Hepatic GSK3β-Dependent CRY1 Degradation Contributes to Diabetic Hyperglycemia

The liver plays important roles in glucose homeostasis by balancing glucose production and storage in the form of glycogen (1). During fasting, the liver maintains adequate levels of blood glucose by promoting hepatic glucose production (HGP) and glycogenolysis to meet metabolic demands in the peripheral tissues. In contrast, during the postprandial state, the liver suppresses HGP and converts excess glucose into glycogen (2). In type 2 diabetes, aberrant HGP regulation is one of the key contributors to hyperglycemia (1,3), which results from hepatic insulin resistance, which is characterized by the inability of insulin to suppress HGP (4).

HGP is regulated by several key enzymes, including phosphoenolpyruvate carboxykinase 1 (PEPCK) and glucose-6-phosphatase (G6Pase), which convert pyruvate to glucose (5). Several transcription factors and coactivators modulate the expression of gluconeogenic enzymes in response to hormones such as glucagon and insulin (1). Forkhead box O1 (FOXO1), a pivotal transcription factor for gluconeogenic gene expression, mediates the effect of glucagon on HGP during fasting. In contrast, in response to insulin, Akt-dependent phosphorylated (phospho)-FOXO1 is ubiquitinated and degraded via proteasomal degradation pathways during postprandial periods (6). In animal models of diabetes, the level of hepatic FOXO1 is elevated, accompanied by increased gluconeogenic gene expression (7). As the suppression of FOXO1 is a major route of insulin to inhibit HGP, it is likely that elevated FOXO1 leads to surplus HGP (8).

Cryptochrome 1 (CRY1) is an essential molecular clock component for the generation of circadian rhythms (9). Mammalian circadian clocks are tightly regulated by negative feedback loops (10). In addition to transcriptional regulation, posttranslational modifications of circadian clock proteins accompanied by altered levels of clock proteins affect circadian clock activities (11). Besides serving as circadian regulators, peripheral circadian clocks are highly associated with glucose metabolism for systemic energy homeostasis (12). Recent studies have shown that dysregulated levels of clock proteins are associated with type 2 diabetes (13). It has been reported that CRY1 represses HGP by inhibiting glucagon receptor and/or glucocorticoid receptor signaling pathways (14,15), indicating that CRY1 plays certain roles in glucose metabolism (16). In addition, we and others have demonstrated that hepatic CRY1 protein downregulates HGP by promoting MDM2-mediated nuclear FOXO1 degradation in response to feeding or insulin (17,18). Although decreased hepatic CRY1 protein is closely linked to hyperglycemia in diabetic animals (17,19), how hepatic CRY1 dysregulation induces excessive HGP in diabetes remains largely unknown.

In this study, we aimed to elucidate the mechanism(s) underlying hepatic CRY1 protein reduction in diabetic animals. We performed proteomic analysis to identify regulatory factor(s) for hepatic CRY1 protein degradation in diabetic animals. Using mass spectrometry and CRY1 mutagenesis, we investigated the signaling pathway(s) involved in hepatic CRY1 protein degradation. Moreover, the effect of enhanced CRY1 on HGP was investigated in a diabetic animal model established through high-fat, high-cholesterol, and high-fructose diet (HFCF) feeding. Collectively, our data suggest that enhanced GSK3β-dependent hepatic CRY1 degradation is crucial for promoting HGP in diabetic animals.

Male db/+ and db/db mice were obtained from Dae Han Bio Link (Seoul, Korea). CRY1 knockout (KO) mice were provided by Dr. Sancar Aziz from the University of North Carolina. All animals were maintained under a 12 h/12 h light/dark cycle in a pathogen-free animal facility. For establishing a mouse model of diabetes, 8-week-old mice were fed the HFCF diet, consisting of 40% of kilocalories from fat, 2% cholesterol, and 20% fructose (D09100310; Research Diets, New Brunswick, NJ), for the next 12–16 weeks. For intraperitoneal glucose tolerance test (GTT), mice were fasted for 16 h and glucose was administered (1 g/kg body wt). For pyruvate tolerance test (PTT), normal chow (NC)- or HFCF-fed mice were fasted for 18 h and then injected with pyruvate (1 g/kg body wt i.p.). GSK3β inhibitors were treated for 2 h before performance of PTT. HFCF-fed wild-type (WT) and CRY1 KO mice were intraperitoneally injected with vehicle (3% DMSO in PBS or PBS). Dissected tissue specimens were immediately stored at −80 °C until analysis. For in vivo FBXL3 and GSK3β knockdown, vehicle (PBS) or siRNA complex was injected (2.0 mg/kg i.v.) with use of invivofectamine 3.0 (IVF 3001; Invitrogen) according to the manufacturer’s protocol. After 3 days, mice were subjected to the PTT and analyzed. All animal experiments were approved by the Seoul National University Institutional Animal Care and Use Committee.

