Arrestin domain-containing 3 (Arrdc3) modulates insulin action and glucose metabolism in liver

Significance

Insulin resistance in liver is a key component in the pathogenesis of type 2 diabetes, obesity-associated nonalcoholic fatty liver disease, and metabolic syndrome. Here we demonstrate that hepatic expression of Arrdc3, a molecular adaptor previously linked to human obesity, is potently induced by physiological insulin and obesity-related hyperinsulinemia. Liver-specific deletion of Arrdc3 increases hepatic insulin sensitivity. This is mechanistically linked to a direct interaction between the insulin receptor (IR) and ARRDC3 proteins controlling IR levels in the plasma membrane and downstream insulin signaling involving the transcription factor FOXO1 and expression of the rate-limiting enzymes in the glycolysis and gluconeogenesis pathways. Thus, Arrdc3 may provide a therapeutic target for diseases associated with hepatic insulin resistance.

Abstract

Insulin action in the liver is critical for glucose homeostasis through regulation of glycogen synthesis and glucose output. Arrestin domain-containing 3 (Arrdc3) is a member of the α-arrestin family previously linked to human obesity. Here, we show that Arrdc3 is differentially regulated by insulin in vivo in mice undergoing euglycemic-hyperinsulinemic clamps, being highly up-regulated in liver and down-regulated in muscle and fat. Mice with liver-specific knockout (KO) of the insulin receptor (IR) have a 50% reduction in Arrdc3 messenger RNA, while, conversely, mice with liver-specific KO of Arrdc3 (L-Arrdc3 KO) have increased IR protein in plasma membrane. This leads to increased hepatic insulin sensitivity with increased phosphorylation of FOXO1, reduced expression of PEPCK, and increased glucokinase expression resulting in reduced hepatic glucose production and increased hepatic glycogen accumulation. These effects are due to interaction of ARRDC3 with IR resulting in phosphorylation of ARRDC3 on a conserved tyrosine (Y382) in the carboxyl-terminal domain. Thus, Arrdc3 is an insulin target gene, and ARRDC3 protein directly interacts with IR to serve as a feedback regulator of insulin action in control of liver metabolism.

The liver plays a critical role in adapting to changes in nutrient status and in maintaining blood glucose levels during feeding and fasting. After a meal, excess glucose is stored in liver as glycogen which can be broken down to form glucose in the fasted state. In addition, during prolonged fasting when glycogen is depleted, the liver maintains glucose levels by gluconeogenesis from amino acids, glycerol, and lactate, thus preventing hypoglycemia (1).

In hepatocytes, the fate of glucose is largely determined by insulin, which, upon binding to insulin receptor (IR), increases receptor tyrosine kinase activity and initiates a broad signaling network by phosphorylating insulin receptor substrates −1 and −2, activating the PI 3-kinase/Akt (PI3K/AKT) pathway that mediates most of the metabolic effects of insulin, including glycogen synthesis (2). Insulin action also induces phosphorylation and nuclear exclusion of FOXO transcription factors, thus preventing FOXO activation of the transcription of the rate-limiting enzymes of the gluconeogenesis pathway, especially phosphoenolpyruvate pyruvate carboxykinase 1 (PEPCK) and glucose 6-phosphatase (G6Pase) (3, 4). Failure of insulin signaling results in fasting hyperglycemia and depletion of liver glycogen, both typical features of uncontrolled diabetes (5).

Arrestin domain-containing 3 (ARRDC3) is a member of the family of alpha arrestins in the superfamily of arrestin adaptor proteins (6). ARRDC3 has been linked to regulation of adrenergic signaling through interaction and regulation of ubiquitination and trafficking of the β2-adrenergic receptor (7, 8). In humans, ARRDC3 expression in omental adipose tissue positively correlates with body mass index, and whole body ablation of Arrdc3 in mice prevents age-related obesity, improves hepatic steatosis, and increases insulin sensitivity (9). However, mice lacking Arrdc3 only in adipocytes do not fully recapitulate these effects (10), indicating that other tissues play a key role in the phenotype of Arrdc3-null mice. Recently, we identified Arrdc3 as one of the most up-regulated transcripts in the liver of mice during a euglycemic-hyperinsulinemic clamp, suggesting an important role of Arrdc3 in insulin action in the liver (11). In the present study, we investigate the role of Arrdc3 in hepatic regulation of glucose homeostasis and insulin action.

We find that Arrdc3 gene expression in liver is induced by insulin and by high fat diet and genetic obesity, and is reduced in mice with liver-specific IR knockout. Mice lacking Arrdc3 specifically in the hepatocytes (L-Arrdc3 KO) display increased hepatic insulin sensitivity as demonstrated by increased insulin-induced phosphorylation of FOXO1. This translates into lower PEPCK and higher Gck expression, lower endogenous glucose production, and higher glycogen synthesis following insulin stimulation. Mechanistically, increased insulin sensitivity in livers from L-Arrdc3 KO mice is associated with increased IR protein in membrane fractions. This involves a direct interaction of ARRDC3 with IR and requires phosphorylation of a conserved tyrosine 382 residue in the C-terminal tail of ARRDC3. Thus, ARRDC3 is a target of the insulin-signaling pathway in liver that regulates insulin action and endogenous glucose production.

