Hyperlipidemia-Induced MicroRNA-155-5p Improves β-Cell Function by Targeting Mafb

Failure of pancreatic β-cells to enhance insulin secretion in response to reduced systemic insulin sensitivity plays a key role in the development of hyperglycemia and type 2 diabetes (T2D) (1,2). Inflammatory macrophage recruitment into visceral adipose tissue during obesity frequently contributes to insulin resistance and adipocyte dysfunction through secretion of inflammatory cytokines (3). Although β-cells can maintain glucose homeostasis in insulin-resistant states by increasing circulating insulin levels, chronically elevated insulin secretion may result in exhaustion of β-cell function because of apoptosis or dedifferentiation (4,5). Obesity and T2D may promote β-cell failure by decreasing the secretion of glucagon-like peptide 1 (GLP-1), which enhances insulin secretion and β-cell function, from intestinal L cells (6–11). In addition, pancreatic α-cells can be a source of GLP-1, for instance, in response to interleukin-6 (IL-6)–mediated upregulation of proprotein convertase subtilisin/kexin type 1/3 (PC1/3; encoded by the proprotein convertase subtilisin/kexin-type [Pcsk1] gene) and thereby improve β-cell function during obesity (7,12,13).

Lipopolysaccharide (LPS), which leaks into the circulation after a high-fat meal because of increased intestinal permeability (14,15), also promotes insulin secretion by upregulating GLP-1 production (16,17). In the circulation, LPS binds primarily to lipoproteins such as LDL and VLDL (18,19). Notably, patients with familial hypercholesterolemia have a reduced risk for T2D and an increased LPS binding capacity because of the elevated lipoprotein levels (19,20). However, chronically elevated LPS levels during high-fat diet feeding also induce adipose tissue inflammation, insulin resistance, and obesity (21). In macrophages, many LPS effects are mediated through the highly conserved vertebrate microRNA-155-5p (miR-155-5p), which is preferentially upregulated upon Toll-like receptor 4 activation (22,23). Moreover, hyperlipidemia induces miR-155-5p expression in macrophages and thereby changes its effect from inhibiting macrophage proliferation in early atherosclerosis to impairing efferocytosis and promoting inflammatory activation in advanced lesions (23–28). In adipocytes, inflammatory cytokines, such as tumor necrosis factor-α, upregulate miR-155-5p expression, which may contribute to obesity progression in female mice by limiting brown adipose tissue differentiation (29,30). The role of miR-155-5p, however, in obesity and glucose homeostasis during hyperlipidemia-associated endotoxemia is unclear.

We found that endotoxemia induces miR-155-5p expression in pancreatic β-cells, which increases insulin secretion by targeting v-maf musculoaponeurotic fibrosarcoma oncogene family, protein B (Mafb) in hyperlipidemic mice. MafB represses IL-6 expression in β-cells and thereby inhibits intraislet GLP-1 production. Through this mechanism, miR-155-5p improved the adaptation of β-cells to hyperlipidemic stress and the compensation for obesity-induced insulin resistance and likely limited the progression of obesity and atherosclerosis.

For further details, refer to the Supplementary Data online.

Mir155^−/− mice were crossed with LDL receptor knockout (Ldlr^−/−) or apolipoprotein E knockout (Apoe^−/−) mice (all on a C57BL/6J background; The Jackson Laboratory) to obtain Mir155^−/−Ldlr^−/− mice and Mir155^−/−Apoe^−/− mice. Mir155^−/−Ldlr^−/− mice and Mir155^+/+Ldlr^−/− mice at 10–12 weeks of age were fed a diabetogenic diet supplemented with cholesterol (DDC; 35.5% fat and 36.3% carbohydrates with 0.15% weight-for-weight total cholesterol; ssniff Spezialdiäten) or a normal diet (ND; 3.3% fat; ssniff Spezialdiäten).

