Retinol-Binding Protein 4 Activates STRA6, Provoking Pancreatic β-Cell Dysfunction in Type 2 Diabetes

Retinol-binding protein 4 (RBP4), primarily known as a principal retinol transporter in plasma that delivers retinol from hepatocytes to peripheral target tissues, is expressed mainly in hepatocytes and relatively lower in adipocytes and skeletal myocytes (1). It was first identified as a diabetogenic adipokine in 2005 when Yang et al. (2) reported that RBP4 levels were elevated in different murine insulin resistance models as well as in obese patients with diabetes. Thereafter, ample evidence from epidemiologic investigations has supported that RBP4 is a diabetogenic factor that is significantly elevated in patients with type 2 diabetes (2–4) and could be an independent predictor of the incidence of type 2 diabetes in patients with prediabetes (5,6). Decreasing circulating RBP4 with sitagliptin could improve glycemia in a rat model of type 2 diabetes (7) as well as in women with gestational diabetes mellitus (8).

Although RBP4 is well recognized as a diabetogenic factor, the underlying molecular mechanisms are still far from clear. Until now, efforts made to elucidate the mechanisms mediating the diabetogenic effect of RBP4 have mainly focused on RBP4’s influence on insulin resistance. The underlying molecular mechanisms involve the impairment of insulin signaling pathways, such as phosphatidylinositol 3-kinase activity and insulin receptor substrate 1 phosphorylation in muscle, liver, and adipose tissue (2,9,10), as well as the activation of antigen-presenting cells and inflammation leading to the aggravation of systemic insulin resistance (11–13). Pancreatic β-cell dysfunction plays a decisive role in the onset and progression of type 2 diabetes. Hyperglycemia does not occur unless β-cell function is no longer sufficient to overcome insulin resistance (14). However, limited attention, except for a few epidemiologic studies, has been paid to the association of RBP4 with β-cell function, and the results have been highly inconsistent (15–18). The effect of RBP4 on β-cell function remains elusive, and the underlying mechanisms are unknown. In the present study, we sought to explore the possible effect of RBP4 on pancreatic β-cell function in mouse models and further decipher its underlying molecular mechanisms.

All animal experiments were approved by the animal care and use committee of Sun Yat-sen University. Transgenic mice expressing human RBP4 on a C57BL/6J background (RBP4-Tg) were constructed with CAG promotor as previously described (19). Male mice were used in all mouse studies to minimize the effect of confounders, such as sex steroids, on RBP4 levels and β-cell function (3,20). The expression levels of mouse and human RBP4 mRNAs were analyzed by quantitative real-time PCR with the primers indicated in Supplementary Table 1. Circulating RBP4 levels were measured with commercial kits (AG-45A-0012YEK-KI01 for mouse RBP4 and AG-45A-0010YEK-KI01 for human RBP4; AdipoGen) according to the manufacture’s guidelines.

Mouse RBP4 silencing and null adenoviruses were constructed by HanBio Technology (Shanghai, China). A mouse RBP4-shRNA (NM-011255.2-504s1c1) with superior silencing efficiency (>90%) was cloned into the adenovirus shuttle plasmid U6. Viruses were purified, dialyzed, and stored at −80°C to avoid repeated freezing and thawing. Six-week-old BKS.Cg-Dock 7m^+/+Leprdb/J (db/db) mice were randomly assigned to tail vein injection of adeno-associated virus (AAV) encoding shRBP4 or green fluorescent protein (GFP) control group (1 × 10¹² genomic copy number per mouse), and its wild-type (WT) littermates were used as control.

To construct AAV-β-short hairpin stimulated by retinoic acid 6 (shSTRA6), DNA fragments encoding mouse shSTRA6 (TGCAGTTGCTACAGACCAA) were inserted into a phosphorylated adenovirus-insulin 1-EGFP-mir30-shRNA vector (vector map shown in Supplementary Material). AAV serotype 8 was used in this study. After packaged, purified, and sequenced to ensure accuracy, AAV-β-shSTRA6 and its negative control AAV-β-GFP were intraperitoneally injected into 10-week-old RBP4-Tg or WT mice (4 × 10¹² genomic copy numbers per mouse).

