miR-210-3p Promotes Obesity-Induced Adipose Tissue Inflammation and Insulin Resistance by Targeting SOCS1-Mediated NF-κB Pathway

Increased accumulation of body fat in obesity is considered the highest risk factor for the development of several metabolic diseases, including insulin resistance and type 2 diabetes (1–4). Excess storage of fat inside the adipocytes leads to the formation of hypertrophied adipocytes, which by releasing higher levels of circulatory free fatty acids, induce a state of chronic low-grade inflammation in adipocytes and adipose tissue macrophages (ATMs), resulting in the onset of obesity-induced adipose tissue inflammation (5,6). Obesity-induced enhancement of free fatty acids is accompanied by the reduction of adipose tissue vascularization, creating a zone deprived of adequate oxygen and nutrients. Accumulating evidence has revealed a considerable drop in pO₂, referred to as hypoxia in the subcutaneous and visceral adipose tissue (VAT) of obese patients with diabetes compared with lean subjects without diabetes (7–9). It has been shown that adipose tissue hypoxia plays a major role in adipose tissue inflammation and insulin resistance during the state of chronic obesity (10). Hypoxic adipocytes promote adipose tissue inflammation by secreting various proinflammatory cytokines and chemokines (10). The inflammatory chemokines, such as CCL2/MCP1, aggravate the adipose tissue inflammation by heightened infiltration of macrophages and its polarization toward the M1 phenotype, as evident in the obese adipose tissue of patients and mice (7,11). Although these studies have unveiled the significance of the obese adipose tissue microenvironment (ATenv) on inflammation and insulin resistance, the underlying mechanism is still elusive. Thus, it is important to elucidate how the hypoxic lipid-rich microenvironment in obese ATenv promotes adipose tissue inflammation and leads to the onset of insulin resistance.

Exploring the involvement of noncoding RNAs, particularly miRNAs, in obese hypoxic adipose tissue, it appears that miR-210 is a well-known hypoxia-inducible miRNA (12–15). Some evidence has shown that patients with diabetes, as well as diabetic animal models, have abnormally high levels of miR-210 (16–18). In clinical studies, elevated levels of miR-210 have been detected in adolescent patients with type 1 diabetes and type 2 diabetes (19,20). Thus, we are interested in investigating any potential role of miR-210-3p in ATMs for causing obesity-induced chronic low-grade inflammation.

Although the lone effects of palmitate or hypoxia on adipocyte function and insulin resistance have been well studied, the combined impact of high lipid content along with hypoxia on ATM inflammation is not clearly understood (10,21,22). In this study, we found that the obese hypoxic ATenv stimulates increased expression of miR-210-3p in the ATMs, provoking inflammation and polarization toward the M1 phenotype via the suppressor of cytokine signaling 1 (SOCS1)/nuclear factor-κB (NF-κB) pathway. Suppression of miR-210-3p markedly attenuated adipose tissue inflammation with improvement of insulin sensitivity. Therefore, targeting ATM-specific miR-210-3p could be an excellent strategy to regulate obesity-induced adipose tissue inflammation for the management of insulin resistance and type 2 diabetes.

All tissue culture materials were purchased from Life Technologies/Gibco (Grand Island, NY) and Nunc (Corning, NY). Description of all the antibodies, including catalog numbers and the dilutions used in different experiments, are given in Supplementary Table 2. We purchased the Dual-Luciferase Reporter Assay System (cat. no. E1910) from Promega (Madison, WI); VECTASHIELD Antifade Mounting Medium with DAPI (cat. no. H-1500) from Vector Laboratories (Burlingame, CA); QuikChange Lightning Multi Site-Directed Mutagenesis Kit (cat. no. 210515) from Agilent Technologies (Santa Clara, CA); Glucose Uptake Cell-Based Assay Kit (cat. no. 600470) from Cayman Chemicals (Ann Arbor, MI); mirVana miRNA Isolation Kit (cat. no. AM1560), Lipofectamine LTX Reagent with PLUS Reagent (cat. no. 15338100), Lipofectamine RNAiMAX Transfection Reagent (cat. no. 13778-075), and NP40 cell lysis buffer (cat. no. FNN0021) from Invitrogen, Thermo Fisher Scientific (Grand Island, NY); Pierce BCA Protein Assay Kit (cat. no. 23227) and Halt Protease and Phosphatase Inhibitor Cocktail (cat. no. 78441) from Thermo Fisher Scientific (Waltham, MA); Clarity Western ECL Substrate (cat. no. 1705060) and iScript Reverse Transcription Supermix (cat. no. 1708891) from Bio-Rad Laboratories (Hercules, CA); polyvinylidene fluoride membranes (cat. no. 548IPVH00010) from Merck (Darmstadt, Germany); PowerUp SYBR Green Master Mix (cat. no. A25742) from Applied Biosystems, Thermo Fisher Scientific; and TaqMan MicroRNA Reverse Transcription Kit (cat. no. 4366596) and TaqMan Multiplex Master Mix (cat. no. 4461881) from Applied Biosystems, Thermo Fisher Scientific. We purchased miRIDIAN miRNA miRNAmmu-miR-210-3p mimic (cat. no. C-310570-05-0005; sequence CUGUGCGUGUGACAGCGGCUGA; accession no. MIMAT0000658), miRIDIAN miRNA hairpin mimic negative control (cat. no. CN-001000-01-05; accession no. MIMAT0000039), miRIDIAN miRNA hsa-miR-210-3p hairpin inhibitor (cat. no. IH-310570-07-0005; sequence GACACGCACACUGUCGCCGACU; accession no. MIMAT0000267), and miRIDIAN miRNA hairpin inhibitor negative control (cat. no. IN-001005-01-05; accession no. MIMAT0000039) from GE Dharmacon (Lafayette, CO). The miRNA-specific primers for miR-210-3p (cat. no. 4427975; assay ID 000512) and U6 small nuclear RNA (cat. no. 4427975; assay ID 001973) were procured from Applied Biosystems (Foster City, CA). We procured CON siRNA (sc-37007) and SOCS1 siRNA(m) (sc-40997) from Santa Cruz Biotechnology. Image-iT Green Hypoxia Reagent (cat. no. I14833) was procured from Invitrogen, Thermo Fisher Scientific. Various gene-specific primers were procured from Integrated DNA Technologies, India, and the sequence details are presented in Supplementary Table 3.

