Targeting IRE1α improves insulin sensitivity and thermogenesis and suppresses metabolically active adipose tissue macrophages in obesity

Reviewer #1 (Public review):

First, the authors confirm the up-regulation of the main genes involved in the three branches of the Unfolded Protein Response (UPR) system in diet-induced obese mice in AT, observations that have been extensively reported before. Not surprisingly, IRE1a inhibition with STF led to an amelioration of the obesity and insulin resistance of the animals. Moreover, non-alcoholic fatty liver disease was also improved by the treatment. More novel are their results in terms of thermogenesis and energy expenditure, where IRE1a seems to act via activation of brown AT. Finally, mice treated with STF exhibited significantly fewer metabolically active and M1-like macrophages in the AT compared to those under vehicle conditions. Overall, the authors conclude that targeting IRE1a has therapeutical potential for treating obesity and insulin resistance.

The study has some strengths, such as the detailed characterization of the effect of STF in different fat depots and a thorough analysis of macrophage populations. However, the lack of novelty in the findings somewhat limits the study´s impact on the field.

Reviewer #3 (Public review):

Summary:

The manuscript by Wu D. et al. explores an innovative approach in immunometabolism and obesity by investigating the potential of targeting macrophage Inositol-requiring enzyme 1α (IRE1α) in cases of overnutrition. Their findings suggest that pharmacological inhibition of IRE1α could influence key aspects such as adipose tissue inflammation, insulin resistance, and thermogenesis. Notable discoveries include the identification of High-Fat Diet (HFD)-induced CD9+ Trem2+ macrophages and the reversal of metabolically active macrophages' activity with IRE1α inhibition using STF. These insights could significantly impact future obesity treatments.

Strengths:

The study's key strengths lie in its identification of specific macrophage subsets and the demonstration that inhibiting IRE1α can reverse the activity of these macrophages. This provides a potential new avenue for developing obesity treatments and contributes valuable knowledge to the field.

Weaknesses:

The research lacks an in-depth exploration of the broader metabolic mechanisms involved in controlling diet-induced obesity (DIO). Addressing this gap would strengthen the understanding of how targeting IRE1α might fit into the larger metabolic landscape.

Impact and Utility:

The findings have the potential to advance the field of obesity treatment by offering a novel target for intervention. However, further research is needed to fully elucidate the metabolic pathways involved and to confirm the long-term efficacy and safety of this approach. The methods and data presented are useful, but additional context and exploration are required for broader application and understanding.

Comments on revisions:

The author has revised the manuscript and addressed the most relevant comments raised by the reviewers. The paper is now significantly improved, though two minor issues remain.

(1) Studies were limited to male mice; this should be mentioned in the paper's Title.
(2) Please include the sample size (n=) in all provided tables in the main manuscript and supplementary tables.

Author response:

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

eLife Assessment

The study presents important findings on inositol-requiring enzyme (IRE1α) inhibition on diet-induced obesity (overnutrition) and insulin resistance where IRE1α inhibition enhances thermogenesis and reduces the metabolically active and M1-like macrophages in adipose tissue. The evidence supporting the conclusions is convincing but can be enhanced with information/data on the validity, specificity, selectivity, and toxicity of the IRE1α inhibitor and supported with more detail on the mechanisms by which adipose tissue macrophages influence adipocyte metabolism. The work will be of interest to cell biologists and biochemists working in metabolism, insulin resistance, and inflammation.

We thank the editors for the assessment and appreciation of our findings in this study. In the revision, we have added the information on the validity, selectivity and toxicity of IRE1α inhibitor. In addition, we also discussed the likelihood that suppression of metabolically activated proinflammatory macrophage population in adipose tissue on the reversal of adipose remodeling and thermogenesis. In the revision, we have improved the manuscript significantly throughout the text and figures following the recommends by the reviewers.

