Tmem117 in AVP neurons regulates the counterregulatory response to hypoglycemia

  1. Sevasti Gaspari
  2. Gwenaël Labouèbe
  3. Alexandre Picard
  4. Xavier Berney
  5. Ana Rodriguez Sanchez‐Archidona
  6. Bernard Thorens

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

The counterregulatory response to hypoglycemia (CRR), which ensures a sufficient glucose supply to the brain, is an essential survival function. It is orchestrated by incompletely characterized glucose‐sensing neurons, which trigger a coordinated autonomous and hormonal response that restores normoglycemia. Here, we investigate the role of hypothalamic Tmem117 , identified in a genetic screen as a regulator of CRR. We show that Tmem117 is expressed in vasopressin magnocellular neurons of the hypothalamus. Tmem117 inactivation in these neurons increases hypoglycemia‐induced vasopressin secretion leading to higher glucagon secretion in male mice, and this effect is estrus cycle phase dependent in female mice. Ex vivo electrophysiological analysis, in situ hybridization, and in vivo calcium imaging reveal that Tmem117 inactivation does not affect the glucose‐sensing properties of vasopressin neurons but increases ER stress, ROS production, and intracellular calcium levels accompanied by increased vasopressin production and secretion. Thus, Tmem117 in vasopressin neurons is a physiological regulator of glucagon secretion, which highlights the role of these neurons in the coordinated response to hypoglycemia.

Article activity feed

  1. Version published to 10.15252/embr.202357344
  2. Review Commons

    Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    Reviewer #1:

    Major comments:

    1. In figure 4 and throughout the text they refer to AVP neurons as being "directly" inhibited by glucose but they do not show this. They do show membrane depolarization in response to decreased glucose form 2.5 to 0.1 mM; however this could easily be an indirect effect. To demonstrate that they directly sense glucose, it is necessary to block presynaptic input (generally done with the sodium channel blocker tetrodotoxin). In addition, the example recordings are not very good. The neurons seem to just decrease action potential frequency throughout the recording, despite the membrane depolarization. In the methods section, they state that cell resistance was measured but they do not report it. Looking at the traces, it appears that resistance does not change in response to decreased glucose which further supports a presynaptic action. In order to characterize these neurons as directly glucose sensing, they must use a blocker of presynaptic activity such as tetrodotoxin and the resistance values should be reported. They should also choose better representative traces.

    Answer: By “directly inhibited” we do not refer to a cell-autonomous effect but to a direct glucose responsiveness. Our cFOS experiments (Fig 3) and preexisting in vivo fiber photometry data (Kim et al., PMID: 34787082; Mandelblat-Cerf et al., PMID: 27989461) reported activation of AVP neurons in conditions of hypoglycemia. We further wanted to verify that the activation was triggered directly by the decreased glucose levels and not by accompanying systemic signals (for example insulin levels, stress response, etc). Furthermore, we aimed at identifying if inactivation of Tmem117 in AVP neurons affects their glucose responsiveness (at the circuit level). Therefore, we decided to take into account the synaptic inputs. Of course, this approach provides information at the circuit level (that we consider essential for CRR characterization) but is also accompanied with highly variable firing patterns. We focused our analysis on the membrane potential, as an indicator of cell excitability state that based on our previous experience was proven to be consistently affected in glucose sensing neurons regardless of cell-type, inputs or topology (Strembitska et al., PMID: 36180454; Kessler et al., PMID: 34622169; Quenneville et al., PMID: 32839348; Quenneville et al., PMID: 32839348; Labouèbe et al., PMID: 29218531; Steinbusch et al., PMID: 27422385; Labouèbe et al., PMID: 27322418; Lamy et al., PMID: 24606905). We used membrane resistance as an indicator of our patch-clamp quality. As specified in the Materials and Methods section (pages 33, lines 702-704), neurons with an access resistance >25 MΩ or changing by >20% during the recording were excluded from the analysis.

