeta plaques. Cell Rep, 27, 1293-1306 (2019)

Reviewer #2 (Public Review):

Summary:

The goal of the paper was to trace the transitions hippocampal microglia undergo along aging. ScRNA-seq analysis allowed the authors to predict a trajectory and hypothesize about possible molecular checkpoints, which keep the pace of microglial aging. E.g. TGF1b was predicted as a molecule slowing down the microglial aging path and indeed, loss of TGF1 in microglia led to premature microglia aging, which was associated with premature loss of cognitive ability. The authors also used the parabiosis model to show how peripheral, blood-derived signals from the old organism can "push" microglia forward on the aging path.

Strengths:

A major strength and uniqueness of this work is the in-depth single-cell dataset, which may be a useful resource for the community, as well as the data showing what happens to young microglia in heterochronic parabiosis setting and upon loss of TGFb in their environment.

Weaknesses:

That said, given what we recently learned about microglia isolation for RNA-seq analysis, there is a danger that some of the observations are a result of not age, but cell stress from sample preparation (enzymatic digestion 10min at 37C; e.g. PMID: 35260865). Changes in cell state distribution along aging were made based on scRNA-seq and were not corroborated by any other method, such as imaging of cluster-specific marker expression in microglia at different ages. This analysis would allow confirming the scRNA-seq data and would also give us an idea of where the subsets are present within the hippocampus, and whether there is any interesting distribution of cell states (e.g. some are present closer to stem cells?). Since TGFb is thought to be crucial to microglia biology, it would be valuable to include more analysis of the mice with microglia-specific Tgfb deletion e.g. what was the efficiency of recombination in microglia? Did their numbers change after induction of Tgfb deletion in Cx3cr1-creERT2::Tgfb-flox mice.

Overall:

In general, I think the authors did a good job following the initial observations and devised clever ways to test the emerging hypotheses. The resulting data are an important addition to what we know about microglial aging and can be fruitfully used by other researchers, e.g. those working on microglia in a disease context.

https://taz.de/EZB-kritisiert/!6016842/

eLife assessment

This valuable study showing that sleep deprivation increases functional synapses while depleting silent synapses supports previous findings that excitatory signaling, in particular via AMPA receptors, increases during wakefulness. The consistency with the literature increases confidence in the conclusions, which otherwise are supported by incomplete evidence. An interesting aspect of this manuscript is the inclusion of a model for the accumulation of sleep need that is based upon the MEF2C transcription factor but also links to the sleep-regulating SIK3-HDAC4/5 pathway. As such, the manuscript is as much of a perspective as a primary research paper.

Reviewer #1 (Public Review):

Summary:

This manuscript by Vogt et al examines how the synaptic composition of AMPA and NMDA receptors changes over sleep and wake states. The authors perform whole-cell patch clamp recordings to quantify changes in silent synapse numbers across conditions of spontaneous sleep, sleep deprivation, and recovery sleep after deprivation. They also perform single nucleus RNAseq to identify transcriptional changes related to AMPA/NMDA receptor composition following spontaneous sleep and sleep deprivation. The findings of this study are consistent with a decrease in silent synapse number during wakefulness and an increase during sleep. However, these changes cannot be conclusively linked to sleep/wake states. Measurements were performed in the motor cortex, and sleep deprivation was achieved by forced locomotion, raising the possibility that recent levels of neuronal activity/induction of plasticity, rather than sleep/wake states, are responsible for the observed results.

Strengths:

This study examines an important question. Glutamatergic synaptic transmission has been a focus of studies in the sleep field, but AMPA receptor function has been the primary target of these studies. Silent synapses, which contain NMDA receptors but lack AMPA receptors, have important functional consequences for the brain. Exploring the role of sleep in regulating silent synapse numbers is important to understanding the role of sleep in brain function. The electrophysiological approach of measuring the failure rate ratio, supported by AMPA/NMDA ratio measurements, is a rigorous tool to evaluate silent synapse numbers.

The authors also perform snRNAseq to identify genes differentially expressed in the spontaneous sleep and sleep deprivation groups. This analysis reveals an intriguing pattern of upregulated genes controlled by HDAC4 and Mef2c, along with synaptic shaping component genes and genes associated with autism spectrum disorder, across cell types in the sleep deprivation group. This unbiased approach identifies candidate genes for follow-up studies.

Weaknesses:

A major weakness of this study is the experimental design. Measurements are made from the motor cortex, and sleep deprivation was achieved using forced locomotion on a treadmill. Therefore, the effects observed could be due to recent high levels of activity or plasticity induction in the motor cortex from locomotion, rather than lack of sleep per se. In support of this interpretation, other groups have failed to find a difference in AMPA/NMDA ratio in mice with different spontaneous sleep/wake histories, although sleep deprivation was not performed (Bridi et al., Neuron 2020).