Protein sampling and proteomics analyses were performed as previously reported (20). Briefly, CRY1 peptides were generated by in-gel or in-solution digestion. For in-gel digestion, gel slices were destained in 50% acetonitrile solution, followed by in-gel alkylation of cysteine residues with dithiothreitol and iodoacetamide. Then, the samples were digested with sequencing-grade trypsin at a ratio of 1:50 (w/w) at 37°C overnight. For in-solution digestion, purified CRY1 samples were denatured in 8 mol/L urea and subjected to reduction and alkylation, followed by trypsinization after dilution to obtain 1 mol/L urea. Liquid chromatography–tandem mass spectrometry (LC-MS/MS) was performed with a Q Exactive mass spectrometer (Thermo Fisher Scientific) coupled with a nanoACQUITY Ultra Performance LC instrument (Waters) equipped with an in-house packed capillary trap column (150-μm i.d., 3 cm) and analytical column (75-μm i.d., 100 cm) with 3-μm Jupiter C18 particles (Phenomenex) at a flow rate of 300 nL/min. A linear gradient (100 min) was applied to each biological replicate. The data were analyzed in MaxQuant (version 1.5.3.30) with the Andromeda search engine at 10 ppm precursor ion mass tolerance against the Mus musculus proteome database at a protein false discovery rate of <1%. Extracted ion chromatograms were plotted with Qual Browser in the Xcalibur software (Thermo Scientific).

Raw sequence reads were trimmed and quality controlled with Trim Galore! (version 0.6.7) (21). Trimmed reads were mapped to the mouse reference genome GRCm39/mm39 with HISAT2 (version 2.2.1) (22). Mapped reads were assembled and quantified with StringTie (version 2.1.7) (23). Differential expression analysis was conducted with use of Limma (version 3.42.2) (24). Significance threshold was adjusted P value < 0.05 for CRY1 KO samples versus WT samples. Identified differentially expressed genes (DEGs), 198, consisted of 135 upregulated genes and 63 downregulated genes. Gene ontology and pathway enrichment analysis of the upregulated DEGs in CRY1 KO samples was conducted with Enrichr (25).

Whole transcription factor–target gene network was downloaded from the TRRUST (version 2) database (26). After filtering and preprocessing, the network contained 1,729 genes and 4,262 interactions. 198 DEGs were mapped on the whole transcription factor–target gene network, and network propagation (NP) was conducted using a random walk with restart algorithm to interpret individual gene-level perturbations at the network-level associations (27). NP is a graph-based analysis method that propagates information of a node to nearby nodes through the edges at each iteration for a fixed number of steps or until convergence, allowing estimation of gene interactions. We denoted the value of each node after NP as the NP score.

MG132 was purchased from Calbiochem (San Diego, CA). SB-415286 was purchased from Enzo Life Sciences (Farmingdale, NY). Antibodies against MYC (2276, 1:1,000 dilution; Cell Signaling Technology), HA (3724, 1:1,000; Cell Signaling Technology), V5 (3792, 1:1,000; Millipore), phospho-serine (9332, 1:500; Abcam), FOXO1 (2880, 1:1,000; Cell Signaling Technology), AKT (9272, 1:1,000; Cell Signaling Technology), phospho–AKT-S473 (9271, 1:1,000; Cell Signaling Technology), GSK3β (610201, 1:1,000; BD Biosciences), phospho–GSK3β-S9 (9336, 1:1,000; Cell Signaling Technology), FLAG (F3165, 1:1,000; Sigma-Aldrich, St. Louis, MO), actin (A5316, 1:2,000; Sigma-Aldrich), GAPDH (LF-PA0018, 1:1,000; LabFrontier Co.), PEPCK (sc-32879, 1:500; Santa Cruz Biotechnology), FBXL3 (123116, 1:500; Abcam), and CRY1 (104736, 1:1,000; Abcam [laboratory-made antibody obtained from Dr. Aziz Sancar (28), 1:500 dilution]) were used. MYC-CRY1, V5-GSK3β, 2SA CRY1-MYC, 2SD CRY1-MYC, and V5-CA-GSK3β cDNAs were cloned into the pcDNA3.1 vector, and Flag-FBXL3 cDNA was cloned into pCMV-3 Flag.

