Research ArticleCancer

Squalene epoxidase drives NAFLD-induced hepatocellular carcinoma and is a pharmaceutical target

See allHide authors and affiliations

Science Translational Medicine  18 Apr 2018:
Vol. 10, Issue 437, eaap9840
DOI: 10.1126/scitranslmed.aap9840

Antifungal drug may help fight cancer

The rates of nonalcoholic fatty liver disease are increasing with the rising prevalence of obesity and are associated with an increase in the incidence of hepatocellular carcinoma. By studying the mechanism of hepatocellular carcinoma arising in the setting of nonalcoholic fatty liver disease, Liu et al. discovered a role for squalene epoxidase, a metabolic enzyme involved in cholesterol biosynthesis. Using a combination of human cancer cells and mouse models of disease, the authors examined the pathway by which squalene epoxidase promotes the development of hepatocellular carcinoma and identified terbinafine, a clinically approved antifungal drug, as a potential intervention.

Abstract

Nonalcoholic fatty liver disease (NAFLD)–induced hepatocellular carcinoma (HCC) is an emerging malignancy in the developed world; however, mechanisms that contribute to its formation are largely unknown, and targeted therapy is currently not available. Our RNA sequencing analysis of NAFLD-HCC samples revealed squalene epoxidase (SQLE) as the top outlier metabolic gene overexpressed in NAFLD-HCC patients. Hepatocyte-specific Sqle transgenic expression in mice accelerated the development of high-fat, high-cholesterol diet–induced HCC. SQLE exerts its oncogenic effect via its metabolites, cholesteryl ester and nicotinamide adenine dinucleotide phosphate (NADP+). Increased SQLE expression promotes the biosynthesis of cholesteryl ester, which induces NAFLD-HCC cell growth. SQLE increased the NADP+/NADPH (reduced form of NADP+) ratio, which triggered a cascade of events involving oxidative stress–induced DNA methyltransferase 3A (DNMT3A) expression, DNMT3A-mediated epigenetic silencing of PTEN, and activation of AKT-mTOR (mammalian target of rapamycin). In human NAFLD-HCC and HCC, SQLE is overexpressed and its expression is associated with poor patient outcomes. Terbinafine, a U.S. Food and Drug Administration–approved antifungal drug targeting SQLE, markedly inhibited SQLE-induced NAFLD-HCC cell growth in NAFLD-HCC and HCC cells and attenuated tumor development in xenograft models and in Sqle transgenic mice. Suppression of tumor growth by terbinafine is associated with decreased cholesteryl ester concentrations, restoration of PTEN expression, and inhibition of AKT-mTOR, consistent with blockade of SQLE function. Collectively, we established SQLE as an oncogene in NAFLD-HCC and propose that repurposing SQLE inhibitors may be a promising approach for the prevention and treatment of NAFLD-HCC.

INTRODUCTION

Nonalcoholic fatty liver disease (NAFLD) affects 30 to 40% of the adult population (1, 2), and its incidence is very high in obese individuals (75 to 100%) (3, 4). About 2.4 to 12.8% of NAFLD patients with cirrhosis progress to NAFLD-associated hepatocellular carcinoma (HCC) (4, 5). Given its prevalence in the general population and the rising obesity epidemic, NAFLD is an emerging risk factor for HCC (6). NAFLD is associated with high hepatic triglyceride and cholesterol content (7). Excess cholesterol is lipotoxic, and it induces nonalcoholic steatohepatitis, a severe form of NAFLD that strongly predisposes to HCC development (8). To probe the molecular mechanism of NAFLD-HCC, we performed RNA sequencing (RNA-seq) analysis of 17 paired human NAFLD-HCC and adjacent normal tissues, which revealed squalene epoxidase (SQLE) as an outlier gene markedly up-regulated in NAFLD-HCC (25.2-fold). SQLE encodes a monooxygenase and is the second rate-limiting enzyme in cholesterol biosynthesis by catalyzing the first oxygenation step in sterol biosynthesis (9). Because hepatic cholesterol accumulation underlies the development of NAFLD, we hypothesized that SQLE might play a crucial role in NAFLD-HCC. However, the pathological role of SQLE in NAFLD-HCC remains unclear.