Results

Liver Arrdc3 Expression Is Up-Regulated by Hyperinsulinemia in an IR-Dependent Manner.

Using RNA-seq analysis, we have previously shown that, during a 3-h euglycemic-hyperinsulinemic clamp in mice, insulin regulates the expression of over 1,000 genes in liver (11). Of these genes, one of the most up-regulated transcripts was Arrdc3. qPCR analysis of the clamp samples revealed that, as early as 20 min after initiation of the insulin infusion in the clamp, Arrdc3 mRNA was up-regulated by twofold, and this increased to an eightfold elevation by 180 min (Fig. 1A). This induction was specific to liver. Indeed, in muscle and white adipose tissue (WAT) of mice from the same experiment, Arrdc3 mRNA was suppressed during the hyperinsulinemic clamp by 70 to 80%, respectively (SI Appendix, Fig. S1). Arrdc3 expression was also increased by 2.5- to 3.5-fold in liver of mice with other hyperinsulinemic states, including those associated with the insulin resistance of high-fat-diet (HFD)–induced or genetic obesity (ob/ob mice) (Fig. 1 B and C). On the other hand, total hepatic insulin resistance caused by liver-specific IR knockout (LIRKO) resulted in an over 50% reduction in Arrdc3 mRNA compared to floxed controls (Fig. 1D). Thus, Arrdc3 expression in the liver is up-regulated in states of physiological or pathophysiological hyperinsulinemia and is dependent on insulin receptor signaling.

Fig. 1.

Arrdc3 mRNA is induced in liver by hyperinsulinemia. (A) Arrdc3 mRNA levels in livers of fasted WT mice during a euglycemic-hyperinsulinemic clamp (12 mU/kg/min) for 20 or 180 min (n = 6). (B) Relative Arrdc3 mRNA levels in livers of C57/BL/6J mice fed on a normal chow (20% calories from fat) versus a HFD (60% calories from fat) for 10 wk (n = 5 to 6). (C and D) Arrdc3 mRNA levels in 10-wk-old Ob/Ob mice compared to WT controls (n = 4 to 6) and in 3-mo-old LIRKO mice compared to IR^fl/fl controls (n = 8). Data are means ± SEM; *P < 0.05, **P < 0.01, ***P < 0.001, Student’s t test.

Liver-Specific Arrdc3 Deletion Increases Hepatic Insulin Sensitivity.

To understand the contribution of Arrdc3 in liver to systemic glucose control, we ablated Arrdc3 expression specifically in hepatocytes by crossing transgenic mice expressing CRE recombinase under control of an albumin promoter with mice in which exon 3 of the Arrdc3 gene was flanked with loxP sites (Arrdc3^fl/fl). The resulting liver-specific Arrdc3 knockout mice (L-Arrdc3 KO) had a 90% reduction in Arrdc3 mRNA levels in liver, with no change in Arrdc3 expression in muscle, WAT, and kidneys, confirming the specificity of the knockout (Fig. 2A). L-Arrdc3 KO mice showed similar body weights as controls from birth to 18 wk of age (SI Appendix, Fig. S2A). Body composition assessed by ¹H-magnetic resonance spectroscopy (MRS) was also comparable between genotypes (SI Appendix, Fig. S2B). At 16 wk of age, blood glucose during random feeding, overnight fasting, or fasting followed by 6 h refeeding was not different between L-Arrdc3 KO and Arrdc3^fl/fl mice (Fig. 2B). Likewise, plasma insulin and glucagon levels after a 5-h fast (Fig. 2C) and glucose excursion during an intraperitoneal (i.p.) glucose tolerance test were unchanged (Fig. 2D). Following an i.p. insulin challenge, blood glucose tended to decrease more in L-Arrdc3 KOs compared to Arrdc3^fl/fl mice, suggesting increased insulin sensitivity, but this did not reach statistical significance (Fig. 2E). Glucose production from pyruvate, amino acids, lactate, and glycerol via gluconeogenesis is a major function of the liver for prevention of hypoglycemia during fasting (12). An i.p. pyruvate tolerance test to assess gluconeogenesis revealed a significantly (P < 0.05) lower blood glucose response in the L-Arrdc3 KO mice at 15, 30, and 60 min and a significant decrease in the area under the curve in mice with L-Arrdc3 KO (Fig. 2F).

Fig. 2.

Liver-specific Arrdc3 deletion regulates glucose homeostasis. (A) Arrdc3 mRNA levels in liver, muscle, WAT, and kidney of Arrdc3^fl/fl and L-Arrdc3 KO mice (n = 4 to 6). (B) Blood glucose levels in random-fed conditions after an overnight fast or 6 h refeeding at ∼16 wk of age (n = 5 to 6). (C) Fasting insulin and glucagon levels at ∼16 wk of age (n = 4 to 6). (D) Glucose, (E) insulin, and (F) pyruvate tolerance tests at 15 to 17 wk of age (n = 4 to 6). Data are means ± SEM; *P < 0.05, **P < 0.01, Student’s t test.