COS1 cells (CRL1650; ATCC) were transfected with plasmids encoding CRY1-MYC, FBXL3-Flag, and ubiquitin-HA in the presence of 20 μmol/L MG132 for 6 h. Total cell lysates were prepared with Tris-Glycine Novel (TGN) buffer (50 mmol/L Tris-HCl, pH 7.5; 150 mmol/L NaCl; 1% Tween-20; and 0.3% NP-40). CRY1-MYC was immunoprecipitated with an anti-MYC antibody (Cell Signaling Technology), and after washing in TGN buffer, the proteins were separated by SDS-PAGE followed by Western blotting with an anti-HA antibody.

Mouse primary hepatocytes were isolated from 10-week-old mice with the collagenase perfusion method (17) and seeded in medium 199 (M199) supplemented with 10% FBS. After 6 h of attachment, cells were transfected with siRNA or vectors in serum-free medium.

After washing in cold PBS, cells were treated with the TGN buffer supplemented with 0.1% protease inhibitor cocktail (P3100; GenDEPOT, Katy, TX). The lysates were incubated with primary antibodies at 4°C for 16 h and then in a 50% slurry of protein A sepharose presaturated with lysis buffer for 1 h. After three washes with lysis buffer, the immunoprecipitated proteins were recovered from the beads and analyzed with SDS-PAGE and immunoblotting.

RNA isolation and quantitative RT-PCR (qRT-PCR) analysis were performed as previously described (29). Total RNA was isolated with TRIzol reagent (Invitrogen). cDNA was synthesized with a reverse transcriptase kit (Toyobo, Osaka, Japan) according to the manufacturer’s instructions. The primers were generated at Bioneer (Daejeon, South Korea) and are listed in Supplementary Table 2. Relative mRNA levels were measured with a CFX real-time quantitative PCR detection system (Bio-Rad Laboratories).

siRNA duplexes for FBXL3 and GSK3β were purchased from Bioneer. Primary hepatocytes were transiently transfected with Lipofectamine 3000 reagent (Invitrogen, Carlsbad, CA) according to the manufacturer’s protocol.

Glucose production by mouse primary hepatocytes was measured with a glucose oxidase assay (St. Louis, MO) according to the manufacturer’s protocol. Briefly, cells were treated with forskolin (10 μmol/L) and incubated in Krebs-Ringer buffer (115 mmol/L NaCl, 5.9 mmol/L KCl, 1.2 mmol/L MgCl₂, 1.2 mmol/L NaH₂PO₄, 2.5 mmol/L CaCl₂, 25 mmol/L NaHCO₃, 10 mmol/L lactate, and 2 mmol/L pyruvate, pH 7.4) equilibrated with 5% CO₂ at 37°C for 6 h. The assays were performed at least in duplicate.

Data are means ± SD or SEM. All n values indicated in the figures refer to biological replicates. Two-tailed Student t tests were used to compare the means of two groups. One-way ANOVA followed by Tukey post hoc tests was used to compare means of more than two groups. Two-way ANOVA followed by Sidak multiple comparisons tests was used to compare two independent variables. Statistical analyses were performed with GraphPad Prism 7 software (GraphPad Software, La Jolla, CA).

The data sets generated or analyzed during the current study are available from Gene Expression Omnibus, accession no. GSE197839.