Primary islets were isolated, and INS-1E cells were cultured as previously described (21). For glucose-stimulated insulin secretion (GSIS), the isolated islets and INS-1E cells were incubated with Krebs-Ringer bicarbonate buffer containing 0.1% fatty acid–free BSA for 30 min followed by treatment with 2.0 mmol/L (basal) or 16.8 mmol/L (stimulated) glucose. Dynamic insulin secretion of the isolated islets was analyzed using a perfusion system according to a previous publication (21), while eluted fractions were collected at indicated time points and the first-phase defined as 0–10 min. Insulin concentration of the supernatant was measured by commercial ELISA kits (80-INSMSU-E10, Mouse Insulin Ultrasensitive ELISA Kit [ALPCO Diagnostics], and 10-1250-01, Rat Insulin ELISA Kit [Mercodia]).

Human retinol–bound RBP4 (holo-RBP4, NP_006735.2, Met 1-Leu 201) was expressed in HEK293 cells with a COOH-terminal polyhistidine tag obtained from Sino Biological Inc. (Beijing, China). The endotoxin content was <1.0 endotoxin units per 1 g of the protein as determined by the Limulus amebocyte assay. The recombinant human RBP4 consists of 194 amino acids after removal of the signal peptide and migrates as an ∼23-kD protein as predicted (22). RBP4 used in the in vitro experiment was holo-RBP4, unless it was specifically marked as retinol-free RBP4 (apo-RBP4).

The human stimulated by retinoic acid 6 (STRA6) cDNA (Open Biosystems) was subcloned into EcoRI/XhoI of pCMV-Tag2 (Stratagene). STRA6-T644M and STRA6-Y643M point mutants were generated using the QuikChange Site-Directed Mutagenesis Kit (Stratagene). Transfections were accomplished using Lipofectamine 3000 (Invitrogen) according to the manufacturer’s guidelines.

Mouse pancreas tissues were embedded in Tissue-Tek embedding medium and then cut into 8-μm sections for immunofluorescence staining. Formalin-fixed paraffin-embedded pancreas tissues were cut into 8-μm sections for immunohistochemistry staining. Slides were incubated with indicated primary antibodies (insulin [ab8304, 1:400; Abcam], glucagon [ab36232, 1:400; Abcam], STRA6 [orb158514, 1:200; Biorbyt]) overnight at 4°C and then with the corresponding secondary antibody after pretreatment. Immunohistochemical staining sections were observed with a Leica DMRXE upright brightfield microscope. Immunofluorescence images were captured with a laser scanning confocal microscope (TCS SP5 II; Leica). Insulin and glucagon density were then calculated by ImageJ software (National Institutes of Health, Bethesda, MD) with the same settings.

After mouse pancreas biopsy samples were sectioned and mounted, in situ hybridization was carried out using fluorescently labeled RNAScope probes for insulin (mRNA region 282–371) and STRA6 mRNA (region 782–961) with DAPI counterstain according to kit instructions. Images were captured using an inverted fluorescence microscope with oil immersion lens (Leica). Negative control probes generated no fluorescent signals.

Specimens intended for transmission electron microscopy scan were fixed in 4% paraformaldehyde/2.5% glutaraldehyde and then postfixed in 2% osmium tetroxide, ethanol dehydrated, and embedded in Epon 812 resin. Semithin cross sections (1 μm) were stained with toluidine blue before selecting representative areas for further thin sectioning. Ultrathin cross sections (40–60 nm) were cut and mounted on bare copper grids and strained with uranyl acetate and lead citrate. Finally, the sections were analyzed and photographed with a JEOL-100SX transmission electron microscope (JEOL Ltd., Tokyo, Japan).

