Adipocyte-Specific Hnrnpa1 Knockout Aggravates Obesity-Induced Metabolic Dysfunction via Upregulation of CCL2

In past decades, the prevalence of obesity worldwide has risen dramatically (1). Obesity-related metabolic disorders (insulin resistance, type 2 diabetes, and cardiovascular disease) represent a severe health issue. The critical characteristic of obesity is chronic low-grade inflammation and metabolic disorders (2). Adipose tissue is now recognized as a critical metabolic organ that regulates whole-body energy homeostasis via the regulation of energy storage and dissipation (3) and secretion of adipokines (4). During obesity, increased proinflammatory adipokines activate the immune system and drive it toward a proinflammatory phenotype (5). Among the adipokines, immune-modulating adipokines, such as interleukin-6 (IL-6), IL-8, C-X-C motif chemokine ligand 5, and C-C motif chemokine ligand 2 (CCL2, also known as MCP-1), play important roles in adipose tissue macrophage (ATM) recruitment and activation (6), resulting in a feedback loop of adipose tissue inflammation. Recently, many of these inflammatory signals generated by adipose tissue were reported to block insulin sensitivity that finally resulted in systemic metabolic dysfunction (7,8). Although extensive research and several mechanisms have been proposed, the causal processes promoting adipose tissue inflammation in obesity are continuously being expanded and have always been a hot spot for research (9).

It is well established that many different types of RNA-binding proteins exist in eukaryotic cells. RNA-binding proteins could combine with specific sequences or secondary structures of mRNA transcript variants to regulate their synthesis, expression, and function (10). A growing amount of research indicates that RNA processing prepares the transcript so that it can function in adipogenesis (11), adipocyte lipolysis (12), adipocyte thermogenesis (13), etc. Heterogeneous nuclear ribonucleoprotein A1 (HNRNPA1), a ubiquitously expressed member of the HNRNP family of RNA-binding proteins, contains highly conserved RNA recognition domains and a nuclear targeting sequence (14). It takes part in multiple cell functions, including regulation of transcription factors, alternative splicing, mRNA stability, mRNA nuclear export, and miRNA processing (15). Emerging evidence indicates that HNRNPA1 plays important roles in regulating metabolic balance in the liver and muscle through multiple target genes (16–18). However, it remains unclear whether HNRNPA1 in adipose tissue would exert any effect on metabolic disorders. In this study, we comprehensively investigated the role of HNRNPA1 in adipose tissue inflammation and obesity-related metabolic disorders and deciphered its mechanisms via regulating mRNA stability of CCL2, which could provide therapeutic strategies against obesity and metabolic diseases.

Hnrnpa1 flox/flox (fl/fl) C57BL/6J mice were purchased from GemPharmatech Co. Ltd., in which exons 2–11 of the Hnrnpa1 allele were flanked by loxP site. Adipoq-Cre C57BL/6J mice were purchased from The Jackson Laboratory. Adipocyte-specific Hnrnpa1 knockout (Hnrnpa1 ako) mice were generated by intercrossing Hnrnpa1 fl/fl mice with heterozygous adipoq-Cre mice. Hnrnpa1 ako mice were genotyped by PCR using DNA isolated from tails. The genotyping primers are shown in Supplementary Table 1. Detailed procedures for mouse modeling, metabolic phenotype, and serum analysis can be found in the Supplementary Methods.

For tissue staining, dissected adipose and liver tissues were fixed in 4% paraformaldehyde (G1101; Servicebio) and embedded by paraffin. Tissue sections were stained with hematoxylin-eosin (H-E) (G1076; Servicebio) or antibodies, including anti-F4/80 and anti-CD11c, followed by species-specific secondary antibodies. The images were acquired by a microscope (Leica) using a 20× objective.

The mouse stromal vascular fraction (SVF) was isolated from the included inguinal white adipose tissue (iWAT) depot of 4–6-week-old male mice. Detailed steps for cell isolation, culture, and differentiation are presented in the Supplementary Material, and mature adipocytes were used for further studies. Adipoq-Cre adenovirus (Cre AdV) and negative control were provided by Genechem Co. Ltd., which were used to knockdown Hnrnpa1 in primary adipocytes from Hnrnpa1 fl/fl mice.