    We understand that the term “directly affected” has created some confusion since it could be interpreted as cell-autonomous effect. Therefore, we have exchanged the term as follows:

    • Results section (page 14, line 281): “To determine whether AVP neurons were directly sensitive to hypoglycemia and whether Tmem117 would modify this sensitivity” was converted to “To determine whether AVP neurons were sensitive to the decreased glucose availability and whether Tmem117 would modify this sensitivity”
    • Discussion section (page 27, lines 548-553): “Thus, AVP neurons can directly respond to hypoglycemia. Interestingly, a recent study reported that hypoglycemia can also activate AVP neurons indirectly, through afferent connections arising from GI neurons of the basolateral medulla (Kim et al., 2021). Thus, AVP neurons are part of a brainstem-hypothalamus neuronal circuit where hypoglycemia can be sensed by neurons located at multiple sites to activate the secretion of AVP leading to increased secretion of GCG.” was converted to “Thus, AVP neurons can respond to decreased glucose levels. Interestingly, a recent study reported that hypoglycemia can also activate AVP neurons through afferent connections arising from GI neurons of the basolateral medulla (Kim et al., 2021). Thus, AVP neurons are part of a brainstem-hypothalamus neuronal circuit where hypoglycemia can be sensed by neurons located at multiple sites to activate the secretion of AVP leading to increased secretion of GCG.”

    Regarding the membrane resistance measurements, 6 out of 7 AVPTM117KO GI cells and 6 out of 9 AVPTM117WT GI cells show a decrease in conditions of low glucose suggesting a cell autonomous effect, at least in a subpopulation of AVP GI neurons. We realized, based on the reviewer’s comment, that not including these data can lead to misinterpretation of the results, therefore we included the membrane resistance measurements in figure 4 and added the corresponding explanation in the Results section (page 15, lines 295-298).

    Regarding the representative traces, those included in the figure clearly show the increase in membrane potential that was consistent in all our GI cells. Choosing a representative trace for firing patterns is unrealistic given the high variability observed without the use of synaptic blockers.

    1. The relationship between the AVP/Tmem117 knockout and glucagon secretion is also not very convincing - especially in terms of the data showing that the effect seen at 1-week reverses by 3 weeks post knockdown in males when the neurons have died. First, the effect is very slight to begin with. Second, and more of a concern, the over secretion is still evident at the later time point, it just did not reach significance. This could have been due to the difference in sample size. Sample size in the early timepoint is 15-16 and in the late time point is reported as 8-9 (almost half the size), and in fact when I try to count data points, it really looks like the WT sample size is only 5. To make the conclusion that the effect disappears at the later time it is necessary to increase sample size so that the studies are equivalent. This is very important because these data are what is supporting the hypothesis that Tmem117 in AVP neurons regulates (inhibits) glucagon secretion. It is also concerning that the increased glucagon secretion in the knockout did not translate into differences in the depth of hypoglycemia raising concerns about physiological relevance.

    Answer: The sample size required based on our power calculations for these traits (plasma glucagon and copeptin) in male mice is 8 animals per group. Figure 2 consists of data merged from two individual experiments, both of which showed a significant effect. Figure 6 is the 3rd replication of our phenotypic observations with the addition of a later timepoint measurement. The number of animals included is 9 AVPTM117WT and 8 AVPTM117KO. Only for copeptin levels (panel F), as stated in the figure legend, the n is 8 mice per group, since for one of the AVPTM117WT mice the baseline levels of copeptin were undetectable by ELISA. Counting the datapoints in a 2D graph can cause misinterpretation since similar values can be depicted as overlapping points. We had decided to present in Fig 6 only the later timepoint that corresponds to the IF results reporting AVP cell death. We now realize based on the reviewer’s comment that this can cause misinterpretation of the phenotypic outcome, therefore we have exchanged graphs E-G (containing 2 timepoints) for the corresponding graphs that contain all 3 timepoints. We believe that with this addition the concern will be addressed since with the same sample size there is a significant increase in glucagon and copeptin secretion in AVPTM117KO mice at the early timepoint that is no longer evident at the later timepoint.

