DR3 stimulation of adipose resident ILC2s ameliorates type 2 diabetes mellitus

Disturbances in glucose homeostasis and low-grade chronic inflammation culminate into metabolic syndrome that increase the risk for the development of type 2 diabetes mellitus (T2DM). The recently discovered group 2 innate lymphoid cells (ILC2s) are capable of secreting copious amounts of type 2 cytokines to modulate metabolic homeostasis in adipose tissue. In this study, we have established that expression of Death Receptor 3 (DR3), a member of the TNF superfamily, on visceral adipose tissue (VAT)-derived murine and peripheral blood human ILC2s is inducible by IL-33. We demonstrate that DR3 engages the canonical and/or non-canonical NF-κB pathways, and thus stimulates naïve and co-stimulates IL-33-activated ILC2s. Importantly, DR3 engagement on ILC2s significantly ameliorates glucose tolerance, protects against insulin-resistance onset and remarkably reverses already established insulin-resistance. Taken together, these results convey the potent role of DR3 as an ILC2 regulator and introduce DR3 agonistic treatment as a novel therapeutic avenue for treating T2DM.

1) The authors used C57BL/6J mouse as a wild type control mice. However, Rag2 knockout (KO) mice, Rag2Il2rg double KO mice and Il13 KO mice were on a BALB/c background. Fasting glucose level of HFD-fed DR3 agonist-treated mice was lower than HFD-fed PBS-treated mice in ITT shown in Fig. 2e but such difference was not observed in Rag2 KO or Rag2Il2rg double KO mice in Figs. 3f and 4f. Such difference could be due to the difference in genetic backgrounds. The authors should use BALB/c mouse as a wild type control mouse to show data currently shown in Fig. 2.
2) In addition, it is unclear why the authors used C57BL/6J mice instead of C57BL/6N mice for both in vivo and in vitro experiments. It has been widely known that C57BL/6J mice possess an nicotinamide nucleotide transhydrogenase (Nnt) gene mutation and are known to develop obesity and metabolic diseases more readily than C57BL/6N mice can on HFD feeding. Therefore, people usually use C57BL/6N mice for metabolism study.
3) According to the Method section, the authors obtained only C57BL/6J mice as wild type mice. Did the authors transfer ILC2 from C57BL/6J wild type mice into Rag2Il2rg double KO mice on a BALB/c background in Fig. 5? 4) Because activation of ILC2 to induce IL-5 and IL-13 contributes to the pathophysiology of allergic inflammation such as asthma, DR3 engagement leading to the secretion of IL-5 and IL-13 could deteriorate certain diseases. The authors should discuss such potential drawback of treatment of T2DM by DR3 engagement.
Following are other points. 1) In Fig. 2f, the authors showed that the mRNA levels were increased upon HFD feeding. Because IL-33 secretion depends on the cellular damages, amounts mRNA do not necessarily correlate with those of secreted proteins. Is there any evidence that secreted IL-33 was also increased by HFD feeding?
2) The headline on the line 154 of page 7 should be modified because the experimental results in this section showed that the effects of DR3 engagement is independent of T/B cells but did not show the dependency on ILC2.
3) In Fig. 5c, was the difference between wild type (red squares) and Il5 KO (purple diamonds) significant? 6) Stock numbers should be provided for mice obtained from the Jackson Laboratory. 6) In the discussion section on page 14, the authors implied that new flux of naïve ILC2s from the bone marrow. However, parabiosis analyses had shown that new flux contributes little to the number of ILC2 in the adipose tissues and even in the lung after IL-33 administration. It is possible that number of tissue-resident ILC2 in adipose tissues simply decreases under the type 1 environment rather than impaired survival of naïve ILC2 from the bone marrow.
Reviewer #2: Remarks to the Author: Activation of Type 2 innate lymphoid cells (ILC2s) has been reported to improve obesity and glucose intolerance, but the mechanisms still remain unclear. The authors found that Death Receptor 3 (DR3) was expressed in ILC2s and induced by IL-33 treatment. DR3 agonist improved insulin resistance and glucose intolerance in high fat diet-fed mice. When the ILC2s of wild-type mice were transferred into the Rag / IL2rg double-deficient mice, DR3 agonist improved insulin resistance and glucose intolerance, but not obesity, in the high-fat fed Rag / IL2rg double-deficient mice. When the ILC2s of IL-13 deficient mice were transferred into the Rag / IL2rg doubledeficient mice, these phenotypes such as insulin resistance and glucose intolerance were not improved by DR3 agonist. In human ILC2s, DR3 expression was increased by IL-33.