The electrophysiological measurements are problematic in several ways. First, the methods lack crucial details such as inclusion/exclusion criteria for each cell based on input and series resistance, stability of input/series resistance, polysynaptic responses, etc. that make it difficult to interpret the data. The holding potential (-90mV) used for AMPA receptor current recordings is much more hyperpolarized than typically used for these measurements. The statistical analysis of these experiments is also problematic. The number of mice used is low (3/group) and more should be added to account for inter-animal variability. Comparing the raw data with the statistical tests in supplementary table 1 (FR ratio), it appears that a data point has been dropped from the analysis, but it is unclear why. In addition, a false discovery rate (FDR) correction for multiple comparisons is used to evaluate group differences following the ANOVAs. Correcting for the FDR is less stringent and is typically used when a large number of hypotheses are tested and false positives are more acceptable. In this analysis, few comparisons are made, and the standard approach of correcting for the family-wise error rate is more appropriate.

The snRNAseq data are intriguing, but a more thorough discussion of the candidate genes and pathways that are upregulated during sleep deprivation is warranted. Several genes relevant to the AMPA/NMDA ratio are mentioned, but upregulation of most of these genes would not be expected to increase the AMPA/NMDA ratio based on the literature cited. The model presented in Figure 4C is not consistent with the data (e.g. many candidate genes could alter NMDAR function without receptor insertion/removal), and it is unclear how the current study fits into the model presented in 4D.

Reviewer #2 (Public Review):

Summary:

Here Vogt et al., provide new insights into the need for sleep and the molecular and physiological response to sleep loss. The authors expand on their previously published work (Bjorness et al., 2020) and draw from recent advances in the field to propose a neuron-centric molecular model for the accumulation and resolution of sleep need and the basis of restorative sleep function. While speculative, the proposed model successfully links important observations in the field and provides a framework to stimulate further research and advances on the molecular basis of sleep function. In my review, I highlight the important advances of this current work, and the clear merits of the proposed model, and indicate areas of the model that can serve to stimulate further investigation.

Strengths:

Reviewer comment on new data in Vogt et al., 2024<br /> Using classic slice electrophysiology, the authors conclude that wakefulness (sleep deprivation (SD)) drives a potentiation of excitatory glutamate synapses, mediated in large part by "un-silencing" of NMDAR-active synapses to AMPAR-active synapses. Using a modern single nuclear RNAseq approach the authors conclude that SD drives changes in gene expression primarily occurring in glutamatergic neurons. The two experiments combined highlight the accumulation and resolution of sleep need centered on the strength of excitatory synapses onto excitatory neurons. This view is entirely consistent with a large body of extant and emerging literature and provides important direction for future research.

Consistent with prior work, wakefulness/SD drives an LTP-type potentiation of excitatory synaptic strength on principle cortical neurons. It has been proposed that LTP associated with wake, leads to the accumulation of sleep need by increasing neuronal excitability, and by the "saturation" of LTP capacity. This saturation subsequently impairs the capacity for further ongoing learning. This new data provides a satisfying mechanism of this saturation phenomenon by introducing the concept of silent synapses. The new data show that in mice well rested, a substantial number of synapses are "silent", containing an NMDAR component but not AMPARs. Silent synapses provide a type of reservoir for learning in that activity can drive the un-silencing, increasing the number of functional synapses. SD depletes this reservoir of silent synapses to essentially zero, explaining how SD can exhaust learning capacity. Recovery sleep led to restoration of silent synapses, explaining how recovery sleep can renew learning capacity. In their prior work (Bjorness et al., 2020) this group showed that SD drives an increase in mEPSC frequency onto these same cortical neurons, but without a clear change in pre-synaptic release probability, implying a change in the number of functional synapses. This prediction is now born out in this new dataset.

The new snRNAseq dataset indicates the sleep need is primarily seen (at the transcriptional level) in excitatory neurons, consistent with a number of other studies. First, this conclusion is corroborated by an independent, contemporary snRNAseq analysis recently available as a pre-print (Ford et al., 2023 BioRxiv https://doi.org/10.1101/2023.11.28.569011). A recently published analysis on the effects of SD in drosophila imaged synapses in every brain region in a cell-type dependent manner (Weiss et al., PNAS 2024), concluding that SD drives brain wide increases in synaptic strength almost exclusively in excitatory neurons. Further, Kim et al., Nature 2022, heavily cited in this work, show that the newly described SIK3-HDAC4/5 pathway promotes sleep depth via excitatory neurons and not inhibitory neurons.

The new experiments provided in Fig1-3 are expertly conducted and presented. This reviewer has no comments of concern regarding the execution and conclusions of these experiments.