    Regarding the concern of effect size, we acknowledge the fact that we are reporting a small effect. But the CRR is controlled by a highly complex network of hypoglycemia activated neurons, which are located in different brain regions and act in an integrated manner to induce glucagon secretion through autonomic control, the HPA axis, and the secretion of AVP (reviewed in Steinbush et al., PMID: 26163755; Thorens, PMID: 34989155). Therefore, in this highly redundant and robust system, suppressing only one hypoglycemia sensing node, cannot be expected to have major consequence on the glucagon secretion. Nevertheless, we provide solid evidence that inactivation of Tmem117 in AVP neurons leads to a slight upregulation of AVP secretion that is accompanied by a subsequent increase in the glucose-stimulating hormone glucagon. To further highlight this observation, we added in figure 6 the correlation analysis between glucagon and copeptin levels of each tested sample that is reporting a significant positive correlation in both genotypes.

    Regarding the concern for physiological relevance, we would like to clarify that in this model we have not followed the reestablishment of normoglycemia over time. The timepoint of blood collection and glycemic measurement was selected based on our previous experience. We have performed multiple times full insulin tolerance tests in various mouse models, and we selected here (and in other papers: Picard et al., PMID: 35339728; Strembitska et al., PMID: 36180454) the one-hour time point because this is always the nadir of the glycemic curves, where CRR has the strongest effect in stimulating the secretion of CRR hormones. Glycemia at this time point depends on the insulin sensitivity of the animal. Thus, finding similar glycemia at 60 minutes after insulin injection indicates similar insulin sensitivity in AVPTM117WT and AVPTM117KO mice, further supporting that the increased glucagon secretion is directly correlated to the increased AVP secretion.

    1. It is also curious that the females in proestrus were similar to males but that females in other stages showed no effect. If, as mentioned in the discussion, estradiol was involved one would expect that males would be more similar to females in stages with low estradiol. The sex difference should be discussed more fully since it does seem to be one of the strong conclusions in the study

    Answer: We acknowledge that the discussion of the sex-specific phenotype was minimal in the initial version of our manuscript. We consider this finding very important for the general understanding of sex-differences in CRR, therefore we have included it in our main figures. But given that our observations remain in the phenotypic level our intention was to keep the discussion minimal and not add a lot of speculations that would not be supported by experimental data. We have performed all our further mechanistic analysis only in male mice and therefore we are not able to comment about this sex specific effect with evidence on the underlying mechanism. Of course, we have our literature-based explanation/hypothesis for the observed phenotype and given that this point was raised by all 3 reviewers, we have expanded this section of the discussion (pages 26, lines 520-542).

    1. The discussion needs to be adjusted in line with the findings.

    Answer: Several adjustments both in the Results and Discussion sessions have been incorporated based on the reviewers’ comments. Details on the adjustments for each point raised can be found in the corresponding answer.

    Minor comments:

    The BXD mice should be defined and the reason for their use explained

    Answer: We did not include further information in the first place because all the details and rational for the use of BXD mice were included in the original publication of this genetic screen (Picard et al., PMID: 35339728). We have now added further details on the origin of the BXD mouse lines accompanied with the corresponding reference (page 4, lines 83-85).

    The abbreviation cQTL needs to be defined at first use in the text of the results

    Answer: We added the abbreviation (page 5, line 102).

    Reviewer #2:

    Minor comments:

    1. Page 5, line 111: Fig1 shows Tmem117 expression in the hypothalamus. Tmem117 expression in AVP neurons in the PVN and SON are highlighted, however, AVP is not only expressed in these areas in the hypothalamus and in extra-hypothalamic areas. Was Tmem117 also expressed in AVP neurons beyond these areas?

    Answer: Based on our immunofluorescent detection Tmem117 is only observed in AVP neurons of the PVN and SON. This pattern (Fig 1A), the percentage of AVP-Tmem117 co-positive cells in these two areas (Fig 1H, I) and the presence of strong positive staining in the neurohypophysis (Fig 1B) points towards specific expression in magnocellular neurosecretory AVP cells. We do not observe staining in other AVP positive neurons. An example can be found in Fig 1A where Tmem117 staining is localized at the PVN and SON, but not at the Suprachiasmatic nucleus.

    1. Page 10, line 219: Fig2 shows CRR responses after conditional Tmem117 ko in AVP neurons. A sex-dependent difference was observed in the increased CRR after Tmem117 inactivation. A similar pattern between males and females during pro-estrus in copeptin and glucagon was observed. Was AVP expression and AVP neuron response to hypoglycemia also analyzed in both sexes, and in the case of females, were the results analyzed during the pro-estrus phase?