Public Reviews:

Reviewer #1 (Public review):

First, the authors confirm the up-regulation of the main genes involved in the three branches of the Unfolded Protein Response (UPR) system in diet-induced obese mice in AT, observations that have been extensively reported before. Not surprisingly, IRE1a inhibition with STF led to an amelioration of the obesity and insulin resistance of the animals. Moreover, non-alcoholic fatty liver disease was also improved by the treatment. More novel are their results in terms of thermogenesis and energy expenditure, where IRE1a seems to act via activation of brown AT. Finally, mice treated with STF exhibited significantly fewer metabolically active and M1-like macrophages in the AT compared to those under vehicle conditions. Overall, the authors conclude that targeting IRE1a has therapeutical potential for treating obesity and insulin resistance.

The study has some strengths, such as the detailed characterization of the effect of STF in different fat depots and a thorough analysis of macrophage populations. However, the lack of novelty in the findings somewhat limits the study´s impact on the field.

We thank the reviewer for the appreciation of our findings and the comments about the novelty. Regarding the novelty, we would emphasize several novelties presented in this manuscript. First, as the reviewer correctly pointed out, we discovered that IRE1 inhibition by STF activates brown AT and promotes thermogenesis and that IRE1 inhibition not only significantly attenuated the newly discovered CD9+ ATMs and the “M1-like” CD11c+ ATMs but also diminished the M2 ATMs for the first time. These discoveries are very important and novel. In obesity, it was originally proposed that ATM undergoes M1/M2 polarization from an anti-inflammatory M2 to a classical pro-inflammatory M1 state. It was further reported that IRE1 deletion improves thermogenesis by boosting M2 population which then synthesize and secrete catecholamines to promote thermogenesis. It is now known that M2 macrophages do not synthesize catecholamines or promote thermogenesis. In this study, we discovered that IRE1 inhibition doesn’t increase (but instead decrease) the M2 population and that IRE1 inhibition promotes thermogenesis likely by suppressing pro-inflammatory macrophage populations including the M1-like ATMs and most importantly the newly identified metabolically active macrophages, given that ATM inflammation has been reported to suppress thermogenesis. Second, this study presented the first characterization of relationship between the more classical M1-like ATMs and the newly discovered metabolically active ATMs, showing that the CD11c+ M1-like ATMs are largely overlapping with but yet non-identical to CD9+ ATMs in the eWAT under HFD. Third, although upregulation of ER stress response genes in the adipose tissues of diet-induced obese mice have been extensively reported, it doesn’t necessarily mean that targeting IRE1a or ER stress can reverse existing insulin resistance and obesity. It is not uncommon that a therapy doesn’t yield the desired effect as expected. For instance, amyloid plaques are a hallmark of Alzheimer's disease (AD), interventions that prevent or reverse beta amyloid deposition have been expected to prevent progression or even reverse cognitive impairment in AD patients. However, clinical trials on such therapies have been disappointing. In essence, experimental demonstration of effectiveness or feasibility for any potential therapeutic targets is a first step for any future clinical implementation.

Reviewer #2 (Public review):

The manuscript by Wu et al demonstrated that IRE1a inhibition mitigated insulin resistance and other comorbidities through increased energy expenditure in DIO mice. In this reviewer's opinion, this timely study has high significance in the field of metabolism research for the following reasons.

(1) The authors' findings are significant and may offer a new therapeutic target to treat metabolic diseases, including diabetes, obesity, NAFLD, etc.

(2) The authors carefully profiled the ATMs and examined the changes in gene expression after STF treatment.

(3) The authors presented evidence collected from both systemic indirect calorimetry and individual tissue gene expression to support the notion of increased energy expenditure.

Overall, the authors have presented sufficient background in a clear and logically organized structure, clearly stated the key question to be addressed, used the appropriate methodology, produced significant and innovative main findings, and made a justified conclusion.

We thank the reviewer for the appreciation of our work.

Reviewer #3 (Public review):

Summary:

The manuscript by Wu D. et al. explores an innovative approach to immunometabolism and obesity by investigating the potential of targeting macrophage Inositol-requiring enzyme 1α (IRE1α) in cases of overnutrition. Their findings suggest that pharmacological inhibition of IRE1α could influence key aspects such as adipose tissue inflammation, insulin resistance, and thermogenesis. Notable discoveries include the identification of High-Fat Diet (HFD)-induced CD9+ Trem2+ macrophages and the reversal of metabolically active macrophages' activity with IRE1α inhibition using STF. These insights could significantly impact future obesity treatments.