To assess the insulin-induced AKT pathway, primary adipocytes were stimulated with insulin (10 nmol/L for 10 min). The glucose uptake ability of adipocytes was measured using the Glucose Uptake-Glo Assay Kit (J1342; Promega). For lipolysis assay, adipocytes were stimulated with tumor necrosis factor-α (TNF-α) (10 ng/mL). The Mouse CCL2/JE/MCP-1 Quantikine ELISA Kit (MJE00B; R&D Systems) was used to analyze the level of CCL2 according to the instructions provided in the manual. To assess the mRNA stability of CCL2 under HNRNPA1 knockdown, differentiated 3T3-L1 cells were treated with 5 μg/mL actinomycin D (HY-17559; MedChemExpress) and harvested at indicated time points.

pLKO.1-puro empty vector plasmid, HNRNPA1 shRNA plasmids, pSLenti-CMV-7Myc-EGFP-PGK-Puro-WPRE empty vector plasmid, pSLenti-CMV-7Myc-Slc2a4-linker-EGFP-PGK-Puro-WPRE plasmid, psPAX2, and pMD2.G were used for the lentiviral package. The procedure of lentivirus package, concentration, and shRNA sequences are shown in the Supplementary Methods.

There were 236 individuals included for adipose HNRNPA1 mRNA expression analysis. These individuals underwent either elective abdominal surgery for cholecystectomy or weight reduction bariatric surgery at Shanghai Jiao Tong University School of Medicine Affiliated Sixth People’s Hospital between July 2019 and August 2020. Detailed standards about recruitment, exclusion, clinical data, and sample acquisition were described previously (19). The human study was approved by the ethics committee of Shanghai Jiao Tong University School of Medicine Affiliated Sixth People’s Hospital.

According to the manufacturer’s instructions, the RNA immunoprecipitation (RIP) assay was performed using Magna RIP RNA-Binding Protein Immunoprecipitation Kit (17-700; Millipore). Detailed steps are provided in the Supplementary Material.

Tissue and cell samples were lysed with radioimmunoprecipitation assay lysis buffer containing EDTA-free protease and phosphatase inhibitor cocktail. Detailed steps are provided in the Supplementary Material.

Total RNA was extracted from cells and tissues using TRIzol reagent (15596018; Invitrogen). A total of 500 ng RNA was reversed to cDNA using the PrimeScript RT Reagent Kit with gDNA Eraser (RR047B; Takara Bio) according to the manufacturer’s instructions. Quantitative RT-PCR (RT-qPCR) was performed using SYBR qPCR Master Mix (R222; Vazyme) on a Bio-Rad C1000 thermal cycler. Experiments were repeated three times. Primers for RT-qPCR are listed in Supplementary Table 2.

For RNA sequencing (RNA-seq), total RNA samples of primary adipocytes and iWAT from Hnrnpa1 fl/fl and ako mice were extracted using TRIzol reagent. The specific sequencing procedure is shown in the Supplementary Methods.

All data are presented as mean ± SEM, median (interquartile range), and n (%). Statistical analysis was performed using unpaired two-tailed Student t test or two-way ANOVA with Šidák multiple comparisons test. Correlational analyses were conducted by nonparametric Spearman correlation. ImageJ version 1.53 software was used for Western blot image densitometry analysis.

The data sets generated and/or analyzed during the current study are available from the corresponding authors upon reasonable request.

We analyzed HNRNPA1 expression in human metabolic tissues and found that its expression was much more abundant in WAT than liver, muscle, pancreas, and hypothalamus based on the Genotype-Tissue Expression (GTEx) database (20) (Supplementary Fig. 1). We therefore measured HNRNPA1 expression in human subcutaneous adipose tissue (SAT) and visceral adipose tissue (VAT) from 10 nonobese and 10 obese individuals. Representative expression and quantitative analysis showed an obvious reduction in HNRNPA1 protein levels among obese humans compared with the control subjects (Fig. 1A and B). Moreover, weight loss caused by bariatric surgery upregulated mRNA levels of HNRNPA1 in human SAT according to the data set from Poitou et al. (21) (Fig. 1F).

