Defective FXR-SHP Regulation in Obesity Aberrantly Increases miR-802 Expression, Promoting Insulin Resistance and Fatty Liver

Obesity substantially increases the risk for many human diseases, ranging from type 2 diabetes, nonalcoholic fatty liver disease (NAFLD), and cardiovascular disease, to infertility and certain types of cancer (1). The global obesity pandemic and the urgent need for effective drugs to treat obesity-associated metabolic disorders have greatly increased research interest in farnesoid X receptor (FXR; also known as NR1H4), the primary bile acid nuclear receptor and a key regulator of hepatic lipid and glucose metabolism (2–6).

FXR is activated by bile acids in organisms in the fed state and maintains bile acid homeostasis by transcriptional control of numerous genes, including small heterodimer partner (SHP; also known as NR0B2), an orphan nuclear receptor that acts as a corepressor (4,7,8). In addition to bile acid metabolism, FXR-SHP also plays an important role in regulation of glucose and lipid metabolism (9,10) and hepatic autophagy (11,12). Activation of FXR improves hyperglycemia and hyperlipidemia in diabetic mice (13); conversely, FXR-knockout (KO) mice have hyperglycemia and dyslipidemia (14). Interestingly, metabolic benefits of vertical sleeve gastrectomy on body weight and glucose tolerance are substantially reduced by the absence of FXR (5,15). These previous studies demonstrated the potential of FXR for treating metabolic disorders. An FXR agonist, obeticholic acid (OCA), recently approved for treating primary biliary cholangitis (16), has potential efficacy in other metabolic diseases (5,17) and OCA treatment of NAFLD, nonalcoholic steatohepatitis, lipodystrophy, liver cirrhosis, and fibrosis is being evaluated in active clinical trials.

MicroRNAs (miRNAs) have also emerged as promising potential targets for treating obesity-related diseases (18). miRNAs are posttranscriptional gene silencers that partially base pair to 3′-untranslated regions of their target mRNAs and inhibit translation and/or destabilize the mRNAs. Consistent with their critical functions in diverse biological pathways, expression of numerous miRNAs is aberrantly increased or decreased in diseases (19–25). For example, obesity-induced overexpression of miR-802 impairs hepatic insulin sensitivity and glucose metabolism (23). However, the mechanisms by which expression of miR-802, as well as many other miRNAs, is altered in obesity are poorly understood. In published miRNA expression profile studies, hepatic levels of miR-802 are highly elevated in both FXR-KO (26) and liver-specific SHP-downregulated (SHP-LKD) mice (27), suggesting an FXR-SHP cascade may have a role in the regulation of miR-802 gene expression.

In this study, we show that FXR-SHP and aromatic hydrocarbon receptor (AHR) reciprocally regulate hepatic expression of miR-802 physiologically and that this regulation is abnormal in obese mice. Remarkably, in patients with NAFLD, occupancy of SHP and AHR at the miR-802 gene promoter is altered in a manner consistent with increased hepatic miR-802 levels, suggesting the FXR-SHP–miR-802 axis is dysregulated in obese humans as well as in mice.

Antibodies for SHP (sc-30169), AHR (sc-133088), and RNA polymerase II (RNA pol II) (sc-9001) were purchased from Santa Cruz Biotech; for ARNT (NB100–110), purchased from NOVUS; for KLF4 (AF3158), purchased from R&D System; for histone H3 (ab1791), purchased from Abcam; for H3K9/14-Ac (#06–599), purchased from Millipore; for β-actin (#4970), p-AKT (#9271), AKT (#9272), p-GSK (#9323), GSK (#9315), and FOXO1 (#2880), purchased from Cell Signaling; and for HNF-1β (SAB2501022), purchased from Sigma. GW4064 (#2473) was obtained from Tocris Bioscience and OCA from MedChemExpress. ON-TARGET plus mouse siRNAs for AHR, KLF4, and FOXO1 were purchased from Dharmacon; purified adenovirus expressing miR-802 or antisense miR-802 was purchased from ABM; lentivirus-expressing miR-802-sponge from was purchased VectorBuilder; and AAV8-TBG-Cre and control AAV8-TBG-GFP was purchased from VectorBiolabs.