Strengths:

The study's key strengths lie in its identification of specific macrophage subsets and the demonstration that inhibiting IRE1α can reverse the activity of these macrophages. This provides a potential new avenue for developing obesity treatments and contributes valuable knowledge to the field.

Weaknesses:

The research lacks an in-depth exploration of the broader metabolic mechanisms involved in controlling diet-induced obesity (DIO). Addressing this gap would strengthen the understanding of how targeting IRE1α might fit into the larger metabolic landscape.

Impact and Utility:

The findings have the potential to advance the field of obesity treatment by offering a novel target for intervention. However, further research is needed to fully elucidate the metabolic pathways involved and to confirm the long-term efficacy and safety of this approach. The methods and data presented are useful, but additional context and exploration are required for broader application and understanding.

We thank the reviewer for the appreciation of strengths in our manuscript. In particular, we appreciate the reviewer’s recommendation on the exploration of broader metabolic landscape, such as the effect of IRE1 inhibition on non-adipose tissue macrophages and metabolism. We agree that achieving these will certainly broaden the therapeutic potential of IRE1 inhibition to larger metabolic disorders and we will pursue these explorations in future studies.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

A list of recommendations for the authors is presented below:

(1) Please, update the literature review to include more recent studies relevant to the topic.

We thank the reviewer’s suggestions. We have added more references from recent studies.

(2) Please, provide a detailed explanation of how STF functions, including potential off-target effects or issues related to specificity.

We thank the reviewer’s suggestions. STF is a small-molecule inhibitor designed to selectively inhibit the RNase activity of IRE1a. Once IRE1a is activated (e.g., in obesity), its RNase domain initiates the unconventional splicing of the transcription factor X-box binding protein 1 (XBP1) mRNA and the Regulated IRE1-Dependent Decay (RIDD) of microRNAs, which is detrimental if prolonged. IRE1a RNase inhibitors including STF engage the RNase-active site of IRE1a with high affinity and specificity by exploiting a shallow complementary pocket through pi-stacking interactions with His910 and Phe889 and an essential Schiff base interaction between the aldehyde moiety of the inhibitor and the side chain amino group of Lys907 (Sanches et al., NComm 2014, PMID: 25164867). This specific and high affinity binding blocks the IRE1a RNase activity, preventing the splicing of XBP1 mRNA and RIDD. As IRE1a has been shown to be activated in multiple tissues under various pathological conditions and to be responsible for the progression of the pathological conditions, inhibition of IRE1a by pharmacological agents including STF has the great potential for the treatment of various pathological disorders. Several studies have reported that STF shows no overt toxicity when administered systemically (Madhavan, Aparajita, et al.2022, PMID 35105890; Herlea-Pana et al., 2021, PMID 34675883; Papandreou et al., 2011, PMID 21081713; Tufanli et al., 2017, PMID 28137856).

(3) Lines 263-266 require a reference.

We thank the reviewer’s suggestion. A reference has been added.

(4) Stromal vascular fraction (SVF) also contains a significant amount of preadipocytes and stem cells, not only macrophages, which might affect the conclusions reached by the authors.

We thank the reviewer’s comments. It is true that SVF consists of multiple cell types, including endothelial cells, macrophages, preadipocytes, and various stem cell populations. In HFD-induced obesity, adipose tissue undergoes significant remodeling, and the percentage of macrophages in the SVF of obese adipose tissue increases significantly relative to other cell types. In our studies, SVFs from adipose tissues of obese mice were isolated, cultured, and treated with STF for overnight. We observed that IRE1 RNase activity in SVFs was inhibited by STF treatment, and that ATM population and the expression of pro-inflammatory genes were downregulated by STF. Given the short-term treatment, the parsimonious interpretation of the data would be that STF directly acts on ATMs. However, we note that the possibility that the effect of STF on other cell types might influence the ATM and inflammatory gene expression can’t be totally ruled out. As such, we have modified our conclusion from “these results indicate that STF acts directly on ATMs to regulate inflammation” to “these results indicate that STF likely acts directly on ATMs to regulate inflammation”.

