DDB1-Mediated CRY1 Degradation Promotes FOXO1-Driven Gluconeogenesis in Liver

Maintaining steady levels of glucose is crucial for the survival and health of mammals during feeding and fasting cycles. During fasting, the liver serves as the major site to keep blood glucose steady by activating de novo gluconeogenesis. During feeding, insulin reduces the glucose surge by promoting glycogen synthesis while inhibiting gluconeogenesis in the liver (1–3). The hepatic insulin signaling cascade has been shown to be indispensable in maintaining glucose homeostasis in animal models. Mice lacking insulin receptors in hepatocytes quickly develop hyperglycemia and hyperinsulinemia independently of obesity (4). In patients with diabetes, severe hepatic insulin resistance has been considered to be a major driver in excessive hepatic glucose production and fasting-associated hyperglycemia (2). Therefore, the ability of insulin to suppress hepatic glucose production has been targeted to treat diabetes.

Previous studies have established FOXO1 as a key transcription factor in integrating insulin signaling and hepatic glucose metabolism (5,6). Under normal physiological conditions, insulin activates the phosphatidylinositol 3-kinase–AKT pathway, which in turn phosphorylates the transcription factor FOXO1 at T24/S253/S316 (7). Subsequently, phosphorylated FOXO1 is excluded from the nucleus, leading to reduced mRNA levels of key gluconeogenic enzymes such as G6pase (glucose-6-phosphatase) and Pepck (phosphoenolpyruvate carboxykinase) (8–11). However, in the case of insulin resistance, the nuclear accumulation of FOXO1 promotes transcription of gluconeogenic enzymes and eventually results in hyperglycemia (12,13). FOXO1 has been found to be targeted for proteasomal degradation via ubiquitination downstream of insulin signaling (14–16). Several E3 ligases including MDM2, SKP2, and COP1 promote FOXO1 ubiquitination and degradation at least in vitro (17–19). Since inactivation of FOXO1 transcriptional activity is the major route for insulin to inhibit gluconeogenesis, inhibition of FOXO1-mediated gluconeogenesis by targeting its degradation could offer a promising avenue to treat diabetes without activating the insulin-AKT–dependent lipogenic pathway.

DDB1 (DNA-damage–binding protein 1) is a scaffolding component of the DDB1-CUL4A ubiquitin E3 ligase complex (20–22). Within this complex, DDB1 serves as the linker protein between CUL4A and substrate-binding proteins (21,23). Ddb1 deletion disrupts the DDB1-CUL4A complex and subsequently abolishes its E3 ligase actions (21,24,25). DDB1-CUL4A E3 ligase promotes ubiquitination and degradation of a variety of substrates, including p27, c-JUN, and CDT1 (23,26,27). We recently discovered that DDB1-CUL4A E3 ligase targets CRY1 for ubiquitination and degradation via DDB1 and CUL4-associated factors (DCAF) protein CDT2 (28). Ddb1 deletion increases total levels of CRY1 protein in mouse hepatocytes and mouse liver. Thus far, the role of DDB1 in liver metabolism is largely unknown, since Ddb1 global knockout is embryonically lethal (24). Given the emerging role of CRY1 in liver metabolism (29,30), it is likely that DDB1 could function as a metabolic regulator at least by manipulating the CRY1 protein stability.

CRY1 is an evolutionarily conserved clock protein with diverse functions (31). Besides its classic function as a negative regulator of the circadian network, CRY1 has been shown to be a novel regulator of glucose metabolism. Ectopic overexpression of CRY1 in liver reverses hyperglycemia in db/db mice, a type 2 diabetes model, by interfering with CREB-dependent glucagon signaling (30). CRY1 was also found to modulate glucocorticoid receptor function during the induction of gluconeogenesis (29). Recently, SREBP-1c–induced Cry1 has been shown to regulate hepatic glucose production (32). All these studies have highlighted the critical role of CRY1 in hepatic glucose metabolism under both normal and pathological conditions.