    Answer: Unfortunately, a detailed analysis of the mechanistic underpins of the estrus phase-dependent phenotype in female mice would exceed the spectrum of this initial characterization. We do believe that this finding is of great importance for the general understanding of sex-differences in CRR, therefore we have included it in our main figures. But our observations remain at the phenotypic level. We have performed all our further mechanistic analysis only in male mice. Of course, we have our literature-based explanation/hypothesis for the observed phenotype and given that this point was raised by all 3 reviewers, we have expanded this section of the discussion (pages 26, lines 520-542).

    1. Many of the experiments performed after Fig 3 (glucose responsiveness of AVP neurons, ER stress and ROS production, Ca2+ imaging) are mainly analyzed in the SON, but not the PVN. What is the rationale for analyzing only the SON in some experiments? Analyzing the PVN is of relevance due to its known role in CRR triggering.

    Answer: The reason for focusing our single-cell level analysis only in SON is its homogeneous population of AVP-Tmem117 double positive cells (Fig 1I). In the PVN about 10% of AVP neurons are negative for Tmem117 (Fig 1H). Therefore, we would be unable to distinguish between the cells with Tmem117 inactivation and those that did not express it at the first place. We agree with the reviewer that PVN is of great relevance for CRR induction, and this is why for all our in vivo phenotyping experiments we aimed at targeting both areas. Our AVPTM117KO mouse model is generated by injection of the AAV6-AVP-icre virus in the posterior pituitary and is targeting both regions (Fig S2E). This model was used for all our phenotypic analyses (Fig 2, 6) and for the fiber photometry recordings (Fig 5H-M). The Ca2+ signal we obtain is a collective output of AVP magnocellular terminals in the posterior pituitary deriving from both PVN and SON. But when it comes to the analysis of responses at the single-cell level, not having the ability to identify the subpopulation of Tmem117 negative cells could have a big impact on the results and their interpretation.

    1. Page 16, line 346: Again here, only BIP expression in AVP neurons is shown in the SON. I would be interested in looking at expression in AVP neurons in the PVN. Same goes for activation of these neurons (measured by Ca2+) after insulin administration.

    Answer: As mentioned above, we did not quantify these single-cell events in AVP neurons of the PVN due to the inability of effectively distinguishing between cells with Tmem117 inactivation and those that do not express Tmem117. But based on our supporting data (activation of PVN AVP cells in Fig 3 and decreased number of AVP PVN cells upon Tmem117 inactivation in Fig 6) we believe that the effect of Tmem117 inactivation is comparable between the Tmem117-positive AVP cells of both areas.

    1. What would be the role of Tmem117 in CRR triggering after hypoglycemia? Would its expression change during hypoglycemic conditions versus normoglycemia? Would that be different in models of pathology, like diabetes? Speculation of its physiological role in the discussion might be of help to visualize its role beyond inactivation.

    Answer: We focused our discussion strictly on the acquired results, but we agree with the reviewer than including speculations regarding the physiological role of Tmem117 would enhance the communication of our results. We have incorporated some extra points focusing on the possible role of Tmem117 expression in physiological conditions and pathological states in the Discussion section (pages 27-28, lines 562-571). Furthermore, we have included some extra data from microdissected PVN and SON tissue reporting downregulation of Tmem117 in the SON as a physiological response to insulin-induced hypoglycemia (Fig 3G-I).

    Reviewer #3:

    Major comments:

    1. The in vivo calcium data is not convincing (e.g. Figure 5j). Why is the activity not similar to that normally seen from populations of neurons (and in particular AVP neurons) in vivo - bursting/spikes. This is just a background increase in calcium... not an increase in activity?