FXR-KO and SHP-KO mice have been described previously (14,28). FXR–liver-specific KO mice were generated by breeding FXR floxed mice with Albumin-Cre mice (The Jackson Laboratory). For SHP-LKD mouse studies, we injected 8-week-old, male SHP floxed mice via the tail vein with AAV8-TBG-Cre or control AAV8-TBG-GFP (1–2 × 10¹¹ genome copies), as described previously (29,30). In C57BL/6 or SHP-KO mice, which had been fasted overnight, FXR signaling was activated by either intraperitoneal injection with GW4064 (30 mg/kg in corn oil) or by feeding 0.5% cholic acid–containing chow 6 h before sacrifice or by administration of 10 mg/kg OCA daily for 1 week by oral gavage. To induce dietary obesity, mice were fed a high-fat diet (TD8813; Harlan Teklad) or a high-fat diet with 25% fructose in water (high fat/high fructose) for 9–12 weeks.

Dietary obese mice were injected via the tail vein with adenovirus expressing mature miR-802 or antisense RNA for miR-802 (0.5–1.0 × 10⁹ active viral particles) and were treated with vehicle or OCA by oral gavage daily for 2 weeks. To examine insulin signaling, these mice were treated additionally with insulin (0.25 units/kg) for 10 min and then, p-AKT and p-GSK levels in liver and soleus muscle extracts were detected by IB. All experiments were approved by the Institutional Animal Care and Use and Biosafety Committees of the University of Illinois at Urbana-Champaign.

For the glucose tolerance test, the mice were fasted overnight and injected i.p. with d-glucose (2 g/kg body wt), and for the insulin tolerance test, the mice were fasted for 6 h and injected i.p. with insulin (1 units/kg). Blood glucose levels were determined using a portable Accu-Chek glucose meter (Roche). Plasma insulin levels were determined using a mouse insulin ELISA Kit (Crystal Chem).

Primary mouse hepatocytes (PMHs) were isolated by collagenase (0.8 mg/mL) (Sigma) perfusion. Hepatocytes were filtered through a cell strainer, centrifuged through 45% Percoll (Sigma), and cultured in M199 medium containing 10% FBS. PMHs were infected with Ad-miR-802 and, after 24 h, were incubated overnight in serum-, glucose-, and phenol red–free DMEM containing 20 mmol/L sodium lactate and 2 mmol/L sodium pyruvate. Fresh medium containing palmitic acid (300 μmol/L) and OCA (10 μmol/L) was added, and 24 h later, glucose concentrations in the medium were measured with a glucose assay kit (Sigma).

Liver specimens from 15 normal organ donors or 15 patients with NAFLD-associated steatosis were obtained from the Liver Tissue Procurement and Distribution System. The samples were unidentifiable; thus, ethical approval was not required. Hepatic miR-802 levels and mRNA levels of genes were measured by RT-quantitative PCR (qPCR) and the occupancies of SHP, AHR, and RNA pol II at the miR-802 promoter were determined by chromatin immunoprecipitation (ChIP).

RNA was isolated from mouse liver or Hepa1c1c7 cells using miRNeasy or RNeasy mini kits (Qiagen) for miRNAs or mRNAs, respectively. Levels of mRNAs or miRNAs were determined by RT-qPCR using primers listed in Supplementary Table 1. Levels of miR-802 were normalized to small nucleolar RNA, and mRNA amounts were normalized to 36B4 mRNA.

Mice were fasted overnight, then refed for 1–6 h, and ChIP assays were performed as described previously (11,30). For re-ChIP, chromatin samples were immunoprecipitated with the first antibody, washed, eluted with 10 mmol/L dithiothreitol, diluted 20× with 20 mmol/L Tris-HCl, at pH 8.0, 150 mmol/L NaCl, 2 mmol/L EDTA, and 1% Triton X-100, and immunoprecipitated with the second antibody. Enrichment of target-gene sequences was determined by qPCR. Sequences of primers are listed in Supplementary Table 2.