Major concerns 1. Were the expression of DR3 and its downstream signaling impaired in ILC2s of obesity? 2. As described in the discussion, the number of ILC2s has been reported to be decreased in adipose tissue of obesity. Did the number of ILC2s in adipose tissue change before and after administration of DR3 agonist?
3. Did ILC2-specific DR3 deficient mice exhibit glucose intolerance or insulin resistance? 4. Although DR3 agonist did not change the body weight under high fat-diet condition, did it affect the size and number of fat cells in adipose tissue? Were insulin resistance in liver and skeletal muscle improved by DR3 agonist? 5. ILC2s have been reported to promote beiging of white adipose tissue (Nature 519 :242-6, 2015). Please examine the oxygen consumption and the expression of tyrosine hydroxylase (TH) in adipose tissue when DR3 agonist is injected into the high fat-diet fed mice. Minor concern 1. In page 12 lines 278-279, does "canonical and canonical NF-kB" mean "canonical and uncanonical NF-kB"? 2. In page 10 line 227, "Lastly" is redundant.
Reviewer #3: Remarks to the Author: Shafiei-Jahani and colleagues describe a role for DR3 in a model of type 2 diabetes mellitus (T2DM), proposing that stimulation of ILC2s via agonistic anti-DR3 Abs impairs the formation of visceral adipose tissue, ameliorating glucose intolerance and protecting against insulin resistance in animals on a high fat diet and in an IL-13-dependent fashion. DR3 and its capacity to drive IL-5 and IL-13 production from ILC2s is not novel as it has been shown by others in models of other inflammatory diseases, but this is the first demonstration in T2DM. The data shows a very clear phenotype that is potentially very exciting, but there are a number of points, addressing of which would give a better context to their data.
General Points (i) The authors need to be a bit more careful with the description of their data. The published functions and expression patterns for DR3 are more extensive than described by the authors in their Introduction (this requires additional references) and this needs coverage and consideration when the authors discuss their data and its potential as a therapeutic avenue. This is not an exhaustive list, but a role for DR3 on stroma in general but also including fibroblasts and epithelial cells specifically, and also myeloid lineage cells (macrophages, neutrophils) and neurones all have been described, so a phenotype following the application of a DR3 agonist to RAG2-/-or RAG2-/-IL2rg-/-mice may not just be due to its impact on ILC2s as described by the authors on lines 179-184 (although the later adoptive transfer expts in Fig 5 are consistent with that effect). Stand alone, there is no ILC2 specific data for the expts described in Figures 2-4. Did the authors look at VAT ILC2s numbers in their DR3 agonist expts in BL/6, RAG2-/-and RAG2-/-IL2rg-/-mice to show accumulation in the VAT of endogenous ILC2s? The ideal expt would be one on ILC2 deficient mice (Rorasg/floxIl7rCre/+). Have the authors also considered performing high fat diet expts on DR3-/mice (which should then have impaired control of glucose)? (ii) Relating to the above but perhaps tangential to the main conclusions of the manuscript, there are also aspects of their phenotype which could be caused by the interplay between the cytokine production induced by DR3 activation on ILC2s and non-lymphoid tissue, particularly the reduced numbers of VAT ILC2s when adding IL-5-/-or IL-13-/-ILC2s. The authors comment that this is intriguing, but do not seem to discuss it. Did the authors consider adoptive transfer of DR3-/-ILC2 to give an indication of the underlying mechanisms behind this (ie. to what extent does DR3 signalling in the ILC2s and interactions elsewhere contribute to their accumulation in VAT?). (iii) Statistically, t-Tests for individual time points are incorrect if the samples come longitudinally from the same animal as the data suggests. This should be corrected where required.
Specific minor points (iv) Line 77 -TL1A is not the only published ligand for DR3 -it is the only known ligand that is a member of the TNF superfamily. Progranulin/attstrin has also been reported to bind DR3 and this requires some coverage/discussion. We thank the reviewers for their valuable time and thoughtful feedbacks. We appreciate all of the comments that have helped us enhance the context of the results and strengthen the conclusions of this study. Please see our point-by-point response bellow.

