Hepatic F-Box Protein FBXW7 Maintains Glucose Homeostasis Through Degradation of Fetuin-A

Type 2 diabetes mellitus (T2DM), caused by insulin resistance and/or impaired insulin secretion, has become a severe public health problem owing to its high prevalence and propensity for causing various complications, such as cardiovascular disease, diabetic nephropathy, and retinopathy. Despite decades of progress, the molecular mechanisms underlying insulin resistance are still not completely understood.

Recent studies in obesity research identified several proteins secreted by the liver, collectively termed hepatokines (1,2), which enter into circulation and regulate systemic glucose homeostasis. Manipulating the expression levels of these proteins can improve insulin resistance and/or glucose intolerance in obese rodents (3–6), demonstrating that they are important in the development of T2DM. The first hepatokine that has been proven to play a major role in the pathophysiology of T2DM is α2-HS-glycoprotein (fetuin-A) (1,2,7). Fetuin-A strongly promotes insulin resistance and hyperglycemia by perturbing insulin signaling pathways, inducing inflammatory cytokine expression, and repressing adiponectin expression (7). Conversely, mice lacking the Ahsg gene are protected from high-fat diet (HFD)-induced insulin resistance and weight gain (8). In support of this, many cross-sectional studies have shown that elevated fetuin-A levels strongly and positively correlate with hyperglycemia, insulin resistance, and dyslipidemia in humans (9–11); follow-up studies further demonstrated that high circulating levels of fetuin-A are strongly predictive of incident T2DM (10,11–13). However, the molecular basis for the increased fetuin-A concentrations in obesity has not been comprehensively addressed and requires further study.

Ubiquitination is the process by which ubiquitin is covalently coupled to lysine residues on target proteins via a cascade of enzymatic reactions carried out by activating (E1), conjugating (E2), and ligating (E3) enzymes, in which E3 ligase specifically recognizes substrates (14,15). Among several different E3 enzyme families, Skp1–Cullin1 (Cul1)–F-box protein ubiquitin ligases (SCFs), consisting of Skp1, Cul1, Rbx1, and one member of a family of F-box proteins, are the largest and best characterized family (16,17). F-box proteins contain a 40-amino-acid F-box domain, which docks with Skp1 to create a link to Cul1 (16,17). By recognizing and mediating the degradation of a large number of substrates, F-box proteins play a crucial role in diverse biological processes, such as cell proliferation and apoptosis, DNA damage and repair, and tumorigenesis (18,19).

In the current study, we identified the F-box and WD repeat domain-containing 7 (FBXW7) (also named Cdc4) as an E3 ubiquitin ligase for fetuin-A. We show that FBXW7 can interact with fetuin-A to promote its ubiquitination and degradation in a manner dependent on the phosphorylation of Serine 305 and Serine 309. Consequently, depletion of FBXW7 in the normal liver leads to accumulation of fetuin-A and impaired glucose homeostasis. On the other hand, expression of FBXW7 in the obese liver confers several beneficial metabolic effects including protection against hyperglycemia, insulin resistance, and glucose intolerance. Furthermore, we identify a functional human FBXW7 coding variant associated with elevated fetuin-A and blood glucose levels. Therefore, our results indicate that hepatic FBXW7 is a major regulator of fetuin-A expression and glucose metabolism.

Male C57BL/6 and ob/ob mice aged 8–12 weeks were purchased from the Shanghai Laboratory Animal Company (Shanghai, China). FBXW7-floxed mice were obtained from The Jackson Laboratory. All mice were housed at 21°C ± 1°C with a humidity of 55% ± 10% and a 12-h light/dark cycle. The HFD contained 60% kcal from fat, 20% kcal from carbohydrate, and 20% kcal from protein. The normal diet (ND) contained 10% kcal from fat, 70% kcal from carbohydrate, and 20% kcal from protein. The animal protocol was reviewed and approved by the Animal Care Committees of Zhongshan Hospital, Fudan University, and Shanghai Jiao Tong University School of Medicine.