The SHP binding region in the miR-802 promoter (−3/−252) was amplified by PCR from mouse genomic DNA and was inserted into pGL3 luciferase reporter vector (Promega). Hepa1c1c7 cells were cotransfected with 200 ng of luciferase reporter plasmids, 100 ng of cytomegalovirus β-galactosidase plasmid, and expression plasmids for ARNT (50 ng), AHR (5 ng), or constitutively active form of AHR (CA-AHR; 5 ng), and for SHP (5–50 ng). Two days after transfection, luciferase activities were determined and normalized to β-galactosidase activities.

Data were analyzed by the two-tailed Student t test, one- or two-way ANOVA with false discovery rate test for single or multiple comparisons, as appropriate, using Prism (GraphPad Software). P < 0.05 was considered statistically significant.

No data sets were generated during this study. The SHP ChIP-sequencing data set in mouse liver, which was analyzed, is available in the National Center for Biotechnology Information Gene Expression Omnibus database (accession number, GSE74913). Other data and resources are available on reasonable request from the corresponding author.

To explore the role of FXR in miR-802 expression, miR-802 levels were measured in FXR-KO mice. Hepatic miR-802 levels were increased approximately twofold in FXR-KO and FXR–liver-specific KO mice compared with control mice (Fig. 1A and B). Conversely, activation of FXR signaling by treatment with an agonist, GW4064 or OCA, or feeding chow containing cholic acid all reduced hepatic miR-802 levels by approximately 50% (Fig. 1C). These results suggest FXR inhibits hepatic expression of miR-802.

FXR indirectly represses expression of genes in part through induction of SHP, which acts as a corepressor (4,7,8). Hepatic miR-802 levels were increased approximately twofold in SHP-KO and in SHP-LKD mice (Fig. 1D and E). Importantly, the repression of miR-802 mediated by GW4064-activated FXR was not observed in SHP-KO mice (Fig. 1F), indicating the FXR-mediated repression of miR-802 is SHP dependent. These results suggest hepatic expression of miR-802 gene is likely inhibited by FXR-SHP.

To examine the mechanism of SHP repression of miR-802, we first examined whether SHP binds to miR-802. Analysis of published liver ChIP-sequencing data in mice (31) revealed SHP binding at miR-802 (Supplementary Fig. 1), which was confirmed by ChIP, in which SHP occupancy at the miR-802 promoter was similar to that at Cyp7a1, a well-known direct SHP target (4), and no binding was detected at a control region (Fig. 2A–C).

SHP does not have a DNA-binding domain and is recruited to genes by interaction with other transcription factors (8,27,31–33). In motif analysis, several sites for transcription factors, including KLF4 and AHR with the highest scores, were detected within the SHP binding region at miR-802 (Fig. 2D). We initially focused on KLF4 and AHR, because they are known SHP-interacting proteins (27,32). In Hepa1c1c7 cells, downregulation of AHR, but not KLF4, decreased miR-802 levels (Fig. 2E). In luciferase reporter assays, overexpression of CA-AHR (32,34) increased the activity of the miR-802 promoter threefold, and coexpression of its DNA binding partner, ARNT, further increased the activity, which was inhibited by expression of SHP (Fig. 2F). The increases in activity were not observed if the AHR binding site was mutated (Fig. 2F). These results suggest SHP inhibits AHR transactivation of miR-802.

To examine whether AHR and SHP reciprocally regulate miR-802 expression physiologically, we used a fasting/feeding model in mice. In liver re-ChIP assays, occupancy of SHP was increased in AHR-bound chromatin 2 h after refeeding, indicating co-occupancy of AHR and SHP at the miR-802 promoter (Fig. 3A). Binding of RNA pol II and levels of a gene-activation histone mark, acetylated H3K9/14-Ac, were decreased (Fig. 3B), consistent with SHP repression of miR-802 (Fig. 2).