While hepatic CRY1 protein was downregulated in genetically diabetic db/db mice (17), it is unclear whether hepatic CRY1 expression is altered in diet-induced diabetic animal models. For generation of a diet-induced diabetes mouse model with elevated HGP, C57BL/6J mice were fed an HFCF diet (Supplementary Fig. 1A–E) (30). Similar to in db/db mice (Fig. 1A and Supplementary Fig. 1F), the level of hepatic CRY1 protein, but not mRNA, was decreased in HFCF-fed mice compared with NC-fed mice (Fig. 1B and Supplementary Fig. 1G), implying that hepatic CRY1 protein would be modulated at the posttranscriptional level. It is well established that various posttranslational modifications modulate CRY1 protein stability, and proteasomal degradation is a key determinant of CRY1 protein levels (9,11). Consistently, the proteasomal degradation blocker MG132 attenuated the degradation of CRY1 protein in hepatocytes (Supplementary Fig. 1H). To investigate which factor(s) are responsible for hepatic CRY1 protein degradation, we overexpressed myc-tagged CRY1 in the livers of db/db mice, pulled down CRY1-binding proteins, and performed LC-MS/MS (Supplementary Fig. 1I). The results indicated that FBXL3 and deltex E3 ubiquitin ligase 3L (DTX3L), which are E3 ligases, interacted with CRY1 (Fig. 1C and Supplementary Tables 1 and 4). Although FBXL3 reportedly is involved in CRY1 degradation (31), whether it could affect hepatic CRY1 protein stability has not been studied. As shown in Fig. 1D, FBXL3 potently increased the extent of CRY1 polyubiquitination, whereas DTX3L was less active in inducing CRY1 protein polyubiquitination. For affirmation of the physical interaction between FBXL3 and CRY1, a coimmunoprecipitation assay was performed with liver extract. Endogenous FBXL3 formed a protein complex with CRY1 (Supplementary Fig. 2A), and hepatic CRY1 associated with endogenous Skp, Cullin, F-box containing (SCF) complex (SKP1-CUL1-FBXL3) (Supplementary Fig. 2B). When we evaluated FBXL3 expression in the liver, both protein and mRNA levels of FBXL3 were increased in db/db mice (Fig. 1E and F) and HFCF mice (Fig. 1G and H). To further examine the effect of FBXL3 on hepatic CRY1 protein stability, a cycloheximide-chase experiment was performed in mouse primary hepatocytes. FBXL3 suppression via siRNA mitigated CRY1 protein degradation (Fig. 1I) and attenuated polyubiquitination of CRY1 protein in hepatocytes (Fig. 1J). In contrast, FBXL3 overexpression stimulated polyubiquitination and degradation of CRY1 protein (Supplementary Fig. 2C and D). These data propose that elevated FBXL3 expression promotes hepatic CRY1 protein degradation in diabetic animals.

CRY1 acts as a negative regulator of HGP, whose dysregulation leads to excessive HGP (17,18). To identify key factor(s) involved in CRY1-mediated HGP regulation, we conducted unbiased bioinformatics analyses using bulk RNA-sequencing (RNA-seq) data of the liver from WT and CRY1 KO mice. As shown in Supplementary Fig. 3, in silico analysis revealed that FOXO1 was one of key transcription factors that could affect gene expression profile in the liver of CRY1 KO mice. For testing of whether FBXL3 modulates HGP through CRY1 degradation, FBXL3 was ectopically expressed in primary hepatocytes. In WT hepatocytes, FBXL3 overexpression boosted HGP (Fig. 2A). In addition, ectopic expression of FBXL3 markedly increased FOXO1 protein levels and gluconeogenic gene expression, concomitant with decreased CRY1 protein levels (Fig. 2B and C). On the contrary, in CRY1 KO hepatocytes, FBXL3 overexpression did not affect HGP or gluconeogenic gene expression (Fig. 2A–C). Further, we studied whether FBXL3 suppression would downregulate HGP via an increase in CRY1. When FBXL3 was suppressed via siRNA, knockdown of FBXL3 repressed HGP and gluconeogenic gene expression in WT hepatocytes (Fig. 2D and E and Supplementary Fig. 4A). Simultaneously, the level of FOXO1 protein decreased (Fig. 2F). However, these effects of FBXL3 suppression were nullified in CRY1 KO hepatocytes (Fig. 2D–F). These findings led us to investigate the in vivo effect of FBXL3 on HGP. To address this, FBXL3 was suppressed via siRNA in the liver of WT mice through tail vein injection. siFbxl3 treatment did not affect body weight or expression of FBXL3 mRNA in other tissues except the liver (Supplementary Fig. 4B and C). Consistent with above data, PTT showed that in vivo FBXL3 suppression significantly decreased the level of blood glucose following pyruvate challenge (Fig. 2G). Moreover, hepatic FBXL3 suppression upregulated the level of CRY1 protein, resulting in suppression of FOXO1 protein and gluconeogenic genes in the liver (Fig. 2H and I and Supplementary Fig. 4D). To further investigate the role(s) of FBXL3-CRY1 axis in diabetic animals, we explored the effect of FBXL3 suppression on HGP in db/db mice. As shown in Fig. 2J, hepatic FBXL3 suppression decreased blood glucose level during PTT in db/db mice. Additionally, hepatic FBXL3 suppression increased the level of CRY1 protein in diabetic animals (Fig. 2K). Together, these data suggest that hepatic FBXL3 suppression would ameliorate hyperglycemia by promoting CRY1-mediated FOXO1 degradation in diabetic animals.

