MAD2-Dependent Insulin Receptor Endocytosis Regulates Metabolic Homeostasis

Insulin binds to and activates insulin receptor (IR), a receptor tyrosine kinase (1–6). Insulin-stimulated IR triggers two distinct signaling cascades: the phosphatidylinositol 3-kinase (PI3K)-AKT pathway and the mitogen-activated protein kinase (MAPK) pathway. These two pathways maintain glucose, lipid, and amino acid homeostasis and regulate cell growth and proliferation (4,7–9). Insulin-activated IR undergoes endocytosis and is either recycled to the cell surface to initiate a new round of insulin signaling or degraded in the lysosome (10–13) (Fig. 1A). Perturbations of insulin signaling cause metabolic disorders, such as diabetes and severe insulin resistance syndromes (4,8,14).

IR endocytosis, particularly in the liver, facilitates insulin clearance from portal circulation (15,16). Therefore, in addition to insulin secretion by pancreatic β-cells, hepatic insulin clearance by IR endocytosis plays a crucial role in maintaining proper insulin levels and its action in peripheral insulin-target tissues (15). Hepatic IR knockout (KO) or carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM1) KO mice exhibited defects in IR endocytosis, leading to hyperinsulinemia and systemic insulin resistance (17–19). Conversely, hepatic inhibition of EPH receptor B4 (EPHB4)-dependent IR endocytosis and lysosomal degradation improved insulin and glucose tolerance in obese mice (20). Furthermore, inhibition of INCEPTOR (insulin inhibitory receptor) enhanced insulin signaling in β-cells and improved glucose tolerance (21). We previously showed that accelerated IR endocytosis in liver caused whole-body insulin resistance (22), whereas delayed IR endocytosis in the liver by Src homology phosphatase 2 (SHP2) inhibition prolonged insulin signaling and increased insulin sensitivity in mice (23). Our findings support the idea that IR endocytosis terminates and redistributes insulin signaling. At present, the function of IR endocytosis in whole-body insulin sensitivity is not well understood. It is unclear whether the metabolic phenotypes described above depend solely on IR endocytosis or on other IR functions, such as the IR kinase activity, or even IR-independent functions.

Cell division regulators MAD2, BUBR1, and p31^comet regulate IR endocytosis and signaling (10,22,23). Mitosis arrest deficiency 2 (MAD2) directly binds to IR through the MAD2-interacting motif (MIM) in the COOH terminus of IR and collaborates with budding uninhibited by benomyl 1–related 1 (BUBR1) to recruit clathrin adaptor protein AP2 to IR (Fig. 1A). p31^comet inhibits the association of BUBR1-AP2 to IR, thereby preventing IR endocytosis. Here, we report that genetic ablation of the MIM of IR delays IR endocytosis and prolongs insulin action at the cell surface, while simultaneously increasing insulin and glucose counterregulatory factors and altering whole-body metabolic homeostasis. Our results demonstrate the importance of IR endocytosis in metabolic regulation.

Animal work described in this article was approved and conducted under the oversight of the Columbia University Institutional Animal Care and Use Committee. Mice were fed a standard rodent chow (no. 5053; LabDiet) or high-fat diet (HFD) (D12492; Research Diets) for indicated periods. All animals were maintained in a specific antigen-free barrier facility with 12-h light/dark cycles (6:00 a.m. on and 6:00 p.m. off). Two- to three-month-old male mice were used in this study unless otherwise noted. For inducing insulin resistance, mice were fed an HFD (60%) (D12492; Research Diets). Gene targeting strategies, hyperinsulinemic-euglycemic clamp, and metabolic cage studies are described in Supplementary Material.

