Polarity Protein AF6 Controls Hepatic Glucose Homeostasis and Insulin Sensitivity by Modulating IRS1/AKT Insulin Pathway in an SHP2-Dependent Manner

Obesity-associated insulin resistance is a hallmark of type 2 diabetes and plays a central role in the metabolic syndrome (1–3). In the state of insulin resistance, the effect of insulin is repressed; the resulting excess of hepatic glucose production contributes to hyperglycemia (4). Liver is the dominant organ in the maintenance of glucose homeostasis, a function that occurs primarily through multiple insulin-mediated events (5). Thus, identifying genes and pathways that govern hepatic glucose metabolism is critical for the development of effective therapies for type 2 diabetes.

Recently, studies have implied a novel role of polarity proteins in the regulation of cell metabolism. For example, tissue-specific deletions of liver kinase B1 (LKB1) in liver (6), skeletal muscle (7,8) and pancreas (9,10) have revealed that LKB1 functions to control glucose homeostasis and energy metabolism in various tissues. Moreover, the polarity kinases PAR-1a and PAR-1b have also been reported to regulate glucose tolerance and adiposity in vivo (11,12). Linkage between cell polarity and regulation of metabolism is likely important in normal cell development and metabolic disease, which needs further investigation. AF6 (also known as afadin and MLLT4) is a multidomain F-actin–binding protein that is expressed in almost all epithelial tissues (13). AF6 plays an essential role in cell-cell junction organization and development (13–15). As a polarity protein, AF6 usually interacts with various proteins (receptor tyrosine kinases, cell adhesion molecules, regulatory and signaling molecules) to exert its effect (16–19). Notably, AF6 has been reported to attenuate phosphorylation of AKT, the major component of insulin signaling in epithelial homeostasis and oncogenesis (20–22). Because of its potent connection to PI3K/AKT signaling, we began to consider the possibility that the polarity protein AF6 might function as a regulator of glucose metabolism and insulin sensitivity.

In the current study, we show that AF6 is elevated in the livers of high-fat diet (HFD)–induced diabetic mice and db/db mice. We performed in vivo and in vitro experiments to test our hypothesis regarding the function of AF6 in glucose homeostasis and insulin sensitivity. Using gain-of-function and loss-of-function approaches, we demonstrate that AF6 exacerbates insulin resistance through its interaction with Src homology 2 domain-containing phosphatase 2 (SHP2) and its regulation of SHP’s tyrosine phosphatase activity. Furthermore, we assessed the effect of AF6 knockdown on glucose metabolism in two diabetic mouse models. Our findings reveal a novel role for the polarity protein AF6 in the regulation of insulin sensitivity, suggesting AF6 as a potential target for the treatment of diabetes.

All mice were males aged 8–12 weeks, maintained on a C57BL/6J background under a 12-h light/dark cycle at 25°C, and provided free access to water and food. Wild-type (WT) and leptin receptor–mutated (db/db) mice were obtained from Shanghai Laboratory Animal Co. (Shanghai, China). AF6^flox/+ mice were generated by standard homologous recombination at Shanghai Biomodel Organism (Shanghai, China) and backcrossed to WT C57BL/6J mice for at least six generations. AF6^flox/flox (AF6 fl/fl) mice were crossed with albumin-Cre mice to generate liver-specific AF6 knockout (AF6 LKO) offspring. Primers used for genotyping AF6 knockout (KO) are as follows: AF6-KO-p1, 5′-TGACAGGATGGCAAACTCT-3′; AF6-KO-p2, 5′-GAGGGACCGTGTAGGAGAC-3′. Mice were fed an HFD (D12492; Research Diets) or control chow (Shanghai Laboratory Research Center, Shanghai, China) for the same period. Food intake and body weight were measured weekly. All animals were anesthetized and subjected to terminal bleeding before euthanasia. The resulting serum was stored at −80°C; tissues were collected and snap frozen in liquid nitrogen pending further analysis. All experimental procedures were approved by the institutional animal care and use committee of the Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences.