Given that FBXL3 promotes the degradation of its substrates in a phosphorylation-dependent manner (32), we questioned whether FBXL3-dependent CRY1 degradation might be linked to its phosphorylation. For answering this question, the level of hepatic CRY1 phosphorylation was examined in db/db mice. As indicated in Fig. 3A, the level of CRY1 phosphorylation was enhanced in the livers of db/db mice. To determine which kinase(s) are responsible for hepatic CRY1 phosphorylation, we performed in silico analyses (33,34) and found that there were at least five potential kinases, including AMPK, MAPK, GSK3β, DNA-PK, and CK-1 (Supplementary Fig. 5A). For elucidation of the key kinase(s) responsible for modulating CRY1 protein stability, primary hepatocytes were treated with an inhibitor of each kinase (i.e., SB415286, GSK3β; NU-7441, DNA-PK; IC261, CK-1; SB203580, p38 MAPK). As shown in Fig. 3B and C, the level of hepatic CRY1 protein was increased by the GSK3β inhibitor SB415286 but not by the other kinase inhibitors. Consistent with the previous findings that the enzymatic activity of GSK3β is augmented in diabetic animals (35,36), the level of S9 phosphorylation of GSK3β, a marker of inhibitory GSK3β activity, was decreased in db/db and HFCF-fed mice (Fig. 3D and Supplementary Fig. 5B). To evaluate whether GSK3β might be involved in CRY1 phosphorylation, we examined the level of CRY1 phosphorylation using GSK3β inhibitors. The GSK3β inhibitors LiCl and SB415286 decreased CRY1 phosphorylation (Supplementary Fig. 5C and D). Accordingly, the level of CRY1 phosphorylation was increased on GSK3β overexpression, whereas these effects were diminished after GSK3β inhibitor treatment (Fig. 3E). For investigation of whether GSK3β might promote CRY1 protein degradation, a cycloheximide-chase assay was performed. In hepatocytes, CRY protein degradation was suppressed by GSK3β inhibitors (Fig. 3F and Supplementary Fig. 5E). To study whether CRY1 phosphorylation by GSK3β might facilitate FBXL3-mediated ubiquitination, we performed ubiquitination assays using GSK3β inhibitors. As shown in Fig. 3G and Supplementary Fig. 5F, FBXL3 accelerated CRY1 polyubiquitination, whereas GSK3β inhibitors abolished such effect. Therefore, these data suggest that GSK3β would phosphorylate hepatic CRY1 protein and promote its FBXL3-dependent degradation in diabetic animals.

To identify potential phosphorylation residue(s) in the CRY1 protein by GSK3β, we performed mass spectrometry (Supplementary Fig. 6A). Compared with control (Fig. 4A), overexpression of constitutively active GSK3β (CA-GSK3β; serine 9 to alanine mutation) induced CRY1 phosphorylation at S280 (Fig. 4B). In silico analysis revealed that S281 in CRY1 might be another potential phosphorylation residue corresponding to the GSK3β consensus phosphorylation site (S-XXX-S) (Supplementary Fig. 6B) (37). As both S280 and S281 CRY1 are well conserved in various species (Supplementary Fig. 6C), we examined whether these residues might be important for CRY1 degradation by FBXL3. To address this, we mutated both S280 and S281 to alanine (phospho-dead mutation; 2SA mutant CRY1) or aspartate (phospho-mimicking mutation; 2SD mutant CRY1) and examined CRY1 protein stability. The degradation rate of the 2SA mutant CRY1 protein was significantly decreased compared with that of WT CRY1 protein (Fig. 4C), whereas that of the 2SD mutant CRY1 protein was increased (Fig. 4D). For further affirmation that these serine residues are crucial for FBXL3-dependent polyubiquitination, a cell-based ubiquitination assay was performed. As shown in Fig. 4E, FBXL3-dependent CRY1 polyubiquitination was downregulated in 2SA CRY1, whereas it was promoted in 2SD CRY1 (Fig. 4F). Collectively, these data suggest that CRY1 phosphorylation at S280/S281 would be mediated by GSK3β, leading to FBXL3-dependent polyubiquitination and proteasomal degradation of hepatic CRY1 protein.

