To explore the effect of intermittent fasting on intestinal immune cells, we applied alternate day fasting to mice fed normal chow diet (NCD-IF group) or high-fat diet (HFD-IF group) (Figure 1A). Intermittent fasting significantly reduced the body weight of mice fed HFD, while demonstrating no effect on food intake rate which was normalized to body weight (Figure 1—figure supplement 1A and B). Moreover, intermittent fasting decreased the respiratory quotient (RQ) on the fasting day while increased the energy consumption on the feeding day in mice fed HFD (Figure 1—figure supplement 1C and D), improved glucose and lipid metabolism in mice fed HFD (Figure 1—figure supplement 1E and F), and promoted white adipose tissue beigeing in mice fed NCD or HFD (Figure 1—figure supplement 2).

To explore whether gut immune system contributes to the effects of intermittent fasting on white adipose tissue beigeing, we examined levels of various cytokines in intestine. Notably, mRNA level of Il22 in the NCD-IF group was significantly higher relevant to the control group (Figure 1B). Consistently, plasma concentration of IL-22 detected by enzyme-linked immunosorbent assay (ELISA) was also increased (Figure 1C). These results suggest that intermittent fasting increases levels of IL-22 in the intestine and plasma. Since ILC3s are the main source of IL-22 in the small intestine (Seillet et al., 2020; Victor et al., 2017; Gronke et al., 2019), we detected the proportion of ILC3s in the lamina propria of small intestine using flow cytometry. In mice fed NCD, intermittent fasting significantly increased proportion of IL-22 positive ILC3s in the small intestine lamina propria (siLP) (Figure 1D), while demonstrating no effect on the percentile of total ILC3s (Figure 1E). Interestingly, intermittent fasting did not influence the secretion of IL-22 of T cells marked as lineage⁺ Rorγt⁺ cells (Figure 1F). Besides, intermittent fasting didn’t alter the levels of ILC3s and IL-22 in mouse adipose tissue (Figure 1G and Figure 1—figure supplement 3). Similar to NCD mice, intermittent fasting significantly increased mRNA levels of IL-22 in the intestine of HFD mice while didn’t influence the secretion of IL-22 by T cells (Figure 1H and I). Furthermore, mice fed HFD showed an obvious reduction in the plasma IL-22 (Figure 1J) and percentage of total and IL-22 positive ILC3s (Figure 1K). Further, the decrement of plasma IL-22 (Figure 1J) as well as IL-22 positive ILCs in intestine (Figure 1K) was attenuated by 30 days’ intermittent fasting.

In order to explain the chronological relationship between ILC3s secreting IL-22 and beigeing of white adipose tissue, the mice were exposed to one cycle of intermittent fasting for 2 days. At this time, the body weight of mice didn’t change and beige adipocytes haven’t been induced (Figure 1—figure supplement 4A and B). However, significant increase in proportion of IL-22 positive ILC3s was induced by intermittent fasting for 2 days whereas total percentile of ILC3s remained unaltered (Figure 1—figure supplement 4C). These results indicate that the beiging of white adipose tissue are subsequent to its effect on ILC3s secreting IL-22.