The 293T cells were obtained from the American Type Culture Collection (Manassas, VA) and grown in DMEM (Thermo Fisher Scientific) containing 10% FBS (HyClone) and 1% penicillin/streptomycin (Thermo Fisher Scientific). Mouse primary hepatocytes were prepared by collagenase perfusion as described previously (23). The isolated primary hepatocytes were cultured in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin precoated overnight with 20 μg/mL collagen type I (EMD Millipore). Primary hepatocytes fasted overnight were treated with various concentrations of insulin, glucose, oleic acid (OA), or palmitic acid (PA) to examine the expression of AF6.

Double-stranded small interfering RNA (siRNA) targeting mouse SHP2 was purchased from Ribobio (Guangzhou, China). The siRNA sequences specific for mouse SHP2 were as follows: si-SHP2-1, 5′-CCACUUUGGCUGAACUGGUUCAGUA-3′; si-SHP2-2, 5′-GTTAGGAACGTCAAAGAAA-3′.

The AF6 cDNA was amplified from an AF6 overexpression vector described previously (24) and cloned into pAdeno-MCMV adenoviral backbone vector (OBiO Technology, Shanghai, China). The SHP2 cDNA was amplified from mouse hepatic cDNA and cloned into pAdeno-MCMV adenoviral backbone vector. Plasmid constructions and PCR were performed using standard molecular biology techniques. Recombinant adenoviruses expressing AF6 (Ad-AF6), SHP2 (Ad-SHP2), and short hairpin RNA (shRNA) for mouse AF6 (Ad-shAF6) were then constructed using the AdMax system (OBiO Technology) according to the manufacturer’s instructions. Purified high-titer stocks of amplified recombinant adenoviruses were diluted in PBS and administered at a dose of 1 × 10⁷ plaque-forming units (PFU)/well in 12-well plates or injected at a dose of 1 × 10⁹ PFU/mouse through a single injection into the tail vein. Injection of the adenoviruses above or control viruses did not affect food consumption compared with that of control-treated animals.

Levels of blood glucose and serum insulin were measured with a Glucometer Elite monitor or Rat Insulin ELISA Kit (EMD Millipore), respectively. The HOMA for insulin resistance (HOMA-IR) index was calculated according to the following equation: (fasting glucose levels [mmol/L] × fasting serum insulin [μU/mL]) / 22.5.

For the glucose tolerance test (GTT), mice fasted overnight for 16 h were injected intraperitoneally (i.p.) with glucose 2 g/kg body weight. For the insulin tolerance test (ITT), mice fasted for 4 h were injected with insulin 0.75 or 1 unit/kg i.p.

Cell glycogen content was determined as previously described (25). Liver glycogen content was determined as previously described (26).

Mouse primary hepatocytes infected with adenovirus for 48 h or liver tissues were lysed in lysis buffer (25 mmol/L HEPES at PH 7.4, 150 mmol/L NaCl, 2 mmol/L EDTA, 0.5% Triton X-100) and then centrifuged at 12,000g for 10 min at 4°C. The supernatant was incubated with the anti-SHP2 antibody for 2 h at 4°C. Protein G-Sepharose beads (GE Healthcare) were added and further incubated for 2 h at 4°C. Immunocomplexes were extensively washed twice with lysis buffer and twice with protein tyrosine phosphatase buffer (20 mmol/L HEPES at PH 7.5, 1 mmol/L EDTA, 5% glycerol, 1 mmol/L dithiothreitol). Beads then were resuspended in 50 μL of protein tyrosine phosphatase buffer supplemented with 250 μmol/L phosphopeptide (TSTEPQpYQPGENL; EMD Millipore) and incubated for 30 min at 37°C. To assess the amount of free phosphate from reactions, aliquots of supernatants were added to 96-well plates with 100 μL of malachite green solution (EMD Millipore) and incubated for 15 min at room temperature. Absorbances were read at 620 nm with a microplate reader. The absorbance was compared with a phosphate standard curve to determine the release of phosphate in picomoles. To control the expression levels of SHP2 in the immune complex in the assays, aliquots of the immune complexes were Western blotted using the anti-SHP2 antibody.

Mice were fasted overnight for 16 h and injected with 5 units/kg i.p. human insulin (Novo Nordisk) or PBS; 10 min after injection, mice were killed, liver samples were snap frozen in liquid nitrogen, and proteins were extracted for Western blot analysis. Mouse primary hepatocytes infected with adenovirus for 32 h were serum starved for 16 h and treated with or without 100 nmol/L insulin for 5–15 min (as indicated in the figures), and cells were then harvested for Western blot analysis.