In addition, high-fat diet (HFD) feeding resulted in decreased mRNA and protein expression of HNRNPA1 in mouse iWAT and epididymal WAT (eWAT), while this effect was not observed in brown adipose tissue (BAT) (Fig. 1D–F). These data suggest that obesity status could downregulate HNRNPA1 expression of adipose tissue.

The specific role of HNRNPA1 in adipose tissue was investigated by crossing Hnrnpa1 fl/fl mice with adipoq-Cre mice (Fig. 2A). As expected, HNRNPA1 was significantly reduced in adipose tissues, while there was no change in mRNA and protein levels in the hypothalamus, liver, and muscle (Fig. 2B–D), verifying the specificity of the Hnrnpa1 ako mice. After isolating SVF and mature adipocyte fraction, Western blot experiments provided further evidence for adipocyte-specific knockout of Hnrnpa1 (Supplementary Fig. 2). We first observed the phenotypes in the setting of normal chow diet (NCD) feeding for 16 weeks. The body weight, tissue weight, and food intake of Hnrnpa1 ako mice were comparable with the control group (Supplementary Fig. 3A–C). There was also no difference in glucose tolerance and insulin sensitivity (Fig. 2E and F). However, pyruvate tolerance test analysis showed that blood glucose levels were mildly higher in Hnrnpa1 ako mice than in Hnrnpa1 fl/fl, indicating impaired pyruvate tolerance (Fig. 2G).

To further investigate the functional significance of adipocyte-specific Hnrnpa1 deletion, mice were fed a 60% HFD for 20 weeks, starting at 8 weeks of age. Both Hnrnpa1 fl/fl and Hnrnpa1 ako mice showed a similar age-dependent increase in body weight (Fig. 3A). Besides, the food intake (Fig. 3B), fat mass, and lean mass showed no difference (Fig. 3C). Remarkably, Hnrnpa1 ako mice displayed significantly impaired glucose intolerance and insulin sensitivity (Fig. 3D and E). For a better understanding of the role of HNRNPA1 in insulin action, we harvested adipose, liver, and muscle tissue after 10 min of intraperitoneal insulin injection. Consistent with the insulin tolerance studies, insulin-stimulated AKT phosphorylation was significantly lower in iWAT, eWAT, and liver but not in muscle tissues of Hnrnpa1 ako versus Hnrnpa1 fl/fl mice (Fig. 3F). The ability of insulin to suppress lipolysis also indicated insulin sensitivity of adipose tissue, and we found that hormone-sensitive lipase phosphorylation inhibition and serum free fatty acid suppression were both impaired in Hnrnpa1 ako mice (Fig. 3G and H).

In addition, we examined the impact of HNRNPA1 on liver metabolism under HFD feeding. Hnrnpa1 ako mice showed significantly higher levels of glycemia than control mice during pyruvate tolerance testing (Fig. 3I), indicating overactivated hepatic gluconeogenesis. We analyzed mouse liver tissues for the key gluconeogenic gene expression and found that Pepck and G6pc were expressed higher in the liver of Hnrnpa1 ako mice (Fig. 3J). The liver showed a light gray color, and H-E staining demonstrated increased lipid accumulation in Hnrnpa1 ako mice (Fig. 3K and L), indicating that adipocyte HNRNPA1 deletion promoted hepatic steatosis. Consistently, hepatic triglyceride levels were significantly increased (Fig. 3M). Overall, the data demonstrated that adipocyte HNRNPA1 plays important roles in metabolic disorders.

Even though no obvious changes in global metabolic phenotypes were observed in the NCD-fed mice between genotypes, we found that Hnrnpa1 ako mice showed elevated proinflammatory gene expression and a trend toward decreased expression of anti-inflammatory genes in WAT (Fig. 4A and B), while no differences were observed in adipogenesis, lipolysis, and glucose metabolism genes in WAT between Hnrnpa1 fl/fl and Hnrnpa1 ako mice (Supplementary Fig. 3D and E). Since adipose tissue inflammation is known to promote lipolysis (22), we also detected a significant increase in lipolysis of Hnrnpa1 ako mice whether at room temperature or under cold stimulation (Supplementary Fig. 3F and G).

