Interleukin-36 cytokines alter the intestinal microbiome and can protect against obesity and metabolic dysfunction

Members of the interleukin-1 (IL-1) family are important mediators of obesity and metabolic disease and have been described to often play opposing roles. Here we report that the interleukin-36 (IL-36) subfamily can play a protective role against the development of disease. Elevated IL-36 cytokine expression is found in the serum of obese patients and negatively correlates with blood glucose levels among those presenting with type 2 diabetes. Mice lacking IL-36Ra, an IL-36 family signalling antagonist, develop less diet-induced weight gain, hyperglycemia and insulin resistance. These protective effects correlate with increased abundance of the metabolically protective bacteria Akkermansia muciniphila in the intestinal microbiome. IL-36 cytokines promote its outgrowth as well as increased colonic mucus secretion. These findings identify a protective role for IL-36 cytokines in obesity and metabolic disease, adding to the current understanding of the role the broader IL-1 family plays in regulating disease pathogenesis.

The paper is interesting and sheds light on a class of molecules underexplored in context of metabolic diseases and microbiota. However, in my view there are several controls that are missing, a better description of the glucose phenotype is needed, and the evidence concerning the mechanism is too descriptive.
Main points: -What are the organs responsible for the improved glucose phenotype in the Il36rn KO mice? Hyperinsulinemic-euglycemic clamp, or at least a 2-deoxyglucose tracer studies coupled with pAKT measurements could help elucidating this. The insulin sensitivity should be also investigated in the mice on chow diet (shown in Figure 2).
-To support the claims that the elevated IL36 levels and signalling are important in regulating the metabolic homeostasis, the IL36 alpha, beta and gamma need to be supplemented alone, or in combination in the WT and in the KO mice. This is a critical experiment that could also suggest therapeutic relevance of these findings.
-Authors dedicate substantial effort in speculating the mechanisms related to the IL36, but this is not sufficiently addressed experimentally. Does supplementing WT mice with IL-9 promote a KO-like phenotype on the abundance of Akkermansia? Or conversely, does CD4+ depletion in the KO mice reduce mucin (and therefore A. muciniphila's) levels? -The Verucomicrobia in the WT mice shown in Fig. 4c is almost not present. This is surprising, and is not consistent with the literature. These inconsistencies might explain the remarkable increase of the Verucomicrobia abundance (> 1600 fold) in the KO mice. These results need to be confirmed in an independent groups of mice (and preferably also at different ages). The interpretation of these results is difficult without specifying the exact age of the mice -and this should be done in all the figures throughout the study.
-The colon mucin layer should be quantified in the WT vs. the KO mice.
Additional points: -Authors need to confirm that the Il36rn KO mice indeed show increased IL36 downstream signalling.
-The study suggests that there is no difference of the macrophage subsets per gram tissue. Notably, despite the lack of calculated significance, in some cases there more than a double fold difference. Based on what power calculation were these animal numbers chosen? Each value should be presented as a data point, and I suggest increasing the animal number (coupled to power calculations) to allow proper interpretation of these results. -In relation to the above comments -there is a large discrepancy between the levels of A. muciniphila detected in Fig 4D,E with the ones in figure 6D. Is there an error in this figure?
Reviewer #2 (Remarks to the Author): This study identified that IL-36 may protect against metabolic dysfunction in both human and mice. Mechanistically, the authors suggested that the protective effects are dependent on the metabolically protective bacteria A. muciniphila. Overall, this manuscript reported an interesting and novel hostmicrobiota interaction that promotes metabolic health. However, there are some moderate concerns that the authors could address.
1. In Fig. 4E, the A. muciniphila level at the various time points in WT and IL-36rn-/-mice without IL-36Ra treatment should be provided as controls. This is particularly important as rIL-36Ra injections seemed to bring up the A. muciniphila level in WT mice (the opposite of what you would expect from their results).
2. Fig. 6 co-housing result provided a strong positive association of A. muciniphila level and the metabolically protective phenotype. To prove a causal effect of A. muciniphila, the authors could gavage mice with A. muciniphila and examine whether this approach can produce protective phenotype. This would be the most convincing data for their claim.

Response to Reviewers' comments:
We thank the reviewers for their insightful and helpful comments and suggestions. Based upon their recommendations, we have undertaken considerable additional experimentation which we believe significantly improves our study through a deeper mechanistic understanding of how IL-36 alters the colonic microenvironment to facilitate protection from disease, inclusion of improved control experiments, and an analysis and discussion of possible confounding factors. All individual points raised are addressed below in a point for point fashion.