For investigation of whether GSK3β-induced CRY1 phosphorylation and degradation might modulate HGP, the effects of GSK3β suppression on HGP were examined with use of RNA interference and LiCl. In WT hepatocytes, GSK3β suppression downregulated glucose production (Fig. 5A and Supplementary Fig. 7A). In accordance with this, the levels of Pck1 and G6pc mRNA and FOXO1 protein were decreased by GSK3β suppression and LiCl treatment, accompanied by increased CRY1 protein (Fig. 5B and C and Supplementary Fig. 7B and C). In contrast, in CRY1 KO hepatocytes, GSK3β suppression had little effect on glucose production and gluconeogenic gene expression (Fig. 5A–C and Supplementary Fig. 7A–C), indicating that hepatic CRY1 would be one of key factors mediating the effect of GSK3β on HGP.

Next, we decided to study whether decreased HGP through GSK3β inhibition might be associated with degradation of phospho-CRY1 protein. In CRY1 KO hepatocytes, ectopic expression of CRY1 WT attenuated glucose production and gluconeogenic gene expression, while that of 2SA CRY1 mutant further potently repressed them (Fig. 5D and E). Moreover, GSK3β inhibition failed to repress glucose production and gluconeogenic gene expression induced by 2SA CRY1 overexpression (Fig. 5D and E). Additionally, the FOXO1 protein level was lowered by WT CRY1 and appeared to be further downregulated by 2SA CRY1 (Fig. 5F). In WT CRY1-overexpressing hepatocytes, there was a clear difference in the levels of FOXO1 and CRY1 proteins after GSK3β inhibition (Fig. 5F). However, these differences were abolished in 2SA CRY1–overexpressing hepatocytes (Fig. 5F). For confirmation that GSK3β modulates HGP through FBXL3-dependent CRY1 degradation, FBXL3 was ectopically expressed in primary hepatocytes with or without GSK3β inhibitors. As shown in Fig. 5G, FBXL3 overexpression elevated glucose production, while FBXL3 did not promote glucose production in the presence of a GSK3β inhibitor. In primary hepatocytes, FBXL3 overexpression increased the levels of FOXO1 protein and gluconeogenic gene expression, which were abrogated by GSK3β inhibition (Fig. 5H and I). Furthermore, FBXL3 overexpression did not decrease CRY1 protein expression in the presence of LiCl (Fig. 5I). Together, these findings suggest that GSK3β-induced CRY1 phosphorylation could elevate HGP through FBXL3-dependent CRY1 degradation in the liver.

As CRY1 KO mice exhibit impaired HGP (17), we explored whether CRY1 reduction by activated GSK3β might be attributable to HFCF-mediated hyperglycemia. Compared with WT mice, CRY1 KO mice exhibited higher levels of fasting glucose and HOMA of insulin resistance (HOMA-IR) (Fig. 6A and B). Interestingly, on HFCF, CRY1 KO mice became glucose intolerant (Fig. 6C). In HFCF-fed CRY1 KO mice, the levels of pyruvate-induced blood glucose production and gluconeogenic gene expression were significantly increased (Fig. 6D and E). Together, these results suggest that CRY1 deficiency would lead to more glucose intolerance and elevated hepatic gluconeogenesis in diabetic animals.