Western blot analysis and immunoprecipitation assay were performed as previously described (24). For Western blotting, target proteins were detected using an enhanced chemiluminescence kit (Pierce), quantified by National Institutes of Health Image J software, and normalized to HSP90 or the corresponding total proteins. The primary antibodies were obtained as follows: anti-AF6 antibody (BD Transduction Laboratories); anti-p-insulin receptor (IR) (Tyr1150/1151), anti-IR-β, anti-insulin receptor substrate 1 (IRS1), anti-p-AKT (Ser473), anti-p-AKT (Thr308), anti-AKT, anti-p-glycogen synthase kinase-3β (GSK3β) (Ser9), anti-GSK3β, anti-p-STAT3 (Tyr705), anti-STAT3, anti-p-extracellular signal–regulated kinase (ERK) 1/2 (Thr202/Tyr204), anti-ERK1/2, anti-SHP2, anti-glycogen synthase (GS), anti-HSP90, and anti-HA (Cell Signaling Technology); anti-p-IRS1 (Tyr612/608 [human/mouse]) and anti-phosphotyrosine (clone 4G10) (EMD Millipore); anti-glycogen phosphorylase (PYGL) and anti-glucose-6-phosphatase (G6Pase) (Abcam); anti-PEPCK and anti-IRS2 (Santa Cruz Biotechnology); and anti-FLAG (Sigma).

Total RNA was extracted from tissues or cells using TRIzol reagent (Invitrogen), according to the manufacturer’s instructions. RNA was reverse transcribed with PrimeScript RT-PCR Kit (Takara Bio) to generate cDNA for real-time PCR using SYBR Premix Ex Taq (Takara Bio) in an ABI Prism 7900HT Sequence Detection System (Thermo Fisher Scientific). Transcriptional levels for mRNA were normalized to m36B4. The relative expression of mRNA was calculated using the 2^−ΔΔCT method. Primers used for real-time PCR are shown in Supplementary Table 1.

Liver tissues were excised and fixed with 4% paraformaldehyde for 24–36 h. For periodic acid Schiff (PAS) staining, the paraffin-embedded specimens were sectioned at 5 μm and then stained. Immunofluorescence staining was performed with anti-AF6 antibody (BD Transduction Laboratories) as described previously (24). Fluorescence was monitored by confocal laser microscopy.

Statistical analyses were performed using GraphPad Prism 7 software. Unpaired two-tailed Student t test was used for comparisons between two groups. One-way or two-way ANOVA followed by Bonferroni post hoc test was used for multiple comparisons. P < 0.05 was considered statistically significant. Results are expressed as mean ± SEM. All results are representative of at least three independent experiments.

We examined the expression of AF6 in liver tissues in two different types of diabetic mouse models: HFD-fed mice and db/db mice. Mice fed HFD for 8 weeks were obese and hyperglycemic (data not shown). Both mRNA and protein levels of AF6 were significantly increased in the livers of HFD-fed mice compared with control chow-fed mice (Fig. 1A and B). Similarly, hepatic AF6 levels were higher in db/db mice than in WT mice (Fig. 1C and D). These data suggest that hepatic expression of AF6 is significantly upregulated in dietary and genetic mouse models of diabetes.

To study the potential role of AF6 in liver metabolism, we generated a liver-conditional AF6-KO mouse with loxP sites flanking exon 2 of AF6 (Fig. 2A). AF6 LKO mice (albumin-Cre/AF6^flox/flox) were obtained by crossing AF6^flox/flox mice with mice that expressed Cre recombinase driven by the albumin promoter. Littermate AF6^flox/flox (AF6 fl/fl) mice were used as controls. AF6 LKO and control mice were born in accordance with Mendelian ratio. Real-time PCR, immunofluorescence, and Western blot results confirmed efficient AF6 deletion in the livers of AF6 LKO mice (Fig. 2B and Supplementary Fig. 1). Although body weight and food intake were comparable between AF6 LKO and control mice maintained on the standard chow diet (Supplementary Fig. 2A and B), levels of blood glucose were significantly lower in AF6 LKO mice under both fed and fasting conditions (Fig. 2C), with no difference in serum insulin levels (Fig. 2D). The HOMA-IR index was also decreased in AF6 LKO mice (Fig. 2E). Consistent with these changes, blood glucose levels decreased much more quickly after challenge with glucose or insulin in AF6 LKO mice than in control mice as measured by GTT and ITT, indicating improved systemic glucose tolerance and insulin sensitivity by deletion of AF6 (Fig. 2F and G).