Wild-type (WT) C57BL/6J male mice aged 4–5 weeks and weighing 18–22 g were procured from the Indian Institute of Science Education and Research Mohali animal facility and kept in the National Institute of Pharmaceutical Education and Research Mohali animal housing facility for 5–6 days under a 12-h light/dark cycle at 23 ± 2°C with relative humidity 55 ± 5%. The mice were fed a normal pellet diet (standard diet) and water ad libitum. For the development of the diet-induced obese insulin-resistant C57BL/6J mice were fed high-fat diet (HFD) pellets (60% kcal from fat; D12492; Research Diets, New Brunswick, NJ) for 12 weeks. All other mice were fed standard diet pellets (10% kcal from fat) for 12 weeks. All experimental animals had free access to sterilized water and food. Blood glucose levels were measured regularly with an Accu-Chek glucometer (Roche).

Mice fed an HFD diet for 12 weeks were considered for anti–miR-210-3p locked nucleic acid (LNA) delivery. Briefly, mice were anesthetized by low-dose isoflurane inhalation per standard recommendations. We then created a small incision on the abdominal site to remove the omental fat pads from the abdominal cavity. A total of 100 nmol/L miRCURY LNA miRNA Power Inhibitor (anti)mmu-miR-210-3p (Gene Globe ID YI04103147-DDA; QIAGEN, Germantown, MD) was injected in five different sites of each side of the abdominal omental fat pad. Similarly, 100 nmol/L of miRCURY LNA Control Inhibitor (Gene Globe ID YI00199006-ADA; QIAGEN) was injected in five different sites on each side of the abdominal omental fat pad of the HFD mice. The skin layer was stitched carefully using Ethicon absorbable surgical suture (Johnson & Johnson, New Brunswick, NJ). We tested glucose tolerance by measuring blood glucose levels before and after oral gavages of 1 g glucose/kg body weight at the indicated time points. Similarly, we tested insulin tolerance by injecting 1 IU insulin/kg. Mice were then sacrificed, and omental adipose tissue was collected. For the FACS study, adipose tissue was immediately processed for experiments. Standard diet and HFD mice were administered with 250 nmol/L Image-iT Green dye per mouse in the tail vein (intravenous), and 2 h from injection, VATs were harvested in a dark place. Immediately after collection, tissues were snap frozen in optimal cutting temperature compound (Leica Biosystems) for cryosectioning and imaging. All animal experiments followed the guidelines prescribed by and with the approval of the institutional animal ethics committee of the National Institute of Pharmaceutical Education and Research Mohali (project no. IAEC/19/37-ext1).

Seven men and 10 women participated in this study. The study population was categorized into two groups based on BMI and blood glucose level. Study participants with a BMI of 22.2–27.4 kg/m² and fasting serum glucose level ≤4.35 mmol/L were considered as the nonobese nondiabetic group (n = 7), whereas patients with a BMI of 32.2–44 kg/m² and fasting serum glucose level ≥5.4 mmol/L were considered as the obese diabetic group (n = 10). In this study, surgically dissected VAT samples were collected from the patients who underwent abdominal surgery at Dayanand Medical College and Hospital. The study protocol for the use of human blood and tissue samples was approved by the institutional ethics committee of Dayanand Medical College and Hospital (protocol no. DMCH/R&D/2021/64; IEC no. 2021-658). We obtained written informed consent from all participants in this study.

RAW264.7 macrophage, C2C12 myoblast cells were obtained from the National Centre for Cell Science (Pune, India) and cultured in DMEM (cat. no. 11995073; Life Technologies/Gibco) supplemented with 10% FBS (cat. no. 10082147; Life Technologies/Gibco) and 1% penicillin-streptomycin (cat. no. 15140122; Life Technologies/Gibco) at 37°C in a humidified atmosphere with 5% CO₂. We procured the 3T3-L1 preadipocyte (cat. no. CL-173) cell line from ATCC and cultured it in ATCC-formulated DMEM (cat. no. 30-2002) supplemented with 10% BCS (cat. no. 30-2030; ATCC) and 1% penicillin-streptomycin. To differentiate preadipocytes into mature adipocytes, we followed a chemically induced differentiation protocol provided by ATCC. Briefly, preadipocytes were seeded with at a density of 8 × 10⁴ cells in each well of six-well plates having preadipocyte expansion medium (DMEM containing 10% BCS, and 1% penicillin-streptomycin) until it achieved 100% confluency. The media were changed after full confluency and continued with preadipocyte expansion medium for another 48 h. The existing growth media were then removed, and differentiation media (DMEM containing 10% FBS, 1.0 μmol/L dexamethasone, 0.5 mmol/L 3-isobutyl-1-methylxanthine, and 1.0 μg/mL insulin) were added for 48 h. The differentiation media were replaced with adipocyte maintenance media (DMEM containing 10% FBS and 1.0 μg/mL insulin) for 72 h until visible lipid accumulation was observed. We incubated the macrophages and other mentioned cells with a fixed concentration of palmitate (0.75 mmol/L) and exposed them to hypoxic conditions (1% O₂ and 5% CO₂) for various time periods in a Heracell VIOS 160i incubator (Thermo Fisher Scientific). We performed primary cell culture with ATMs isolated from the VAT of lean and obese mice using a BD FACSAria cell sorter (BD Biosciences, Franklin Lakes, NJ). The ATMs were cultured in RPMI medium (cat. no. A1049101; Life Technologies/Gibco) supplemented with 10% FBS and 1% penicillin-streptomycin at 37°C in a humidified atmosphere with 5% CO₂. The ATMs were used in various treatment conditions. All treatments were given in serum- and antibiotics-free media.