Next, we examined whether adoptive transfer of intestinal ILC3s can increase beigeing of white adipose tissue in DIO mice. Intestinal ILC3s were isolated and purified from the siLP of NCD mice (Figure 2—figure supplement 1A). These cells were defined as lineage^-CD127⁺KLRG1^-c-Kit⁺ cells (Figure 2—figure supplement 1B). As shown in Figure 2A, DIO mice transferred with ILC3s demonstrated a significant increment in the proportion of small intestinal ILC3s. Previous researches report that intestinal ILC3s specifically express gut homing receptors CCR7, CCR9, and α4β7 (Kim et al., 2015; Mackley et al., 2015; Yu et al., 2021), which may explain transplantation of intestinal ILC3s can migrate mainly to the intestine instead of adipose tissue. Plasma concentration of IL-22 also increased significantly relevant to phosphate-buffered saline (PBS) control (Figure 2B). Relevant to PBS control, adoptive transfer of intestinal ILC3s decreased body weight and weight of sWAT (subcutaneous white adipose tissue) slightly, while had no impact on food intake and the liver weight (Figure 2—figure supplement 1C–E). In addition, ILC3s from CD45.1 mouse intestinal lamina propria lymphocytes were adoptively transferred into recipient mice, and CD45.1 positive immune cells were significantly increased in intestine but not in adipose tissue in mice transferred with ILC3s (Figure 2—figure supplement 1F), indicating the feasibility of ILC3s adoptive transfer. Notably, DIO mice transferred with intestinal ILC3s showed improved glucose tolerance (Figure 2C) and decreased levels of random blood glucose (Figure 2D). Cold exposure experiment showed that DIO mice transferred with intestinal ILC3s maintained significantly higher rectal temperatures during a 6 hr cold challenge (Figure 2E). Consistently, key thermogenic genes in sWAT including Ucp1 and Cidea were significantly increased (Figure 2F). Size of adipocytes in the sWAT and eWAT was markedly reduced (Figure 2G and H). Furthermore, differentiated stromal vascular fraction (SVF) cells co-cultured with intestinal ILC3s in vitro (Figure 2I) demonstrated a significant increment in the expression of Ucp1 and Pparg (Figure 2J), as well as a concurrent decrement in the size of lipid droplets (Figure 2K). On the other hand, co-culture with CD127^- cells demonstrated no effect (Figure 2J and K). The in vitro experiment indicates the direct effect of ILC3s on adipocytes.

The role of type 3 immunity in DIO and metabolic syndrome is complex. ILC3-derived IL-22 are beneficial in metabolic syndrome (Wang et al., 2014; Zou et al., 2018) but can also contribute to metabolic disease (Sasaki et al., 2019; Upadhyay et al., 2012; Wang et al., 2017). In order to examine the role of promoting beigeing of white adipose tissue of IL-22 in the context of NCD and HFD mice, we intraperitoneally administrated IL-22 at the dose of 4µg/kg body weight/every other day for 6weeks into mice fed NCD or HFD. Saline was used as control. Administration of IL-22 increased its plasma concentration in mice fed either NCD or HFD (Figure 3—figure supplement 1A). Exogenous IL-22 significantly increased oxygen consumption, carbon dioxide production, and energy expenditure in mice fed either NCD or HFD (Figure 3A–F). RQ was significantly reduced in mice fed NCD at dark and mice fed HFD at light (Figure 3G and H). No significant difference was observed for animal activity (Figure 3—figure supplement 1). In addition, exogenous IL-22 significantly improved glucose tolerance in mice fed HFD (Figure 4A). Interestingly, body weight and food intake were not altered (Figure 3—figure supplement 1), indicating that the metabolic benefit of IL-22 is not dependent on food intake. Consistent with previous research, IL-22 administration improves insulin sensitivity without change in body weight (Qi et al., 2019). In addition, IL-22 can increase Akt phosphorylation in muscle, liver, and adipose tissues, leading to improvement in insulin sensitivity (Wang et al., 2014).

We next explored the effect of exogenous IL-22 on thermogenesis induced by 4°C cold exposure. Exogenous IL-22 rendered mice fed HFD resistant to core temperature drop induced by cold exposure (Figure 4B). Relevant to the pale yellow color in the control animals, subcutaneous fat of obese mice treated with exogenous IL-22 appeared dark yellowish (Figure 4C). The mRNA levels of thermogenic genes such as Ucp1 were significantly increased by IL-22 (Figure 4D). Adipocyte size of subcutaneous adipose tissue and epididymal adipose tissue was significantly reduced in the animals treated with IL-22 (Figure 4E and F). These observations suggest that intraperitoneal injection of IL-22 promotes the beigeing of white adipose tissue in mice.