In this study, we examined the role of DDB1 E3 ligase in hepatic glucose metabolism. We established for the first time that DDB1 E3 ligase is a novel positive regulator of hepatic gluconeogenesis. Hepatocyte deficiency of Ddb1 not only suppresses hepatic gluconeogenesis during fasting but also protects mice from high-fat diet (HFD)-induced fasting hyperglycemia. At the mechanistic level, DDB1 increases FOXO1 protein via CRY1 degradation and promotes the FOXO1-driven gluconeogenesis in hepatocytes. In Cry1-depleted hepatocytes, insulin fails to reduce nuclear FOXO1 abundance and repress gluconeogenic gene expression. Therefore, our study discovered a novel pathway to regulate the FOXO1-driven gluconeogenesis via DDB1-mediated CRY1 degradation.

Animal experiments were conducted in accordance with the guidelines of the institutional Animal Care and Use Committee of University of Michigan Medical School. Male C57BL/6J mice and albumin-Cre mice were purchased from The Jackson Laboratory. The Ddb1^flox/flox mice were backcrossed to the C57BL/6J background for at least nine generations. The liver-specific Ddb1 knockout (Ddb1-LKO) mice were generated by crossing Ddb1^flox/flox mice with albumin-Cre mice. All mice were housed on a 12-h:12-h light:dark cycle at 25°C with free access to water and regular chow or HFD (45% kcal from fat; Research Diets). For fasting experiments, mice fasted for 16 h from 6:00 p.m. to 10:00 a.m.

Levels of blood glucose were measured with a Contour glucometer (Bayer). Aliquots of serum were analyzed for insulin levels by an insulin ELISA kit (R&D Systems). Pyruvate tolerance tests (PTTs) were performed in fasting animals at the indicated time points after intraperitoneal injection of sodium pyruvate in saline at 2 g/kg.

The RNA isolation and quantitative RT-PCR (RT-qPCR) analysis were performed as previously described (28). The primer sequences used in qPCR analysis are listed in Supplementary Data. All gene expression experiments were performed in at least two independent experiments in biological triplicates.

For preparation of cytosolic and nuclear proteins, liver tissues or cell pellets were homogenized in hypotonic buffer, incubated on ice for 15–20 min, and centrifuged at 3,000 rpm for 10 min at 4°C. The supernatant was saved as cytosolic fraction. The pellet was washed once with hypotonic buffer and resuspended in radioimmunoprecipitation assay buffer prior to sonication for 5 s. The nuclear protein was then collected after centrifugation at 13,000 rpm × 10 min. Immunoprecipitation protocol to detect protein-protein interaction has previously been described (28). FLAG-M2 beads or streptavidin beads were added into the lysate to capture immunocomplex containing FLAG-CRY1 or CBP-CRY1. The protocol for detecting protein ubiquitination has previously been described (33). Anti-FOXO1 was used to pull down FOXO1-ubiquitin conjugates.

Proximity ligation assay (PLA) was performed using Duolink In Situ PLA reagent according to the instructions from Sigma-Aldrich. After overnight transduction with adenovirus expressing Foxo1 (Ad-Foxo1) and Ad-Cry1 in chamber slides, Cry1^−/−/Cry2^−/− mouse embryonic fibroblast (MEF) cells were fixed in 4% paraformaldehyde at 4°C for 10 min. After quenching with 0.1 mol/L glycine and permeabilization with 0.25% Triton-X 100, cells were blocked in 10% BSA and then incubated in 1:250 diluted primary antibody solutions (anti-FOXO1 H-128 [SC-11350] anti-CRY1 W-L5 [SC-101006]; Santa Cruz Biotechnology). Afterward, the slide was incubated in PLA mixture of Duolink In Situ Probe anti-Rabbit PLUS and Duolink In Situ Probe anti-Mouse MINUS in a preheated humidity chamber at 37°C for 1 h. After ligation and amplification at 37°C with Duolink In Situ Detection Reagents Far Red, the slide was mounted with coverslips using Duolink In Situ Mounting Medium with DAPI. Imaging was performed on a Zeiss Axio Imager M2 microscope with filters set at λEx = 360 nm/λEm = 460 nm for DAPI and at λEx = 647 nm/λEm = 669 nm for far red, where Ex is excitation and Em is emission. Nine individual fields under ×63 oil objective were captured for each condition and quantified using the ImageJ software.