The nuclear localization of AHR is increased early after feeding by insulin-activated AKT/PKB, whereas nuclear levels of SHP are substantially increased by FGF15/19 signaling in the late fed state (32). Refeeding of mice after fasting increased occupancy of both AHR and ARNT at miR-802 as early as 1 h after feeding, and occupancy remained elevated up to 4 h before decreasing at 6 h (Fig. 3C). In contrast, the occupancy of SHP at miR-802 did not increase until 2 h and remained elevated up to 4–6 h (Fig. 3C). Consistent with these results, hepatic levels of both primary miR-802 and mature miR-802 transiently increased 1 h after feeding, decreased by 2 h, and decreased to levels lower than the fasting levels by 6 h (Fig. 3D). These results support a temporal model in which AHR activates transcription early after feeding by binding to the miR-802 promoter, and SHP later binds to repress it. Consistent with these findings, the decrease in hepatic miR-802 levels in C57BL/6 mice refed for 6 h was not observed in SHP-KO mice (Fig. 3E). Furthermore, occupancy of SHP at the miR-802 promoter was increased in control mice, but not in SHP-KO mice, as expected, whereas the occupancy of AHR and ARNT was detected 6 h after refeeding in both control and SHP-KO mice (Fig. 3F). These results suggest SHP inhibits the AHR transactivation of miR-802 expression in a physiologically relevant manner.

Kornfeld et al. (23) reported that miR-802 levels in the liver are increased in obese mice, impairing glucose metabolism. We confirmed that hepatic miR-802 levels are elevated in diet-induced obese mice (Fig. 4A) and that downregulation of miR-802 in obese mice improved glucose tolerance, decreased liver triglyceride levels and expression of a lipogenic gene, Fasn, but increased expression of a fatty acid β-oxidation gene, Mcad, and decreased expression of direct SHP targets in bile acid and cholesterol metabolism: Cyp7a1, Hmgcr, and Insig1 (31) (Supplementary Fig. 2).

Conversely, overexpression of miR-802 in chow-fed mice resulted in increased glucose intolerance and liver triglyceride levels and increased mRNA levels of Fasn and gluconeogenic genes G-6-pase and Pck1 but decreased levels of Mcad (Supplementary Fig. 3). For genes involved in bile acid metabolism, mRNA levels of Shp and Cyp7a1 were modestly changed. These results, together with those of previous studies (23), strongly suggest that obesity-induced miR-802 impairs glucose metabolism and promotes fatty liver.

Because AHR and SHP reciprocally regulate miR-802 expression physiologically (Figs. 2 and 3), we next tested if the AHR-SHP regulation is abnormal in obesity, which may contribute to increased miR-802 expression (23). Consistent with elevated miR-802 levels (Fig. 4A), SHP occupancy at miR-802 was substantially reduced in obese mice, whereas AHR occupancy was modestly increased and RNA pol II occupancy was increased (Fig. 4B). These results suggest reduced binding of SHP impairs SHP repression of AHR-mediated transactivation, resulting in increased miR-802 expression in obese mice.

To examine further the reciprocal regulation of miR-802 by AHR and SHP in obese mice, we examined whether overexpression of CA-AHR increases hepatic miR-802 levels and whether coexpression of SHP attenuates the AHR-mediated increase (Fig. 4C and D). Hepatic miR-802 levels, which are already elevated in obese mice, were only modestly increased by expression of CA-AHR, and coexpression of SHP blocked the increase (Fig. 4E). Conversely, in PMHs that had been treated with palmitic acid to mimic obesity conditions, miR-802 levels were increased approximately twofold, and downregulation of AHR, but not KLF4, blocked the increase (Supplementary Fig. 4). These results suggest AHR and SHP reciprocally regulate miR-802 expression in obesity.

Because obesity-induced miR-802 impairs glucose metabolism by silencing Hnf-1β (23), we next examined effects of AHR and SHP on the expression of Hnf-1β and gluconeogenic genes. Forced expression of AHR increased mRNA levels of G6pc and Pck1 and those of Hnf-4 only modestly, and coexpression of SHP increased mRNA (Fig. 4F) and protein (Supplementary Fig. 5) levels of Hnf-1β and decreased mRNA levels of G6pc, Pck1, Foxo1, Hnf-4, and Crtc2, and decreased mRNA levels of the lipogenic genes Fasn and Srebp-1 (Fig. 4F). These results strongly suggest changes in SHP and AHR binding in obesity alter miR-802 expression and, consequently, expression of gluconeogenic and lipogenic genes.

