Hyperlipidemia-Induced Microrna-155-5P Improves Β-Cell Function by Targeting Mafb

Page 1 of 74 Diabetes

Hyperlipidemiainduced microRNA1555p improves βcell function by targeting Mafb

Short title: microRNA1555p improves βcell adaptation

Mengyu Zhu1, Yuanyuan Wei1, 2, Claudia Geißler1, Kathrin Abschlag1, Judit Corbalán

Campos1, Michael Hristov1, Julia Möllmann3, Michael Lehrke3, Ela Karshovska1, 2,

Andreas Schober1, 2

1Institute for Cardiovascular Prevention, LudwigMaximiliansUniversity Munich, Munich,

Germany

2DZHK (German Centre for Cardiovascular Research), partner site Munich Heart Alliance,

Munich, Germany

3Department of Internal Medicine I; University Hospital Aachen, Germany

Word count: 4186

Tables and Figures in main text: 1 Table and 7 Figures

Corresponding author: Andreas Schober, MD

Institute for Cardiovascular Prevention

LudwigMaximiliansUniversity Munich

Pettenkoferstrasse 9b, 80336 Munich, Germany

Tel: 4989440055151; Fax: 4989440054740

Email: [email protected]

Diabetes Publish Ahead of Print, published online September 29, 2017 Diabetes Page 2 of 74

ABSTRACT

A highfat diet increases bacterial lipopolysaccharide (LPS) in the circulation, and thereby stimulates glucagonlike peptide1 (GLP1)mediated insulin secretion by upregulating interleukin (IL)6. Although microRNA1555p (miR1555p), which increases IL6 expression, is upregulated by LPS and hyperlipidemia, and patients with familial hypercholesterolemia less frequently develop diabetes, the role of miR1555p in the islet stress response to hyperlipidemia is unclear. Here, we demonstrate that hyperlipidemia associated endotoxemia upregulates miR1555p in murine pancreatic βcells, which improved glucose metabolism and the adaptation of βcells to obesityinduced insulin resistance. This effect of miR1555p is due to suppression of vmaf musculoaponeurotic fibrosarcoma oncogene family, protein B (Mafb), which promotes βcell function through IL

6induced GLP1 production in αcells. Moreover, reduced GLP1 levels are associated with increased obesity progression, dyslipidemia, and atherosclerosis in hyperlipidemic Mir155 knockout mice. Hence, induction of miR1555p expression in βcells by hyperlipidemia associated endotoxemia improves the adaptation of βcells to insulin resistance and represents a protective mechanism in the islet stress response.

2 Page 3 of 74 Diabetes

Failure of pancreatic βcells 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 βcells can maintain glucose homeostasis

in insulin resistant states by increasing circulating insulin levels, chronically elevated insulin

secretion may result in exhaustion of βcell function due to apoptosis or dedifferentiation (4;

5). Obesity and T2D may promote βcell failure by decreasing the secretion of glucagonlike

peptide1 (GLP1), which enhances insulin secretion and βcell function, from intestinal L

cells (611). In addition, pancreatic αcells can be a source of GLP1, for instance, in response

to interleukin (IL) 6mediated upregulation of proprotein convertase 1 (PC1/3, encoded by

Pcsk1 gene), and thereby improve βcell function during obesity (7; 12; 13).

Lipopolysaccharide (LPS), which leaks into the circulation after a highfat meal due to

increased intestinal permeability (14; 15), also promotes insulin secretion by upregulating

GLP1 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 due to the elevated lipoprotein

levels (19; 20). However, chronically elevated LPS levels during highfat 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 (miRNA)

miR1555p, which is preferentially upregulated upon Tolllike receptor (TLR) 4 activation

(22; 23). Moreover, hyperlipidemia induces miR1555p 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 (2328).

In adipocytes, inflammatory cytokines, such as tumor necrosis factor (TNFα), upregulate

miR1555p expression, which may contribute to obesity progression in female mice by

3 Diabetes Page 4 of 74

limiting brown adipose tissue differentiation (29; 30). The role of miR1555p, however, in obesity and glucose homeostasis during hyperlipidemiaassociated endotoxemia is unclear.

We found that endotoxemia induces miR1555p expression in pancreatic βcells, which increases insulin secretion by targeting vmaf musculoaponeurotic fibrosarcoma oncogene family, protein B (Mafb) in hyperlipidemic mice. MafB represses IL6 expression in βcells and thereby inhibits intraislet GLP1 production. Through this mechanism, miR1555p improved the adaptation of βcells to hyperlipidemic stress and the compensation for obesity induced insulin resistance, and likely limited the progression of obesity and atherosclerosis.

4 Page 5 of 74 Diabetes

RESEARCH DESIGN AND METHODS

For further details, refer to Supplementary Experimental Procedures in Supplementary Data.

Animals. Mir155–/– mice were crossed with Ldlr–/– or Apoe–/– mice (all on a C57BL/6J

background, Jackson Laboratory) to obtain Mir155–/–Ldlr–/– mice and Mir155–/–Apoe–/– mice.

Mir155–/–Ldlr–/– mice and Mir155+/+Ldlr–/– mice at 1012 wks of age were fed a diabetogenic

diet supplemented with cholesterol (DDC; 35.5% fat, 36.3% carbohydrates with 0.15% w/w

total cholesterol, ssniff Spezialdiäten GmbH) or a normal diet (3.3% fat, ssniff Spezialdiäten).

Isolation of pancreatic islets. Murine pancreatic islets were isolated by collagenase digestion

and density gradient centrifugation. Briefly, collagenase P solution (4 mL, 1 mg/mL, Roche

Diagnostics GmbH) 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, SigmaAldrich).

Cell culture and transfection. MIN6 cells (kindly provided by Prof. Ingo Rustenbeck,

University of Braunschweig, Germany), human islets (Pelo Biotech) and GLUTag cells were

transfected with LNAmiR1555p inhibitors (50 nM; Exiqon), miR1555p mimics (15 nM;

Thermo Fisher Scientific), 155/Mafb target site blockers (50 nM; Exiqon), or scrambled

controls using Lipofectamine2000 (Thermo Fisher Scientific).

MicroRNA target identification and quantification system. MIN6 cells and human islets

were cotransfected with miR1555p mimics and the pMirTrap vector using the XfectTM

microRNA transfection reagent in combination with Xfect Polymer (all from Clontech). The

pMirTrap vector expresses a DYKDDDDKtagged GW182 protein. Cell lysates were

incubated with antiDYKDDDDK beads (Clontech) and RNA was isolated from input and

immunoprecipitated samples, and analysed by qPCR. Fold enrichment of the target genes in

the GW182immunoprecipitates was normalized to the enrichment of Gapdh.

5 Diabetes Page 6 of 74

In vivo TSB treatment. 10weekold Ldlr–/– mice fed a normal diet were injected intravenously via the tail vein with 155/Mafb TSBs or control TSBs (each 0.4 mg/20 g per injection; miRCURY LNATM Target Site Blocker, in vivo use; Exiqon), as described in

Supplementary Experimental Procedures.

Statistical analysis. Data represent the mean ± SEM. Statistical analysis of microarray data was performed by a modified ttest using GeneSpring software (GX13, Agilent

Technologies). Student’s ttests and oneway ANOVAs followed by the NewmanKeuls post hoc test were used for statistical comparisons between groups using Prism 6 software

(GraphPad). The variance is similar between the groups that are being statistically compared.

P < 0.05 was considered statistically significant.

6 Page 7 of 74 Diabetes

RESULTS

Deletion of Mir155 deteriorates metabolic disease in Ldlr–/– mice.

To study the effect of miR1555p on atherosclerosis in the context of obesity and T2D, we

deleted the miR1555p coding gene in hyperlipidemic lowdensity lipoprotein receptor

knockout (Ldlr–/–) mice that develop atherosclerosis, obesity and diabetes after a cholesterol

enricheddiabetogenic diet (DDC) feeding (31). After a 24wks 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 24wks DDC feeding period (Supplementary Fig. 1B). In

Mir155–/–Ldlr–/– mice, the cholesterol level was increased in the VLDL and LDL lipoprotein

fraction and reduced in the HDL fraction (Fig. 1B). While Mir155–/–Ldlr–/– mice and

Mir155+/+Ldlr–/– mice gained similar body weight in the first 20 wks of DDC feeding, the

body weight of Mir155–/–Ldlr–/– mice increased more than that of Mir155+/+Ldlr–/– mice

during the last 4 wks of the 24wks 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 downregulated

and upregulated, respectively, in the eWAT of Mir155–/–Ldlr–/– mice (Supplementary Fig.

1C). The expression of the proinflammatory macrophagerelated gene nitric oxide synthase 2

(Nos2) and the antiinflammatory macrophage marker mannose receptor, C type 1 (Mrc1) was

not different between the groups (Supplementary Fig. 1C). Deletion of Mir155 did not alter

7 Diabetes Page 8 of 74

interleukin 6 (Il6) mRNA expression, but increased tumor necrosis factor (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 wks, Fig. 1G), indicating that the hyperglycemia in Mir155–/–Ldlr–/– mice is not due to the increased weight gain. These findings indicate that miR1555p improves glucose homeostasis in hyperlipidemic mice and thereby limits obesity and atherosclerosis.

Mir155 knockout inhibits insulin production in hyperlipidemic mice.

To investigate the mechanism by which miR1555p 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 normal diet (ND, 0 wks) and after the 24 wks DDC feeding period (Fig. 2A), indicating that loss of miR1555p compromises islet function. In islets from NDfed Mir155–/–Ldlr–/– mice, the percentage of insulinexpressing cells and the insulin content were reduced compared with Mir155+/+Ldlr–/– mice (Fig. 2B).

Conversely, the percentage of glucagonexpressing cells and the glucagon protein content were higher in islets from Mir155–/–Ldlr–/– mice (Fig. 2C).

Proglucagon is processed to GLP1 and glucagon by prohormone convertase (PC) 1/3

[encoded by the proprotein convertase subtilisin/kexin type (Pcsk1) gene] and PC2 (encoded by the Pcsk2 gene), respectively (6; 8). GLP1 can be generated locally in pancreatic αcells, and increases insulin and reduces glucagon secretion (6). Therefore, we studied the effect of

Mir155 knockout on GLP1 expression. The intraislet GLP1 protein content was reduced in

8 Page 9 of 74 Diabetes

NDfed Mir155–/–Ldlr–/– mice (Fig. 2D). Plasma GLP1 levels were also lower in Mir155–/–

Ldlr–/– mice than in Mir155+/+Ldlr–/– mice fed a ND (0 wks) or the DDC for 24 wks (Fig. 2E).

These effects were associated with decreased insulin (Ins) expression in islets (Fig. 2F) and in

βcells (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

βcell 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 Tcells

in islets were negligible in both groups of mice (Supplementary Fig. 2B).

In vitro, miR1555p 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, miR1555p mimic treatment increased the cellular insulin and

GLP1 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 αcells after treatment with miR1555p mimics compared with control mimics

(Supplementary Fig. 2E). In human islets, miR1555p mimic decreased the expression of

GCG in αcells and increased INS expression in βcells, whereas PCSK1 expression was up

regulated in αcells (Supplementary Fig. 2F). Conversely, miR1555p 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 GLP1 content, and

increased glucagon content (Supplementary Fig. 2D).

9 Diabetes Page 10 of 74

In addition, overexpression of miR1555p promoted GLP1 secretion from murine islets

(Fig. 2I). By contrast, treatment of an enteroendocrine Lcell line with miR1555p mimics did not affect GLP1 protein content and secretion, and GCG and PCSK1 mRNA expression

(Supplementary Fig. 2G and H). Hence, miR1555p promotes intraislet GLP1 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 apolipoprotein E knockout (Apoe–/–) mice (Supplementary Fig. 3B and C) fed a ND, whereas glucose tolerance was not affected by Mir155 knockout in normal lipidemic male NDfed Ldlr+/+ mice (Fig. 2K). Thus, miR1555p improved glucose homeostasis only under hyperlipidemic conditions.

Hyperlipidemiaassociated endotoxemia induces islet miR1555p expression.

Next, we studied the regulation of islet miR1555p expression by hyperlipidemia and LPS.

Feeding Ldlr–/– mice the DDC for 24 wks increased plasma endotoxin activity and islet miR

1555p expression compared to ND feeding (Fig. 3A). In 10–12 wksold, NDfed mice, knockout of Ldlr increased plasma cholesterol and triglyceride levels (Supplementary Fig.

4A), circulating endotoxin activity, and islet miR1555p expression (Fig. 3B and C). miR

1555p 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 (moxLDL) up regulated miR1555p expression in MIN6 cells compared with vehicle treatment (Fig. 3D).

LPS stimulation increased miR1555p expression in MIN6 cells (Fig. 3E) and in human islet cells (Fig. 3F). Moreover, mild oxidation increased the endotoxin activity in LDL (Fig. 3G).

Knockout of Ldlr in NDfed mice resulted in deposition of oxidized LDL in islets (Fig. 3H).

10 Page 11 of 74 Diabetes

Thus, enhanced endotoxin activity of oxidized LDL deposited in islets during hyperlipidemia

may induce miR1555p expression in βcells.

LPSinduced insulin release from islets following glucose stimulation was decreased in

islets from Mir155–/–Ldlr–/– mice (Fig. 3I). Treatment of Ldlr–/– mice with lowdose LPS up

regulated islet miR1555p expression (Fig. 3J), and increased insulin and GLP1 plasma

levels (Supplementary Fig. 4B). The glucose lowering effect of lowdose LPS following

intraperitoneal glucose injection in Ldlr–/– mice (Fig. 3K), was partially abolished by Mir155

knockout (Fig. 3K). Together these data suggest that hyperlipidemiainduced miR1555p

expression improves βcell adaptation to hyperlipidemiaassociated endotoxemia stress.

miR1555p promotes IL6 expression in βcells.

To determine how miR1555p regulates βcell function, we analyzed the effect of Mir155

knockout on islet gene expression by microarray analysis. In NDfed 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 per 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 βcell failure in mouse models of T2D (33), was increased in Mir155–/–Ldlr–/–

mice (Fig. 4B). Moreover, Glut2dependent pathways and pathways related to cyclic AMP,

GLP1, and glucosedependent insulinotropic polypeptide (GIP) signaling were inhibited,

suggesting impaired glucose uptake and insulin secretion (34).

11 Diabetes Page 12 of 74

Among the inflammatory pathways, IL6 receptor activation was reduced and signaling pathways downstream of the IL6 receptor, such as the JAK/STAT and ERK1/2 pathways, was inhibited in Mir155–/–Ldlr–/– mice (Fig. 4B). Accordingly, islet IL6 mRNA and protein expression, and the number of IL6producing βcells were reduced in Mir155–/–Ldlr–/– mice compared with Mir155+/+Ldlr–/– mice (Fig. 4C and D). Whereas Il6 expression was unchanged in sorted αcells, it was downregulated in βcells from in Mir155–/–Ldlr–/– mice compared with

Mir155+/+Ldlr–/– mice (Fig. 4E). Moreover, IL6 expression was upregulated by miR1555p mimic treatment in human βcells, but was not affected in αcells compared with control mimic (Fig. 4F). In vitro, gainandlossoffunction experiments demonstrated that miR155

5p upregulates IL6 mRNA and protein expression in MIN6 cells (Supplementary Fig. 5A and B). Inhibition of IL6 secreted from MIN6 cells using a blocking IL6 antibody reduced

Ins and Pcsk1 expression and increased Pcsk2 expression (Fig. 4G). In addition, treatment of sorted human αcells with conditioned medium from miR1555p mimic treated human β cells enhanced GLP1 secretion and the cellular GLP1 content (Supplementary Fig. 5C).

Taken together, these results indicate that miR1555p in βcells stimulates the expression and secretion of IL6, which in turn increases GLP1 production by upregulating Pcsk1 expression in αcells. miR1555p upregulates IL6 by targeting Mafb.

To determine the targets that mediate the effect of miR1555p on IL6 expression in βcells, we screened the 3’untranslated region (3’UTR) of the genes upregulated in islets from

Mir155–/–Ldlr–/– mice for miR1555p 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 12like (Med12l), contained miR1555p binding sites (Table 1). The miR1555p 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

12 Page 13 of 74 Diabetes

poorly conserved sites were also found in humans, including the site in the AU RNA binding

protein/enoylcoenzyme A hydratase (Auh), stathminlike 2 (Stmn2) and Med12l mRNAs.