Moreover, in obese mice, we observed increased proinflammatory and decreased anti-inflammatory gene expression in WAT of Hnrnpa1 ako mice (Fig. 4C and D). Previous studies reported that ATMs mainly accumulate in obese WAT, forming crown-like structures and leading to the development of chronic inflammation (23). Accordingly, we found a higher number of crown-like structures measured by F4/80 immunohistochemical staining in iWAT and eWAT from Hnrnpa1 ako mice than Hnrnpa1 fl/fl mice (Fig. 4E). CD11c is a classical marker of proinflammatory macrophages (24). Consistently, the intensity of CD11c staining in iWAT and eWAT was elevated in Hnrnpa1 ako mice (Fig. 4F and G and Supplementary Fig. 4). As accumulative macrophage infiltration leads to adipocyte death and eventual fibrosis (25), Hnnrpa1 ako mice showed increased expression of fibrotic genes in WAT (Supplementary Fig. 5), accelerating pathological remodeling of adipose tissue. Furthermore, Hnrnpa1 ako mice showed high serum levels of inflammatory factors, including IL-1α, IL-9, IL-12p40, interferon-γ (IFN-γ), CCL2, TNF-α, MIP-1β, and granulocyte colony-stimulating factor (Fig. 4H). Additionally, we quantified immune cells using the FACS technique. SVFs were isolated from obese mice after 12 weeks of HFD feeding. The F4/80⁺CD11b⁺ cell numbers in iWAT and eWAT, as measured by FACS, were significantly higher in Hnrnpa1 ako mice (Supplementary Fig. 6A and B). Taken together, these data suggest that adipocyte-specific Hnrnpa1 knockout leads to increased adipose tissue inflammation.

A previous study reported that silencing of HNRNPA1 impaired insulin sensitivity in C2C12 myotubes (26). Thus, we explored whether the deletion of HNRNPA1 could affect adipocyte function. The SVF from the iWAT of Hnrnpa1 fl/fl mice was harvested, and the differentiated adipocytes were infected with control or AdV. As expected, Cre AdV infection caused the knockdown of HNRNPA1 in primary adipocytes (Fig. 5A). HNRNPA1 knockdown did not affect the expression of genes involved in differentiation (Adipoq, Fabp4), lipolysis (Lipe, Pnpla2), and glucose metabolism (Slc2a1, Slc2a4) (Fig. 5B). Notably, HNRNPA1 knockdown impaired insulin-stimulated AKT phosphorylation, indicating impaired insulin signaling pathway (Fig. 5C). As insulin-AKT signaling stimulates glucose uptake via GLUT4 plasma membrane translocation (27), the immunofluorescence analysis showed that insulin-stimulated GLUT4 plasma membrane translocation was obviously decreased in HNRNPA1 knockdown adipocytes as visualized using confocal microscopy (Fig. 5D). The impaired ability of glucose uptake was also determined (Fig. 5E). Additionally, HNRNPA1 knockdown induced adipocyte lipolysis (Fig. 5F and G). These findings were consistent with the results in Hnrnpa1 ako mice under HFD feeding, which further demonstrated that HNRNPA1 deficiency destroyed adipocyte function.

To dissect the molecular mechanism of HNRNPA1 in adipose tissue, we examined the transcriptional profiles of iWAT from Hnrnpa1 fl/fl and Hnrnpa1 ako mice by performing RNA-seq analysis. Principal component analysis showed that iWAT from Hnrnpa1 ako mice was separated from the control group (Fig. 6A). The pathway enrichment data showed that differential genes mainly enriched in cytokine-cytokine receptor interaction and chemokine signaling pathways (Fig. 6B). From the volcano plot, chemokines (including Ccl2, Ccl3, Ccl8, Ccl9, Cxcl10, and Cxcl13) were increased in the iWAT of Hnrnpa1 ako mice (Fig. 6C). Then, we verified the expression of chemokines in primary adipocytes and found that there was a significant difference in Ccl2 expression between control and HNRNPA1 knockdown adipocytes (Fig. 6D). Besides, shRNA was used to knockdown Hnrnpa1 mRNA in 3T3-L1 cells (Supplementary Fig. 7A and B), and decreased Ccl2 expression was also identified in Hnrnpa1 knockdown cells (Fig. 6E). After knocking down Hnrnpa1 in primary adipocytes, the secretion of CCL2 in the cell culture supernatant was also reduced (Supplementary Fig. 8). From these results, we conclude that HNRNPA1 deletion resulted in increased expression and secretion of CCL2.