To determine whether IL-22 can directly act on adipocytes to promote their beigeing, adipose tissue SVF was isolated and induced for beige differentiation. Il22ra1 mainly expressed on mature adipocytes (Figure 4—figure supplement 1A). IL-22 at the dose of 100 ng/mL was continuously administered during beige differentiation. IL-22 significantly increased SVF beige differentiation evidenced by cell morphology, increment in mRNA levels of genes relevant to thermogenesis, including Ucp1 and Cidea, and protein level of UCP1 (Figure 4G–I). IL-22 did not alter the levels of genes related to adipogenesis (Figure 4—figure supplement 1B). As expected, IL-22 increased the phosphorylation of STAT3 and MAPK (Figure 4I, Figure 4—figure supplement 1C), which can increase the expression of thermogenic genes (Li et al., 2021; Zhang et al., 2016). These results suggest that IL-22 can directly stimulate beigeing of white adipose tissue.

To explore whether IL-22R mediates the effect of intermittent fasting on the beigeing of white adipose tissue, IL-22R knockout (IL-22RKO) mice and wild-type (WT) littermates were subjected to alternate day fasting diet for 30 days. Rectal temperature of WT-IF and IL-22RKO-IF mice was monitored during 2 consecutive days of intermittent fasting. On the fasting day, the rectal temperature of IL-22RKO-IF mice was significantly lower than that of WT-IF mice at 16:00 time point (Figure 5A). On the feeding day, the rectal temperature of IL-22RKO-IF mice was lower at three time points: 8:00, 12:00, and 20:00 (Figure 5B). These results indicate that IL-22RKO attenuates the thermogenesis induced by intermittent fasting. Knockout of IL-22R demonstrated no effect on the glucose tolerance and insulin sensitivity in mice fed NCD (Figure 5C and D). However, the weight of subcutaneous adipose tissue in IL-22RKO-IF mice increased significantly (Figure 5E). mRNA levels of the thermogenic gene Ucp1 decreased substantially, whereas Fabp4 increased (Figure 5F). Furthermore, knockout of IL-22R significantly attenuated the increment of multilocular lipid droplets and the decrement of adipocyte size in the subcutaneous fat induced by intermittent fasting (Figure 5G and H), indicating a reduction in beigeing.

To further explore whether IL-22R mediates the effect of ILC3s on beigeing of adipocytes, intestinal ILC3s isolated from intermittent fasting mice were co-cultured with beige adipocytes from subcutaneous SVF of WT or IL-22RKO mice (Figure 5I and J). Co-culture with ILC3s significantly increased the mRNA levels of thermogenic genes in beige adipocytes derived from WT mice, while demonstrating no effect on beige adipocytes derived from IL-22RKO mice (Figure 5K). These observations indicate that IL-22R mediates the effect of intestinal ILC3s on beigeing of white adipocytes.