To determine if the lower hepatic glucose production in L-Arrdc3 KO mice was due to changes in insulin sensitivity, we performed hyperinsulinemic-euglycemic clamps (Fig. 3A). Glucose infusion rates required to maintain euglycemia during the clamp were higher in L-Arrdc3 KO mice than in controls, indicating increased insulin sensitivity (Fig. 3B). Metabolic labeling using a [1-¹⁴C] 2-deoxyglucose isotope tracer during the clamp revealed no significant differences in uptake rates in skeletal muscle, WAT, and brown adipose tissue (BAT) (Fig. 3C). There were also no significant differences in glucose turnover, glycolytic rates, or whole-body glycogen synthesis (Fig. 3D). On the other hand, during clamp conditions, while mice of both genotypes showed a significant suppression of hepatic glucose production (HGP) (P < 0.0001), the suppressive effects of insulin were more pronounced in L-Arrdc3 KO mice compared to Arrdc3^fl/fl controls (Fig. 3E), indicating that Arrdc3 deletion in hepatocytes leads to increased insulin sensitivity specifically in the liver.

Fig. 3.

Liver-specific Arrdc3 deletion increases insulin sensitivity. Euglycemic-hyperinsulinemic clamps were performed on 20-wk-old L-Arrdc3 KO and Arrdc3^fl/fl controls. (A) Blood glucose levels at baseline and during the last 30 min of the clamp (average from 90 to 120 min). (B) Average glucose infusion rate during the entire 120 min of the clamp. Data are means ± SEM; *P < 0.05, Student’s t test. (C) Insulin-stimulated 2-deoxy-d-[1-¹⁴C]glucose uptake in gastrocnemius muscle, epididymal WAT, and BAT. (D) Whole-body glucose turnover, glycolysis, glycogen synthesis, and (E) insulin-stimulated suppression of hepatic glucose production were assessed during the clamp as described in Methods (n = 6 to 7). Data are means ± SEM; *P < 0.05, **P < 0.01, ****P < 0.0001, two-way ANOVA.

L-Arrdc3 KOs Display Higher Insulin-Induced FOXO1 Phosphorylation and Lower PEPCK Expression.

To explore the molecular pathways leading to increased insulin sensitivity in L-Arrdc3 KO mice, we performed in vivo insulin signaling experiments in livers of L-Arrdc3 KOs and Arrdc3^fl/fl littermates. In both groups, insulin administration induced robust phosphorylation of early signaling steps including IR and its substrate IRS-1, as well as the downstream targets AKT, GSK3, and FOXO1 compared to saline-injected controls (Fig. 4A). Proximal insulin signaling at the level of IR, IRS-1, AKT, and GSK3 was slightly elevated or unchanged in the liver of L-Arrdc3 KOs, and none of these changes reached statistical significance (SI Appendix, Fig. S3A). However, insulin-induced phosphorylation of the transcription factor FOXO1 at Thr24 was enhanced by 65% (P < 0.05) in L-Arrdc3 KOs (Fig. 4B). FOXO1 is a critical regulator of hepatic glucose production, and, following inhibitory phosphorylation in response to insulin, is excluded from the nucleus. This prevents FOXO1 binding and stimulation of promoters of gluconeogenic genes such as Pck1 and G6pc and suppression of glucose utilization genes such as Gck (13, 14).

Fig. 4.

Arrdc3 regulates insulin signaling and PEPCK expression in liver. (A) Western blot analysis of insulin-signaling proteins in livers from overnight fasted L-Arrdc3 KO mice and Arrdc3^fl/fl controls collected 5 min after intravenous insulin (2 IU/mouse) or phosphate-buffered saline (minus insulin) injection (n = 2 to 4). (B) Relative phosphorylation of FOXO1 calculated as the ratio of phosphoprotein/total protein from data in A (n = 2 to 4). Data are means ± SEM; *P < 0.05, two-way ANOVA. (C) Liver glycogen levels were assessed after an overnight fast followed by 6 h of refeeding (n = 4 to 5). (D) Relative Pck1 and (E) Gck mRNA expression normalized to TBP in livers from overnight-fasted or 6 h refed Arrdc3^fl/fl and L-Arrdc3 KO mice (n = 4 to 7). Data are means ± SEM; *P < 0.05, ***P < 0.001, ****P < 0.0001, two-way ANOVA. (F) PEPCK protein expression normalized to vinculin was assessed by Western blotting in livers of overnight-fasted L-Arrdc3 KO and Arrdc3^fl/fl mice (n = 4). Data are means ± SEM; *P < 0.05, Student’s t test.