Reviewer comment on the model in Vogt et al., 2024

In the view of this reviewer the new model proposed by Vogt et al., is an important contribution. The model is not definitively supported by new data, and in this regard should be viewed as a perspective, providing mechanistic links between recent molecular advances, while still leaving areas that need to be addressed in future work. New snRNAseq analysis indicates that SD drives the expression of synaptic shaping components (SSCs) consistent with the excitatory synapse as a major target for the restorative basis of sleep function. SD-induced gene expression is also enriched for autism spectrum disorder (ASD) risk genes. As pointed out by the authors, sleep problems are commonly reported in ASD, but the emphasis has been on sleep amount. This new analysis highlights the need to understand the impact on sleep's functional output (synapses) to fully understand the role of sleep problems in ASD.

Importantly, SD-induced gene expression in excitatory neurons overlaps with genes regulated by the transcription factor MEF2C and HDAC4/5 (Figure 4). In their prior work, the authors show loss of MEF2C in excitatory neurons abolished the SD transcriptional response and the functional recovery of synapses from SD by recovery sleep. Recent advances identified HDAC4/5 as major regulators of sleep depth and duration (in excitatory neurons) downstream of the recently identified sleep-promoting kinase SIK3. In Zhou et al., and Kim et al., Nature 2022, both groups propose a model whereby "sleep-need" signals from the synapse activate SIK3, which phosphorylates HDAC4/5, driving cytoplasmic targeting, allowing for the de-repression and transcriptional activation of "sleep genes". Prior work shows that HDAC4/5 are repressors of MEF2C. Therefore, the "sleep genes" derepressed by HDAC4/5 may be the same genes activated in response to SD by MEF2C. The new model thereby extends the signaling of sleep need at synapses (through SIK3-HDAC4/5) to the functional output of synaptic recovery by expression of synaptic/sleep genes by MEF2C. The model thereby links aspects of the expression of sleep need with the resolution of sleep need by mediating sleep function: synapse renormalization.

Weaknesses:

Areas for further investigation

In the discussion section Vogt et al., explore the links between excitatory synapse strength, arguably the major target of "sleep function", and NREM slow-wave activity (SWA), the most established marker of sleep need. SIK3-HDAC4/5 have major effects on the "depth" of sleep by regulating NREM-SWA. The effects of MEF2C loss of function on NREM SWA activity are less obvious, but clearly impact the recovery of glutamatergic synapses from SD. The authors point out how adenosine signaling is well established as a mediator of SWA, but the links between adenosine and glutamatergic strength are far from clear. The mechanistic links between SIK3/HDAC4/5, adenosine signaling, and MEF2C, are far from understood. Therefore, the molecular/mechanistic links between a synaptic basis of sleep need and resolution with NREM-SWA activity require further investigation.

Additional work is also needed to understand the mechanistic links between SIK3-HDAC4/5 signaling and MEF2C activity. The authors point out that constitutively nuclear (cn) HDAC4/5 (acting as a repressor) will mimic MEF2C loss of function. This is reasonable, however, there are notable differences in the reported phenotypes of each. Notably, cnHDAC4/5 suppresses NREM amount and NREM SWA but had no effect on the NREM-SWA increase following SD (Zhou et al., Nature 2022). Loss of MEF2C in CaMKII neurons had no effect on NREM amount and suppressed the increase in NREM-SWA following SD (Bjorness et al., 2020). These instances indicate that cnHDAC4/5 and loss of MEF2C do not exactly match suggesting additional factors are relevant in these phenotypes. Likely HDAC4/5 have functionally important interactions with other transcription factors, and likewise for MEF2C, suggesting areas for future analysis.

One emerging theme may be that the SIK3-HDAC4/5 axis is a major regulator of the sleep state, perhaps stabilizing the NREM state once the transition from wakefulness occurs. MEF2C is less involved in regulating sleep per se, and more involved in executing sleep function, by promoting restorative synaptic modifications to resolve sleep need.

Finally, advances in the roles of the respective SIK3-HDAC4/5 and MEF2C pathways point towards transcription of "sleep genes", as clearly indicated in the model of Figure 4. Clearly, more work is needed to understand how the expression of such genes ultimately leads to the resolution of sleep need by functional changes at synapses. What are these sleep genes and how do they mechanistically resolve sleep need? Thus, the current work provides a mechanistic framework to stimulate further advances in understanding the molecular basis for sleep need and the restorative basis of sleep function.

eLife assessment

The important study by Ding and colleagues identifies subpopulations of neurons recorded in the monkey subthalamic nucleus (STN) with distinct activity profiles and causal contributions during perceptual decision-making. The combination of neuronal recording, microstimulation, and computational methods provides convincing evidence for a heterogenous neural population that could support multifaceted roles in decision formation. This study should be of wide interest to computational and experimental neuroscientists interested in cognitive function.