INS-1E cells treated with or without RBP4 were used to perform chromatin immunoprecipitation (ChIP) assays. After lysis and sonication, samples were incubated overnight at 4°C with anti-STAT1 antibody or rabbit anti-IgG antibody as negative control. Immunocomplexes were then collected with magnetic beads and incubated for 3 h at 62°C in high-salt solution to reverse the crosslink reaction. DNA fragments were analyzed by real-time PCR with the following primer pair for the STAT1 binding sites 21,262–21,143 base pairs (bp) of the rat Isl-1 promoter: 5′-CTCAGAAAGACGGTAGATTT-3′ and 5′-AGCTTTCTAATTTGTTTCC-3′. The resulting products were separated on 2% agarose gel and stained with ethidium bromide while 10% input DNA solution was analyzed simultaneously for normalization.

Tissues or cells were lysed in radioimmunoprecipitation assay buffer with protease inhibitor (Ventana Medical Systems, Tucson, AZ). The total amount of protein was determined using bicinchoninic acid colorimetric method and separated by SDS-PAGE. After incubation with primary antibodies and indicated secondary antibodies, immunoblots were visualized with the chemiluminescence Western blot detection kit (ECL; Thermo Fisher Scientific, Waltham, MA).

Immunoprecipitation analysis was conducted with the Co-immunoprecipitation Kit (Thermo Fisher Scientific) according to the manufacturer’s instruction. Briefly, INS-1E cells treated with 80 μg/mL RBP4 were solubilized in lysis buffer, and then 1,000-μg protein extracts were used to incubate with anti-STRA6 antibodies (ab77616; Abcam). The antigen sample and antibody mixture were collected with protein A/G magnetic beads and eluted from the beads for immunoblot analysis with anti-RBP4 (ab188230; Abcam) and anti-STRA6 (ab77616; Abcam). The data were quantified using ImageJ software.

Total RNA was extracted using TRIzol reagent (Life Technologies, Burlington, ON, Canada) from tissues, isolated islets, or INS-1E cells, and cDNA was synthesized from 0.5 mg total RNA by reverse transcription using PrimeScript RT Master Mix (Takara, Kyoto, Japan). Quantitative real-time PCR was performed using SYBR Green qPCR Master Mix with an Applied Biosystems Prism 7000 sequence detection system. The specific primers used in this study are shown in Supplementary Table 1.

The data are expressed as means ± SEMs. Unpaired Student t test was used to compare the difference between two groups, and one-way ANOVA followed by Bonferroni post hoc test was applied for comparisons among multiple experimental groups. SPSS version 22.0 statistical software (IBM Corporation, Chicago, IL) was used for all analyses. P < 0.05 was considered to be statistically significant.

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

First, we evaluated the association of circulating RBP4 with β-cell function in the progression of hyperglycemia in db/db mice. The 5–7-week-old db/db mice with modestly elevated glucose concentrations (150–250 mg/dL) were recognized as prediabetic and glucose concentrations >500 mg/dL as diabetic, as previously described (23,24). Insulin expression level in islets (Fig. 1A and B) and insulinogenic index calculated from GSIS (Fig. 1C) were gradually and significantly deteriorated in prediabetic and diabetic mice. Circulating RBP4 levels were steadily increased with the progression of hyperglycemia in db/db mice compared with control mice (Fig. 1D) and inversely correlated with the insulinogenic index (r = −0.582; P < 0.05) (Fig. 1E).