To analyze the physiological relationship between insulin action and gluconeogenesis in mice lacking Arrdc3 in liver, tissue samples were collected from mice subjected to a 16-h overnight fast followed by a 6-h refeeding period. Consistent with the increase of insulin sensitivity in L-Arrdc3 KO livers, glycogen accumulation induced by refeeding was higher in these mice compared to controls (Fig. 4C). Likewise, in control Arrdc3^fl/fl mice, refeeding promoted a nonsignificant decrease in Pck1 expression, whereas in livers of L-Arrdc3 KOs, refeeding significantly suppressed Pck1 expression by 55% (Fig. 4D). Conversely, mRNA levels of Gck, which catalyzes the first step of glucose utilization and is increased upon feeding or insulin action, showed only a nonsignificant 1.4-fold increase upon refeeding in livers of control mice (Fig. 4E), but a significant 4-fold increase in Arrdc3 KO livers (P < 0.0001), which was also significantly higher than observed in refed Arrdc3^fl/fl mice (P < 0.001). Expression of other important gluconeogenesis genes, such as G6pc and Pcx, and expression of genes involved in glycogen synthesis (Gys2) and breakdown (Pygl) were not changed by either feeding conditions or genotype (SI Appendix, Fig. S3 B–E). Interestingly, despite no changes of mRNA levels for Pck1 in the fasted state, PEPCK protein was reduced by nearly 60% in livers of L-Arrdc3 KO mice (Fig. 4F). Thus, increased insulin sensitivity in livers of L-Arrdc3 KO mice leads to increased FOXO1 phosphorylation, associated with lower glucose output, increased glucose storage, and reciprocal regulation of Pck1 and Gck genes.

Arrdc3 interacts with the Insulin Receptor and Regulates Its Expression in the Plasma Membrane and Autophosphorylation Activity.

ARRDC3 has been reported to interact with and regulate trafficking or activity of transmembrane proteins, such as integrin subunit beta 4 (ITGB4) (15, 16) and β-adrenergic receptors (7–9, 17). To explore whether ARRDC3 modulates insulin sensitivity by regulation of IR at the plasma membrane, we analyzed IR protein expression in membrane fractions isolated from livers of L-Arrdc3 KO mice and floxed controls (Fig. 5A). We found that membranes isolated from livers of L-Arrdc3 KO mice collected in the fed state showed an increase in IR protein levels by 60% (P = 0.016) when compared to control Arrdc3^fl/fl mice (Fig. 5B). This occurred despite no alterations in IR expression in total liver lysates (Fig. 4A).

Fig. 5.

ARRDC3 interacts with the IR and regulates tyrosine kinase activity. (A) IR protein levels in membrane fractions isolated from livers of overnight-fasted and 6 h refed Arrdc3^fl/fl and L-Arrdc3 KO mice (n = 3). (B) Quantification of IR levels normalized to total protein indicated by ultraviolet imaging of a stain-free gel. Data are means ± SEM; *P < 0.05, Student’s t test. (C) Coimmunoprecipitation assay of FLAG-tagged hIR and HA-tagged hARRDC3 transiently overexpressed in HEK 293 cells. (D) Insulin-induced tyrosine phosphorylation (pY) of HA-hARRDC3. (E and F) IR/IGF1R tyrosine phosphorylation in total lysates of HEK 293 cells transiently overexpressing HA-hARRDC3. Cells were serum starved for 3 h and stimulated with 100 nM insulin for 10 min. Data are means ± SEM; **P < 0.01, Student’s t test. All in vitro experiments were done at least twice.

To determine whether the regulation of insulin sensitivity by ARRDC3 may involve interaction with the insulin receptor, we transiently expressed the human IR with a FLAG-tag either alone or in combination with HA-tagged human ARRDC3 in HEK 293 cells and performed coimmunoprecipitation experiments. ARRDC3 was readily detected in immunoprecipitates of the FLAG-tagged insulin receptor, and, conversely, IR was detected in HA-ARRDC3 precipitates (Fig. 5C), thus confirming an interaction. Although this interaction between ARRDC3 and IR was unaffected by ligand-induced receptor activation (Fig. 5C, lane 3), insulin stimulation promoted a robust increase in tyrosine phosphorylation of ARRDC3 in HA immunoprecipitates and total lysates (Fig. 5D). To determine if ARRDC3 overexpression could affect the function of IR, we stimulated cells transiently overexpressing ARRDC3 with insulin and analyzed endogenous IR tyrosine phosphorylation. Under these conditions, ARRDC3 overexpression led to a nearly 45% reduction (P < 0.01) in IR tyrosine phosphorylation (Fig. 5 E and F).

A Conserved Tyrosine (Y³⁸²) Residue in ARRDC3 C-Terminal Tail Regulates Its Binding to the Insulin Receptor.