HNRNPA1 is an RNA-binding protein that has been reported to be critical for mRNA processing (15). Therefore, to decipher the exact regulatory mechanism of HNRNPA1, RIP assay was performed, and the results suggested that HNRNPA1 could interact with Ccl2 mRNA (Fig. 6F and G). Importantly, HNRNPA1 knockdown increased the mRNA stability of Ccl2 in 3T3-L1 cells treated with actinomycin D (a specific RNA synthesis inhibitor) (Fig. 6H), which indicated that HNRNPA1 functioned via binding to mRNA of Ccl2 to inhibit its mRNA stability. Furthermore, we performed RNA-seq of mouse primary adipocytes derived from iWAT of Hnrnpa1 fl/fl mice infected with control or Cre AdV, separately. The differentially expressed genes are shown in Supplementary Table 3, which included 66 upregulated genes and 122 downregulated genes (absolute log₂ |fold change| ≥0.58, P < 0.05). Although other differentially expressed genes were identified, CCL2 was still one of the top differentially downregulated genes.

Since CCL2 functioned mainly through CCR2 (28), several studies explored the role of the CCL2-CCR2 pathway in metabolic complications based on CCL2-neutralized antibody or antagonist (29,30). Thus, we aimed to detect whether CCL2-CCR2 signaling antagonist could rescue the metabolic disorders of Hnrnpa1 ako mice under HFD feeding. INCB3344 was developed as a selective small-molecule antagonist of CCR2, and several studies have identified its protective roles in inflammatory disorders (31,32). We treated 16-week DIO Hnrnpa1 fl/fl and Hnrnpa1 ako mice with daily subcutaneous injection of INCB3344 (30 mg/kg) for 14 days (Fig. 7A). At the end of the treatment period, body and tissue weight showed no significant difference between Hnrnpa1 fl/fl and Hnrnpa1 ako mice with or without INCB3344 injection (Supplementary Fig. 9A and B). Remarkably, the numbers of infiltrated ATMs decreased in iWAT and eWAT when treated with INCB3344 (Fig. 7B). Consistently, the reduction of CD11c intensity in iWAT and eWAT was also determined in INCB3344-injected mice (Fig. 7C and D and Supplementary Fig. 10), demonstrating that INCB3344 could abolish the proinflammatory effect of HNRNPA1 deficiency. We assessed the overall state of glucose metabolism, and the results showed that INCB3344 rescued the glucose metabolic abnormalities and had a tendency to improve insulin sensitivity in adipocyte-specific knockout of HNRNPA1 (Fig. 7E and F). Collectively, our data further suggest that the obesity-related metabolic disorders of Hnrnpa1 ako mice are mainly mediated by CCL2.

To further explore whether HNRNPA1 is correlated with human clinical characteristics, we analyzed its expression in paired SAT and VAT of individuals undergoing either elective abdominal surgery for cholecystectomy or weight reduction bariatric surgery. The clinical characteristics are shown in Supplementary Table 4. We observed a negative correlation of HNRNPA1 expression with BMI, fat percentage, and subcutaneous fat area in SAT (Fig. 8A–C). Notably, LEP gene expression was significantly correlated with HNRNPA1 expression (Fig. 8D). Among individuals with detectable HNRNPA1 expression, 33 were followed up 1 year after bariatric surgery. The percentage of total weight loss (TWL%) is an accurate indicator to assess efficiency of weight loss surgery (33). We found that a higher level of HNRNPA1 correlated with better TWL% (Fig. 8E), indicating that HNRNPA1 expression in SAT might predict the outcome of bariatric surgery (as illustrated in Fig. 8F).