Reviewer #1 (Public Review):

The study reports that STN neurons recorded while monkeys performed a random-dot motion task show diverse activation timecourses relative to task events and dependencies on coherence, reaction time, and saccade-choice direction. Different neuron types could be grouped into functional subpopulations, e.g., coherence sensitivity emerging early only in choice-coding neurons. Clustering techniques identified three functionally defined neuron clusters whose dynamic activity profiles related to computational predictions of different decision models in the literature. Microstimulation at different STN recording sites affected behavioral performance in varying but well-conceptualized ways that were captured by the parameters of drift-diffusion models and related to the presence of STN functional clusters at recording sites. The authors conclude that their results validate key aspects of decision models and identify novel aspects of decision-related STN activity.

This is an interesting and high-quality paper that will be of interest across computational and decision neuroscience fields. The recordings and data analyses seem carefully conducted. The study has an attractive theoretical starting point of three specific computational signals that are then mapped onto identified neuron clusters. The combination of single-cell recordings, microstimulation, and computational modelling is a distinct strength of the paper. I only have a few questions and suggestions for clarification.

(1) It would be helpful to explain the criteria for choosing a given number of clusters and for accepting the final clustering solution more clearly. The quantitative results (silhouette plots, Rand index) in Supplementary Figure 2 should perhaps be included in the main figure to justify the parameter choices and acceptance of specific clustering solutions.

(2) It would be helpful to show how the activity profiles in Figure 3 would look like for 3 or 5 (or 6) clusters, to give the reader an impression of how activity profiles recovered using different numbers of clusters would differ.

(3) The authors attempt to link the microstimulation effects to the presence of functional neuron clusters at the stimulation site. How can you rule out that there were other, session-specific factors (e.g., related to the animal's motivation) that affected both neuronal activity and behavior? For example, could you incorporate aspects of the monkey's baseline performance (mean reaction time, fixation breaks, error trials) into the analysis?

(4) Line 84: What was the rationale for not including both coherence and reaction time in one multiple regression model?

Reviewer #2 (Public Review):

This study uses single-unit recordings in the monkey STN to examine the evidence for three theoretical models that propose distinct roles for the STN in perceptual decision-making. Importantly, the proposed functional roles are predictive of unique patterns of neural activity. Using k-means clustering with seeds informed by each model's predictions, the current study identified three neural clusters with activity dynamics that resembled those predicted by the described theoretical models. The authors are thorough and transparent in reporting the analyses used to validate the clustering procedure and the stability of the clustering results. To further establish a causal role for the STN in decision-making, the researchers applied microstimulation to the STN and found effects on response times, choice preferences, and latent decision parameters estimated with a drift diffusion model. Overall, the study provides strong evidence for a functionally diverse population of STN neurons that could indeed support multiple roles involved in perceptual decision-making. The manuscript would benefit from stronger evidence linking each neural cluster to specific decision roles in order to strengthen the overall conclusions.

The interpretation of the results, and specifically, the degree to which the identified clusters support each model, is largely dependent on whether the artificial vectors used as model-based clustering seeds adequately capture the expected behavior under each theoretical model. The manuscript would benefit from providing further justification for the specific model predictions summarized in Figure 1B. Further, although each cluster's activity can be described in the context of the discussed models, these same neural dynamics could also reflect other processes not specific to the models. That is, while a model attributing the STN's role to assessing evidence accumulation may predict a ramping up of neural activity, activity ramping is not a selective correlate of evidence accumulation and could be indicative of a number of processes, e.g., uncertainty, the passage of time, etc. This lack of specificity makes it challenging to infer the functional relevance of cluster activity and should be acknowledged in the discussion.

Additionally, although the effects of STN microstimulation on behavior provide important causal evidence linking the STN to decision processes, the stimulation results are highly variable and difficult to interpret. The authors provide a reasonable explanation for the variability, showing that neurons from unique clusters are anatomically intermingled such that stimulation likely affects neurons across several clusters. It is worth noting, however, that a substantial body of literature suggests that neural populations in the STN are topographically organized in a manner that is crucial for its role in action selection, providing "channels" that guide action execution. The authors should comment on how the current results, indicative of little anatomical clustering amongst the functional clusters, relate to other reports showing topographical organization.

Overall, the association between the identified clusters and the function ascribed to the STN by each of the models is largely descriptive and should be interpreted accordingly. For example, Figure 3 is referenced when describing which cluster activity is choice/coherence dependent, yet it is unclear what specific criteria and measures are being used to determine whether activity is choice/coherence "dependent." Visually, coherence activity seems to largely overlap in panel B (top row). Is there a statistically significant distinction between low and high coherence in this plot? The interpretation of these plots and the methods used to determine choice/coherence "dependence" needs further explanation.