All plasmids are constructed by our group. For generation of FBXW7 liver-specific knockout mice (LKO), FBXW7-floxed mice were injected via the tail vein with adenoviruses expressing Cre recombinase. For silencing of fetuin-A expression in FBXW7 LKO mice, short hairpin (sh)RNA knockdown adenoviruses were generated that target fetuin-A or express a scrambled negative control shRNA with the BLOCK-IT Adenoviral RNAi Expression System (Invitrogen). The shRNA designed to knock down fetuin-A had the following target sequence: 5′-CCGGAGCTTTGTTGGCCGTGGACTACTCGAGTAGTC CACGGCCAACAAAGC-3′. The adeno-associated virus (AAV)8 vector system was used to create AAV-FBXW7 or AAV-GFP vector. C57BL/6 mice were injected with 100 µL virus containing 2 × 10¹¹ AAV8 vector genomes via the tail vein. Overexpression of adenoviruses FBXW7 (wild type [WT] or A204T) or GFP was achieved by means of tail vein injection of adenoviruses containing FBXW7 or GFP (1 × 10⁹ plaque-forming units) in ob/ob or HFD mice. All viruses were purified by the cesium chloride method and dialyzed in PBS containing 10% glycerol prior to animal injection.

Glucose tolerance tests were performed by intraperitoneal injection of d-glucose (Sigma-Aldrich, St. Louis, MO) at a dose of 2.0 mg/g body wt after a 16-h fast. For insulin tolerance tests, mice were injected with regular human insulin (Eli Lilly, Indianapolis, IN) at a dose of 0.75 units/kg body wt after a 6-h fast. Blood glucose was determined using a portable blood glucose meter (LifeScan, Johnson & Johnson, New Brunswick, NJ).

For ELISA measurements of hepatic fetuin-A, freshly frozen liver tissues were homogenized in buffer containing 150 mmol/L NaCl, 10 mmol/L HEPES (pH 7.4), and 0.5% Triton X-100 plus antiprotease cocktail (Sigma-Aldrich). Fetuin-A concentrations in the liver and plasma were measured by commercial kits (R&D Systems).

Hepatic tissues or cells were lysed in radioimmunoprecipitation (RIPA) buffer containing 50 mmol/L Tris-HCl (pH 8), 150 mmol/L NaCl, 5 mmol/L MgCl₂, 2 mmol/L EDTA, 1 mmol/L NaF, 1% NP40, and 0.1% SDS. The following antibodies were purchased: anti–Flag M2 antibody (Sigma-Aldrich), anti–human influenza virus hemagglutinin (HA)-tag antibody (Cell Signaling Technology), anti–Flag tag antibody (Cell Signaling Technology), anti-FBXW7 antibody (Abcam, Cambridge, MA), anti–REV-ERBα antibody (Cell Signaling), anti-GFP antibody (Abcam), and anti-GAPDH antibody (Cell Signaling Technology).

Human subjects were drawn from the Shanghai Diabetes Study (SHDS) and FAt Distribution and diseasE (FADE) cohort. The characteristics of subjects in two cohorts are presented in Supplementary Table 1, and detailed information has previously been published (20–22). We genotyped the p.Ala204Thr (c.610G>A [rs189772026]) variant by using the MassARRAY Compact Analyzer (Sequenom, San Diego, CA), and the single nucleotide polymorphism passed through quality-control analysis (call rate >98%, concordant rate >99%, and consistent with the Hardy-Weinberg equilibrium). The statistical analysis was performed using SAS for Windows (version 8.0; SAS Institute, Cary, NC). Descriptive statistics were calculated for both WT (FBXW7^WT) and A204T variant (FBXW7^A204T) groups and compared by t test. Quantitative traits with a skewed distribution were logarithmically transformed and analyzed by linear regression adjusted for age and sex as confounding factors if appropriate. A two-tailed P value of <0.05 was considered statistically significant. For the analysis of hepatic fetuin-A expression, liver biopsy was performed in subjects who donated their partial livers for liver transplantation as previously described (23). The human study was approved by the Human Research Ethics Committee of Zhongshan Hospital, Fudan University, and Shanghai Jiaotong University Affiliated Sixth People’s Hospital. Written informed consent was obtained from each subject.

All values were shown as mean ± SEM. Statistical differences were determined by a two-tailed Student t test. Statistical significance is displayed as *P < 0.05, **P < 0.01, or ***P < 0.001.