Both 293T and Hepa1c1c-7 cells were purchased from American Type Culture Collection and maintained according to the instructions. Transient transfection in 293T or Hepa1 cells was performed using polyethylenimine (28). Primary mouse hepatocyte isolation has previously been described (28). Wild-type (WT) MEFs and Cry1/2 double knockout MEF were generously provided by John Hogenesch (University of Cincinnati). Adenoviral transductions in cells were performed using 1.0 × 10⁸ plaque-forming units per well of six-well plates for 16 h. For FOXO1 transcriptional activity assay, 293T cells in a 24-well plate were transfected with the G6Pase-luc construct (contains 1.2 kb promoter sequence upstream transcription starting site) alongside various combinations of expression vectors. Thirty-six hours posttransfection, cells were lysed for luciferase activity measurement on a Bio Tek Synergy 2 microplate reader. A β-galactosidase construct was cotransfected in each well for normalization of luciferase activity.

Adenoviruses including Ad–short hairpin (sh)LacZ, Ad-shDdb1, Ad-Flag-Cry1-WT and Ad-Flag-Cry1–585KA, and Ad-Ddb1 have previously been described (28). Ad-FOXO1-ADA was generously provided by Henry Dong at the University of Pittsburg. Ad-Myc-Foxo1-WT was generated using previously described Gateway technology (28). Adeno-associated virus (AAV)-thyroxin binding globulin (TBG)-CRE and AAV-TBG-GFP were purchased from the University of Pennsylvania Vector Core. For adenoviral injections, 1 × 10¹² plaque-forming units per adenovirus were administrated via tail vein injection. For each virus, a group of four to five mice were injected with the same dose of viral particles. Ten to fourteen days after injection, mice were sacrificed at Zeitgeber time (ZT) 8 after overnight fasting and liver tissues were harvested for protein analysis.

Western blot analysis was performed using the following primary antibodies: anti-DDB1 (Abcam), anti-CRY1 (sc-101006), anti-GAPDH (sc-25778), anti-Lamin A/C (sc-20681), anti-PEPCK (sc-32879), anti-G6PASE (sc-25840), anti-FOXO1 (sc-11350), anti-CUL4A (sc-10782), anti-AKT1/2 (sc-1619), anti-CBP (sc-33000) (Santa Cruz Biotechnology), anti–phosphorylated (phospho)-AKT (S473) and anti–phospho-glycogen synthase kinase (GSK)3β (S9) (Cell Signaling Technology), and anti-ubiquitin and anti–β-tubulin (Sigma-Aldrich). Anti-p85 was a gift from Liangyou Rui at the University of Michigan.

All data are reported as mean ± SD. Differences between two groups were assessed by two-tailed Student t test. Difference between more than two groups was analyzed by ANOVA followed by Tukey post hoc testing. P < 0.05 was deemed statistically different.

Thus far, the metabolic actions of DDB1 in the liver have never been tested. We set out to investigate whether Ddb1 deficiency in hepatocytes could impact glucose metabolism in mice. We firstly confirmed Ddb1 deletion in the Ddb1-LKO mouse liver and primary mouse hepatocytes by measuring DDB1 protein expression (Fig. 1A and Supplementary Fig. 1A). The residual DDB1 signal in Ddb1-LKO mouse liver was likely to be from other cell types within the liver, since the Ddb1 mRNA is ubiquitously expressed (Supplementary Fig. 1B). In the same liver tissues, we observed increased levels of CRY1 protein (Fig. 1A), consistent with our previous report (28). Hepatocyte deletion of Ddb1 reduced blood glucose levels during fasting and refeeding without affecting body weight (Fig. 1B–E), suggesting that DDB1 protein could act as a physiological regulator of glucose metabolism. The liver maintains steady levels of blood glucose during fasting by enhancing hepatic gluconeogenesis (34). For assessment of the role of DDB1 in this process, we performed pyruvate tolerance tests. Compared with Ddb1^flox/flox mice, Ddb1-LKO mice showed impaired hepatic gluconeogenic response after gluconeogenic substrate pyruvate was injected (Fig. 1F). This reduction in gluconeogenic response is consistent with lowered protein and mRNA levels of PEPCK and G6Pase, two key enzymes in gluconeogenesis (Fig. 1G and H). These data, for the first time, demonstrated DDB1 as a new metabolic regulator of hepatic glucose metabolism.

