c-JUN inhibits mTORC2 and glucose uptake to promote self-renewal and obesity

Abstract

Metabolic syndrome is associated with obesity, insulin resistance, and the risk of cancer. We tested whether oncogenic transcription factor c-JUN metabolically reprogrammed cells to induce obesity and cancer by reduction of glucose uptake, with promotion of the stemness phenotype leading to malignant transformation. Liquid alcohol, high-cholesterol, fat diet (HCFD), and isocaloric dextrin were fed to wild-type or experimental mice for 12 months to promote hepatocellular carcinoma (HCC). We demonstrated 40% of mice developed liver tumors after chronic HCFD feeding. Disruption of liver-specific c-Jun reduced tumor incidence 4-fold and improved insulin sensitivity. Overexpression of c-JUN downregulated RICTOR transcription, leading to inhibition of the mTORC2/AKT and glycolysis pathways. c-JUN inhibited GLUT1, 2, and 3 transactivation to suppress glucose uptake. Silencing of RICTOR or c-JUN overexpression promoted self-renewal ability. Taken together, c-JUN inhibited mTORC2 via RICTOR downregulation and inhibited glucose uptake via downregulation of glucose intake, leading to self-renewal and obesity.

Keywords: Biological sciences; Cancer; Cell biology; Diabetology; Endocrinology; Molecular biology.

Graphical abstract

Figure 1

c-Jun disruption in hepatocytes reduces liver tumor development and restores glucose intolerance and insulin resistance phenotypes (A) (Left) Experimental design to define the role of c-Jun in liver tumors induced by HCV NS5A and alcohol/diabetes. HCV NS5A Tg mice or their control non-Tg littermates were fed HCFD from 8 weeks of age for 12 months. These transgenic mice express Cre recombinase driven by liver-specific albumin-promoter, which induces the recombination of lox sites that flank the c-Jun gene and its deletion (“floxed”). (A) (Right) Mice were euthanized after 12 months of feeding for analysis of tumor incidence. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (B) Quantitative data are shown for tumor size (ratios of liver tumors weights versus total liver weights) of HCFD-fed Tg mice. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (C) (Top) Representative images of HCFD-fed NS5A Tg mice. (C, Bottom) H&E-stained liver sections from mice. Note the widespread hemorrhaging in the liver of NS5A Tg mice. Representative H&E-stained sections of mouse organs. Scale bars represent 10 μm. (D) Serum endotoxin levels were quantified. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (E, left) NS5A transgene expression was analyzed in c-Jun^fl/fl;NS5A and Alb-Cre;c-Jun^fl/fl;NS5A mice. (E, Right) Immunoblot analyses validated that c-JUN knockout in hepatocytes reduced stemness marker Nanog and malignant HCC marker Vimentin expression in mouse tissues. (F) Tumor-initiating stem-like cells (TICs) are induced in NS5A Tg mice fed HCFD. CD133 + CD49f + cells are considered as TICs. Double staining of liver sections from NS5A Tg mice showed that some of the hepatocytes have both CD133 and CD49f expression, indicating that HCV NS5A increased cancer stem cells in liver. To determine if NS5A expression induces Nanog with higher ligand levels caused by HCFD feeding, liver sections from NS5A Tg mice at the age of 12 months were stained with antibodies directed to Nanog and CD49f. Most of Nanog-expressing hepatocytes have high levels of CD49f, which is a marker of hepatic progenitor cells. Scale bars represent 10 μm. (Bottom) Quantitative data are shown for TIC formation/Number in HCFD-fed Tg mice. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (G) Glucose tolerance test. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk represents statistical significance (p < 0.05). (H) Insulin resistance test. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (I) c-JUN overexpression inhibited glucose uptake in the HepG2 cells. FACs analysis using fluorescent d-glucose analog 2-[N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl) amino]-2-deoxy-D-glucose (2-NBDG) showed that c-Jun overexpression in HepG2 cells decreased glucose uptake. N = 6. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (J) c-JUN overexpressed HepG2 cells also showed the increase in the gluconeogenesis/glucose production, which was performed using glucose assay kit (Sigma). N = 5. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 2