Murine pancreatic islets were isolated by collagenase digestion and density-gradient centrifugation. Briefly, collagenase P solution (4 mL, 1 mg/mL; Roche Diagnostics) was slowly injected into the common bile duct after occlusion of the ampulla in the duodenum. Islets were purified by gradient separation using sodium diatrizoate (Histopaque 1119 and Histopaque 1077; Sigma-Aldrich).

MIN6 cells (kindly provided by Dr. Ingo Rustenbeck, University of Braunschweig, Braunschweig, Germany), human islets (Pelobiotech), and GLUTag cells were transfected with locked nucleic acid (LNA)–miR-155-5p inhibitors (50 nmol/L; Exiqon), miR-155-5p mimics (15 nmol/L; Thermo Fisher Scientific), 155/Mafb target site blockers (TSBs; 50 nmol/L; Exiqon), or scrambled controls using Lipofectamine 2000 (Thermo Fisher Scientific).

MIN6 cells and human islets were cotransfected with miR-155-5p mimics and the pMirTrap vector using the Xfect miR transfection reagent in combination with Xfect Polymer (all from Clontech Laboratories). The pMirTrap vector expresses a DYKDDDDK-tagged GW182 protein. Cell lysates were incubated with anti-DYKDDDDK beads (Clontech Laboratories), and RNA was isolated from input and immunoprecipitated samples and analyzed by quantitative real-time PCR (qPCR). Fold enrichment of the target genes in the GW182 immunoprecipitates was normalized to the enrichment of Gapdh.

Ten-week-old Ldlr^−/− mice fed an ND were injected intravenously via the tail vein with 155/Mafb TSBs or control TSBs (each 0.4 mg/20 g/injection; miRCURY LNA microRNA Target Site Blocker for in vivo use; Exiqon), as described in Supplementary Data.

Data represent the mean ± SEM. Statistical analysis of microarray data were performed by a modified t test using GeneSpring software (GX13; Agilent Technologies). Student t tests and one-way ANOVAs followed by the Newman-Keuls post hoc test were used for statistical comparisons between groups using Prism 6 software (GraphPad Software). The variance is similar between the groups that are being statistically compared. A P value <0.05 was considered statistically significant.

To study the effect of miR-155-5p on atherosclerosis in the context of obesity and T2D, we deleted the miR-155-5p coding gene in hyperlipidemic Ldlr^−/− mice that develop atherosclerosis, obesity, and diabetes after DDC feeding (31). After a 24-week DDC feeding period, the development of atherosclerosis and the necrotic core formation in the lesions were increased in Mir155^−/−Ldlr^−/− mice compared with Mir155^+/+Ldlr^−/− mice (Fig. 1A). Lesions in Mir155^−/−Ldlr^−/− mice contained less macrophages than in Mir155^+/+Ldlr^−/− mice, whereas the lesional smooth muscle cell content was similar in both groups (Supplementary Fig. 1A). Total cholesterol and triglyceride plasma levels were higher in Mir155^−/−Ldlr^−/− mice than those in Mir155^+/+Ldlr^−/− mice after the 24-week DDC feeding period (Supplementary Fig. 1B). In Mir155^−/−Ldlr^−/− mice, the cholesterol level was increased in the VLDL and LDL fraction and reduced in the HDL fraction (Fig. 1B). Although Mir155^−/−Ldlr^−/− mice and Mir155^+/+Ldlr^−/− mice gained similar body weight in the first 20 weeks of DDC feeding, the body weight of Mir155^−/−Ldlr^−/− mice increased more than that of Mir155^+/+Ldlr^−/− mice during the last 4 weeks of the 24-week DDC feeding period (Fig. 1C). This effect in Mir155^−/−Ldlr^−/− mice was associated with an increase in epididymal white adipose tissue (eWAT) weight (Fig. 1D), adipocyte size (Fig. 1E), and macrophage infiltration in adipose tissue (Fig. 1F). Moreover, the expression of adiponectin (Adipoq) and leptin (Lep) was down- and upregulated, respectively, in the eWAT of Mir155^−/−Ldlr^−/− mice (Supplementary Fig. 1C). The expression of the proinflammatory macrophage-related gene nitric oxide synthase 2 (Nos2) and the anti-inflammatory macrophage marker mannose receptor, C type 1 (Mrc1) was not different between the groups (Supplementary Fig. 1C). Deletion of Mir155 did not alter Il6 mRNA expression, but increased Tnf expression in eWAT (Supplementary Fig. 1C).