Reviewers' comments:
Reviewer #1 (Remarks to the Author): 1) The authors used C57BL/6J mouse as a wild type control mice. However, Rag2 knockout (KO) mice, Rag2Il2rg double KO mice and Il13 KO mice were on a BALB/c background. Fasting glucose level of mice was lower than HFD-fed PBS-treated mice in ITT shown in Fig. 2e but such difference was not observed in Rag2 KO or Rag2Il2rg double KO mice in Figs. 3f and 4f. Such difference could be due to the difference in genetic backgrounds. The authors should use BALB/c mouse as a wild type control mouse to show data currently shown in Fig. 2.
We agree genetic background in pre-clinical models may influence the severity of dietinduced obesity 1 . However, we utilized an established protocol as described by our group and others before 2-6 . In Fig. 2 we used C57BL/6J males, which are known to be susceptible to induction of HFD-induced T2DM. For studies related to Rag2 -/and Rag2 -/γc -/mice, every strain used is from the BALB/c background. We have now clarified the genetic backgrounds of all mice in the methods section.
We agree with the reviewer that the difference in the fasting glucose levels (between the HFD-Fed Isotype-treated cohort and others) could be due to the genetic profiles of the different backgrounds and experimental variability. It is important to note that although the basal levels may vary, the slopes (response), trends, and overall phenotypes associated with amelioration of T2DM via DR3-dependent stimulation of ILC2s in different cohorts are significant and consistent. To address this concern, we have performed a novel set of experiments using BALB/c mice as suggested and compared the results to our observations presented in Fig. 2. The new data is now incorporated to the manuscript as Supplementary Figure 1. Briefly, in these experiments a cohort of BALB/c mice were fed either a normal chow diet (NCD) or an HFD for 14 weeks. During this period, the mice were intraperitoneally treated with either DR3 agonist (1mg/mouse) or the isotype control, according to the protocol presented as Supplementary Figure  1a. These new results importantly confirm our observations in C57BL/6J as the DR3 agonist-treated cohort showed significant reduction in fasting blood glucose levels when compared to isotype control (Supplementary Figure 1c). Furthermore, the glucose tolerance (Supplementary Figure 1d) and insulin sensitivity (Supplementary Figure  1e) of the DR3 agonist treated BALB/c cohort significantly improved when compared to the isotype control. Altogether, these results suggest that DR3-dependent stimulation of ILC2s significantly improved glucose homeostasis and insulin resistance regardless of the genetic background.
2) In addition, it is unclear why the authors used C57BL/6J mice instead of C57BL/6N mice for both in vivo and in vitro experiments. It has been widely known that C57BL/6J mice possess an nicotinamide nucleotide transhydrogenase (Nnt) gene mutation and are known to develop obesity and metabolic diseases more readily than C57BL/6N mice can on HFD feeding. Therefore, people usually use C57BL/6N mice for metabolism study.
We appreciate the reviewer's comment regarding the differences between the substrains of C57BL/6. In this study, we followed an established model of C57BL/6J that has been previously published by our group and several others 2-6 . We agree the M35T missense mutation alters RNA splicing of Nnt gene and allows for a more rapid development of metabolic distress 7,8 . Therefore, we believe C57BL/6J and C57BL/6N should not be used interchangeably within the same study. We have now clarified the genetic backgrounds of all mice in the manuscript and included the therapeutic efficacy of DR3 engagement in two independent genetic backgrounds (BALB/c and C57BL/6J).
3) According to the Method section, the authors obtained only C57BL/6J mice as wild type mice. Did the authors transfer ILC2 from C57BL/6J wild type mice into Rag2Il2rg double KO mice on a BALB/c background in Fig. 5?