Here, we evaluated the oncogenic role of SQLE in normal liver and NAFLD-HCC cell lines in vitro and developed hepatocyte-specific Sqle transgenic (tg) mice to assess its effect on NAFLD-HCC development in vivo. We also elucidated the mechanism of SQLE action in promoting NAFLD-HCC and assessed the therapeutic efficacy of an SQLE inhibitor, terbinafine, for the treatment of this disease.

RESULTS

SQLE is overexpressed in NAFLD-HCC tissues

We performed RNA-seq analysis of 17 paired NAFLD-HCC tumor and adjacent normal tissues. Reactome analysis of differentially expressed genes identified metabolism as a core set of pathways altered in NAFLD-HCC. Among up-regulated metabolic genes, SQLE was a top outlier gene and it was overexpressed in 16 of 17 paired NAFLD-HCC samples (Fig. 1, A and B), with 25.2-fold up-regulation. The increase in SQLE mRNA was validated in an independent cohort of 10 paired NAFLD-HCC samples (Fig. 1C). SQLE protein was also increased in NAFLD-HCC (Fig. 1D). We next determined the expression of Sqle in two obesity-associated NAFLD-HCC mouse models. Sqle was up-regulated in all HCC tumors (six of six) from N,N-diethylnitrosamine (DEN) and high-fat, high-cholesterol (HFHC) diet–treated C57BL/6 mice (Fig. 1E). Similarly, Sqle was up-regulated in 8 of 10 HCC tumors from DEN-treated db/db mice (Fig. 1E). SQLE is therefore commonly overexpressed in human NAFLD-HCC and experimental NAFLD-HCC mouse models. In addition, we observed that the copy number amplification of SQLE was positively correlated with its mRNA expression in human NAFLD-HCC (Fig. 1F), indicating that copy number gain contributes to up-regulation of SQLE. Moreover, in silico analysis indicated that transcription factors MEIS1, SREBP2, and EVI1 might bind to the SQLE promoter (fig. S1A). In the NAFLD-HCC RNA-seq cohort, mRNA expression of SQLE was positively correlated with MEIS1 (R = 0.559, P = 0.03) and SREBP2 (R = 0.542, P = 0.03), but not EVI1 (fig. S1B), suggesting that SQLE expression may also be regulated by transcription factors, especially SREBP2 and MEIS1. SREBP2 is a known transcriptional regulator of SQLE (10). Knockdown of SREBP2 in LO2 (normal hepatocyte) and HKCI2 (NAFLD-HCC) (11, 12) cells inhibited SQLE protein expression (fig. S2A). Predicted binding motif of MEIS1 is TGACA (fig. S2B), and it is located −669 base pairs from the transcription start site (fig. S2C). The interaction of MEIS1 with SQLE promoter was confirmed using chromatin immunoprecipitation (ChIP)–polymerase chain reaction (PCR) and luciferase reporter assays (fig. S2, D and E). MEIS1 overexpression in LO2 and HKCI2 cells induced SQLE mRNA and protein expression, whereas knockdown of MEIS1 in LO2 and HKCI2 cells suppressed SQLE (fig. S2, F and G). Gene amplification and transcription factors are hence the main drivers of SQLE overexpression in NAFLD-HCC.

Fig. 1 SQLE is overexpressed in NAFLD-HCC.