For transfection of miR-210-3p mimic/inhibitor and their respective control mimic/inhibitor, Lipofectamine RNAiMAX transfection reagent was used according to the manufacturer’s protocol. Briefly, RAW264.7 macrophages were seeded (0.1 × 10⁶ cells/well) in a 12-well plate in an antibiotic-free complete growth medium for 24 h before transfection. For each well, 50 nmol/L miR-210-3p mimic/control mimic (miRIDIAN miRNA mmu-miR-210-3p mimic/miRIDIAN miRNA hairpin mimic negative control) or 100 nmol/L miR-210-3p inhibitor/control inhibitor (miRIDIAN miRNA mmu-miR-210-3p hairpin inhibitor/miRIDIAN miRNA hairpin inhibitor negative control) and Lipofectamine RNAiMAX reagent were added separately into the Opti-MEM serum-free medium (Thermo Fisher Scientific). Both these solutions were mixed and incubated for 5 min. The transfection mixture was added to the cells containing complete growth medium and incubated for 48 h. After 48 h of transfection, cells were washed, and fresh media was added and used for various treatments.

RAW264.7 macrophages (0.5 × 10⁵ cells/well) were cultured on a transwell cell culture insert (0.4-μm pore size; Corning) transfected with control mimic/inhibitor or miR-210-3p mimic/inhibitor in the absence or presence of a hypoxia (1% O₂) and lipid (palmitate 0.75 mmol/L) (H + L) microenvironment for 16 h. On termination of incubations, cells were washed several times with PBS to remove the residual palmitate, if any, and placed on a 24-well plate containing 3T3-L1 adipocytes (0.5 × 10⁵ cells/well) serum starved overnight in Krebs-Ringer bicarbonate buffer (cat. no. TL-1097; HiMedia Laboratories, Mumbai, India) supplemented with 0.2% BSA and incubated for 6 h. After 6 h, the 3T3-L1 adipocytes were used for glucose uptake assay using a Glucose Uptake Cell-Based Assay Kit (cat. no. 600470; Cayman Chemicals) according to the manufacturer’s instructions. Briefly, insulin (100 nmol/L) was added to the control and treated adipocytes and incubated for 30 min. Fluorescent-labeled glucose analog 2-deoxy-2-[(7-nitro-2,1,3-benzoxadiazol-4-yl)-amino]-d-glucose (2-NBDG) was added to each of the incubations for 10 min before the termination of the experiment. Cells were then lysed, and fluorescence intensity was measured by a microplate reader (BMG LABTECH, Ortenberg, Germany).

A WT SOCS1 3′-untranslated region (UTR) plasmid construct was used as the template for the generation of mutated SOCS1 3′-UTR plasmids by using QuikChange Lightning Multi Site-Directed Mutagenesis Kit according to the manufacturer’s protocol. Primers used to generate the mutated SOCS1 3′-UTR plasmids were designed with the help of the QuikChange Primer Design program available online at https://www.agilent.com/genomics/qcpd. Primer sequences used for mutated SOCS1 3′-UTR plasmid construction are listed in Supplementary Table 3.

RAW264.7 macrophages were cotransfected with 500 ng of WT or mutated SOCS1 3′-UTR plasmid and with either control or miR-210-3p mimic or control or miR-210-3p inhibitor (GE Dharmacon) using Lipofectamine LTX with PLUS Reagent for 48 h in a 24-well plate. Cells were then treated without or with palmitate (0.75 mmol/L) and 1% O₂. On termination of incubations, cells were lysed, and luciferase activity was determined using the Dual Luciferase Reporter Assay System (Promega, Madison, WI) in a GloMax Navigator Microplate Luminometer (Promega) according to the manufacturer’s protocol. Data normalization was achieved by cotransfecting cells with Renilla plasmid (10 ng). Relative luciferase activity was plotted as a ratio of firefly-to-Renilla luciferase activity.

Control and treated RAW264.7 macrophages and primary ATMs isolated from the VAT of human participants or mice were harvested, centrifuged at 350g for 5 min, and washed with PBS. The cell pellets were resuspended in cell staining buffer (PBS containing 0.2% FBS and 0.09% NaNO₃) and blocked with TruStain FcX (Fcγ blocker, mouse anti-CD16/32 antibody; BioLegend) for 15 min at 4°C. Cells were then stained with fluorochrome-labeled primary antibodies against F4/80 (anti-mouse), CD80 (anti-mouse/anti-human), CD206 (anti-mouse), CD64 (anti-human), CD163 (anti-human), and CD11b (anti-human/anti-mouse) (BioLegend) for 1 h on ice. Cells were then washed twice with chilled PBS, resuspended in cell-staining buffer, and analyzed in a flow cytometer (BD Accuri C6 Plus; BD Biosciences, San Jose, CA) using FlowJo version 10.6.1 software.