We apologize for not being clear and have now clarified the genetic background of donors and recipients in the methods section. WT, Il5 -/-, Il13 -/donors and Rag2 -/γc -/recipients were all on BALB/c background as described in the methods. 4) Because activation of ILC2 to induce IL-5 and IL-13 contributes to the pathophysiology of allergic inflammation such as asthma, DR3 engagement leading to the secretion of IL-5 and IL-13 could deteriorate certain diseases. The authors should discuss such potential drawback of treatment of T2DM by DR3 engagement. This is a great point raised by the reviewer and we have now discussed this potential adverse effect in the manuscript. We believe local administration and hydrophobic modifications need to be developed and explored to minimize any adverse effects associated with DR3 treatment for T2DM. 1) In Fig. 2f, the authors showed that the mRNA levels were increased upon HFD feeding. Because IL-33 secretion depends on the cellular damages, amounts mRNA do not necessarily correlate with those of secreted proteins. Is there any evidence that secreted IL-33 was also increased by HFD feeding?
Unfortunately, we made numerous attempts to detect local IL-33 at protein level but it appears that the levels are below the threshold limits of ELISA and Luminex. Therefore, we designed RT-PCR methods and observed a significant increase of IL-33 at transcriptome level (Fig. 2f).
2) The headline on the line 154 of page 7 should be modified because the experimental results in this section showed that the effects of DR3 engagement is independent of T/B cells but did not show the dependency on ILC2.
We have now modified and improved the headline.
3) In Fig. 5c, was the difference between wild type (red squares) and Il5 KO (purple diamonds) significant?
The trend observed between wild type ILC2 and Il5 -/-ILC2 receipt cohorts did not reach statistical significance. We now added ns (not significant) to the Figure for  6) Stock numbers should be provided for mice obtained from the Jackson Laboratory.
We have included the stock numbers in the methods.

7)
In the discussion section on page 14, the authors implied that new flux of naïve ILC2s from the bone marrow. However, parabiosis analyses had shown that new flux contributes little to the number of ILC2 in the adipose tissues and even in the lung after IL-33 administration. It is possible that number of tissue-resident ILC2 in adipose tissues simply decreases under the type 1 environment rather than impaired survival of naïve ILC2 from the bone marrow. This is an interesting topic of discussion. Trafficking of ILC2s remains under investigation and is an area of active research. We are aware of parabiosis experiments demonstrating that tissue resident ILC2s are programmed and destined for specific tissues. However, the results of those experiments clearly suggest that bone marrow supply is necessary for homeostatic maintenance of tissue resident ILC2s. Please note that although tissue-residents ILC2s do not traffic to other tissues, there has been several lines of work that show BM derived ILC2s have the ability to traffic and replenish ILC2s in various tissues [15][16][17] . We now discuss this point in the manuscript.
We agree with the reviewer that type 1 environment is detrimental to survival of ILC2s and contributes to the low number and insufficient activation of ILC2s in VAT. However, we also believe that BM ILC2 flux is needed to adequately sustain homeostasis of tissue resident ILC2s. We have now clarified this point in the manuscript.

Reviewer #2 (Remarks to the Author):
Major concerns 1. Were the expression of DR3 and its downstream signaling impaired in ILC2s of obesity?
This is a great question. We have assessed and compared DR3 expression in naïve and activated ILC2s of mice (HFD and NCD) and incorporated the results as Figures 1a-c. Our data suggest that DR3 expression is comparable on VAT ILC2s. Moreover, we also assessed the expression levels of downstream transcription factors p65 and p52 NF-κB pathways and compared the levels in ILC2s derived from adipose tissue of HFD and NCD groups (Supplementary Figure 4). Our data suggest that DR3 dependent NF-κB signaling pathways are intact and also comparable in both groups.
2. As described in the discussion, the number of ILC2s has been reported to be decreased in adipose tissue of obesity. Did the number of ILC2s in adipose tissue change before and after administration of DR3 agonist?
We quantified the number of ILC2s in the WT and Rag2 -/mice treated with DR3 agonist or isotype control. We added the quantification results for WT mice as 3. Did ILC2-specific DR3 deficient mice exhibit glucose intolerance or insulin resistance?
We considered to explore DR3 -/-ILC2s at the beginning of this project. However, it became apparent that DR3 -/mice have a very low number of ILC2 as DR3 plays an important role in ILC2 homeostasis and cytokine production 18,19 . Therefore, isolation of adequate number of VAT ILC2s from DR3 -/mice was not technically feasible for performing adoptive transfer experiments. We need to point out that our results highlight the benefit and phenotype of DR3 engagement and signaling on ILC2s. The results of experiments with DR3 deficient mice would only provide information regarding the specificity of murine DR3 agonist, which was addressed by other groups before 20-23 . 4. Although DR3 agonist did not change the body weight under high fat-diet condition, did it affect the size and number of fat cells in adipose tissue? Were insulin resistance in liver and skeletal muscle improved by DR3 agonist?