During the DDC feeding period, fasting blood glucose (FBG) levels increased constantly at a similar rate in both groups of mice, whereas FBG levels in Mir155^−/−Ldlr^−/− mice were always higher than those in Mir155^+/+Ldlr^−/− mice (Fig. 1G). Surprisingly, FBG levels were also higher in Mir155^−/−Ldlr^−/− mice compared with Mir155^+/+Ldlr^−/− mice before feeding of the DDC (0 weeks) (Fig. 1G), indicating that the hyperglycemia in Mir155^−/−Ldlr^−/− mice is not because of the increased weight gain. These findings indicate that miR-155-5p improves glucose homeostasis in hyperlipidemic mice and thereby limits obesity and atherosclerosis.

To investigate the mechanism by which miR-155-5p affects glucose homeostasis, we determined the effect of Mir155 knockout on insulin and glucagon plasma levels. Fasting insulin levels were lower, whereas glucagon levels were higher in the plasma from Mir155^−/−Ldlr^−/− mice than in Mir155^+/+Ldlr^−/− mice fed the ND (0 weeks) and after the 24-week DDC feeding period (Fig. 2A), indicating that loss of miR-155-5p compromises islet function. In islets from ND-fed Mir155^−/−Ldlr^−/− mice, the percentage of insulin-expressing cells and the insulin content were reduced compared with Mir155^+/+Ldlr^−/− mice (Fig. 2B). Conversely, the percentage of glucagon-expressing cells and the glucagon protein content were higher in islets from Mir155^−/−Ldlr^−/− mice (Fig. 2C).

Proglucagon is processed to GLP-1 and glucagon by PC 1/3 (encoded by the Pcsk1 gene) and PC2 (encoded by the Pcsk2 gene), respectively (6,8). GLP-1 can be generated locally in pancreatic α-cells and increases insulin and reduces glucagon secretion (6). Therefore, we studied the effect of Mir155 knockout on GLP-1 expression. The intraislet GLP-1 protein content was reduced in ND-fed Mir155^−/−Ldlr^−/− mice (Fig. 2D). Plasma GLP-1 levels were also lower in Mir155^−/−Ldlr^−/− mice than in Mir155^+/+Ldlr^−/− mice fed an ND (0 weeks) or the DDC for 24 weeks (Fig. 2E). These effects were associated with decreased insulin (Ins) expression in islets (Fig. 2F) and in β-cells (Fig. 2G) and upregulation of glucagon (Gcg) expression in islets (Fig. 2F) and in α- cells (Fig. 2G) from Mir155^−/−Ldlr^−/− mice compared with those from Mir155^+/+Ldlr^−/− mice. Moreover, like in whole islets, Pcsk1 expression was downregulated (Fig. 2G) in α- and β-cells from Mir155^−/−Ldlr^−/− mice. By contrast, the expression of somatostatin (Sst) and of the β-cell transcription factors ISL LIM homeobox 1 (Isl1), aristaless-related homeobox (Arx), pancreatic and duodenal homeobox 1 (Pdx1), paired box 6 (Pax6), neurogenic differentiation 1 (Neurod1), and forkhead box A1 (Foxa1) in islets was not different between the groups (Supplementary Fig. 2A). Islet cell apoptosis and the accumulation of macrophages or T cells in islets were negligible in both groups of mice (Supplementary Fig. 2B).