In the current study, we demonstrate HNRNPA1 as a regulator that links adipose tissue inflammation to metabolic dysfunction. Hnrnpa1 ako mice were generated and used in metabolic studies to clarify the role of HNRNPA1 in insulin sensitivity, glucose tolerance, and hepatic glucose and lipid metabolism, as well as adipose tissue inflammation. Cell experiments revealed that HNRNPA1 knockdown led to adipocyte dysfunction. Mechanistically, HNRNPA1 regulated the mRNA stability of chemokine CCL2, which influenced the recruitment and activation of ATMs. Moreover, HNRNPA1 expression in human WAT was negatively correlated with indicators of fat accumulation and indicated the outcome of metabolic surgery.

Previous studies have reported that HNRNPA1 is involved in metabolic regulation. An earlier study found that HNRNPA1 was one of the RNA-processing genes with decreased expression in human liver and muscle samples (34). Hepatic HNRNPA1 overexpression could relieve hyperglycemia and liver steatosis from HFD feeding by binding and regulating calmodulin mRNAs (17). Studies in skeletal muscle showed that HNRNPA1 improves insulin resistance via regulating mRNA expression of CPT1b (18) or glycogen synthase 1 (26). The metabolic protective roles in the liver and muscle prompted us to explore the potential function of HNRNPA1 in adipose tissue. Although it was identified that loss of HNRNPA1 in murine skeletal muscle exacerbated HFD-induced insulin resistance and hepatic steatosis (26), we revealed the new role of HNRNPA1 in adipocytes via an entirely different inflammatory mechanism, emphasizing the contribution of adipocyte HNRNPA1 to systemic metabolism.

Apart from regulating alternative mRNA splicing by binding mRNA elements, HNRNPA1 also plays important roles via regulating mRNA stability (15). HNRNPA1 could bind with adenylate uridylate–rich sequences to modulate mRNA stability of IL-2 (35) and granulocyte-macrophage colony-stimulating factor, which maintains the function of T cells (36). In addition, HNRNPA1 could affect cellular senescence and apoptosis via regulating cIAP1 mRNA stability (37). Recent studies revealed that HNRNPA1 exerts antiaging effects in human lung fibroblast and vascular cells via regulating the degradation of SIRT1 or octamer-binding transcriptional factor 4 (38,39). Our study initially identified a downregulation of HNRNPA1 expression in obesity. A previous study reported a decrease in HNRNPA1 following insulin treatment (26), which suggested that chronic hyperinsulinemia may regulate the expression of HNRNPA1. This further highlighted the close relationship between HNRNPA1 and metabolism. Then, we identified that adipocyte HNRNPA1 deficiency disturbed the metabolic homeostasis in both in vivo and in vitro experiments. Considering that HNRNPA1 is a vital posttranscriptional regulator, we speculate that HNRNPA1 might function by regulating mRNA stability in adipose tissues.

Notably, proinflammatory factor CCL2 was hypothesized to be a linker between HNRNPA1 depletion and adipose tissue proinflammatory response in this study. CCL2 is secreted not only by macrophages and endothelial cells but also by adipocytes (40). It is well established that CCL2 mediates macrophage recruitment and activation (41). HFD-induced and ob/ob obese mice had high expression levels of CCL2 in WAT, which has been extended to humans (42). AP2 promoter–mediated Ccl2 transgene mice exhibited ATM infiltration, insulin resistance, and hepatic steatosis under HFD feeding (40). Furthermore, CCL2 or CCR2 homozygous knockout mice showed improved inflammation and metabolic dysfunction (43). These studies identified the connection among CCL2, adipose tissue inflammation, and metabolic system disorders. Although the research on adipose tissue inflammation began earlier, emerging research continued to reveal its new mechanism and clinical significance (44). In our sequencing data of iWAT in Hnrnpa1 ako mice, the most noticeable changes were cytokine-cytokine receptor interaction and chemokine pathway. Increased CCL2 expression was identified in adipose tissue, primary differentiated adipocytes, and 3T3-L1 cells. Thus, we reveal a new regulatory mechanism of ATM infiltration through the HNRNPA1-CCL2 axis. Besides, the function of adipocyte secretion on systemic inflammation has also been continuously explored (45,46). Apart from its role in recruiting immune cells at the tissue level, previous studies found that CCL2 could directly regulate the function of adipocytes (42), pancreatic islet cells (47), skeletal muscle cells (48), and tubular epithelial cells (49) independent of CCR2. So, impaired adipocyte function may be due to the autocrine regulatory role of CCL2. This study clarifies the regulation and function of CCL2 derived from adipocytes.