We thank the reviewer for their comment. We performed additional histological analysis of VAT in both preventive (Fig. 3j-l) and therapeutic models (Fig4. g-i). Consistent with the improved insulin resistance and glucose tolerance previously shown in our study, the DR3-treated cohort in both models exhibited reduced hypertrophy of adipocytes and abrogated adipose tissue hyperplasia. Since hypertrophy and hyperplasia of adipocytes have been implicated in obese patients [24][25][26] , these new results add yet another supporting evidence for the efficacy and exciting therapeutic potential of DR3 agonistic treatment.
Regarding the diversity of mechanisms by which activation of ILC2s ameliorates phenotypes associated with T2DM, several groups have already demonstrated that activation of ILC2s mainly ameliorates T2DM by promoting beiging of white adipose tissue through production of Th2 associated cytokines 5,6,27 . We closely checked the liver mass in our experimental Rag2 -/groups and did not detect a significant difference in hepatic mass. This result is now added as Supplementary Figure 2d. Unfortunately, we do not have the data or samples available to assess the skeletal muscle and lean mass. However, we recently showed that activation of ILC2s via GITR, another member of the TNF superfamily increased the lean mass of the mice and ameliorated T2DM 6 . We now discuss this possibility and cite the relevant literature in the manuscript. Additionally, we have added new data to further examine the mechanisms of beiging and increased metabolic rate in response to the reviewer's comments bellow. 5. ILC2s have been reported to promote beiging of white adipose tissue (Nature 519 :242-6, 2015). Please examine the oxygen consumption and the expression of tyrosine hydroxylase (TH) in adipose tissue when DR3 agonist is injected into the high fat-diet fed mice.
We believe our results are in agreement with Brestoff's report that ILC2 activation can induce beiging of the adipose tissue and in turn increase thermogenesis and caloric expenditure. In this study, we previously showed the expression of uncoupling protein 1 (a characteristic gene of brown adipose tissue) and presence of VAT-associated alternatively activated macrophages (AAMs) are augmented, suggesting that DR3 treatment increases beiging of white adipose tissue (Fig. 3g-h). To further strengthen our observations, we performed additional RT-qPCR experiments and have now quantified five additional thermogenic genes in the VAT. These five genes (Cidea, Prdm16, Pgc1a, Cox7a, Dio2) have been shown to be molecular markers of brown adipose tissue. Moreover, Dio2, Prdm16 and Pgc1a expressions are also positively associated with a higher metabolic rate [28][29][30] . Our results suggest that the expression of aforementione alternatively activated macrophages d genes in VAT lysate was significantly increased after DR3 treatment. These results are now added as Fig. 3i to the manuscript.
Next, we assessed the expression of tyrosine hydroxylase (TH) and 5 well-known genes encoding respiratory chain complexes (I, III, IV, and V) in the adipose tissue of the mice treated with DR3 agonist or isotype. We chose this approach due to logistical constraints. These new RT-qPCR results demonstrate that DR3 agonist treatment significantly increased tyrosine hydroxylase (TH); as well as complex I (ND1 and ND5), complex III (mitochondria-encoded NADH dehydrogenase I (Cytb)), complex IV (mitochondria-encoded cytochrome c oxidase I (Cox1)), and complex V (mitochondriaencoded ATP synthase 6 (Atp6)) in the adipose tissue. These results are now added to the manuscript as Supplementary Figure 2e-j. Taken together, these results are in agreement with the findings by Brestoff et al. and support the notion that activation of ILC2s induces beiging of the adipose tissue and increase the metabolic rate of the VAT. 6. By what mechanism does IL-33 increase the DR3 expression?