Insulin signaling and IR endocytosis in vivo analyses were performed as previously described with some modifications (22–24). Male mice 2–3 months old were fasted overnight. Following anesthesia, mice were injected with 6 nmol Humulin (Eli Lilly) per mouse via inferior vena cava. Livers were removed at 3 min after injection. Adipose tissues and skeletal muscle were removed at 5 min and 7 min after injection, respectively. Hypothalamus was collected at 10 min after 30 nmol/mouse Humulin injection via inferior vena cava. Adipose tissues were homogenized in radioimmunoprecipitation assay buffer (50 mmol/L Tris [pH 8.0], 150 mmol/L NaCl, 1% [v/v] NP-40, 0.5% [w/v] sodium deoxycholate, 0.1% [w/v] SDS, and 1 mmol/L EDTA), and other tissues were mixed with lysis buffer B (50 mmol/L HEPES, 150 mmol/L NaCl, 10% [v/v] glycerol, 1% [v/v] triton X-100, 1 mmol/L EDTA, 0.5 mmol/L dithiothreitol, and 2 mmol/L phenylmethylsulfonyl fluoride) supplemented with cOmplete Protease Inhibitor Cocktail (Roche), PhosSTOP (Sigma-Aldrich), and 25 units/mL TurboNuclease (Accelagen), homogenized with Fisherbrand Bead Mill homogenizer, and then incubated on ice for 1 h. After centrifuge at 20,817g at 4°C for 30 min, the concentrations of cell lysate were measured with Micro BCA Protein Assay Kit (Thermo Fisher Scientific). The lysates were then analyzed with quantitative Western blotting (LI-COR, Lincoln, NE). The antibodies used for this study are listed in Supplementary Table 1.

Mouse primary hepatocytes were isolated from 2- to 3-month-old male mice with a standard two-step collagenase perfusion procedure as previously described (22,24). Isolated hepatocytes were resuspended with attached medium (Williams’ Medium E supplemented with 10% [v/v] FBS, 100 pmol/L insulin, 100 nmol/L dexamethasone, 5.5 µg/mL transferrin, 6.7 ng/mL sodium selenite, and 1% penicillin/streptomycin) and plated on collagen (no. C3867; Sigma-Aldrich)-coated dishes. After 4 h, the medium was changed to serum-free low-glucose DMEM. After 14–16 h, the cells were treated with insulin for analysis of IR signaling.

For serum preparation, blood was centrifuged and stored at −80°C after forming clots at room temperature for 30 min. Serum insulin and C-peptide were measured with ultrasensitive mouse insulin ELISA kits (no. 90080; Crystal Chem) and mouse C-peptide ELISA kits (80-CPTMS-E01; ALPCO). Plasma samples were collected from facial vein bleeding with venous blood collection tubes (41.13950.105; Sarstedt). Plasma glucagon, noradrenaline, adrenaline, and fatty acids (FA) were measured with mouse glucagon ELISA kits (81518; Crystal Chem), Bi-CAT adrenaline and noradrenaline ELISA kits (17-BCTHU-E02-RES; ALPCO), and FA quantification colorimetric/fluorometric kits (ab65341; Abcam). Blood glucose and HbA_1c levels from tail bleeding were measured with a glucometer (AlphaTrak) and A1C Now+ (Bayer Vital) test kits. Plasma and hepatic triglyceride levels were measured with triglyceride quantification colorimetric/fluorometric kits (MAK266; Sigma-Aldrich).

For glucose tolerance tests, mice were fasted for 6 h and their blood glucose levels (T = 0) were measured with tail bleeding. Then glucose (2 g/kg body wt) was injected intraperitoneally or using oral gavage. Blood glucose levels were measured at indicated time points after glucose injection. For insulin tolerance tests (ITT), mice fasted for 2 h were injected intraperitoneally with Humulin (6 nmol/kg body wt). For pyruvate tolerance tests (PTT), mice fasted for 14 h were injected with pyruvate (1 g/kg body wt) (P5280; Sigma-Aldrich).

Hepatic diacylglycerols (DAGs) levels and PKCε translocation were analyzed as previously described (25). Briefly, DAGs were extracted from tissues with 2:1 chloroform:methanol (v/v) with 0.01% butylated hydroxytoluene, dried down, and redissolved in 95:5:0.5 hexane/methylene chloride/ethyl ether (v/v/v) before analysis with liquid chromatography–tandem mass spectrometry.

