Loss of POMC-mediated antinociception contributes to painful diabetic neuropathy

Painful neuropathy is a frequent complication in diabetes. Proopiomelanocortin (POMC) is an endogenous opioid precursor peptide, which plays a protective role against pain. Here, we report dysfunctional POMC-mediated antinociception in sensory neurons in diabetes. In streptozotocin-induced diabetic mice the Pomc promoter is repressed due to increased binding of NF-kB p50 subunit, leading to a loss in basal POMC level in peripheral nerves. Decreased POMC levels are also observed in peripheral nervous system tissue from diabetic patients. The antinociceptive pathway mediated by POMC is further impaired due to lysosomal degradation of μ-opioid receptor (MOR). Importantly, the neuropathic phenotype of the diabetic mice is rescued upon viral overexpression of POMC and MOR in the sensory ganglia. This study identifies an antinociceptive mechanism in the sensory ganglia that paves a way for a potential therapy for diabetic neuropathic pain.

1. Fig.1b, POMC immunostaining is weak and almost covered by the red channel, which make it difficult to see POMC expression clearly. 2. Fig.3c, in the control panel, it seems that nearly all Tuj-1neurons express MOR in this picture, however the quantitative data shows only 60% neurons are MOR positive. The contrast setting of images may result in statistical errors. Can the authors provide raw confocal image files? Similar concern in Fig. 2d and Fig.s2. 3. Regarding to the specificity of POMC immunostaining in DRG, can the authors provide a positive/negative control of the antibody they used? For example, using the hypothalamic arcuate nuclei or the pituitary gland as a positive control and other brain areas as a negative control.
Reviewer #3 (Remarks to the Author): In this work, Deshpande and colleagues have identified a pain-promoting pathway which involves a reduction of proopiomelanocortin (POMC) signaling (at both the ligand and receptor levels) in hyperglycaemic conditions encountered in diabetic neuropathy. The authors trace back these changes to the periphery, namely the dorsal root ganglion (DRG) and associated axons innervating the skin. Rescuing the diabetes-induced downregulation of POMC signalling via viral overexpression of the ligand and/or its receptor in the DRG provided analgesic effects across a variety of pain modalities. The authors propose this mechanism could form the basis for new antinociceptive therapies in humans.
The study contains a significant amount of data, including extensive behavioural pain testing. The existing experiments are conducted carefully using all the necessary controls and the results are explained in detail. The addition of human samples further substantiates the data obtained in mice. The methods are written in sufficient detail to allow replication of the experiments. The statistical analysis employed for each comparison is denoted clearly and is appropriate. Overall, the paper is very well written -experiments follow a logical flow and the narrative is clear. The conclusions are supported by the experimental data and the findings are novel.
While the results are interesting and solid, there is some key information missing which could extend the applicability of the conclusions and make it more translational.
1) The authors focused on female diabetic mice on the basis that this sex develops thermal hypersensitivity (as opposed to hyposensitivity in males) at the examined time points after diabetes induction. Given the recently emerged knowledge on sexual dimorphism in pain responses, it is important to characterise this pathway a bit more with regards to sex. Obvious questions are: -Does the downregulation of POMC and MOR occur in the same way in male diabetic mice? -Does viral overexpression of POMC/MOR reverse pain behaviours in male diabetic mice? This may not be possible to assess for thermal hypersensitivity, but it is definitely an option for mechanical pain as well as spontaneous pain tests.
2) The authors conclude that the site of interest is the periphery where they detect downregulation of the ligand/receptor. The evidence for this seems clear. However we cannot exclude an additional site of action in the CNS, where MOR is also expressed. Is there evidence for downregulation of this pathway in the spinal cord? Viral overexpression in DRG has antinociceptive effects but can the authors exclude that a) ligand is transported to central terminals to act on postsynaptic MOR in the spinal cord b) the injected AAV does not transduce spinal cord neurons?
3) It would be very interesting to ascertain whether this pathway is specific to metabolic neuropathies such as diabetic neuropathy, or a more generalised mechanism that also extends to traumatic neuropathies. The authors could use a model such as spinal nerve ligation to quickly check whether POMC/MOR downregulation also occurs in the context. 4) While the PKC-mediated degradation of MOR is a plausible mechanism, there could be other pathways at play eg phosphorylation of ion channels. The authors should acknowledge this possibility in the discussion. Do PKA inhibitors also have an effect on MOR internalisation and/or diabetic pain? Some more minor points: While the authors provide a control staining (pre-incubation with antigen) to demonstrate the specificity of the POMC antibody, unfortunately this control is of limited use as it only shows that the antibody correctly recognises this antigen -it is however still possible that it also recognises distinct but similar proteins. The ideal control would be testing the antibody on tissue taken from POMC KO mice, which appear to be available (e.g. https://doi.org/10.1073/pnas.0306931101 and https://www.jax.org/strain/003191). If the authors get some of these mice, it would also be very interesting to check their baseline pain responses as well as how they develop diabetic pain.
The presented work identifies a pain-protective pathway that could be exploited for treatment of diabetic pain, but presumably the POMC/MOR downregulation is not the source of diabetic pain. It would be useful to have a very brief discussion of established and emerging candidate molecules that could be responsible for the pathogenesis of pain such as Nav1.8, HCN2, TRPV1, Cav, especially since there might be overlap of pathways involving cAMP/PKA/PKC (eg https://stm.sciencemag.org/content/9/409/eaam6072, https://www.nature.com/articles/nm.2750, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2735619/) In terms of applicability of results in the clinic, the authors could perhaps extend the discussion. How are these results likely to change the already established use of opioids for pain control in diabetes? Since we need to restore expression of ligand/receptor, would the proposed way forward be introducing copies of the protein via gene therapy?