c-Jun inhibits Akt phosphorylation via Rictor downregulation and inhibition of Torc2 complex formation (A) Western blots of signaling molecules showing dysregulation in the AKT pathway by decreased p-Akt-S473 and p-Foxo-1 in c-Jun wild-type mice fed high-fat diet. In contrast, the c-Jun knockout samples showed that p-Akt-S473 and p-Foxo-1 signals increased. (B, left) Western Blots of signaling molecules showing similar dysregulation in the AKT pathway by decreased p-AKT-S473 and p-FOXO-1 in c-JUN overexpressed HepG2 samples. (B, right) c-JUN downregulates the RICTOR levels in mTORC2 complex. Western blot analysis shows that RICTOR levels were decreased in c-JUN overexpressed HepG2 cells, whereas the other components of mTORC2 (MTOR, SIN-1) complex were the same. (C) Real-time RT-PCR analysis also showed that RICTOR mRNA expression was significantly reduced in c-JUN overexpressed HepG2 cells, whereas the expression of the components remained the same in all the samples. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (D, left) Immunoprecipitation analysis revealed that there was no complex formation between SIN-1 and RICTOR. (D, right) Reverse IP-western blot was performed to show if c-JUN overexpression inhibited RICTOR-mTLR1-SIN1 complex formation. c-JUN expression inhibited the interactions between mTOR1 and SIN1. (E) (left) Immunohistochemistry of c-Jun wild-type and c-Jun knockout HCFD-fed mice showed increased Rictor expression in the knockout samples. These liver tissue sections were collected from HCFD-fed Alb-Cre;Jun^fl/fl or Alb-Cre;Jun^fl/fl;NS5A Tg mice. (Right) Immunoreactivity score was calculated based on both staining frequency and intensity. Scale bars represent 10 μm. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (F) Dysregulation of AKT pathway disrupts downstream pathways. Activation of FOXO-1 activates the gluconeogenic genes. Real-time RT-PCR analysis revealed that PEPCK and G6PASE are increased in c-Jun overexpressed HepG2 cells. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (F, the 3^rd to the left) Real-time RT-PCR analysis also showed that the expression of glucokinases decreased in c-Jun overexpressed HepG2 cells. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (F, right) Western blot analysis showed that GSK3β was activated as it is hypophosphorylated in c-JUN overexpressed samples; it also showed that activation of GSK3β inactivates glycogen synthase, as it was hyperphosphorylated in c-Jun overexpressed HepG2 cells. N = 5. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 3

c-Jun downregulation promotes transactivation of RICTOR promoter in TICs and human HCC specimens (A) Knocking down of c-Jun restored Rictor expression. Western blot and RT-PCR showing restoration of Rictor and p-Akt-S473 expression when c-Jun was knocked down in mouse TICs using shRNA lentivirus construct. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk represents statistical significance (p < 0.05). (B) ChIP-qPCR analysis showed c-Jun binding sites on two fragments of Rictor promoter between bases −751 and −529. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (C) Luciferase analysis of RICTOR promoter-pGL3-basic demonstrated that knockdown of c-JUN strongly activated RICTOR expression in a particular region, between bases −751 and −529, and both binding sites were important for Rictor downregulation. N = 5. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (D) In vitro mutagenesis of the two binding sites showed that after mutating the site between bases −622 and −529, the Rictor expression increased in TICs. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (E) c-JUN silencing in HepG2 was confirmed by immunostaining. Scale bars represent 10 μm. (F) RICTOR restoration was confirmed by immunostaining. c-JUN silencing promoted RICTOR expression in HepG2 cells. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Scale bars represent 10 μm. (G) Immunofluorescence staining showed higher levels of c-JUN and lower levels of RICTOR in HCC areas in comparison to those of noncancerous counterparts. Double staining of RICTOR and c-JUN showed that c-JUN expressing cells inversely have less RICTOR staining. Quantitative immunoreactivity scores (IRS) were plotted from many human HCC and nontumorous specimen analyses. Note: c-JUN staining was inversely correlated with that of RICTOR-stained cells in HCC tissues. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Scale bars represent 10 μm.

Figure 4

Metabolomic analyses of NANOG+ c-JUN+ cells showed lower glycolysis and gluconeogenesis pathways in tumors (A) Heatmap of gene expression profiles of human cirrhotic liver tissues and HCCs. (B) Dot plots of gene expression profiles of normal tissues versus HCC tissues. (C) Immunostaining data of human HCC and adjacent tumor tissues of c-JUN and NANOG staining. Scale bars represent 10 μm. (D) Gene set enrichment analyses (GSEA) in HCV transgenic mice fed high-cholesterol high-fat diet. Mouse liver microarray data were analyzed by gene-set enrichment analyses. WT: nontransgenic littermates fed high-cholesterol high-fat diet (HCFD) for 12 months; HCFD: HCV NS5A transgenic mice fed HCFD for 12 months. (E) The mRNA expression levels of TLR4 and/or c-JUN are elevated in one-thirds of HCC patients (32%) from TCGA data analysis by use of cBioPortal. (F) TCGA data analysis by use of cBioPortal showed that the HCC patients who have higher mRNA levels of TLR4 and/or c-JUN have poorest survival rate (Log-rank test p value: 0.0250). HCC patients with altered TLR4 expression have poor prognosis.