The miRNA sequencing data set of ATM-derived exosomes (ATM-Exos), obtained from lean and obese mice (n = 7 in each group) was downloaded from the Gene Expression Omnibus (GEO) database (accession no. GSE97652). Quality assessment of miRNA sequencing reads was performed using the FastQC toolkit. Furthermore, Illumina 3′ small RNA sequencing adapter was clipped by the cutadapt tool from the sequencing libraries (23). Trimmed short sequencing reads were subsequently aligned by bowtie to the mouse reference genome (GRCm38) using prebuilt genomic indices downloaded from the Illumina iGenomes collection site. The feature Counts tool from the Subread package was used to calculate per-miRNA count across all the samples (24,25). miRNA with very low read counts (i.e., sum of read counts <5 across all samples) were excluded from further analysis. The DESeq2 package was used to identify differentially expressed miRNAs between obese and lean conditions using R Bioconductor (26). miRNAs with log₂ fold change (log₂FC) of 1.5 and false discovery rate (FDR) <0.1 were considered as significant expression changes between contrasting conditions. Volcano and box plots of normalized read counts for corresponding miRNAs of specific interest were done using the ggplot2 package in R.

All data analyses were performed using GraphPad Prism 8.0 software (GraphPad Software, La Jolla, CA). Data are presented as mean ± SD. Student t test and one-way or two-way ANOVA were used to determine statistical significance, with P < 0.05 considered significant. The rest of the methods, including immunocytochemistry, immunofluorescence, Oil Red O staining, immunoblotting, RNA extraction, and quantitative PCR, ELISA, and FACS are described in the Supplementary Material.

The data are available upon request from the corresponding author. The resources are available in the article and Supplementary Tables 2 and 3.

By investigating the pathophysiological ATenv of the obese diabetic state, we found a marked induction of hypoxia and lipid accumulation in the VAT of obese patients type 2 diabetes (Supplementary Table 1) and an HFD-induced obese diabetic mouse model (Supplementary Fig. 1A) as evident from hypoxia-inducible factor-1α (HIF-1α) levels (Fig. 1A and B and Supplementary Fig. 1C) and lipid staining (Oil Red O and BODIPY F16) (Fig. 1A and B and Supplementary Fig. 1B) compared with their control counterparts. This was further confirmed by the significant enhancement of triglyceride levels in the VAT of obese patients with diabetes and HFD mice (Fig. 1C) along with the lowering of oxygen tension indicated by Image-iT Green hypoxia staining in the VAT of HFD mice (Fig. 1D) compared with their respective controls. These results indicate the dyslipidemic and hypoxic state of the obese diabetic ATenv. We also noticed a significant accumulation of macrophage population in the VAT of obese patients with diabetes and HFD mice as indicated by CD68 and F4/80 staining (Fig. 1E and F and Supplementary Fig. 1D–F), which coincided with the inflammatory landscape as evident from the increased level of inducible nitric oxide synthase (iNOS) over ARG1 (Supplementary Fig. 1G and H) and gene expression of various proinflammatory cytokines (Supplementary Fig. 1I). Since ATMs play a crucial role in obesity-associated chronic inflammation and insulin resistance (27), we were interested in exploring the involvement of specific miRNAs therein. To investigate the specific hypoxia-induced miRNAs in the ATMs of the obese diabetic ATenv, we searched existing literature and considered a GEO database (GSE97652) that comprised an miRNA data set of ATM-derived exosomes (ATM-Exos) obtained from lean and obese mice. We reanalyzed the data set and filtered out the primary miRNAs associated with hypoxic conditions (28,29). Interestingly, the volcano plot derived from the DESeq2 analysis of the GSE97652 data set (Fig. 1G) showed six differentially expressed hypoxia-regulated miRNAs (miR-210-5p, miR-210-3p, miR-27a-5p, miR-27b-5p, miR-101b-5p, and miR-128-1-5p) in ATM-Exos of the obese mice compared with the lean controls. Figure 1H shows read count distribution box plots for these miRNAs in the ATM-Exos of lean and obese mice. Among these miRNAs, miR-210-3p exhibited maximum induction of an approximately fourfold increase in obese mice compared with lean controls. This finding led us to examine the miR-210-3p level in the ATMs of obese patients with diabetes and HFD mice. As expected, we observed a profound increase of miR-210-3p expression in ATMs isolated from the VAT of obese patients with diabetes and 12-week-old HFD mice compared with their respective controls (Fig. 1I and J). Taken together, these results demonstrate that a higher level of miR-210-3p is a key signature in the ATMs of the pathophysiological obese diabetic condition.