Loss of POMC-mediated antinociception contributes to painful diabetic neuropathy
We are grateful to the editor and all of the reviewers for their encouraging and constructive comments for our initial manuscript. We have now performed a series of new experiments to address each comment successfully. The reviewers' suggestions have helped improve the study and expand the scope of the findings. Below, we have presented our new data in a point-by-point response. The newly added text is highlighted in the revised version of the manuscript. Author response: We thank the reviewer for the positive comments.

Reviewers
The reviewer raises a valid conceptual point regarding the importance of the POMC pathway in the holistic picture of painful diabetic peripheral neuropathy (DPN), which we discuss below and have included this important point in the Discussion section of the Revised Manuscript on pg 21.
DPN can be viewed as a disease consisting of a complex interplay of several aberrant mechanisms ultimately resulting in neuropathic pain during diabetes. Hyperexcitability of the injured sensory neurons generating action potentials in the absence of a 'painful' stimulus, is a major determinant underlying diabetic neuropathic pain 1,2 . At the subspinal level, this hyperexcitability may occur in two separate ways: Firstly, the altered expression or modification of ion channels on the nociceptors that results in 'gain-of-function', culminating in increased neuronal excitability. For instance, an increased expression of Na v 1.8 (ion channel amplifying action potential in neurons) in the sensory neurons of diabetic animal model has been reported 3 . Increased activity of Na v 1.8 4 and Ca v 3.2 5 due to glycation and glycosylation respectively has also been shown to promote painful DPN. Phosphorylation of Na v 1.8 by protein kinase C (PKC) is yet another mechanism known to underlie hypersensitivity in animal models of neuropathic pain 6 . We demonstrate increased levels of PKC in the DRG of diabetic mice (Revised Manuscript Figure 5a), which may contribute to increased Na v 1.8 activity linked with painful DPN.
We propose that the impaired POMC antinociceptive pathway constitutes the 'loss-offunction', which also promotes neuropathic pain phenotype. For the termination of a nociceptive response, inhibitory signalling is vital, the dysfunction of which can lead to persistent generation of action potential and result in hyperexcitability. This is shown to occur during diabetes due to reduced expression and activity of the inhibitory potassium ion channels K v an animal model of diabetes 7 .
The summation of the nociceptive inputs from the PNS neurons influences synaptic transmission within the spinal cord. Therefore, enhanced input from the persistently active nociceptors further amplifies the nociceptive signalling leading to central sensitization, a phenomenon observed during painful DPN 8 .
Downregulated POMC-mediated antinociception may exacerbate the local functional imbalance between facilitation and inhibition of the nociceptive signalling, which can collectively influence the ascending pain pathways during diabetes, a theme discussed in greater detail by Marshall A et al 9 , Tesfaye et al 10 among others.
The dysregulation of the POMC pathway, observed in this study, co-exists with other aberrant pathways, which would together promote hypersensitivity in DPN. However, we showed that reinstating the POMC-mediated antinociception helps overcome painful DPN demonstrating the importance of this inhibitory subspinal mechanism.