During the development of insulin resistance and type 2 diabetes caused by an HFD, unchecked gluconeogenesis in the liver is a major contributor to fasting hyperglycemia (35,36). Given the inhibitory effects of Ddb1 deficiency on hepatic gluconeogenesis (Fig. 1F–H), we asked whether hepatic DDB1 level is affected by high-fat feeding. As shown in Fig. 2A, 12 weeks of HFD (45% calories from fat) markedly elevated the protein level of DDB1 in the mouse liver without affecting its binding partner CUL4A. This seems to be a liver-specific phenomenon, since DDB1 protein expression remained similar in adipose tissues from the same cohort of mice fed either regular chow or HFD (Fig. 2A). Moreover, induction of liver DDB1 protein by HFD possibly occurs through posttranslational modifications, since its mRNA levels are comparable between the regular chow and HFD groups (Supplementary Fig. 2A and B). Taken together, these studies suggest that the DDB1 protein abundance is sensitive to nutrient status in hepatocytes.

For testing of whether deletion of hepatic Ddb1 could impact gluconeogenesis in mouse liver after HFD feeding, both Ddb1^flox/flox and Ddb1-LKO mice were challenged with an HFD for 12 weeks. Compared with Ddb1^flox/flox mice fed an HFD for 6–12 weeks, Ddb1-LKO mice displayed similar body weight gain (Fig. 2B) but lower levels of fasting glucose (Fig. 2C). Furthermore, Ddb1-LKO mice showed suppressed hepatic glucose production in PTTs after HFD feeding after 4 weeks of HFD feeding (Fig. 2D). Consistent with reduced glucose production in the liver, the mRNA and protein levels of G6Pase and PEPCK were reduced in Ddb1-LKO mice (Fig. 2E and F). Meanwhile, the nuclear CRY1 protein was enhanced in Ddb1-LKO mice (Fig. 2F). Taken together, our data suggested that hepatic Ddb1 deficiency reduces blood glucose and suppresses hepatic gluconeogenesis during the course of an HFD. Of note, hepatic Ddb1 deficiency shows no impact on systemic glucose utilization (measured by glucose tolerance test) and insulin sensitivity (by insulin tolerance test) after chronic HFD (Supplementary Fig. 3).

To further test how hepatocyte DDB1 expression affects gluconeogenic response in adult mice, we used AAV-Cre to delete hepatic Ddb1 in 8-week Ddb1^flox/flox mice and then subjected them to HFD feeding for another 6 weeks (Fig. 3A). Acute Ddb1 deletion did not affect body weight (Fig. 3B) but severely blunted hepatic gluconeogenesis response upon PTT (Fig. 3C). Even after 6 h of fasting we observed that mice injected with AAV-Cre showed higher blood glucose (134 mg/dL) in comparison with mice with AAV-GFP (77 mg/dL). Moreover, we detected significant reduction in G6pase and Pepck mRNA and protein expression in the liver of HFD-treated AAV-Cre–injected Ddb1^flox/flox mice (Fig. 3D and E). These data suggest that hepatic DDB1 in adult mice plays a crucial role in elevating hepatic glucose production in the context of HFD challenge.