Figure 5

Glycolysis, glucose metabolism, and immunofluorescence analyses (A) TIC versus hepatocytes for metabolomic analyses. Metabolomic analyses of culture media and cells of mouse TICs and primary hepatocytes. Mouse CD133+ TICs and primary hepatocytes were compared for metabolomics data. (B) Illustration of TCA cycle is marked by up- and downregulation of cellular metabolism of TICs over the values of primary hepatocytes. (C) (Left) Seahorse assays showed that c-Jun silencing promoted ECAR (glycolysis) in TICs, but not in scrambled shRNA-transduced cells. (C, right) Silencing of c-Jun promoted glycolysis. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (D) Upregulation and downregulation of metabolites in TICs in comparison to those in mouse primary hepatocytes. Hypothetical model of Nanog-mediated glucose metabolic shift. (E) Glucose assays and gene expression analyses of Nanog and c-Jun in the presence or absence of silencing of c-Jun or Nanog. (C, lower right) Silencing effects of shRNA transduction on c-Jun were validated by qRT-PCR analyses. N = 5. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (F) Immunofluorescence staining showed higher levels of c-JUN and lower levels of GLUT2 in patient HCC areas in comparison to noncancerous tissues. Double staining of GLUT2 and c-JUN showed that c-JUN expressing cells inversely have less GLUT2 staining. Quantitative immunoreactivity scores (IRS) were plotted from many human HCC and nontumorous specimen analyses. Note: c-JUN staining was inversely correlated with that of GLUT2 stained cells in HCC tissues. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Scale bars represent 10 μm.

Figure 6

Glucose metabolism and GLUT2 gene regulatory mechanism (A) (Left) Silencing effects of shRNA transduction on c-JUN were validated by immunoblot analyses. (Middle) Immunofluorescence staining analyses showed restoration of GLUT2 expression in c-JUN-silenced HepG2 cells. Scale bars represent 10 μm. (Right) Silencing c-JUN induced GLUT1 and GLUT3 protein levels. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (B) Effects of cellular c-JUN expression on various regions of the GLUT2 promoter. Silencing of c-JUN increased GLUT2 promoter activity compared with the control in Huh7 cells transfected with all GLUT2 promoter-luciferase constructs compared with the control. No significant difference in luciferase activity was observed when −1100 AP-1 binding region was truncated. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (C) Effects of cellular c-JUN expression on various regions of the GLUT2 promoter. Overexpression of c-JUN produced no significant changes in luciferase expression in the series of GLUT2 promoter-luciferase constructs. No significant difference in luciferase activity was observed when −1100 AP-1 binding region was truncated. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. ∗p < 0.05, ∗t∗p < 0.01, ∗∗∗p < 0.001. (D) (Top) In vitro mutagenesis of AP-1 consensus sequences in the GLUT2 promoter. The (−206/+308) GLUT2 promoter-luciferase constructs were used to produce mutated AP-1 sites at −144 or −123. (Bottom) Mutation at the −144 AP-1 site resulted in significantly greater luciferase activity. Overexpression of c-JUN reduced luciferase expression in the unmutated condition. In contrast, overexpression continued to produce a significant increase in luciferase activity in the mutant condition. Mutation of one c-JUN/AP1 binding site abrogated the c-JUN-mediated repression of GLUT2 promoter activity. N = 6. Asterisk represents statistical significance (p < 0.05). Constructs containing the full-length (−1291) GLUT2 promoters were mutated at the −1100 AP-1 site. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (E) Mutagenesis of the −1100 AP-1 site resulted in decreased luciferase activity, whereas the overexpression of c-JUN produced a greater luciferase expression compared with the empty vector control. Activation of c-JUN at serine 63/73 increased GLUT2 promoter luciferase reporter activities and protein expression in HepG2 cells. HepG2 cells were seeded into T-25 flasks and transfected with 500 ng of expression vectors containing WT S63/73S His₆-tagged c-JUN (MT35), phosphomimetic D63/73D His₆-tagged c-JUN (MT112), or empty vector control (PBSK). Overexpression of MT35 c-JUN produced a mild increase in GLUT2 promoter luciferase reporter activities (p < 0.05) and protein. Overexpression of MT112 c-JUN produced a pronounced increase in GLUT2 mRNA (p < 0.001) and protein. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (F) Immunoblot confirmed the c-JUN overexpression by use of anti-c-JUN or anti-His-tag antibody. Overexpression of nonphosphorylatable c-JUN mutant promoted GLUT1 and GLUT3 expression. (G) Activation of c-JUN at serine 63/73 increased GLUT2 mRNA and protein expression in HepG2 cells. HepG2 cells were seeded into T-25 flasks and transfected with 500 ng of expression vectors containing WT S63/73S His₆-tagged c-JUN (MT35), nonphosphorylatable A63/73A His₆-tagged c-JUN (MT111), phosphomimetic D63/73D His₆-tagged c-JUN (MT112), or empty vector control (PBSK). Overexpression of MT35 c-JUN produced a mild increase in GLUT2 mRNA (p < 0.05) and protein. Overexpression of MT112 c-JUN produced a pronounced increase in GLUT2 mRNA (p < 0.001) and protein. However, MT111 c-JUN overexpression produced a decrease in GLUT2 mRNA (p < 0.01) with mild change in protein levels. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (H) LPS stimulation promoted c-JUN phosphorylation. (I) Overexpression of c-JUN produced no significant changes in luciferase expression in the series of GLUT2 promoter-luciferase constructs. No significant difference in luciferase activity was observed when −1100 AP-1 binding region was truncated. Data are represented as mean ± SD. Student’s t test was used to calculate statistical significance ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (J) HepG2 cells with low levels of c-JUN have high levels of GLUT3 proteins. Overexpression of c-JUN or phosphomimetic mutant of c-JUN (D63/73D) reduced GLUT2 levels, whereas nonphosphorylatable mutant of c-JUN overexpression restored the GLUT3 protein levels. Scale bars represent 10 μm.