Reviewer 1, Comment 2:
Are the findings related to diabetes or to nerve lesion as such? I.e., would the same findings occur in a model of nerve injury or chemotherapy induced painful neuropathy?
Author response: We thank the reviewer for raising this crucial question.
According to our findings, the downregulation of POMC occurs in diabetes via NF-kB mediated promoter suppression, while elevated PKC levels underlie MOR degradation. Both NF-kB and PKC activation have been shown to be a consequence of metabolically-triggered changes during diabetes 11 .
For the current study, we employed a low-dose Streptozotocin (STZ) model, which does not cause direct neurotoxicity 12 . Using this mouse model, our group recently published a study on extensive characterization of morphological changes in the peripheral nerves of STZinduced mice, using a high-resolution magnetic resonance neurography. This study showed that there were no focal lesions in the peripheral nerves of the diabetic mice. The first occurrence of subtle ultrastructural changes were observed no earlier than 24 weeks post STZ-induction 13 . These changes were associated with a hypoalgesic phenotype in the diabetic mice. In the current study, our observations are made at no later than 12 weeks post STZ-induction and correlate with a hyperalgesic phenotype, making their association with metabolically triggered changes stronger than the ultrastructural changes in the PNS of our diabetic mouse model.
In order to answer the second part of the question, we performed molecular analysis of the DRG of mice with spared nerve injury (SNI) 14 . We measured POMC and MOR protein levels in the L3 and L4 DRG of the ipsilateral and contralateral side of SNI and sham-operated mice at 7 days post-operation. No changes were observed in the DRG of the sham-operated mice.
Surprisingly, POMC and MOR were both downregulated in the ipsilateral DRG of SNI mice (Reviewer Figure 1). The downregulation of MOR in the DRG after SNI has been previously reported by Wieskopf et al 15 . However, our preliminary findings show that the malfunction exists also upstream of the receptor, i.e. at the ligand level. To investigate the mechanisms underlying as the downregulation of POMC-MOR in SNI, is beyond the scope of this study. The well-studied chronic neuroinflammatory changes (including NF-kB [16][17][18] and PKC 19,20 activation) occurring in the DRG during SNI could be a contributing factor. These findings present an interesting basis for future studies with different focus.
Taken together, our preliminary findings in the SNI mouse model indicate that the POMCpathway may represent a more generalized mechanism and possibly be extended to other chronic neuroinflammatory conditions. Representative blots and densitometric quantification of POMC (∼ 26KDa band normalized to actin) and MOR protein (∼ 55KDa band normalized to actin) in total lysates of L3 and L4 DRG (n=4-5 per group; two tailed t-test). Data represents mean ± SEM. *p<0.05, **p<0.01, ***p<0.001.

Reviewer 1, Comment 3: Another major question is the point about hyperalgesia only occurring in female mice. If this is so, and if the POMC pathway is very important for
hyperalgesia, the downstream findings should be different in male mice. If the biochemistry is the same in male and female mice, and only female mice develop hyperalgesia, this casts some doubt on the causative connection. Furthermore, if the mechanism is only valid in females, the human data and the importance for male and female human diabetic patients should be discussed.