The C-terminal tail of ARRDC3 is highly conserved among species (SI Appendix, Fig. S4A) and contains two conserved proline-rich PPXY motifs (PPLY and PPSY) that have been shown to be important sites of interaction with HECT domain-containing E3 ubiquitin ligases and to regulate protein ubiquitination (8, 17, 18). These PPXY motifs flank a conserved tyrosine residue (Y³⁸²) that is phosphorylated in tumors with epidermal growth factor receptor-activating mutations (19), raising the possibility that this is an important site for regulation by receptor tyrosine kinases. To determine the contribution of these tyrosine-containing regions as potential sites for IR-mediated tyrosine phosphorylation of ARRDC3 and their interaction, we created ARRDC3 expression constructs with C-terminal truncations of one or both PPXY motifs and/or the intervening Y³⁸² residue (Fig. 6A). Transfection of HEK 293 cells with these constructs followed by extraction of cell lysates and Western blotting showed a downward shift of the HA band consistent with the predicted size of the three truncated proteins (Fig. 6B, “Input”). Following immunoprecipitation of FLAG-tagged IR, we confirmed the previously observed interaction of IR with full-length wild-type (WT) ARRDC3 (1 to 414^WT) (Fig. 6B). While removal of the C-terminal PPLY motif (1 to 390) did not affect binding to the IR, further deletion of nine amino acids including the Y³⁸² residue (1 to 381) or deletion of the entire C terminus to residue 345, which led to removal of the PPSY motif, reduced the interaction almost completely (Fig. 6B). Furthermore, the 1 to 381 truncation which removed Y³⁸² and the C-terminal PPLY motif also reduced insulin-induced tyrosine phosphorylation of ARRDC3 in HA precipitates by nearly 60% (Fig. 6C). To determine specifically the role of Y382 in interaction with IR, we introduced a Y-to-A mutation at position 382 (SI Appendix, Fig. S4B) and transfected cells with the mutant or WT ARRDC3. As observed in the truncation experiments, binding between ARRDC3 and IR was almost completely abolished with the Y382A mutant compared to WT ARRDC3 (Fig. 6D). Thus, ARRDC3 interacts with IR specifically through a region in its C terminus tail containing the Y³⁸² residue.

Fig. 6.

Tyrosine 382 residue in ARRDC3 C-tail regulates its interaction with IR. (A) schematics depicting structure and truncation mutations of the ARRDC3 C-terminal tail. (B) Coimmunoprecipitation assay of FLAG-tagged hIR and HA-tagged hARRDC3 truncation constructs transiently expressed in HEK 293 cells. (C) Tyrosine phosphorylation in anti-HA precipitates of full-length or 1 to 381 truncation mutant of HA-tagged hARRDC3. (D) Coimmunoprecipitation assay of FLAG-tagged hIR and HA-tagged WT or Y382A hARRDC3 in HEK 293 cells.

Discussion

The liver is a major target of insulin action, in which insulin controls the balance between glucose storage and production. Insulin resistance in the liver with unsuppressed hepatic glucose output is the major driver of fasting hyperglycemia in type 2 diabetes (20). Although there has been progress in management of type 2 diabetes (T2D), identifying new targets for treatment of hepatic insulin resistance in T2D and other conditions with associated with hepatic insulin resistance, such as nonalcoholic fatty liver disease, remains important as these disorders continue to increase in prevalence. Here, we show that the α-arrestin Arrdc3 is an insulin-regulated gene that may be such target. We demonstrate that Arrdc3 is a highly insulin-regulated gene in liver and is induced by hyperinsulinemia in both physiological and insulin resistance contexts. Liver-specific Arrdc3 deletion increases hepatic insulin sensitivity, leading to higher glycogen levels and lower endogenous glucose production. Mechanistically, this is associated with increased insulin-induced FOXO1 phosphorylation and suppression of PEPCK expression. Liver-specific Arrdc3 deletion also increased IR levels in membrane fractions due to a direct interaction between these proteins involving the C-terminal tail of ARRDC3 and phosphorylation of tyrosine 382.

The arrestin superfamily of proteins are molecular adapters that have multiple cellular functions. The β-arrestins are the best studied and regulate G-protein–coupled receptor (GPCR) inactivation and recycling, as well as scaffolding of multiple signaling complexes (21). The less-studied α-arrestin subfamily of proteins has also been shown to modulate intracellular trafficking of GPCRs and integrin receptors (8, 15–17), but a role in insulin receptor signaling has not been shown. Rodent and human studies, however, have demonstrated a metabolic role of α-arrestin proteins, including regulation of muscle, adipose tissue, and fibroblast glucose uptake, resistance to age-related obesity, and thermogenesis (9, 22, 23). The best-characterized member of the α-arrestins, Txnip, has been implicated in regulation of insulin sensitivity and adipogenesis in mice and humans (23, 24).

Transcriptional profiling studies by our group have shown that Arrdc3 is one of the most up-regulated transcripts with a nearly eightfold increase in liver during a hyperinsulinemic-euglycemic clamp (11). This was confirmed in this study and shown to be a tissue-specific effect. Indeed, in contrast to up-regulation in liver, Arrdc3 expression in WAT and skeletal muscle decreased during hyperinsulinemic clamp conditions. Consistent with this differential regulation by insulin, Arrdc3 expression has been shown to increase by two- to fivefold in response to fasting in adipose tissue and muscle, while it tends to decrease in liver during a fast (9). Arrdc3 is also induced in livers of hyperinsulinemic HFD-induced or genetically obese (ob/ob) mice, despite the presence of hepatic insulin resistance and hepatic steatosis (25).