(5) Figures 1A and G: It is common practice to present the XBP1s/XBP1u ratio; consider using this standard measure.

We thank the reviewer’s comments. Regarding the XBP1 mRNA splicing, we see both ways of presentation in publications. There are quite a number of papers, for instance, PMID25018104, 2014, Cell; PMID23086298, 2012, NCB, that used the XBP1s/ (XBP1s+XBP1u) ratio. We preferred this way of presentation as it shows the ratio of spliced XBP1 (XBP1s) relative to the total XBP1 mRNA (XBP1s+XBP1u).

(6) Figure 1F: please indicate the type of AKT phosphorylation assessed.

We thank the reviewer’s comments. We have added Ser473 as the phosphorylation site at in both figure legend and figure.

(7) Figures 2E-H: please clearly indicate the specific fat depots analyzed in each figure.

We thank the reviewer’s comments. We have added the information in the figure legends and figures.

(8) Figures 1I and 3A, and Supplementary Figures 6D-E: please include a quantification analysis of the images presented.

We thank the reviewer’s suggestion. We have added the quantifications of the images.

(9) In Figure 3D the image corresponding to the merge for the STF condition is a duplication of the control, please correct this.

We thank the reviewer for pointing this out. We have replaced it with the correct image.

(10) Figures 4B-F: please provide individual data points in the graphs to show variability and sample distribution.

We thank the reviewer’s suggestion. We have re-plotted the graphs in Fig. 4B-F with the individual data points.

(11) Figure 4I: it is rather unusual to have such a strong signal of UCP1 in ND conditions, please explain.

We thank the reviewer for the comment. We wish to point out that the images were taken from BAT slides. UCP1 is expected to show strong staining in BAT under DN condition, which as expected is weakened under HFD condition. STF treatment was able to correct the HFD-induced weakening of UCP1 staining in BAT.

(12) Supplementary Figures 2C-D: please provide representative images for better clarity and interpretation.

We thank the reviewer for the comment. The representative images for Supplementary Figures 2C-D were actually shown in Figures 2C and F. Supplementary Figures 2C-D were the mere quantification for adipocyte areas for Figures 2C and F.

(13) Supplementary Table 3 is repeated, please remove.

We thank the reviewer for the comment. We have deleted this repetition.

Reviewer #2 (Recommendations for the authors):

The manuscript can be further strengthened with more clarification on the following points.

(1) The use of IRE1a pharmacological inhibitor STF-083010 (STF) needs to be validated. How was the dose determined? Were there any dose-dependent studies? Under the current dosing regimen, what are the specificity, selectivity, and toxicity of STF? Also, were the serine/threonine kinase and RNase activities measured in the adipocytes and ATMs of the animals dosed with the compound? What's the PK data?

We thank the reviewer for the comments. In the animal study, we used STF 10 mg/kg for intraperitoneal injection. This dose was adopted from several recent studies (Madhavan, Aparajita, et al.2022, PMID 35105890; Herlea-Pana et al., 2021, PMID 34675883; Papandreou et al., 2011, PMID 21081713; Tufanli et al., 2017, PMID 28137856), in which STF treatment showed beneficial effect in their respective disease models. STF didn’t compromise cell viability or induce any other toxicity at the dose or concentration used in these studies (Papandreou I, et al., 2011; Upton JP, et al., 2012; Lerner AG, et al., 2012; Kemp KL, et al., 2013; Cross BC, et al., 2012). In our study, we didn’t observe any apparent toxicity on mice at this dose. Importantly, we did observe that STF inhibited IRE1 RNase activity in adipose tissues (F1G, S1D) and ATMs (F6Q, S8C, G, I) of the animals at this dose. As the IRE1 inhibitors including STF has been extensively examined and shown to have no effect on the kinase function of IRE1 (Cross et al., 2012, PMID: 22315414; Tufanli et al., 2017, PMID 28137856), we didn’t perform the assay on Ire1 kinase activity. Additionally, as the chemical has been administered into several animal models, with significant beneficial effects, one would assume decent pharmacokinetic parameters being achieved with the current dose. It would be important and necessary to have systematic PK studies in the future if clinical trials are to be considered.