Contralateral Ipsilateral
Author response: We agree with the issue raised by the reviewer. To satisfactorily clarify this concern regarding the gender differences, we carried out extensive characterization of male diabetic mice.
In our initial manuscript, to determine the time point of maximum hypersensitivity, the male mice were tested for thermal hyperalgesia using Hotplate, which is a relatively broad measure. In our revised manuscript, we performed a thorough pain phenotyping over the course of time.
We examined the male mice for mechanical hyperalgesia (using Von Frey filaments) and thermal hyperalgesia (using Hargreaves) every 2 weeks post STZ-induction and compared them with age-matched healthy controls. In our setting using a low-dose STZ protocol, a brief thermal hypersensitivity phase was detectable using Hargreaves at 6 weeks, after which the male diabetic mice tended towards hypoalgesia (Reviewer Figure 2a, Revised Manuscript Figure S5). This small but significant difference in thermal hypersensitivity was detectable using Hargreaves, but not in the Hotplate method. The male mice displayed a comparatively prolonged period of mechanical hypersensitivity with a peak at 6 weeks post-induction (Reviewer Figure 2b, Revised Manuscript Figure S7a). The pain phenotyping we conducted for our initial manuscript was at 4, 8 and 12 weeks post-induction. We therefore had missed this brief phase of hyperalgesia at 6 weeks detectable using Hargreaves and von Frey filaments.
Having identified hypersensitivity peak at 6 weeks post-induction, we harvested the DRG and quantified POMC and MOR protein levels in the diabetic male mice. Consistent with our findings in the female diabetic mice, the POMC and MOR protein levels were significantly decreased in the DRG of male diabetic mice at this time point (Reviewer Figure 2c, Revised Manuscript Figure S7).
In terms of pain sensitivity, the disease progressed differently in both sexes over the course of time, with females tending towards hyperalgesia at 12 week post-induction, whilst the males displayed a hyperalgesic phase comparatively earlier. Though both genders displayed a neuropathic pain phenotype, we noted that the hypersensitivity (thermal and mechanical) was more pronounced in the female diabetic mice than in the male diabetic mice (Reviewer Figure 3, Revised Manuscript Figure 2a, 2b, S7a, S7b). Similar gender differences showing higher propensity of female sex to neuropathic pain and earlier onset of neuropathy in the males, have been previously reported in patients [21][22][23][24][25] and animal models of diabetic neuropathic pain 26,27 . To investigate why there is a gender-specific difference with respect to the DPN onset and progression is beyond the scope of this study. However, the key finding is that when both male and female mice display the hypersensitivity, there is a concurrent loss of POMC and MOR regardless of the gender. This would suggest that gender difference is not responsible for this downregulation. This is further supported by the observation that the decrease of POMC or MOR was observed in both, female and male patients. This issue has been addressed on pg 20 of the Revised Manuscript.
In addition, we were also able to show that overexpression of POMC-MOR in the male diabetic mice ameliorated thermal and mechanical hypersensitivity, as well as, the gait anomalies (Reviewer Figure 4, Revised Manuscript Figure S11, S12 and S13). This further demonstrated that the antinociceptive effect of the proposed therapeutic strategy is not gender-specific.  In our initial manuscript, to determine the time point of maximum hypersensitivity, the male mice were tested for thermal hyperalgesia using Hotplate, which is a relatively broad measure. In our revised manuscript, we performed a thorough pain phenotyping over the course of time.
We examined the male mice for mechanical hyperalgesia (using Von Frey filaments) and thermal hyperalgesia (using Hargreaves) every 2 weeks post STZ-induction and compared them with age-matched healthy controls. In our setting using a low-dose STZ protocol, a brief  Figure S5). This small but significant difference in thermal hypersensitivity was detectable using Hargreaves, but not in the Hotplate method. The male mice displayed a comparatively prolonged period of mechanical hypersensitivity with a peak at 6 weeks post-induction (Reviewer Figure 2b, Revised Manuscript Figure S7a). The pain phenotyping we conducted for our initial manuscript was at 4, 8 and 12 weeks post-induction. We therefore had missed this brief phase of hyperalgesia at 6 weeks detectable using Hargreaves and von Frey filaments.
Having identified hypersensitivity peak at 6 weeks post-induction, we harvested the DRG and quantified POMC and MOR protein levels in the diabetic male mice. Consistent with our findings in the female diabetic mice, the POMC and MOR protein levels were significantly decreased in the DRG of male diabetic mice at this time point (Reviewer Figure 2c, Revised Manuscript Figure S7).
In terms of pain sensitivity, the disease progressed differently in both sexes over the course of time, with females tending towards hyperalgesia at 12 week post-induction, whilst the males displayed a hyperalgesic phase comparatively earlier. Though both genders displayed a neuropathic pain phenotype, we noted that the hypersensitivity (thermal and mechanical) was more pronounced in the female diabetic mice than in the male diabetic mice (Reviewer Figure 3, Revised Manuscript Figure 2a, 2b, S7a, S7b). Similar gender differences showing higher propensity of female sex to neuropathic pain and earlier onset of neuropathy in the males, have been previously reported in patients [21][22][23][24][25] and animal models of diabetic neuropathic pain 26,27 . To investigate why there is a gender-specific difference with respect to the DPN onset and progression is beyond the scope of this study. However, the key finding is that when both male and female mice display the hypersensitivity, there is a concurrent loss of POMC and MOR regardless of the gender. This would suggest that gender difference is not responsible for this downregulation. This is further supported by the observation that the decrease of POMC or MOR was observed in both, female and male patients. This issue has been addressed on pg 20 of the Revised Manuscript.
Finally, we overexpressed the POMC-MOR bicistronic construct in the DRG of diabetic male mice using adeno-associated viral particles. We performed functional analyses for thermal and mechanical hypersensitivities, as well as, gait parameters at 6 weeks post STZinduction. We observed a significant amelioration of the evoked pain responses and associated gait anomalies in the mice overexpressing POMC and MOR protein, upon comparison with the diabetic male mice expression only GFP viral constructs (Reviewer Figure 4, Revised Manuscript Figure S11, S12 and S13).
Thus, taken together, despite time-related differences in the genders with respect to progression of painful DPN, neuropathic pain phenotype was observed in females and males, which could be rescued upon re-instating the dysfunctional POMC-MOR axis.