Reviewer #1:
The paper is interesting and sheds light on a class of molecules underexplored in context of metabolic diseases and microbiota. However, in my view there are several controls that are missing, a better description of the glucose phenotype is needed, and the evidence concerning the mechanism is too descriptive.
Author response: We thank the reviewer for their interest in our paper and their extremely helpful critique of our study. We have carried out extensive new experimentation aimed at addressing the specific critiques and recommendations. These are outlined in a point for point fashion below.
Main points: -What are the organs responsible for the improved glucose phenotype in the Il36rn KO mice? Hyperinsulinemic-euglycemic clamp, or at least a 2-deoxyglucose tracer studies coupled with pAKT measurements could help elucidating this. The insulin sensitivity should be also investigated in the mice on chow diet (shown in Figure 2).
Author response: As suggested, we have included data demonstrating improved insulin sensitivity in aged mice, now shown in Figure 2f & g. We agree with the reviewer that identifying the organs responsible for the improved glucose phenotype observed would add to the study. Although, we were unable to carry out clamp studies and 2-deoxyglucose tracer studies as suggested, due to technical constraints and a lack of ethical and regulatory approval for such studies, we did examine relative insulin sensitivity in wt and il36rn-/ mice subjected to HFD by examining levels of p-Akt in the livers of mice 10 mins after insulin challenge. These new data, included as Figure 3i, demonstrate that insulin sensitivity is indeed increased in the livers of Il36rn-/-mice under these conditions.
-To support the claims that the elevated IL36 levels and signalling are important in regulating the metabolic homeostasis, the IL36 alpha, beta and gamma need to be supplemented alone, or in combination in the WT and in the KO mice. This is a critical experiment that could also suggest therapeutic relevance of these findings.
Author response: We agree with the reviewer that supplementation with IL-36 cytokines should be investigated to support the manuscript and point towards therapeutic relevance. Indeed, as suggested, we carried out further experimentation to determine whether the administration of recombinant IL-36 cytokines (a combination of IL-36, , and  (500ng of each/ per mouse)) in 5 doses, every other day for 8 days, would improve glucose tolerance in wt mice which had been on a HFD for 12 weeks. The dose of recombinant cytokine administered was chosen based upon recent studies in the literature (Scheibe, K. et al., Gastroenterology, 2019). However, we were unable to demonstrate any effect on glucose tolerance in IL-36 ligand treated mice, over that observed for control mice in these experiments (see figure 1 attached with this rebuttal below). While this lack of an effect is disappointing, it may stem from a relatively short half-life of IL-1 family ligands in vivo, such has been reported for IL-18 and IL-1 (Hosohara, K et al., Clin Diagn Lab Immunol. 2002)(Kudo, S. et al., Cancer Res, 1990). In addition, it is conceivable that administered cytokines would have to be directly delivered to the colonic mucosa in order to alter the tissue microenvironment there, and influence the pathogenesis of metabolic disease in mice.
-Authors dedicate substantial effort in speculating the mechanisms related to the IL36, but this is not sufficiently addressed experimentally. Does supplementing WT mice with IL-9 promote a KO-like phenotype on the abundance of Akkermansia? Or conversely, does CD4+ depletion in the KO mice reduce mucin (and therefore A. muciniphila's) levels?
Author response: We agree with the reviewers assertion that the mechanisms described in our study could have been addressed more fully experimentally. Guided by their suggestions, we have undertaken significant new experimentation to improve this central part of our study. Firstly, in order to further investigate the role of IL-9 in mediating this phenotype we chose to administer an anti-IL-9 neutralising mAb to determine whether this approach would reverse the observed effects on mucin levels/goblet cell numbers. We chose this approach, ahead of administering recombinant IL-9 as suggested, given our lack of success with IL-36,, supplementation as described above. Importantly, administration of this Ab did not alter the relative goblet cell hyperplasia observed in Il36rn-/-mice suggesting that elevated IL-9 does not contribute significantly to the phenotype observed. In contrast, administration of recombinant IL-36Ra, not only reversed the outgrowth of A. muciniphila, but also reversed the increased numbers of goblet cells observed in Il36rn-/mice. Furthermore, in extension of these observations, we have also now demonstrated that IL-36 supplementation in vitro led to significant increase in Muc2 gene expression in wt colon explant cultures after 4hrs. We believe that these new data add significantly to the mechanistic analysis component of our study and indicate a direct role for IL-36 in mediating these effects as opposed to indirectly through increased IL-9. These new data are now shown as an additional figure 6 in the revised manuscript. Given the apparent lack of a role for IL-9, this data is now shown in the new supplementary Figure 4. In addition, we have now also demonstrated that Il36rn-/-mice display enhanced intestinal barrier function consistent with the observed changes in the colonic tissue environment (now Fig. 5f).
-The Verucomicrobia in the WT mice shown in Fig. 4c is almost not present. This is surprising, and is not consistent with the literature. These inconsistencies might explain the remarkable increase of the Verucomicrobia abundance (> 1600 fold) in the KO mice. These results need to be confirmed in an independent groups of mice (and preferably also at different ages). The interpretation of these results is difficult without specifying the exact age of the mice -and this should be done in all the figures throughout the study.