To clarify the mechanism underlying the effect of hepatic AF6 knockout on glucose homeostasis, we examined glycogen storage in the livers of AF6 LKO and control mice. PAS staining suggested that the amount of glycogen was increased in the livers of AF6 LKO mice (Fig. 2H). Consistent with the histology, glycogen content analysis showed that the glycogen level was approximately doubled in the AF6 LKO mice (Fig. 2I). These data indicate that knockout of AF6 controls glycogen metabolism in the liver. We further examined hepatic genes involved in glucose metabolism. Relevant to the glycogen alternations, AF6 deletion in the liver was found to cause a significant decrease in the mRNA level of PYGL, the key hepatic glycogenolysis gene (27), having an opposite effect on that of GS, the key glycogen synthesis gene, and glucokinase (GK), which catalyzes the first step of glucose metabolism (28), but scoring almost no effect on genes related to gluconeogenesis (PEPCK, G6Pase, and peroxisome proliferator–activated receptor γ coactivator-1α [PGC-1α]) (29) (Supplementary Fig. 3A). Alternations of PYGL and GS protein levels in AF6 LKO livers further confirmed the negative regulation of glycogen synthesis programmed by AF6 (Supplementary Fig. 3C). Given that hepatic gluconeogenic gene expression is facilitated in response to fasting, we assessed the protein levels of PEPCK and G6Pase in AF6 LKO mice during the fasting-refeeding transition but only detected a significant decrease in G6Pase expression after refeeding (Supplementary Fig. 3E). Additionally, we analyzed the phosphorylation of crucial participants in the insulin signaling cascade. Insulin-stimulated phosphorylation of IRS1 (Tyr608), AKT (Ser473), and GSK3β (Ser9), but not that of IR, was evidently increased in the livers of AF6 LKO mice (Fig. 2J). In line with increased IRS1 tyrosyl phosphorylation, there was more p85 subunit of PI3K in IRS1 immunoprecipitates from AF6 LKO livers compared with controls (Supplementary Fig. 5A). In addition, hepatic AF6 deficiency also resulted in increased tyrosyl phosphorylation of IRS2 (Supplementary Fig. 6A). These results demonstrate that hepatic knockout of AF6 could increase clearance of blood glucose most likely by promoting conversion of blood glucose into liver glycogen.

To further confirm the role of AF6 in insulin sensitivity, we constructed an adenovirus vector encoding the AF6 protein (Ad-AF6) and injected this construct into male C57BL/6J mice. Ad-AF6 led to increased hepatic AF6 expression compared with negative control adenovirus (Ad-NC) mice (Fig. 3A and I). Body weight and food intake did not differ significantly between the Ad-AF6 and Ad-NC animals (Supplementary Fig. 2C and D), nor did fasting blood glucose and serum insulin levels (Fig. 3B and C). However, mice infected with Ad-AF6 exhibited significantly elevated fed blood glucose and serum insulin levels as well as HOMA-IR index (Fig. 3B–D). Consistently, hepatic overexpression of AF6 resulted in impaired glucose tolerance and insulin sensitivity (Fig. 3E and F). Moreover, glycogen content was decreased by Ad-AF6 infection as evidenced by PAS staining and glycogen content analysis (Fig. 3G and H). In agreement with these findings, PYGL expression was enhanced and GS expression decreased in the livers of Ad-AF6 mice (Supplementary Fig. 3B and D), whereas genes related to gluconeogenesis were not affected (Supplementary Fig. 3B and F). Examination of insulin signaling showed that the insulin-induced phosphorylation of IRS1, IRS2, AKT, and GSK3β was notably decreased in the livers of Ad-AF6–infected mice (Fig. 3I and Supplementary Fig. 6B). We further examined whether re-expression of AF6 in the livers of AF6 LKO mice rescues the metabolic phenotype. As expected, restoration of AF6 expression back to a normal level totally reversed the improved effect of AF6 deletion on blood glucose, glucose tolerance, and insulin sensitivity (Supplementary Fig. 4). These results indicate that hepatic AF6 can directly induce insulin resistance in vivo.