Reviewer 2, Comment 2: Since POMC expression is downregulated under diabetic conditions, it is also interesting to know whether POMC expression will be increased in response to acute pain stimulus. That is whether DRG POMC plays an antinociceptive role under normal conditions.
Author response: To address this question, we employed the capsaicin-induced acute pain model in healthy mice (Reviewer Figure 5, Revised Manuscript Figure S4). We injected capsaicin in one hindpaw (ipsilateral) and vehicle in the other (contralateral). We then quantified Pomc mRNA and protein levels in the DRG (site of POMC synthesis) and footpads (site of POMC proteolysis and release). We observed that capsaicin evoked acute nocifensive behavior, which lasted for ca. 5 minutes post injection (Reviewer Figure 5b), suggesting the resolution of the acute pain thereafter. At 30 minutes post injection, we observed significantly decreased POMC protein levels in the ipsilateral footpads and DRG as  Figure 5d). The gene expression, however, was unchanged at this time point (Reviewer Figure 5c). At 4 hours post injection, the Pomc Pomc mRNA in the ipsilateral DRG was significantly increased compared to the contralateral DRG. This was reflected in the POMC protein level, which was normalized (Reviewer Figure  5e). Taken together, these findings indicate that capsaicin induces an initial release of POMC peptides from the nerves around the time the acute pain is resolved, which is then replenished by an increased Pomc gene expression. Minor points: Fig.1b, POMC immunostaining is weak and almost covered by the red channel, which make it difficult to see POMC expression clearly. Fig.3c, Fig. 2d and Fig.s2.

Reviewer 2, Comment 3:
Author response: We agree with the reviewer's concerns and have replaced the mentioned figures with more representative images. The raw confocal image files are also provided. Please note that Fig. S2 of the initial manuscript is Fig. S1 in the revised manuscript.

Reviewer 2, Comment 4:
Regarding to the specificity of POMC immunostaining in DRG, can the authors provide a positive/negative control of the antibody they used? For example, using the hypothalamic arcuate nuclei or the pituitary gland as a positive control and other brain areas as a negative control.

Author response:
We are grateful to the reviewer for providing this suggestion and an opportunity to strengthen our study.
As recommended, we have performed the staining of brain sections, which shows that the anti-POMC antibodies detect signal in the neuronal somata of hypothalamus but not in the cortex region, thereby demonstrating their specificity (Reviewer Figure 6 and Revised Manuscript Figure S3a). Reviewer Figure 6: Specificity of the anti-POMC antibodies shown in immunofluorescence using mouse brain Naïve control mouse brain sections immunostained using either anti-POMC a Goat (Gt; #PA518368) antibody or Rabbit (Rb; #23499) antibody detect specific signal in hypothalamus region known to express POMC, whereas no signal in neuronal somata of the hippocampal region. Inset shows a magnified image of neuronal soma.
In addition, we were also able to obtain frozen DRG tissues from Pomc conditional knockout (cKO) mice lacking Pomc to further test the antibody specificity. This mouse carries a neuronal specific deletion of Pomc gene, generated using cre-lox system (Snap 25 cre x Pomc fl/fl). Upon immunostaining the DRG of the Pomc cKO mice and wild-type mice, we observed a loss of POMC immunoreactivity detectable in the wild-type DRG, confirming the specificity of the anti-POMC antibodies used in this study (Reviewer Figure 7a and Revised