IL-33 is a pluripotent cytokine that is capable of activating multiple cellular signaling networks 31 . Our results suggest that DR3 engagement can act on ILC2s independent of IL-33 signaling, as naive ILC2s in both mice and humans showed higher cytokine production after treatment with DR3 agonist or TL1A-L (Fig 1d. and Fig 7d.). However, IL-33 upregulates a variety of pathways including NF-κB signaling via MyD88/IRAK/TRAF6, ERK1/2, JNK, p38 and PI3K/AKT, and all these pathways have been previously shown to upregulate costimulatory molecules including (but not limited to) CD40, CD80, CD86, OX40, ICAM-1, PD-1, TNFR2 and now DR3 [32][33][34][35][36][37][38][39][40][41][42][43][44][45] . We now discuss the effect of IL-33 on enhancement of costimulatory molecules in the discussion and cite relevant articles. The expression of DR3 on intestinal resident ILCs was previously reported and data suggest that DR3 expression is comparable to the VAT resident ILC2s. We believe tissue-specific programing and DR3-dependent stimulation need to be investigated in another study. However, we agree with the reviewer and we now discuss tissue-specific programing of ILC2s, and the studies conducted by Sasaki et al. and Li et al. in the manuscript.
We have revised the manuscript.
We have adapted the manuscript accordingly.

Reviewer #3 (Remarks to the Author):
General Points (i) The authors need to be a bit more careful with the description of their data. The published functions and expression patterns for DR3 are more extensive than described by the authors in their Introduction (this requires additional references) and this needs coverage and consideration when the authors discuss their data and its potential as a therapeutic avenue. This is not an exhaustive list, but a role for DR3 on stroma in general but also including fibroblasts and epithelial cells specifically, and also myeloid lineage cells (macrophages, neutrophils) and neurones all have been described, so a phenotype following the application of a DR3 agonist to RAG2-/-or RAG2-/-IL2rg-/mice may not just be due to its impact on ILC2s as described by the authors on lines 179-184 (although the later adoptive transfer expts in Fig 5 are consistent with that effect). Stand alone, there is no ILC2 specific data for the expts described in Figures 2-4. Did the authors look at VAT ILC2s numbers in their DR3 agonist expts in BL/6, RAG2-/-and RAG2-/-IL2rg-/-mice to show accumulation in the VAT of endogenous ILC2s? The ideal expt would be one on ILC2 deficient mice (Rorasg/floxIl7rCre/+). Have the authors also considered performing high fat diet expts on DR3-/-mice (which should then have impaired control of glucose)?
The reviewer 3's inquiry regarding the number of ILC2s was addressed in response to reviewer 2's second comment above. Briefly, we have added new data that suggest DR3 stimulation of new flux of naïve ILC2s and co-stimulation of endothelial-derived IL-33-activated VAT ILC2s (through the canonical and/or non-canonical NF-κB pathways) efficiently combats the decline in number of VAT ILC2s reported in animal models and obese patients 5,63 . As additionally mentioned above, we now added new data that demonstrate DR3 expression and signaling is not impaired in the HFD model, further underscoring the therapeutic viability of targeting this pathway in obese patients.
Unfortunately, we do not have access to Rorasg/floxIl7rCre/+, which would have been an excellent tool for confirming previous studies that have showed importance of ILC2s in T2DM 5,27,64 . The reviewer 3's inquiry regarding DR3 -/mice was addressed in our response to reviewer 2's third comment above. As discussed above DR3 -/mice have extremely low numbers of ILC2s as reported by several groups 18,19 . Therefore, it is not technically feasible to isolate sufficient numbers of VAT DR3 -/-ILC2s to perform experiments. Nevertheless, the results of adoptive transfer with DR3 -/-ILC2s treated with DR3 agonist or isotype control will only provide information regarding the specificity of the murine DR3 agonist, which was described by several other groups before [20][21][22][23] . We need to point out that our results highlight the benefit and phenotype of DR3 engagement on ILC2s and provide a potentially therapeutic approach for the patients with T2DM.
(ii) Relating to the above but perhaps tangential to the main conclusions of the manuscript, there are also aspects of their phenotype which could be caused by the interplay between the cytokine production induced by DR3 activation on ILC2s and non-lymphoid tissue, particularly the reduced numbers of VAT ILC2s when adding IL-5-/-or IL-13-/-ILC2s. The authors comment that this is intriguing, but do not seem to discuss it. Did the authors consider adoptive transfer of DR3-/-ILC2 to give an indication of the underlying mechanisms behind this (ie. to what extent does DR3 signalling in the ILC2s and interactions elsewhere contribute to their accumulation in VAT?).