Liver lysates with buffer A (20 mmol/L Tris-HCl, pH 7.4, 1 mmol/L EDTA, 0.25 mmol/L EGTA, 250 mmol/L sucrose, and freshly added protease and phosphatase inhibitors; Roche Diagnostics) were centrifuged (60 min, 100,000g, 4°C), and the supernatant was saved as the cytosolic fraction. The pellet was resuspended in buffer B (250 mmol/L Tris-HCl, pH 7.4, 1 mmol/L EDTA, 0.25 mmol/L EGTA, 2% Triton X-100, and freshly added protease and phosphatase inhibitors) and centrifuged (60 min, 100,000g, 4°C). An aliquot of the supernatant was saved as the membrane fraction. The resulting protein samples were subjected to Western blot analysis with anti-PKCε, anti–Na-K ATPase, and anti-GAPDH antibodies. After washing, membranes were incubated with horseradish peroxidase–conjugated secondary antibody (Cell Signaling Technology). Detection was performed with enhanced chemiluminescence.

Mouse tissues were fixed in buffered 10% neutral buffered formalin for hematoxylin-eosin (H-E) staining, oil red O staining, or periodic acid Schiff staining by Molecular Pathology Core at Columbia University. Stained slides were scanned with a Leica SCN400 scanner. For IR endocytosis assays, the livers were fixed in 10% neutral buffered formalin and embedded in paraffin blocks. Sections were deparaffinized, subjected to antigen retrieval with 10 mmol/L sodium citrate (pH 6.0), incubated with 0.3% H₂O₂, blocked with 0.3% BSA, and then incubated with anti-IR antibodies. The slides were counterstained with DAPI. Adipocyte sizes were assessed with Adiposoft software in ImageJ on scanned images of formalin-fixed adipose tissue.

Prism 9 was used for the generation of graphs and for statistical analyses. Results are presented as mean ± SD or mean ± SEM. Two-tailed unpaired t tests were used for pairwise significance analysis. Two-way ANOVA followed by Tukey multiple comparisons test was used. Sample sizes were determined based on the maximum number of mice. Power analysis for sample sizes was not performed. Randomization and blinding methods were not used, and data were analyzed after the completion of all data collection in each experiment.

All data are available in the main text or Supplementary Material. RNA-sequencing data are available in the Gene Expression Omnibus (GEO) (GSE240578).

We previously showed that the IR mutant with four residues in its MIM mutated to alanine (IR-4A) was deficient in MAD2 binding and insulin-induced IR endocytosis (10,22,23). To determine the physiological function of the IR-MAD2 interaction, we generated IR-4A knock-in (IR-4A) mice with CRISPR/Cas9 (Supplementary Fig. 1A). IR-4A mice survived and did not show discernable differences from wild-type (WT) mice. Metabolic cage analyses of male mice fed normal chow diet did not show significant effects on whole-body oxygen consumption, carbon dioxide production, respiratory quotient, energy expenditure, caloric intake, or activity in IR-4A mice compared with WT mice (Fig. 1A and Supplementary Fig. 2).

Insulin-activated IR is internalized, ultimately leading to downregulation of insulin signaling (26in vivo, overnight-fasted WT and IR-4A mice were injected with insulin via inferior vena cava. The livers were collected from these mice and subjected to analysis of localization of the endogenous IR. As expected, insulin treatment promoted IR internalization in the WT liver, whereas IR in IR-4A liver retained its plasma membrane (PM) localization (Fig. 1B and C). Consequently, the basal IR level in IR-4A mice was higher than that of WT mice (Fig. 1D and Supplementary Fig. 1B–F). These data establish the importance of the IR-MAD2 interaction in IR endocytosis and basal level of IR in vivo.

Next, we monitored the activating phosphorylation of IR (pY962 in the juxtamembrane region and pY1152/1153 in the kinase domain, with amino acid numbering for mature mouse IR-A isoform), and downstream phosphorylation events, including AKT and ERK1/2 in liver, skeletal muscle, and epididymal white adipose tissue (eWAT) (Fig. 2A and Supplementary Fig. 1E and F) on insulin stimulation. The levels of phosphorylated (p)IR and pAKT in IR-4A mice were slightly increased, as compared with WT mice. pERK levels were similar between the two groups. Consistent with in vivo findings, freshly isolated IR-4A primary hepatocytes showed slightly enhanced and extended pIR and pAKT at multiple time points and various insulin concentrations (Fig. 2B–E). No significant differences in pERK levels were observed between WT and IR-4A hepatocytes. Furthermore, the pIR and pAKT levels were mildly elevated in IR-4A in particular muscle under the refeeding condition (Supplementary Fig. 3A and B). These data suggest that disruption of the IR-MAD2 interaction delays IR endocytosis and degradation of IR/insulin complexes, marginally prolonging IR signaling.