In general, the association between cluster activity and each model could be more directly tested. At least two of the models assume coordination with other brain regions. Does the current dataset include recordings from any of these regions (e.g., mPFC or GPe) that could be used to bolster claims about the functional relevance of specific subpopulations? For example, one would expect coordinated activity between neural activity in mPFC and Cluster 2 according to the Ratcliff and Frank model. Additionally, the reported drift-diffusion model (DDM) results are difficult to interpret as microstimulation appears to have broad and varied effects across almost all the DDM model parameters. The DDM framework could, however, be used to more specifically test the relationships between each neural cluster and specific decision functions described in each model. Several studies have successfully shown that neural activity tracks specific latent decision parameters estimated by the DDM by including neural activity as a predictor in the model. Using this approach, the current study could examine whether each cluster's activity is predictive of specific decision parameters (e.g., evidence accumulation, decision thresholds, etc.). For example, according to the Ratcliff and Frank model, activity in cluster 2 might track decision thresholds.

Reviewer #3 (Public Review):

Summary:

The authors provide compelling evidence for the causal role of the subthalamic nucleus (STN) in perceptual decision-making. By recording from a large number of STN neurons and using microstimulation, they demonstrate the STN's involvement in setting decision bounds, scaling evidence accumulation, and modulating non-decision time.

Strengths:

The study tested three hypotheses about the STN's function and identified distinct STN subpopulations whose activity patterns support predictions from previous computational models. The experiments are well-designed, the analyses are rigorous, and the results significantly advance our understanding of the STN's multi-faceted role in decision formation.

Weaknesses:

While the study provides valuable insights into the STN's role in decision-making, there are a few areas that could be improved. First, the interpretation of the neural subpopulations' activity patterns in relation to the computational models should be clarified, as the observed patterns may not directly correspond to the specific signals predicted by the models. Second, the authors could consider using a supervised learning method to more explicitly model the pattern correlations between the three profiles. Third, a neural population model could be employed to better understand how the STN population jointly contributes to decision-making dynamics. Finally, the added value of the microstimulation experiments should be more directly addressed in the Results section, as the changes in firing patterns compared to the original patterns are not clearly evident.

eLife assessment

This important study uses calcium imaging to show an increase in the selectivity of the sensory-evoked response in the apical dendritic tuft of layer 5 barrel cortex neurons as mice learn a whisker-dependent discrimination task. The evidence supporting the conclusions is compelling, and this work will be of great interest to neuroscientists working on reward-based learning and sensory processing.

Reviewer #1 (Public Review):

What neurophysiological changes support the learning of new sensorimotor transformations is a key question in neuroscience. Many studies have attempted to answer this question at the neuronal population level - with varying degrees of success - but few, if any, have studied the change in activity of the apical dendrites of layer 5 cortical neurons. Neurons in layer 5 of the sensory cortex appear to play a key role in sensorimotor transformations, showing important decision and reward-related signals, and being the main source of cortical and subcortical projections from the cortex. In particular, pyramidal track (PT) neurons project directly to subcortical regions related to motor activity, such as the striatum and brainstem, and could initiate rapid motor action in response to given sensory inputs. Additionally, layer 5 cortical neurons have large apical dendrites that extend to layer 1 where different neuromodulatory and long-range inputs converge, providing motor and contextual information that could be used to modulate layer 5 neurons output and/or to establish the synaptic plasticity required for learning a new association.

In this study, the authors aimed to test whether the learning of a new sensorimotor transformation could be supported by a change in the evoked response of the apical dendrites of layer 5 neurons in the mouse whisker primary somatosensory cortex. To do this, they performed longitudinal functional calcium imaging of the apical dendrites of layer 5 neurons while mice learned to discriminate between two multi-whisker stimuli. The authors used a simple conditioning task in which one whisker stimulus (upward or backward air puff, CS+) is associated with a reward after a short delay, while the other whisker stimulus (CS-) is not. They found that task learning (measured by the probability of anticipatory licking just after the CS+) was not associated with a significant change in the average population response evoked by the CS+ or the CS-, nor a change in the average population selectivity. However, when considering individual dendritic tufts, they found interesting changes in selectivity, with approximately equal numbers of dendrites becoming more selective for CS+ and dendrites becoming more selective for CS-.