Since ATMs are the main source of inflammatory cytokine production in obesity-induced inflamed adipose tissue and cause its dysfunction (27), we explored how increased expression of miR-210-3p in obese ATenv governs ATM inflammation and adipocyte dysfunction. We first set up an in vitro condition where RAW264.7 macrophages were coincubated with 1% O₂ and 0.75 mmol/L palmitate for various time periods to simulate the pathophysiological condition of obese ATenv and analyzed miR-210-3p expression. H + L coexposure significantly increased miR-210-3p and HIF-1α levels at 16 h in RAW264.7 cells (Fig. 2A and Supplementary Fig. 2A and B). Stimulation with hypoxia or lipid alone augmented miR-210-3p expression in macrophages, but the H + L coexposure potentiated miR-210-3p expression in macrophages to a greater extent (Fig. 2B). Moreover, H + L coexposure significantly stimulated gene expression of various proinflammatory cytokines (Fig. 2C) compared with the lone effect of hypoxia or lipid (Supplementary Fig. 2C). Immunofluorescence staining of iNOS and ARG1 further confirmed that H + L coexposure caused iNOS levels to soar concomitant with the reduction of ARG1 levels in RAW264.7 macrophages (Fig. 2D and E and Supplementary Fig. 2D–F). For a better understanding of the inflammatory nature of H + L–treated macrophages, we examined the expression profile of 21 different inflammatory genes and found a significant upregulation of NOS2, TNF-α, CXCL5, IL6, CCL2, IL1β, CD80, CD86, MHCII, and CCR2 gene expressions and downregulation of IL10, IL4, Ym1, CD206, and CD163 gene expressions (Fig. 2F), which depict the M1 macrophage phenotype. Furthermore, a significant enhancement of gene expression of proinflammatory cytokines was observed because of H + L in IL-4–induced M2 macrophages (Supplementary Fig. 2G), suggesting an M2-to-M1 polarization shift. For more clarification, we have performed flow cytometric analysis of M1 (CD80) and M2 (CD206) phenotypic markers. A massive induction of the CD80⁺CD206⁻ population (∼80-fold) was noticed in H + L–treated RAW264.7 cells compared with control (Supplementary Fig. 2H and I). For further validation, we isolated ATMs (CD11b⁺CD64⁺) from VAT of participants without diabetes and obese patients with diabetes and analyzed CD80 (M1) and CD163 (M2) surface markers. Sorted ATMs of obese patients with diabetes displayed an enhanced level of M1 over the M2 marker by expressing high CD80 and low CD163 (Supplementary Fig. 2J and K). Similarly, high CD80 with low CD206 was observed in F4/80⁺ ATM populations of HFD mice compared with standard diet mice (Supplementary Fig. 2L and M). All these results indicate that the obese hypoxic state of the ATenv influences macrophages’ M1 state with a higher level of miR-210-3p expression, suggesting a possible link of miR-210-3p with obesity-induced adipose tissue inflammation.

To investigate the direct effect of miR-210 on macrophage M1 polarization and inflammation, RAW264.7 macrophages were transfected with miR-210-3p mimic or inhibitor in the absence and presence of an H + L microenvironment, followed by the analysis of M1/M2 populations through flow cytometry. Interestingly, miR-210-3p inhibition significantly reduced CD80 (M1) levels along with the induction of CD206 (M2) in H + L–treated macrophages (Fig. 2G and Supplementary Fig. 2N and O). Concomitantly, we noticed suppression of proinflammatory cytokines genes (TNF-α, iNOS, and IL6) expression along with the upregulation of anti-inflammatory cytokines gene (IL4, IL13, and Ym1) expression in miR-210-3p–inhibited H + L–treated macrophages (Fig. 2H). ELISA further confirmed higher levels of IL-6 and TNF-α cytokine secretion from the H + L–stimulated macrophages, and this was significantly prevented in the presence of miR-210-3p inhibitor compared with the control inhibitor (Fig. 2I). Since miR-210-3p–inhibited macrophages were protected from H + L-induced inflammation, iNOS and ARG1 levels remained unaltered under the H + L microenvironment (Fig. 2J). On the contrary, introducing miR-210-3p mimic markedly upregulated the CD80 level in macrophages in the absence of H + L stimulation (Fig. 2K and Supplementary Fig. 2P). Furthermore, induction of proinflammatory TNF-α gene expression along with reduction of anti-inflammatory Ym1 gene expression (Fig. 2L) coincided with the increased levels of iNOS and subdued levels of ARG1 (Fig. 2M) observed in miR-210-3p mimic–transfected cells. Since obese adipose tissue had a profound accumulation of macrophages, which releases a considerable amount of miR-210-3p in the exosomes, we evaluated the impact of miR-210-3p in nearby adipocytes. We transfected miR-210-3p mimic in 3T3-L1 adipocytes and found that miR-210-3p mimic significantly increased NF-κB activation and upregulation of proinflammatory cytokine (iNOS, MCP1) gene expression without any notable alteration of anti-inflammatory cytokine (IL4, IL10) gene expression (Supplementary Fig. 2Q and R). These results revealed a direct role of miR-210-3p in ATM and adipocyte inflammation, favoring a state of inflamed adipose tissue that facilitates macrophage polarity switching toward the M1 proinflammatory phenotype.