To explore the mechanism by which intermittent fasting promotes the secretion of IL-22 by ILC3s, live CD45⁺ lineage (CD3, CD5, B220, CD19, Gr1)^- cells isolated from the siLP of mice fed NCD, HFD, or HFD-IF were subjected for single-cell sequencing (Figure 6—figure supplement 1A). Following quality controls, we analyzed 7455, 4803, 5954 single cells for the NCD, HFD, and HFD-IF groups using Seurat-V3.1, respectively. Based on singleR, we identified 25 distinct clusters of immune cells (Figure 6A and Figure 6—figure supplement 1B), including ILC1s (cluster 7 and cluster 16), ILC2s (cluster 2, cluster 12, cluster 13, and cluster 14), ILC3s (cluster 3 and cluster 5), DCs (cluster 1, cluster 4, cluster 9, cluster 10, cluster 11, cluster 17, cluster 18, and cluster 21), eosinophils (cluster 0 and cluster 6), B cells (cluster 8 and cluster 24), macrophages (cluster 15, cluster 20, and cluster 23), NKT cells (cluster 22) and mast cells (cluster 19). As expected, ILC3s expressed high RNA levels of Il7r (CD127), Rorc, and IL-22 (Figure 6B, C and Figure 6—figure supplement 1C, D). ILC3s contain two distinct cell types with highly similar expression profiles: NCR⁺ and NCR^-. These ILC3s were distinguished in our analysis as cluster 5 and cluster 3 respectively. NCR^- ILC3s (cluster 3) were CCR6 positive and contained lymphoid tissue inducer cells (Figure 6B). Interestingly, NCR^- ILC3s were characterized by high levels of vasoactive intestinal peptide receptor 2 (Vipr2) (Figure 6C), which is critical for the migration and function of ILC3s (Seillet et al., 2020; Yu et al., 2021; Talbot et al., 2020). The heatmap of cellular composition showed the differences in cell number and percentage among NCD, HFD, and HFD IF mice (Figure 6—figure supplement 2A). The change in ILC3 number was consistent with the flow cytometry results. The percentage of IL-22⁺ cells in NCR⁺ ILC3s was significantly increased by IF in mice fed HFD (Figure 6—figure supplement 2B). Moreover, GSEA revealed that HFD-induced decrement of cytokine-cytokine receptor interaction, in which IL-22 is included, was reversed by IF (Figure 6—figure supplement 2C). Besides, neuroactive ligand-receptor interaction, in which vipr2 is included, was upregulated in HFD-IF (Figure 6—figure supplement 2). The mRNA levels of Vipr2 had a tendency to increase in sorted ILC3s from the small intestine of HFD-IF mice compared with that of HFD mice (Figure 6—figure supplement 2E). Further, gene difference analysis revealed that HFD-induced increase of Zmat4 was significantly attenuated by IF (Figure 6D, Figure 6—figure supplement 2F, G).

Further, the expression level of Hsp90ab1 in both NCR^- ILC3s and NCR⁺ was significantly increased by IF in mice fed HFD (Figure 6D, Figure 6—figure supplement 2H). Transcription factor analysis using the JASPAR database and TFBS Tools revealed that the expression of Hsp90ab1 may be regulated by aryl hydrocarbon receptor (AHR) (Figure 6D). AHR is located in the Hsp90:XAP2:p23:Src chaperone protein complex in the cytoplasm without stimulation. Upon binding with the ligand, AhR translocates to the nucleus and heterodimerizes with AhR nuclear translocator to control the transcription of target genes, including AhR repressor (Ahrr), Cyp1a1, Cyp1b1, and Il22 (Stockinger et al., 2014). Consistently, gene ontology analysis of the top 20 increased difference genes revealed that AHR signaling was one of the key pathways affected by intermittent fasting in HFD mice (Figure 6E and F). Furthermore, the AHR target genes, including Il22, Ahrr, Cyp1a1, Cyp1b1, increased significantly in ILC3s sorted from HFD-IF mice compared with HFD mice. These results indicate that intermittent fasting increases the production of IL-22 likely via activation of AhR.

Because DCs can influence the production of IL-22 by ILC3s, we next analyzed the change in DCs induced by intermittent fasting. IF significantly attenuated the increment of inflammatory factors, such as Ccl4, Ccl17, and Ccl22, of DCs in mice fed HFD (Figure 7A). Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis of the top 20 differentially expressed genes in cluster 17 revealed that the increase of inflammation-related pathways induced by HFD was obviously attenuated by IF (Figure 7B and C). These observations indicate that intermittent fasting may ameliorate the inflammatory state of the intestine by decreasing the NOD-like receptor signaling pathway in DCs (Figure 7B and C).