In vitro, miR-155-5p mimics treatment downregulated Gcg and Pcsk2 mRNA expression and upregulated Pcsk1 mRNA expression in MIN6 cells (Fig. 2H and Supplementary Fig. 2C). At the protein level, miR-155-5p mimic treatment increased the cellular insulin and GLP-1 content and reduced the glucagon level in MIN6 cells compared with control mimics (Fig. 2H). Accordingly, Pcsk1 and Pcsk2 expression was increased and reduced, respectively, in sorted α-cells after treatment with miR-155-5p mimics compared with control mimics (Supplementary Fig. 2E). In human islets, miR-155-5p mimic decreased the expression of GCG in α-cells and increased INS expression in β-cells, whereas PCSK1 expression was upregulated in α-cells (Supplementary Fig. 2F). Conversely, miR-155-5p inhibitor treatment increased Gcg and Pcsk2 expression and reduced Pcsk1 expression in MIN6 cells (Supplementary Fig. 2C and D), which resulted in decreased insulin and GLP-1 content and increased glucagon content (Supplementary Fig. 2D).

In addition, overexpression of miR-155-5p promoted GLP-1 secretion from murine islets (Fig. 2I). By contrast, treatment of an enteroendocrine L-cell line with miR-155-5p mimics did not affect GLP-1 protein content and secretion and GCG and PCSK1 mRNA expression (Supplementary Fig. 2G and H). Hence, miR-155-5p promotes intraislet GLP-1 production by upregulating Pcsk1 expression and may thereby improve glucose homeostasis.

Next, we studied glucose tolerance in Mir155^−/− mice in the absence and presence of hyperlipidemia. Notably, Mir155 knockout increased blood glucose levels following intraperitoneal glucose challenge in hyperlipidemic male and female Ldlr^−/− (Fig. 2J and Supplementary Fig. 3A) or Apoe^−/− mice (Supplementary Fig. 3B and C) fed an ND, whereas glucose tolerance was not affected by Mir155 knockout in normal, lipidemic, male ND-fed Ldlr^+/+ mice (Fig. 2K). Thus, miR-155-5p improved glucose homeostasis only under hyperlipidemic conditions.

Next, we studied the regulation of islet miR-155-5p expression by hyperlipidemia and LPS. Feeding Ldlr^−/− mice the DDC for 24 weeks increased plasma endotoxin activity and islet miR-155-5p expression compared with ND feeding (Fig. 3A). In 10- to 12-week-old ND-fed mice, knockout of Ldlr increased plasma cholesterol and triglyceride levels (Supplementary Fig. 4A), circulating endotoxin activity, and islet miR-155-5p expression (Fig. 3B and C). miR-155-5p was mainly detectable in glucagon⁻ cells in islets by combined immunostaining and in situ PCR (Fig. 3C). In contrast to native LDL (nLDL), mildly oxidized LDL (moxLDL) upregulated miR-155-5p expression in MIN6 cells compared with vehicle treatment (Fig. 3D). LPS stimulation increased miR-155-5p expression in MIN6 cells (Fig. 3E) and human islet cells (Fig. 3F). Moreover, mild oxidation increased the endotoxin activity in LDL (Fig. 3G). Knockout of Ldlr in ND-fed mice resulted in deposition of oxidized LDL (oxLDL) in islets (Fig. 3H). Thus, enhanced endotoxin activity of oxLDL deposited in islets during hyperlipidemia may induce miR-155-5p expression in β-cells.

LPS-induced insulin release from islets following glucose stimulation was decreased in islets from Mir155^−/−Ldlr^−/− mice (Fig. 3I). Treatment of Ldlr^−/− mice with low-dose LPS upregulated islet miR-155-5p expression (Fig. 3J) and increased insulin and GLP-1 plasma levels (Supplementary Fig. 4B). The glucose-lowering effect of low-dose LPS following intraperitoneal glucose injection in Ldlr^−/− mice (Fig. 3K) was partially abolished by Mir155 knockout (Fig. 3K). Together, these data suggest that hyperlipidemia-induced miR-155-5p expression improves β-cell adaptation to hyperlipidemia-associated endotoxemia stress.