One of the major challenges when assessing changes in neural representation during the learning of such Go/NoGo tasks is that the movements and rewards themselves may elicit strong neural responses that may be a confounding factor, that is, inexperienced mice do not lick in response to the CS+, while trained mice do. In this study, the authors addressed this issue in three ways: first, they carefully monitored the orofacial movements of mice and showed that task learning is not associated with changes in evoked whisker movements. Second, they show that whisking or licking evokes very little activity in the dendritic tufts compared to whisker stimuli (CS+ and CS-). Finally, the authors introduced into the design of their task a post-conditioning session after the last conditioning session during which the CS+ and the CS- are presented but no reward is delivered. During this post-session, the mice gradually stopped licking in response to the CS+. A better design might have been to perform the pre-conditioning and post-conditioning sessions in non-water-restricted, unmotivated mice to completely exclude any lick response, but the fact that the change in selectivity persists after the mice stopped licking in the last blocks of the post-conditioning session (in mice relying only on their whiskers to perform the task) is convincing.

The clever task design and careful data analysis provide compelling evidence that learning this whisker discrimination task does not result in a massive change in sensory representation in the apical dendritic tufts of layer 5 neurons in the primary somatosensory cortex on average. Nevertheless, individual dendritic tufts do increase their selectivity for one or the other sensory stimulus, likely enhancing the ability of S1 neurons to accurately discriminate the two stimuli and trigger the appropriate motor response (to lick or not to lick).

One limitation of the present study is the lack of evidence for the necessity of the primary somatosensory cortex in the learning and execution of the task. As the authors have strongly emphasized in their previous publications, the primary somatosensory cortex may not be necessary for the learning and execution of simple whisker detection tasks, especially when the stimulus is very salient. Although this new task requires the discrimination between two whisker stimuli, the simplicity and salience of the whisker stimuli used could make this task cortex-independent. Especially when considering that some mice seem to not rely entirely on their whiskers to execute the task.

Nevertheless, this is an important result that shows for the first time changes in the selectivity to sensory stimuli at the level of individual apical dendritic tufts in correlation with the learning of a discrimination task. This study sheds new light on the cortical cellular substrates of reward-based learning and opens interesting perspectives for future research in this area. In future studies, it will be important to determine whether the change in selectivity of dendritic calcium spikes is causally involved in the learning of the task or whether it simply correlates with learning, as a consequence of changes in synaptic inputs caused by reward. The dendritic calcium spikes may be involved in the establishment of synaptic plasticity required for learning and impact the output of layer 5 pyramidal neurons to trigger the appropriate motor response. It would be important also to study the changes in selectivity in the apical dendrite of the identified projection neurons.

Reviewer #2 (Public Review):

Summary:

The authors did not find an increased representation of CS+ throughout reinforcement learning in the tuft dendrites of Rbp4-positive neurons from layer 5B of the barrel cortex, as previously reported for soma from layer 2/3 of the visual cortex.

Alternatively, the authors observed an increased selectivity to both stimuli (CS+ and CS-) during reinforcement learning. This feature:

(1) was not present in repeated exposures (without reinforcement),<br /> (2) was not explained by the animal's behaviour (choice, licking, and whisking), and<br /> (3) was long-lasting, being present even when the mice disengaged from the task.

Importantly, increased selectivity was correlated with learning (% correct choices), and neural discriminability between stimuli increased with learning.

In conclusion, the authors show that tuft dendrites from layer 5B of the barrel cortex increase the representation of conditioned (CS+) and unconditioned stimuli (CS-) applied to the whiskers, during reinforcement learning.

Strengths:

The results presented are very consistent throughout the entire study, and therefore very convincing:

(1) The results observed are very similar using two different imaging techniques (2-photon -planar imaging- and SCAPE-volumetric imaging). Figure 3 and Figure 4 respectively.

(2) The results are similar using "different groups" of tuft dendrites for the analysis (e.g. initially unresponsive and responsive pre- and post-learning). Figure 5.

(3) The results are similar from a specific set of trials (with the same sensory input, but different choices). Figure 7.

(4) Additionally, the selectivity of tuft dendrites from layer 5B of the barrel cortex was higher in the mice that exclusively used the whisker to respond to the stimuli (CS+ and CS-).<br /> The results presented are controlled against a group of mice that received the same stimuli presentation, except for the reinforcement (reward).

Additionally, the behaviour outputs, such as choice, whisking, and licking could not account for the results observed.

Although there are no causal experiments, the correlation between selectivity and learning (percentage of correct choices), as well as the increased neural discriminability with learning, but not in repeated exposure, are very convincing.

Weaknesses:

The biggest weakness is the absence of causality experiments. Although inhibiting specifically tuft dendritic activity in layer 1 from layer 5 pyramidal neurons is very challenging, tuft dendritic activity in layer 1 could be silenced through optogenetic experiments as in Abs et al. 2018. By manipulating NDNF-positive neurons the authors could specifically modify tuft dendritic activity in the barrel cortex during CS presentations, and test if silencing tuft dendritic activity in layer 1 would lead to the lack of selectivity and an impairment of reinforcement learning. Additionally, this experiment will test if the selectivity observed during reinforcement learning is due to changes in the local network, namely changes in local synaptic connectivity, or solely due to changes in the long-range inputs.

eLife assessment:

This study reported that cold exposure induced mRNA expression of genes related to lipid metabolism in the paraventricular nucleus of the hypothalamus (PVH). The authors provide useful data highlighting the potential role of lipid metabolism in the brain during cold exposure. However, the study is incomplete and would require specific experiments to solidify the claims being made.