(A) RNA-seq analysis of 18 paired NAFLD-HCC and adjacent normal tissues (left). SQLE was the top outlier gene among the up-regulated metabolic genes (right). FPKM, fragments per kilobase of transcript per million mapped reads. (B) SQLE mRNA expression in 17 individual paired NAFLD-HCC and adjacent normal samples (one paired sample was not available for analysis). (C) Increased SQLE mRNA and (D) protein expression in human NAFLD-HCC were validated in an independent cohort. Arrows show SQLE protein. (E) Sqle mRNA expression was up-regulated in dietary and genetic NAFLD-HCC animal models: DEN-injected and high-fat diet–treated wild-type (WT) mice (left) and DEN-treated db/db mice (right). (F) Analysis of correlation between SQLE gene copy number and mRNA expression in 17 paired NAFLD-HCC. Data are means ± SEM. (C to E) Paired two-tailed Student’s t tests were used. (F) The Pearson correlation coefficient was used.

SQLE promotes cell growth, regulates cell cycle progression, and inhibits apoptosis in NAFLD-HCC cell lines

To assess the oncogenic effect of SQLE in human HCC, we constructed LO2 and HKCI10 (NAFLD-HCC) (11, 12) cell lines stably expressing SQLE (Fig. 2A). SQLE expression promoted cell growth, as evidenced by cell growth curve and colony formation assays (Fig. 2, A and B). Conversely, knockdown of SQLE in HKCI2 (also NAFLD-HCC) cells by short hairpin RNA (shRNA) inhibited cell viability compared with shControl (Fig. 2, C and D). Apart from NAFLD-HCC cell lines, SQLE also promoted the growth and colony formation of Huh7 HCC cell line, whereas the silencing of SQLE had an opposite effect in BEL-7404 and HepG2 HCC cells (fig. S3, A and B). Moreover, LO2 and HKCI10 cells overexpressing SQLE showed an increase in S-phase cell population, with a concomitant decrease in cells in the G1 phase (Fig. 2E). In contrast, SQLE knockdown in HKCI2 cells induced G1 arrest (Fig. 2F). In HCC cell lines, manipulation of SQLE expression had the same effect on cell cycle (fig. S4, A and B), indicating that SQLE accelerated G1-S progression. Consistent with these observations, SQLE induced protein expression of cyclin D1 and proliferating cell nuclear antigen (PCNA) (Fig. 2G). SQLE also regulates apoptosis, and its overexpression suppressed apoptosis in LO2 and HKCI10 cells and reduced the protein expression of cleaved forms of caspase-7 and caspase-9, whereas SQLE knockdown had an inverse effect in HKCI2 and BEL-7404 cells (Fig. 2, H to J, and fig. S4C). These results suggest that SQLE promoted cell growth by activating cell cycle progression and inhibiting apoptosis.

Fig. 2 SQLE promotes NAFLD-HCC cell growth.

(A) Overexpression of SQLE in LO2 (normal liver) and HKCI10 (NAFLD-HCC) cells was confirmed by Western blot analysis. Ectopic SQLE expression promoted cell viability (A) and colony formation (B) (n = 3, performed in triplicate). (C) SQLE knockdown in HKCI2 (NAFLD-HCC) cells was confirmed by Western blot analysis. SQLE knockdown suppressed cell viability (C) and colony formation (D) (n = 3, performed in triplicate). (E) Cells were stained with propidium iodide (PI) and analyzed by flow cytometry. Flow cytometry analysis indicated that SQLE expression decreased the number of cells in G1 phase but increased the number of cells in S phase (n = 3, performed in triplicate). (F) Cells were stained with PI and analyzed by flow cytometry. Flow cytometry analysis indicated that knockdown of SQLE in HKCI2 induced G1 arrest (n = 3, performed in triplicate). (G) Western blot analysis indicated that SQLE overexpression increased PCNA and cyclin D1 expression, whereas SQLE knockdown had the opposite effects. (H) SQLE overexpression reduced apoptosis, as determined by annexin V–phycoerythrin and 7-aminoactinomycin D (7-AAD) staining and flow cytometry (n = 3, performed in triplicate). (I) SQLE knockdown in HKCI2 cells induced apoptosis, as determined by annexin V–phycoerythrin and 7-AAD staining and flow cytometry (n = 3, performed in triplicate). (J) Western blot analysis showed that SQLE overexpression reduced the protein expression of cleaved forms of caspase-7 and caspase-9, whereas SQLE knockdown had the opposite effects. Data are means ± SEM. Difference in cell viability between two groups was determined by repeated-measures ANOVA. Mann-Whitney U test or Student’s t test was performed to compare the variables in two groups (colony formation, cell cycle, and apoptosis). *P < 0.05, **P < 0.01, ***P < 0.001.