To determine how miR-155-5p regulates β-cell function, we analyzed the effect of Mir155 knockout on islet gene expression by microarray analysis. In ND-fed Mir155^−/−Ldlr^−/− mice, 239 genes were upregulated (Supplementary Table 1), and 420 genes were downregulated (Supplementary Table 2) compared with Mir155^+/+Ldlr^−/− mice (P < 0.05; absolute fold change ≥1.5; n = 3 samples/group). Differentially regulated genes were enriched in the carbohydrate and lipid metabolism pathways and in pathways related to endocrine system function, cellular growth, DNA replication, and cell survival, as determined by Ingenuity Pathway Analysis software (Fig. 4A). Analysis of potential upstream regulators of differential gene expression in islets indicated Cdkn1 activation and Cdk4 inhibition in Mir155^−/−Ldlr^−/− mice, which may reduce islet cell proliferation (Fig. 4B) (32). PTEN activation, which contributes to β-cell failure in mouse models of T2D (33), was increased in Mir155^−/−Ldlr^−/− mice (Fig. 4B). Moreover, Glut2-dependent pathways and pathways related to cyclic AMP, GLP-1, and glucose-dependent insulinotropic polypeptide signaling were inhibited, suggesting impaired glucose uptake and insulin secretion (34).

Among the inflammatory pathways, IL-6 receptor activation was reduced, and signaling pathways downstream of the IL-6 receptor, such as the Janus kinase/STAT and extracellular signal–regulated kinase 1/2 pathways, were inhibited in Mir155^−/−Ldlr^−/− mice (Fig. 4B). Accordingly, islet IL-6 mRNA and protein expression and the number of IL-6–producing β-cells were reduced in Mir155^−/−Ldlr^−/− mice compared with Mir155^+/+Ldlr^−/− mice (Fig. 4C and D). Whereas Il6 expression was unchanged in sorted α-cells, it was downregulated in β-cells from Mir155^−/−Ldlr^−/− mice compared with Mir155^+/+Ldlr^−/− mice (Fig. 4E). Moreover, IL6 expression was upregulated by miR-155-5p mimic treatment in human β-cells, but was not affected in α-cells compared with control mimic (Fig. 4F). In vitro, gain- and loss-of-function experiments demonstrated that miR-155-5p upregulates IL-6 mRNA and protein expression in MIN6 cells (Supplementary Fig. 5A and B). Inhibition of IL-6 secreted from MIN6 cells using a blocking IL-6 antibody reduced Ins and Pcsk1 expression and increased Pcsk2 expression (Fig. 4G). In addition, treatment of sorted human α-cells with conditioned medium from miR-155-5p mimic–treated human β-cells enhanced GLP-1 secretion and the cellular GLP-1 content (Supplementary Fig. 5C). Taken together, these results indicate that miR-155-5p in β-cells stimulates the expression and secretion of IL-6, which in turn increases GLP-1 production by upregulating Pcsk1 expression in α-cells.

To determine the targets that mediate the effect of miR-155-5p on IL-6 expression in β-cells, we screened the 3′-untranslated region (UTR) of the genes upregulated in islets from Mir155^−/−Ldlr^−/− mice for miR-155-5p binding sites. According to the TargetScan (v7.0) prediction algorithm, 27 out of the 239 upregulated genes, including Mafb, semaphorin 5A (Sema5a), and mediator complex subunit 12-like (Med12l), contained miR-155-5p binding sites (Table 1). The miR-155-5p target sites in the Mafb and Sema5a 3′-UTRs were conserved among species, whereas the other 25 sites were poorly conserved. However, three of the poorly conserved sites were also found in humans, including the site in the AU RNA-binding protein/enoyl-CoA hydratase (Auh), stathmin-like 2 (Stmn2), and Med12l mRNAs. Upregulation of islet Auh, Mafb, Med12l, Sema5a, and Stmn2 expression in Mir155^−/−Ldlr^−/− mice was confirmed by qPCR (Fig. 5A). Treatment of MIN6 cells with miR-155-5p mimics (Fig. 5B) and inhibitors (Supplementary Fig. 6A) reduced and increased the expression of Auh, Mafb, Med12l, Sema5a, and Stmn2, respectively.