To investigate the underlying mechanism of miR-210-3p involvement in obesity-induced ATM inflammation, we searched for the putative targets of miR-210-3p that can potentially regulate inflammatory pathways. We found that SOCS1 could be a potential target of miR-210-3p as indicated by the miRWalk miRNA target prediction database (Supplementary Fig. 3A) and that may be involved in the impairment of NF-κB signaling. The nucleotide sequence of SOCS1 3′-UTR is conserved in the transcript sequence of humans and mice that critically offers the binding site of miR-210-3p (Fig. 3A). We used the RNAhybrid web server to predict the minimum free energy for the interaction of various SOCS1 mRNA transcripts with miR-210-3p. The miR-210-3p sequence complementary to the 3′-UTR of SOCS1 forms a hybrid sequence having a minimum free energy of −32.2 kcal/mol (mouse SOCS1 transcript) and −31.7 kcal/mol (human SOCS1 transcript) (Supplementary Fig. 3B). To validate the binding of miR-210-3p with the 3′-UTR of SOCS1, we performed a luciferase reporter assay with WT and mutant SOCS1 3′-UTR. H + L coexposure significantly inhibited luciferase activity in WT SOCS1 3′-UTR–transfected cells; however, such an effect was compromised in cells transfected with mutated SOCS1 3′-UTRs (Fig. 3B), indicating a direct interaction of miR-210-3p with the SOCS1 3′-UTR in the pathophysiological condition. We then examined SOCS1 expression and observed a significant reduction of its level in the stromal vascular fraction (SVF) of adipose tissue of obese patients with diabetes compared with lean participants without diabetes (Fig. 3C and D). Among the members of the SOCS family, only SOCS1 acts as a ubiquitin ligase that is capable of interacting with NF-κB p65 through its SOCS box domain, which leads to polyubiquitination and proteasomal degradation of NF-κB p65, resulting in termination of NF-κB–inducible gene expression (30). We therefore analyzed the SOCS1-mediated NF-κB activation in pathophysiological conditions. Western blot analysis showed a significant reduction of SOCS1 expression along with the enhanced levels of phosphorylated NF-κB (pNF-κB) in the adipose tissue SVF of obese patients with diabetes compared with lean participants without diabetes (Fig. 3E). This was also evident from H + L–treated RAW264.7 cells exhibiting a time-dependent enhancement of NF-κB activation (Supplementary Fig. 3C) and its increased nuclear translocation (Supplementary Fig. 3D and E). To examine the role of macrophage-specific SOCS1 in macrophage inflammation, we silenced SOCS1 in RAW264.7 cells using siRNA (Supplementary Fig. 3F and G) and incubated them in the absence or presence of H + L stimulation. Silencing of SOCS1 significantly increased NF-κB activation, whereas costimulation of H + L in SOCS1 silenced cells and further aggravated NF-κB activation (Supplementary Fig. 3H). Furthermore, to exclude the possibility of other targets of miR-210-3p that could contribute to the suppression of NF-κB signaling, we analyzed RelA (p65 subunit of NF-κB)-interacting partners using the EMBL INACT server (https://www.ebi.ac.uk/intact/search). The RelA-interacting partners were searched against the 16,853 mmu-miR-210-3p target genes identified through the miRWALK database search (Supplementary Fig. 3I and J), and BRD4, NF-κBIB, SOCS1, CUL2, COMMD1, XPO1, and MYOCD were found to be predicted targets of mmu-miR-210-3p, of which SOCS1, NF-κBIB, CUL2, COMMD1, and MYOCD are negative regulators of NF-κB. To find out the most efficient target of miR-210-3p that can potentially inhibit NF-κB under lipid-rich hypoxic conditions, we evaluated expression of these selected candidate genes in control mimic– and miR-210-3p mimic–transfected macrophages and observed a profound decline of SOCS1 gene expression (∼70%) compared with other candidate genes in miR-210-3p mimic–transfected cells (Supplementary Fig. 3K).

To decipher the direct role of miR-210-3p on SOCS1-mediated NF-κB activation in the ATMs, miR-210-3p mimic–transfected cells were examined and found to significantly downregulate SOCS1 gene and protein expression concomitant with the upregulation of NF-κB activation (Supplementary Fig. 3L and M). Also, we analyzed SOCS1 and NF-κB activation in control and miR-210-3p inhibitor–transfected RAW264.7 macrophages cotreated with H + L. Exposure of H + L strikingly reduced SOCS1 expression, which coincided with the increased level of pNF-κB and its nuclear translocation; however, such attributes were significantly attenuated in miR-210-3p inhibitor–transfected cells (Fig. 3F–H). Moreover, to assess whether the SOCS1 inhibitory effect on NF-κB activation is mediated through the proteasomal degradation of pNF-κB, we transfected macrophages with control or miR-210-3p inhibitor in the absence or presence of the proteasome inhibitor MG-132 and treated with H + L. Our findings demonstrated that MG-132 treatment notably protects NF-κB p65 phosphorylation-dependent proteasomal degradation in macrophages transfected with miR-210-3p inhibitor (Fig. 3I). Moreover, a higher level of pNF-κB and its nuclear localization was fanned when miR-210-3p mimic transfected cells were treated with MG-132 (Supplementary Fig. 3N). Since the obesity-induced inflammatory milieu in adipose tissue is known to be associated with the impairment of insulin sensitivity (31), we cocultured 3T3-L1 adipocytes with macrophages transfected with a control inhibitor or miR-210-3p inhibitor in the absence or presence of H + L stimulation. Interestingly, attenuation of adipocytes’ insulin sensitivity in response to its coculture with H + L–incubated macrophages was markedly prevented when transfected with miR-210-3p inhibitor (Fig. 3J). Since greater enrichment of miR-210-3p was evident from ATM-Exos, we analyzed the paracrine and endocrine action of miR-210-3p on the insulin sensitivity of 3T3-L1 adipocytes and C2C12 myotubes. Interestingly, we found that delivery of miR-210-3p mimic significantly attenuated insulin sensitivity in 3T3-L1 adipocytes and C2C12 myotubes as evident from the reduction of 2-NBDG uptake study (Supplementary Fig. 3O and P). All these results suggest that miR-210-3p plays a key role in obesity-induced ATM inflammation by targeting the SOCS1/NF-κB pathway and could also influence adipocytes’ insulin sensitivity.