    Answer: Our aim with the fiber photometry experiments was not to detect the activity of AVP neurons but to evaluate the Ca2+ fluctuations in AVP neuronal terminals, directly at the site of AVP exocytosis. The activity of AVP neurons in conditions of hypoglycemia has been previously detected with similar techniques at the level of cell bodies in the SON (Kim et al., PMID: 34787082; Mandelblat-Cerf et al., PMID: 27989461) and was further supported by our cFOS and ephys experiments (Fig 3, 4). These experiments showed that inactivation of Tmem117 did not affect the activation of AVP neurons in conditions of hypoglycemia. But our phenotypic characterization revealed higher secretion of AVP (Fig 2) and our in vitro (Fig S3) and in vivo (Fig 5A-G) data demonstrated increased ER stress and ROS in AVPTM117KO cells. Therefore, we wanted to assess if this increased stress is associated with increased Ca2+ at the level of neuronal terminals which would enhance AVP exocytosis. The acquired signal is a collective output of AVP magnocellular neurons from PVN and SON and although it is not resembling the pattern observed from cell body recordings, we believe it provides important information for the output of this circuit in conditions of hypoglycemia. Our results do not only report that Tmem117 inactivation is accompanied by an increase in intracellular Ca2+ at the level of nerve terminals/exocytosis, but further demonstrate (for the first time to our knowledge) that the activity reported in conditions of hypoglycemia is linked, as expected, to an increase in local Ca2+ levels at the posterior pituitary. All the essential controls for normalization and interpretation have been taken into account as mentioned in the methods section (page 31, lines 655-668). GCaMP7 signal recordings are corrected for artifacts and photobleaching. The isosbestic channel is subtracted to correct for non-GCaMP7 related signal artifacts and then at a second stage (during the DF/F0 normalization) the traces are corrected for photobleaching by the running average algorithm. Furthermore, 60min recordings of mice injected with saline are used as control for determining the increased signal after insulin injection. We have now added a clarification on the origin of the recorded signal in the Results section (page 18, line 368).

    1. Why would this response be oestrus cycle dependent? This is given very little thought in the manuscript.

    Answer: We acknowledge that the sex-specific phenotype observed was not thoroughly discussed in the initial version of our manuscript. We included the data in a main figure since we consider them as an important piece of information supporting the crucial role of sex hormones in the regulation of CRR. On the other hand, since our observations remain in the phenotypic level, we thought it would be better to keep the discussion on the topic short and not include various literature-based hypothesis that would not be supported by experimental data. We have performed all our further mechanistic analysis only in male mice and therefore we are not able to comment about this sex specific effect with evidence on the underlying mechanism. We have now realized based on the reviewers’ comments that further discussion on the topic is needed, therefore we have expanded this section of the Discussion (page 26, lines 520-542).

    Minor comments:

    1. How does the regulation make sense in the context of AVPs normal role in whole body physiology?

    Answer: We focused our discussion strictly on the acquired results, but it has become evident based on the reviewers’ comments that a discussion of the broad picture regarding the role of AVP and Tmem117 in physiological and pathological conditions would strengthen the interpretation of our results. We have now incorporated some extra points focusing on the possible role of Tmem117 expression in physiological conditions and pathological states in the Discussion section (pages 27-28, lines 562-571).

    1. What do the authors think about the role of AVP in islets, some authors suggest it actually has the opposite effect.

    Answer: The literature on the role of AVP on pancreatic islets shows indeed some controversies. A couple of studies suggest an action on beta cells and insulin secretion without any effect on glucagon secretion. On the other hand, the evidence on the stimulation of glucagon secretion is supported by numerous studies over the years and has been convincingly demonstrated with various ex vivo and in vivo models. Our study, in line with these findings, verifies the increased AVP secretion in conditions of hypoglycemia and reports a positive correlation between the levels of circulating copeptin and glucagon (Fig 6H).

    1. Why have the authors left the recent study by Kim, Rorsman and colleagues et al. 2022 (eLife) to the discussion? This manuscript is an important piece in this field and should be introduced in the introduction.

    Answer: We have now added the AVP induced glucagon secretion, accompanied by the supporting references (including the reference mentioned by the reviewer) as integral part of the CRR in our Introduction section (pages 3-4, lines 68-71).

    1. The intro should also mention that islets also can intrinsically regulate glucagon release.

    Answer: We focused our introduction exclusively to the brain triggered glucagon secretion, but in accordance with the reviewer’s comment we have now incorporated the intra-islet regulation and supporting literature in the Introduction (page3, line 55).