Next, we performed immunoprecipitation of the microRNA-induced silencing complex using extracts from MIN6 cells and human islets overexpressing FLAG-tagged GW182 (35). Among the potential target genes, miR-155-5p mimic treatment most strongly enriched Mafb in the miR-induced silencing complex of MIN6 cells and human islets. In contrast to Auh, Med12l, and Sema5a, Stmn2 was also enriched by miR-155-5p in both cell types (Fig. 5C and Supplementary Fig. 6B). In Ldlr^−/− mice, Mir155 knockout increased the number of MafB-expressing cells in islets and, in contrast to α-cells, upregulated Mafb expression in sorted β-cells (Fig. 5D and E). Moreover, miR-155-5p mimic treatment reduced MAFB expression in human β-cells, but not in α-cells (Supplementary Fig. 6C).

The miR-155-5p binding site in the MAFB 3′-UTR has been previously verified in B-cell lymphoma cells (Supplementary Fig. 6D) (36). To test the function of this site, we designed LNA-modified oligonucleotides that selectively inhibit the interaction between miR-155-5p and Mafb (155/Mafb TSB) (Supplementary Fig. 6D). In MIN6 cells, treatment with 155/Mafb TSBs increased Mafb, Gcg, and Pcsk2 expression and reduced Ins and Pcsk1 expression compared with control TSBs (Fig. 5F). Notably, 155/Mafb TSB treatment reduced IL-6 expression at the mRNA and protein level (Fig. 5G). These data indicate that the effects of miR-155-5p on β-cells are mainly mediated by the targeting of Mafb.

To study how Mafb regulates IL-6 expression, MIN6 cells were transfected with a luciferase reporter vector containing the wild-type Il6 promoter or the Il6 promoter containing mutations in the predicted Mafb binding sites Mafb1 and Mafb2 (Supplementary Fig. 6E). Treatment with miR-155-5p inhibitors reduced the luciferase activity in cells expressing the wild-type promoter (Fig. 5H and Supplementary Fig. 6F), but not in cells expressing the promoter containing the mutated Mafb binding sites (Fig. 5H). These findings suggest that reduced MafB-mediated transcriptional repression of IL-6 contributes to the effect of miR-155-5p on β-cell function.

To study whether the effect of hyperlipidemia-induced miR-155-5p in β-cells on glucose homeostasis is mediated by the suppression of Mafb, Ldlr^−/− mice were treated with 155/Mafb TSBs or nontargeting, LNA-modified oligonucleotides (control TSB). Body weights and differential blood counts were not different between the groups at 21 days after the treatment (Supplementary Fig. 7A and B). Mafb mRNA expression levels were increased in islets and spleen, but not in heart, liver, and eWAT in 155/Mafb TSB-treated mice (Fig. 6A). 155/Mafb TSB treatment did not affect islet Auh, Med12l, Sema5a, and Stmn2 expression levels (Supplementary Fig. 7C). The percentage of MafB-expressing cells in islets was higher in 155/Mafb TSB-treated mice than in mice treated with control TSBs (Fig. 6B). Treatment with 155/Mafb TSBs increased Gcg mRNA expression and the percentage of α-cells and reduced Pcsk1 and Il6 expression and the percentage of β-cells compared with control (Fig. 6C and D). This effect in 155/Mafb TSB-treated mice was associated with reduced insulin and GLP-1 plasma levels and increased glucagon plasma levels (Fig. 6E). Moreover, 155/Mafb TSB treatment elevated FBG levels (Fig. 6F) and impaired glucose tolerance following intraperitoneal glucose injection (Fig. 6G). These data indicate that hyperlipidemia-induced miR-155-5p expression improves β-cell adaptation and maintains glucose hemostasis by suppressing Mafb.