To inspect the therapeutic potential of miR-210-3p inhibitor in the rescue of obesity-induced adipose tissue inflammation and insulin resistance, we administered control LNA or anti–miR-210-3p LNA directly to the VAT of HFD mice (Fig. 4A) and validated their efficacy by analyzing miR-210-3p expression (Supplementary Fig. 4A). Also, we estimated the body weight and food intake for the entire 7 days of postsurgery (Supplementary Fig. 4B and C) to examine the postsurgery recovery of mice. On day 7, adipose tissue was harvested for immunofluorescence analysis of F4/80, CD163, iNOS, and ARG1. A profound increase in the anti-inflammatory CD163⁺F4/80⁺ macrophage population was observed in the HFD adipose tissue injected with anti–miR-210-3p LNA (Fig. 4B). Similarly, a reduced level of iNOS and an enhanced level of ARG1 were also seen in the anti–miR-210-3p LNA adipose tissue of HFD mice (Fig. 4C). Moreover, we collected the ATMs from these mice on day 7 and analyzed for ATM polarization and inflammation. HFD mice administered anti–miR-210-3p LNA exhibited an abundance of the M2 (F4/80⁺CD206⁺) over the M1 (F4/80⁺CD80⁺) ATM population (Fig. 4D and Supplementary Fig. 4D), with a notable reduction of proinflammatory (iNOS, IL6) over anti-inflammatory (IL13, Ym1) cytokine gene expression (Fig. 4E) compared with control LNA–injected HFD mice. To exclude the contributions of other cell types in adipose tissue in response to anti–miR-210-3p LNA treatment, we checked miR-210-3p and SOCS1 expression in the adipocytes isolated from mice injected with control LNA and anti–miR-210-3p LNA. We found that miR-210-3p expression was reduced in isolated adipocytes of mice administered anti–miR-210-3p LNA, but not significantly compared with controls (Supplementary Fig. 4E). Also, adipocytes’ SOCS1 expression and inflammatory cytokine levels were not significantly altered in response to anti–miR-210-3p LNA administration (Supplementary Fig. 4F and G). These results suggest that adipocytes’ contribution has not been substantial for altering the adipose tissue inflammatory state upon anti–miR-210-3p LNA treatment; rather, it depicts the exclusive role of miR-210-3p in ATM inflammation, and, therefore, ATM-specific inhibition of miR-210-3p would be beneficial to improve the inflammatory setting of obese adipose tissue.

As we found that miR-210-3p promotes NF-κB activation–dependent inflammation in RAW264.7 macrophages by targeting the SOCS1/NF-κB pathway (Fig. 3), we examined the level of SOCS1 expression and NF-κB activation in the adipose tissue of standard diet and HFD mice (Supplementary Fig. 4H) treated with control LNA or anti-miR-210-3p LNA. Immunofluorescence analysis showed a profound induction of SOCS1 in the VAT of HFD mice treated with anti–miR-210-3p LNA (Fig. 4F). Upregulation of SOCS1 expression (Fig. 4G and H) coincided with the subdued levels of pNF-κB p65 (Fig. 4H) in the ATMs of HFD mice administered anti–miR-210-3p LNA. However, NF-κB p65 activation and SOCS1 expression were not altered significantly in standard diet mice treated with anti-miR-210-3p LNA or control LNA (Supplementary Fig. 4I). Moreover, by investigating the insulin sensitivity of HFD mice treated with control LNA or anti–miR-210-3p LNA, we found that anti–miR-210-3p LNA delivery significantly restored HFD mice from diet-induced insulin resistance as indicated by glucose tolerance test (Fig. 4I), insulin tolerance test (Fig. 4J), and HOMA of insulin resistance (Fig. 4K). These results suggest that the application of anti-miR-210-3p LNA could be beneficial in rescuing obesity-induced inflammation and insulin resistance.

Alteration of the ATenv is an important feature in obesity characterized by adipose tissue expansion along with increased infiltration and activation of immune cells, particularly macrophages (32). Compelling evidence has shown that the obesity-associated hypertrophic expansion of visceral adipocytes leads to the deprivation of oxygen, resulting in adipose tissue hypoxia (33,34). Moreover, adipose tissue hypoxia positively correlates with the increased rate of necrotic adipocytes (crown-like structures), which by secreting various chemoattractants and related proteins, strongly influences macrophage recruitment and polarization toward a proinflammatory state in hypertrophic adipose tissue (35,36).

Studies from past decades focused mainly on the role of either increased lipid level or deprived oxygen tension on adipose tissue dysfunction and provided evidence that both these phenomena cause the onset of adipose tissue inflammation that leads to insulin resistance and type 2 diabetes (37–39). In reality, both lipid burden and hypoxia coincide in the obese condition because of hypertrophied adipocytes and poor vasculature, creating a unique pathophysiological ATenv. We also noticed the coexistence of lipid burden and hypoxia in the VAT of obese patients with diabetes and mouse models. Therefore, we weighted both lipid and hypoxic insults on ATMs to understand their synergic effect and the underlying mechanism of adipose tissue inflammation and insulin resistance. Increased accumulation of ATMs in the VAT of patients with diabetes or mice exhibited proinflammatory features, which coincide with the results of an in vitro model of RAW264.7 macrophages costimulated with H + L. These findings distinctly depict that obesity-associated ATenv strongly correlates with ATM inflammation.