    1. Has tmem117 shown up in any human screens of hypogylcaemia risk? Do tmem117 variants show in any osmoregulation or insipidus risk?

    Answer: Not to our knowledge.

    1. Please clarify what N= is? What is a statistical unit in each case in the graphs. It is not clear when it is a mouse, cell or experiment. This should be justified in each case.

    Answer: We have now justified the corresponding values in all figure legends.

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #3

    Evidence, reproducibility and clarity

    Glucagon secretion counters hypogylcaemia, and this is in part regulated by the CNS.

    The authors investigated the role of hypothalamic Tmem117, which was previously identified in a genetic screen as a regulator of the counter-regulatory response to hypoglycaemia. They show that Tmem117 is expressed in vasopressin neurons and that inactivation of Tmem117 in AVP neurons increases hypoglycemia-induced vasopressin secretion leading to higher glucagon secretion. They show this is oestrus cycle phase-dependent.

    The manuscript is generally well written and the results of good quality. The findings are important.

    I have some comments:

    Major:

    1. The in vivo calcium data is not convincing (e.g. Figure 5j). Why is the activity not similar to that normally seen from populations of neurons (and in particular AVP neurons) in vivo - bursting/spikes. This is just a background increase in calcium... not an increase in activity?
    2. Why would this response be oestrus cycle dependent? This is given very little thought in the manuscript.

    Minor:

    1. How does the regulation make sense in the context of AVPs normal role in whole body physiology?
    2. What do the authors think about the role of AVP in islets, some authors suggest it actually has the opposite effect.
    3. Why have the authors left the recent study by Kim, Rorsman and colleagues et al. 2022 (eLife) to the discussion? This manuscript is an important piece in this field and should be introduced in the introduction.
    4. The intro should also mention that islets also can intrinsically regulate glucagon release.
    5. Has tmem117 shown up in any human screens of hypogylcaemia risk? Do tmem117 variants show in any osmoregulation or insipidus risk?
    6. Please clarify what N= is? What is a statistical unit in each case in the graphs. It is not clear when it is a mouse, cell or experiment. This should be justified in each case.

    Significance

    This is important, but some of the in vivo calcium data is weak. The manuscript is also riddled with self-citations and doesn't really put the results into a broader context. In addition, the finding is in a sense piggybacking off a recent revelation in the field (AVP neuron importance) using a technique that is available to the authors (Tmem117 manipulation) rather than trying to address the hypothesis/question with THE BEST approach to answer the question. The question feels like: "how can we use our mouse model to probe this AVP neuron hypothesis ourselves". Strange. The finding that the results are only relevant during a particular phase of the oestrus cycle feels a bit random and lacks adequate explanation. I still like the manuscript very much, the experiments are very technically challenging, the results are very interest, the data (mostly) of excellent quality, and the relevance to the field highly important, but I feel there is still room for improvement (unlikely more experiments, definitely some thinking and rewritting).

  4. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    The present study led by Gaspari, et al shows a new role for Tmem117 as a major regulator in the CRR. The authors used a variety of techniques to demonstrate that Tmem117 innactivation modulates hypoglycemia-induced responses by regulating AVP action in the hypothalamus. I found the rationale of the study interesting, the results seem to support the author's claims and conclusions seem to be proper. However there are some minor issues that needs to addressed:

    1. Page 5, line 111: Fig1 shows Tmem117 expression in the hypothalamus. Tmem117 expression in AVP neurons in the PVN and SON are highlighted, however, AVP is not only expressed in these areas in the hypothalamus and in extra-hypothalamic areas. Was Tmem117 also expressed in AVP neurons beyond these areas?
    2. Page 10, line 219: Fig2 shows CRR responses after conditional Tmem117 ko in AVP neurons. A sex-dependent difference was observed in the increased CRR after Tmem117 inactivation. A similar pattern between males and females during pro-estrus in co-peptin and glucagon was observed. Was AVP expression and AVP neuron response to hypoglycemia also analyzed in both sexes, and in the case of females, were the results analyzed during the pro-estrus phase?
    3. Many of the experiments performed after Fig 3 (glucose responsiveness of AVP neurons, ER stress and ROS production, Ca2+ imaging) are mainly analyzed in the SON, but not the PVN. What is the rationale for analyzing only the SON in some experiments? Analyzing the PVN is of relevance due to its known role in CRR triggering.
    4. Page 16, line 346: Again here, only BIP expression in AVP neurons is shown in the SON. I would be interested in looking at expression in AVP neurons in the PVN. Same goes for activation of these neurons (measured by Ca2+) after insulin administration.
    5. What would be the role of Tmem117 in CRR triggering after hypoglycemia? Would its expression change during hypoglycemic conditions versus normoglycemia? Would that be different in models of pathology, like diabetes? Speculation of its physiological role in the discussion might be of help to visualize its role beyond inactivation.