Reviewer #1 (Public Review):

Summary:

This study focuses on metabolic changes in the paraventricular hypothalamic (PVH) region of the brain during acute periods of cold exposure. The authors point out that in comparison to the extensive literature on the effects of cold exposure in peripheral tissues, we know relatively little about its effects on the brain. They specifically focus on the hypothalamus, and identify the PVH as having changes in Atgl and Hsl gene expression changes during cold exposure. They then go on to show accumulation of lipid droplets, increased Fos expression, and increased lipid peroxidation during cold exposure. Further, they show that neuronal activation is required for the formation of lipid droplets and lipid peroxidation.

Strengths:

A strength of the study is trying to better understand how metabolism in the brain is a dynamic process, much like how it has been viewed in other organs. The authors also use a creative approach to measuring in vivo lipid peroxidation via delivery of a BD-C11 sensor through a cannula to the region in conjunction with fiber photometry to measure fluorescence changes deep in the brain.

Weaknesses:

Although the topic and findings are of interest, there are a few key weaknesses in the study that would improve the work if addressed. One weakness was many of the experiments were done in a manner that could not distinguish between the contributions of neurons and glial cells, limiting the extent of conclusions that could be made. While this is not easily doable for all experiments, it can be done for some. For example, the Fos experiments in Figure 3 would be more conclusive if done with the labeling of neuronal nuclei with NeuN, as glial cells can also express Fos. To similarly show more conclusively that neurons are being activated during cold exposure, the calcium imaging experiments in Figure S3 can be done with cold exposure. Additionally, many experiments are only done with the minimal three animals required for statistics and could be more robust with additional animals included. Another weakness is that the authors do not address whether manipulating lipid droplet accumulation or lipid peroxidation has any effect on PVH function (e.g. does it change neuronal activity in the region?).

Reviewer #2 (Public Review):

Summary:

Cold-induced lipid metabolism is well-established in adipose tissues. The authors set out to determine whether cold could alter brain lipid metabolism. By QPCR analysis of brain punches after acute cold, they found that mRNA expressions of several lipolysis-related genes were upregulated compared to RT controls. By combining fluorescent sensors and in vivo fiberphotometry, they observed cold-induced lipid peroxidation/lipolysis, which could be blocked by pharmacological inhibitors of neuronal activity (muscimol and kynurenic acid). The brain is not traditionally considered an organ with high lipid metabolism (vs carbohydrate); therefore, the observation and hypothesis proposed by the authors are unexpected and can be interesting. However, the experiments and data were rather preliminary and superficial and did not support the authors' conclusions. In addition, the main hypothesis, in relationship to the role of cold/temperature, remains incoherent and needs a major update.

Strengths:

A set of relatively novel and interesting observations.

Creative use of several in vivo sensors and techniques.

Weaknesses:

(1) The physiological relevance of lipolysis and thermogenesis genes in the PVH. The authors need to provide quantitative and substantial characterizations of lipid metabolism in the brain beyond a panel of qPCRs, especially considering these genes are likely expressed at very low levels. mRNA and protein level quantification of genes in Fig 1, in direct comparison to BAT/iWAT, should be provided. Besides bulk mRNA/protein, IHC/ISH-based characterization should be added to confirm to cellular expression of these genes.

(2) The fiberphotometry work they cited (Chen 2022, Andersen 2023, Sun 2018) used well-established, genetically encoded neuropeptide sensors (e.g., GRABs). The authors need to first quantitatively demonstrate that adapting BD-C11 and EnzCheck for in vivo brain FP could effectively and accurately report peroxidation and lipolysis. For example, the sensitivity, dynamic range, and off-time should all be calibrated with mass spectrometry measurements before any conclusions can be made based on plots in Figures 4, 5, and 6. This is particularly important because the main hypothesis heavily relies on this unvalidated technique.

(3) Generally, the histology data need significant improvement. It was not convincing, for example, in Figure 3, how the Fos+ neurons can be quantified based on the poor IF images where most red signals were not in the neurons.