Earlier in vitro studies have shown that mRNA levels of fetuin-A could be induced by NF-κB and endoplasmic reticulum (ER) stress pathways (24,25). Here, to test whether this regulation occurs in vivo, C57BL/6 mice were treated with lipopolysaccharide to activate NF-κB, as evidenced by upregulation of inflammatory cytokines (Supplementary Fig. 1A). Unexpectedly, mRNA levels of fetuin-A showed a time-dependent manner and were stable after 2 or 4 h of lipopolysaccharide challenge and reduced at 8 h, as shown by quantitative real-time PCR (Supplementary Fig. 1B). Hepatic and plasma fetuin-A protein concentrations remained unaffected, as determined by ELISA (Supplementary Fig. 1C and D). Similar results were also observed in mice treated with tunicamycin (Supplementary Fig. 1E), which activates ER stress response (Supplementary Fig. 1F). Therefore, we hypothesized that in obese subjects, fetuin-A might be upregulated by other mechanisms rather than by NF-κB and ER stress.

To verify this hypothesis, we examined fetuin-A expression in the livers and plasma of mice that were fed an HFD or an ND (Supplementary Fig. 2A–C). We found that circulating and hepatic fetuin-A protein concentrations were elevated in HFD mice (Fig. 1A and B). Interestingly, fetuin-A mRNA levels were not significantly changed in the livers of HFD mice (Fig. 1C), while mRNA levels of SREBP-1c, a master regulator of de novo lipogenesis in obesity (26), were elevated. Similar results were observed in leptin-deficient ob/ob mice and obese humans (Fig. 1D–I and Supplementary Fig. 2D–H). Therefore, we speculate that elevated circulating fetuin-A in obesity is attributed to a posttranscriptional regulatory mechanism.

Protein degradation often plays a crucial role in regulating gene expression posttranscriptionally. Thus, we investigated whether fetuin-A is degraded via proteasome. To test this, we treated human embryonic kidney (HEK)293T cells overexpressing HA-tagged fetuin-A with cycloheximide (CHX) to block protein synthesis. Western blot analysis showed that fetuin-A proteins were rapidly degraded; however, the degradation of fetuin-A was blocked by MG132, a proteasome inhibitor (Supplementary Fig. 3A). High–molecular weight species of ubiquitinated fetuin-A protein were also detected in cells transfected with an expression plasmid for ubiquitin in the presence of MG132 (Supplementary Fig. 3B), suggesting that fetuin-A is a target of ubiquitin-mediated proteasomal degradation.

As mentioned above, plenty of proteins regulated by ubiquitin-mediated degradation are recognized by the E3 ligases of SCF complexes, which contain a Cul1 protein to form a scaffold for the Skp1 and F-box proteins (16–18). For testing of whether fetuin-A is a target of SCF complexes, 293T cells were treated with MLN4924, an inhibitor of Cullin protein (27). We found that protein levels of fetuin-A were enhanced by MLN4924 in a dose-dependent manner (Supplementary Fig. 3C). Moreover, six dominant-negative (DN) Cullins were cotransfected with fetuin-A in 293T cells. Fetuin-A protein levels were specifically increased in cells expressing dominant-negative Cul1 (Supplementary Fig. 3D), further indicating that the ubiquitin-mediated degradation of fetuin-A is controlled by an SCF complex.

Next, for identification of which member of the F-box proteins could regulate fetuin-A stability, HA–fetuin-A was coexpressed in 293T cells with several members of the F-box protein family. As shown in Fig. 2A, expression of the F-box protein FBXW7 reduced the protein levels of fetuin-A. Other F-box proteins, such as FBXW1, FBXW2, and FBXW5, did not have this effect (Fig. 2A). Abundance of fetuin-B, which shares 22% homology with fetuin-A and is also increased in obese humans (28), was not affected by FBXW7 coexpression (Fig. 2B), further supporting the notion that FBXW7 specifically regulates the stability of fetuin-A.