Upregulation of islet Auh, Mafb, Med12l, Sema5a and Stmn2 expression in Mir155–/–Ldlr–/–

mice was confirmed by realtime PCR (qPCR) (Fig. 5A). Treatment of MIN6 cells with miR

1555p 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 miRNAinduced silencing complex

(miRISC) using extracts from MIN6 cells and human islets overexpressing FLAGtagged

GW182 (35). Among the potential target genes, miR1555p mimic treatment most strongly

enriched Mafb in the miRISC of MIN6 cells and human islets. In contrast to Auh, Med12l and

Sema5a, Stmn2 was also enriched by miR1555p 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 αcells, upregulated Mafb expression in sorted β

cells (Fig. 5D and E). Moreover, miR1555p mimic treatment reduced MAFB expression in

human βcells, but not in αcells (Supplementary Fig. 6C).

The miR1555p binding site in the MAFB 3’UTR has been previously verified in Bcell

lymphoma cells (Supplementary Fig. 6D) (36). To test the function of this site, we designed

locked nucleic acid (LNA)modified oligonucleotides that selectively inhibit the interaction

between miR1555p and Mafb (155/Mafb target site blocker, 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 IL6 expression at the mRNA and protein level (Fig. 5G).

These data indicate that the effects of miR1555p on βcells are mainly mediated by the

targeting of Mafb.

To study how Mafb regulates IL6 expression, MIN6 cells were transfected with a

luciferase reporter vector containing the wildtype Il6 promoter or the Il6 promoter containing

13 Diabetes Page 14 of 74

mutations in the predicted Mafb binding sites Mafb1 and Mafb2 (Supplementary Fig. 6E).

Treatment with miR1555p inhibitors reduced the luciferase activity in cells expressing the wildtype 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 MafBmediated transcriptional repression of IL6 contributes to the effect of miR

1555p on βcell function.

Role of the miR1555pMafb interaction in glucose homeostasis in vivo.

To study whether the effect of hyperlipidemiainduced miR1555p in βcells on glucose homeostasis is mediated by the suppression of Mafb, Ldlr–/– mice were treated with 155/Mafb

TSBs or nontargeting, LNAmodified 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 TSBtreated mice (Fig. 6A). 155/Mafb

TSB treatment did not affect islet Auh, Med12l, Sema5a and Stmn2 expression levels

(Supplementary Fig. 7C). The percentage of MafBexpressing cells in islets was higher in

155/Mafb TSBtreated mice than in mice treated with control TSBs (Fig. 6B). Treatment with

155/Mafb TSBs increased Gcg mRNA expression and the percentage of αcells, and reduced

Pcsk1 and Il6 expression, and the percentage of βcells compared with control (Fig. 6C and

D). This effect in 155/Mafb TSBtreated mice was associated with reduced insulin and GLP1 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). This data indicates that hyperlipidemiainduced miR1555p expression improves βcell adaptation and maintains glucose hemostasis by suppressing Mafb.

14 Page 15 of 74 Diabetes

DISCUSSION

We found that hyperlipidemia and LPS upregulate miR1555p expression in βcells, which

improved glucose homeostasis by targeting MafB. In the absence of miR1555p, up

regulation of MafB by hyperlipidemia inhibits IL6 expression and thereby reduces IL6

mediated GLP1 production in αcells (Fig. 7). In obese mice, miR1555pinduced GLP1

production may limit atherosclerosis, dyslipidemia, and the progression of adiposity, and

improve the adaptation of βcells to insulin resistance.

Upregulation of miR1555p in response to LPS plays an essential role in inflammatory

macrophage activation (37). In addition, moxLDL promotes inflammatory activation and

miR1555p expression in macrophages in a TLR4dependent manner (26). In line with these

results, we found that LPS and moxLDL increase miR1555p expression in βcells. 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 due to altered interactions between lipids from LDL and LPS, suggesting

that LPS mediates the effect of moxLDL on miR1555p expression. Moreover, our finding

that the deposition of oxidized LDL in islets and endotoxemia were increased in Ldlr–/– mice

indicates that LPS contributes to the upregulation of miR1555p in βcells by

hyperlipidemia. Lowdose LPS improves insulin secretion by upregulating GLP1 production

(17) and knockout of Mir155 reduced the effect of LPS on glucose metabolism, suggesting

that βcell miR1555p contributes to LPSinduced insulin secretion. Accordingly, Mir155

knockout elevated plasma glucose levels in Ldlr–/– mice due to reduced insulin and increased

glucagon production in islets. The effect of miR1555p on islet function is likely mediated by

the upregulation of intraislet GLP1, which improves βcell function and inhibits glucagon

expression (6; 7).

15 Diabetes Page 16 of 74

The main mechanisms of βcell failure in T2D development involve dedifferentiation, apoptosis, and impaired regeneration of βcells (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 βcells. In adult rodent islets, MafA is only expressed in βcells and promotes insulin expression, whereas MafB is exclusively expressed in αcells and induces

Gcg transcription (4144). In pregnant or obese mice, however, MafB expression is up regulated in βcell (45). Moreover, derepression of MafB in the absence of the βcellspecific transcription factor pancreatic and duodenal homeobox 1 (Pdx1) leads to βtoαcell reprogramming, which may contribute to βcell failure in T2D (40; 43; 46). Notably, highfat diet feeding induced reexpression of MafB in βcells, suggesting that hyperlipidemia promotes βtoαcell conversion (45). Accordingly, our findings indicate that hyperlipidemia induced expression of miR1555p in βcells limits the upregulation of MafB and thereby improves βcell function, probably due to enhanced GLP1 production in αcells. In mouse models of obesity and diabetes, IL6 increases intraislet GLP1 expression in αcell by up regulating Pcsk1 expression (12). Notably, miR1555p mediates LPSinduced IL6 expression in macrophages (22; 47), and highfat diet feeding and inflammatory cytokines up regulate IL6 in βcells (48; 49). Our findings indicate that increased LPS levels during high fat diet feeding induces IL6 in βcells, which contributes to the autocrine stimulation of GLP

1 production by IL6 in αcells under normal conditions (50). Moreover, our results show that miR1555p increases IL6 expression in βcells by targeting Mafb that acts as a repressor of

IL6 gene transcription. Taken together, our data indicate that hyperlipidemiainduced miR

1555p expression in βcells reduces βtoαcell reprogramming through suppression of MafB.

Both GLP1 and glucagon are processed from the proglucagon precursor through the PC1/3 and PC2, respectively (6; 8). Although PC1/3 expression in αcells is low under normal conditions, lipotoxic stress and glycemia upregulate Pcsk1 expression and GLP1 production, which enhances insulin secretion (12; 13; 51; 52). Our data indicate that miR1555p

16 Page 17 of 74 Diabetes

increases Pcsk1 expression in acells by IL6 secreted from βcells and thereby shifts

proglucagon processing from glucagon to GLP1 production.

In addition to reduced intraislet GLP1 expression, we found that Mir155 knockout

decreased plasma GLP1 levels in Ldlr–/– mice. Although postprandial increases of circulating

GLP1 levels are due to its secretion by intestinal Lcells, the source of fasting plasma GLP1

is unclear. Insulin can trigger GLP1 secretion from Lcells, and plasma GLP1 levels are

elevated in hyperinsulinemic mice (53). Hence, reduced fasting insulin levels in Mir155–/–

Ldlr–/– mice may decrease GLP1 secretion from Lcells and lower basal GLP1 plasma levels.

Notably, GLP1 receptor agonists and overexpression of GLP1 reduce obesity in humans and

adipose tissue inflammation in mice, respectively (54; 55). Moreover, treatment with GLP1

receptor agonists improves obesityrelated dyslipidemia, probably by inhibiting hepatic

VLDL production (54; 56). Therefore, reduced GLP1 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 due to the low islet miR1555p

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 miR1555p 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 oxidized LDLassociated LPS on β

cell function during hyperlipidemia by inducing miR1555p, which prevents the up

regulation of MafB and βtoαcell reprogramming. Hence, upregulation of miR1555p

17 Diabetes Page 18 of 74

represents a selfprotective mechanism in the stress response of βcells and improves the adaptation of βcells to insulin resistance.

18 Page 19 of 74 Diabetes

Acknowledgments. The authors thank Dr. Ingo Rustenbeck (University of Braunschweig,

Germany) for providing the MIN6 cells.

Funding. This work was funded by the German Federal Ministry of Education and Research

(01KU1213A), by the the German Research Foundation (DFG) as part of the Collaborative

Research Center 1123 (B04), and by 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, 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, assisted and

advised on GLUTag cell culture and GLP1 secretion assay. E.K. discussed and revised the

manuscript. A.S. designed the project and supervised the experiments, wrote, and revised the

manuscript. A.S. and M.Z. are the guarantors 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.

19 Diabetes Page 20 of 74

REFERENCES

1. Hu FB: Globalization of diabetes: the role of diet, lifestyle, and genes. Diabetes Care 2011;34:12491257 2. Prentki M, Nolan CJ: Islet beta cell failure in type 2 diabetes. J Clin Invest 2006;116:1802 1812 3. McNelis JC, Olefsky JM: Macrophages, immunity, and metabolic disease. Immunity 2014;41:3648 4. Talchai C, Xuan S, Lin HV, Sussel L, Accili D: Pancreatic beta cell dedifferentiation as a mechanism of diabetic beta cell failure. Cell 2012;150:12231234 5. Butler AE, Janson J, BonnerWeir S, Ritzel R, Rizza RA, Butler PC: Betacell deficit and increased betacell apoptosis in humans with type 2 diabetes. Diabetes 2003;52:102110 6. Cho YM, Fujita Y, Kieffer TJ: Glucagonlike peptide1: glucose homeostasis and beyond. Annu Rev Physiol 2014;76:535559 7. Campbell JE, Drucker DJ: Pharmacology, physiology, and mechanisms of incretin hormone action. Cell Metab 2013;17:819837 8. Sandoval DA, D'Alessio DA: Physiology of proglucagon peptides: role of glucagon and GLP1 in health and disease. Physiol Rev 2015;95:513548 9. Lugari R, Dei Cas A, Ugolotti D, Finardi L, Barilli AL, Ognibene C, Luciani A, Zandomeneghi R, Gnudi A: Evidence for early impairment of glucagonlike peptide 1 induced insulin secretion in human type 2 (non insulindependent) diabetes. Horm Metab Res 2002;34:150154 10. Tuduri E, Lopez M, Dieguez C, Nadal A, Nogueiras R: Glucagonlike peptide 1 analogs and their effects on pancreatic islets. Trends Endocrinol Metab 2016;27:304318 11. Muscelli E, Mari A, Casolaro A, Camastra S, Seghieri G, Gastaldelli A, Holst JJ, Ferrannini E: Separate impact of obesity and glucose tolerance on the incretin effect in normal subjects and type 2 diabetic patients. Diabetes 2008;57:13401348 12. Ellingsgaard H, Hauselmann I, Schuler B, Habib AM, Baggio LL, Meier DT, Eppler E, Bouzakri K, Wueest S, Muller YD, Hansen AM, Reinecke M, Konrad D, Gassmann M, Reimann F, Halban PA, Gromada J, Drucker DJ, Gribble FM, Ehses JA, Donath MY: Interleukin6 enhances insulin secretion by increasing glucagonlike peptide1 secretion from L cells and alpha cells. Nat Med 2011;17:14811489 13. Kilimnik G, Kim A, Steiner DF, Friedman TC, Hara M: Intraislet production of GLP1 by activation of prohormone convertase 1/3 in pancreatic alphacells in mouse models of beta cell regeneration. Islets 2010;2:149155 14. Graham C, Mullen A, Whelan K: Obesity and the gastrointestinal microbiota: a review of associations and mechanisms. Nutr Rev 2015;73:376385 15. Winer DA, Luck H, Tsai S, Winer S: The intestinal immune system in obesity and insulin resistance. Cell Metab 2016;23:413426 16. Kahles F, Meyer C, Mollmann J, Diebold S, Findeisen HM, Lebherz C, Trautwein C, Koch A, Tacke F, Marx N, Lehrke M: GLP1 secretion is increased by inflammatory stimuli in an IL6dependent manner, leading to hyperinsulinemia and blood glucose lowering. Diabetes 2014;63:32213229

20 Page 21 of 74 Diabetes

17. Nguyen AT, Mandard S, Dray C, Deckert V, Valet P, Besnard P, Drucker DJ, Lagrost L, Grober J: Lipopolysaccharidesmediated increase in glucosestimulated insulin secretion: involvement of the GLP1 pathway. Diabetes 2014;63:471482 18. Vreugdenhil AC, Snoek AM, van 't Veer C, Greve JW, Buurman WA: LPSbinding protein circulates in association with apoBcontaining lipoproteins and enhances endotoxin LDL/VLDL interaction. J Clin Invest 2001;107:225234 19. Schwartz Y, Dushkin MI: Endotoxinlipoprotein complex formation as a factor in atherogenesis: associations with hyperlipidemia and with lecithin:cholesterol acyltransferase activity. Biochemistry (Mosc) 2002;67:747752 20. Besseling J, Kastelein JJ, Defesche JC, Hutten BA, Hovingh GK: Association between familial hypercholesterolemia and prevalence of type 2 diabetes mellitus. Jama 2015;313:10291036 21. Cani PD, Amar J, Iglesias MA, Poggi M, Knauf C, Bastelica D, Neyrinck AM, Fava F, Tuohy KM, Chabo C, Waget A, Delmee E, Cousin B, Sulpice T, Chamontin B, Ferrieres J, Tanti JF, Gibson GR, Casteilla L, Delzenne NM, Alessi MC, Burcelin R: Metabolic endotoxemia initiates obesity and insulin resistance. Diabetes 2007;56:17611772 22. Androulidaki A, Iliopoulos D, Arranz A, Doxaki C, Schworer S, Zacharioudaki V, Margioris AN, Tsichlis PN, Tsatsanis C: The kinase Akt1 controls macrophage response to lipopolysaccharide by regulating microRNAs. Immunity 2009;31:220231 23. NazariJahantigh M, Wei Y, Noels H, Akhtar S, Zhou Z, Koenen RR, Heyll K, Gremse F, Kiessling F, Grommes J, Weber C, Schober A: MicroRNA155 promotes atherosclerosis by repressing Bcl6 in macrophages. J Clin Invest 2012;122:41904202 24. Tian FJ, An LN, Wang GK, Zhu JQ, Li Q, Zhang YY, Zeng A, Zou J, Zhu RF, Han XS, Shen N, Yang HT, Zhao XX, Huang S, Qin YW, Jing Q: Elevated microRNA155 promotes foam cell formation by targeting HBP1 in atherogenesis. Cardiovasc Res 2014;103:100110 25. Wei Y, Schober A: MicroRNA regulation of macrophages in human pathologies. Cell Mol Life Sci 2016;73:34733495 26. Du F, Yu F, Wang Y, Hui Y, Carnevale K, Fu M, Lu H, Fan D: MicroRNA155 deficiency results in decreased macrophage inflammation and attenuated atherogenesis in apolipoprotein Edeficient mice. Arterioscler Thromb Vasc Biol 2014;34:759767 27. Schober A, Weber C: Mechanisms of microRNAs in atherosclerosis. Annu Rev Pathol 2016;11:583616 28. Wei Y, Zhu M, CorbalanCampos J, Heyll K, Weber C, Schober A: Regulation of Csf1r and Bcl6 in macrophages mediates the stagespecific effects of microRNA155 on atherosclerosis. Arterioscler Thromb Vasc Biol 2015;35:796803 29. Liu S, Yang Y, Wu J: TNFalphainduced upregulation of miR155 inhibits adipogenesis by downregulating early adipogenic transcription factors. Biochem Biophys Res Commun 2011;414:618624 30. Gaudet AD, Fonken LK, Gushchina LV, Aubrecht TG, Maurya SK, Periasamy M, Nelson RJ, Popovich PG: miR155 deletion in female mice prevents dietinduced obesity. Sci Rep 2016;6:22862 31. Subramanian S, Han CY, Chiba T, McMillen TS, Wang SA, Haw A, 3rd, Kirk EA, O'Brien KD, Chait A: Dietary cholesterol worsens adipose tissue macrophage accumulation and atherosclerosis in obese LDL receptordeficient mice. Arterioscler Thromb Vasc Biol 2008;28:685691