For examination of whether GSK3β-induced CRY1 degradation is involved in hyperglycemia with enhanced HGP in diabetic animals, in vivo effects of a GSK3β inhibitor on HGP were evaluated. GSK3β inhibitors did not largely affect body weights in either WT or CRY1 KO mice (Supplementary Fig. 8A and C). Also, there was little histological difference in the liver with or without GSK3β inhibitors (Supplementary Fig. 8B and D). In WT mice, the level of blood glucose was mitigated by GSK3β inhibition during PTT, while GSK3β inhibitors did not significantly affect the level of blood glucose on pyruvate challenge in CRY1 KO mice (Fig. 6F and Supplementary Fig. 8E). In accord with these findings, the expression of Pck1 and G6pc was downregulated by GSK3β inhibitors in WT mice, whereas it was not affected in CRY1 KO mice (Fig. 6G and Supplementary Fig. 8F). Furthermore, in the presence of a GSK3β inhibitor, the level of hepatic FOXO1 protein was downregulated, which was not observed in CRY1 KO mice (Fig. 6H and Supplementary Fig. 8G). Moreover, HGP was examined with in vivo GSK3β suppression in animal models. Similar to results with pharmacological intervention of GSK3β, siGsk3b-treated WT mice had decreased HGP during PTT and reduced hepatic FOXO1 protein, which was not observed in CRY1 KO mice (Supplementary Fig. 8H–J). Together, these data evidently suggest that maintaining CRY1 protein stability would be crucial to attenuate hyperglycemia by repressing HGP in diabetic mice.

Emerging evidence suggests that a decrease in hepatic CRY1 protein is responsible for dysregulated glucose metabolism in diabetes (17,19). Also, several studies have shown that single nucleotide polymorphisms of CRY1 genes are associated with hyperglycemia, insulin resistance, and diabetes risk (38,39). However, the underlying mechanism(s) by which CRY1 protein would be reduced to confer hyperglycemia is largely unknown. In this study, we demonstrated that elevated GSK3β-dependent hepatic CRY1 degradation is attributable to hyperglycemia due to upregulated HGP. Hepatic FBXL3 expression was elevated, promoting CRY1 degradation in diabetes. Mechanistically, activated GSK3β-induced CRY1 phosphorylation at S280/S281 residues facilitated FBXL3-mediated CRY1 degradation, resulting in elevated HGP. Collectively, our findings suggest that the dysregulation of CRY1 protein stability through GSK3β-dependent CRY1 degradation would be one of the etiological features of hyperglycemia.

FBXL3-mediated CRY1 polyubiquitination appears to be important for CRY1 protein stability in the regulation of circadian rhythms (31). Recently, it was demonstrated that FBXL3 is involved in various cellular processes beyond circadian rhythm, including cell proliferation, cell cycle progression, and cancer cell metabolism, by manipulating CRY1 protein stability (40,41). In hepatocytes, we found that FBXL3 modulates CRY1 protein stability by promoting polyubiquitination and proteasomal degradation. Given that CRY1 suppresses HGP through repression of FOXO1 levels, we hypothesized that FBXL3-dependent CRY1 degradation in hepatocytes might elevate the level of FOXO1, leading to enhanced HGP. In WT hepatocytes, FBXL3 overexpression stimulated HGP and the levels of FOXO1 protein and gluconeogenic gene expression, which was abolished in CRY1 KO hepatocytes, indicating that FBXL3-dependent CRY1 degradation stimulates FOXO1-mediated HGP. Here, we found that hepatic FBXL3 expression was elevated in diabetic animals. Although further studies are needed to determine which factors upregulate hepatic FBXL3 in diabetic animals, our data suggest that aberrantly increased FBXL3 would contribute to excessive HGP and metabolic disorders.

The FBXL3-containing SCF complex is important for recognizing and degrading phosphorylated substrate proteins (42). We observed that the level of hepatic CRY1 protein phosphorylation was increased in diabetic animals, implying that FBXL3-dependent CRY1 degradation might be augmented by CRY1 phosphorylation. To date, it is unclear which kinase(s) are involved in the regulation of CRY1 protein stability in diabetic animals. In silico analysis revealed that CRY1 protein might be phosphorylated by several kinases. Among them, hepatic GSK3β likely negatively modulates CRY1 protein stability. We also found that phosphorylation of CRY1 at S280/S281 residues by GSK3β promoted FBXL3-dependent CRY1 polyubiquitination. Of interest, the reduction in CRY1 protein induced by FBXL3 overexpression was attenuated by GSK3β inhibition, suggesting that CRY1 phosphorylated by GSK3β is a preferential target for FBXL3-dependent degradation. Furthermore, the enzymatic activity of hepatic GSK3β is elevated in obese and diabetic animals (35,36). In this regard, we suggest that GSK3β could operate as a rheostat of hepatic CRY1 protein stability in response to pathological cues by modulating the level of CRY1 phosphorylation, leading to FBXL3-dependent degradation in diabetes.