Given that expression of FBXW7 decreases the levels of fetuin-A, we hypothesized that FBXW7 can interact with and promote the ubiquitination and degradation of fetuin-A. The interaction between FBXW7 and fetuin-A was confirmed by coimmunoprecipitation assays in transfected 293T cells (Fig. 2C). Coexpression of FBXW7 also decreased the half-life of fetuin-A to <10 min (Fig. 2D). Furthermore, ectopic expression of ubiquitin and coexpression of FBXW7 increased fetuin-A ubiquitination in 293T cells (Fig. 2E), which was further confirmed by in vitro ubiquitination assays (Fig. 2F). Together, these results demonstrate that FBXW7 promotes the ubiquitination and degradation of fetuin-A.

Usually, FBXW7 targets substrates by recognizing a short, phosphothreonine-containing motif known as Cdc4 phosphodegron (CPD) (29,30). Interestingly, the sequence surrounding Ser305 and Ser309 in fetuin-A protein conforms to a highly conserved CPD found in other well-established FBXW7 substrates (Supplementary Fig. 3E), suggesting that this motif is required for degradation of fetuin-A. The CPD motif did not exist in the protein sequence of fetuin-B, consistent with FBXW7 failing to regulate fetuin B stability. This hypothesis was further supported by the results that substitution of Ser305, Ser309, or both residues with an alanine increased the abundance of the mutant proteins, and the three mutants were resistant to FBXW7-mediated protein degradation (Fig. 2G). In agreement, while coexpression of FBXW7 enhanced the turnover of WT fetuin-A in CHX-treated cells, these alanine mutants (S305A, S309A, and S305A/S309A) were stable and insensitive to FBXW7 (Supplementary Fig. 3F), confirming the essential role of an intact CPD in controlling the degradation of fetuin-A.

Previous studies have shown that substrates of FBXW7 are usually phosphorylated by several kinases, such as casein kinase II (CK2), glycogen synthase kinase (GSK)3β, and cyclin-dependent kinases (CDK)s (31–33). We therefore sought to determine which kinase is responsible for fetuin-A phosphorylation. As shown in the Supplementary Fig. 3G, protein abundance of fetuin-A was dramatically increased in 293T cells treated with silmitasertib (CK2 inhibitor), but not SB216763 (GSK3β inhibitor), CGP74514A (CDK1 inhibitor), or PNU112455A (CDK2/5 inhibitor). In contrast, overexpression of CK2 reduced fetuin-A protein levels (Fig. 2H). Additionally, knockdown of CK2 by small interfering RNA dramatically blocked the FBXW7-induced protein degradation (Supplementary Fig. 3H). Finally, for determination of whether S305 and S309 are phosphorylated by CK2, 293T cells were transfected with WT fetuin-A or S305A/S309A mutant, along with CK2. Upon probing the fetuin-A immunoprecipitates with an anti-phosphoserine/threonine antibody, we observed that CK2-mediated phosphorylation of fetuin-A was greatly diminished when the S305 and S309 residues were mutated to alanine (Fig. 2I). Thus, these data implicate CK2 as responsible for fetuin-A phosphorylation and indicate that FBXW7-mediated degradation of fetuin-A is dependent on S305 and S309.

Next, we determined the role of an FBXW7–fetuin-A axis in vivo. First, we analyzed the expression levels of FBXW7 in obese mice. Although mRNA expression was unchanged (Supplementary Fig. 4A–C), FBXW7 protein levels were lower in the livers of both obese mice (ob/ob and HFD) and humans than in livers of the corresponding normal control subjects (Fig. 3A–C). Moreover, in human subjects, hepatic FBXW7 protein levels were negatively correlated with fasting plasma glucose and hemoglobin A_1c (HbA_1c) levels (Supplementary Fig. 5A and B).

A key prediction, based on the reduction of FBXW7 in the livers of obese mice, is that FBXW7 deficiency might promote insulin resistance. For testing of this, liver-specific FBXW7 knockout mice (FBXW7 LKO) were created by injection of adenovirus-expressing Cre recombinase into FBXW7-floxed mice through tail vein. Hepatic FBXW7 expression was found to be largely abolished, while FBXW7 expression in skeletal muscles (SKMs) and white adipose tissues (WAT) remained unaffected (Fig. 3D and E). As a result, liver-specific depletion of FBXW7 increased hepatic and plasma fetuin-A concentrations without any changes in its mRNA expression (Fig. 3F–H). FBXW7 LKO mice displayed higher blood glucose and insulin concentrations than the corresponding WT mice (Fig. 3I and J). Glucose tolerance test (GTT) and insulin tolerance test (ITT) also revealed that a significantly impaired whole-body insulin sensitivity occurred in FBXW7 LKO mice (Fig. 3K and L). Consistently, liver triglyceride contents and expression levels of proinflammatory cytokines in WAT were markedly increased in FBXW7 LKO mice (Supplementary Fig. 6A–C), and plasma levels of alanine transaminase (ALT) and aspartate transaminase (AST) were also significantly increased (data not shown). Body weight, food intake, locomotor activity, CO₂ production and O₂ consumption, and fat and lean mass were comparable between two groups of mice (Supplementary Fig. 6D–I).