Hepatic gluconeogenesis is mainly driven by transcription activators such as FOXO1, CREB, GR, and HNF-4α (2,3). It is possible that Ddb1 deficiency could lead to impaired abilities of these transcription factors to promote gluconeogenesis. We focused on the potential role of FOXO1 because a recent study identified CRY1 as a suppressor of FOXO1 downstream of SREBP-1c (32). In a luciferase assay, Ddb1 deficiency repressed the induction of the G6pase promoter–driven luciferase activity by Foxo1, whereas DDB1 overexpression augmented the luciferase activity (Supplementary Fig. 4). Although DDB1 functions as a scaffolding protein in the DDB1-CUL4A-CDT2 E3 ligase complex, we observed a great reduction in nuclear FOXO1 protein in Ddb1-deleted liver or hepatocytes (Fig. 3F and G). Moreover, inhibition of proteasomes by MG132 blocked the FOXO1 reduction caused by Ddb1 depletion (Fig. 3H). These results suggest that DDB1 may control FOXO1 protein stability, and particularly nuclear FOXO1 abundance, to regulate gluconeogenic response.

Since DDB1 serves as a linker protein in the CUL4A ubiquitin E3 ligase complex, it is counterintuitive that Ddb1 deficiency reduces FOXO1 protein abundance in our study. We speculated that this regulation could be mediated through one of the DDB1 substrates. We reported that the DDB1-CUL4A E3 ligase targets CRY1 for ubiquitination-dependent degradation to regulate the molecular clock activity and CRY1 level is increased in the Ddb1-deficient mouse liver (28). CRY1 has also been shown to be a negative regulator of hepatic gluconeogenesis mediated by GR, CREB, and, most recently, FOXO1 (29,30,32). To test whether CRY1 protein might be a likely candidate to mediate FOXO1 degradation, we firstly compared the overall levels of the endogenous FOXO1 in WT versus Cry1^−/−/Cry2^−/− MEF cells. Indeed, total FOXO1 protein expression is inversely correlated to total CRY1 protein (Fig. 4A). Next, we compared the nuclear abundance of FOXO1 protein in WT versus Cry1^−/−/Cry2^−/− MEF cells, since it is the nuclear FOXO1 that drives the transcription of its target genes. The nuclear FOXO1 was also markedly increased in Cry1^−/−/Cry2^−/− MEF (Fig. 4B). Conversely, restoring CRY1 expression in Cry1^−/−/Cry2^−/− MEF was sufficient to reduce the nuclear FOXO1 abundance (Fig. 4B). It is known that nuclear FOXO1 abundance is influenced by either stability and/or localization. To distinguish these possibilities, we performed a similar experiment using the mutant FOXO1 (FOXO1-ADA) that is constitutively localized in the nucleus owing to the loss of AKT-dependent phosphorylation (7). Similar to FOXO1-WT, the levels of FOXO1-ADA were also increased in the Cry1^−/−/Cry2^−/− MEF (Fig. 4C). Taken together, our data suggest that CRY1 downregulates mainly the nuclear FOXO1 protein abundance.

To directly address whether CRY1 could be a downstream mediator linking DDB1 and FOXO1, we performed acute knockdown of Ddb1 in Ad-Foxo1–transduced WT and Cry1^−/−/Cry2^−/− MEF cells. In agreement with AAV-Cre–transduced Ddb1^flox/flox mouse liver and primary mouse hepatocytes (PMHs) (Fig. 3F and G), Ddb1 depletion reduced the nuclear FOXO1 protein in WT MEF but not in Cry1^−/−/Cry2^−/− MEF (Fig. 4D). Conversely, overexpression of DDB1 increased nuclear FOXO1 in WT MEF but not in Cry1^−/−/Cry2^−/− MEF (Fig. 4E). These data support that DDB1 promotes the nuclear FOXO1 abundance via degrading CRY1.

To further explore how CRY1 regulates FOXO1 protein turnover, we measured levels of poly-ubiquitinated FOXO1 protein in WT or Cry1^−/−/Cry2^−/− MEF cells by immunoblotting with anti-ubiquitin after immunoprecipitation (IP) with anti-FOXO1. Ubiquitination of FOXO1 was greatly reduced in Cry1^−/−/Cry2^−/− MEF (Fig. 5A). In contrast, Cry1 overexpression enhanced the formation of polyubiquitinated FOXO1 in WT MEF cells (Fig. 5B). These data suggest that CRY1 downregulates FOXO1 protein stability by promoting its ubiquitination and degradation.