OCA is an FXR agonist in clinical trials for treating various liver diseases, including NAFLD (5,6,17). Because activation of FXR inhibits miR-802 expression, we examined if OCA-mediated beneficial effects are diminished by overexpression of miR-802 and enhanced by downregulation of miR-802. Obese mice fed a high-fat and high-fructose diet were injected with adenovirus expressing miR-802, antisense RNA for miR-802, or control virus, and then were treated daily for 2 weeks with OCA (Fig. 5A).

Administration of OCA with or without downregulation of miR-802 had similar effects. In mice fed a high-fat and high-fructose diet, OCA treatment decreased miR-802 levels approximately twofold, similar to levels in lean mice. Adenoviral-mediated expression of antimiR-802 did not further decrease the miR-802 levels (Fig. 5B) although infection with this virus effectively decreased miR-802 levels in obese mice (Supplementary Fig. 6). Body weight was modestly decreased in OCA-treated groups, although food intake was not changed (Supplementary Fig. 7). OCA treatment improved glucose and insulin tolerance (Fig. 5C and D) and decreased fasting glucose and insulin levels (Fig. 5E and F). Consistent with these results, the HOMA–insulin resistance index was decreased with OCA treatment (Fig. 5G). In each case, overexpression of miR-802 reversed the OCA-mediated beneficial effects, whereas expression of antimiR-802, as expected, had little effect because miR-802 levels were not changed (Fig. 5C–G). Hepatic mRNA (Fig. 5H) and protein (Supplementary Fig. 8) levels of Hnf-1β and mRNA levels of the gluconeogenic genes Pck1 and G6pc were increased by OCA, and the increases were reversed by overexpression of miR-802 (Fig. 5H). Consistent with these results, OCA treatment in PMHs reduced glucose production, and overexpression of miR-802 reversed the reduction (Fig. 5I).

Because the OCA-mediated beneficial effects on glucose metabolism were blocked by overexpression of miR-802 (Fig. 5B–I), we next examined insulin signaling. Administration of OCA resulted in both increased basal and insulin-stimulated phosphorylation of AKT and GSK in liver and muscle, and the OCA-mediated increases in p-AKT and p-GSK levels were reduced by miR-802 overexpression (Fig. 5J) but were not affected by expression of antisense miR-802. Similarly, in PMHs, overexpression of miR-802 markedly reduced p-AKT and p-GSK levels (Supplementary Fig. 9). These results, together, indicate that activation of FXR by OCA improves insulin sensitivity and glucose metabolism in obese mice and that these OCA-mediated beneficial effects are largely abolished by overexpression of miR-802.

Hepatic insulin resistance is manifested selectively in obese animals, such that insulin does not inhibit hepatic gluconeogenesis but continues to stimulate lipogenesis (35,36). In obese mice, administration of OCA resulted in a dramatic decrease in liver size (Fig. 6A) and the ratio of liver to body weight (Fig. 6B), an indicator of fatty liver. Indeed, OCA treatment reduced neutral lipid levels in liver sections (Fig. 6C) and decreased liver triglyceride levels (Fig. 6D). Consistent with these results, mRNA levels of lipogenic genes Fasn and Srebp-1 and the fat transporter Cd36 were all decreased, whereas those of β-oxidation genes, such as Cpt1 and Mcad, were modestly increased by OCA administration (Fig. 6E). Changes in protein levels of Fasn were consistent with those of mRNA levels (Supplementary Fig. 8). Furthermore, OCA treatment led to increased Shp and decreased Cyp7a1 mRNA levels (Supplementary Fig. 10). Each of these OCA-mediated effects that improved fatty liver was reversed by overexpression of miR-802 (Fig. 6A–E) and was not further improved by expression of antisense miR-802. Collectively, these results demonstrate activation of FXR by OCA ameliorates fatty liver in obese mice and that the OCA-mediated beneficial effects on hepatic lipid and glucose metabolism are largely dependent on decreased miR-802 levels.