Furthermore, to address whether the effects of FBXW7 depletion on glucose homeostasis are mainly mediated by increased fetuin-A levels, we injected the fetuin-A shRNA adenovirus into FBXW7 LKO mice to knock down fetuin-A expression (Fig. 3M). Our data showed that inactivation of endogenous fetuin-A expression reversed the FBXW7 deficiency–induced hyperglycemia, glucose intolerance, insulin resistance, and hepatic triglyceride accumulation (Fig. 3N–P and Supplementary Fig. 6J), suggesting that FBXW7 improves glucose metabolism by promoting fetuin-A degradation.

The above results suggest a protective role of FBXW7 in the regulation of glucose homeostasis. We thus investigated what effect upregulation of FBXW7 has in livers of obese mice using different experimental approaches. Firstly, we used AAV containing FBXW7 or GFP to overexpress FBXW7 in the livers of C57BL/6 mice. Those mice were then fed with an ND or HFD for 12 weeks (Supplementary Fig. 7A and B and Fig. 4A). Blood glucose and insulin levels were increased by HFD in GFP-treated mice compared with those in mice fed an ND (Fig. 4B and C). By contrast, there was no dramatic impairment in HFD mice expressing FBXW7 over the entire period (Fig. 4B and C). Results of GTT and ITT from the end of the 12-week period revealed a major improvement of insulin resistance in FBXW7-expressing mice (Fig. 4D and E). Additionally, hepatic triglyceride contents and plasma ALT and AST levels were significantly reduced (Supplementary Fig. 7C–E).

We next investigated whether fetuin-A levels were affected when FBXW7 was increased. As shown in Fig. 4F and G, plasma and hepatic concentrations of fetuin-A were higher with HFD treatment in GFP-injected mice compared with those in mice fed an ND. However, AAV-mediated expression of FBXW7 led to a significant reduction in fetuin-A protein contents (Fig. 4F and G), without any changes in the corresponding mRNA levels (Fig. 4H). Collectively, these observations suggested a potential role of FBXW7 to prevent the development of HFD-induced insulin resistance.

Additionally, for exploration of the effect of FBXW7 in a more severe model of obesity and diabetes, ob/ob mice were administered an adenoviral FBXW7 or GFP via tail vein injection. As shown in Fig. 5A, FBXW7 was uniquely overexpressed in the livers but not in SKMs or WATs (Supplementary Fig. 8A and B). Forced expression of FBXW7 reduced fetuin-A protein levels in the liver and plasma, while mRNA levels remained unaffected (Fig. 5B–D). Although body weight and food intake were not altered by FBXW7 overexpression (Supplementary Fig. 8C and D), blood glucose and plasma insulin levels were decreased (Fig. 5E and F). GTT and ITT further revealed that whole-body insulin sensitivity was significantly enhanced in ob/ob mice expressing FBXW7 (Fig. 5G and H). In agreement, insulin signaling in the liver and skeletal muscle was improved in the group overexpressing FBXW7 compared with that in GFP controls (Fig. 5I and Supplementary Fig. 8E). Finally, ALT and AST levels were reduced in the FBXW7 group (Supplementary Fig. 8F and G), confirming a beneficial role of FBXW7 in the obese livers.