Hypothalamus Cortex
Hypothalamus Cortex Manuscript Figure S3c). In addition, we also used total protein extracts from these tissues to verify the POMC antibody specificity in immunoblotting. The 26KDa band detectable in the total DRG lysates from the wild-type mice, was significantly diminished in the conditional knockout mice, not only validating the evidence for antibody specificity, but also corroborating that neurons are the major cell type to express POMC in the DRG under basal condition Reviewer Figure 7b and Revised Manuscript Figure S3d).

✱✱✱
Author response: We thank the reviewer for the kind words of appreciation and for providing suggestions that have helped expand the scope of the study.

Reviewer 3, Comment 1:
The authors focused on female diabetic mice on the basis that this sex develops thermal hypersensitivity (as opposed to hyposensitivity in males) at the examined time points after diabetes induction. Given the recently emerged knowledge on sexual dimorphism in pain responses, it is important to characterise this pathway a bit more with regards to sex. Obvious questions are: -Does the downregulation of POMC and MOR occur in the same way in male diabetic mice? -Does viral overexpression of POMC/MOR reverse pain behaviours in male diabetic mice? This may not be possible to assess for thermal hypersensitivity, but it is definitely an option for mechanical pain as well as spontaneous pain tests.
Author response: We agree with the reviewer and have conducted a series of new experiments to answer all the questions asked.
In our initial manuscript, to determine the time point of maximum hypersensitivity, the male mice were tested for thermal hyperalgesia using Hotplate, which is a relatively broad measure. In our revised manuscript, we performed a thorough pain phenotyping over the course of time.
We examined the male mice for mechanical hyperalgesia (using Von Frey filaments) and thermal hyperalgesia (using Hargreaves) every 2 weeks post STZ-induction and compared them with age-matched healthy controls. In our setting using a low-dose STZ protocol, a brief thermal hypersensitivity phase was detectable using Hargreaves at 6 weeks, after which the male diabetic mice tended towards hypoalgesia (Reviewer Figure 2a, Revised Manuscript Figure S5). This small but significant difference in thermal hypersensitivity was detectable using Hargreaves, but not in the Hotplate method. The male mice displayed a comparatively prolonged period of mechanical hypersensitivity with a peak at 6 weeks post-induction (Reviewer Figure 2b, Revised Manuscript Figure S7a). The pain phenotyping we conducted for our initial manuscript was at 4, 8 and 12 weeks post-induction. We therefore had missed this brief phase of hyperalgesia at 6 weeks detectable using Hargreaves and von Frey filaments.
Having identified hypersensitivity peak at 6 weeks post-induction, we harvested the DRG and quantified POMC and MOR protein levels in the diabetic male mice. Consistent with our findings in the female diabetic mice, the POMC and MOR protein levels were significantly decreased in the DRG of male diabetic mice at this time point (Reviewer Figure 2c, Revised Manuscript Figure S7).
In terms of pain sensitivity, the disease progressed differently in both sexes over the course of time, with females tending towards hyperalgesia at 12 week post-induction, whilst the males displayed a hyperalgesic phase comparatively earlier. Though both genders displayed a neuropathic pain phenotype, we noted that the hypersensitivity (thermal and mechanical) was more pronounced in the female diabetic mice than in the male diabetic mice (Reviewer Figure 3, Revised Manuscript Figure 2a, 2b, S7a, S7b). Similar gender differences showing higher propensity of female sex to neuropathic pain and earlier onset of neuropathy in the males, have been previously reported in patients [21][22][23][24][25] and animal models of diabetic neuropathic pain 26,27 . To investigate why there is a gender-specific difference with respect to the DPN onset and progression is beyond the scope of this study. However, the key finding is that when both male and female mice display the hypersensitivity, there is a concurrent loss of POMC and MOR regardless of the gender. This would suggest that gender difference is not responsible for this downregulation. This is further supported by the observation that the decrease of POMC or MOR was observed in both, female and male patients. has been addressed on pg 20 of the Revised Manuscript.
Finally, we overexpressed the POMC-MOR bicistronic construct in the DRG of diabetic male mice using adeno-associated viral particles. We performed functional analyses for thermal and mechanical hypersensitivities, as well as, gait parameters at 6 weeks post STZinduction. We observed a significant amelioration of the evoked pain responses and associated gait anomalies in the mice overexpressing POMC and MOR protein, upon comparison with the diabetic male mice expression only GFP viral constructs (Reviewer Figure 4, Revised Manuscript Figure S11, S12 and S13).
Thus, taken together, despite small differences in the genders with respect to progression of painful DPN, neuropathic pain phenotype was observed in females and males, which could be rescued upon re-instating the dysfunctional POMC-MOR axis.  36,37 , which causes MOR desensitization, degradation and induces opioid tolerance 38 .