The expression levels of RBP4 in β-cells is still controversial, although RBP4 mRNA was detected in human and mouse islets, but RBP4 protein was not stained in β-cells (25–28). In this study, we detected extremely low RBP4 expression levels in primary islets and β-cells compared with that in liver (Supplementary Fig. 1A). To assess the possibility of the direct effect of RBP4 on GSIS, primary isolated islets from C57BL/J mice and β-cell line (INS-1E) were then treated either with different doses of exogeneous recombinant RBP4 protein (0, 40, 80, 120 μg/mL), which cover the range of previously reported circulating RBP4 levels in subjects with type 2 diabetes (4), for 24 h or with 80 μg/mL RBP4 for different times (0, 12, 18, 24 h). We found that RBP4 could dose- and time-dependently suppress GSIS in primary isolated islets (Fig. 2A–D). The stimulatory index was decreased by 38%, 47%, and 70% in 40, 80, and 120 μg/mL RBP4, respectively (Fig. 2B). RBP4 80 μg/mL decreased GSIS with prolonged treatment, reaching significance at 18 h (Fig. 2C and D). Similar responses were observed in RBP4-treated INS-1E cells (Fig. 2E–H). However, 80 μg/mL apo-RBP4 had no significant effect on GSIS in isolated primary islets (Supplementary Fig. 2).

To further characterize the role of RBP4 on β-cell function in vivo, RBP4-Tg mice expressing human RBP4 were generated and bred as previously reported (19). The dynamic changes of parameters related to glucose homeostasis were monitored (Supplementary Fig. 3A–H). RBP4-Tg mice exhibited a dynamic decrease of GSIS, which appeared as early as 8 weeks of age (Fig. 3A–C), while insulin sensitivity was not changed until 16 weeks of age as measured by intraperitoneal insulin tolerance test (Supplementary Fig. 3F). Moreover, a significant decrease of random fed and fasting insulin levels was detected at 24 and 28 weeks of age (Supplementary Fig. 3G and H). In line with these in vivo observations, the islets isolated from RBP4-Tg mice exhibited a significant decrease of insulin secretion (∼40%) in response to the glucose stimulation (Fig. 3D and E), with the first-phase GSIS dramatically inhibited by up to 80% (Fig. 3F). Insulin density was markedly decreased in RBP4-Tg islets (Fig. 3G and H), while no significant changes of glucagon density (Fig. 3I), pancreatic β-cell area (insulin-positive cells) (Fig. 3J), and α-cell area (glucagon-positive cells) (Fig. 3K) were detected by immunofluorescent staining. Downregulation of RBP4 by administration of shRBP4 could effectively rescue β-cell dysfunction in RBP4-Tg mice (Supplementary Fig. 4A–E). Taken together, these data provide strong evidence that RBP4 could directly impair the pancreatic β-cell function in vitro and in vivo.

To determine how RBP4 decreases insulin secretion in β-cells, we first screened a panel of genes involved in the process of GSIS, including glucose uptake; glucose metabolism; ATP-dependent potassium channels; voltage-gated calcium channels; soluble N-ethylmaleimide attachment protein receptor (SNARE) complex proteins, which mediate exocytosis of insulin granules; and gap junction channels, which mediate cell-cell communication and synchronization (29). Unexpectedly, no significant alteration of these genes was detected in both islets isolated from RBP4-Tg mice (Fig. 4A) and RBP4-treated INS-1E cells (Fig. 4B). No significant changes of proinflammatory and anti-inflammatory cytokines were found in islets isolated from RBP4-Tg mice (Supplementary Fig. 5) compared with WT littermates, despite the previously reported proinflammatory effects of RBP4 in adipose tissue (11,13).

In contrast, a dramatic reduction of Ins-1 and Ins-2 transcription was detected in both primary islets isolated from RBP4-Tg mice (65% and 80%, respectively) (Fig. 4C) and RBP4-treated INS-1E cells (20% and 40%, respectively) (Fig. 4D). Consistently, the significant reduction of insulin granules in pancreatic β-cells of RBP4-Tg mice was also detected by transmission electron microscope imaging (Fig. 4E). Insulin synthesis is under the regulation of a batch of transcription factors (30). Isl-1 was the only transcription factor downregulated in response to RBP4 stimulation, with a decrease by 40% in islets from RBP4-Tg mice (Fig. 4F) and 30% in RBP4-treated INS-1E cells (Fig. 4G), while no changes of other screened transcription factors (including MafA, Pdx1, NeuroD, Pax6, Lmx1a, Hnf4α, and Nkx6.1) were detected (Fig. 4F and G). Overall, these results suggest that the deleterious effect of RBP4 on β-cell function was directly through repressing insulin transcription.