To confirm the role of AF6 in insulin signaling, we isolated mouse primary hepatocytes from C57BL/6J mice and infected these cells in vitro with Ad-AF6 or Ad-NC. AF6 overexpression significantly impaired insulin-stimulated phosphorylation of IRS1 (Tyr608), AKT (Ser473), and GSK3β (Ser9), which suggests that Ad-AF6 induces insulin resistance (Fig. 4A). The opposite effects were observed when AF6 was knocked down by infection with Ad-shRNA specific for AF6 (Ad-shAF6) compared with control cells infected with scrambled adenovirus (Ad-scramble) (Fig. 4C). Insulin-stimulated phosphorylation of IR, however, was not affected by infection with either Ad-AF6 or Ad-shAF6 (Fig. 4A and C). In addition, glycogen content was significantly changed upon infection of mouse primary hepatocytes with these adenoviruses, specifically yielding decreased glycogen storage with Ad-AF6 in the presence of insulin and increased glycogen storage with Ad-shAF6 in the presence or absence of insulin (Fig. 4B and D).

The potent function of AF6 in insulin sensitivity prompted us to explore the mechanisms underlying its effects. AF6 has been reported to regulate platelet-derived growth factor receptor activation through the protein tyrosine phosphatase SHP2 (30), and SHP2 has been shown to be a negative regulator of hepatic insulin sensitivity (31). We postulated that AF6 might regulate insulin signaling through SHP2. To test this hypothesis, we first used coimmunoprecipitation assays to assess the interaction between AF6 and SHP2. When both AF6 and SHP2 were overexpressed in 293T cells, immunoprecipitation of SHP2 could pull down AF6 (Supplementary Fig. 7A). Conversely, immunoprecipitation of AF6 could also pull down SHP2 (Supplementary Fig. 7B). Furthermore, we found that endogenous AF6 did form a complex with SHP2 in mouse primary hepatocytes (Fig. 5A and B). Interestingly, the amount of SHP2 coimmunoprecipitated with AF6 was increased in the presence of insulin in mouse primary hepatocytes (Fig. 5C). We speculated that insulin might stimulate AF6-mediated recruitment of SHP2 and contribute to the feedback inhibition loop. We further examined the effect of AF6 on the tyrosine phosphatase activity of SHP2. The endogenous SHP2 was immunoprecipitated from mouse primary hepatocytes infected with Ad-AF6 or Ad-NC. Next, the tyrosine phosphatase activity of SHP2 was assessed using a small phosphopeptide, TSTEPQpYQPGENL, as a model substrate. AF6 overexpression significantly increased the phosphatase activity of SHP2 (Fig. 5D). Consistently, the phosphatase activity was impaired when endogenous AF6 levels were reduced by Ad-shAF6 (Fig. 5E). In line with these in vitro results, phosphatase activity of hepatic SHP2 was enhanced in Ad-AF6–infected mice and attenuated in AF6-KO mice (Fig. 5F and G). Taken together, these data indicate that AF6 interacts with SHP2 and is sufficient for the activation of SHP2 tyrosine phosphatase activity.

We then investigated whether SHP2 is involved in AF6-regulated insulin sensitivity. Mouse primary hepatocytes from C57BL/6J mice were isolated and transfected with two different siRNAs to downregulate SHP2 expression. Ad-AF6 significantly impaired insulin-stimulated phosphorylation of IRS1, AKT, and GSK3β, and this effect was rescued by knockdown of SHP2 (Fig. 6A). Furthermore, we isolated mouse primary hepatocytes from AF6 fl/fl mice and infected these cells with the combination of Ad-cre (or the control Ad-NC) and Ad-SHP2 (or the control Ad-null). As predicted, overexpression of SHP2 significantly blocked the increase in insulin-stimulated phosphorylation of IRS1, AKT, and GSK3β seen upon deletion of AF6 (Fig. 6B). However, SHP2 overexpression did not have an apparent effect on insulin-stimulated phosphorylation of IR in cells subjected to any of these treatments. To confirm this epistasis in vivo, we injected AF6 LKO or AF6 fl/fl mice with Ad-SHP2 and examined the effects on AF6 deficiency–improved insulin sensitivity. As we supposed, infection with Ad-SHP2 significantly elevated blood glucose and serum insulin levels in AF6 LKO mice (Fig. 6C and D). Overexpression of SHP2 also counteracted the changes of HOMA-IR index observed in AF6 LKO mice (Fig. 6E) as well as the improved glucose tolerance and insulin sensitivity seen with AF6 deletion (Fig. 6F and G). We next analyzed insulin signaling in the livers of these mice. Notably, SHP2 overexpression abrogated the effect of AF6 deletion on insulin-stimulated phosphorylation of IRS1, AKT, and GSK3β as well as IRS1-p85 interaction (Fig. 6H and Supplementary Fig. 5B). Considered together, these experiments clearly demonstrate that overexpression of SHP2 is able to restore insulin signaling in hepatocytes with AF6 deficiency and block the enhanced insulin sensitivity in AF6 LKO mice.