We next examined the effects of FBXW7 overexpression in mice with HFD-induced obesity. Consistent with the results observed in ob/ob mice, mice with HFD-induced obesity administered with adenoviral FBXW7 displayed reduced fetuin-A levels along with lower blood glucose and insulin levels compared with the GFP group (Fig. 5J–O). GTTs and ITTs showed that whole-body insulin sensitivity was also improved in the FBXW7 group (Fig. 5P and Q). In addition, expression levels of PEPCK, a key gluconeogenic enzyme, were downregulated in ob/ob and HFD mice overexpressing FBXW7 (Supplementary Fig. 8H). Therefore, we conclude that FBXW7 overexpression can reduce fetuin-A protein levels in obese mice, resulting in an improvement in glucose homeostasis.

With our results taken together, we have identified a pivotal role of FBXW7 in the regulation of fetuin-A expression and glucose metabolism in mice, on the basis of which we further explored the association of FBXW7 with blood glucose levels in human. Interestingly, we identified a p.Ala204Thr (c.610G>A [rs189772026]) variant in exon 3 of the FBXW7 gene (Supplementary Fig. 9A–C), which led to missense mutation (GA heterozygotes) in 14 of 3,711 (0.37%) subjects from the FADE and SHDS cohorts (Supplementary Table 1). Of note, subjects with the A204T variant of FBXW7 had higher fasting, 2-h plasma glucose, and 2-h insulin levels compared with WT controls (Fig. 6A–C) (all P < 0.05). Linear regression models also found that rs189772026 was associated with 2-h insulin levels after adjustment for age and sex (P = 0.046). In addition, plasma fetuin-A concentrations of A204T carriers were higher than in WT controls after age, sex, and BMI adjustment (Fig. 6D).

In vitro and in vivo functional studies were further performed to explore the biological function of FBXW7 variant. The A204T site is highly conserved among different species (Fig. 6E) and predicted to be “possibly damaged” by PolyPhen-2. We next cloned this variant and found that it had a minimal effect on the abundance and half-life of fetuin-A (Fig. 6F and G) and failed to promote fetuin-A ubiquitination (Fig. 6H) compared with WT FBXW7. These findings suggest that it is a functional variation. If this notion is correct, the A204T variant should be unable to reduce blood glucose levels and ameliorate insulin resistance in obese mice. For testing of this possibility, ob/ob mice were administered adenoviruses expressing GFP, WT FBXW7, or the A204T variant via tail vein (Fig. 6I). Although WT FBXW7 expression significantly reduced blood glucose and insulin concentrations, improved insulin resistance and glucose intolerance, and inhibited TG accumulation and lipogenic genes expression in ob/ob mice, this effect was dramatically impaired in A204T-overexpressing livers (Fig. 6J–M and Supplementary Fig. 9D–F). In agreement, hepatic and plasma fetuin-A concentrations were significantly reduced in WT-expressing mice but not in the A204T-injected group (Fig. 6N and P). Taken together, these results implicated the contribution of the FBXW7 variant to elevated fetuin-A and blood glucose levels in humans.

It has been well established that elevated fetuin-A levels lead to insulin resistance and T2DM (7,34). However, despite compelling animal and human data, the mechanism by which fetuin-A levels are increased in obesity has not been fully explored. In the current study, we identified FBXW7 as a novel regulator that controls fetuin-A expression in obese mice and humans. This is supported by multiple lines of evidence. First, we observed increased fetuin-A protein expression in obese livers, while its mRNA levels are not altered. Second, FBXW7 interacts with and promotes the ubiquitin-mediated degradation of fetuin-A. Third, we found that two residues in the CPD motif of fetuin-A, S305 and S309, are critical for its stability and degradation. Fourthly, loss- or gain-of-function studies indicate that FBXW7 affects whole-body glucose metabolism through the regulation of fetuin-A levels. Finally, a functional missense sequence variation (A204T) in the human FBXW7 gene is associated with increased blood glucose and fetuin-A levels in a Chinese population. Therefore, our study provides an important and novel mechanism for high circulating fetuin-A levels in obese individuals (Fig. 7). However, mechanisms underlying the degradation of fetuin-A by FBXW7 remain unclear. It has been reported that Fbxw7β, one of the FBXW7 isoforms, resides in the ER membrane (35), in which cytokines are synthesized and then transported to the Golgi complex or cell surface (36). Therefore, further studies are still needed to investigate whether fetuin-A is degraded or secreted during trafficking pathways in hepatocytes.