We found that hyperlipidemia and LPS upregulate miR-155-5p expression in β-cells, which improved glucose homeostasis by targeting Mafb. In the absence of miR-155-5p, upregulation of MafB by hyperlipidemia inhibits IL-6 expression and thereby reduces IL-6–mediated GLP-1 production in α-cells (Fig. 7). In obese mice, miR-155-5p–induced GLP-1 production may limit atherosclerosis, dyslipidemia, and the progression of adiposity and improve the adaptation of β-cells to insulin resistance.

Upregulation of miR-155-5p in response to LPS plays an essential role in inflammatory macrophage activation (37). In addition, moxLDL promotes inflammatory activation and miR-155-5p expression in macrophages in a Toll-like receptor 4–dependent manner (26). In line with these results, we found that LPS and moxLDL increase miR-155-5p expression in β-cells. LPS binds to LDL in the circulation, which reduces the biological activity of LPS and promotes endotoxin removal (18,38,39). However, mild oxidation of LDL increased its endotoxin activity, presumably because of altered interactions between lipids from LDL and LPS, suggesting that LPS mediates the effect of moxLDL on miR-155-5p expression. Moreover, our finding that the deposition of oxLDL in islets and endotoxemia were increased in Ldlr^−/− mice indicates that LPS contributes to the upregulation of miR-155-5p in β-cells by hyperlipidemia. Low-dose LPS improves insulin secretion by upregulating GLP-1 production (17) and knockout of Mir155 reduced the effect of LPS on glucose metabolism, suggesting that β-cell miR-155-5p contributes to LPS-induced insulin secretion. Accordingly, Mir155 knockout elevated plasma glucose levels in Ldlr^−/− mice because of reduced insulin and increased glucagon production in islets. The effect of miR-155-5p on islet function is likely mediated by the upregulation of intraislet GLP-1, which improves β-cell function and inhibits glucagon expression (6,7).

The main mechanisms of β-cell failure in T2D development involve dedifferentiation, apoptosis, and impaired regeneration of β-cells (2,4,40). The members of the large Maf protein transcription factor family, MafA and MafB, play critical roles in the development and function of α- and β-cells. In adult rodent islets, MafA is only expressed in β-cells and promotes insulin expression, whereas MafB is exclusively expressed in α-cells and induces Gcg transcription (41–44). In pregnant or obese mice, however, MafB expression is upregulated in β-cells (45). Moreover, derepression of MafB in the absence of the β-cell–specific transcription factor pancreatic and duodenal homeobox 1 (Pdx1) leads to β-to-α-cell reprogramming, which may contribute to β-cell failure in T2D (40,43,46). Notably, high-fat diet feeding induced re-expression of MafB in β-cells, suggesting that hyperlipidemia promotes β-to-α-cell conversion (45). Accordingly, our findings indicate that hyperlipidemia-induced expression of miR-155-5p in β-cells limits the upregulation of MafB and thereby improves β-cell function, probably because of enhanced GLP-1 production in α-cells. In mouse models of obesity and diabetes, IL-6 increases intraislet GLP-1 expression in α-cell by upregulating Pcsk1 expression (12). Notably, miR-155-5p mediates LPS-induced IL-6 expression in macrophages (22,47), and high-fat diet feeding and inflammatory cytokines upregulate IL-6 in β-cells (48,49). Our findings indicate that increased LPS levels during high-fat diet feeding induces IL-6 in β-cells, which contributes to the autocrine stimulation of GLP-1 production by IL-6 in α-cells under normal conditions (50). Moreover, our results show that miR-155-5p increases IL-6 expression in β-cells by targeting Mafb that acts as a repressor of IL-6 gene transcription. Taken together, our data indicate that hyperlipidemia-induced miR-155-5p expression in β-cells reduces β-to-α-cell reprogramming through suppression of MafB. Both GLP-1 and glucagon are processed from the proglucagon precursor through the PC1/3 and PC2, respectively (6,8). Although PC1/3 expression in α-cells is low under normal conditions, lipotoxic stress and glycemia upregulate Pcsk1 expression and GLP-1 production, which enhances insulin secretion (12,13,51,52). Our data indicate that miR-155-5p increases Pcsk1 expression in α-cells by IL-6 secreted from β-cells and thereby shifts proglucagon processing from glucagon to GLP-1 production.