To further uncover the molecular mechanisms underlying the inhibitory effect of RBP4 on insulin synthesis, we sought to explore whether STRA6, the only identified specific cell surface receptor for RBP4, was involved in this effect. There are as-yet no reports regarding its expression and function in islets and β-cells. In this study, STRA6 was detected to be expressed in islets as well as in INS-1E cells on both mRNA (Fig. 5A) and protein levels (Fig. 5B). Interestingly, STRA6 was specifically colocalized with insulin-positive β-cells in islets (Fig. 5C and D) and INS-1E cells (Fig. 5E) as indicated by immunofluorescence staining or in situ hybridization. The direct interaction between exogeneous RBP4 and STRA6 in β-cells was confirmed by coimmunoprecipitation analysis (Fig. 5F). We found that RBP4 could increase both STRA6 protein expression and STRA6 phosphorylation in a dose-dependent manner (Fig. 5G and H). At 40 μg/mL, RBP4 increased the total amount of STRA6 protein (Fig. 5G) but not the ratio of STRA6 phosphorylation to total STRA6 (Fig. 5H). Genetic inhibition of STRA6 with specific siRNA totally blocked its upregulation induced by RBP4 and reversed the suppression effect on GSIS (Fig. 5I and J and Supplementary Fig. 6A and B), suggesting that STRA6 was involved in RBP4-mediated β-cell dysfunction.

Janus kinase (JAK)2/STATs signaling has been reported to be driven by RBP4-STRA6 to exert diverse effects in various tissues (9,31,32). Therefore, we proposed that the JAK2/STATs pathway might be involved in RBP4-STRA6–induced β-cell dysfunction. Herein, we found that treatment of RBP4 on INS-1E cells significantly augmented JAK2 phosphorylation, which subsequently promoted phosphorylation of transcription factor STAT1 at Tyr701, not STAT3 or STAT5 (Fig. 6A). We then performed site-directed mutagenesis of STRA6 in tyrosine residue (Y643F) and threonine residue (T644M) of the putative SH2 domain-binding motif, the site involved in the activation of JAKs and docking of the transcription factors STATs (9). Mutations of the receptor abolished RBP4-induced JAK2/STAT1 activation (Fig. 6B) and GSIS reduction (Fig. 6C and D). Pharmacological inhibition of STAT1 phosphorylation using fludarabine, a specific chemical inhibitor of STAT1, effectively prevented the RBP4-induced reduction of GSIS and preserved ISL-1 expression (Fig. 6E and F and Supplementary Fig. 7A).

Activation of JAK2/STATs signaling in β-cells has been reported to transcriptionally inhibit insulin synthesis by several cytokines or adipokines through binding to TT(N)nAA (n-bp-spacing elements between the TT-AA core sequence) element in the promoter region (33–36). We next explored whether RBP4-induced activation of STAT1 had an increased ability to bind the element of Isl-1 promoter. We performed a ChIP analysis by using DNA fragments precipitated with anti-STAT1 from RBP4-treated cells. RBP4 treatment significantly increased the binding ability of STAT1 to the Isl-1 promoter region compared with untreated control cells (Fig. 6G and H).