How does CRY1 promote FOXO1 ubiquitination and degradation? It is possible that CRY1 could interact with FOXO1 and recruit its E3 ligase. Indeed, a strong interaction of CRY1 was detected with either FOXO1-WT or FOXO1-ADA in cotransfected 293T cells and transduced PMHs (Fig. 5C–E). The protein interaction between CRY1 and FOXO1 in the nucleus was further validated by a proximity ligation assay with anti-FOXO1 and anti-CRY1 in Cry1^−/−/Cry2^−/−MEF cells transduced with both Ad-Cry1 and Ad-Foxo1 (Fig. 5F). Furthermore, such interaction was found to require the CRY1-terminal region (300–614 aa) after domain mapping via a series of CRY1 truncation mutants (Supplementary Fig. 6).

Upon food intake, insulin suppresses FOXO1 transcription via insulin receptor (InsR)-AKT signaling (5,7). Whether CRY1 contributes to insulin-induced suppression of FOXO1 action has not been tested yet. We observed that CRY1 protein in primary hepatocytes was quickly induced within 1 h of treatment with insulin (Fig. 6A). Such regulation was also observed in the fasted mouse livers 2 h after insulin injection (Fig. 6B). In both cases, the Cry1 mRNA remained unchanged (Supplementary Fig. 7) within 2 h of insulin treatment, suggesting that insulin is likely to promote CRY1 protein stabilization independently of transcriptional activation during the first 2 h of insulin exposure.

Next, we asked whether CRY1 could affect insulin-induced nuclear FOXO1 protein degradation in mouse hepatocytes (15,16). Consistent with the literature (6,7), insulin treatment reduces the nuclear FOXO1 in Ad-shLacZ–transduced PMHs. However, its effect was abolished in the Ad-shCry1–transduced PMHs (Fig. 6C), suggesting that CRY1 is required for insulin-induced nuclear FOXO1 degradation. To further test the role of CRY1 in insulin suppression of gluconeogenesis, we compared the mRNA levels of G6Pase and Pepck in PMHs transduced with Ad-shCry1 versus Ad-shLacZ. As shown in Fig. 6D and E, insulin potently suppresses mRNA of G6Pase and Pepck in a dose-dependent manner in cells transduced with Ad-shLacZ but not in cells transduced with Ad-shCry1. Intriguingly, Pepck mRNA was significantly increased in cells transduced with Ad-shCry1 after insulin treatment. Thus, CRY1 is required for insulin-induced suppression of nuclear FOXO1 and gluconeogenic gene expression in hepatocytes.

To further evaluate the impact of Cry1 depletion on glucose metabolism in vivo, we injected WT mice with Ad-shCry1 versus Ad-shLacZ via tail vein. The CRY1 protein was measured in the mouse liver 14 days postinjection. CRY1 protein levels were reduced in Ad-shCry1–injected liver, whereas FOXO1 protein was elevated (Fig. 7A). Seven days after injection, we detected a significant increase in blood glucose level in Ad-shCry1–injected mice after overnight fasting compared with Ad-shLacZ–injected mice (Fig. 7B), a phenotype similar to that in Cry1 and Cry2 double knockout mice (29,37). To assess the gluconeogenic activity in the Cry1-depleted liver, we performed PTTs in both groups of mice at 7 days postinjection. Consistently, we observed a significant increase in blood glucose in response to pyruvate in mice injected with Ad-shCry1 (Fig. 7C). The serum insulin levels were slightly reduced in the shCry1 group without statistical significance (Fig. 7D). Meanwhile, the levels of G6Pase, Pepck, and Pgc-1α (coactivator for gluconeogenesis) were significantly induced (Fig. 7E). Taken together, our data support CRY1 as a negative regulator of hepatic gluconeogenesis in vivo.