21 Diabetes Page 22 of 74

32. Rane SG, Dubus P, Mettus RV, Galbreath EJ, Boden G, Reddy EP, Barbacid M: Loss of Cdk4 expression causes insulindeficient diabetes and Cdk4 activation results in betaislet cell hyperplasia. Nat Genet 1999;22:4452 33. Wang L, Liu Y, Yan Lu S, Nguyen KT, Schroer SA, Suzuki A, Mak TW, Gaisano H, Woo M: Deletion of Pten in pancreatic betacells protects against deficient betacell mass and function in mouse models of type 2 diabetes. Diabetes 2010;59:31173126 34. Ammala C, Ashcroft FM, Rorsman P: Calciumindependent potentiation of insulin release by cyclic AMP in single betacells. Nature 1993;363:356358 35. Cambronne XA, Shen R, Auer PL, Goodman RH: Capturing microRNA targets using an RNAinduced silencing complex (RISC)trap approach. Proc Natl Acad Sci USA 2012;109:2047320478 36. Zhang Y, Roccaro AM, Rombaoa C, Flores L, Obad S, Fernandes SM, Sacco A, Liu Y, Ngo H, Quang P, Azab AK, Azab F, Maiso P, Reagan M, Brown JR, Thai TH, Kauppinen S, Ghobrial IM: LNAmediated antimiR155 silencing in lowgrade Bcell lymphomas. Blood 2012;120:16781686 37. Jablonski KA, Gaudet AD, Amici SA, Popovich PG, GueraudeArellano M: Control of the inflammatory macrophage transcriptional signature by miR155. PLoS One 2016;11:e0159724 38. Van Lenten BJ, Fogelman AM, Haberland ME, Edwards PA: The role of lipoproteins and receptormediated endocytosis in the transport of bacterial lipopolysaccharide. Proc Natl Acad Sci USA 1986;83:27042708 39. Topchiy E, Cirstea M, Kong HJ, Boyd JH, Wang Y, Russell JA, Walley KR: Lipopolysaccharide is cleared from the circulation by hepatocytes via the low density lipoprotein receptor. PLoS One 2016;11:e0155030 40. Spijker HS, Song H, Ellenbroek JH, Roefs MM, Engelse MA, Bos E, Koster AJ, Rabelink TJ, Hansen BC, Clark A, Carlotti F, de Koning EJ: Loss of betacell identity occurs in type 2 diabetes and is associated with islet amyloid deposits. Diabetes 2015;64:29282938 41. Artner I, Le Lay J, Hang Y, Elghazi L, Schisler JC, Henderson E, SosaPineda B, Stein R: MafB: an activator of the glucagon gene expressed in developing islet alpha and betacells. Diabetes 2006;55:297304 42. Artner I, Hang Y, Mazur M, Yamamoto T, Guo M, Lindner J, Magnuson MA, Stein R: MafA and MafB regulate genes critical to betacells in a unique temporal manner. Diabetes 2010;59:25302539 43. Gao T, McKenna B, Li C, Reichert M, Nguyen J, Singh T, Yang C, Pannikar A, Doliba N, Zhang T, Stoffers DA, Edlund H, Matschinsky F, Stein R, Stanger BZ: Pdx1 maintains beta cell identity and function by repressing an alpha cell program. Cell Metab 2014;19:259271 44. Nishimura W, Kondo T, Salameh T, El Khattabi I, Dodge R, BonnerWeir S, Sharma A: A switch from MafB to MafA expression accompanies differentiation to pancreatic betacells. Dev Biol 2006;293:526539 45. Lu J, Hamze Z, Bonnavion R, Herath N, Pouponnot C, Assade F, Fontaniere S, Bertolino P, CordierBussat M, Zhang CX: Reexpression of oncoprotein MafB in proliferative beta cells and Men1 insulinomas in mouse. Oncogene 2012;31:36473654 46. Spijker HS, Ravelli RB, MommaasKienhuis AM, van Apeldoorn AA, Engelse MA, Zaldumbide A, BonnerWeir S, Rabelink TJ, Hoeben RC, Clevers H, Mummery CL, Carlotti

22 Page 23 of 74 Diabetes

F, de Koning EJ: Conversion of mature human betacells into glucagonproducing alphacells. Diabetes 2013;62:24712480 47. Wei Y, NazariJahantigh M, Chan L, Zhu M, Heyll K, CorbalanCampos J, Hartmann P, Thiemann A, Weber C, Schober A: The microRNA3425p fosters inflammatory macrophage activation through an Akt1 and microRNA155dependent pathway during atherosclerosis. Circulation 2013;127:16091619 48. Ehses JA, Perren A, Eppler E, Ribaux P, Pospisilik JA, MaorCahn R, Gueripel X, Ellingsgaard H, Schneider MK, Biollaz G, Fontana A, Reinecke M, HomoDelarche F, Donath MY: Increased number of isletassociated macrophages in type 2 diabetes. Diabetes 2007;56:23562370 49. Campbell IL, Cutri A, Wilson A, Harrison LC: Evidence for IL6 production by and effects on the pancreatic betacell. J Immunol 1989;143:11881191 50. Timper K, Dalmas E, Dror E, Rutti S, Thienel C, Sauter NS, Bouzakri K, Bedat B, Pattou F, KerrConte J, BoniSchnetzler M, Donath MY: Glucosedependent insulinotropic peptide stimulates glucagonlike peptide 1 production by pancreatic islets via interleukin 6, produced by alpha cells. Gastroenterology 2016;151:165179 51. Wideman RD, Yu IL, Webber TD, Verchere CB, Johnson JD, Cheung AT, Kieffer TJ: Improving function and survival of pancreatic islets by endogenous production of glucagon like peptide 1 (GLP1). Proc Natl Acad Sci U S A 2006;103:1346813473 52. Nie Y, Nakashima M, Brubaker PL, Li QL, Perfetti R, Jansen E, Zambre Y, Pipeleers D, Friedman TC: Regulation of pancreatic PC1 and PC2 associated with increased glucagonlike peptide 1 in diabetic rats. J Clin Invest 2000;105:955965 53. Lim GE, Huang GJ, Flora N, LeRoith D, Rhodes CJ, Brubaker PL: Insulin regulates glucagonlike peptide1 secretion from the enteroendocrine L cell. Endocrinology 2009;150:580591 54. PiSunyer X, Astrup A, Fujioka K, Greenway F, Halpern A, Krempf M, Lau DC, le Roux CW, Violante Ortiz R, Jensen CB, Wilding JP, Obesity S, Prediabetes NNSG: A randomized, controlled trial of 3.0 mg of liraglutide in weight management. N Engl J Med 2015;373:1122 55. Lee YS, Park MS, Choung JS, Kim SS, Oh HH, Choi CS, Ha SY, Kang Y, Kim Y, Jun HS: Glucagonlike peptide1 inhibits adipose tissue macrophage infiltration and inflammation in an obese mouse model of diabetes. Diabetologia 2012;55:24562468 56. Taher J, Baker CL, Cuizon C, Masoudpour H, Zhang R, Farr S, Naples M, Bourdon C, Pausova Z, Adeli K: GLP1 receptor agonism ameliorates hepatic VLDL overproduction and de novo lipogenesis in insulin resistance. Mol Metab 2014;3:823833

23 Diabetes Page 24 of 74

Table 1. Putative miR1555p target genes in pancreatic islets.

Gene PCT Conservation Mafb 0.39 conserved Sema5a 0.3 conserved Med12l 0.15 poorly conserved* Stmn2 0.13 poorly conserved* Auh 0.12 poorly conserved* F13a1 < 0.1 poorly conserved# Dhfr < 0.1 poorly conserved# Klhl42 < 0.1 poorly conserved# Ppp1r9a < 0.1 poorly conserved# Phf21a < 0.1 poorly conserved# Rab3c < 0.1 poorly conserved# Nedd4l < 0.1 poorly conserved# Homez < 0.1 rodentspecific Nrp1 < 0.1 rodentspecific Pde4d < 0.1 rodentspecific Zkscan3 < 0.1 rodentspecific Zfp14 < 0.1 rodentspecific Bik < 0.1 mousespecific Camkk2 < 0.1 mousespecific Clec1a < 0.1 mousespecific Gpr179 < 0.1 mousespecific Htra3 < 0.1 mousespecific Myct1 < 0.1 mousespecific Scai < 0.1 mousespecific Zfp111 < 0.1 mousespecific Zfp937 < 0.1 mousespecific Zscan20 < 0.1 mousespecific

Among the genes significantly upregulated (≥1.5fold and p<0.05) in pancreatic islets of Mir155–/– Ldlr–/– mice compared to Mir155+/+Ldlr–/– mice (as determined by global gene expression analysis), miR1555p targets and the conservation of the putative miR1555p binding sites across species were predicted by Targetscan software (http://www.targetscan.org/).The probability of conserved targeting # was indicated by PCT. *: Conserved between human and rodent; : Seed sequences are different in mouse and human.

24 Page 25 of 74 Diabetes

Figure legends for primary figures

Figure 1─ Effects of Mir155 knockout on atherosclerosis, obesity, and metabolism in Ldlr−/− mice. Male Mir155+/+Ldlr−/− and Mir155−/−Ldlr−/− mice were fed a DDC for 24 wks. A: Lesion and necrotic core areas in aortic roots in mice after the 24wk DDC feeding period (n = 10 mice per group). Scale bars: 100 µm. B: Cholesterol levels in VLDL, LDL and HDL fractions from mice after the 24wk DDC feeding period analyzed by highperformance liquid chromatography (n = 8 mice per group). C: Body weight gain of mice during the 24wk DDC feeding period (n = 10 mice per group). D and E: Quantitation of epididymal white adipose tissue (eWAT) weight (D) and adipocyte size in the eWAT (E) from mice after the 24wk DDC feeding period (n = 10 mice per group). Scale bars: 100 µm. F: Macrophage accumulation in eWAT from mice after the 24wk DDC feeding period assessed by Mac2 immunostaining (n = 9 mice per group). The nuclei were counterstained with DAPI. Scale bars: 50 µm. G: Fasting blood glucose (FBG) concentrations in mice during the 24wk DDC feeding period (n = 10 mice per group). Data are represented as mean ± SEM. *P < 0.05, **P < 0.01 and ***P < 0.001.

Figure 2─ Effect of miR1555p on pancreatic islets. A: Fasting insulin and glucagon plasma concentrations in mice fed a normal diet (ND, 0 wks DDC) and after the 24wks DDC feeding period (n = 6 mice per group). B: Quantitation of the percentage of insulinexpressing βcells per total islet cells (n = 15 or 18 mice per group) and islet insulin concentrations (n = 6 per group) by immunostaining and Luminex multiplex analysis, respectively, in 10–12 wks old mice fed a normal diet. The nuclei were counterstained with DAPI. Scale bars: 50 µm. C: Quantitation of the percentage of glucagonexpressing αcells per total islet cells (n = 15 or 18 mice per group) and islet glucagon concentrations (n = 6 per group) by immunostaining and Luminex multiplex analysis, respectively, in 10–12 wks old mice fed a normal diet. The nuclei were counterstained with DAPI. Scale bars: 50 µm. D: Islet GLP1 protein concentration in 10–12 wks old mice fed a normal diet determined by Luminex multiplex analysis (n = 6 per group). E: Fasting GLP1 plasma concentrations in mice fed a normal diet (0 wks DDC) and after the 24wks DDC feeding period (n = 6 mice per group). F: Islet insulin (Ins), glucagon (Gcg), and proprotein convertase subtilisin/kexin type (Pcsk) 1 and Pcsk2 mRNA expression levels in 10–12 wks old mice fed an ND determined by realtime PCR (qPCR, n = 6 or 8 per group). G: Quantitation of gene expression by qPCR in α and β cells sorted from islets of NDfed Mir155−/−Ldlr−/− mice and Mir155+/+Ldlr−/− mice (10–12 wks of age) (n = 3–4 per group). H: Ins, Gcg, Pcsk1 and Pcsk2 mRNA expression levels (n =

25 Diabetes Page 26 of 74

4 or 6 per group) determined by qPCR, and insulin, glucagon, and GLP1 protein levels (n = 4 per group) measured by Luminex multiplex analysis in MIN6 cells treated with miR1555p mimics or nontargeting oligonucleotides (control mimics). I: GLP1 secretion from KCl stimulated islets isolated from NDfed Mir155+/+Ldlr−/− mice (10–12 wks of age) after treatment with miR1555p mimics or control mimics (n = 4 per group). J: Intraperitoneal glucose tolerance test (IPGTT) in male Mir155−/−Ldlr−/− mice and Mir155+/+Ldlr−/− mice at 10–12 wks of age fed an ND (n = 10 mice per group). K: IPGTT in male Mir155+/+Ldlr+/+ mice (n = 6 mice per group) and Mir155−/−Ldlr+/+ mice (n = 8 mice per group) at 10–12 wks of age fed an ND. Data are represented as mean ± SEM. *P < 0.05, **P < 0.01 and ***P < 0.001.

Figure 3─ miR1555p mediates the effects of LPS and hyperlipidemia on glucose homeostasis. A: Serum endotoxin levels in mice fed an ND or the DDC for 24 wks (left, n = 6 per group). Quantitation of miR1555p expression in lasermicrodissected islets from mice fed the ND or the DDC for 24 wks (right, n = 4 per group). **P < 0.01 and ***P < 0.001. B: Serum endotoxin levels in 10–12 wks old mice fed an ND determined by LAL test (left, n = 6 mice per group). Quantitation of miR1555p expression by qPCR in isolated murine islets from NDfed mice (1012 wks of age) (right, n = 6 mice per group). *P < 0.05. C: Localization of miR1555p expression in islets from NDfed Ldlr+/+ and Ldlr−/− mice (10–12 wks of age) determined by in situ PCR and glucagon immunostaining. The nuclei were counterstained with DAPI. Scale bars: 50 µm. D: Quantitation of miR1555p expression in MIN6 cells treated with native LDL (nLDL), mildly oxidized LDL (moxLDL) or vehicle for 6 h (n = 5–6 per group). *P < 0.05. E and F: Quantitation of miR1555p expression in MIN6 cells (E) and human islets (F) treated with LPS or vehicle for 6 h (n = 5–6 per group). *P < 0.05. G: Endotoxin activity in nLDL or moxLDL determined by LAL test (n = 3 per group). ***P < 0.001. H: oxLDL immunostaining in islets from NDfed mice (10–12 wks of age). The nuclei were counterstained with DAPI. Scale bars: 50 µm. I: Glucoseinduced insulin secretion from islets isolated from NDfed Mir155−/−Ldlr−/− mice and Mir155+/+Ldlr−/− mice (10–12 wks of age) with or without LPS (50 ng/ml) stimulation. Insulin concentrations in the medium were measured by ELISA (n = 4 per group). *P < 0.05. J: Quantitation of miR155 5p expression by qPCR in islets isolated from NDfed Ldlr−/− mice 6 h after injection of LPS (2 mg/kg) or vehicle (n = 6 mice per group). *P < 0.05. K: IPGTT in Ldlr−/− mice 6 h after injection of LPS (2 mg/kg) or vehicle (n = 6 mice per group). *P < 0.05 and **P < 0.01 between LPS Mir155−/−Ldlr−/− and LPS Mir155+/+Ldlr−/−; #P < 0.01 and ##P < 0.001 between LPS Mir155+/+Ldlr−/− and vehicle Mir155+/+Ldlr−/−. Data are represented as mean ± SEM.

26 Page 27 of 74 Diabetes

Figure 4─ Mir155 deficiency reduces IL6 expression in βcells. A and B: Gene expression profiling by microarrays in islets isolated from NDfed Mir155+/+Ldlr−/− and Mir155−/−Ldlr−/− mice (10–12 wks of age) (n = 3 samples per group). Biological processes enriched with differentially regulated genes (A) and upstream regulators (B) of differential gene expression predicted by Ingenuity Pathway Analysis software (p < 0.05; fold change cutoff = 1.5). C: Quantitation of IL6 expression at the mRNA and protein level in islets isolated from NDfed mice (10–12 wks of age) by qPCR and ELISA, respectively (n = 6 per group). D: Combined IL6 and insulin immunostaining in pancreatic sections from NDfed mice (10–12 wks of age). Arrows indicate insulin+ cells expressing IL6. Nuclei were counterstained with DAPI. Scale bars: 50 µm. E: Quantitation of Il6 mRNA expression by qPCR in α and βcells sorted from islets of NDfed Mir155+/+Ldlr−/− mice and Mir155−/−Ldlr−/− mice (n = 3–4 per group). F: Quantitation of Il6 mRNA expression by qPCR in sorted human α and βcells treated with miR1555p mimics or control mimics (n = 3–4 per group).G: Effect of antiIL6 antibody treatment on the expression levels of Ins, Gcg, Psck1, and Pcsk2 in MIN6 cells compared to treatment with isotype control antibodies (n = 4 per group). Data are represented as mean ± SEM. *P < 0.05, **P < 0.01 and ***P < 0.001.

Figure 5─ Targeting of MafB by miR1555p promotes IL6 expression in islets. A: Quantitation of predicted miR1555p target gene expression in islets isolated from NDfed Mir155−/−Ldlr−/− mice and Mir155+/+Ldlr−/− mice (10–12 wks of age) by qPCR (n = 6–8 per group). B: Expression of predicted miR1555p targets in MIN6 cells after transfection with miR1555p mimics by qPCR (n = 6 per group). Nontargeting oligonucleotides were used as control. C: Enrichment of potential miR1555p targets in the Argonaute/RISC complexes from MIN6 cells determined by GW182 immunoprecipitation (MirTrapIP) and qPCR (n = 3 per group). The results are expressed as fold enrichment of the transcripts in miR1555p mimics treated MIN6 cells compared to those treated with nontargeting, control mimics. The fold enrichment of the AcGFP1 control in miR132 mimics treated MIN6 cells was used as positive control. D: Quantitation of MafBexpressing cells in islets from NDfed Mir155+/+Ldlr−/− mice and Mir155−/−Ldlr−/− mice (10–12 wks of age) by combined MafB and insulin immunostaining (n = 10 mice per group). The nuclei were counterstained with DAPI. Scale bars: 50 µm. E: Quantitation of Mafb mRNA expression by qPCR in α and βcells sorted from islets of NDfed Mir155+/+Ldlr−/− mice and Mir155−/−Ldlr−/− mice (n = 3–4 per group). F: Quantitation of Mafb, Ins, Gcg, Pcsk1 and Pcsk2 mRNA expression in MIN6 cells treated with oligonucleotides that block the interaction between miR1555p and the 3’UTR of Mafb (target site blocker, 155/Mafb TSB) or nontargeting TSBs (control TSB) by qPCR (n

27 Diabetes Page 28 of 74

= 5 per group). G: Quantitation of IL6 expression at the mRNA and protein level in MIN6 cells treated with 155/Mafb TSB or control TSB by qPCR and ELISA, respectively (n = 5 per group). H: Luciferase activity in MIN6 cells cotransfected with the empty luciferase reporter (control vector) or luciferase reporter constructs harboring the Il6 promoter region with or without sitedirected mutations in the predicted Mafb binding sites (Il6 promoter vector, Il6 promoter∆Mafb1 and Il6 promoter∆Mafb2) and miR1555p LNA inhibitors or non targeting LNA oligonucleotides (n = 4 per group). The luminescence intensities of Gaussia luciferase (GLuc) were normalized to the activity of secreted alkaline phosphatase (SEAP). Data are represented as mean ± SEM. *P < 0.05, **P < 0.01 and ***P < 0.001.

Figure 6─ Effect of the interaction between miR1555p and Mafb on glucose homeostasis in Ldlr–/– mice. A: Quantitation of Mafb mRNA expression by qPCR in various tissues of ND fed mice 21 days after the injection of 155/Mafb TSBs or control TSBs (n = 4 per group). B: Quantitation of MafBexpressing cells in murine islets 21 days after the injection of 155/Mafb TSBs or control TSBs by combined MafB and insulin immunostaining (n = 6 or 7 mice per group). The nuclei were counterstained with DAPI. Scale bars: 50 µm. C: Quantitation of gene expression by qPCR in islets isolated from NDfed mice 21 days after injection of 155/Mafb TSBs or control TSBs (n = 4 per group). D: Quantitation of insulin and glucagon producing cells in islets from NDfed mice 21 days after injection of 155/Mafb TSBs or control TSBs by immunostaining (n = 6 or 7 mice per group). The nuclei were counterstained with DAPI. Scale bars: 50 µm. E: Fasting insulin, glucagon, GLP1 plasma concentrations by Luminex multiplex analysis in NDfed mice 21 days after treatment with 155/Mafb TSBs or control TSBs (n = 7 mice per group). F and G: FBG levels (F) and glucose tolerance (G) in NDfed mice 21 days after injection of 155/Mafb TSBs or control TSBs (n = 7 mice per group). Data are represented as mean ± SEM. *P < 0.05 and **P < 0.01.

Figure 7─ miR1555p improves islets adaptation to lipotoxic stress. Induction of miR155 5p expression in βcells by hyperlipidemiaassociated endotoxemia promotes a βcell phenotype by targeting the transcription factor Mafb, which results in derepression of IL6 gene transcription and increased production of intraislet GLP1.

28 Page 29 of 74 Diabetes

Figure 1─ Effects of Mir155 knockout on atherosclerosis, obesity, and metabolism in

Ldlr−/− mice.

Diabetes Page 30 of 74

Figure 2─ Effect of miR1555p on pancreatic islets.

Page 31 of 74 Diabetes

Figure 3─ miR1555p mediates the effects of LPS and hyperlipidemia on glucose homeostasis.

Diabetes Page 32 of 74

Figure 4─ Mir155 deficiency reduces IL6 expression in βcells.

Page 33 of 74 Diabetes

Figure 5─ Targeting of MafB by miR1555p promotes IL6 expression in islets.

Diabetes Page 34 of 74

Figure 6─ Effect of the interaction between miR1555p and Mafb on glucose homeostasis in Ldlr–/– mice.

Page 35 of 74 Diabetes

Figure 7─ miR1555p improves islets adaptation to lipotoxic stress.

Diabetes Page 36 of 74

Figure legends for primary figures

Page 37 of 74 Diabetes

Figure 3─ miR1555p mediates the effects of LPS and hyperlipidemia on glucose homeostasis. A: Serum endotoxin levels in mice fed an ND or the DDC for 24 wks (left, n = 6 per group). Quantitation of miR1555p expression in lasermicrodissected islets from mice fed the ND or the DDC for 24 wks (right, n = 4 per group). **P < 0.01 and ***P < 0.001. B: Serum endotoxin levels in 10–12 wks old mice fed an ND determined by LAL test (left, n = 6 mice per group). Quantitation of miR1555p expression by qPCR in isolated murine islets from NDfed mice (1012 wks of age) (right, n = 6 mice per group). *P < 0.05. C: Localization of miR1555p expression in islets from NDfed Ldlr+/+ and Ldlr−/− mice (10–12 wks of age) determined by in situ PCR and glucagon immunostaining. The nuclei were counterstained with DAPI. Scale bars: 50 µm. D: Quantitation of miR1555p expression in MIN6 cells treated with native LDL (nLDL), mildly oxidized LDL (moxLDL) or vehicle for 6 h (n = 5–6 per group). *P < 0.05. E and F: Quantitation of miR1555p expression in MIN6 cells (E) and human islets (F) treated with LPS or vehicle for 6 h (n = 5–6 per group). *P < 0.05. G: Endotoxin activity in nLDL or moxLDL determined by LAL test (n = 3 per group). ***P < 0.001. H: oxLDL immunostaining in islets from NDfed mice (10–12 wks of age). The nuclei were counterstained with DAPI. Scale bars: 50 µm. I: Glucoseinduced insulin secretion from islets isolated from NDfed Mir155−/−Ldlr−/− mice and Mir155+/+Ldlr−/− mice (10–12 wks of age) with or without LPS (50 ng/ml) stimulation. Insulin concentrations in the medium were measured by ELISA (n = 4 per group). *P < 0.05. J: Quantitation of miR155 5p expression by qPCR in islets isolated from NDfed Ldlr−/− mice 6 h after injection of LPS (2 mg/kg) or vehicle (n = 6 mice per group). *P < 0.05. K: IPGTT in Ldlr−/− mice 6 h after injection of LPS (2 mg/kg) or vehicle (n = 6 mice per group). *P < 0.05 and **P < 0.01 between LPS Mir155−/−Ldlr−/− and LPS Mir155+/+Ldlr−/−; #P < 0.01 and ##P < 0.001 between +/+ −/− +/+ −/− LPS Mir155 Ldlr and vehicle Mir155 Ldlr . Data are represented as mean ± SEM. 9

Diabetes Page 38 of 74

Page 39 of 74 Diabetes

Figure 7─ miR1555p improves islets adaptation to lipotoxic stress. Induction of miR 1555p expression in βcells by hyperlipidemiaassociated endotoxemia promotes a βcell phenotype by targeting the transcription factor Mafb, which results in derepression of IL6 gene transcription and increased production of intraislet GLP1.

Diabetes Page 40 of 74

SUPPLEMENTARY DATA

Hyperlipidemiainduced microRNA1555p improves βcell function by targeting Mafb

Short title: microRNA1555p improves βcell adaptation

Mengyu Zhu1, Yuanyuan Wei1, 2, Claudia Geißler1, Kathrin Abschlag1, Judit Corbalán

Campos1, Michael Hristov1, Julia Möllmann3, Michael Lehrke3, Ela Karshovska1, 2,

Andreas Schober1, 2

1Institute for Cardiovascular Prevention, LudwigMaximiliansUniversity Munich, Munich,

Germany

2DZHK (German Centre for Cardiovascular Research), partner site Munich Heart Alliance,

Munich, Germany

3Department of Internal Medicine I; University Hospital Aachen, Germany

Word count: 4186

Tables and Figures in main text: 1 Table and 7 Figures

Corresponding author: Andreas Schober, MD

Institute for Cardiovascular Prevention

LudwigMaximiliansUniversity Munich

Pettenkoferstrasse 9b, 80336 Munich, Germany

Tel: 4989440055151; Fax: 4989440054740

Email: [email protected]

Page 41 of 74 Diabetes

Inventory of Supplementary Data

Supplementary Figures

Supplementary Figure 1. Effects of miR1555p on atherosclerosis, lipid profile and

adipokines in Ldlr−/− mice.

Supplementary Figure 2. Effects of miR1555p on pancreatic islets.

Supplementary Figure 3. Effects of miR1555p on glucose tolerance.

Supplementary Figure 4. miR1555p mediates the effects of LPS and hyperlipidemia on

glucose homeostasis.

Supplementary Figure 5. Effects of miR1555p on IL6 and GLP1.

Supplementary Figure 6. miR1555p promotes IL6 expression by targeting of MafB.

Supplementary Figure 7. Effects of 155/Mafb TSB on Ldlr−/− mice.

Supplementary Tables

Supplementary Table 1. Genes upregulated in islets from Mir155–/–Ldlr–/– mice compared

with Mir155+/+Ldlr–/– mice.

Supplementary Table 2. Genes downregulated in islets from Mir155–/–Ldlr–/– mice compared

with Mir155+/+Ldlr–/– mice.

Supplementary Table 3. Mouse primer sequences for qPCR analysis.

Supplementary Table 4. Human primer sequences for qPCR analysis.

Supplementary Table 5. Primer sequences for in situ PCR.

Supplementary Table 6. Primer sequences for luciferase reporter assays.

Supplementary Experimental Procedures

Supplementary References

Diabetes Page 42 of 74

Supplementary Figures

Supplementary Figure 1. Effects of miR1555p on atherosclerosis, lipid profile and adipokines in Ldlr−/− mice. A: Accumulation of macrophages and smooth muscle cells in aortic root lesions from Mir155+/+Ldlr−/− and Mir155−/−Ldlr−/− mice fed a DDC for 24–wks determined by combined Mac2 and smooth muscle actin (SMA) immunostaining, respectively (n = 10 mice per group). The nuclei were counterstained with DAPI. Scale bars: 100 µm. B: Plasma cholesterol and triglyceride levels in Mir155+/+Ldlr−/− and Mir155−/−Ldlr−/− mice after a 24–wks DDC feeding period (n = 10 mice per group). C: Quantitation of gene expression by realtime PCR (qPCR) in eWAT from Mir155+/+Ldlr−/− and Mir155−/−Ldlr−/− mice fed a diabetogenic diet containing cholesterol (DDC) for 24 wks (n = 4 or 6 mice per group). Adipoq, Adiponectin; Lep, leptin; Tnf, tumor necrosis factor; Il6, interleukin 6; Nos2, nitric oxide synthase 2; Mrc1, mannose receptor, C type 1. Data are represented as mean ± SEM; *P < 0.05 and **P < 0.01.

Page 43 of 74 Diabetes

Supplementary Figure 2. Effects of miR1555p on pancreatic islets. A: Quantitation of gene expression by qPCR in murine islets from normal diet (ND)fed Mir155+/+Ldlr−/− mice and Mir155−/−Ldlr−/− mice (10–12 wks of age, n = 6 or 8 mice per group). Sst, somatostatin; Isl1, ISL LIM homeobox 1; Arx, aristaless related homeobox; Pdx1, pancreatic and duodenal

Diabetes Page 44 of 74

homeobox 1; Pax6, paired box 6; Neurod1, neurogenic differentiation 1; Foxa1, forkhead box A1. B: Islet cell apoptosis determined by immunostaining of activated Caspase3, and macrophages and Tcells detected by combined Mac2 and CD3 immunostaining, respectively, in murine islets from NDfed Mir155+/+Ldlr−/− mice and Mir155−/−Ldlr−/− mice (10–12 wks of age) (n = 10 mice per group). Representative images are shown. The nuclei were counterstained with DAPI. Scale bars: 50 µm. C: Quantitation of miR1555p expression in MIN6 cells treated with miR1555p mimics or miR1555p LNAinhibitors (miR1555p inhibitors) by qPCR (n = 6 per group). Nontargeting oligonucleotides were used as control (control inhibitors or control mimics). D: Quantitation of gene expression and protein levels by qPCR (n = 6 per group) and Luminex multiplex analysis (n = 4 per group), respectively, in MIN6 cells treated with miR1555p LNAinhibitors or nontargeting LNAoligonucleotides (control inhibitors). Ins, insulin; Gcg, glucagon; Pcsk, proprotein convertase subtilisin/kexin type. E: Quantitation of Pcsk1 and Pcsk2 mRNA expression in miR1555p mimics treated α cells sorted from islets of NDfed Mir155+/+Ldlr−/− mice (12 wks of age) (n = 3–4 per group). F: Quantitation of gene expression by qPCR in sorted human α and βcells treated with miR 1555p mimics or control mimics (n = 3 – 4 per group). G and H: Quantitation of GLP1 secretion (G, n = 6 per group, right), the GLP1 cellular protein content (G, n = 4 per group), and of the GCG and PCSK1 mRNA expression (H, n = 6 per group) after transfection of intestinal Lcells (GLUTag cells) with miR1555p mimics or control mimics. Data are represented as mean ± SEM; *P < 0.05, **P < 0.01 and ***P < 0.001.

Page 45 of 74 Diabetes

Supplementary Figure 3. Effects of miR1555p on glucose tolerance. AC: Comparison of glucose tolerance between female Mir155+/+Ldlr−/− mice and Mir155−/−Ldlr−/− mice (A, n = 4 mice per group), male Mir155+/+Apoe−/− mice and Mir155−/−Apoe−/− mice (B, n = 4–5 mice per group), and female Mir155+/+Apoe−/− mice and Mir155−/−Apoe−/− mice (C, n = 4–6 mice per group) at 10–12 wks of age fed a ND. Data are represented as mean ± SEM; *P < 0.05 and **P < 0.01.

Diabetes Page 46 of 74

Supplementary Figure 4. miR1555p mediates the effects of LPS and hyperlipidemia on glucose homeostasis. A: Plasma cholesterol and triglyceride levels in Ldlr+/+and Ldlr−/− mice at 10–12 wks of age fed a ND (n = 10 mice per group). B: Fasting insulin and GLP1 plasma concentrations in NDfed Ldlr−/− mice 6 h after injection of LPS (2 mg/kg) or vehicle determined by Luminex multiplex analysis (n = 6 mice per group). Data are represented as mean ± SEM. *P < 0.05, **P < 0.01 and ***P < 0.001.

Page 47 of 74 Diabetes

Supplementary Figure 5. Effects of miR1555p on IL6 and GLP1. A and B: Quantitation of Il6 mRNA expression (n = 6 per group) and IL6 protein secretion (n = 4 per group) by qPCR and ELISA, respectively, after transfection of MIN6 cells with miR1555p mimics (A) or miR1555p LNA inhibitors (B). Nontargeting oligonucleotides were used as control. C: Human αcells were incubated with conditioned medium from βcells that had been treated with miR1555p mimics or control mimics. GLP1 was quantitated in the supernatant (n = 3– 4 per group) and cell lysates (n = 4 per group) of the αcells stimulated with the conditioned medium. Data are represented as mean ± SEM. *P < 0.05 and **P < 0.01.

Diabetes Page 48 of 74

Supplementary Figure 6. miR1555p promotes IL6 expression by targeting of MafB. A: Expression of putative miR1555p targets in MIN6 cells after transfection with miR1555p inhibitors by qPCR (n = 5 or 6 per group). Nontargeting oligonucleotides were used as control. B: Enrichment of potential miR1555p targets in the miRISC from human islet cells determined by GW182 immunoprecipitation (MirTrapIP) and qPCR (n = 2 per group). The results are expressed as fold enrichment of the transcripts in miR1555p mimictreated human islets compared to those treated with nontargeting, control mimics. The fold enrichment of AcGFP1 in miR132 mimics treated human islet cells was used as a positive control. C: Quantitation of MAFB mRNA expression by qPCR in sorted human α and βcells following treatment with miR1555p mimics or control mimics (n = 3–4 per group). D: Conserved binding site of miR1555p in the 3′ untranslated region (3’–UTR) of Mafb mRNA in human (hsa) and mouse (mmu). The boxes indicate the interaction between miR1555p and Mafb. Sequence of the Mafbtarget site blocker oligonucleotide (155/Mafb TSB) complementary to the miR1555p binding site and its 5’ flanking region in the Mafb3′UTR.

Page 49 of 74 Diabetes

E: Wildtype (Il6 promoterMafb1 and Il6 promoterMafb2) and mutated (Il6 promoter ∆Mafb1 and Il6 promoter∆Mafb2, underlined) sequences of putative Mafb binding sites in the Il6 promoter region. Data are represented as mean ± SEM. *p < 0.05, **p < 0.01 and ***p < 0.001.

Diabetes Page 50 of 74

Supplementary Figure 7. Effects of 155/Mafb TSB on Ldlr−/− mice. A: Body weight of Ldlr−/− mice before (Day 0) and 21 d (Day 21) after treatment with 155/Mafb TSBs or control TSBs (n = 7 mice per group). B: White blood cell (WBC), lymphocyte (LYM), monocyte (MO) and granulocyte (GRA) count in the blood of Ldlr−/− mice 21 d after injection of 155/Mafb TSBs or control TSBs (n = 7 mice per group). C: Quantitation of putative miR155 5p targets genes by qPCR in islets from Ldlr−/− mice 21 d after treatment with 155/Mafb TSBs or control TSBs (n = 4 per group). Data are represented as mean ± SEM.

Page 51 of 74 Diabetes

Supplementary Table 1. Genes upregulated in islets from Mir155–/–Ldlr–/– mice compared with Mir155+/+Ldlr–/– mice.

GeneSymbol EntrezGeneID p value Fold change Lrrn2 16980 0.003 5.01 Rnu3b1 19858 0.050 4.54 Rbp7 63954 0.037 3.95 Aldob 230163 0.002 3.47 Zfp354a 21408 0.001 3.37 Mycn 18109 0.009 3.35 D9Wsu90e 27962 0.019 3.26 Npas4 225872 0.040 2.98 Pisdps3 66776 0.004 2.97 Stmn2 20257 0.032 2.78 Agfg2 231801 0.001 2.70 Gm3325 100041420 0.021 2.61 Zfp354a 21408 0.003 2.56 Med12l 329650 0.009 2.52 Ppp1r17 19051 0.000 2.51 Ushbp1 234395 0.039 2.50 Piezo2 667742 0.000 2.49 Casq2 12373 0.002 2.49 Hist1h2bq 665596 0.009 2.41 Pde4d 238871 0.004 2.36 AU015680 552875 0.006 2.31 Pdk4 27273 0.023 2.31 4932431P20Rik 114675 0.020 2.29 Zfp454 237758 0.008 2.28 Cmya5 76469 0.020 2.27 Tbx2 21385 0.006 2.25 Gpr179 217143 0.021 2.22 Zfp36 22695 0.027 2.21 Scai 320271 0.001 2.19 B020031M17Rik 333467 0.042 2.14 Plcb4 18798 0.010 2.13 Olfr1372ps1 257871 0.002 2.09 Gcm1 14531 0.019 2.07 Nsg1 18196 0.030 2.07 Nrp1 18186 0.050 2.06 Lama3 16774 0.026 2.05 Notch1 18128 0.038 2.05 Sncg 20618 0.016 2.04 F13a1 74145 0.003 2.03 Pclo 26875 0.042 2.02 Zfp111 56707 0.011 2.01 Sema5a 20356 0.012 2.01 Neurog3 11925 0.020 2.00 C2cd4a 244911 0.036 2.00 A630081D01Rik 98754 0.043 1.99 C2cd4a 244911 0.038 1.99 Saysd1 67509 0.034 1.98 LOC552901 552901 0.005 1.97 Zfp174 385674 0.023 1.95 Zfp280d 235469 0.042 1.94 Gabbr2 242425 0.032 1.94 Prkg2 19092 0.031 1.93 Asb9 69299 0.013 1.93 Gm13271 435791 0.026 1.91 Purg 75029 0.033 1.90 Gm5868 545758 0.028 1.90 Rhox8 434768 0.034 1.90 BC052040 399568 0.006 1.90 Hyls1 76832 0.007 1.89 Glrx2 69367 0.007 1.89 12

Diabetes Page 52 of 74

Cldn15 60363 0.005 1.89 Slc2a4rgps 329584 0.024 1.89 Pisd 320951 0.005 1.89 9330177L23Rik 77246 0.010 1.88 Snx29 74478 0.044 1.88 Airn 104103 0.020 1.87 A130078K24Rik 399616 0.014 1.87 Cmpk2 22169 0.032 1.87 4930598N05Rik 75375 0.010 1.86 Cecr5 214932 0.008 1.86 Sgms1 208449 0.030 1.86 Zkscan3 72739 0.008 1.85 Kif17 16559 0.006 1.85 Alox12 11684 0.015 1.85 9930024M15Rik 399602 0.031 1.84 Pcyox1l 240334 0.014 1.84 L1td1 381591 0.021 1.83 Trpc3 22065 0.012 1.83 Prrt1 260297 0.019 1.82 Nkapl 66707 0.049 1.82 Zfp950 214368 0.004 1.82 Lrrc3b 218763 0.036 1.82 Gm13286 100041062 0.010 1.81 Ppp2r2c 269643 0.013 1.81 Zfp606 67370 0.037 1.81 Dcc 13176 0.042 1.80 Olfr1384 258464 0.043 1.79 Zfp108 54678 0.015 1.78 Nek3 23954 0.029 1.77 Mnd1 76915 0.019 1.77 AA388235 433100 0.015 1.76 Reep1 52250 0.009 1.75 Xntrpc 102443351 0.021 1.75 Auh 11992 0.034 1.75 Rab13 68328 0.015 1.74 Phf21a 192285 0.022 1.74 Rgs11 50782 0.028 1.74 Rab3c 67295 0.028 1.74 Malat1 72289 0.024 1.73 Ltbp4 108075 0.020 1.73 5930433N17Rik 399623 0.007 1.73 4930461C15Rik 74883 0.031 1.73 Otud1 71198 0.015 1.72 Ccl3 20302 0.022 1.72 Nnat 18111 0.024 1.72 Zfp937 245174 0.038 1.71 Gatm 67092 0.033 1.71 Ghsr 208188 0.033 1.71 Acrbp 54137 0.038 1.71 Kif5a 16572 0.032 1.71 C030039L03Rik 112415 0.044 1.70 Pigr 18703 0.044 1.70 1600020E01Rik 72012 0.011 1.70 S100a4 20198 0.012 1.70 Gap43 14432 0.042 1.69 Dhfr 13361 0.022 1.69 Cdo1 12583 0.027 1.68 Mkks 59030 0.046 1.68 Zbtb12 193736 0.031 1.68 Gm3604 100041979 0.015 1.68 Npas3 27386 0.026 1.68 1700029I15Rik 75641 0.028 1.68 Pde4dip 83679 0.029 1.67 D630039A03Rik 242484 0.030 1.67 Doc2g 60425 0.021 1.67 13

Page 53 of 74 Diabetes

9630039A02Rik 319254 0.036 1.67 Ribc1 66611 0.020 1.67 1700023D09Rik 69445 0.029 1.67 Smarca1 93761 0.034 1.66 Sdad1 231452 0.041 1.66 F5 14067 0.016 1.66 Lama2 16773 0.013 1.65 Rabif 98710 0.031 1.65 Bcl7a 77045 0.023 1.65 Pnma2 239157 0.032 1.65 Zfp14 243906 0.050 1.65 Fbxo41 330369 0.034 1.65 4931429I11Rik 70989 0.017 1.65 Fev 260298 0.023 1.64 2310016D23Rik 76443 0.025 1.64 Bik 12124 0.009 1.64 Ccdc30 73332 0.015 1.64 Ppp1r9a 243725 0.019 1.64 Ly6c1 17067 0.028 1.64 Syce1 74075 0.048 1.64 Podxl 27205 0.022 1.64 Ccl4 20303 0.027 1.64 LOC552902 552902 0.013 1.63 Gm609 208166 0.012 1.63 Pinx1 72400 0.048 1.63 Cep68 216543 0.010 1.63 5730407M17Rik 70480 0.024 1.62 Mnx1 15285 0.047 1.62 Pisdps1 236604 0.009 1.62 A330023F24Rik 320977 0.029 1.62 Nrep 27528 0.046 1.62 B4galt4 56375 0.034 1.61 Dnmt3b 13436 0.021 1.61 Ces2e 234673 0.024 1.61 Gm11123 100169876 0.033 1.60 Dnmt3b 13436 0.030 1.60 Slc27a2 26458 0.013 1.60 Gpr4 319197 0.013 1.60 Nnat 18111 0.030 1.60 Mafb 16658 0.017 1.67 Lrrc4b 272381 0.020 1.60 Iqcf4 67320 0.028 1.60 Pcdh12 53601 0.028 1.59 2900024J01Rik 72850 0.037 1.59 Sall4 99377 0.031 1.59 Smarca1 93761 0.027 1.59 Ppnr 26930 0.031 1.59 Snrpn 20646 0.024 1.59 B3gnt8 232984 0.041 1.59 Ccdc77 67200 0.022 1.58 Timeless 21853 0.015 1.58 Cpt1c 78070 0.035 1.58 Srrm3 58212 0.018 1.58 Dpp10 269109 0.034 1.57 Nedd4l 83814 0.038 1.57 Plscr2 18828 0.025 1.57 4930579K19Rik 75881 0.041 1.56 Zfp786 330301 0.022 1.56 4930412L05Rik 73941 0.015 1.56 Tmem145 330485 0.025 1.56 Apof 103161 0.025 1.56 Zfp672 319475 0.039 1.56 Cpb2 56373 0.032 1.55 Dysf 26903 0.023 1.55 Ssu2 243612 0.037 1.55 14

Diabetes Page 54 of 74

AW551984 244810 0.043 1.55 Nrsn1 22360 0.040 1.54 Timeless 21853 0.022 1.54 Prr22 100504446 0.017 1.54 Zfp296 63872 0.049 1.54 5330421C15Rik 78279 0.028 1.53 Zscan20 269585 0.027 1.53 Zfp473 243963 0.029 1.53 Esm1 71690 0.021 1.53 Fam120b 67544 0.019 1.53 Rufy3 52822 0.028 1.53 Pard3 93742 0.044 1.53 Nav2 78286 0.042 1.53 LOC548102 548102 0.033 1.53 C030005K06Rik 78695 0.041 1.53 Enox1 239188 0.029 1.52 Klhl42 232539 0.029 1.52 Ncoa7 211329 0.046 1.52 Mblac1 330216 0.039 1.52 Kank3 80880 0.027 1.52 Zfp93 22755 0.042 1.52 9430015G10Rik 230996 0.031 1.52 Camkk2 207565 0.038 1.52 Gria3 53623 0.035 1.52 Gchfr 320415 0.035 1.51 Klhl17 231003 0.044 1.51 Htra3 78558 0.043 1.51 Clec1a 243653 0.040 1.51 Homez 239099 0.050 1.51 Myct1 68632 0.041 1.51 1600014C10Rik 72244 0.044 1.50 Cnot11 52846 0.023 1.50 Itih2 16425 0.042 1.73 Scn2b 72821 0.019 3.22 Nfatc2 18019 0.039 2.77 Cep350 74081 0.046 2.61 Rhox4g 664608 0.025 2.52 Cdkl5 382253 0.003 2.44 Camsap2 67886 0.010 2.10 Paqr6 68957 0.035 2.00 Cpne6 12891 0.021 1.92 Fstl5 213262 0.018 1.83 Maged2 80884 0.022 1.82 Nxph1 18231 0.024 1.75 Oscp1 230751 0.047 1.72 Cyp2d40 71754 0.032 1.64 Cyb5rl 230582 0.017 1.57 Rhox4a 664609 0.043 3.06 Ptgds 19215 0.001 2.63

Page 55 of 74 Diabetes

Supplementary Table 2. Genes downregulated in islets from Mir155–/–Ldlr–/– mice compared with Mir155+/+Ldlr–/– mice.

GeneSymbol EntrezGeneID p value Fold change Ptgr1 67103 0,00001 13,32 Psapl1 76943 0,00001 57,83 Car2 12349 0,00005 22,73 S100a14 66166 0,00005 13,45 Ces2c 234671 0,00009 67,71 Sult1b1 56362 0,00011 9,27 Ces2c 234671 0,00014 7,58 Aldh1a3 56847 0,00018 2,80 Mrpl39 27393 0,00019 2,78 Mst1r 19882 0,00021 4,92 Krt19 16669 0,00021 12,86 Akr1b8 14187 0,00022 3,16 Gsto1 14873 0,00024 4,33 Fam83e 73813 0,00024 5,81 Krt42 68239 0,00030 12,01 Spink4 20731 0,00034 7,04 Mal 17153 0,00034 17,46 Sdc1 20969 0,00035 2,55 Aqp3 11828 0,00039 13,24 Flnc 68794 0,00042 7,37 Sh3bgrl2 212531 0,00043 2,76 Slc39a4 72027 0,00044 3,62 Crispld2 78892 0,00044 3,90 Fcgbp 215384 0,00048 7,29 Areg 11839 0,00048 3,46 Tst 22117 0,00052 4,24 Crispld2 78892 0,00075 4,15 Rab27b 80718 0,00076 3,13 Vill 22351 0,00080 4,91 Nqo1 18104 0,00082 4,04 Tspan1 66805 0,00082 7,22 Vsig2 57276 0,00086 5,76 Bcas1 76960 0,00087 4,11 Fabp2 14079 0,00087 10,90 AA467197 433470 0,00103 5,46 Krt20 66809 0,00109 12,14 Wdfy1 69368 0,00114 2,69 Fez1 235180 0,00121 4,23 Cbr3 109857 0,00128 2,24 Gkn3 68888 0,00144 3,44 Pifo 100503311 0,00144 2,33 Macc1 238455 0,00172 5,90 Pttg1 30939 0,00178 4,53 Wdfy1 69368 0,00194 2,44 Cyp2s1 74134 0,00217 4,15 Myo1a 432516 0,00257 3,45 Gdf15 23886 0,00257 2,39 Tmem254c 100039192 0,00263 2,21 Itga2 16398 0,00267 2,00 Gsta3 14859 0,00271 4,33 Fads3 60527 0,00271 2,36 Lgals4 16855 0,00274 3,58 Noxa1 241275 0,00278 2,20 Abcc3 76408 0,00299 2,42 Nkx63 74561 0,00340 3,67 Hrh3 99296 0,00340 1,87 Gast 14459 0,00348 4,85 1700007K13Rik 69327 0,00357 2,00 Mgst3 66447 0,00364 2,52 Ren2 19702 0,00369 3,73 16

Diabetes Page 56 of 74

Itgb4 192897 0,00375 2,07 Gsta3 14859 0,00377 4,26 Dbp 13170 0,00386 2,16 Smim24 72273 0,00390 2,13 Mgst3 66447 0,00391 2,51 Tst 22117 0,00395 3,88 Gzma 14938 0,00399 5,64 Lgals4 16855 0,00405 4,10 Mmp3 17392 0,00414 36,23 Ly6g6c 68468 0,00423 2,85 Kdm5b 75605 0,00457 2,44 Egln3 112407 0,00462 3,46 Kctd14 233529 0,00478 2,54 Rep15 66532 0,00507 3,26 Lgals6 16857 0,00516 3,21 Hk2 15277 0,00539 3,67 Lgals4 16855 0,00541 3,88 Mc3r 17201 0,00543 2,15 Lyar 17089 0,00551 1,99 Slc22a18 18400 0,00581 2,47 Lgals3 16854 0,00581 2,75 Rbp1 19659 0,00609 2,88 Gkn1 66283 0,00613 3,87 Ldha 16828 0,00649 3,10 Krt23 94179 0,00662 3,40 Mid1 17318 0,00671 2,96 Smim3 106878 0,00686 3,11 Syt8 55925 0,00693 2,20 Gm13315 625342 0,00714 3,33 S100a6 20200 0,00731 2,78 Bmp2 12156 0,00790 2,80 Selp 20344 0,00793 5,62 Popdc2 64082 0,00795 3,42 Ldha 16828 0,00802 3,35 Dgat2 67800 0,00811 2,08 Lypd6b 71897 0,00830 4,55 Etnk2 214253 0,00835 2,24 Islr2 320563 0,00840 2,41 Pyroxd2 74580 0,00852 2,08 Gjb1 14618 0,00855 2,07 Timp1 21857 0,00858 8,05 Steap1 70358 0,00859 1,77 Grwd1 101612 0,00873 1,79 Il6 16193 0,00874 3,85 1700026L06Rik 69987 0,00890 1,81 Foxa1 15375 0,00911 3,62 Dbp 13170 0,00931 1,96 9530036O11Rik 654796 0,00934 2,65 Cald1 109624 0,00960 1,97 Sgcd 24052 0,00961 1,68 Cartpt 27220 0,00965 1,93 Fut1 14343 0,01013 1,65 Anapc15 75430 0,01046 1,67 Hmga2ps1 15365 0,01060 1,70 Alpl 11647 0,01064 1,92 Cyp26b1 232174 0,01081 2,36 5830468F06Rik 76082 0,01095 2,43 E130012A19Rik 103551 0,01135 2,62 Tppp 72948 0,01148 3,22 Lamb3 16780 0,01152 3,61 Egln3 112407 0,01176 3,55 Psat1 107272 0,01193 1,84 Wfdc13 408190 0,01195 1,63 Odf3b 70113 0,01255 1,77 1700016K19Rik 74230 0,01262 1,64 17

Page 57 of 74 Diabetes

3300002A11Rik 1E+08 0,01274 3,01 Nt5e 23959 0,01275 3,00 Yap1 22601 0,01278 1,84 Cabyr 71132 0,01289 2,32 Gprc5a 232431 0,01309 1,66 D630013G24Rik 319825 0,01317 2,43 Kcnd3 56543 0,01328 2,50 Npvf 60531 0,01362 1,86 Pgm5 226041 0,01372 5,50 Ppargc1a 19017 0,01379 2,08 Mlf1 17349 0,01380 2,30 2200002D01Rik 72275 0,01385 2,65 Cyp4a31 666168 0,01388 1,56 Tgtp2 100039796 0,01397 3,67 Tmem54 66260 0,01406 2,74 Slco2a1 24059 0,01419 1,60 Odc1 18263 0,01437 1,82 Slc52a3 69698 0,01441 1,84 Serinc2 230779 0,01491 1,99 Usp2 53376 0,01505 1,95 Mgst2 211666 0,01515 19,26 Tnfaip6 21930 0,01544 3,32 Kiss1 280287 0,01566 1,61 Mpzl2 14012 0,01578 1,95 Ecm1 13601 0,01609 1,76 Col18a1 12822 0,01613 1,61 Fgf7 14178 0,01625 3,19 Ogn 18295 0,01633 4,66 Ckb 12709 0,01635 1,81 Plac8 231507 0,01642 5,01 Fcna 14133 0,01659 2,01 Vmn1r204 632793 0,01663 1,58 Vcam1 22329 0,01674 4,05 Olfr1466 258689 0,01697 1,53 Slc38a1 105727 0,01698 2,20 Scin 20259 0,01698 1,91 Sbno2 216161 0,01715 1,86 Vwf 22371 0,01732 1,96 Trp53inp1 60599 0,01743 1,69 Tekt4 71840 0,01787 1,58 Chn2 69993 0,01822 2,72 Des 13346 0,01836 3,91 Tnni2 21953 0,01847 4,08 Il11 16156 0,01859 5,78 Dusp16 70686 0,01863 2,27 Synpo2 118449 0,01874 2,10 Plac9a 211623 0,01880 2,63 Calml4 75600 0,01882 1,80 Pgm5 226041 0,01886 6,78 Ckb 12709 0,01887 1,60 2610318N02Rik 70458 0,01904 1,64 Mif 17319 0,01910 1,68 Rab4a 19341 0,01912 1,60 Cenpa 12615 0,01922 1,71 Wdr96 101711286 0,01925 3,39 Smim6 68528 0,01934 1,83 Wisp1 22402 0,01940 2,20 Peli1 67245 0,01972 1,76 Lcn2 16819 0,02024 9,09 D7Wsu130e 28017 0,02031 2,38 Ccdc19 71870 0,02031 1,59 Ecm1 13601 0,02068 1,52 Ppp1r32 67752 0,02091 2,98 Pde3a 54611 0,02113 1,54 Ppat 231327 0,02120 1,56 18

Diabetes Page 58 of 74

Plac8 231507 0,02133 4,40 Stra6 20897 0,02143 3,15 Lypd5 76942 0,02164 1,56 Tnfrsf22 79202 0,02166 1,73 Cntfr 12804 0,02228 2,92 Ipo9 226432 0,02234 1,52 3930401B19Rik 77570 0,02249 2,71 Gfpt2 14584 0,02265 3,21 Itpripl2 319622 0,02319 1,82 Timm8a1 30058 0,02322 1,55 Stbd1 52331 0,02352 2,23 Pde4b 18578 0,02382 1,60 Sp140 434484 0,02386 3,26 Map3k9 338372 0,02411 1,71 Inhba 16323 0,02414 2,10 Efhb 211482 0,02438 3,55 Fosl1 14283 0,02439 3,15 Tdp1 104884 0,02442 1,60 Ethe1 66071 0,02476 1,58 Gata4 14463 0,02484 2,52 Rffl 67338 0,02498 2,01 Hk1 15275 0,02518 1,75 Penk 18619 0,02520 2,61 Tmem254a 66039 0,02556 1,92 Svopl 320590 0,02595 1,79 6430548M08Rik 234797 0,02604 1,84 Srf 20807 0,02640 1,77 Fam129a 63913 0,02657 2,35 B4galt5 56336 0,02670 1,62 Ifnab 15974 0,02700 1,94 Acss1 68738 0,02716 2,49 Slpi 20568 0,02721 14,57 Nr1d1 217166 0,02770 1,88 Nr4a2 18227 0,02772 1,58 Smim24 72273 0,02774 1,74 Trib3 228775 0,02779 2,06 Ptgs2 19225 0,02792 5,17 Mif 17319 0,02800 1,64 Mapkapk2 17164 0,02838 1,79 Mbd1 17190 0,02890 2,21 Cald1 109624 0,02911 1,70 Acot10 64833 0,02928 1,83 Plet1 76509 0,02929 1,70 Fads3 60527 0,02936 1,72 Lrrc36 270091 0,02961 1,92 Pgd 110208 0,03035 1,62 LOC102639083 102639083 0,03069 2,06 Has1 15116 0,03094 3,45 Degs2 70059 0,03113 1,62 Ackr1 13349 0,03118 4,50 Adora3 11542 0,03120 1,57 Lpin2 64898 0,03121 1,66 Tnfrsf19 29820 0,03135 2,77 Zbp1 58203 0,03148 3,83 4732456N10Rik 239673 0,03163 2,26 Clmp 71566 0,03171 3,46 Myl9 98932 0,03172 2,86 Cystm1 66060 0,03200 1,57 Psat1 107272 0,03218 1,68 Clmp 71566 0,03221 4,75 Bub1b 12236 0,03226 1,61 Tpm2 22004 0,03232 4,53 Gpbar1 227289 0,03239 1,52 Slc36a1 215335 0,03261 1,63 Procr 19124 0,03281 3,18 19

Page 59 of 74 Diabetes

Gpr18 110168 0,03297 5,96 Gm13547 433416 0,03314 1,55 Angptl6 70726 0,03315 1,51 Als2 74018 0,03317 1,59 Reg3b 18489 0,03331 5,08 Sh3bp1 20401 0,03332 1,54 Ppp2r1b 73699 0,03340 1,53 Stard5 170460 0,03364 1,53 Hmox1 15368 0,03407 1,77 Ccl7 20306 0,03408 4,96 Tpmt 22017 0,03428 1,55 Cxcr5 12145 0,03471 3,44 Cnn1 12797 0,03491 13,40 Col18a1 12822 0,03504 1,56 Capn9 73647 0,03516 1,66 Ceacam2 26367 0,03527 1,55 Gpnmb 93695 0,03592 3,31 Mdfi 17240 0,03594 2,61 Dynlrb2 75465 0,03623 1,63 Tpm2 22004 0,03624 4,65 Tpm1 22003 0,03633 1,86 Sntg2 268534 0,03690 2,35 Map3k13 71751 0,03697 1,53 Itga5 16402 0,03736 3,98 Diras2 68203 0,03738 1,82 Serinc2 230779 0,03750 1,83 Tmem79 71913 0,03758 1,59 Smim5 66528 0,03769 1,75 Trabd2b 666048 0,03816 1,99 Clec4a1 269799 0,03824 1,77 Fosl2 14284 0,03832 1,73 H2Q10 15007 0,03854 1,79 Cxcl5 20311 0,03860 6,83 Itgb6 16420 0,03864 2,39 Fam107b 66540 0,03866 1,66 Kcne4 57814 0,03875 2,30 Anxa2 12306 0,03880 1,83 Tubb6 67951 0,03914 1,98 Layn 244864 0,03918 1,64 Tfcp2l1 81879 0,03926 1,71 Ifngr1 15979 0,03941 1,53 Mustn1 66175 0,03970 1,89 Mif 17319 0,03988 1,67 Akap2 11641 0,03993 1,58 Capg 12332 0,04000 1,80 Nr1d2 353187 0,04040 1,62 Synm 233335 0,04041 3,31 Stat4 20849 0,04042 3,32 Gm6981 629557 0,04042 1,54 Fjx1 14221 0,04043 1,87 Irg1 16365 0,04059 3,20 Muc13 17063 0,04066 2,30 Pdpn 14726 0,04084 3,57 Grem2 23893 0,04111 2,11 Adm2 223780 0,04133 2,11 Zfp385a 29813 0,04143 1,64 Gm15645 626055 0,04150 1,89 Camkk1 55984 0,04162 1,69 6330403K07Rik 103712 0,04194 1,83 Actn2 11472 0,04210 1,90 Fabp5 16592 0,04213 1,69 Sult1c2 69083 0,04217 1,73 Tnfrsf18 21936 0,04252 5,71 Mif 17319 0,04284 1,68 Cryab 12955 0,04286 1,53 20

Diabetes Page 60 of 74

Hk1 15275 0,04312 2,08 Lfng 16848 0,04346 2,79 Agpat9 231510 0,04354 2,34 Tpm1 22003 0,04377 1,79 Ldha 16828 0,04382 2,40 Slc22a2 20518 0,04386 1,65 Acta1 11459 0,04387 1,99 Fblim1 74202 0,04406 1,67 Pgk1 18655 0,04411 1,62 Dapk2 13143 0,04448 1,94 Pgk1 18655 0,04515 1,63 Cytip 227929 0,04519 3,20 Serinc2 230779 0,04523 1,93 Kcnmb1 16533 0,04524 2,25 Mep1a 17287 0,04588 2,86 LOC102639105 1E+08 0,04595 2,91 Sparc 20692 0,04603 2,05 Gipc2 54120 0,04613 2,04 Sdcbp2 228765 0,04702 1,57 Adamts4 240913 0,04736 2,67 Zscan18 232875 0,04738 1,57 Nccrp1 233038 0,04772 1,90 Dgat1 13350 0,04861 1,54 Slc7a3 11989 0,04887 1,60 Anxa8 11752 0,04909 1,62 Ppp1r16b 228852 0,04933 2,84 Cd151 12476 0,04937 2,09 Tuft1 22156 0,04988 1,69

Page 61 of 74 Diabetes

Supplementary Table 3. Mouse primer sequences for qPCR analysis. Gene Primer Primer sequences Mafb Forward 5'GTAGTGTGGAGGACCGCTTC3' Reverse 5'TTATACCTGCACGACTGGGC3' Glucagon Forward 5'CTGGTGAAAGGCCGAGGAAG3' Reverse 5'GAGAAGGAGCCATCAGCGTG3' Insulin Forward 5'TGGCTTCTTCTACACACCCA3' Reverse 5'TCTAGTTGCAGTAGTTCTCCA3' Il6 Forward 5'CCACTTCACAAGTCGGAGGC3' Reverse 5'TGCCATTGCACAACTCTTTTCT3' Pcsk1 Forward 5'GTGAATGTTGTGGAGAAGCGG3' Reverse 5'TTGTAGGAGTCGCAGCATGG3' Pcsk2 Forward 5'GGTACTGACCCTCAAAACAAATGCATGTG3' Reverse 5'GGAGGTCATGTTGATGTTCAGGTCTCC3' Arx Forward 5'GGCCGGAGTGCAAGAGTAAAT3' Reverse 5'TGCATGGCTTTTTCCTGGTCA3' Med12l Forward 5'CAGAATCAGGGGTTGGGGAC3' Reverse 5'GGATGTTCCAGACGCAAAGC3' Somatostatin Forward 5'ATGCTGTCCTGCCGTCTC3' Reverse 5'TTCTCTGTCTGGTTGGGCTC3' Sema5a Forward 5'CAGGACCCTTACTGTGGCTG3' Reverse 5'ATTTCTGGTCGGACAGGTGG3' Stmn2 Forward 5'CTTGAAGCCACCATCTCCCAT3' Reverse 5'CTCTTGAGACTTTCTTCGCTCCT3' Auh Forward 5'CTCTGCAAAAATGGGCCTGG3' Reverse 5'TAATCTCTGTGTCCCTCCTCCG3' Nos2 Forward 5'TCATTGGGCCTGGTACGGGCA3' Reverse 5'ACACCAAGCTCATGCGGCCTC3' Mrc1 Forward 5'AATGCTGACCTCCTGAGTGT3' Reverse 5'CAGTTCAGATACCGGAATGG3' Arg1 Forward 5'TGGGCAACCTGTGTCCTTTCTCCT3' Reverse 5'TTCCCCAGGGTCTACGTCTCGCA3' Gapdh Forward 5'CATGGCCTTCCGTGTTCCTA3' Reverse 5'CCTGCTTCACCACCTTCTTGAT3' Actb Forward 5'GGCTGTATTCCCCTCCATCG3' Reverse 5'CCAGTTGGTAACAATGCCATGT3' Hprt Forward 5'TCAGTCAACGGGGGACATAAA3' Reverse 5'GGGGCTGTACTGCTTAACCAG3'

Diabetes Page 62 of 74

Supplementary Table 4. Human primer sequences for qPCR analysis. Gene Primer Primer sequences MAFB Forward 5'AACTTTGTCTTGGGGCACAC3' Reverse 5'GGGACCTCTCGGTTCTCTCT3' GLUCAGON Forward 5'AAGCATTTACTTTGTGGCTGGATT3' Reverse 5'TGATCTGGATTTCTCCTCTGTGTCT3' INSULIN Forward 5'CTCTCTACCTAGTGTGCGGG3' Reverse 5'TGTTCCACAATGCCACGCTT3' IL6 Forward 5'AATTCGGTACATCCTCGACGG3' Reverse 5'GGTTGTTTTCTGCCAGTGCC3' PCSK1 Forward 5'ACACCGACCAGAGAATCACG3' Reverse 5'ATTTGGGTTTGCTTCCAGGG3' PCSK2 Forward 5'CCTGGGAGCTGGGATACACA3' Reverse 5'CCGAGGGTAAGGATAGGGGT3' MED12L Forward 5'TACCTCAAGCAACGGGCAAA3' Reverse 5'AACACGGCTCCCACTTCAAT3' SEMA5A Forward 5'TCCTCACCCTGCTCGTCTAT3' Reverse 5'CACCGAGTCGTACTTGTCCA3' STMN2 Forward 5'CTTCTCTCTCGCTCTCTCCG3' Reverse 5'TCCTTGTAGGCCATTGCTGT3' AUH Forward 5'ATCAGGAGTGAAGTCCCAGG3' Reverse 5'GCGAGTCCATCTATTGCTGC3' GAPDH Forward 5'ACCCACTCCTCCACCTTTGAC3' Reverse 5'TCCACCACCCTGTTGCTGTAG3' HPRT Forward 5'AGATGGTCAAGGTCGCAAGC3' Reverse 5'TCAAGGGCATATCCTACAACAAAC3'