Previous studies have identified a myriad of FBXW7 substrates, such as c-myc, cyclin E, SRC-3, P100, and MCL1 (37–41), which are important for cell proliferation, differentiation, and apoptosis. Therefore, much attention has been focused on the relationship between FBXW7 and tumorigenesis. Indeed, mutation or reduced expression of FBXW7 is frequently observed in human neoplasms, including breast cancers, colon cancers, and T-cell acute lymphocytic leukemia (42–44). However, its physiological and pathological role in glucose metabolism is poorly understood. Although PGC-1α and mTOR, two important regulators of glucose homeostasis, have been identified as substrates of FBXW7 in primary neurons and tumor cells (45,46), their expression profiles were not significantly altered in our study (Supplementary Fig. 10A and B), suggesting that the function and substrates of FBXW7 might be cell or tissue specific. Furthermore, FBXW7 was shown to regulate the stability and function of all members of the SREBP family of proteins, which control triglyceride synthesis (47). As a result, long-term deletion of hepatic FBXW7 in mice using the Cre-loxP system led to hepatomegaly and steatohepatitis, with massive depositions of triglycerides (48). In agreement, hepatic triglyceride contents were increased in our FBXW7 LKO mice, which was partially attenuated by knockdown of endogenous fetuin-A expression. Thus, in addition to controlling glucose metabolism, we speculate that the FBXW7/fetuin-A regulatory axis is also involved in hepatic triglyceride homeostasis.

One question that remains to be answered is that why hepatic FBXW7 expression is decreased in obesity. Here, we found that FBXW7 protein levels were markedly reduced in obese livers, while its mRNA expression was affected, indicating that the downregulation of FBXW7 may also be attribute to a posttranscriptional regulatory mechanism. Earlier studies have shown that FBXW7 could be targeted by multiple microRNAs in tumors (49). Interestingly, some of these microRNAs are upregulated in the livers of HFD or ob/ob mice (Supplementary Fig. 11). Therefore, we speculate that dysregulation of those microRNAs might lead to the downregulation of FBXW7 in obese livers.

Zhao et al. (50) demonstrated that hepatic ablation of FBXW7 by FBXW7-floxed crossed with albumin-Cre mice can stabilize REV-ERBα and disrupt circadian transcriptome, leading to glucose metabolism opposite to the ones reported in our study. Although the reason behind these conflicting data are not clear, one possible explanation is that chronic and long-term depletion of FBXW7 using albumin-Cre mice may elicit a compensatory response, leading to the outcome observed in the results of Zhao et al. (50). Besides, the results we present here are based on several different experimental approaches including mouse and human studies, and all of the data keep consistently showing that hepatic FBXW7 is beneficial for preservation of glucose homeostasis. In addition, the protein levels of REV-ERBα were not changed when comparing WT control and liver-specific FBXW7-overexpressing or knockout mice (Supplementary Fig. 12A and B).

In summary, our data demonstrate that FBXW7 is a key regulator of fetuin-A expression in the liver. Reduction of FBXW7 leads to elevated hepatic and blood fetuin-A protein levels, promoting systemic insulin resistance and hyperglycemia in obesity (Fig. 7). Our findings provide clinically relevant evidence that a functional FBXW7 variant (A204T) is associated with T2DM in humans and that FBXW7 expression is markedly reduced in obese livers. This study has clear implications for understanding the pathobiology of T2DM and supports that strategies to activate or upregulate FBXW7 in liver might be a potential therapeutic approach for improving metabolic disorders.

Funding. This study is supported by grants from Natural Science Foundation of China (nos. 31530033, 81322010, and 81570770), Science and Technology Commission of Shanghai Municipality (16JC1404100), and Shanghai Municipal Education Commission–Gaofeng Clinical Medicine Grant Support (20152527).

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

Author Contributions. J.Z., X.X., and Y. Li performed experiments. X. Liu and Y.J. analyzed the metabolic phenotypes of mice. T.W., Ho.Z., and C.H. performed analysis and interpretation of human data. J.J., Hu.Z., Q.T., X.G., and Xu. Li provided technical assistance and reagents. J.Z., Y. Lu, and Xi. Li wrote the manuscript. Y. Lu, B.L., and Xi. Li designed the project and coordinated the execution of the experimental plan. Xi. Li 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.