To further confirm the critical role of STRA6 signaling in RBP4-induced β-cell dysfunction in vivo, RBP4-Tg mice were intraperitoneally injected with AAV expressing shSTRA6 under the control of a modified mouse insulin promoter to specifically silence the expression of STRA6 in β-cells (AAV-β-shSTRA6). As expected, STRA6 protein expression was markedly decreased in islets (Fig. 7A), not in epididymal adipose tissue (Fig. 7B and C), of RBP4-Tg mice after AAV-β-shSTRA6 infection. Body weight and visceral weight of RBP4-Tg mice were not affected by AAV-β-shSTRA6 treatment during the experiment period (Supplementary Table 2). In parallel with increased glucose clearance (Fig. 7D), β-cell dysfunction was effectively reversed as indicated by GSIS both in vivo and in vitro (Fig. 7E–H). Moreover, insulin expression in the islets of RBP4-Tg mice was preserved by STRA6 depletion (Fig. 7I and J). Activated JAK2/STAT1 cascade in islets of RBP4-Tg mice was also attenuated by the AAV-β-shSTRA6 treatment (Fig. 7K and L). Taken together, these results suggest that STRA6 was expressed in pancreatic β-cells, which mediated the inhibitory effect of RBP4 on insulin synthesis through the JAK2/STAT1/ISL-1 pathway.

Next, we evaluated the therapeutic role of RBP4 on β-cell dysfunction in the progression of type 2 diabetes. AAV expressing shRBP4 (AAV-shRBP4) was generated and injected into 6-week-old db/db mice with early-onset diabetes using the tail vein. RBP4 expression levels in liver (Fig. 8A) and circulating RBP4 concentrations (Fig. 8B) in AAV-shRBP4–injected mice were decreased to a level comparable to the control group. The mice with AAV-shRBP4 infection gained less weight (Fig. 8C and Supplementary Table 3), and both fasting glucose levels (Fig. 8D) and glucose clearance (Fig. 8E) were significantly improved, which were paralleled with increased random fed insulin levels (Fig. 8F) and GSIS (Fig. 8G) compared with AAV-GFP–treated db/db mice. Consistent with the increase of serum insulin levels, both insulin expression levels in islets and β-cell mass were partially preserved in AAV-shRBP4–infected db/db mice (Fig. 8H and I).

In the current study, we demonstrate that RBP4 could directly cause pancreatic β-cell dysfunction through STRA6/JAK2/STAT1/ISL-1–dependent inhibition of insulin synthesis (Fig. 8J). Targeting circulating RBP4 could preserve pancreatic β-cell function and thereby ameliorate hyperglycemia in murine type 2 diabetes models. Thus, our findings indicated that β-cell dysfunction is a new link between RBP4 and type 2 diabetes.

RBP4 has been reported to exert diverse effects in different cell types, including inducing adiposity in hepatocytes (37), activating antigen presentation in macrophages (11,13), and inhibiting insulin signaling in both adipocytes and cardiomyocytes (9,10,12,38). Here, we first demonstrated that the pancreatic β-cells are also the target cells of RBP4 that can directly suppress GSIS in a dose- and time-dependent manner in both primary isolated islets and the β-cell line. Moreover, RBP4-Tg mice exhibit a dynamic pancreatic β-cell dysfunction in vivo and decreased GSIS in isolated islets. It is important to note that RBP4 overexpression mice also exhibited multiple metabolic disorders, including increased body weight and plasma fatty acid concentration and the acceleration of hepatic steatosis (19), which might have had a secondary deleterious effect on β-cell function in vivo.

STRA6, the only known specific membrane receptor of RBP4, has been reported to not only mediate the intake of extracellular retinol into target cells (39,40) but also function as a signaling receptor of RBP4 to regulate lipid accumulation in adipocytes and tumorigenesis in the intestines (9,32,41). Here, using the combination of in situ hybridization and immunostaining, we provided an expression profile of STRA6 in pancreas and demonstrated that STRA6 is predominantly expressed in β-cells, which provides further support that β-cells could be direct targets of RBP4. The physiological role of STRA6 in β-cells is largely unknown. Whether it plays an essential role in the retinol homeostasis of β-cell needs further investigation. In this study, we found that RBP4 could both upregulate STRA6 expression and activate its phosphorylation in β-cells. Downregulation of STRA6 and mutation of the phosphorylation site of STRA6 could effectively reverse the inhibitory effect GSIS by RBP4. These observations support the critical role of STRA6 in mediating RBP4’s direct effect on β-cells. We also found that it was holo-RBP4 that inhibits insulin secretion through STRA6, while apo-RBP4 had no significant effect on insulin secretion. Consistent with previous reports, RBP4 only activates STRA6 when complexed to its binding partner retinol (9). To our knowledge, this is the first report on the direct effect of RBP4 on β-cells through its membrane receptor STRA6, which further expands the current understanding of the biological effects of RBP4 and provides an explanation for the previously reported association between STRA6 polymorphisms and insulin resistance or type 2 diabetes (42–44).