Given the ability of AF6 to affect endogenous SHP2 activity, we were interested in investigating whether overexpression of AF6 could enhance the inhibitory effect of Ad-SHP2 on glucose tolerance and insulin sensitivity. Five times less Ad-SHP2 (2 × 10⁸ PFU/mouse) was used for in vivo experiments. SHP2 expression levels were moderately elevated in the livers of mice after Ad-SHP2 infection. A moderate amount of Ad-SHP2 had no obvious effect on levels of blood glucose, glucose tolerance, and insulin sensitivity as well as on the activation of IRS1/AKT signaling; however, additional injection of Ad-AF6 significantly impaired these phenotypes (Supplementary Fig. 8). On the basis of these findings, we propose that the action of AF6 on glucose metabolism and hepatic insulin signaling is mediated through its regulation of SHP2’s tyrosine phosphatase activity.

The observations that AF6 depletion promotes liver glycogen storage and improves insulin sensitivity led us to hypothesize that knockdown of hepatic AF6 under diabetic conditions may be helpful for reducing blood glucose level. To address this hypothesis, we first fed AF6 LKO and control mice an HFD for an interval of 2 months, starting from the age of 8 weeks. Real-time PCR and Western blot results demonstrated almost completely abolished expression of AF6 in HFD-fed AF6 LKO mice compared with that from control mice (Supplementary Figs. 9A and 11A). These HFD-fed AF6 LKO mice exhibited significantly decreased levels of fed and fasting blood glucose and fed serum insulin, as well as a deceased HOMA-IR index, compared with HFD-fed control mice (Fig. 7A–C). Importantly, the glucose tolerance and clearance of HFD-fed mice were markedly improved by deletion of AF6 (Fig. 7D and E). In addition, AF6 LKO resulted in enhanced hepatic insulin sensitivity as evidenced by the significantly increased levels of phosphorylated IRS1, AKT, and GSK3β in the LKO mice (Supplementary Fig. 11A). Taken together, these results demonstrate that AF6 deficiency could alleviate hyperglycemia and insulin resistance in HFD-induced diabetic mice.

To further validate the functional role of AF6 in insulin resistant conditions, we injected db/db mice with Ad-shAF6 or Ad-scramble to examine whether decreased AF6 expression could reverse insulin resistance in these mice. As shown in Supplementary Figs. 9B and 11B, Ad-shAF6 infection caused efficient knockdown of endogenous AF6 in the liver. As expected, infection of db/db mice with Ad-shAF6 yielded significant decreases in the levels of fed and fasting blood glucose and serum insulin as well as in the HOMA-IR index (Fig. 7F–H). Consistent with the effect of AF6 LKO in HFD-induced diabetic mice, db/db mice injected with Ad-shAF6 exhibited markedly enhanced glucose tolerance, insulin sensitivity, and hepatic insulin signaling (Fig. 7I and J and Supplementary Fig. 11B). Knockdown of AF6 in both HFD-induced diabetic mice and db/db mice had no significant effect on body weight and food intake (Supplementary Fig. 10). Collectively, these results suggest that depletion of AF6 in liver ameliorates hyperglycemia and insulin resistance in diabetic mice.

Because hepatic AF6 was elevated in diabetic mice (Fig. 1), which have hyperinsulinemia, hyperglycemia, and high levels of plasma free fatty acids (FFAs), we examined whether various concentrations of insulin, glucose, or FFAs could regulate hepatic AF6 expression in primary hepatocytes. Our results suggest that the treatment of insulin or glucose has no effect on AF6 expression, but the treatment of OA and PA could upregulate both the mRNA and the protein levels of AF6 in a dose-dependent manner in primary hepatocytes (Fig. 8A–D and Supplementary Fig. 12). These results indicate that hepatic AF6 might be upregulated by FFAs in diabetes and that downregulation of hepatic AF6 may be helpful for lowering blood glucose levels.