Hepatocyte-specific tg SQLE expression in mice accelerates NAFLD-HCC formation

To determine the relevance of SQLE in NAFLD-HCC development in vivo, we constructed Sqle tg mice. Crossing Sqle tg mice to Albumin-Cre (Alb-Cre) mice results in hepatocyte-specific Sqle expression (Fig. 3A). To evaluate the role of Sqle in NAFLD-HCC, we injected WT and Sqle tg mice with a single dose of DEN at days 10 to 13. Starting at the age of 6 weeks, mice were fed with HFHC diet (Fig. 3B). At 25 weeks of age, mice were sacrificed and the liver was analyzed. Significantly more Sqle tg mice developed tumors (9 of 10) as compared to WT mice (2 of 10) (P = 0.003), and histological examination [hematoxylin and eosin (H&E) staining] confirmed HCC formation in the livers of Sqle tg mice, together with hallmarks of fatty liver disease such as hepatocyte ballooning and inflammatory cell infiltration (Fig. 3B). Sqle tg mice also showed increased liver weight and liver/body weight ratio but not body weight (Fig. 3C). Consistent with development of HCC, α-fetoprotein (AFP), a serum biomarker for liver cancer, was increased in Sqle tg mice (Fig. 3D). Serum concentrations of alanine aminotransferase (ALT) and aspartate aminotransferase (AST), markers for liver inflammation and injury, were also significantly higher in Sqle tg mice (P < 0.01; Fig. 3D). We next performed Ki-67 staining to assess cell proliferation (Fig. 3E). Compared to WT mouse liver tissues, nontumorous liver tissues from Sqle tg mice showed increased cell proliferation, and tumors derived from Sqle tg mice had the highest Ki-67 scores (Fig. 3E), which is consistent with our in vitro observations. We also performed TUNEL [terminal deoxynucleotidyl transferase (TdT)–mediated deoxyuridine triphosphate nick end labeling] staining and Western blots to evaluate liver apoptosis. Compared to WT mouse liver tissues, nontumor liver tissues from Sqle tg mice had higher TUNEL scores (fig. S5A) and increased amounts of cleaved forms of caspase-3 and caspase-7 (fig. S5B). Cytokine and Chemokines PCR array indicated that Sqle overexpression increased the mRNA expression of proinflammatory mediators, including Ccl-12, Ccl-20, Ccr-5, Cxcl-9, Cxcr-4, Il-2, Il-4, Il-18r1, Osm, and Spp1 (fig. S5C). Collectively, these data demonstrate that SQLE overexpression exacerbates HFHC diet–induced NAFLD and promotes NAFLD-HCC formation in mice by inducing inflammation and cell proliferation.

Fig. 3 Hepatocyte-specific transgenic SQLE expression in mice accelerates HFHC diet–associated NAFLD-HCC.

(A) Scheme for the generation of mice with hepatocyte-specific Sqle overexpression. Western blot confirmed overexpression of SQLE in the livers of Sqle tg mice. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. (B) Experimental design for DEN-injected and HFHC diet–induced mouse models of NAFLD-HCC. At the age of 10 to 13 days, WT or Sqle tg mice were injected with a single dose of DEN. Starting at 6 weeks of age, mice were fed with HFHC diet until week 25 (top). H&E staining of WT and Sqle tg mouse livers (middle). HCC tumor incidence and multiplicities in WT and Sqle tg mice (bottom). The red arrows show the tumor. Results are means ± SEM (n = 9 to 10). (C) Hepatocyte-specific Sqle expression increased liver weight (left) and liver/body weight ratio (middle), but not body weight (right), in DEN-injected HFHC diet–treated mice. (D) Serum AFP (left), ALT (middle), and AST (right) concentrations of WT and Sqle tg mice treated with DEN and HFHC diet. Results are means ± SEM (n = 9 to 10). (E) Ki-67 staining of livers from DEN-injected and HFHC diet–treated WT and Sqle tg mice. The red arrows show the Ki-67–positive cells. T, tumor; N, adjacent normal. Results are means ± SEM (n = 9 to 10). Mann-Whitney U test was used. Scale bars, 50 μm. ***P < 0.001.