We next analyzed the intestinal cellular communication networks using CellPhoneDB analysis based on homologous gene transformation on all immune cells acquired. CellPhoneDB ligand-receptor analysis revealed hundreds of immune-to-immune interactions (Figure 7—figure supplement 1A). Connectome web analysis of siLP immune cells revealed strong interactions among CD8⁺ DCs (cluster 4), CD127⁺ ILC1s (cluster 16), DC8^- DC-1 (cluster 9), macrophages-1 (cluster 15), CD8^- DC-2 (cluster 1), macrophages-2 (cluster 23), CD8^- DC-3 (cluster 10), CD8^- DC-4 (cluster 11), and CD8^- DC-5 (cluster 17). Notably, NCR⁺ ILC3s strongly interacted with CD8⁺ DCs (cluster 4), CD127⁺ ILC1 s (cluster 16), DC8^- DC (cluster 9), macrophages (cluster 15), CD8^- DC (cluster 1), macrophages (cluster 23), CD8^- DC (cluster 10), CD8^- DC (cluster 11), and CD8^- DC (cluster 17). ILC subsets, DCs, and macrophages were in the central communication hubs of the healthy small intestine (Figure 7—figure supplement 1B and C). Analysis of highly expressed interactions uncovered various uncharacterized and validated signaling pathways implicated in intestine homeostasis in mice (Figure 7—figure supplement 1D). Analysis of the NCD mouse interactome suggests that macrophages (cluster 15) may interact with NCR⁺ ILC3s through CD74_COPA, CD74_MIF, CD44_HBEGF, CD44_FGFR2, CCL4_SLC7A1, and IL1B_ADRB2. Similarly, DCs (cluster 9) may interact with NCR⁺ILC3s through CD74_COPA, CD74_MIF, IL1B_ADRB2, CD44_HBEGF, and CD44_FGFR2. All these observations indicate that macrophages and DCs play critical roles in the recruitment and maintenance of ILC3s.

HFD significantly reduced the interaction between ILC subsets, DCs, and macrophages in the small intestine (Figure 7D and Figure 7—figure supplement 1B). This reduction was reversed by IF (Figure 7E and Figure 7—figure supplement 1B). Analysis of the interactomes in mice fed HFD suggested that IF significantly altered the interacting proteins (Figure 7F and G). To further test this observation, we sorted macrophages and ILC3s from HFD and HFD-IF mice (Figure 7—figure supplement 2). As shown in Figure 7—figure supplement 2B and C, Ccl4 and Cd74 increased significantly in macrophages, while the receptors expressed on ILC3s remained largely unchanged (Figure 7—figure supplement 2). Together, these data suggest that IF may promote the production of IL-22 from ILC3s by altering the interactome in intestinal myeloid cells and ILC3s.

Four-week-old male C57BL/6 and CD45.1 mice were obtained from Charles River Laboratories (Peking, China). IL-22RKO mice (KO-00115) were purchased from BRL Medicine Inc Mice were housed in standard rodent cages and maintained in a regulated environment (21–24°C, humidity at 40–70%, 12:12 hr light:dark cycle with lights on at 8:00 AM) at the Department of Experimental Animal Science, Peking University Health Science Center. An NCD (D12450H; Research Diets) and water were available ad libitum. Obese mice (DIO) were induced with an HFD (60% fat, D12492; Research Diets) for 12 weeks. Eight-week-old C57BL/6 mice received intraperitoneal injection of saline control or IL-22 (R&D Systems, 582-ML) at a dose of 4 μg/kg/day every other day for 6 weeks. All of the animal experiments complied with the protocols for animal use, treatment and euthanasia approved by Peking University (Permit Number: LA2017099).

For isolation of siLP cells, small intestines from euthanized mice were emptied of the contents, excised of Peyer’s patches, opened longitudinally and cut into 1 cm pieces. The intraepithelial lymphocytes were dissociated from the intestine fragments by first shaking the fragments for 20 min at 37°C in PBS containing 0.3% BSA, 5 mM EDTA, and 1 mM dithiothreitol. Vortex the fragments three times with PBS containing 2 mM EDTA. To isolate lamina propria cells, the remaining fragments were minced and digested at 37°C for 50 min in RPMI 1640 medium containing 0.04 mg/mL collagenase IV (Sigma), 0.1 mg/mL deoxyribonuclease (DNase) I (Roche), and 0.5 mg/mL dispase. The digestion suspension was then filtered through a 40 μm cell strainer and centrifuged at 540×g for 6 min. Cell pellets were resuspended in PBS containing 2% fetal bovine serum (FBS) for further analysis.