(2) The statistical method for individual panels in each figure needs to be specified.

We thank the reviewer for the suggestion. We have specified the statistical method in the figure legends.

(3) In Figure 1E, there's no difference in fasting insulin levels, though a difference was detected after the glucose load. This suggests an effect on insulin secretion but not insulin sensitivity.

We thank the reviewer for the comments. The insulin levels are still different between Veh and STF groups at fasting, just not reaching statistically significant. Under glucose stimulation, the insulin levels all showed the same trend, which is, the STF group is lower than the Veh group. Even if the fasting insulin levels showed no difference between the two groups regardless of glucose stimulation, the fact that the blood glucose levels at all the time points are lower in STF group than Veh group (Fig. 1C) indicates that insulin sensitivity is improved. In our study, the insulin levels were lower in STF group, but the blood glucose levels were still lowered by STF, further strengthening the notion that STF treatment improves insulin sensitivity. This is indeed further corroborated by the ITT results (Fig. 1D).

(4) Figure 2 and S2A did not show a decrease in BW but rather BW gain. The statement (line 308) needs to be edited. As a result of this, the relative fat mass measurement (% of BW) needs to be presented in addition to Figure 2B.

We thank the reviewer for the comments/suggestions. As shown in Figs. 2A and S2A, we observed a slight decrease in body weight (~2g reduction) in STF-treated mice while Veh group increased body weight by ~3.5g, at the end of 4 weeks of treatment. As shown in Fig. 2B, this difference in body weight between Veh and STF groups was primarily due to a reduction in fat tissue. In the revision, we also added the percentages of fat and lean masses over total body weight in Supplemental Fig. 2B, which show the similar trend.

(5) The measurement of blood lipid levels in Figure 3F-H is informative. More importantly, hepatic lipid content needs to be measured.

We thank the reviewer for and agree on the comments. As this study is more focused on the insulin resistance and adipose tissue remodeling, we didn’t go deep into the comorbidities beyond the reported observations. It will be interesting to explore the effects of IRE1 inhibition on the obesity/insulin resistance comorbidities including hepatic lipid content measurement in future study.

Minor corrections:

(1) Line 261: "(spliced".

Done. We have corrected it.

(2) Line 334: spell out "PEPCK".

We have added the full name “Phosphoenolpyruvate carboxykinase”. Thanks!

(3) Line 478: please rephrase.

We thank the reviewer for the comment. We have rephrased the sentence as following: “These results reveal that STF treatment suppresses the adipose tissue inflammation and the accumulation of pro-inflammatory ATM with augmenting (suppressing instead) M2-like ATMs.”

(4) Figure 4L: "pGC1-a".

We thank the reviewer for pointing this out. We have corrected the name.

(5) Figure 4O: missing Y-axis label.

We have added the label. Thanks!

Reviewer #3 (Recommendations for the authors):

The observations presented by Wu D. et al. in the manuscript are potentially interesting and relevant. The current study seeks to build upon previous findings, specifically from the work titled, "Silencing IRE1α using myeloid-specific cre suppresses alternative activation of macrophages and impairs energy expenditure in obesity." By using a pharmacological inhibitor to modulate IRE1α activity in adipose tissue macrophages (ATMs), the authors aim to develop therapeutics that could significantly impact the treatment of obesity and metabolic disease.

The authors have performed some satisfactory experiments related to liver steatosis. However, the manuscript would benefit from a more comprehensive exploration of the mechanisms by which ATMs influence adipocyte metabolism, particularly in epididymal white adipose tissue (eWAT). In particular, the study should investigate how adiposity and lipid droplet size change in response to alterations in lipolysis and adipogenesis, as this could provide insights into how these processes contribute to the amelioration of the obesity phenotype.

Several issues should be addressed to strengthen the manuscript and make the study more convincing. Below are specific comments and recommendations:

Major:

(1) The indirect calorimetric data should be normalized for dependent variables such as body weight, lean mass, and fat mass+ lean mass to accurately interpret the results. The results for 24-hour energy expenditure should be included in Figure 4B-F to provide a more comprehensive analysis. It is recommended to plot bar graphs with all individual data points for the energy expenditure (EE) results shown in Figure 4B-F, to offer a clearer and more detailed presentation of the data (Figure 4B-F).