To test if abnormal SHP and AHR regulation may also occur in obese humans, we examined hepatic miR-802 levels and occupancy of SHP and AHR at miR-802 in patients with NAFLD-associated steatosis. Levels of hepatic miR-802 were substantially elevated in liver samples from the patients with NAFLD (Fig. 7A). Consistent with increased miR-802 levels, hepatic mRNA levels of HNF-1β were decreased and those of PCK1 and G6PC were increased (Fig. 7B). Recently, we showed that hepatic expression of AHR is increased and that of SHP is not changed in patients with NAFLD compared with normal individuals (8,32), but nuclear levels of SHP are substantially reduced in the patients (37). Indeed, mRNA levels of SHP were not changed, and those of FXR were modestly decreased in the patients with NAFLD (Supplementary Fig. 11).

To determine if changes in SHP and/or AHR function might underlie the increases in miR-802 expression in the patients, occupancies of these factors at miR-802 were examined by ChIP. Consistent with reduced nuclear levels of SHP in livers of patients with NAFLD (37), SHP occupancy was substantially reduced, AHR occupancy was only modestly increased, RNA pol II occupancy was robustly increased (Fig. 7C), and miR-802 levels were increased (Fig. 7A). These results demonstrate that hepatic expression of miR-802 is increased in patients with NAFLD and suggest that changes in SHP and AHR occupancy may underlie the increased miR-802 levels in patients with NAFLD, as was observed in obese mice.

Numerous miRNAs, including miR-802, are aberrantly upregulated in obesity, contributing to obesity-associated metabolic abnormalities (19–23). Although most studies have focused on miRNA function and therapeutic potential, few have examined the mechanisms for their aberrant expression. In this study, we have shown that a pathway involving FXR and SHP, and a transcription activator, AHR, reciprocally regulated hepatic expression of the miR-802 gene physiologically, and this regulation is abnormal in obesity, leading to elevated miR-802 levels (Fig. 7D).

Our findings in liver ChIP studies in patients with NAFLD and in obese mice suggest the elevated levels of miR-802 in obesity are primarily due to decreased repression of miR-802 expression by SHP, with lesser contributions from increased activation by AHR. The reduced repression of SHP is likely associated with reduced FXR functions, because the global FXR cistrome was substantially reduced in obese mice compared with lean mice (38). FXR induces Shp expression but also enhances SHP repression function (27,31–33) through induction of Lsd1, a SHP-interacting epigenomic corepressor (8,12), and FGF15 signaling genes, including Fgf15 and the coreceptor bKL (39). Notably, nuclear localization of SHP is substantially reduced in obese animals, in part due to impaired FGF15 signaling (8,32,37). Thus, decreased global FXR function in obesity likely leads to decreases in both SHP nuclear levels and repression function. In this study, the inhibition of miR-802 by FXR was dependent on SHP, because similar elevated levels of miR-802 were observed in FXR-KO and SHP-KO mice, and FXR inhibition of miR-802 was absent in SHP-KO mice. Together, these findings suggest decreased transcription function of FXR-SHP likely plays an important role in elevated miR-802 expression in obesity.

We identified AHR in this study as a transcription activator of miR-802. AHR is activated by environmental toxicants to regulate xenobiotic metabolism (34), but it also regulates energy metabolism (32,40). The nuclear localization of AHR is dramatically increased by binding of exogenous ligands (34), but it also is increased by insulin-PKB signaling upon feeding (32). We observed that early after feeding, AHR and ARNT occupancy at miR-802 increased to a maximum by 1 h, which is consistent with the initial increase in expression of miR-802 after eating. Occupancy of SHP at miR-802 increased later, reaching a maximum by 2–3 h after eating, which is consistent with the decrease in miR-802 expression after the initial increase. Multiple signaling pathways are involved in the response to feeding and are abnormal in obesity. Thus, in this study, we focused on the FXR-SHP-AHR regulation of miR-802 but cannot rule out the possibility that other obesity-associated factors and/or aberrant metabolic signaling also increase levels of miR-802. Recently, Foxo-1 was shown to increase pancreatic expression of miR-802 in obesity, which is associated with impaired insulin secretion (41). However, downregulation of Foxo1 in hepatocytes only modestly and statistically insignificantly reduced miR-802 levels (Supplementary Fig. 12), suggesting expression of miR-802 is likely tissue specific.