In addition to reduced intraislet GLP-1 expression, we found that Mir155 knockout decreased plasma GLP-1 levels in Ldlr^−/− mice. Although postprandial increases of circulating GLP-1 levels are because of its secretion by intestinal L cells, the source of fasting plasma GLP-1 is unclear. Insulin can trigger GLP-1 secretion from L cells, and plasma GLP-1 levels are elevated in hyperinsulinemic mice (53). Hence, reduced fasting insulin levels in Mir155^−/−Ldlr^−/− mice may decrease GLP-1 secretion from L cells and lower basal GLP-1 plasma levels. Notably, GLP-1 receptor agonists and overexpression of GLP-1 reduce obesity in humans and adipose tissue inflammation in mice, respectively (54,55). Moreover, treatment with GLP-1 receptor agonists improves obesity-related dyslipidemia, probably by inhibiting hepatic VLDL production (54,56). Therefore, reduced GLP-1 plasma levels may contribute to adipose tissue inflammation, obesity progression, and dyslipidemia in Mir155^−/−Ldlr^−/− mice. Consequently, elevated LDL and VLDL levels can promote the progression of atherosclerosis in obese Mir155^−/−Ldlr^−/− mice. By contrast, Mir155 knockout in mice with normal lipoprotein levels did not affect glucose tolerance, presumably because of the low islet miR-155-5p expression level in these mice. Accordingly, Mir155 knockout did not affect obesity in Ldlr^+/+ mice; however, female Mir155 knockout mice were protected from obesity by increased adipose tissue browning and reduced inflammatory macrophage activation (30). Hence, the effect of miR-155-5p on obesity differs between mice with normal lipid levels and hyperlipidemia, likely because different cell types are affected.

In conclusion, our results indicate a protective role of oxLDL-associated LPS on β-cell function during hyperlipidemia by inducing miR-155-5p, which prevents the upregulation of MafB and β-to-α-cell reprogramming. Hence, upregulation of miR-155-5p represents a self-protective mechanism in the stress response of β-cells and improves the adaptation of β-cells to insulin resistance.

Acknowledgments. The authors thank Dr. Ingo Rustenbeck (University of Braunschweig, Braunschweig, Germany) for providing the MIN6 cells.

Funding. This work was supported by the German Federal Ministry of Education and Research (01KU1213A), the German Research Foundation as part of the Collaborative Research Center 1123 (B04), and the German Centre for Cardiovascular Research (MHA VD1.2).

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

Author Contributions. M.Z. performed in vitro experiments and mouse experiments, analyzed the results, performed statistical analysis, and wrote the manuscript. Y.W. assisted in the analysis of the data and discussed and interpreted the results from the study. C.G. performed RNA extraction, qPCR assays, and luciferase reporter assays. K.A. and J.C.C. assisted in immunostainings and animal experiments for high-fat diet study. M.H. assisted in the cell sorting and flow cytometric analyses. J.M. and M.L. provided GLUTag cells and assisted and advised on GLUTag cell culture and GLP-1 secretion assay. E.K. discussed and revised the manuscript. A.S. designed the project, supervised the experiments, and wrote and revised the manuscript. A.S. and M.Z. are the guarantors of this work and, as such, had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented orally at the 52nd Annual Meeting of the European Association for the Study of Diabetes, Munich, Germany, 12–16 September 2016.