The polarity protein AF6 is expressed in almost all tissues in both embryos and adults and in a variety of cell types, including epithelial cells, neurons, and fibroblasts (13,15,32). Recent studies have implicated AF6 as playing a role in the pathogenesis of several diseases, including Parkinson disease (33,34), cardiac hypertrophy (35), and various types of cancer (24,36,37). However, the regulatory effect of AF6 on diabetes as well as its specific functions in the liver remain largely unknown. As reported here, AF6 expression is markedly increased in liver tissues of two diabetic animal models. Liver-specific AF6-KO mice exhibited improved glucose tolerance and insulin sensitivity, whereas Ad-AF6 overexpression in mice had the opposite effect, indicating that AF6 is a negative regulator of insulin signaling in the liver. In addition, our in vitro studies showed that AF6 overexpression and knockdown impairs and improves insulin signaling, respectively, in primary hepatocytes. Mechanistically, AF6 directly interacts with SHP2 and regulates SHP2’s tyrosine phosphatase activity, thereby eliciting the inactivation of downstream IRS1/AKT signaling. Finally, AF6 knockdown significantly ameliorates hyperglycemia and insulin resistance in the two diabetic mouse models. In summary, our findings highlight a critical role of hepatic AF6 in the regulation of glucose homeostasis, suggesting new opportunities for treating diabetes.

The liver plays a critical role in glucose metabolism and can store or produce glucose, depending on physiological conditions. Glycogen, which is considered the principal storage form of glucose, is found primarily in liver and skeletal muscle (29). Glycogen homeostasis involves the concerted regulation of the rate of glycogen synthesis and the rate of glycogenolysis. Defects in these two processes can be major contributors to hyperglycemia and insulin resistance, and the liver glycogen content is decreased in individuals with type 2 diabetes (38). In the current study, liver glycogen storage was significantly increased in AF6 LKO mice. We examined mRNA and protein levels of key target genes implicated in glucose metabolism and found that expression of PYGL was lower and expression of GS higher in AF6 LKO livers compared with controls. We further assessed the protein levels of gluconeogenic genes PEPCK and G6Pase in AF6 LKO mice during the fasting-refeeding transition. Our results suggest that protein expression of PEPCK was slightly suppressed by deletion of AF6 after refeeding (P = 0.051), and the expression of G6Pase was significantly reduced in AF6 LKO mice after refeeding. We also demonstrated in vitro effects of AF6 deficiency to increase cell glycogen content in isolated mouse primary hepatocytes. In addition to deletion of AF6, we carried out Ad-AF6 administration studies in vitro and in vivo. AF6 overexpression led to lower glycogen storage as expected. These observations revealed that AF6 is a critical regulator of glycogen metabolism and suggest that AF6 induces hyperglycemia most likely by inhibiting liver glycogen storage.

The insulin signaling pathway comprises three essential mediators: the IR and related IRS, the PI3K heterodimer, and AKT (39). Our studies showed that AF6 affects phosphorylation of IRS1 and AKT but not that of IR; therefore, we predicted that the target of AF6 likely lies downstream of IR in the insulin signaling cascade. Indeed, prior work has shown that AF6 regulates platelet-derived growth factor receptor activation through the protein tyrosine phosphatase SHP2 (30). SHP2 has been reported to regulate insulin signaling by binding to and dephosphorylating the PI3K binding sites of IRS1 (31,40,41). On the basis of these studies, we hypothesized that AF6 might regulate insulin signaling by SHP2. As we show in the present work, AF6 interacts with SHP2 and increases SHP2’s phosphatase activity. It was reported that SHP2 deletion in hepatocytes could repress Ras/ERK signaling and drive cytokine-induced STAT3 signaling (42,43). We further designed experiments to observe the possibility that AF6 contributes to insulin sensitivity by activation of ERK1/2 or STAT3. We examined phosphorylation of ERK1/2 and STAT3 in liver lysates of AF6 LKO mice or Ad-AF6–infected mice, and our results demonstrated that AF6 has no effect on ERK2 or STAT3 activation in the liver (Supplementary Fig. 13). Taken together, our observations suggest that AF6 controls hepatic glucose homeostasis and insulin sensitivity by modulating the insulin pathway in an SHP2-dependent manner. Interestingly, the interaction between AF6 and SHP2 was increased in insulin-stimulated primary hepatocytes. This raises the possibility that insulin induces the recruitment of an AF6-SHP2 complex and reduces IRS1/AKT signaling, forming a negative feedback loop.