    Significance

    The study of Gaspari, et al proposes a new player in the hypoglycemia-induced CRR, specially affecting AVP modulation of CRR. The results and conclusions are of relevance to the field, as the mechanisms mediating hypoglycemia-induced CRR and how these are affected during pathology are not completely understood. The study is relevant in the metabolic field and for a neuroendocrinology public.

    Field of expertise: Neuroendocrinology, hypoglycemia, diabetes.

  5. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    In this study the authors convincingly demonstrate that the hypothalamic protein Tmem117 protects against elevated ROS and cell death in AVP neurons, and that knocking out Tmem117 in these neurons leads to cell death. However, the major thrust of this paper was to demonstrate that Tmem117, in AVP neurons plays a role in hypoglycemia counter regulation (per the title). The data supporting this hypothesis are much less convincing.

    Major comments

    There are 2 major findings of this study that are problematic (#1 and 2 below). Further experiments are needed to support their conclusions.

    1. In figure 4 and throughout the text they refer to AVP neurons as being "directly" inhibited by glucose but they do not show this. They do show membrane depolarization in response to decreased glucose form 2.5 to 0.1 mM; however this could easily be an indirect effect. To demonstrate that they directly sense glucose, it is necessary to block presynaptic input (generally done with the sodium channel blocker tetrodotoxin). In addition, the example recordings are not very good. The neurons seem to just decrease action potential frequency throughout the recording, despite the membrane depolarization. In the methods section, they state that cell resistance was measured but they do not report it. Looking at the traces, it appears that resistance does not change in response to decreased glucose which further supports a presynaptic action. In order to characterize these neurons as directly glucose sensing, they must use a blocker of presynaptic activity such as tetrodotoxin and the resistance values should be reported. They should also choose better representative traces.
    2. The relationship between the AVP/Tmem117 knockout and glucagon secretion is also not very convincing - especially in terms of the data showing that the effect seen at 1-week reverses by 3 weeks post knockdown in males when the neurons have died. First, the effect is very slight to begin with. Second, and more of a concern, the over secretion is still evident at the later time point, it just did not reach significance. This could have been due to the difference in sample size. Sample size in the early timepoint is 15-16 and in the late time point is reported as 8-9 (almost half the size), and in fact when I try to count data points, it really looks like the WT sample size is only 5. To make the conclusion that the effect disappears at the later time it is necessary to increase sample size so that the studies are equivalent. This is very important because these data are what is supporting the hypothesis that Tmem117 in AVP neurons regulates (inhibits) glucagon secretion. It is also concerning that the increased glucagon secretion in the knockout did not translate into differences in the depth of hypoglycemia raising concerns about physiological relevance.
    3. It is also curious that the females in proestrus were similar to males but that females in other stages showed no effect. If, as mentioned in the discussion, estradiol was involved one would expect that males would be more similar to females in stages with low estradiol. The sex difference should be discussed more fully since it does seem to be one of the strong conclusions in the study
    4. The discussion needs to be adjusted in line with the findings.

    Minor comments:

    The BXD mice should be defined and the reason for their use explained
    The abbreviation cQTL needs to be defined at first use in the text of the results

    Significance

    The observation that Tmem117 is protective against ROS and cell death is interesting. However, the other conclusions are not supported by the data which decreases the overall significance of the study

  6. Version published to 10.1101/2022.10.21.513159v3 on bioRxiv
  7. Version published to 10.1101/2022.10.21.513159v2 on bioRxiv
  8. Version published to 10.1101/2022.10.21.513159v1 on bioRxiv