Several studies have revealed that noncoding RNAs, particularly miRNAs, play a key role in the regulation of inflammatory responses (40). To decipher the involvement of specific miRNAs on macrophage inflammation in the obese ATenv, we first explored a GEO data set of miRNA profiles in ATM-Exos isolated from the VAT of lean and obese mice. Reanalysis of GEO data set GSE97652 in the context of hypoxia revealed six differentially expressed miRNAs: miR-210-5p, miR-210-3p, miR-27a-5p, miR-27b-5p, miR-101b-5p, and miR-128-1-5p. Of these, miR-210-3p showed a profound increase in the ATM-Exos of obese mice compared with lean mice. This is also evident in the ATMs of obese patients with diabetes and HFD mice in our study. This observation led us to examine the role of miR-210-3p in ATM inflammation and insulin resistance.

In obesity, various immune cell types are accumulated in adipose tissue, and among them, macrophages are the most abundant in nature and constitute up to 40% of stromal vascular fraction (35). Obesity-associated inflamed adipose tissue is characterized by an increased ratio of M1 proinflammatory-to-M2 anti-inflammatory phenotypic state of ATMs, which is considered a major source of various proinflammatory mediators responsible for adipocyte dysfunction and insulin resistance (41). Investigating the involvement of miR-210-3p on macrophage polarization and inflammation, we used miR-210-3p mimic– and inhibitor–transfected cells and treated without or with an H + L environment. While the miR-210-3p mimic notably enhanced the M1 polarization with the secretion of proinflammatory cytokines in the absence of any treatment, the miR-210-3p inhibitor rescued H + L–mediated M1 polarization and inflammation by favoring the M2 polarization state with increased anti-inflammatory cytokines. To investigate the molecular target of miR-210-3p in the regulation of macrophage inflammation, different miRNA analysis tools were used. In searching for the putative target of miR-210-3p, we found that SOCS1 is one of the targets of miR-210-3p. The SCOS1 3′-UTR luciferase assay confirmed that miR-210-3p directly binds at 3′-UTR of SCOS1. Since SOCS1 is known to regulate the NF-κB inflammatory pathway by proteasomal degradation of activated NF-κB p65 (30,42,43), we compared SOCS1 expression and NF-κB activation in the ATMs of patients with diabetes and HFD mice. The pathophysiological obese ATenv stimulates miR-210-3p expression in ATMs and promotes NF-κB inflammatory signaling by downregulating SOCS1 expression. However, it will be interesting to see the impact of miR-210-3p on other cell types of adipose tissue. A plethora of studies highlighted that inflamed ATenv negatively correlates with insulin sensitivity (44). H + L–treated RAW264.7 macrophages, when cocultured with adipocytes, significantly impaired insulin-stimulated glucose uptake in adipocytes, but such an effect was compromised when macrophages were transfected with miR-210-3p inhibitor. On the basis of all these observations, we evaluated the therapeutic potential of miR-210-3p inhibitor on adipose tissue inflammation and insulin resistance in vivo. Direct administration of anti–miR-210-3p LNA into the VAT of HFD-induced diabetic mice exhibited a lower level of F4/80⁺-sorted ATMs, which coincided with a lower level of CD80 and a higher level of CD206⁺ macrophage phenotypic markers. Moreover, anti–miR-210-3p LNA delivery remarkably increased SOCS1 levels in HFD mice, which coincided with the attenuation of NF-κB activation compared with control LNA–treated HFD mice. As suppression of adipose tissue inflammation is known to be associated with improvement of insulin sensitivity (45), we also found an improvement of insulin sensitivity associated with the impairment of ATM inflammation in HFD mice when anti–miR-210-3p LNA was administered in the adipose tissue. However, an in-depth study will be required to understand the role of ATM-derived miR-210-3p in impairing adipocytes’ insulin sensitivity.

In conclusion, our study revealed that miR-210-3p plays a crucial role in ATM polarization and inflammation under the influence of a lipid-enriched hypoxic microenvironment of obese adipose tissue. However, inclusion of a larger sample size of patients will strengthen the results. Moreover, the use of macrophage-specific transgenic in vivo models will provide firm and compelling evidence about the vital role of miR-210-3p and SOCS1. Altogether, these shreds of evidence will authenticate and fortify the conclusive statement. Hence, targeted inhibition of miR-210-3p could be beneficial for the management of obesity-induced adipose tissue inflammation and insulin resistance in patients with type 2 diabetes.

Acknowledgments. The authors acknowledge the help of Dr. Himangshu Kumar (Department of Biological Sciences, Indian Institute of Science Education and Research) for providing the mouse SOCS1 3′-UTR construct. The authors also thank the National Institute of Pharmaceutical Education and Research, Sahibzada Ajit Singh Nagar, for the animal housing facility to perform mouse experiments and Indian Institute of Technology Ropar for providing instrumentation facilities.

Funding. This study was supported by a Science and Engineering Research Board, Government of India, Early Career Research grant (ECR/2017/000892) and the Department of Biotechnology, Ministry of Science and Technology, Government of India, twinning project (BT/PR24700/NER/95/819/2017). D.Pat., S.R., and L.A. acknowledge Indian Institute of Technology Ropar and Ministry of Human Resource Development for research fellowships.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. D.Pat. and D.Pal conceived the idea, planned the experiments, and wrote the original draft of the manuscript. D.Pat., S.R., L.A., S.W.K., U.D., D.B., A.S., and D.Pal contributed to the investigation and validation. D.Pat., S.R., L.A., S.W.K., U.D., and D.Pal contributed to the formal analysis. D.Pat., S.D., K.T., A.K., and D.Pal reviewed and edited the manuscript. S.S., S.D., K.T., A.K., and D.Pal contributed resources. D.Pal acquired funding and provided supervision. All authors edited and reviewed the manuscript. D.Pal is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.