While a role of ARRDC3 in diabetes has not been established, genome-wide linkage studies in humans have shown a sex-specific association between morbid obesity and a haplotype in the 5q13-15 chromosome region surrounding the ARRDC3 gene (9). On the other hand, global Arrdc3 deletion in mice protects against age-induced obesity, insulin resistance, and fatty liver, leading to improved glucose tolerance (9). While adipose-specific deletion of Arrdc3 causes some improvement in glucose tolerance and adiposity, the higher insulin sensitivity phenotype is not recapitulated (10), indicating that additional tissues contribute to the phenotype of the whole-body knockout. Here, we demonstrate that loss of Arrdc3 in hepatocytes increases whole-body and hepatic insulin sensitivity as assessed by euglycemic-hyperinsulinemic clamp. The insulin-sensitizing effects of Arrdc3 deletion in the liver are associated with increased expression of glucokinase and lower expression of PEPCK, resulting in higher glycogen accumulation and lower glucose output. The liver may also contribute to enhanced glucose disposal given the negative hepatic glucose production, suggesting a switch from net glucose production to net uptake (26). This notion is supported by higher glycogen accumulation in the liver over the transition to the fed state.

The metabolic improvements seen upon Arrdc3 deletion in the liver are associated with enhanced downstream insulin signaling to phosphorylate and inhibit FOXO1, a key transcription factor regulating glucose production in liver (13, 14). Indeed, liver-specific ablation of FOXO1 is sufficient to rescue hyperglycemia and other metabolic derangements caused by loss of AKT1 and AKT2 or IR (27, 28). The fact that no changes were detected in Ser473 phosphorylation of AKT, a major regulator of FOXO1, or GSK3 at sites linked to AKT activity (29), suggests that kinases other than AKT may contribute to FOXO1 Thr24 regulation in L-Arrdc3 KO mice as has been previously suggested (30). Serum and glucocorticoid-inducible kinase, which is activated by insulin in a PI3K-dependent manner (31), has been shown to also contribute to regulatory phosphorylation on Thr32 of FOXO3a which is homologous to Thr24 of FOXO1 (32), indicating that Arrdc3 deletion in the liver likely modulates specific pathways downstream of IR signaling.

Exactly how this enhanced insulin signaling occurs will require further study; however, our observations that IR and ARRDC3 coimmunoprecipitate together and that ARRDC3 is tyrosine-phosphorylated in response to insulin stimulation argue in favor of a direct interaction between these proteins. In addition, finding that IR protein levels are increased in membrane fractions of L-Arrdc3 KO mice suggests that ARRDC3 may regulate membrane trafficking of IR, similar to its effects on integrin ITGB4 (15, 16) and the β2-adrenergic receptor (7–9, 17). This interaction also may regulate IR activity as indicated by decreased IR autophosphorylation upon ARRDC3 overexpression, consistent with the tumor-suppressing functions ascribed to ARRDC3 interactions with integrins (15, 16). Our protein truncation and point-mutagenesis approaches establish that the conserved Y³⁸² residue within the C-terminal tail of ARRDC3 is the critical site for the interaction with IR and that phosphorylation at this residue could create interaction opportunities with additional signaling proteins, such as those containing phosphotyrosine-binding and Src homology 2 domains. Taken together, our data suggest that Arrdc3 may be part of a negative feedback loop and act as a physiological brake on insulin action in states of persistent IR activation due to hyperinsulinemia. The finding that Arrdc3 expression increases in livers of models of insulin resistance raises the possibility that dysregulated Arrdc3 expression may contribute to features of the metabolic syndrome.

In summary, our study demonstrates a role for ARRDC3 in regulating insulin receptor signaling in liver. ARRDC3 interacts with proximal components of the insulin-signaling cascade, including the IR itself, and these interactions could have important roles in liver insulin actions, such as hepatic suppression of gluconeogenesis that may become disrupted under metabolic distress. The identification of further ARRDC3 interactions at the plasma membrane and exploration of the associated mechanisms in different tissues and metabolic contexts will contribute to understanding the roles of α-arrestins in metabolic regulation and signaling crosstalk in physiological conditions and disease states such as insulin resistance and type 2 diabetes.

Methods

Animals.

All procedures described were approved by the Institutional Animal Care and Use Committee of Harvard University, Cambridge, MA, and the Joslin Diabetes Center, Boston, and were in accordance with NIH guidelines. Mice lacking Arrdc3 expression in liver (L-Arrdc3 KO) were generated by crossbreeding Alb-Cre–expressing mice [B6.Cg-Tg(Alb-cre)21Mgn/J, C57Bl6 background, Jackson Laboratories] and Arrdc3-floxed mice (Arrdc3tm1.1Rlee). Arrdc3-floxed mice without Alb-Cre expression were used as controls. Transgenic mice carrying insulin receptor floxed alleles (IR^fl/fl) were generated as previously described (33). LIRKO were generated by crossbreeding IR^fl/fl mice with Alb-Cre–expressing mice. CRE-negative IR^fl/fl mice were used as controls. All mice were housed under controlled temperature (20°C to 21 °C) and a light/dark cycle with free access to chow diet (PicoLab20 Mouse Diet; Labdiet; 23% protein, 21% fat, and 55% carbohydrates) and water. For fasting/refeeding experiments, mice were randomly assigned to one of the following groups: ad libitum, overnight fasting, or overnight fasting and refeeding for 6 h before collecting plasma samples and tissues.