Given the important roles of CCL2, previous studies examined its regulatory mechanisms. It can be transcriptionally induced by inflammatory stimuli (50), platelet-derived growth factor (51), and the mTORC1-FOXK1 axis (52). Furthermore, posttranscriptional regulation plays a crucial role in CCL2 expression. It was identified that oxygen radicals and pyroglutamate enhance CCL2 mRNA transcript stability (53). The glucocorticoid receptor bound directly to the three stem-loops in the 5′ end of CCL2 and destabilized its mRNA, implicating a role in immune responses (54). Platelet-derived growth factor, angiotensin II (55), prostaglandin E receptor subtype 2 signaling (56), and ribosomal protein L22 (57) were also reported to function in stabilizing or degrading CCL2 mRNA. Recently, Xiao et al. (58) reported that tristetraprolin decreased CCL2 mRNA stability by regulating m6A methylation to ameliorate hepatic injury. Parajuli et al. (59) showed that AT-rich interaction domain-containing protein 5a stabilized CCL2 mRNA to regulate the tumor microenvironment. Therefore, as a vital chemokine in adipose tissue and a potential therapeutic target, our findings provide evidence that HNRNPA1 interacts directly with CCL2 and regulates its mRNA stability in differentiated adipocytes.

In concordance with that study based on Ccl2 transgenic mice, Hnrnpa1 ako mice showed a similar phenotype when maintained on an HFD, including impaired glucose tolerance and insulin sensitivity, accumulated ATM infiltration, and aggravated hepatic steatosis under HFD feeding, while the body weight, adipose tissue weight, and food intake showed no change in Hnrnpa1 ako mice. We also show that administration of CCL2-CCR2 signaling inhibitor to obese mice inhibited the ATM infiltration and partly improved glucose homeostasis. All our results highlight that adipocyte HNRNPA1 functions via chemokine CCL2.

A limitation of our study is that we could not define whether CCL2 is the only target that mediates the effect of HNRNPA1 on adipose tissue inflammation. Given the differentially expressed genes revealed by sequencing results, it is conceivable that HNRNPA1 may have other as-yet undiscovered target genes in adipocytes. Additionally, double adipocyte-specific Hnrnpa1/Ccl2 knockout mice would likely be the optimal choice to demonstrate the role of HNRNPA1 acting through CCL2 in adipocytes. Besides, the underlying mechanism needs to be investigated in further studies, since the specific binding site in CCL2 was not elucidated. Unraveling in-depth mechanisms involved in HNRNPA1-regulated inflammation in adipocytes will therefore be the focus of future studies.

In summary, we report that HNRNPA1 was downregulated in WAT of obese individuals and correlated with clinical indicators. The reduced HNRNPA1 inhibited CCL2 mRNA degradation, leading to ATM infiltration, which activated inflammation and metabolic dysfunction. Our data unravel a balance between HNRNPA1-CCL2 and adipose tissue inflammation, which is disturbed in obesity. Further studies are needed to understand the comprehensive mechanism and its potential therapeutic impact.

Acknowledgments. The authors thank Dr. Jiangfeng Ke (People’s Hospital of Fujian University of Traditional Chinese Medicine) for providing adipoq-Cre transgenic mice.

Funding. This research was supported by National Natural Science Foundation of China grants 82200897 and 82170889 and Shanghai Research Center for Endocrine and Metabolic Diseases grant 2022ZZ01002.

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

Author Contributions. X. Li, Y.S., and Y.X. performed experiments. X. Li, X. Lu, J.S., and W.L. interpreted data. X. Li, Y.Y., and Y.B. conceived the idea and developed the study design. X. Li wrote the manuscript. T.H., J.Z., and X.M. analyzed the RNA-seq and clinical data. Y.Y. and Y.B. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.