IR endocytosis in liver plays an important role in insulin clearance. Liver-specific IR KO mice and CEACAM1 KO mice develop hyperinsulinemia (18,27). We thus examined the levels of insulin and a cleavage product of proinsulin, C-peptide. C-peptide level was not altered in IR-4A mice, suggesting that the IR-MAD2 interaction does not affect insulin secretion (Fig. 3A). In IR-4A mice, serum insulin level was significantly increased (Fig. 3B), thus lowering the C-peptide–to–insulin ratio (Fig. 3C). Hepatic CEACAM1 level was not altered by the IR-4A mutation (Supplementary Fig. 3C and D). These data suggest that disruption of the IR-MAD2 interaction delays insulin clearance, thus increasing peripheral insulin levels.

IR-4A mice displayed mild hypoglycemia possibly due to the high insulin levels, but no significant changes were observed after 6 h fasting (Fig. 3D). Plasma glucose concentrations are also regulated by glucose counterregulatory factors, such as glucagon and (nor)adrenaline, which increase glucose and FA in the bloodstream (28–30). In IR-4A mice fed normal chow diet, plasma glucagon and noradrenaline concentrations were increased (Fig. 3E and F), but not plasma adrenaline (Fig. 3G). These data suggest that glucose counterregulatory mechanisms are activated to maintain glucose homeostasis during chronic mild high-insulin and low-glucose conditions in IR-4A mice.

To determine the function of IR-MAD2 in metabolic homeostasis, we monitored metabolic phenotypes under normal chow feeding in young (2- to 3-month-old) and old (13-month-old) WT and IR-4A mice. Young IR-4A mice exhibited improved glucose tolerance and mildly increased insulin levels compared with WT mice (Fig. 4A–C), suggesting that the increased insulin levels facilitate glucose clearance. Strikingly, there were no significant differences in ITT and PTT (Fig. 4D–G).

To directly assess the whole-body insulin sensitivity of young IR-4A mice, we performed hyperinsulinemic-euglycemic clamp studies. During the clamp analysis, plasma glucose levels are adjusted between the groups to reach ∼110 mg/mL (Fig. 4H). By experimental design, the high dose of insulin infusion during the clamp significantly increased insulin levels, and to a similar extent in WT and IR-4A mice (Fig. 4I). Although IR-4A mice have a slightly higher glucose infusion rate (Fig. 4J and K), they displayed suppression of endogenous glucose production (EGP) and plasma FA similar to that of WT mice (Fig. 4L and M). The insulin-stimulated glucose uptake was slightly, but not significantly, increased (Fig. 4N). These data suggest that, despite the prolonged IR signaling in IR-4A mice, the disruption of the IR-MAD2 interaction does not significantly enhance whole-body insulin sensitivity in young mice, possibly due to increased negative feedback pathways on mild but chronic hyperinsulinemia and hypoglycemia.

IR-4A mice were followed longitudinally for examination of the metabolic function of IR-MAD2 in aging mice. With aging, peripheral insulin resistance progressively increases and insulin clearance declines, resulting in elevation of circulating insulin levels (31–33). Consistently, aged WT mice exhibited increased serum insulin levels, while insulin levels for aged IR-4A mice did not further increase (Fig. 3B). Thus, the fact that insulin levels did not differ between young and old IR-4A mice suggests an acquired defect in insulin homeostasis in WT mice rather than a reversal of the IR internalization in IR-4A mice. There was no significant difference in glucagon levels (Fig. 3H) and glucose tolerance between old WT and IR-4A mice (Fig. 4O and P). Furthermore, unlike in young IR-4A mice, mild hypoglycemia was not observed in the IR-4A mice as they aged. Strikingly, ITT demonstrated that old IR-4A mice are more sensitive to insulin with respect to glucose clearance (Fig. 4Q and R). These data suggest that the disruption of IR-MAD2 interaction delays the onset of age-related insulin resistance, possibly by prolonging insulin signaling. The data further suggest that delayed IR endocytosis improves insulin sensitivity in vivo after high-dose insulin injection.

Next, we investigated the metabolic effects of HFD feeding on WT and IR-4A mice. In comparison with WT littermates, IR-4A mice had a trend toward increasing body weight (Fig. 5A). As expected, serum insulin levels in mice fed HFD were substantially higher than in those fed normal chow (Fig. 5B). Both WT and IR-4A mice showed a marked increase in serum insulin concentrations after 8 days of HFD feeding, and this increase continued over time (Fig. 5B). Insulin levels of IR-4A mice were similar to those of WT mice 8 days after HFD feeding (Fig. 5B), with modest hypoglycemia and virtually unchanged glucose tolerance (Fig. 5E and F). Interestingly, IR-4A mice fed HFD for 8 days exhibited increased insulin sensitivity during an ITT (Fig. 5G and H). In contrast, IR-4A mice fed HFD for 11 weeks exhibited decreased fasting serum glucose levels (Fig. 5C and D and Supplementary Fig. 3E) but did not show increased glucose or insulin sensitivity (Fig. 5I–L). These results suggest that the disruption of the IR-MAD2 interaction delays the development of short-term diet-induced insulin resistance but cannot prevent prolonged diet-induced metabolic complications.

Fasting promotes hydrolysis of triacylglycerols (TG) in adipose tissues and increases circulating FA (34,35). FA are taken up by the liver, where they are esterified to TG or oxidized by β-oxidation into acetyl-CoA, which is then condensed to form ketone bodies. In addition, the energy released in this β-oxidation process is used by the liver for gluconeogenesis. Therefore, lipolysis in adipose tissues is essential for energy homeostasis during fasting.

To determine the role of the IR-MAD2 interaction in response to nutrient deprivation, we monitored the metabolic parameters of mice in fasting and refeeding conditions. In both fasting and random feeding conditions, IR-4A mice did not significantly differ from WT mice in body weight, liver mass, spleen mass, heart mass, kidney mass, or brown adipose tissue mass (Fig. 6A and Supplementary Fig. 4A and B). By H-E stain, IR-4A liver appears normal (Supplementary Fig. 4C). However, despite no observable difference in hepatic glycogen levels between WT and IR-4A mice (Supplementary Fig. 4D), there was a substantial increase in Oil red O staining and hepatic TG contents in IR-4A mice after a 20-h fast (Fig. 6B and C). Feeding decreased and similarly restored hepatic TG content in both WT and IR-4A mice (Fig. 6B and C). We found no difference in circulating TG levels during fasting and refeeding conditions (Fig. 6D), while the serum FA level was increased in fasted IR-4A mice (Fig. 6E). Strikingly, white adipose tissue (WAT) mass of IR-4A mice was significantly increased, especially eWAT, which increased by ∼30% in the fed state and by ∼40% after fasting for 20 h (Fig. 6A and Supplementary Fig. 4B). IR-4A mice had larger adipocyte size in fat deposits examined after a 20-h fast (Fig. 6F and G). Furthermore, IR-4A WAT showed a slight increase in hormone-sensitive lipase (HSL) phosphorylation (Supplementary Fig. 4E and F), suggesting that TG hydrolysis is potentially enhanced. These data suggest a role of the IR-MAD2 interaction in lipid metabolism during fasting.

We examined genes involved in lipid regulation in liver and eWAT of WT and IR-4A mice during 20 h of fasting. Although IR-4A livers have elevated TG content, lipogenesis genes including Srebp1 and Scd1 were decreased, whereas TG synthesis (Dgat1) and lipid transport (Cd36 and Slc27a2) genes were increased in IR-4A livers (Fig. 6H). A β-oxidation regulator, Cpt1, was expressed normally. In eWAT of IR-4A mice, expression of genes involved in lipogenesis (Acc1 and Fasn), lipid transport (Cd36, Lpl, and Fabp4), and lipolysis (Lipe and Plin1) was increased; however, expression of genes that control TG synthesis (Agpat3, Dgat1, and Dgat2) was normal (Fig. 6I). Collectively, these results suggest that disruption of the IR-MAD2 interaction may enhance lipogenesis, lipid transport, and lipolysis in the adipose tissue during fasting, thus increasing circulating FA and promoting hepatic fat accumulation.

To investigate the metabolic effect of the IR-MAD2 interaction in adipose tissues, we examined the potency of adipocyte differentiation and insulin-mediated suppression of lipolysis in vitro (Supplementary Fig. 5A and Bin vivo results, no defects were observed in the adipogenic capacity of IR-4A mouse embryonic fibroblasts. Insulin significantly inhibited isoproterenol-stimulated lipolysis in both WT and IR-4A mouse embryonic fibroblast–derived adipocytes (Supplementary Fig. 5C). A similar result was obtained in the ex vivo adipose tissue culture (Supplementary Fig. 5D), indicating that increased FA release from IR-4A adipose tissues is not associated with the defects in differentiation or insulin sensitivity.

IR-mediated transcytosis determines insulin sensitivity in peripheral tissues such as adipose tissue and muscle with a continuous capillary system (36,37). We tested whether disruption of the IR-MAD2 interaction inhibits the IR-mediated uptake and release of insulin in endothelial cells. We analyzed insulin fate in the CD31⁺ primary endothelial cells isolated from WT and IR-4A eWAT, and L6 myoblasts (Supplementary Fig. 6A and B). The insulin taken up by L6 cells gradually disappeared, and there was no detectable insulin in the medium. In contrast, the insulin taken by CD31⁺ primary endothelial cells decreased over time and was released in the supernatant, indicating insulin release from endothelial cells. In this condition, we did not observe significant differences between WT and IR-4A primary endothelial cells. Note that due to the detection limit of this assay, we could not further reduce the insulin concentration in the pulse to near-physiological concentrations.

Brain has a blood-brain barrier, limiting access of circulating insulin (38). Insulin signaling is delayed in brains of endothelial IR KO mice because insulin transcytosis is inhibited (39). Although IR levels were increased in the IR-4A hypothalamus, pIR and pAKT levels did not differ (Supplementary Fig. 6C). These data suggest that the disruption of the IR-MAD2 interaction did not affect the transcytosis of insulin in supraphysiological conditions.

Hepatic PKCε activation is required for the induction of lipid-induced hepatic insulin resistance (40,41). PM sn-1,2-DAGs in the liver activate PKCε, which then phosphorylates IR T¹¹⁶⁰ (T¹¹⁵⁰ in mouse), resulting in reduced IR kinase activity and hepatic insulin resistance (25,42). To examine the role of the IR-MAD2 interaction in hepatic regulation of PKCε, we examined the total DAG levels in PM using differential centrifugation and liquid chromatography–tandem mass spectrometry method (25). Disruption of the IR-MAD2 interaction reduced the sn-1,2-DAG level in the hepatic PM (Fig. 7A and B). In contrast, there were no discernible differences observed for sn-2,3-DAG and sn-1,3-DAG in the hepatic PM. Consistent with the reduction of PM sn-1,2-DAG in IR-4A liver, IR-4A mutation reduced the amount of membrane PKCε content in the liver with ∼30% lower translocation without altering cytosol PKCε content (Fig. 7C–F). Despite marginally prolonged and enhanced hepatic insulin signaling (Fig. 2), improved glucose clearance (Fig. 4A, J, and K), and decreased sn-1,2-DAG contents in the hepatic PM (Fig. 7A and B), IR-4A mutation did not affect EGP during the clamp (Fig. 4L). Previous studies demonstrating a relatively minor role for direct hepatic insulin signaling in regulating hepatic glucose production, in rodents without severe hepatic steatosis (43), might partly explain this inconsistency.

Because disruption of the IR-MAD2 interaction altered PM sn-1,2-DAG levels and fat accumulation in the liver during fasting, we assessed which metabolic pathways were affected under these circumstances. We first used metabolomics analysis in the plasma of WT and IR-4A mice fasted for 6 h. Heat map analysis with hierarchical clustering demonstrated that plasma metabolites in IR-4A mice clearly separated from those in WT mice (Supplementary Fig. 7A), implying their distinct metabolic phenotype in comparison with WT mice. Using a metabolite set enrichment analysis (44) on plasma metabolomics data, we found that the most affected metabolite sets were those related to linolenic acid and linoleic acid pathways (Fig. 7G). Consistent with these findings, linolenic acid, arachidonic acid, and docosahexaenoic acid levels were altered in IR-4A mice (Supplementary Fig. 7A). Further analysis demonstrated that plasma levels of polyunsaturated FA (PUFA) were increased by IR-4A mutation (Fig. 7H). In contrast, the levels of saturated FA, monounsaturated FA, and FA derivatives in the plasma from IR-4A mice did not form a distinct cluster away from that of WT mice (Supplementary Fig. 7B).

We next assessed transcriptomic profile in the liver and eWAT from WT and IR-4A mice after 6 h fasting using RNA sequencing. There is a distinction between differentially expressed (DE) genes with minimal overlap (Fig. 8A–C) and several shared DE genes with opposite fold changes (Fig. 8D), suggesting that IR-4A mice displayed distinct effects on transcriptomic profiles between liver and eWAT. The upregulated genes in IR-4A liver showed a significant enrichment for genes in pathways regulating the catabolism of leucine and valine, including Hibch, Ivd, Acadm, and Dbt, that contribute to FA synthesis (Fig. 8E–G). The downregulated genes in IR-4A liver showed a significant enrichment for genes in pathways regulating protein localization and targeting to the endoplasmic reticulum (Fig. 8E). An analysis of plasma metabolomics revealed that the levels of the branched chain amino acids leucine and valine were not altered in IR-4A mice (Supplementary Fig. 7C). These data suggest that although RNA-sequencing data suggest a potential increase in the catabolism of leucine and valine in IR-4A liver, liver leucine and valine catabolism may not regulate or be influenced by circulating levels of leucine and valine (Fig. 8E–G).

In eWAT, the upregulated genes in IR-4A mice showed an enrichment for genes in pathways regulating protein synthesis and translation (Fig. 8H). The downregulated genes in IR-4A eWAT showed an enrichment for genes in pathways regulating collagen formation and extracellular matrix organization (e.g., Col6a6, Col1a2, Col4a4, and Col3a1) (Fig. 8H and I). We also observed significantly lower expression of several macrophage markers, including Cd209b, Adgre4, Cd209f, and Lyve1, in IR-4A eWAT (Fig. 8J). Given the fact that PUFAs have anti-inflammatory effects (45,46), increased serum PUFA levels in IR-4A mice (Fig. 7H) may regulate macrophages population or gene expression in the eWAT. Altogether, the data support a profound change in the whole-body metabolism of IR-4A mice and the distinct effects on the transcriptomic profiles between the liver and adipose tissue.

In this study, we have determined the physiological function of IR-MAD2 interaction during insulin signaling and metabolic homeostasis. Specifically, we find that disruption of the IR-MAD2 interaction delays insulin-dependent in vivo. This genetic attenuation of IR endocytosis slightly enhances and prolongs insulin signaling through the PI3K-AKT pathway and does not appreciably affect the other branch of insulin signaling, the MAPK pathway. The elevated PI3K-AKT signaling in these mice may account for the improved insulin sensitivity when circulating insulin levels are similar. Alternatively, signaling events proximal to the PM are more effective in mediating metabolic regulation by insulin. At present, we do not know why signaling through the MAPK pathway is not enhanced in IR-4A hepatocytes. One possibility is that the IR-MAD2 interaction facilitates activation of the MAPK pathway by insulin.

in vivo when insulin levels are equal. We propose that increased levels of insulin and surface IR enhance insulin functions in the adipose tissues of IR-4A mice (Supplementary Fig. 8). Indeed, IR-4A mice exhibited enlarged WAT. We speculate that WAT of IR-4A mice may release more FA under nutrient-deficient conditions because they store more fat in the WAT during feeding. This partly explains why IR-4A mice exhibited elevated levels of circulating FA and hepatic TG during long-term fasting.

On the other hand, IR-4A mice exhibited increased glucose counterregulatory factors such as glucagon, thus affecting glucose control (Supplementary Fig. 8). Therefore, early during HFD feeding, disruption of the IR-MAD2 interaction delays insulin resistance, but at a later stage, the compensatory effect and the comparability of serum insulin levels result in diet-induced metabolic complications in IR-4A mice. The underlying mechanism of hyperglucagonemia in IR-4A mice remains unclear. One possibility is that mild hypoglycemia in IR-4A mice causes glucagon secretion. However, IR-4A mice do not reach the hypoglycemic threshold that induces glucagon secretion (47), suggesting additional regulation. Insulin inhibits glucagon release by promoting somatostatin secretion (48), and glucagon stimulates insulin secretion through glucagon receptor and glucagon-like peptide 1 receptor (49). In addition, the sympathetic and parasympathetic branches of autonomic nervous systems control insulin and glucagon secretion (50). It will be intriguing to determine whether or how the IR-MAD2 controls the activation of the counterregulatory factors in islet and nervous system.

Although we did not observe defects in insulin uptake and release in the endothelial cells and insulin signaling in the brain of IR-4A mice, we cannot rule out the possibility that MAD2 binding deficiency diminishes insulin transcytosis and alters local insulin levels because all experiments were conducted under supraphysiological insulin concentrations. While our whole-body knock-in IR-4A mouse is a powerful system for understanding the systemic function of IR-MAD2, the relative contributions of different processes in different tissues to the global phenotype are difficult to ascertain. Future experiments using tissue-specific conditional knock-in IR-4A mouse are needed to further define the tissue-specific functions of this interaction.

Hyperinsulinemia is associated with insulin resistance, although the cause-effect relationship remains obscure. IR-mediated insulin uptake and degradation are essential mechanisms for insulin clearance. Therefore, the rate of IR endocytosis and the extent of surface IR levels are directly related to circulating insulin levels. Previous studies with use of IR or CEACAM1 KO mice support the idea that the impaired insulin clearance causes insulin resistance. Conversely, IR endocytosis defects caused by hepatic EPHB4 or SHP2 inhibition reduce insulin resistance and improve glucose tolerance. p31^comet-deficient mice exhibit reduced basal surface IR levels and defective insulin signaling, while BUBR1 deficiency enhances insulin sensitivity and rescues the metabolic defects of p31^comet-deficient mice (22in vivo and establish a physiological function of IR trafficking in insulin and glucose homeostasis.

Acknowledgments. The authors thank Dr. Caiying Guo and colleagues (HHMI's Janelia Research Campus) for generating the IR-4A mice, Dr. Sei Higuchi for technical assistance, and Drs. Joseph Goldstein, David Mangelsdorf, and Li Qiang for helpful discussion. The authors are grateful to the Animal Resource Center at Columbia University for assistance with mouse maintenance and the Molecular Pathology Core for assistance with tissue processing and sectioning.

Funding. The authors’ research work has received funding from the National Institutes of Health (NIH) (R01DK132361 to E.C.; R00HL130574 and R01HL151611 to H.Z.; R01HL125649 and R01DK115825 to R.A.H.; P30DK063608 to E.C., R.A.H., and D.A.; UL1TR001873 to E.C. and H.Z.; and UC2DK134901, P30DK045735, and R01DK133143 to G.I.S.), Alice Bohmfalk Charitable Trust (to E.C.), American Lung Association (IA-828202), American Cancer Society (RSG-21-153-01-CCB) (to J.K.), and American Heart Association (postdoctoral fellowship 20POST35130003 and Career Development Award 23CDA1052177 to F.L.). The authors acknowledge the NIH funding sources to the Biomarkers Core Laboratory at the Irving Institute for Clinical and Translational Research, home to Columbia University’s Clinical and Translational Science Award (UL1TR001873); the Genomics and High Throughput Screening Shared Resource (NIH/National Cancer Institute Cancer Center Support Grant P30CA013696); and the Columbia University Digestive and Liver Diseases Research Center (P30DK132710).

The content in this manuscript is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

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

Author Contributions. J.P. designed and performed experiments and analyzed data. C.H. measured the body weight of a mouse on a high-fat diet and analyzed data. B.H., T.L., R.G., and A.N. performed hyperinsulinemic-euglycemic clamp study and PKCε analysis. F.L. and H.Z. performed transcriptomic profile analysis. J.K. performed metabolomic profile analysis. R.A.H. and D.A. provided suggestions. G.I.S. supervised hyperinsulinemic-euglycemic clamp study and PKCε analysis and provided suggestions. H.Y. supervised the project and provided suggestions. E.C. supervised the project, performed experiments, and analyzed data. J.P. and E.C. wrote the manuscript with input from other authors. E.C. 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.