GSK3β has been implicated in inactivating glycogen synthase and regulating gluconeogenic gene expression in glucose metabolism (43). In addition, GSK3 inhibitors have been shown to be effective in normalizing blood glucose levels in animal models of type 2 diabetes (44), and their effects appear to occur primarily through an increase in hepatic glycogen synthesis and a decrease in HGP (45). However, the mechanism by which GSK3β inhibition confers HGP suppression is not thoroughly understood. In this study, we show that CRY1 would be a crucial node in the inhibitory effect of GSK3β suppression on hyperglycemia. FOXO1 protein in WT hepatocytes was decreased by GSK3β repression, concomitantly with increased CRY1 protein levels, which was not largely observed in CRY1 KO hepatocytes. Consistently, the effect of GSK3β inhibitors on decreased FOXO1 protein expression was abolished in CRY1 2SA–expressing hepatocytes. Moreover, GSK3β suppression alleviated excessive HGP and improved hyperglycemia through CRY1-mediated FOXO1 degradation. Nevertheless, we cannot exclude the possibility that other GSK3β targets or related signaling pathways might be involved in HGP because GSK3β suppression marginally alleviated excessive HGP in CRY1 KO. Furthermore, it has been reported that other GSK3β targets, such as CREB, FOXO1, β-catenin, and c-MYC (46–49), would be involved in HGP.

It has been reported that HGP accounts for large portion of endogenous glucose production (50). One of the prominent phenotypes of CRY1 KO mice was enhanced HGP accompanied by increased expression of FOXO1 and gluconeogenic genes. In diabetic animals, restoring CRY1 protein by suppression of hepatic FBXL3 or GSK3β mitigated hyperglycemia with reduced HGP. However, hepatic GSK3β suppression had little effect on HGP in CRY1 KO mice. In addition, the level of glucose production was higher in CRY1 KO primary hepatocytes than in WT primary hepatocytes in a cell-autonomous manner. These in vivo and in vitro data imply that CRY1 would play crucial roles in HGP under physiological and pathological conditions. As we analyzed whole-body CRY1 KO mice, it is feasible that confounding effects of CRY1 deletion in other tissues would also affect HGP and systemic glucose metabolism. Thus, further investigation with hepatocyte-specific KO animal models is needed in the future.

In conclusion, we show that hepatic GSK3β-dependent CRY1 degradation plays a key role in modulating HGP (Fig. 7). In diabetic animals, elevated hepatic FBXL3 facilitates CRY1 degradation via GSK3β-induced phosphorylation, resulting in aberrant FOXO1 accumulation and hyperglycemia. Also, restoring the level of CRY1 protein by inhibiting GSK3β alleviated excessive HGP and ameliorated hyperglycemia. Collectively, our data suggest that the hepatic GSK3β-CRY1-FBXL3 axis would be one of the potential therapeutic targets in diabetes.

Acknowledgments. The authors thank all the members of the Laboratory of Adipocyte and Metabolism Research for helpful discussion.

Funding. This study was supported by the National Research Foundation, funded by the Korean government (NRF-2020R1A3B2078617 to J.B.K.) and the Institute for Basic Science from the Ministry of Science and ICT of Korea (IBS-R008-D1 to J.-S.K.). Y.Y.K., Y.G.J., J.H.S., W.T.L., J.P., H.N., and S.M.H. were supported by the BK21 Plus program.

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

Author Contributions. Y.Y.K. and H.J. designed the study. Y.Y.K. conducted most of the experiments and wrote the manuscript. G.L. performed animal experiments. Y.G.J., J.H.S., J.S.H., J.P., J.Y.H., H.N., and S.M.H. helped with analysis of data and discussion of data. W.T.L. performed intravenous injection for invivofectamine 3.0 delivery. M.P. and S.K. analyzed RNA sequencing and performed NP. J.K. and J.-S.K. performed mass spectrometry experiments and analysis. J.B.K. 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.

Prior Presentation. Parts of this study were presented in abstract form at the 2018 American Society for Biochemistry and Molecular Biology (ASBMB) Annual Meeting, San Diego, CA, 21–25 April 2018, and the virtual 2021 ASBMB Annual Meeting, 27–30 April 2021.