Figure 7

c-Jun silencing reduces self-renewal, tumor-initiation property whereas RICTOR silencing promotes self-renewal ability (A) (Top) Colony formation assay. To determine if c-Jun inhibition reduced oncogenicity, colony formation was employed using loss-of-function approach by use of lentivirus expressing shRNA targeting c-Jun. Silencing c-Jun significantly reduced colony numbers in soft agar assays. (Bottom) c-Jun silencing reduced colony numbers. N = 3. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Data are represented as mean +/− SD. (B) CD133+ TICs had higher levels of c-Jun expression than those of non-TIC population. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Data are represented as mean +/− SD. (C) Validation of RICTOR silencing in HepG2 cells. (D) (Top) Representative images of spheroids of Huh7 cells transduced with lentivirus expressing shRNA targeting c-JUN, GLUT2 or scrambled shRNAs. (Bottom) RICTOR silencing promoted spheroid formation in Huh7 cells especially in Huh7 cells that are cultured in low-glucose media. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Data are represented as mean ± SD. (E) Overexpression of c-JUN promoted CD133 protein levels. (F) CD133(+)/CD49F(+) cells have elevated levels of CD133, CD49F, c-Jun, Nanong, and Oct4 proteins. (G) (Left) c-Jun silencing reduced Nanog expression, but induced Rictor expression. The c-Jun silencing promoted Rictor expression in TICs, indicating that c-Jun was required to maintain stemness and oncogenicity of TICs and partly depended on Rictor suppression. (Right) Xenograft tumors with shRNA targeting c-JUN showed significantly smaller tumor sizes, indicating that c-JUN played a key role for self-renewal and tumor-initiation property. N = 8 mice per groups. Student’s T-Test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗: p < 0.05, ∗∗: p < 0.01, ∗∗∗: p < 0.001. Data are represented as mean ± SD. (H) Xenograft tumors with c-JUN overexpression in TIC population (CD133+/CD49F+) showed significantly larger tumor sizes with HCFD feeding. Overexpression of c-JUN further promoted tumor growth. N = 6 mice per groups. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Data are represented as mean ± SD. (I) Xenograft tumors with c-JUN overexpression in non-TIC population (CD133−/CD49F−) showed significantly larger tumor sizes with HCFD feeding, indicating that c-JUN and HCFD feeding effects played key roles for self-renewal and tumor-initiation property. These results indicated that c-Jun overexpression promoted the self-renewal tumor-initiation property of non-TIC population induced by HCFD feeding. N = 7 mice per groups. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Data are represented as mean ± SD. (J) AC133 promoter has five major promoter regions (P1–P5). (K) The P1 promoter region of AC133 promoter had the highest levels of transactivation activity in Huh7 HCC cell line. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Data are represented as mean ± SD. (L) Truncation promoter analyses showed that the region of AC133 promoter between −341 and +50 had the highest levels of transactivation ability. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Data are represented as mean ± SD. (M) To test if c-JUN/AP1 and/or NANOG binding transactivate the AC133 promoter, in vitro mutagenesis abrogated the AP1 binding sites and NANOG promoter binding sites. These AP1 and/or NANOG binding sites reduced AC133 promoter activity, indicating that AP1 and NANOG proteins are required to transactivate AC133 promoter. Student’s t test was used to calculate statistical significance. Asterisk (∗) indicates statistical significance. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Data are represented as mean ± SD.