We agree with the reviewer that the interplay between ILC2s and non-lymphoid tissue is important and needs to be discussed. Recent studies have highlighted the critical interplays between ILC2s and their respective physical tissue niches during homeostasis and inflammation 49,[65][66][67][68] . We agree with the reviewer that cytokines produced by ILC2s (particularly, IL-5 and IL-13) will impact somatic cells such as adipocytes. A study by Dahlgren et al. demonstrated adventitial stromal cells (ASCs) derived IL-33 and TSLP regulate ILC2 effector function and expansion 65 . Additionally, the interactions between stroma and adipose-resident leukocytes may be mediated by adhesion molecules. Recently, Rana et al. showed that white adipose tissue-resident multipotent stromal cells (WAT-MSCs) supports ICAM-1-mediated proliferation and activation of ILC2s 49 . We now discuss the interplay between ILC2s and the local stromal cell niche in the manuscript and cite the relevant studies.
As discussed above, one of the resulting consequences of activating ILC2 effector function is the conversion of white to beige adipose tissue. As previously described, beiging of the adipose tissue increases thermogenesis, and subsequently increases the caloric expenditure 5 . As mentioned in response to reviewer 2, we have added the data for five additional thermogenic markers (Cidea, Prdm16, Pgc1a, Cox7a, Dio2) and 5 known genes encoding respiratory chain complexes (I, III, IV, and V) in VAT. These data are now added as Fig. 3i and Supplementary Figure 2e-j. Since the adoptive transfer experiments in Fig. 5 indicated that DR3 based amelioration of T2DM is IL-13 + ILC2s dependent, our results collectively indicate DR3 agonistic treatment improves insulin sensitivity and glucose tolerance by promoting beiging of the adipose tissue and increasing the metabolic rate of the VAT. Moreover, we have now added new data showing reduced adipose hypertrophy and hyperplasia, further indicating DR3 treatment is effective in both preventative and therapeutic Rag2 -/models ( Fig. 3j-l and Fig4. g-i). Taken all together, we believe DR3 agonistic treatment results in a potent and efficient therapeutic strategy to prevent and reverse the metabolic syndromes associated with T2DM.
(iii) Statistically, t-Tests for individual time points are incorrect if the samples come longitudinally from the same animal as the data suggests. This should be corrected where required.
We thank the reviewer for the thoughtful comment. After discussion with biostatisticians at USC, we have now implemented t-Tests, one-way and two-way ANOVA tests as appropriate for various results based on the nature of experiments. We have updated the manuscript by adding the new statistical results.
Specific minor points (iv) Line 77 -TL1A is not the only published ligand for DR3 -it is the only known ligand that is a member of the TNF superfamily. Progranulin/attstrin has also been reported to bind DR3 and this requires some coverage/discussion. We thank the reviewer for raising this point, and we have now updated the manuscript to address the studies related to DR3 and progranulin-derived Atsttrin.
(v) Fig 2F -it would be nice to have protein data for IL-33 as well if possible (rather than just mRNA).
As mentioned in response to reviewer 1's first minor point above, we made several attempts to detect IL-33 from adipocyte lysate but unfortunately it has become apparent that the tissue levels of cytokines such as IL-33 are very low and below the threshold limits of detection by ELISA or Luminex.
(vi) Line 227 -one of the 'lastly's should be deleted.
We edited line 227.
(vii) Is there any particular reason why the glucose levels in the RAG2-/-mice in Figures  3 and 4 don't seem to match? In the therapeutic expt, they start at twice the level as the initial DR3 engagement expt.
As mentioned in the manuscript, each experiment is a representative of 3 independent studies that were carried out at different times. Although the slope, response and major phenotypic response to treatment has been the same in each of the replicate studies, the glucose levels appear to be variable among the replicates. We encountered similar basal variations in our previous studies 6 .
Although not yet comprehensibly understood, the glucose levels appear mouse batch dependent. To the best of our knowledge, it remains to be unraveled whether this variation is due to background genetic drift or environmental variables. We agree this could lead to interesting future studies, but we believe exploration of the diverse set of factors that contribute to base line glucose variation is out of the scope of the current study. Since DR3 treatment effectively improved the overall metabolic phenotype in all of the different experimental setups, any factors contributing to the mentioned variation are likely independent of DR3 signaling on ILC2s.