(4) The hypothesis regarding the direct role of brain temperature in cold-induced lipid metabolism is puzzling. From the introduction and discussion, the authors seem to suggest that there are direct brain temperature changes in responses to cold, which could be quite striking. However, this was not supported by any data or experiments. The authors should consolidate their ideas and update a coherent hypothesis based on the actual data presented in the manuscript.

eLife assessment

This is an important study to reveal local circuit mechanisms in the POA that control body temperature and also highlight how neurotransmitter GABA and neuropeptide NTS from the same neurons differentially modulate temperature. This study was carefully executed, providing convincing evidence for the conclusions in this paper. The findings have emphasized the importance of considering multiple diverse functions of the same neuron populations and will be of interest to neuroscientists working on central regulations of energy metabolism and temperature homeostasis.

Reviewer #1 (Public Review):

Little is known about the local circuit mechanisms in the preoptic area (POA) that regulate body temperature. This carefully executed study investigates the role of GABAergic interneurons in the POA that express neurotensin (NTS). The principal finding is that GABA-release from these cells inhibits neighboring neurons, including warm-activated PACAP neurons, thereby promoting hyperthermia, whereas NTS released from these cells has the opposite effect, causing a delayed activation and hypothermia. This is shown through an elegant series of experiments that include slice recordings alongside matched in vivo functional manipulations. The roles of the two neurotransmitters are distinguished using a cell-type-specific knockout of Vgat as well as pharmacology to block GABA and NTS receptors. Overall, this is an excellent study that is noteworthy for revealing local circuit mechanisms in the POA that control body temperature and also for highlighting how amino acid neurotransmitters and neuropeptides released from the same cell can have opposing physiologic effects. I have only minor suggestions for revision.

Reviewer #2 (Public Review):

Summary:

The study has demonstrated how two neurotransmitters and neuromodulators from the same neurons can be regulated and utilized in thermoregulation.

The study utilized electrophysiological methods to examine the characteristics and thermoregulation of Neurotensin (Nts)-expressing neurons in the medial preoptic area (MPO). It was discovered that GABA and Nts may be co-released by neurons in MPO when communicating with their target neurons.

Strengths:

The study has leveraged optogenetic, chemogenetic, knockout, and pharmacological inhibitors to investigate the release process of Nts and GABA in controlling body temperature.

The findings are relevant to those interested in the various functions of specific neuron populations and their distinct regulatory mechanisms on neurotransmitter/neuromodulator activities

Weaknesses:

Key points for consideration include:

(1) The co-release of GABA and Nts is primarily inferred rather than directly proven. Providing more direct evidence for the release of GABA and the co-release of GABA and Nts would strengthen the argument. Further in vitro analysis could strengthen the conclusion regarding this co-releasing process.

(2) The differences between optogenetic and chemogenetic methods were not thoroughly investigated. A comparison of in vitro results and direct observation of release patterns could clarify the mechanisms of GABA release alone or in conjunction with Nts under different stimulation techniques.

(3) Neuronal transcripts were mainly identified through PCR, and alternative methods like single-cell sequencing could be explored.

(4) In Figure 6, the impact of GABA released from Nts neurons in MPO on CBT regulation appears to vary with ambient temperatures, requiring a more detailed explanation for better comprehension.

(5) The model should emphasize the key findings of the study.

Reviewer #3 (Public Review):

Summary:

Understanding the central neural circuits regulating body temperature is critical for improving health outcomes in many disease conditions and in combating heat stress in an ever-warming environment. The authors present important and detailed new data that characterizes a specific population of POA neurons with a relationship to thermoregulation. The new insights provided in this manuscript are exactly what is needed to assemble a neural network model of the central thermoregulatory circuitry that will contribute significantly to our understanding of regulating the critical homeostatic variable of body temperature. These experiments were conducted with the expertise of an investigator with career-long experience in intracellular recordings from POA neurons. They were interpreted conservatively in the appropriate context of current literature.

The Introduction begins with "Homeotherms, including mammals, maintain core body temperature (CBT) within a narrow range", but this ignores the frequent hypothermic episodes of torpor that mice undergo triggered by cold exposure. Although the author does mention torpor briefly in the Discussion, since these experiments were carried out exclusively in mice, greater consideration (albeit speculative) of the potential for a role of MPO Nts neurons in torpor initiation or recovery is warranted. This is especially the case since some 'torpor neurons' have been characterized as PACAP-expressing and a population of PACAP neurons represent the target of MPO Nts neurons.

eLife assessment

This technical study presents a novel sampling strategy for detecting synaptic coupling between neurons from dual pipette patch-clamp recordings in acute slices of mammalian brain tissue in vitro. The authors present solid evidence that this strategy, which incorporates automated patch clamp electrode positioning and cleaning for reuse with strategic neuron targeting, has the potential to substantially improve the efficiency of neuronal sampling with paired recordings. This technique and the extensions discussed will be useful for neuroscientists wanting to apply or already conducting automated multi-pipette patch clamp recording electrophysiology experiments in vitro for neuron connectivity analyses.