Single-cell suspensions were preincubated with anti-CD16/32 (clone 2.4G2) for 10 min to block the surface Fc receptors. Then, cell-surface molecules were stained with different antibody combinations for 30 min in cell staining buffer. Dead cells were excluded with Fixable Viability Dye eFluor 606 (Invitrogen). For intracellular transcription factor staining, the cells were fixed and permeabilized with a Foxp3 staining buffer set (eBioscience) according to the manufacturer’s protocol. Transcription factor staining usually lasted for more than 4 hr at 4°C. The gate strategy for ILC3s was live lineage (CD3, CD5, CD19, B220, and Gr-1)^- CD127⁺RORγt⁺. For intracellular cytokine staining, the digested cells were first incubated in RPMI 1640 with 10% FBS and 10 ng/mL recombinant murine IL-7 (PeproTech), and stimulated with 50 ng/mL phorbol 12-myristate 13-acetate, 750 ng/mL ionomycin for 3 hr and added with 2 μM monensin for the last 2.5 hr.

Flow cytometry analyses were performed on an LSR Fortessa (BD Biosciences). The flow cytometry data were analyzed with FlowJo software (Tree Star). The antibodies used in this study are listed in Key resources table.

For ILC3s transfer experiment, co-culture experiment, and single-cell RNA-seq, ILCs were collected by FACS. siLP cells from male C57BL/6J mice were prepared and stained with surface molecules for 30 min at 4°C followed by sorting on a FACS Aria III cell sorter (BD Biosciences). A gating strategy of live lineage^-CD127⁺KLRG1^-c-Kit⁺ was used to sort the ILC3s.

Purification checks were performed after each sort. The cells were suspended in PBS and then intravenously injected into HFD mice (200 μL PBS/mouse). ILC3s were sorted from 20 WT NCD mice, and sorting for the adoptive transfer was performed six times within 4 weeks from the first injection. The total number of ILC3 injected into HFD mice was 2.4×10⁵ cells/mouse/experiment. Mice were placed on an HFD for 16 weeks before ILC3s were transferred. In addition, to confirm that these cells indeed originate from NCD mice, we transferred ILC3s from CD45.1 mice to WT mice and detected the CD45.1 positive cells in recipient mice.

Single cells were captured via the GemCode Single Cell Platform using the GemCode Gel Bead, Chip, and Library Kits (10x Genomics) according to the manufacturer’s protocol. Briefly, flow-sorted cells were suspended in PBS containing 0.4% BSA and loaded at 6000 cells per channel. The cells were then partitioned into a GemCode instrument, where individual cells were lysed and mixed with beads carrying unique barcodes in individual oil droplets. The products were subjected to reverse transcription, emulsion breaking, cDNA amplification, shearing, 5′ adaptor, and sample index attachment. Libraries were sequenced on a HiSeq 2500 (Illumina).

Isolation of SVF and induction of beigeing were performed as reported (Aune et al., 2013). Briefly, sWAT from male C57BL/6 mice was digested with 1 mg/mL type 2 collagenase (Sigma-Aldrich) at 37°C for 40 min with shaking. Digestion was terminated by complete DMEM/F12 medium containing 10% FBS, 100 units/mL penicillin, and 100 units/mL streptomycin (Invitrogen, CA, USA). The cell suspension was centrifuged at 700×g for 10 min to separate floating adipocytes from the SVF pellets. The pellets containing the SVF cells were resuspended in complete medium and filtered using a 70 μm diameter filter. The cell suspension was then centrifuged at 700×g for 10 min. Cell pellets were resuspended and mixed well and then plated on dishes in complete medium. Cells were grown to 95% confluence in complete medium and then differentiated into beige or white adipocytes as previously described (Seale et al., 2011). Briefly, for induction to beige adipocytes, cells were cultured for 2 days with induction medium supplemented with 5 μg/mL insulin, 1 nmol/L T3, 1 μmol/L rosiglitazone, 125 μmol/L indomethacin, 0.5 mmol/L isobutylmethylxanthine, and 5 μmol/L dexamethasone. Cells were then cultured in maintenance medium supplemented with 5 μg/mL insulin, 1 nmol/L T3, and 1 μmol/L rosiglitazone for 4 days. Fresh media were replaced every 2 days.