Body Measurements.

Whole-body fat and lean masses were noninvasively measured using proton 1H-MRS (Echo Medical Systems). Body weight was recorded weekly using a properly calibrated scale.

Glucose, Insulin, and Pyruvate Tolerance Tests.

Mice were fasted for 16 h prior to a glucose tolerance test and a pyruvate tolerance test or for 4 h prior to an insulin tolerance test. Fasted mice were injected intraperitoneally with glucose (2 g/kg), pyruvate (2 g/kg), or insulin (1 IU/kg). Blood samples were withdrawn via the tail vein, and blood glucose levels were measured using glucose strips and glucometer (Contour, Bayer, Germany).

Hyperinsulinemic-Euglycemic Clamp Experiments.

Hyperinsulinemic-euglycemic clamp procedures for gene expression studies have been detailed elsewhere (11). To determine the role of Arrdc3 in insulin sensitivity, survival surgery was performed on 20-wk-old Arrdc3^fl/fl and L-Arrdc3 KO mice at 6 d prior to clamp experiments to insert a catheter in the jugular vein. Mice were fasted overnight (∼17 h) for the clamp, and a 2-h hyperinsulinemic-euglycemic clamp was conducted in conscious mice with a primed and continuous infusion of human insulin [priming at 150 mU/kg body weight, followed by 2.5 mU/kg/min (Humulin; Eli Lilly)]. A 20% glucose solution was infused at variable rates during the clamps to maintain euglycemia. Whole-body glucose turnover was assessed with a continuous infusion of [3-³H]glucose (PerkinElmer). Nonmetabolizable glucose analog 2-deoxy-d-[1-¹⁴C]glucose (2-[¹⁴C]DG) (PerkinElmer) was administered as a bolus (10 μCi) at 75 min of the experiment to detect insulin-stimulated glucose uptake in individual organs. Whole-body glucose turnover was calculated from the ratio between [3-³H]glucose infusion rate and the specific activity of plasma glucose at the end of the basal period. Whole-body glycolysis was determined by linear regression analysis of ³H₂O concentrations in plasma during 80, 90, 100, 110, and 120 min of the clamp to indicate the rate of appearance. Whole-body glycogen synthesis was estimated by subtracting whole-body glycolysis from whole-body glucose uptake, assuming that glycolysis and glycogen/lipid synthesis account for the majority of insulin-stimulated glucose uptake. Hepatic glucose production was determined by subtracting the glucose infusion rate from the whole-body glucose turnover (34). Mice were anesthetized at the end of the experiments, and tissues were collected for biochemical analysis.

Membrane Fractionation.

Membrane fractions were isolated from livers from overnight-fasted or 6-h refed Arrdc3^fl/fl and L-Arrdc3 KO mice using a Mem-PER Plus kit (Thermo Fisher, catalog no. 89842) according to the manufacturer’s instructions. Briefly, 20 to 40 mg of liver tissue was washed, ground, permeabilized, and homogenized. Homogenates were centrifuged at 16,000 × g for 15 min at 4 °C, and supernatants were separated as cytosolic fraction. The remaining pellet was solubilized and centrifuged at 16,000 × g for 15 min at 4 °C, and supernatant was collected as membrane protein fraction.

Biochemical Analyses and Calculation.

Glucose concentrations during the clamp were measured using an Analox GM9 Analyzer (Analox Instruments, London). Plasma concentrations of [3-³H]glucose, 2-[¹⁴C]DG, and ³H₂O were measured following deproteinization of plasma samples by scintillation counter. HGP rates, whole-body glycolysis, and glycogen synthesis rates and insulin-stimulated whole-body glucose turnover rates were calculated as previously described (34). Nonmetabolized 2-[¹⁴C]DG in individual tissues were isolated using ion exchange columns and measured in a scintillation counter. Plasma insulin levels were measured using an ultrasensitive insulin enzyme-linked immunosorbent assay kit (Alpco Diagnostics), and plasma glucagon levels were measured using a Glucagon enzyme immunoassay kit (Sigma-Aldrich). Liver glycogen levels were measured using a glycogen assay kit (Sigma-Aldrich) according to the manufacturer’s protocol.