Page 63 of 74 Diabetes

Supplementary Table 5. Primer sequences for in situ PCR. Gene Primer sequences Taqin sitummumiR155 RT 5'GTCGTATCCAGTGCAGGGTCCGAGG TATTCGCACTGGATACGACTCACACACCCCT3' Taqin sitummumiR155 Forward 5'TGCGGTTAATGCTAATTGTGATA3' Reverse 5'GTGCAGGGTCCGAGGT3' Taqin situcelmiR39 RT 5'GTCGTATCCAGTGCAGGGTCCGAGG TATTCGCACTGGATACGACCAAGC3' Taqin situcelmiR39 Forward 5'GCCCTCACCGGGTGTAAAT3' Reverse 5'GTGCAGGGTCCGAGGT3'

Supplementary Table 6. Primer sequences for luciferase reporter assays. Gene Primer sequences Il6 promoterPG04 Forward 5'AGTTACTTAAGCTCGGGCCC3' Reverse 5'TTGTTCTCGGTGGGCTTGGC3' Il6 promoter∆Mafb1 Forward 5'AGAAGTCTGTTTAAGTTACTGGGTGCCTAGAAGACTTGA3' Reverse 5'ACCCAGTAACTTAAACAGACTTCTTCCCTTTGGTTAG3' Il6 promoter∆Mafb2 Forward 5'TCAAGACATGCTCAAGTGGATCCAGACTTTTAAAGAAA3' Reverse 5'CTGGATCCACTTGAGCATGTCTTGATGGGAAAGAAAACT3'

Diabetes Page 64 of 74

Supplementary Experimental Procedures

Animals. Mir155–/– mice were crossed with Ldlr–/– or Apoe–/– mice (all on a C57BL/6J background, Jackson Laboratory, Bar Harbor, ME, USA) to obtain Mir155–/–Ldlr–/– mice and Mir155–/–Apoe–/– mice. Male Mir155–/–Ldlr–/– mice and Mir155+/+Ldlr–/– mice at 810 wks of age were fed a diabetogenic diet supplemented with cholesterol (DDC; 35.5% fat, 36.3% carbohydrates with 0.15% w/w total cholesterol, ssniff Spezialdiäten GmbH, Soest, Germany) or a normal diet (ND; 3.3% fat, ssniff Spezialdiäten). Animals were housed in cages with microisolator filter tops, maintained on a 12h light/dark cycle in a temperaturecontrolled room, and given free access to food and water. Aortic roots and epididymal white adipose tissue (eWAT) were either fixed using PAXgene tissue containers (Qiagen, Hilden, Germany) followed by paraffin embedding or immediately stored in RNAlater stabilization solution (Ambion, Thermo Fisher Scientific Inc., Waltham, MA, USA). An intraperitoneal glucose tolerance test was performed in mice after an overnight fast by injecting glucose (2 mg/g body weight, Gibco, Thermo Fisher Scientific) intraperitoneally. In addition, mice fed the ND were injected intravenously with lipopolysaccharide (LPS; 2 mg/kg; from Escherichia coli 055:B5, SigmaAldrich, Munich, Germany) or vehicle (phosphatebuffered saline). All animal experimental procedures were reviewed and approved by the local authorities in accordance with German animal protection laws.

Metabolic parameters. Glucose levels were determined in blood samples obtained from the tail vein after a 5–6 h fasting period using a glucometer (Roche Diagnostics GmbH, Basel, Switzerland). Cholesterol concentrations were measured by a fluorometric cholesterol assay kit (Cayman Chemical, Ann Arbor, MI, USA) in plasma, and in VLDL, LDL and HDL fractions separated by highperformance liquid chromatography using Sephadex column (SHIMADZU, Tokyo, Japan). Plasma triglyceride concentrations were measured using an enzymatic colorimetric triglyceride assay kit (GPOPAP method, Roche Diagnostics GmbH).

LaserMicrodissection. Murine pancreata were fixed using PAXgene tissue containers (Qiagen) and embedded in paraffin. Sections (7 µm thick) were collected on polyester membrane frames slides (0.9 µm thickness, Leica) and excised using a laser microdissection system (LMD7000, Leica).

Histology and immunostaining. Serial sections from PAXgene (Qiagen)fixed and paraffin embedded aortic roots (5 m thick; 4–5 sections per mouse) were stained with elastic van Gieson (EvG) stain. eWAT (7 µm thickness, 23 sections per mouse) sections were stained with hematoxylin and eosin. Atherosclerotic lesion size, necrotic core areas and adipocyte

Page 65 of 74 Diabetes

size were quantified using image analysis software (ImageJ 1.43n, NIH, USA). In addition, pancreas tissue was snap frozen on dry ice and embedded in O.C.T. (TissueTek O.C.T. Compound, Sakura Finetek, Staufen, Germany). Sections (10 µm thickness) were obtained using a cryotome (CD3050S, Leica Microsystem, Wetzlar, Germany).

Immunostaining was performed using primary antibodies against insulin (1:200, guinea pig polyclonal Ab, Cat. # ab7842, Abcam, Cambridge, UK), glucagon (1:1500, rabbit monoclonal Ab, clone EP3070, Cat. # ab92517, Abcam), IL6 (1:400, rabbit polyclonal Ab, Cat. # ab6672, Abcam), MafB (1:200, rabbit polyclonal Ab, Cat. # IHC00351, Bethyl Laboratories, TX, USA), cleaved caspase3 (1:200, rabbit polyclonal Ab, Cat. # 9661S, Cell Signaling Technology, MA, USA), Mac2 (Mac2; 1:200, rat monoclonal Ab, clone M3/38, Cat. # CL8942AP, Cedarlane, Burlington, Canada), αsmooth muscle actin (αSMA, 1:200, mouse monoclonal Ab, clone 1A4, Cat. # MO851, Dako, Glostrup, Denmark) and CD3 (1:100, rat monoclonal Ab, Cat. # MCA1477, AbD Serotec, Kidlington, UK) followed by fluorescently conjugated secondary antibodies. In addition, a biotinylated murine monoclonal antibody against oxidized phospholipids (1:200; mouse monoclonal E06, Cat. # 330002S, Avanti Polar Lipids, Inc. Alabaster, Alabama, USA) followed by a Dylight 549labeled streptavidin (1:200; Kirkegaard & Perry Laboratories, Gaithersburg, MD, USA) was used for oxLDL immunostaining. Cell nuclei were counterstained with DAPI (Vectashield, Vector Laboratories, Peterborough, UK). Nonspecific primary antibodies were used as negative controls (Santa Cruz Biotechnology, Santa Cruz, CA, USA). Digital images were acquired using a LeicaDM6000 B light microscope (LAS AF version 3.2.0.9652, Leica) connected to a CCD camera (DFC365FX, Leica).The size of the positively stained area (Mac2+, SMA+) per aortic lesion area (23 sections per mouse) and the positive cell number (Mac2+) per total adipocytes in each section (45 sections per mouse) were determined using image analysis software (ImageJ 1.43n, NIH, USA). In pancreatic islets, the percentage of positive cells (Insulin+, Glucaon+, MafB+) was calculated by dividing the number of positive cells in one islet by the total number of cells in this islet and multiplying this ratio by 100. At least 10 islets were analyzed per each section, and 23 sections were used for each individual. The background of the negative control staining defined the threshold for the positive staining.

RNA isolation and realtime PCR (qPCR) analysis. Total RNA was isolated using mirVana miRNA kit (Thermo Fisher Scientific), NucleoSpin microRNA kit (MachereyNagel GmbH & Co. KG, Düren, Germany) or PAXgene Tissue miRNA kit (Qiagen) and reverse transcribed (TaqMan microRNA reverse transcription kit or the highcapacity cDNA reverse

Diabetes Page 66 of 74

transcription kit, both from Thermo Fisher Scientific). MicroRNAs were quantitated by qPCR using the TaqMan Universal PCR Master Mix and TaqMan microRNA assays (both from Thermo Fisher Scientific). mRNA expression was measured using TaqMan Universal PCR Master Mix and TaqMan gene expression assays (both from Thermo Fisher Scientific) or the GoTaq qPCR Master Mix (Promega GmbH, Mannheim, Germany) and selfdesigned gene specific primers (SigmaAldrich) (Supplementary Table 3 and 4). All qPCRs were run on a 7900HT realtime PCR system (Thermo Fisher Scientific). Relative expression levels were normalized to a single or multiple reference genes (sno135, sno202, or U6 for miRNAs and Gapdh, Actb or Hprt1 for mRNAs), scaled to the sample with the lowest expression and logarithmically transformed (Log10) using QbasePLUS software (Biogazelle NV, Zwijnaarde, Belgium).

In situ reverse transcriptase PCR. Pancreatic cryosections were treated with DNase (Roche Diagnostics GmbH) for 16 h at 37°C. Onestep reverse transcriptase in situ PCR was performed using genespecific Taq in situ primers (SigmaAldrich) (Supplementary Table 5), SuperScript OneStep RTPCR with PlatinumTaq (Thermo Fisher Scientific), and digoxigenin11dUTPs (Roche Diagnotics GmbH) (1; 2). After stringent washing with SSC buffer and blocking of avidin/biotin binding sites (Blocking Kit, Vector Laboratories), sections were incubated with horseradish peroxidaseconjugated antidigoxigenin sheep F’ab fragments (Fab fragments from sheep, 1:100; Roche Diagnostics GmbH) for 1 h at 37°C. Probes were visualized using a tyramidebased amplification system (TSA Plus Biotin; PerkinElmer Inc., Waltham, MA, USA) and Dylight 549labeled streptavidin (1:200; Kirkegaard & Perry Laboratories) combined with glucagon immunostaining. The images were acquired using a LeicaDM6000 B light microscope (LAS AF version 3.2.0.9652, Leica) connected to a CCD camera (DFC365FX, Leica).

Isolation of pancreatic islets. Murine pancreatic islets were isolated by collagenase digestion and density gradient centrifugation. Briefly, collagenase P solution (4 mL, 1 mg/mL, Roche Diagnostics GmbH) was slowly injected into the common bile duct after occlusion of the ampulla in the duodenum. The pancreas was excised and digested in the collagenase P solution at 37°C for 20 min. Next, islets were purified by gradient separation using sodium diatrizoate (Histopaque 1119 and Histopaque 1077, SigmaAldrich), and handpicked and counted using an inverted microscope (IX50, Olympus optical Co., Tokyo, Japan). Purity of islets was assessed by diphenylthiocarbazone (dithizone or DTZ, SigmaAldrich) staining. After purification, islets were either maintained at 37°C in islet culture medium [Roswell Park

Page 67 of 74 Diabetes

Memorial Institute (RPMI) 1640 medium supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin, and 100 g/ml streptomycin (all from SigmaAldrich)] for recovery, or immediately lysed for RNA and protein extraction. Total RNA was isolated by NucleoSpin microRNA Kit (MachereyNagel) and proteins were extracted following lysis of the islets in RIPA buffer (SigmaAldrich) containing protease inhibitors (cOmplete, EDTAfree protease inhibitor cocktail tablet, Roche Diagnostics GmbH). In all experiments, islets from 3–5 mice were pooled for each biological replicate.

Cell culture and transfection. Mouse pancreatic insulinoma cells (MIN6, passage 18–24) (3) (kindly provided by Prof. Ingo Rustenbeck, University of Braunschweig, Germany), which contain primarily insulinproducing cells, but also glucagon and somatostatinproducing cells (4), were cultured in 6well or 12well plates using Dulbecco's modified Eagle's medium (DMEM; 25 mmol/L glucose, Thermo Fisher Scientific) supplemented with 10% FBS, 5 µl/L ßmercaptoethanol (both from SigmaAldrich), 2 mmol/L Lglutamine and 50 µg/mL

gentamicin (both from Thermo Fisher Scientific) and maintained at 37°C under 5% CO2. GLUTag cells, a murine enteroendocrine L cell line, were cultured in DMEM (5.5 mmol/L glucose, Thermo Fisher Scientific) supplemented with 10% FBS, 1% penicillin and streptomycin and 4 mmol/L glutamine (5). Human pancreatic islets (PELO Biotech, Martinsried, Germany) were cultured in extracellular matrix (ECM) precoated plates using Human Islets of Langerhans Cell Complete Growth medium (both from PELO Biotech)

supplemented with 10% FBS and 50 µg/mL gentamicin at 37°C under 5% CO2. The culture medium was changed every 48–72 h and cells were passaged at 70–80 % confluence.

Cells were transfected with locked nucleic acid (LNA)miR1555p inhibitors (50 nmol/L, miRCURY LNATM microRNA inhibitors; Exiqon, Vedbaek, Denmark), miR1555p mimics (15 nmol/L, mirVanaTM mimics; Thermo Fisher Scientific), 155/Mafb target site blockers (50 nmol/L miRCURY LNATM microRNA target site blockers; Exiqon), or scrambled controls (mirVana™ miRNA Mimic, Negative Control, Thermo Fisher Scientific; miRCURY LNATM microRNA inhibitor control and miRCURY LNATM microRNA target site blockers control, Exiqon) using Lipofectamine2000 (Thermo Fisher Scientific). Cells were harvested 24–48 h after transfection. Moreover, MIN6 cells and human islets were treated with LPS (5, 20, or 50 ng/mL, SigmaAldrich) or phosphatebuffered saline (PBS) for 6 h. In addition, MIN6 cells were treated nLDL (100 µg/mL), moxLDL (100 µg/mL) and PBS for 6 h or with neutralizing IL6 antibodies (200 g/mL, rat monoclonal Ab, Thermo Fisher Scientific.) and isotype control antibodies (Santa Cruz Biotechnology) for 24 h.

Diabetes Page 68 of 74

Cell purification by fluorescenceactivated cell sorting (FACS). Pancreatic islets were isolated from NDfed Mir155−/−Ldlr−/− mice and Mir155+/+Ldlr−/− mice (10–12 wks of age) and incubated at 37°C in islet culture medium for overnight recovery. Moreover, islets from Mir155+/+Ldlr−/− mice were transfected with miR1555p mimics (15 nmol/L) or control mimics using Lipofectamine2000 (all from Thermo Fisher Scientific). Murine βcells were purified by FACS as pervious described (6). In brief, pancreatic islets were washed and incubated in the HepesKrebsRinger bicarbonate (HKRB) buffer (119 mmol/L NaCl, 4.7 mmol/L KCl, 1.2 mmol/L MgSO4, 2 mmol/L CaCl2, 1.2 mmol/L KH2PO4, 25 mmol/L

NaHCO3 and 20 mmol/L HEPES, equilibrated with 5% CO2−95% O2, pH 7.4) with 2.8 mmol/L glucose and 0.2% BSA for 1 h. Islet cells were dissociated into single cell suspensions by incubating in Accutase solution (Thermo Fisher Scientific) at 37°C for 10 min, followed by gentle pipetting. The dispersed cells were filtered through 0.22 µm filters to remove cell debris and resuspended in HKRB buffer containing 2.8 mmol/L glucose and 0.2% BSA. The freshly dissociated islet cells were loaded to FACS Aria III (BD biosciences, San Jose, CA, USA) and illuminated with an Argon blue laser (488 nm). The fluorescence emission was collected at 510550 nm (FITC filter, 530/30 nm bandpass filter). The βcells were separated on the basis of high flavin adenine dinucleotide (FAD) fluorescence and high forward scatter (FSC) as compared to nonβcells. The nonβcells with low FAD and FSC were collected as αcells. The cell purity was verified by analysis of the Ins and Gcg mRNA expression using qPCR. Contamination of the βcell fraction with αcells and of the αcell fraction with βcells was < 3% and < 1.5 %, respectively.