We thank the reviewer for the comments. Data analysis on the indirect calorimetric studies has evolved over the years. One common practice was/is to normalize the data by body weight. However, this approach was deemed improper some years ago (Tschop et al Nature Methods 2012, PMID: 22205519). Tschop paper also pointed out the shortcomings associated with normalization by lean mass. Instead, it concludes that “generalized linear model is the most appropriate statistical approach to accommodate discrete (genotype) and continuous (body mass) traits, rather than using a simple division by BW or lean BW”. In our study, we used CalR, an improved generalized linear model (which includes ANOVA and ANCOVA) (Mina et al Cell Metabolism 2018, PMID: 30017358) for all our energy expenditure data analysis (shown in Fig. 4A-E). In the revision, we also included data analysis normalized by BW (Fig. S2F-H’), which actually shows even wider difference between Veh and STF groups than the data shown in Fig. 4A-F. As STF decreased the fat mass and had little effect on lean mass, the difference would be more drastic for normalization with fat mass and with fat mass+ lean mass than the data shown in Fig. 4A-E and would be similar to the data shown in Fig. 4A-E for normalization with lean mass. In addition, we replotted the graphs in Fig. 4B, D, F-H with the individual data points.

(2) At the thermoneutral point (30{degree sign}C), the study could benefit from testing the indirect calorimetric models of human energy physiology. Future studies could also explore this to evaluate the implications for drug development.

We agree with the reviewer on the comments. In the future study, it will be very informative to investigate the effects of STF under thermoneutral conditions, which could provide more consistent data on how drugs affect metabolic processes in humans, improving translational research.

(3) The current study missed the opportunity to investigate the effects of STF on non-adipose tissue (non-AT) resident macrophage populations, such as those in bone marrow or lymph-node macrophages. Understanding how STF modulates macrophage metabolism in these contexts would be valuable.

We thank the reviewer for and agree on the comments. As this study is more focused on the insulin resistance and adipose tissue remodeling, we were mostly restricted to adipose tissue macrophage populations. In the future, it would be interesting to investigate the effect of STF on macrophages in other non-adipose tissues, which will provide a more comprehensive understanding of STF's effects on immune cell metabolism, which could inform its application in various therapeutic areas.

(4) The study should explore how STF influences the expression of CD9, Trem2, (positive subpopulations), and the secretion of pro-inflammatory cytokines by macrophages, particularly in response to LPS and IFNγ activation in stromal vascular fraction (SVF) cells and bone marrow-derived macrophages (BM-Macrophages).

We appreciate the reviewer for the comments. Under obesity, the ATM does not undergo the classical M1/M2 polarization; instead, both M1-like/pro-inflammatory macrophages and M2 macrophages increase drastically in obesity. It will be interesting to investigate the effects of STF on the newly identified CD9- and Trem2-positive macrophage subpopulations in SVF and bone marrow macrophages in response to LPS and IFNγ stimulation in the future, although these studies might not faithfully reflect the changes in adipose tissue under obesity as these stressors typically induce classical M1/M2 polarization.

(5) Additional macrophage gating is necessary better to understand adipose tissue macrophage (ATM) inflammation. Specifically, CD11c−MHC2 low macrophages represent a newly identified inflammatory and dynamic subset in murine adipose tissue. These ATMs accumulate rapidly after ten days of a high-fat diet (HFD) and should increase further with prolonged HFD. For this study, CD11c−MHC2 low ATMs could be subdivided for flow cytometry analysis based on their MHC2 expression, distinguishing them from CD11c−MHC2 high ATMs. All macrophage subtypes categorized here can be studied for metabolic health using seahorse analysis as well.

We appreciate the reviewer for the comments. It will be interesting to investigate the effects of STF on the newly identified CD11c−MHC2 low macrophage subpopulation in the future. Future studies certainly can include metabolic analysis with Seahorse which can corroborate the energy metabolism at the cellular level with organismal thermogenesis.