Reviewer 3, Comment 2:
In our study, we demonstrated that the MOR protein level was decreased in the PNS of diabetic mice in the absence of any exogenous application of MOR agonists. This finding, along with the detection of the MOR ligand derived from POMC in the PNS was interesting, since this suggested that neuropathic pain could be possibly managed at the PNS level, without any adverse side-effects (such as tolerance and addiction) involving the CNS. Therefore, our main focus was the peripheral sensory neurons and the modulation of PNS intrinsic pathway to rescue painful DPN.
For this purpose, we have mainly used direct DRG injections in which, the AAV constructs were injected in L3 and L4 DRG only. Such a delivery technique allows viral expression in all neurons of the DRG and their axons projecting in the sciatic nerve, as well as, those in those going to the spinal cord. Previous studies have shown that virus injected using this technique is not transduced to the spinal neurons [39][40][41] . We have also used intrathecal delivery of the AAV particles by injecting at the level of L4-L5 region of the spinal cord (Revised manuscript Fig S10, S11 and S12). This delivery method has been shown to express the virus maximally in the DRG neurons, minimally in the spinal cord and negligibly in the brain 42 .
Nevertheless, given the expression of MOR in spinal cord and brain, and of POMC in hypothalamus, we agree that further exploration of this pathway in the CNS is interesting, although that would constitute a new study with a new focus.

Reviewer 3, Comment 3:
It would be very interesting to ascertain whether this pathway is specific to metabolic neuropathies such as diabetic neuropathy, or a more generalised mechanism that also extends to traumatic neuropathies. The authors could use a model such as spinal nerve ligation to quickly check whether POMC/MOR downregulation also occurs in the context.
Author response: In order to further explore the POMC-MOR mechanism in other neuropathies, we performed molecular analysis of the DRG of mice with spared nerve injury (SNI) 14 . We measured POMC and MOR protein levels in the L3 and L4 DRG of the ipsilateral and contralateral side of SNI and sham-operated mice at 7 days post-operation.
No changes were observed in the DRG of the sham-operated mice. Surprisingly, POMC and MOR were both downregulated in the ipsilateral DRG of SNI mice (Reviewer Figure 1). The downregulation of MOR in the DRG after SNI has been previously reported by Wieskopf et al 15 . However, our preliminary findings show that the malfunction exists also upstream of the receptor, i.e. at the ligand level. To investigate the mechanisms underlying as the downregulation of POMC-MOR in SNI, is beyond the scope of this study. The well-studied chronic neuroinflammatory changes (including NF-kB [16][17][18] and PKC 19,20 activation) occurring in the DRG during SNI could be a contributing factor. These findings present an interesting basis for future studies with different focus.
Taken together, our preliminary findings in the SNI mouse model show that the POMCpathway may represent a more generalized mechanism and possibly be extended to other chronic neuroinflammatory conditions. PKA, like PKC, is a cAMP-dependant kinase and is also activated during diabetes 11 . That PKA can phosphorylate ion channels leading to neuronal hyperexcitability has been shown 43,46,47 . PKA inhibitors have also been shown to improve painful DPN 48 .