Cry1 and Cry2 double knockout mice develop systemic insulin resistance when challenged with HFD (37). However, the tissue-specific role of CRY1 in insulin sensitivity remains unclear. To gain insights into the impact of chronic Cry1 depletion on liver insulin signaling, we examined the phosphorylation levels of downstream targets of insulin in the Ad-shCry1–injected liver after 2 weeks of HFD feeding. Consistent with elevated gluconeogenesis, the levels of AKT-P^S473 and GSK3β-P^S9 were reduced in the liver with acute Cry1 knockdown in HFD-fed mice (Fig. 7F), suggesting that chronic Cry1 depletion could lead to impairment of insulin-AKT signaling, which further promotes FOXO1 accumulation in the liver. To confirm that this indeed occurs in hepatocytes, we isolated PMH from mice injected with shCry1 for >7 days and found that these hepatocytes were also very resistant to insulin-stimulated AKT-P^S473 (Supplementary Fig. 8).

To gain insights into how chronic Cry1 deficiency may lead to insulin resistance in hepatocytes, we performed RT-qPCR analysis of known genes associated with insulin resistance either in mice injected with Ad-shLacZ or in mice after HFD feeding. In both liver and PMHs of Ad-shCry1–injected mice, the expression of the known negative regulator of insulin signaling Socs3 was significantly increased (Supplementary Fig. 9A and B) (38,39). Induction of Dbp mRNA was used as a marker for Cry1 depletion. Since CRY1 has been also implicated in inhibiting inflammation in various tissues (40,41) and chronic inflammation has been linked to insulin resistance (42,43), we checked the expression of several proinflammatory markers in the liver of Ad-shCry1–injected mice. As shown in Supplementary Fig. 9C, hepatic Cry1 depletion increased the expression of Tnfα and Mcp-1 in the liver of Ad-shCry1–injected mice. Taken together, our data showed that acute Cry1 deficiency results in hepatic insulin resistance by upregulating the pathways that impede insulin signaling and subsequently exacerbate the FOXO1-driven gluconeogenesis.

Uncontrolled hepatic glucose production is a hallmark of type 2 diabetes (2,35). Here, we uncovered a novel metabolic pathway in which DDB1 E3 ligase promotes FOXO1-driven hepatic gluconeogenesis by degrading CRY1. Furthermore, deletion of Ddb1 in hepatocytes protects mice from HFD-induced fasting hyperglycemia and elevated gluconeogenesis without affecting body weight. In addition, we demonstrated that insulin induces CRY1 protein and CRY1 depletion impairs insulin’s ability to suppress FOXO1-mediated gluconeogenesis. In conclusion, our study discovered DDB1-mediated CRY1 degradation as a novel pathway to regulate FOXO1 protein expression and gluconeogenesis in the liver (Fig. 7G). Given that a specific small-molecule activator of CRY1 has been shown to repress glucagon-induced gluconeogenesis in PMHs (44), suppression of DDB1 E3 ligase–mediated ubiquitination and degradation of CRY1 may offer another therapeutic avenue to control hyperglycemia in people with type 2 diabetes.

Targeted protein ubiquitination and degradation have been shown to regulate cell cycle, genomic stability, and DNA replication (45,46). How protein ubiquitination regulates metabolic events is far less understood. Emerging evidence suggests that E3 ligases could be important regulators of glucose metabolism (47). For example, the ubiquitin E3 ligase MG53 promotes IRS-1 ubiquitination and degradation and therefore contributes to obesity and metabolic syndrome upon chronic HFD (48,49). In contrast, the ubiquitin E3 ligase CBL-B promotes ubiquitination and degradation of TLR4 and therefore reduces macrophage activation and infiltration during obesity (50). Collectively, these findings suggest that E3 ligases could be effectively targeted to restore metabolic homeostasis. In our current study, we provide both in vivo and in vitro evidence suggesting an important role of DDB1-CUL4A E3 ligase in hepatic glucose metabolism. Ddb1 deficiency represses FOXO1-driven gluconeogenesis in hepatocytes and in turn reduces blood glucose during fasting. Interestingly, DDB1 protein is elevated in the mouse liver after HFD, raising the possibility that induction of DDB1 might be required to protect FOXO1 stability during insulin resistance. Since CRY1 has also been implicated to suppress the cAMP-CREB and glucocorticoid-mediated gluconeogenesis pathway (29,30), it remains to be tested whether DDB1 could also regulate these two pathways in addition to FOXO1 in both fasting and HFD conditions.