Perturbed JAK/STATs signaling in β-cells has been reported to regulate insulin transcription by multiple stimuli, including cytokines and adipokines (leptin or interferon-γ) (33,45). Here, we found that RBP4/STRA6 can also activate the JAK2/STATs signaling pathway, with special activation of STAT1 but not STAT3 or STAT5 in β-cells. Inhibition of STAT1 phosphorylation by using its chemical inhibitor fludarabine could effectively restore RBP4-induced reduction of GSIS, which in consistent with inactivation of STAT1 preventing interferon-γ–induced inhibition of β-cell insulin transcription and secretion (34). It was JAK2/STAT5 signaling activation driven by RBP4-STRA6 that impaired insulin signaling in adipose tissue and JAK2/STAT3 signaling that promoted tumorigenesis in a mouse model of colon cancer (9,31,32), indicating a cell type response to RBP4-STRA6 stimulation that warrants further research.

Isl-1 is a well-recognized transcription factor in the regulation of insulin synthesis and maintenance of normal β-cell function (46,47), and the single nucleotide polymorphisms of the Isl-1 gene have been reported to be associated with the risk of type 2 diabetes (48). In the present study, we found that the reduced insulin synthesis was attributed to the specific downregulation of Isl-1 in both islets from RBP4-Tg mice and RBP4-treated INS-1E cells. The activation of STAT1 by RBP4 gained increased binding ability to the promoter of Isl-1 and inhibited its expression, which further supports the role of Isl-1 in mediating the effect of adipokines or the JAK/STATs pathway on decreasing insulin transcription (33,36). Although no significant change was detected in the mRNA expression levels of other key molecules involved in the GSIS process except insulin transcription, we could not rule out other possibilities mediating the role of RBP4 on β-cell function, and decreased insulin synthesis and exocytosis might accompany the adaption change of electrical activity, calcium influx, or ATP levels in β-cells.

Finally, we further showed that lowering circulating RBP4 is able to improve β-cell dysfunction and therefore ameliorate hyperglycemia in db/db mice, which is supported by previous findings that sitagliptin decreases circulating RBP4 and improves β-cell function (7,49,50). Because RBP4 is identified to be one of the abundant adipokines in the bloodstream, this suggests that RBP4 may be an excellent therapy choice for diabetes, with a wide therapeutic window. As is the case with many new potential therapies, validation of the effect and safety of decreasing circulating RBP4 in other diabetes models will be needed before potential clinical testing.

In conclusion, this study has shown that RBP4 plays a previously unrecognized pathophysiological role in β-cell dysfunction in type 2 diabetes by repressing insulin synthesis through the STRA6/JAK2/STAT1/ISL-1 signaling pathway. Our findings shed new light on the diabetogenic role of RBP4, and it could be selected as a promising therapeutic target for β-cell dysfunction in type 2 diabetes.

Acknowledgments. The authors thank Dr. Baile Wang and Prof. Aimin Xu (the University of Hong Kong) for technical support. The authors thank Prof. Min Xia (Sun Yat-sen University) for sharing RBP4-Tg mice for this study.

Funding. This work was supported by the “100 Top Talents Program” of Sun Yat-sen University to L.Z. and by the Science and Technology Program of Guangzhou, China (20190401343), to L.Z.

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

Author Contributions. R.H. conducted the experiments. R.H. and L.Z. designed the research and wrote, reviewed, and edited the manuscript. X.B., X.L., and X.W. performed the animal study and provided material support. L.Z. is the guarantor of this work and, as such, has 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.