Human pancreatic islets of Langerhans (PELO Biotech) were cultured in ECM precoated T75 flasks and transfected with hsamiR1555p mimics (15 nmol/L, mirVanaTM mimics) or scrambled controls (both from Thermo Fisher Scientific) at 70% confluence. After 48 h treatment, human βcells were purified by FACS Aria III (BD biosciences) by virtue of their high Zn2+ content using Newport Green DCF Diacetate (Thermo Fisher Scientific) (7; 8). The cells were incubated for 60 min in HKRB buffer containing 8 µmol/L Newport Green and 2.8 mmol/L glucose at 37°C. Excitation was performed with Argon blue laser (488 nm) and emission measured through a 530/30 nm bandpass filter. The purity of βcells was verified by immunofluorescence and insulinpositive cells accounted for 88 ± 5% of all cells in the βcell fractions. The FACSsorted non–βcells fraction was used as αcells (70 ± 10% αcells).

Measurements of insulin and GLP1 secretion from islets. Pancreatic islets isolated from NDfed Mir155−/−Ldlr−/− mice and Mir155+/+Ldlr−/− mice (10–12 wks of age) were maintained

Page 69 of 74 Diabetes

in islet culture medium for overnight recovery, followed by LPS (50 ng/mL) or PBS treatment for 6 h at 37°C. For the insulin secretion assay, each batch of 10 sizematched islets was transferred into a 24well plate and incubated in HKRB buffer with 2.8 mmol/L glucose and 0.2% BSA for stabilization for 1 h. The preincubation medium was then replaced with HKRB buffer supplemented with 5 mmol/L or 20 mmol/L glucose. After incubation for 1 h at 37°C, insulin concentrations were determined in the supernatant by ELISA (Crystal Chem Inc. Downers Grove, IL, USA)

To measure the GLP1 release from islets, pancreatic islets from NDfed Mir155+/+Ldlr–/– mice were incubated at 37°C in islet culture medium and transfected with miR1555p mimics or control mimics. After 48 h treatment, 20 sizematched islets were pooled and incubated in HKRB with 2.8 mmol/L glucose and 0.2% BSA for stabilization for 1 h. The medium was then replaced with HKRB buffer containing 30 mmol/L KCl (9) and supernatants were collected 2 h later. Proteins in the supernatant were concentrated using Amicon Ultra2 mL Centrifugal Filters (Merck Millipore).

To determine the effect βcell conditioned medium on the GLP1 secretion from human α cells, FACSsorted human βcells were treated with hsamiR1555p mimics (15 nmol/L, mirVanaTM mimics; Thermo Fisher Scientific) or control mimics. The supernatant of βcells was collected 48 h after transfection and filtered through 0.22 µm filters (Merck Millipore) to remove cell debris. Human αcells were stabilized for 1 h in HKRB with 2.8 mmol/L glucose, and then incubated with the supernatant collected from βcells. After 4 h, the supernatants form acells were collected and concentrated (Amicon Ultra2 mL Centrifugal Filters, Merck Millipore). The secretion of GLP1 from GLUTag cells treated with miR1555p mimics or control mimics were measured as pervious described (5). Secreted GLP1 concentrations and cellular GLP1 content were determined by using a GLP1 assay kit (ver. 2, Meso Scale Diagnostics, MD, USA).

Protein quantitation. Insulin (ultrasensitive mouse insulin ELISA kit, Crystal Chem) and IL6 proteins levels (Murine IL6 ELISA kit, RayBiotech, Inc., Norcross GA, USA) were determined by ELISA. Absorbance was measured at 450 nm by a microplate reader (Infinite F200 PRO, Tecan Trading AG).

Electrochemiluminescencebased GLP1 assay kits (ver. 2, lower limit of detection: 0.297 pmol/L) were used to quantify GLP1 protein levels in supernatants and cell lysates by a MESO QuickPlex SQ 120 reader (Meso Scale Diagnostics). Results were analyzed by MSD

Diabetes Page 70 of 74

DISCOVER WORKBENCH 4.0 software (all from Meso Scale Diagnostics) according to the manufacturer’s instructions.

In addition, xMAP technology was used to measure insulin, glucagon and GLP1 protein levels in mice (BioPlex Pro Mouse Diabetes assay Luminex kit, BioRad Laboratories, Inc. Hercules, CA, USA). Plasma insulin, glucagon and GLP1 levels were determined after a 5–6 h fasting period. The Luminex assays were run on a MAGPIX multiplex reader and the results were analyzed using xPONENT software v4.2 (both from Luminex Corporation, Austin, TX, USA).

The protein levels in lysates from MIN6 cells, GLUTag cells and islets were normalized to the total protein concentration determined by a modified Lowry assay (DC protein assay kit, BioRad Laboratories).

Endotoxin activity. Endotoxin activity was determined in serum, native LDL (nLDL) and mildly oxidized LDL (moxLDL) by the limulus amebocyte lysate (LAL) assay (Pierce™ LAL endotoxin quantitation kit, Thermo Fisher Scientific) according to the manufacturer’s instructions. Briefly, samples were heatshocked at 70°C for 15 min and incubated with LAL at 37°C for 10 min, followed by incubation with the chromogenic substrate at 37°C for 6 min. The absorbance was measured at 405 nm by a microplate reader (Infinite F200 PRO, Tecan Trading AG, Männedorf, Switzerland).

Preparation of moxLDL. Human LDL (1 mg/mL, Calbiochem, Merck Millipor, Darmstadt,

Germany) was incubated with 5 mol/L CuSO4 at 37°C for 4 h. LDL oxidation was stopped by adding 10 M EDTA and the LDL was passed through PD10 desalting column (GE Healthcare, Uppsala, Sweden). nLDL was treated in the same way as moxLDL except the addition of CuSO4. The protein concentration was measured using a DC protein assay kit (BioRad Laboratories) with bovine serum albumin as a standard. The level of oxidation was determined by spectrophotometric quantification of thiobarbituric acidreactive species formation (TBARS assay kit, Cayman Chemical, Michigan, USA) at 532 nm. moxLDL and nLDL were stored at 4°C and used for experiments within 14 d after preparation.

Global gene expression analysis. Total RNA was isolated from isolated murine pancreatic islets using the NucleoSpin microRNA Kit (MachereyNagel). The RNA quality was determined (Agilent 2100 Bioanalyzer, Agilent Technologies, Santa Clara, CA, USA) and samples with a RNA integrity number > 7.7 were included in the microarray analysis. Agilent SurePrint G3 Mouse Gene Expression microarrays (8x60K format, Agilent Technologies) were used in combination with a onecolor based hybridization protocol (IMGM Laboratories 31

Page 71 of 74 Diabetes

GmbH, Martinsried, Germany). Raw signals on the microarrays were scanned using the Agilent DNA Microarray Scanner (Agilent Technologies). Quantile normalization and analysis of the raw data were performed using GeneSpring GX 13.0 software (Agilent Technologies).

Pathway analysis. Microarray gene expression data were analysed by Ingenuity Pathway Analysis software (IPA, http://www.ingenuity.com/products/ipa, Qiagen) to predict upstream regulators of differentially expressed genes. Upstream regulator analysis was performed by first computing the enrichment of differentially expressed genes among the target genes of an upstream regulator using Fisher’s exact test. Moreover, the activation zscore of an upstream regulator was calculated by comparing the changes of the differentially expressed genes with the expected changes of these genes according to the IPA knowledge base. A positive zscore represents activation and a negative zscore represents inhibition by an upstream regulator.

MicroRNA target identification and quantification system. MIN6 cells or human islets (PELO Biotech) cultured in T75 flasks (Corning Inc., NY, USA) were cotransfected with miR1555p mimics (15 nM, Thermo Fisher Scientific) and the pMirTrap vector using the XfectTM microRNA transfection reagent in combination with Xfect polymer (all from Clontech, aintGermainenLaye, France). The pMirTrap vector expresses a DYKDDDDK tagged dominantnegative GW182 protein, which enables locking of the miRNA/mRNA complex in the miRNAinduced silencing complex (RISC) (10). MIN6 cells and human islets were harvested 24 h after the transfection, washed in icecold phosphatebuffered saline, and incubated in lysis buffer (MirTrap System) supplemented with protease inhibitors (Roche Diagnostics). The cell lysates were centrifuged and part of the input RNA was extracted from the supernatant using the NucleoSpin RNA XS kit (MachereyNagel). AntiDYKDDDDK beads (Clontech) were washed twice with 1× lysis/wash buffer containing 1 mmol/L DTT, 0.1 unit/L RNase inhibitor and protease inhibitors (Roche Diagnostics GmbH), and blocked for 3 h at 4°C with 1.25 mg/ml tRNA solution and 1.25 mg/ml bovine serum albumin. Immunoprecipitation of the miRNA/mRNA complexes was performed by incubation of the cell lysates with antiDYKDDDDK beads for 2 h at 4°C and centrifugation at 1,000 rpm for 1 min. RNA from the input and the immunoprecipitated samples was reverse transcribed using the highcapacity cDNA reverse transcription kit (Thermo Fisher Scientific), followed by qPCR with genespecific primers (Supplementary Table 3 and 4) and SYBR Green PCR Master Mix (Promega GmbH, Mannheim, Germany) using a 7900HT fast realtime PCR system (Thermo Fisher Scientific). Transfection efficiency was determined by a control

Diabetes Page 72 of 74

transfection using miR132 mimics, the pMirTrap positive control vector, which expresses an AcGFP1 fluorescein protein containing a miR132 recognition element in the 3’UTR or the empty pMirTrap vector (all from MirTrap System). Fold enrichment of the AcGFP1 control or miR1555p predicted target genes in the GW182immunoprecipitates was normalized to Gapdh according to the manufacturer’s protocol.

Luciferase reporter assays. Putative binding sites for MafB in the Il6 promoter were predicted at position 30012038 (Il6 promoterMafb1) and at position 30012863 (Il6 promoter Mafb2) by the TRANSFAC database (version 7.0, Qiagen). The Il6 promoterMafb1 and Il6 promoterMafb2 sequences were mutated using QuickChange sitedirected mutagenesis kit (Agilent Technologies), specific primers (SigmaAldrich) (Supplementary Table 6), Q5 high fidelity DNA polymerase (New England Biolabs, Ipswich, MA, USA) and a PCR cycler (Mastercycler nexus, Eppendorf, Hamburg, Germany). The PCR product was treated with DpnI restriction enzyme (Thermo Fisher Scientific) to digest the parental DNA template. The vector DNA containing the desired mutation was transformed into XL10Gold Ultra component cells (Agilent Technologies) and the plasmid was isolated using the EndoFree Plasmid Maxi Kit (Qiagen).

MIN6 cells were cotransfected with the Gaussia luciferase (GLuc)secreted alkaline phosphatase (SEAP) dualreporter vector pEZXPG04 containing the promoter region of mouse Il6 gene (Il6 promoter vector, GeneCopoeia, Vienna, Austria) or luciferase reporter constructs harboring sitedirected mutations in the predicted Mafb binding sites of the Il6 promoter region (Il6 promoter∆Mafb1 and Il6 promoter∆Mafb2) together with miR1555p inhibitors or nontargeting LNA oligonucleotides (control inhibitors) using Lipofectamine 2000 (Thermo Fisher Scientific). The empty vector (dualreporter vector without the Il6 promoter sequence, GeneCopoeia) was used as control. The GLuc and SEAP activities were assayed by microplate reader (Infinite F200 PRO, Tecan Trading AG) 48 h after the transfection using the SecretePair Dual Luminescence Assay Kit (GeneCopoeia). The luminescence intensities of Gaussia luciferase were normalized to the activity of SEAP.

In vivo TSB treatment. 10weekold Ldlr–/– mice fed a normal diet were injected intravenously via the tail vein with 155/Mafb TSBs (5'TTAATGCAGATTTTCG3') or control TSBs (5'GCTCCCTTCAATCCAA3') (each 0.4 mg/20 g per injection; miRCURY LNATM Target Site Blocker, in vivo use; Exiqon). The fasting glucose levels (6h fast) and body weights were measured every 3 d. Blood cell numbers, and plasma insulin, glucagon and GLP1 levels were determined after a 5–6 h fast 21 d after the treatment. Tissues were

Page 73 of 74 Diabetes

harvested after 21 d and either embedded in TissueTek O.C.T. Compound (Sakura Finetek) and immediately frozen on dry ice for cryostat sections or preserved in RNAlater (Thermo Fisher Scientific) for RNA purification.

Statistical analysis. Sample size (number of mice) was determined on the basis of our previous studies(2; 11; 12). Mice were not randomized to experimental groups. For most mouse experiments, the investigators were blinded when assessing the results without knowing the mouse genotypes. In some cases, selected samples were excluded from specific analyses because of technical flaws during sample processing or data acquisition. The number of biological (nontechnical) replicates for each experiment is indicated in the figure legends. Data represent the mean ± SEM. Statistical analysis of microarray data was performed by a modified ttest using GeneSpring software (GX13, Agilent Technologies). Student’s ttests and oneway ANOVAs followed by the NewmanKeuls posthoc test were used for statistical comparisons between groups using Prism 6 software (GraphPad). The variance is similar between the groups that are being statistically compared. P < 0.05 was considered statistically significant.

Diabetes Page 74 of 74

Supplementary References

1. Nuovo GJ, Elton TS, NanaSinkam P, Volinia S, Croce CM, Schmittgen TD: A methodology for the combined in situ analyses of the precursor and mature forms of microRNAs and correlation with their putative targets. Nat Protoc 2009;4:107115 2. Wei Y, Zhu M, CorbalanCampos J, Heyll K, Weber C, Schober A: Regulation of Csf1r and Bcl6 in macrophages mediates the stagespecific effects of microRNA155 on atherosclerosis. Arterioscler Thromb Vasc Biol 2015;35:796803 3. Miyazaki J, Araki K, Yamato E, Ikegami H, Asano T, Shibasaki Y, Oka Y, Yamamura K: Establishment of a pancreatic beta cell line that retains glucoseinducible insulin secretion: special reference to expression of glucose transporter isoforms. Endocrinology 1990;127:126 132 4. Nakashima K, Kanda Y, Hirokawa Y, Kawasaki F, Matsuki M, Kaku K: MIN6 is not a pure beta cell line but a mixed cell line with other pancreatic endocrine hormones. Endocr J 2009;56:4553 5. Kahles F, Meyer C, Mollmann J, Diebold S, Findeisen HM, Lebherz C, Trautwein C, Koch A, Tacke F, Marx N, Lehrke M: GLP1 secretion is increased by inflammatory stimuli in an IL6dependent manner, leading to hyperinsulinemia and blood glucose lowering. Diabetes 2014;63:32213229 6. Van De Winkel M, Pipeleers D: Autofluorescenceactivated cell sorting of pancreatic islet cells: purification of insulincontaining Bcells according to glucoseinduced changes in cellular redox state. Biochem Biophys Res Commun 1983;114:835842 7. Lukowiak B, Vandewalle B, Riachy R, KerrConte J, Gmyr V, Belaich S, Lefebvre J, Pattou F: Identification and purification of functional human betacells by a new specific zinc fluorescent probe. J Histochem Cytochem 2001;49:519528 8. Parnaud G, Bosco D, Berney T, Pattou F, KerrConte J, Donath MY, Bruun C, Mandrup Poulsen T, Billestrup N, Halban PA: Proliferation of sorted human and rat beta cells. Diabetologia 2008;51:91100 9. Chimerel C, Emery E, Summers DK, Keyser U, Gribble FM, Reimann F: Bacterial metabolite indole modulates incretin secretion from intestinal enteroendocrine L cells. Cell Rep 2014;9:12021208 10. Cambronne XA, Shen R, Auer PL, Goodman RH: Capturing microRNA targets using an RNAinduced silencing complex (RISC)trap approach. Proc Natl Acad Sci USA 2012;109:2047320478 11. NazariJahantigh M, Wei Y, Noels H, Akhtar S, Zhou Z, Koenen RR, Heyll K, Gremse F, Kiessling F, Grommes J, Weber C, Schober A: MicroRNA155 promotes atherosclerosis by repressing Bcl6 in macrophages. J Clin Invest 2012;122:41904202 12. Wei Y, NazariJahantigh M, Chan L, Zhu M, Heyll K, CorbalanCampos J, Hartmann P, Thiemann A, Weber C, Schober A: The microRNA3425p fosters inflammatory macrophage activation through an Akt1 and microRNA155dependent pathway during atherosclerosis. Circulation 2013;127:16091619