We provide the first evidence that insulin induces CRY1 protein expression independently of gene expression in both hepatocytes and liver. We propose two possible mechanisms that may account for such an acute action of insulin on CRY1 stability. The first possibility is that the insulin-AKT signaling pathway promotes phosphorylation of CRY1 to block its interaction with DDB1-CUL4A-CDT2 E3 ligase, therefore protecting CRY1 from degradation. However, sequence analysis by Scansite revealed that CRY1 contains no canonical phosphorylation motifs for AKT. The second possibility is that insulin can block the formation of DDB1-CUL4A-CDT2 E3 ligase. Two recent reports highlighted a role of FBOX11 E3 ligase in promoting CDT2 ubiquitination and degradation in response to TGF-β signaling (51,52), implying a possibility of signal-dependent CDT2 protein degradation. Sequence analysis suggests that CDT2 is a preferred substrate for a number of kinases including AKT, GSK3β, ATM, and DNA-dependent protein kinase. Thus, CDT2 could be a potential direct target of AKT to modulate DDB1-CUL4A E3 ligase activity. More detailed biochemical analysis will be needed to determine whether insulin inhibits DDB1-CUL4A-CDT2 E3 ligase assembly and activity via direct phosphorylation.

Unexpectedly, we observed that acute Cry1 deficiency leads to a marked reduction in AKT activation in both liver and primary mouse hepatocytes. This effect might be due to a combination of induction of both negative regulator of insulin signaling (SOCS3) and proinflammatory markers. How CRY1 is involved in suppressing these pathways remains to be addressed. These findings also suggest that CRY1 can inhibit FOXO1 signaling via two distinctive mechanisms in a time-dependent manner: on one hand, CRY1 acutely binds to nuclear FOXO to promote its ubiquitination-dependent degradation. On the other hand, chronic activation of CRY1 could enhance AKT signaling and therefore facilitates the translocation of FOXO1 to the nucleus to further inhibit gluconeogenic gene expression.

In conclusion, our study has identified an intricate mechanism of how DDB1 stabilizes FOXO1 through CRY1 degradation during fasting. Our study highlighted the critical role of DDB1 in promoting FOXO-1–dependent gluconeogenesis in the liver. Given the function of DDB1 in the CUL4A E3 ligase complex, it is conceivable that targeting DDB1 might offer novel therapeutics for the treatment of type 2 diabetes.

Acknowledgments. The authors thank Yong Cang (Zhejiang University, Hangzhou, China) and Stephen Goff (Columbia University) for sharing the Ddb1^flox/flox mice. The authors also thank John Hogenesch (University of Cincinnati) for providing the Cry1^−/−/Cry2^−/− MEF cells. The authors also thank Henry Dong (University of Pittsburg) for providing the Ad-Foxo1-ADA virus and Shaodong Guo (Texas A&M University Health Science Center) for the pcDNA-FOXO-3A expression vector.

Funding. This work was supported by funding from the National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health (K99/R00-DK-077449 and R01-DK-099593), to L.Y. Part of the work was also supported by pilot grants from Michigan Obesity to L.Y. (P30-DK-089503) and the Michigan Diabetes Research Training Center to X.T. (P60-DK-020572).

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

Author Contributions. X.T. and L.Y. designed and performed in vitro experiments. X.T. and L.Y. wrote and edited the manuscript. X.T., D.Z., N.C., and K.V. conducted in vivo experiments. E.J., K.S., and N.G. carried out RT-qPCR and the generation of adenoviral constructs and concentration of adenoviruses. J.S. analyzed and quantified the PLA imaging data. L.Y. supervised the work and analyzed and interpreted data. L.Y. 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.