(6) All flow cytometry histograms - are they showing mean fluorescence intensity or cell# per population? Please specify. All flow cytometry dot plots - It would be helpful for readers to see populations plotted as bar graphs next to respective flow plots, as opposed to being shown as supplemental tables. Additionally, labeling dot plots with the parent population from which cells were gated on would also help readers understand faster what we're looking at.

We appreciate the reviewer for the comments. In flow cytometry histograms, we used “normalized to mode”. The mode is often used to compare the distribution of fluorescence intensity between different samples. It focuses on the shape of the distribution (with a max of 100%) rather than the absolute cell counts, which helps remove variations caused by different cell numbers or sample sizes, making it easier to compare populations based on fluorescence intensity. When normalizing to the mode, the highest peak in the histogram is scaled to 100%, and all other values are scaled relative to that peak. This allows for easy comparison of multiple histograms, even if the total number of cells (or events) differs between samples.

(7) The results appear to confuse the actual sample size and p-value. Please carefully review the statistical analyses to ensure that biological replicates are accurately represented. Additionally, include p-values alongside fold change data in the text for clarity represented.

We appreciate the reviewer for the comments. We have rechecked the statistical analyses confirming that the biological replicates are now properly represented. The exact number of biological replicates for each experiment is now clearly specified in both the methods section and figure legends.

(8) To further validate the findings, consider using Seahorse analysis at the cellular level in future experiments. This could confirm indirect calorimetric data and thermogenesis responses to cold stimulation.

We appreciate the reviewer for the comments. Yes, Seahorse analysis at the cellular level will be conducted in future experiments.

(9) Please ensure the use of person-first language, avoiding labels or adjectives that define individuals based on a condition or characteristic.

We appreciate the reviewer for the comments. We have changed the descriptions by using person-first language.

(10) The manuscript does not demonstrate how STF inhibition of IRE1α in ATM, specifically through CD9 and Trem2, controls diet-induced obesity. This aspect should be further elucidated.

We appreciate the reviewer for the comment. In this study, we observed that STF inhibits IRE1α RNase activity in SVF and in sorted ATMs as well as in adipose tissue. The improvement in diet-induced obesity can be attributable to IRE1α inhibition in both adipocytes and macrophages as shown previously by myeloid and adipocyte-specific knockouts of IRE1α. To conclude whether the IRE1α in CD9- and/or Trem2-positive ATMs controls diet-induced obesity, genetic means would be needed to generate CD9- and/or Trem2-positive ATMs-specific deletion of IRE1α, which will be technically challenging at this moment as there is no CD9 or Trem2-specific Cre lines available.

Minor:

(1) Line 43-44: Update terminology to "MASLD" instead of "NAFLD."

We thank the reviewer for pointing these out. We have changed the terminology in the revision.

(2) Line 58-59: Add a reference for the mentioned text.

We thank the reviewer for the comment. Added a reference in the text in the revision.

(3) Was the antibody used to detect CD9 and Trem2 validated for FACS and other analyses?

We thank the reviewer for the comment. In our studies, we determined CD9 and Trem2 expression through flow cytometry and immunostaining staining. In flow experiment, CD9 and Trem2 were acquired from Biolegend: PE/Dazzle™ 594 anti-mouse CD9 (BioLegend Cat# 124821, RRID:AB_2800601); APC-conjugated Trem2 (R&D Systems Cat# FAB17291N, RRID:AB_3646995), which were validated for FACS. For immunostaining: CD9 (Abcam Cat# ab223052, RRID:AB_2922392). and Trem2 (R&D Systems Cat# MAB17291, RRID:AB_2208679).

(4) Studies were limited to male mice; this should be noted in the title and discussed as a limitation.

We thank the reviewer for the comment. We have modified the wording in the revision.

(5) Ensure all reagents are fully described with preparation details and identifiable numbers for reproducibility and/or submit the FACS protocol to any protocol archives.

We thank the reviewer for the suggestions. Yes, we have modified the wording in the revision.

(6) Provide the correct version numbers for all software used (FlowJo, Prism, etc.).