Authors Response:
We thank the reviewer for their suggestion to specify the exact age of mice included in the studies and we have now addressed this at relevant points throughout the manuscript. We have also analysed the relative abundance of the Verrucomicrobia strain, A.muciniphila by qPCR across multiple cohorts of mice and a range of ages as suggested (see Figure 2 attached to this rebuttal). We agree that we observe a low abundance of Verrucomicrobia in the faeces of WT mice in our facility. Our multiple analyses have consistently shown that the levels of Verrucomicrobia, analysed by 16S RNA sequencing, and A.muciniphila analysed by qPCR, are found in relatively low abundance in wild type mice in our facility. A similar recent study demonstrated that Il33-/-mice have an approx. 2,500 fold increase in relative abundance in A.muciniphila in the intestinal microbiome (Malik, A, et al. JCI 2016). While this study did not specify the exact levels of abundance of Verrucomicrobia in WT mice, their data highlights that IL-1 family members can influence the relative abundance this bacterial phylum to a similarly high degree. Although there are reports of higher abundance in the literature, several further analyses are also consistent with our observations of abundance lower than 0.5% (Seregin, S. et al. Cell Rep., 2017). (Xiaofei X, Micro. Research, 2015)(Langille MG, Microbiome, 2014. Moreover, the relative abundance of specific phyla detected in the intestinal microbiome of mice is recognized as being influenced by the specific facility in which they are housed (Ericsson, A.C. et al., Sci Rep. 2018, Rausch, P. et al., Int J Med Microbiol. 2016. While this is a possible confounding factor in our study, as well as in all microbiome research, it is unfortunately something which could not be addressed within the scope of this study.
-The colon mucin layer should be quantified in the WT vs. the KO mice.
Authors Response: We thank the reviewer for this suggestion, which is directly relevant to our microbiome data. We have now carried out this analysis and included new data as Figure. 5 b&d.
Additional points: -Authors need to confirm that the Il36rn KO mice indeed show increased IL36 downstream signalling. Authors Response: As suggested we have confirmed that signalling pathways downstream of IL-36 are indeed increased in the colons of il36rn-/-mice. These data are now included as Figure 4b.
-The study suggests that there is no difference of the macrophage subsets per gram tissue. Notably, despite the lack of calculated significance, in some cases there more than a double fold difference. Based on what power calculation were these animal numbers chosen? Each value should be presented as a data point, and I suggest increasing the animal number (coupled to power calculations) to allow proper interpretation of these results.

Authors Response:
We thank the reviewer for this suggestion. The animal numbers chosen for these experiments (n=5) were based upon our experience in carrying out the same types of analysis of macrophage subsets in adipose tissue as previously reported in Hams, E. et al. FASEB Journal, 2016. We have now presented data as individual data points, which as suggested, allows a clearer interpretation of the results.
-In relation to the above comments -there is a large discrepancy between the levels of A. muciniphila detected in Fig 4D,E with the ones in figure 6D. Is there an error in this figure?
Authors Response: We apologise for the error made in the scale on the Y axis in Figure 4E as presented in the original manuscript. We thank the reviewer for bringing this to our attention. We have now reanalysed this qPCR data alongside additional controls and corrected the scale in new figure 4f. We have also included new data to original Figure 6d to demonstrate the temporal loss of A.muciniphila outgrowth after cohousing using correct scales. See new Figure 7d.

Reviewer #2:
This study identified that IL-36 may protect against metabolic dysfunction in both human and mice. Mechanistically, the authors suggested that the protective effects are dependent on the metabolically protective bacteria A. muciniphila. Overall, this manuscript reported an interesting and novel host-microbiota interaction that promotes metabolic health. However, there are some moderate concerns that the authors could address.
1. In Fig. 4E, the A. muciniphila level at the various time points in WT and IL-36rn-/-mice without IL-36Ra treatment should be provided as controls. This is particularly important as rIL-36Ra injections seemed to bring up the A. muciniphila level in WT mice (the opposite of what you would expect from their results).

Authors Response:
We thank the reviewer for this important comment. As suggested, we have now included control treatment conditions, and increased animal numbers, to new figure 4f, which more clearly illustrates that rIL-36Ra administration led to a reduction in A. muciniphila outgrowth in il36rn-/-mice faeces.
2. Fig. 6 co-housing result provided a strong positive association of A. muciniphila level and the metabolically protective phenotype. To prove a causal effect of A. muciniphila, the authors could gavage mice with A. muciniphila and examine whether this approach can produce protective phenotype. This would be the most convincing data for their claim.