For qRT-PCR, RNA was extracted from homogenized liver, adipose tissue, gastrocnemius, and kidney samples by TRIzol (Life Sciences, Carlsbad, CA) method according to the manufacturer’s protocols. The complementary DNA (cDNA) was synthesized from 2 μg of total RNA by use of High Capacity cDNA synthesis kit (Applied Biosystems), and reactions were performed using the Sybr Green supermix (Bio-Rad, Hercules, CA) or Taqman gene expression assays (Applied Biosystems). The following primer sequences or probes were used: Pcx F—5′ GGGATGCCCACCAGTCACT 3′; Pcx R—5′CATAGGGCGCAATCTTTTTGA 3′; Pck1 F—5′ AAGCATTCAACGCCAGGTTC 3′; Pck1 R—5′ GGGCGAGTCTGTCAGTTCAAT 3′; TBP F—5′ GCAGTGCCCAGCATCACTAT 3′; TBP R—5′ GCCCTGAGCATAAGGTGGAA 3′; Arrdc3(Mm00626887_m1), Gck(Mm00439129_m1), Gys2(Mm00523953_m1), Pygl(Mm01289790_m1), G6Pc(Mm00839363_m1), and TBP(Mm00446971_m1). Relative gene expressions were calculated in comparison to the level of the housekeeping gene.

Plasmids.

Human ARRDC3 cDNA (ABMgood, catalog no. ORF000657) was subcloned into an EcoR1 restriction site of an empty pKH3 vector carrying an N-terminal 3xHA tag (Addgene, catalog no. 12555) using the In-Fusion HD cloning system (Takara). Human IR cDNA was cloned into the 3xFLAG CMV-14 mammalian expression vector (Sigma) as previously described (35). For truncations on the C terminus tail of pKH3-ARRDC3, a TCA stop codon was introduced on the underlined locations using an InFusion cloning system (Takara). The forward primer common to all constructs was the following: 5′-AATCGCTGGATCCGAATTCATGGTGCTGGGAAAGGT-3′. The reverse primers were the following: 1 to 390 5′-AATGCCATCGATTGAATTTCAAGGCAAGAATCGAAACTCCT-3′; 1 to 381 5′-AATGCCATCGATTGAATTTCATGCAAACAGTGGTCCTTGAAG-3′; and 1 to 345 5′-AATGCCATCGATTGAATTTCATGCTTCAGGTCTTTCAGGAAG-3′. To introduce the Y382A mutation on ARRDC3, the Y382 codon (TAT) was replaced by an A382 (GCT) codon using a Quikchange site-directed mutagenesis kit from Agilent. The primers used were the following: FWD—5′-CTTCAAGGACCACTGTTTGCAGCTATCCAGGAGTTTCGATTCTTG-3′; and REV—5′-CAAGAATCGAAACTCCTGGATAGCTGCAAACAGTGGTCCTTGAAG-3′. Sanger sequencing from a M13 sequence confirmed the mutation.

Immunoblotting.

For coimmunoprecipitation studies, cells were collected in a buffer containing 50 mM Tris⋅HCl, 500 mM NaCl, 1 mM EDTA, 1% Triton-X, pH 7.4, and protease/phosphatase inhibitors (Biotool) and lysed by sonication. For insulin signaling, tissue fragments were lysed in RIPA buffer (EMD Millipore) containing protease and phosphatase inhibitors using a Tissuelyser II device (Qiagen). Equal protein amounts (∼10 μg) were resolved by sodium dodecyl sulfate/polyacrylamide gel electrophoresis and transferred to polyvinylidene fluoride membranes (EMD Millipore). Membranes were immunoblotted with the following antibodies: p-IR/IGF1R (Cell Signaling, #3024), IRβ (Santa Cruz, sc-711), p-IRS-1 (Millipore, 09-432), IRS-1 (BD, 611394), p-AktS473 (Cell Signaling, #9271), Akt pan (Cell Signaling, #4685), p-FOXO1/3 (Cell Signaling, #9464), FOXO1 (Cell Signaling, #9454), p-GSK3 α/β (Cell Signaling, #8566), GSK3α (Cell Signaling, #4818), phosphotyrosine (pY20) (Santa Cruz, #sc508), PEPCK (Cell Signaling, #12940), GAPDH (Cell Signaling, #5174), FLAG (Sigma, #F3165), HA (Santa Cruz, #sc7392), and vinculin (Chemicon, #3574).

Coimmunoprecipitation.

ARRDC3 and IR plasmids were transiently expressed (6 μg) alone or in combination with HEK 293 cells grown in 10-cm plates using Superfect (Qiagen) according to the manufacturer’s instructions. Lysates containing equal protein amounts were incubated with prewashed magnetic beads conjugated with monoclonal anti-FLAG M2 (Sigma) or anti-HA (Thermofisher) antibodies for 1 h followed by serial washes with 1 mL wash buffer (50 mM Tris, 150 mM NaCl, 1% Triton X-100). Precipitated proteins were released from beads by competitive elution with either FLAG or HA (0.15 mg/mL) peptides during 40 min at 4 °C.

Statistical Analysis.

Data are expressed as means ± SEs. The significance of difference in mean values was determined using two-way analysis of variance (ANOVA) and Student’s t test where applicable. The statistical significance was set at a P value of <0.05.

Data Availability.

All plasmids generated for this study will be made available to qualified researchers for their own use. Reagent requests should be directed and will be fulfilled by R.T.L. (richard_lee@harvard.edu) or C.R.K. (c.ronald.kahn@joslin.harvard.edu).