We thank the reviewer for the suggestions. We have provided the correct version numbers for softwares for FlowJo and Prism.

(7) Specify section size (µm) and blocking agent used for eWAT immunofluorescence (Line 207).

We thank the reviewer for the suggestions. We have added this information.

(8) Add gene accession numbers to Supplementary Table 3.

We thank the reviewer for the suggestions. We have added this information.

(9) Figure 2: Clarify HFD and treatment timelines with a schematic diagram.

We thank the reviewer for the suggestions. We have added a schematic diagram in Supplemental Figure 1C.

(10) For histology analysis, the minimum combined data from triplicate images is shown in Figure 2C-2H. For Figures 2E and H, provide complete methods for histology analysis.

We thank the reviewer for the comments. For the histology analysis shown in Figures 2C–2H, we used a minimum of three mice per treatment group. For each mouse, 3–5 images were taken for analysis. All histology analyses were conducted using ImageJ for image quantification, and the data were processed and organized using Excel and Graphpad.

(11) Figure 3D Macrophage markers F4/80 stained differently in Figure 5B; to avoid false positive staining, show isotype control to confirm actual staining. For eWAT immunofluorescence (Figures 3D, 5B, 6E)., counterstaining is needed in addition to macrophages, such as for adipocytes-perilipin, and phalloidin for total cells.

We thank the reviewer for the comments. Yes, Figures 3D macrophage marker F4/80 stained is differently from that of Figure 5B, as they are in different tissues, with Figure 3D in liver samples while Figure 5B in adipose tissues. In the liver, subsets of macrophages are known as Kupffer cells. Kupffer cells have distinct morphology and behavior compared to other tissue-resident macrophages. When stained with F4/80 in the liver, the pattern may reflect the specialized role of Kupffer cells, typically showing a more diffuse or localized staining around blood vessels and sinusoids. In adipose tissue, macrophages tend to accumulate around dead or dying adipocytes, forming what is known as "crown-like structures" (CLS). The F4/80 staining in adipose tissue shows a more clustered pattern, particularly around areas of fat tissue undergoing remodeling or inflammation. In adipose tissue, you can still see clear, defined cells even without counterstaining like perilipin, and importantly, adipocytes are generally way larger than macrophages in size. Yes, we agree that if with counterstaining it would enhance the accuracy. In the future study, we will use perilipin staining to make it easier to differentiate adipocytes from other structures and provide stronger data.

(12) Insert scale bars in the original images for Figures 3D, 4I, 4M, 5B, 6E, S3B, S6D-E, and S7A-B. All images added a scale bar not inserted while acquiring the image or using imaging software.

We thank the reviewer for the suggestions. The resolution for the scale bars in the images obtained during acquisition, somehow, isn’t sufficient enough to be clearly visible and requires the enlargement of the images to be seen clearly. In the revision, we have manually added the scale bars for clarity.

(13) Figure 5E: Please label X-axis as F4/80.

We thank the reviewer for pointing this out. The label has been added in the revision.

(14) Figure 5F: It is specified in the legend that cells were gated on F4/80+CD11b+CD11c+, but there is a CD11c- population shown in the histogram...How is this population appearing if all cells should be CD11c+?

We thank the reviewer for pointing this out. We gated against CD11c in F4/80+CD11b+ population. As such, we have corrected the description in the legend.

(15) Figure 5G: What is the F4/80+CD11b+CD11c-CD206- population gated in quadrants?

We thank the reviewer for the comment. The F4/80+CD11b+CD11c-CD206- population was shown in Figure 5G on the lower left side, with the percentages being 15.7% for ND, 5.54% for Veh-HFD, and 26% for STF-HFD.

(16) Figure 6J: Flow cytometry gates seem slightly misplaced and the sample appears to be overcompensated - were FMOs included in this experiment to establish proper gates? If so, please include.

We thank the reviewer for the comment. In the study, we did include Fluorescence Minus One (FMO) control in the experiment to establish proper gating. We have included this information in the methods section.

(17) Table 1-3: Indicate the number of replicates (n=) used in all tables.

We thank the reviewer for the suggestion. We have provided the specific number of mice used in the study within the figure legends.