We believe that increased IRS1/AKT signaling is likely the primary cause of the enhanced glucose tolerance and insulin sensitivity in AF6 LKO mice, although we cannot entirely rule out other possibilities that may contribute to the phenotype of these mice. Kubota et al. (44) and Dong et al. (45) reported that IRS1 and IRS2 play overlapping roles in insulin action using conditional hepatic KOs of IRS1 and IRS2. Our data indicate that insulin-stimulated tyrosyl phosphorylation of IRS2 is also increased by hepatic AF6 depletion, whereas tyrosyl phosphorylation of IRS1 may be increased to a greater extent.

The regulation of cell polarity and metabolism are two biologically essential processes that have to be strictly controlled. Recent reports have implicated the emerging role of cell polarity proteins in the regulation of metabolism (6–12). With the use of RNA sequencing, significant progress has been made in understanding the transcript expression of a set of polarity genes in meta-inflammation (C.D., X.W., unpublished observations). On the basis of these findings, we found that high expression of the polarity protein AF6 can directly induce hyperglycemia and insulin resistance in vitro and in vivo and that knockdown of AF6 helps to alleviate glucose metabolism disorders in diabetic mice. To investigate whether the effects of AF6 on glucose metabolism link its role in polarity, we performed immunofluorescence analysis for E-cadherin and ZO-1, two key proteins responsible for maintaining apicobasal cell adhesion and polarity, in the livers of AF6 LKO and AF6 fl/fl mice under chow- and HFD-fed states (Supplementary Fig. 14). We found that depletion of AF6 itself could not significantly change the expression or localization of these two polarity markers. The well-organized hepatic structure was significantly disrupted in HFD-fed mouse liver. Importantly, AF6 deletion did not rescue the disrupted apicobasal polarity observed in the livers of HFD-fed mice. Given that AF6 deletion significantly alleviates hyperglycemia and insulin resistance in HFD-induced diabetic mice (Fig. 7A–E), we conclude that the action of AF6 on glucose metabolism and hepatic insulin signaling is mediated through an SHP2-dependent IRS1/AKT pathway, not through the role of AF6 in cell polarity.

In conclusion, the current study reports a novel role for AF6 in insulin resistance and diabetes. Using in vivo and in vitro experiments, we demonstrated that abnormal AF6 expression can directly impair insulin signaling and induce insulin resistance by binding to SHP2 and upregulating SHP2’s tyrosine phosphatase activity (Fig. 8E). These findings will provide new insights into the diverse functions of cell polarity proteins and further suggest that AF6 could be targeted as a therapeutic avenue for the treatment of diabetes, potentially facilitating improved insulin sensitivity and glucose tolerance.

Acknowledgments. The authors thank Dr. Jiali Deng, Dr. Yajie Guo, and Dr. Yi Luan (CAS Key Laboratory of Nutrition, Metabolism and Food Safety) for technical assistance; Dr. Yingying Le (CAS Key Laboratory of Nutrition, Metabolism and Food Safety) for helpful discussion; and Dr. Yong Liu (Hubei Key Laboratory of Cell Homeostasis, College of Life Sciences, Institute for Advanced Studies, Wuhan University, Wuhan, China) for providing the HFD-fed mouse liver samples.

Funding. This work was supported by grants from the National Key R&D Program of China (2017YFC1600100), the Strategic Priority Research Program of the Chinese Academy of Sciences (XDA12010100), and the National Natural Science Foundation of China (81672965, 81872369).

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

Author Contributions. C.D., X.W., Y.W., and S.Z. researched data. C.D. and L.Z. designed the research and wrote, reviewed, and edited the manuscript. Y.X. and S.Z. contributed to the writing of the manuscript. M.Q. and W.J. contributed to the discussion. L.Z. 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.