To determine the direct effect of ILC3s on beige adipocytes differentiation, SVFs were co-cultured with ILC3s using a transwell system (0.4 μm pore size, BD Biosciences). In brief, SVFs isolated from adipose tissue were grown in the bottom chamber of the transwell insert in a 12-well plate and induced into beige adipocytes, while sorted ILC3s were seeded in the upper chamber. SVFs cultured alone and SVF cells co-cultured with CD127^- cells were used as controls. After co-culture for 72 hr, beige adipocytes in the lower chamber were collected for further detection.

RNA was extracted from cells or adipose tissue using RNATrip (Applied Gene, Beijing, China) and reverse-transcribed into cDNAs with Hifair III 1st Strand cDNA Synthesis SuperMix (Yeasen). SYBR Green-based quantitative real-time PCR was performed using the Agilent Aria Mx real-time PCR system. The primer sequences used in this study are listed in Supplementary file 1.

Differentiated SVF cells and sWAT were homogenized using RIPA lysis buffer. Proteins were separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and then transferred onto a nitrocellulose membrane. Membranes were incubated in 5% fat-free milk for 1 hr at room temperature and then incubated with primary antibodies overnight at 4°C. The reaction was detected with IRDye-conjugated secondary antibody and visualized using the Odyssey infrared imaging system (LI-COR Biosciences).

Paraffin-embedded sWAT sections were stained with hematoxylin-eosin (H&E). Images were scanned using a NanoZoomer-SQ (Hamamatsu).

Cells were washed with PBS for three times, and then fixed with 4% paraformaldehyde for 30 min. Next, the cells were stained with oil red O for 60 min, and washed with PBS for 5 min. Cells were then stained with hematoxylin for 5 s and washed with PBS. The dyed cells were photographed under the microscope (Leica, Germany).

Levels of IL-22 were measured by double-antibody sandwich ELISAs (M2200; R&D Systems). Briefly, 100 µL of Assay Diluent were added to each well. 50 µL of blood samples and standards were added and incubated at room temperature for 2 hr. After washing, conjugates were added and incubated for 2 hr. After washing, the substrate solution was added and incubated for 30 min and washed. And stop solution were added finally. Optical density values were measured using a microplate reader (Bio-Rad, Hercules, CA, USA).

Oral glucose tolerance tests (OGTT) and insulin tolerance tests (ITT) were performed after 16 or 6 hr of fasting, respectively. Mice were given with 3 g/kg glucose by gavage or injected intraperitoneally with 0.75 U/kg insulin. Blood samples were collected from the tail vain at 0, 15, 30, 60, 90, and 120 min after glucose or insulin treatment. Blood glucose levels were measured using a glucometer (Roche, Basel, CH).

Mice were placed in a 4°C cold room for 6 hr. The rectal temperature was measured every hour during the cold challenge with a rectal probe (Braintree Scientific, Braintree, MA, USA).

Data were analyzed by GraphPad Prism software v.8.0 and presented as the mean ± s.e.m. The Shapiro-Wilk normality test was used to determine the normal distribution of samples. Unpaired Student’s t test (normal distribution) or Mann-Whitney U tests (non-normal distribution) were used to analyze data between two groups and one-way ANOVA followed by Bonferroni’s multiple-comparisons test (normal distribution) or Kruskal-Wallis test (non-normal distribution) was used for three or more groups. The sample sizes were determined by power analysis using StatMate v.2.0. No data were excluded during the data analysis.