Reviewer 3, Comment 4:
Within the context of the POMC signalling loop, PKA-mediated MOR phosphorylation occurs when MOR binds to its agonist peptide (homologous phosphorylation). During diabetes, we detected lowered ß-endorphin level, suggesting that heterologous phosphorylation is a more likely event. We also observed an increased phosphorylation at the threonine residues of MOR protein in the DRG of diabetic mice. Illing et al have shown that PKA does not participate in heterologous phosphorylation of threonine residues 49 .
As such, although contribution of PKA to neuropathic phenotype is possible, based on our observations, PKC is more relevant with respect to the POMC-MOR axis during diabetes.
Some more minor points: Author response: We concur with the reviewer on this issue. We have now included better controls showing anti-POMC antibody specificity. As suggested by reviewer 2, we have immunostained the hypothalamic region of the brain known to express POMC as a positive control and the cortex region as negative control. Using the anti-POMC antibodies, we observed a clear signal in the neuronal somata of the hypothalamic region, but not the cortex region, thereby showing the specificity of the anti-POMC antibodies (Reviewer Figure 6 and Revised Manuscript Figure S3a).
Since the Pomc KO mice suggested by the reviewer were available only as frozen embryos (Jackson), we were unable to obtain and resurrect them within the time frame of this revision period. Due to unavailability of the live Pomc KO or the cKO mice, we could not perform the pain measurements in these mice.
However, we were able to obtain frozen DRG tissues from Pomc conditional knockout (cKO) mice lacking POMC to test the antibody specificity. This mouse carries a neuron-specific deletion of Pomc gene, generated using cre-lox system (Snap 25 cre x Pomc fl/fl). Upon immunostaining the DRG of the Pomc cKO mice and wild-type mice, we observed a loss of POMC immunoreactivity detectable in the wt DRG, confirming the specificity of the anti-POMC antibodies used in this study (Reviewer Figure 7a and Revised Manuscript Figure  S3c). In addition, we also used these the tissues from these mice to verify POMC antibody specificity in immunoblotting. The 26KDa band detectable in the total DRG lysates of wild-type mice, was significantly diminished in the conditional knockout mice, not only validating the evidence for antibody specificity, but also corroborating that neurons are the major cell type to express POMC in the DRG under basal condition Reviewer Figure 7b and Revised Manuscript Figure S3d).
We are grateful to reviewer for providing this suggestion and an opportunity to strengthen our study.

Reviewer 3, Comment 6:
The presented work identifies a pain-protective pathway that could be exploited for treatment of diabetic pain, but presumably the POMC/MOR downregulation is not the source of diabetic pain. It would be useful to have a very brief discussion of established and emerging candidate molecules that could be responsible for the pathogenesis of pain such as Nav1. 8 Author response: We thank the reviewer for the above two suggestions, both of which have helped to emphasize the translational nature of the findings. This discussion suggested by the reviewer is given below and on pg 21 of the Revised manuscript.
Studies reporting novel disease mechanisms are currently driving the momentum of developing pathogenesis-oriented therapeutics for painful DPN. E.g. blocking the HCN2 channels using ivabradine 48 or scavenging methylglyoxal and preventing Na v 1.8 glycation 4 have shown promise in in vivo studies. Other emerging molecules include the selective sodium and calcium channel blockers, TRPA1 antagonists among others 50 . While most of these approaches target the excitatory ion channels/receptors, we propose a novel strategy of restoring body's natural defense mechanism of controlling pain signaling. It provides a platform for the development of new therapies by boosting the endogenous POMC synthesis and prevention of MOR degradation. Administration of Pomc promoter agonists, such as CRH (routinely administered in the patients with Cushing's disease of POMC deficiency) 51 or melanocortin receptor agonists into the spinal cord may be one of the possibilities to enhance the ß-endorphin level in the PNS. Other compounds, such as the melanocortin receptor agonists, which have shown to enhance POMC expression in patients with genetic POMC deficiency 52,53 , may be another promising option. Gene delivery techniques would also be a potential therapeutic in the future for patients with painful DPN. Lastly, in light of the ongoing opioid crisis, our study offers a promising alternative approach to counter peripheral neuropathic pain.