FXR is a promising therapeutic target for obesity-associated NAFLD and diabetes (5,6,17). In this study, treatment of obese mice with OCA for 2 weeks reduced miR-802 levels similar to those in lean mice and had metabolic benefits in hepatic glucose and lipid metabolism. However, these OCA-mediated beneficial effects were substantially reduced by overexpression of miR-802. FXR mediated its beneficial effects on energy metabolism and protection against liver injury, at least in part, via the SHP-mediated repression of miR-34a (19–21) and miR-210 (27), respectively. For miR-34a, FXR-induced SHP is recruited to the miR-34a promoter and inhibits the transactivation of p53, but this regulation is abnormal in obesity, resulting in increased miR-34a expression (26). Hepatic expression of miR-34a is also increased by the CREB/CRTC2 complex in obesity, leading to inhibition of FGF21, a key regulator of energy metabolism (42). Thus, the multiple miRNAs regulated by FXR-SHP provide potential novel targets for treating obesity-related pathologies. Intriguingly, miR-34a is present in adipocyte-secreted exosomes and functions as a novel mediator of obesity-induced adipose inflammation by acting on the adjacent macrophages (43). Because miR-802 is also detected in blood samples of patients with diabetes (44), miR-802 could have both diagnostic and therapeutic potential.

MiRNAs have relatively small effects on individual genes but may have profound overall impacts by inhibiting multiple targets involved in the same biological and/or disease pathways (18,25). For example, obesity-induced miR-34a inhibits expression of numerous targets in energy metabolism, including SIRT1, NAMPT, and the nuclear receptors PPARα and HNF-4α, and also the coreceptor for FGF15/19 and FGF21, βKL (19–21,24,45). Although miR-802 impairs insulin sensitivity through targeting Hnf-1β (23), all the detrimental effects of miR-802 leading to impaired glucose metabolism and promotion of fatty liver are not likely due to silencing of HNF-1β. On the basis of in silico analysis, numerous genes involved in glucose and lipid regulation, including AMPK, are predicted to have miR-802 binding sites in their 3′ UTRs. It will be interesting to see whether silencing of these target genes by miR-802 contributes to promotion of fatty liver.

In conclusion, we demonstrate that a pathway involving FXR and SHP regulates hepatic glucose and lipid metabolism, in part via inhibition of miR-802 expression. Normal inhibition of miR-802 by FXR-induced SHP is dysregulated in obesity, resulting in elevated miR-802 levels and, consequently, insulin resistance and fatty liver. The FXR-SHP-miR-802 cascade identified in this study may present potential novel targets for treating obesity-associated diabetes and NAFLD.

Acknowledgments. The authors thank the Liver Tissue Cell Distribution System, University of Minnesota (National Institutes of Health contract no. HHSN276201200017C), for providing liver specimens of patients with NAFLD and of normal individuals. The authors also thank Johan Auwerx and Kristina Schoonjans at École Polytechnique Fédérale de Lausanne for providing SHP-flox mice and FXR-flox mice.

Funding. This study was supported by a Basic Science Award from the American Diabetes Association (1-16-IBS-156) and National Institutes of Health grants (DKR01062777 and DKR01095842) to J.K.K.

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

Author Contributions. S.S., H.S., and J.K.K. designed the research; S.S., H.S., and Y.-C.K. performed experiments; Y.-C.K. performed bioinformatic analysis; all the authors contributed to data analysis; and S.S., B.K., and J.K.K. wrote the article. J.K.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis..