SH2B1 Enhances Insulin Sensitivity by Both Stimulating the Insulin Receptor and Inhibiting Tyrosine Dephosphorylation of Insulin Receptor Substrate Proteins

SH2B1 knockout and TgKO mice have been described previously (29,31) and were backcrossed for six generations onto a C57BL/6 genetic background. Mice were housed on a 12-h light/dark cycle in the Unit for Laboratory Animal Medicine at the University of Michigan and fed either normal rodent diet (9% fat; Lab Diet) or high-fat diet (HFD; 45% fat; Research Diets) ad libitum with free access to water. Fat content was measured by dual-energy X-ray absorptiometry (Norland Medical System). Blood glucose levels were determined using glucometers (Bayer Corp). Plasma insulin was measured using a rat insulin enzyme-linked immunosorbent assay kit (Crystal Chem). Glucose tolerance tests (GTT) (2 g d-glucose/kg of body weight) and insulin tolerance tests (ITT) (1 IU/kg of body weight; Eli Lilly) were conducted as previously described (29,–31). To analyze insulin signaling, mice (fasted 16 h) were anesthetized with Averin (0.5 g of tribromoethanol and 0.25 g or tert-amyl alcohol in 39.5 ml of water; 0.02 ml/g of body weight) and treated with PBS or human insulin (3 units per mouse; Eli Lilly) via inferior vena cava injection. Five minutes after stimulation, gastrocnemius muscles, liver, and epididymal fat pads were dissected, frozen in liquid nitrogen, and stored at −80°C. Tissues were homogenized in ice-cold lysis buffer (50 mmol/l Tris HCl [pH 7.5], 1.0% NP-40, 150 mmol/l NaCl, 2 mmol/l EGTA, 1 mmol/l Na₃VO₄, 100 mmol/l NaF, 10 mmol/l Na₄P₂O₇, 1 mmol/l PMSF, 10 μg/ml aprotinin, 10 μg/ml leupeptin), and extracts were immunoblotted or immunoprecipitated with indicated antibodies. Animal protocols were approved by the University Committee on Use and Care of Animals.

COS7 and HEK293 cells were grown in DMEM supplemented with 5% bovine serum and transfected with indicated plasmids using Lipofectamine 2000 (Invitrogen). Chinese hamster ovary (CHO^IR and CHO^IR/IRS−1) cells were cultured in Ham's F-12 media supplemented with 8% FBS. Cells were deprived of serum for 16 h in DMEM (COS7 and HEK293) or F-12 (CHO) containing 0.6% BSA before being treated. Primary liver cells were isolated from male mice (8 weeks) by perfusion of the liver with type II collagenase (Worthington Biochem) and plated on collagen-coated plates in M199 containing 10% FBS, 100 units/ml penicillin, and 100 μg/ml streptomycin. After 2 h, primary cells were rinsed in PBS and cultured for an additional 16 h in Williams' Medium E (Sigma) supplemented with 0.6% BSA, 100 units/ml penicillin, and 100 μg/ml streptomycin.

Immunoprecipitation and immunoblotting were conducted as described previously (29,37). Proteins were visualized using the Odyssey Infrared Imaging System (Li-Cor Biosciences) or ECL (Amersham) and quantified using Odyssey 1.2 software (Li-Cor). Actin, phosphoAkt (Thr³⁰⁸), Akt, insulin receptor β, Myc, Shc, and tubulin antibodies were from Santa Cruz. The phosphotyrosine-specific antibody was from Upstate. The AS160 antibody was from Millipore and phospho Akt substrate antibody was from Cell Signaling. Phospho Akt (Ser⁴⁷³) was from BioSource. SH2B1 and IRS-1 antibodies have been described (14,37).

Cells were serum deprived for 16 h, treated with insulin, and solubilized in kinase lysis buffer (50 mmol/l Tris HCl [pH 7.5], 0.1% Triton X-100, 150 mmol/l NaCl, 1 mmol/l EDTA, 1 mmol/l Na₃VO₄, 1 mmol/l PMSF, 10 μg/ml aprotinin, 10 μg/ml leupeptin). The insulin receptor was precipitated with wheat germ agglutinin (WGA)-conjugated agarose beads, washed three times in wash buffer (50 mmol/l Tris HCl [pH 7.5], 0.5 mol/l NaCl, 0.1% Triton X-100), and twice in kinase reaction buffer (20 mmol/l HEPES [pH 7.6], 0.1% Triton X-100, 5 mmol/l MgCl₂, 100 μmol/l Na₃VO₄). WGA-immobilized proteins were preincubated in kinase reaction buffer supplemented with soluble glutathione S-transferase (GST) alone, GST-SH2B1, or GST-SH2 fusion proteins at 37°C. GST-IRS-1 (5–10 μg) and ATP (50 μmol/l) were added to initiate kinase reactions at 37°C. Reactions were stopped by adding SDS-PAGE loading buffer, and reaction mixtures were boiled immediately. Proteins were separated by SDS-PAGE and immunoblotted with indicated antibodies.

Immunopurified proteins were washed in lysis buffer and preincubated with GST-SH2B1 or GST (2 μg) in phosphatase reaction buffer (50 mmol/l Tris-HCl [pH 8.2], 100 nmol/l NaCl, 10 mmol/l MgCl₂, 1 mmol/l DTT) for 15 min at room temperature with constant mixing. Alkaline phosphatase (New England Biolabs) was added at the indicated concentration, and the mixtures were incubated an additional 30 min at room temperature. Reactions were stopped by adding SDS-PAGE loading buffer and mixtures were boiled immediately. Proteins were separated by SDS-PAGE and immunoblotted with indicated antibodies.

Data are presented as means ± SE. Differences between groups were determined by two-tailed Student's t tests or ANOVA. P < 0.05 was considered significant.

We previously generated SH2B1 transgenic (Tg) mice in which the expression of recombinant SH2B1 is controlled by the neuron-specific enolase promoter (31). Tg mice were crossed with SH2B1 knockout mice to generate SH2B1 transgenic/knockout compound mutant (TgKO) mice. In TgKO mice, recombinant SH2B1 is expressed in brain but not in other tissues, including liver, muscle, and adipose tissue (31). Neuron-specific restoration of SH2B1 in TgKO mice fully corrected leptin resistance and obesity and largely rescued the hyperglycemia and insulin resistance observed in SH2B1 null mice, indicating that neuronal SH2B1 indirectly regulates insulin sensitivity and glucose metabolism by controlling adiposity (31).

To determine whether loss of peripheral SH2B1 exacerbates dietary fat–induced insulin resistance, TgKO and wild-type littermates (7 weeks) were fed HFD. Body weight and adiposity were similar between wild-type and TgKO mice fed HFD (Fig. 1,A and B). However, fasting (16 h) blood glucose levels were 1.3-fold higher in TgKO mice than in wild-type mice fed HFD for 16 weeks (Fig. 1,C). Fasting plasma insulin levels were twofold higher in TgKO mice than in wild-type mice (Fig. 1,D). To examine insulin sensitivity, GTT and ITT were performed. Blood glucose levels were 23–26% higher in TgKO mice than wild-type mice 15 and 30 min after injection of d-glucose (Fig. 1,E). Exogenous insulin markedly reduced blood glucose in wild-type but not in TgKO mice during ITT (Fig. 1 F). These results indicate that loss of peripheral SH2B1 exacerbates HFD-induced insulin resistance, hyperglycemia, and glucose intolerance independent of obesity.

To examine insulin signaling in skeletal muscle, liver, and epididymal fat, mice (7 weeks) were fed HFD for 16 weeks and treated with insulin or PBS vehicle. Insulin markedly stimulated tyrosine phosphorylation of IRS-1 as well as IRS-1 association with p85, the regulatory subunit of the phosphatidylinositol 3-kinase, in skeletal muscle of wild-type mice (Fig. 2,A). Both IRS-1 phosphorylation and IRS-1-p85 association were markedly reduced in TgKO muscle (Fig. 2,A). Loss of peripheral SH2B1 also decreased insulin receptor autophosphorylation and impaired the ability of insulin to stimulate Akt phosphorylation on Thr³⁰⁸ and Ser⁴⁷³ in TgKO muscle (Fig. 2,A). Insulin-stimulated IRS-1 and Akt Thr³⁰⁸ phosphorylation were reduced by 44 and 52%, respectively, in TgKO muscle (Fig. 2 B).

Insulin signaling was also examined in liver and white adipose tissue (WAT) from HFD-fed mice. Relative to wild-type mice, basal IRS-1 phosphorylation was increased in both liver and WAT in TgKO mice; insulin stimulated IRS-1 phosphorylation in these tissues from wild-type but not TgKO mice (Fig. 2,C and E). Akt phosphorylation (Ser⁴⁷³) was also reduced in liver (Fig. 2,D) and WAT (not shown) in TgKO mice. AS160, a Rab-GAP, is an Akt-substrate involved in GLUT4 vesicle trafficking in adipocytes (38,39). To measure AS160 phosphorylation, WAT extracts were immunoprecipitated with anti–phospho-Ser/Thr Akt substrate antibody and immunoblotted with anti-AS160 antibody. Similar to IRS-1, basal AS160 phosphorylation was increased in adipose tissue from TgKO mice (Fig. 2,F); however, insulin failed to further stimulate AS160 phosphorylation (Fig. 2 F). Together, these data indicate that peripheral SH2B1 increases insulin sensitivity in mice by promoting insulin signaling, including the activation of the IRS protein/phosphatidylinositol 3-kinase/Akt pathway, in muscle, liver, and WAT.

To determine whether endogenous SH2B1 directly enhances insulin signaling, primary hepatocyte cultures were prepared from wild-type and SH2B1 knockout littermates, and treated with insulin. SH2B1 was detected in wild-type but not in knockout hepatocytes as expected (Fig. 3,A). Insulin stimulated tyrosine phosphorylation of insulin receptor and IRS-1, IRS-1 association with p85, and phosphorylation of Akt in wild-type hepatocytes; however, insulin receptor autophosphorylation, IRS-1 phosphorylation, IRS-1 association with p85, and Akt phosphorylation were all reduced in knockout hepatocytes (Fig. 3,A). Interestingly, SH2B1 deficiency impaired IRS-1 phosphorylation to a greater extent than insulin receptor autophosphorylation. In HEK293 cells, overexpression of SH2B1 markedly increased insulin-stimulated tyrosine phosphorylation of IRS-1; in contrast, a SH2B1 mutant in which the SH2 domain is disrupted because of replacement of Arg⁵⁵⁵ with Glu (R555E) functioned as a dominant negative to inhibit IRS-1 phosphorylation (Fig. 3,B). To further assess the role of the SH2 domain of SH2B1, IRS-1 and insulin receptor were coexpressed with ΔN504, an NH₂-terminal truncated form of rat SH2B1β (amino acids 504–670) that contained the entire SH2 domain and a minimal number of adjacent amino acids. ΔN504 also promoted IRS-1 phosphorylation in insulin-treated cells (Fig. 3 C). These data suggest that the SH2 domain of SH2B1 is not only required but also sufficient to promote insulin receptor–mediated phosphorylation of IRS-1.

To determine whether SH2B1 also promotes phosphorylation of other IRSs, Shc was coexpressed with SH2B1 or ΔN504. SH2B1 and ΔN504 enhanced insulin stimulation of Shc phosphorylation to similar levels (Fig. 3,D). By contrast, SH2B1 stimulated IRS-1 phosphorylation to a higher level than did ΔN504 (Fig. 3 C), suggesting that SH2B1 promotes phosphorylation of IRS-1 and Shc by different mechanisms.

SH2B1 directly binds via its SH2 domain to Tyr¹¹⁵⁸ within the activation loop of insulin receptor (40,41). To test whether this interaction modulates insulin receptor activation, CHO^IR cells, which stably express insulin receptor, were treated with insulin, and active insulin receptor was purified using WGA-conjugated agarose beads (42). Insulin receptor was then preincubated with purified GST-SH2B1 fusion protein and subsequently subjected to in vitro kinase assays using GST-IRS-1 fusion protein as substrate. Tyrosine phosphorylation of GST-IRS-1 was measured by immunoblotting with anti–phospho-tyrosine antibodies. GST-SH2B1, but not GST alone, dose-dependently stimulated insulin receptor kinase activity as indicated by increased phosphorylation of GST-IRS-1 (Fig. 4,A). In similar experiments, a GST-SH2 fusion protein prepared by fusing the SH2 domain (amino acids 524–670 of SH2B1β) to GST was preincubated with WGA-purified insulin receptor in vitro kinase assays. The SH2 domain of SH2B1 was sufficient to enhance insulin receptor catalytic activity, stimulating IRS-1 phosphorylation by ∼66% (Fig. 4,B). Additionally, the SH2 domain of SH2B1 also promoted the catalytic activity of insulin receptor immunopurified with an anti–phospho-tyrosine antibody, increasing IRS-1 substrate phosphorylation by ∼79% (Fig. 4 C).

To determine whether Tyr¹¹⁵⁸ in insulin receptor is involved in SH2B1 stimulation of insulin receptor activity, Tyr¹¹⁵⁸ was replaced with Phe (Y1158F). COS7 cells were transiently transfected with insulin receptor or Y1158F and treated with insulin. Insulin stimulated autophosphorylation of both insulin receptor and Y1158F, but Y1158F autophosphorylation was reduced (Fig. 4,D). Y1158F phosphorylated IRS-1 in response to insulin (data not shown), indicating that Y1158F retains the ability to be activated and to phosphorylate its substrates. Insulin receptor and Y1158F were purified using WGA-beads, preincubated with GST-SH2B1, and subjected to in vitro kinase assays. SH2B1 stimulated insulin receptor kinase activity by approximately fivefold; however, SH2B1 was unable to stimulate Y1158F catalytic activity (Fig. 4 E). Taken together, these data suggest that the physical interaction between the SH2 domain of SH2B1 and Tyr¹¹⁵⁸ in insulin receptor is required and sufficient for stimulation of insulin receptor catalytic activity.

SH2B1 directly binds to IRS-1 and IRS-2 in vitro (37), and insulin stimulated coimmunoprecipitation of SH2B1 with IRS-1 in cells (Fig. 5,A). To determine whether this physical interaction inhibits IRS-1 dephosphorylation by phosphatases, CHO^IR/IRS-1 cells, which stably express insulin receptor and IRS-1, were stimulated with insulin to promote tyrosine phosphorylation of IRS-1. Phosphorylated IRS-1 was immunopurified, preincubated with GST or GST-SH2B1, and subjected to in vitro dephosphorylation assays. IRS-1 bound to GST-SH2B1 but not to GST (data not shown). Alkaline phosphatase dose-dependently dephosphorylated IRS-1 on tyrosines in the GST-pretreated samples; in contrast, alkaline phosphatase was unable to dephosphorylate SH2B1-bound IRS-1 (Fig. 5 B). Insulin also promoted the association of SH2B1 with IRS-2, and SH2B1 similarly inhibited tyrosine dephosphorylation of IRS-2 (data not shown).

To determine whether SH2B1 inhibits IRS-1 dephosphorylation in cells, IRS-1 was coexpressed with PTP1B (a protein tyrosine phosphatase) in the absence or presence of SH2B1. PTP1B dephosphorylated IRS-1, and SH2B1 dose-dependently attenuated the ability of PTP1B to dephosphorylate IRS-1 (Fig. 5,C). To determine whether SH2B1 is able to promote IRS-1 phosphorylation without stimulating insulin receptor kinase activity, Y1158F was coexpressed with SH2B1. Although SH2B1 was unable to stimulate Y1158F kinase activity (Fig. 4,E), SH2B1 still markedly enhanced tyrosine phosphorylation of IRS-1 in Y1158F-expressing cells (Fig. 5 D). Thus, SH2B1 is likely to augment Y1158F-mediated phosphorylation of IRS-1 by inhibiting IRS-1 dephosphorylation by endogenous protein phosphatase(s).

To determine whether the SH2B1-IRS interaction sterically inhibits the binding of IRS proteins to phosphatidylinositol 3-kinase, IRS-1 and Y1158F were coexpressed with or without SH2B1 in HEK293 cells, and IRS-1-p85 association was examined by coimmunoprecipitation assays. Insulin stimulated coimmunoprecipitation of IRS-1 with p85, the regulatory subunit of phosphatidylinositol 3-kinase; importantly, SH2B1 markedly enhanced insulin-stimulated p85 binding to IRS-1 (Fig. 5,E). These data indicate that the SH2B1-IRS interaction does not interfere with IRS-phosphatidylinositol 3-kinase interaction but rather increases the IRS-phosphatidylinositol 3-kinase association by inhibiting IRS dephosphorylation. Consistent with these observations, SH2B1 also enhanced insulin-stimulated, Y1158F-mediated Akt phosphorylation (Fig. 5 F). Collectively, these data suggest that, in addition to enhancing insulin receptor catalytic activity via binding to Tyr¹¹⁵⁸, SH2B1 also promotes activation of the IRS protein/phosphatidylinositol 3-kinase/Akt pathway by inhibiting IRS dephosphorylation.

Insulin resistance is the primary risk factor for various metabolic diseases, including type 2 diabetes, nonalcoholic fatty liver disease, dyslipidemia, and cardiovascular disease. The molecular mechanisms underlying insulin resistance are extremely complex and not completely understood. It is commonly accepted that impairments in insulin signal transduction play a key role in the development of insulin resistance. We previously observed that insulin signaling is enhanced by SH2B1 (29). SH2B1 overexpression increases insulin receptor autophosphorylation and tyrosine phosphorylation of IRS-1 and IRS-2 in cultured cells (29,43). Similar observations were independently reported by two other groups (44,45). Additionally, we showed that genetic deletion of SH2B1 results in severe insulin resistance and type 2 diabetes in mice (29). However, SH2B1-null mice are also severely obese because of leptin resistance (30,–32), raising the possibility that insulin resistance may be secondary to obesity in SH2B1-null mice. Therefore, it was unclear whether peripheral SH2B1 directly regulates insulin sensitivity in insulin target tissues in vivo.

We generated TgKO mice that express SH2B1 only in the brain but not in peripheral tissues (e.g., liver, muscle, and adipose tissue). Body weight was similar between TgKO and wild-type littermates fed HFD, consistent with our previous conclusion that neuronal SH2B1 controls energy balance and body weight by enhancing leptin sensitivity in the hypothalamus (31). In the current study, we demonstrated that loss of peripheral SH2B1 markedly impaired insulin sensitivity independent of body weight. TgKO mice developed hyperglycemia, hyperinsulinemia, and glucose intolerance to a greater extent than wild-type littermates fed HFD. The ability of exogenous insulin to reduce blood glucose and to stimulate insulin receptor autophosphorylation and phosphorylation of IRS proteins and Akt in muscle, liver, and adipose tissue was significantly reduced in TgKO mice. These results suggest that peripheral SH2B1 serves as an endogenous insulin sensitizer. Insulin signaling has been shown to be attenuated by multiple intracellular signaling molecules (e.g., PTP1B, Grb10, Grb14, SOCS1, SOCS3, JNK, PKCθ, S6K, and IKKβ) that contribute to the development of insulin resistance (6,,,,,,,,,,,,,,,,,–24). Our data suggest that insulin sensitivity is controlled by a balance between these negative regulators and SH2B1 in insulin target cells.

SH2B1 promotes insulin signaling by stimulating insulin receptor catalytic activity. SH2B1 binds via its SH2 domain to phospho-Tyr¹¹⁵⁸ in the activation loop of insulin receptor (40,41). We showed that bacteria-derived SH2B1 markedly increased the ability of purified insulin receptor to tyrosyl phosphorylate IRS-1 in vitro. In contrast, SH2B1 was unable to stimulate the catalytic activity of Y1158F, an insulin receptor mutant lacking the binding site for SH2B1. In cells, SH2B1 overexpression promotes insulin receptor autophosphorylation as well as insulin receptor phosphorylation of its substrates (e.g., IRS-1, IRS-2, and Shc). Conversely, deletion of endogenous SH2B1 impaired insulin stimulation of insulin receptor autophosphorylation and IRS-1 phosphorylation in primary hepatocyte cultures. Consistent with these observations, SH2B1 complexes, which are immunoprecipitated from cell extracts, reportedly promote insulin receptor autophosphorylation by reducing the K_m for ATP (45). The same report also concluded that SH2B1 dimerization was required for its stimulation of insulin receptor autophoshorylation, because treatment of cells with dimerization domain peptide mimetics inhibited insulin receptor autophosphorylation and downstream pathways (45). However, the report did not provide evidence showing that the mimetics disrupted SH2B1 dimerization. In contrast, we observed that the SH2 domain alone was sufficient to stimulate insulin receptor catalytic activity in vitro. Moreover, ΔN504, an NH₂-terminal truncated SH2B1 containing the intact SH2 domain but completely lacking both dimerization and pleckstrin homology domains, still markedly enhanced insulin-stimulated tyrosine phosphorylation of both IRS-1 and Shc. Conversely, R555E, a SH2B1 mutant with a defective SH2 domain, inhibited insulin signaling as a dominant negative mutant. These data indicate that the SH2 domain of SH2B1 is both required and sufficient to stimulate insulin receptor kinase activity. Because Tyr¹¹⁵⁸ phosphorylation occurs early in the activation of the insulin receptor kinase (42,46,47), binding of the SH2 domain of SH2B1 to phospho-Tyr¹¹⁵⁸ may stabilize insulin receptor in an active conformation. Alternatively, SH2B1-insulin receptor interaction may facilitate insulin receptor binding to its substrates.

Insulin stimulated the binding of SH2B1 to IRS-1 or IRS-2. Importantly, SH2B1 directly inhibited tyrosine dephosphorylation of IRS-1 and IRS-2 by recombinant phosphatase in vitro and by PTP1B in cultured cells. Although unable to stimulate Y1158F catalytic activity, SH2B1 still enhanced Y1158F-mediated phosphorylation of IRS-1 in cultured cells, presumably by inhibiting IRS-1 dephosphorylation by endogenous tyrosine phosphatase(s). Consistent with these observations, deletion of endogenous SH2B1 impaired tyrosine phosphorylation of IRS-1 to a greater extent than insulin receptor autophosphorylation in primary hepatocyte cultures. Together, these data suggest that the SH2B1-IRS physical interaction inhibits IRS dephosphorylation by tyrosine phosphatases. Interestingly, the SH2B1-IRS interaction did not inhibit the ability of phosphorylated IRS proteins to bind to p85, the regulatory subunit of phosphatidylinositol 3-kinase; in contrast, it enhanced insulin-stimulated IRS-p85 association and subsequent Akt phosphorylation and activation, presumably by protecting IRS proteins against dephosphorylation. These data suggest that SH2B1 does not compete with p85 for the same binding sites in IRS proteins, and that the SH2B1-IRS interaction does not sterically interfere with the IRS-p85 interaction. Therefore, the SH2B1-IRS interaction may selectively block IRS interaction with tyrosine phosphatases, thereby inhibiting IRS dephosphorylation. Alternatively, the SH2B1-IRS interaction may alter IRS conformation so that multiple tyrosine phosphorylation sites, in addition to SH2B1-bound site(s), are resistant to dephosphorylation, but still retain their ability to bind to downstream signaling molecules and activate downstream pathways including the phosphatidylinositol 3-kinase/Akt pathway.

In conclusion, SH2B1 appears to promote insulin sensitivity in animals by multiple mechanisms (Fig. 6). Neuronal SH2B1 increases insulin sensitivity indirectly by reducing adiposity (31). In muscle, liver, and adipose tissue, SH2B1 binds to insulin receptor and stimulates insulin receptor catalytic activity to globally activate pathways downstream of insulin receptor. SH2B1 binds to both IRS-1 and IRS-2 and protects IRS proteins from tyrosine dephosphorylation, augmenting and/or prolonging IRS protein-mediated pathways. In addition, SH2B1 forms dimers, and each SH2B1 molecule in a SH2B1 dimer may simultaneously bind to insulin receptor and IRS-1 (or IRS-2), thereby stabilizing insulin receptor/IRS-1 (or insulin receptor/IRS-2) complexes. Therefore, SH2B1 and molecules that mimic these functions of SH2B1 are potential therapeutic targets for the treatment of obesity and/or type 2 diabetes.

This study was supported by National Institutes of Health Grants RO1 DK-065122, RO1 DK-073601 (to L.R.), and F31NS056575 (to D.L.M.) and by American Diabetes Association Award 1-09-RA-156 (to L.R.). This work utilized the cores supported by the Michigan Diabetes Research and Training Center (funded by National Institutes of Health 5P60 DK-20572), the University of Michigan's Cancer Center (funded by National Institutes of Health 5 P30 CA-46592), the University of Michigan Nathan Shock Center (funded by National Institutes of Health P30AG013283), and the University of Michigan Gut Peptide Research Center (funded by National Institutes of Health DK-34933).

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

We thank Xiao Jian Sun (University of Chicago) for providing GST-IRS-1 plasmids and Liang Sheng for his assistance in primary hepatocyte cultures.