SQLE promotes cell growth via intracellular cholesterol/cholesteryl ester accumulation

Given that SQLE is the rate-limiting enzyme for cholesterol biosynthesis, we hypothesized that its overexpression may deregulate cholesterol metabolism in NAFLD-HCC. We first determined the concentrations of cholesterol and cholesteryl esters in WT and Sqle tg mice. With a normal diet, Sqle tg mice showed markedly enhanced accumulation of liver free cholesterol and cholesteryl ester compared to WT littermates (Fig. 4A). Moreover, an HFHC diet together with Sqle tg exacerbated liver free cholesterol and cholesteryl ester concentrations (Fig. 4A). In vitro, both intracellular free cholesterol and cholesteryl ester were increased in LO2-SQLE and HKCI10-SQLE cells compared to their control cells, especially for cholesteryl ester (more than sevenfold) (Fig. 4B). Conversely, SQLE knockdown in HKCI2 and BEL-7404 reduced free cholesterol and cholesteryl ester concentrations (Fig. 4B). To determine whether cholesterol contributed to the growth-promoting effect of SQLE, we cultured LO2 and HKCI2 cells in the presence of exogenous cholesterol. Cholesterol promoted cell viability (Fig. 4C). Analysis of intracellular cholesterol content revealed that cholesteryl ester, but not free cholesterol, was increased after cholesterol treatment (Fig. 4C), suggesting that cholesteryl ester is responsible for tumor-promoting effect of SQLE in both LO2 and HKCI2 cell lines. Consistent with our hypothesis, avasimibe (an acyl–coenzyme A: cholesterol acyltransferase inhibitor, which blocks cholesterol ester synthesis) abolished growth-promoting effects of cholesterol in LO2 and HKCI2 cells (Fig. 4D). Moreover, exogenous cholesteryl ester directly induced cell growth in LO2 and HKCI2 cells (Fig. 4E), thereby confirming an oncogenic role of this metabolite in HCC. Finally, the incubation of SQLE-overexpressing HKCI2 and HKCI10 cells with avasimibe repressed cell growth to baseline rates (Fig. 4F). Reduction in cholesteryl ester concentrations confirmed the on-target inhibition by avasimibe (Fig. 4G). To confirm the correlation of cholesteryl ester with NAFLD-HCC, we measured the cholesterol/cholesteryl ester concentrations in 10 paired NAFLD-HCC tumors and adjacent normal tissues (fig. S6, A and B). We found increased concentrations of cholesteryl ester (P < 0.05) and free cholesterol (P < 0.05) in primary NAFLD-HCC compared to adjacent normal tissues. We also observed a positive correlation between free cholesterol and cholesteryl ester in adjacent normal tissues (R = 0.7354, P = 0.015) but not in HCC (R = 0.5379, P = 0.100) (fig. S6C). Our data therefore indicate that SQLE induces cholesteryl ester biosynthesis, which contributes to the pathogenesis of NAFLD-HCC.

Fig. 4 SQLE promotes intracellular cholesterol/cholesteryl ester accumulation, which induces tumor cell growth.

(A) Hepatocyte-specific Sqle expression increased liver free cholesterol (left) and cholesteryl ester (right) concentrations in normal diet or DEN-injected, HFHC diet–treated mice. (B) SQLE overexpression in LO2 and HKCI10 cells increased intracellular free cholesterol and cholesteryl ester (left). Knockdown of SQLE in HKCI2 and an HCC cell line (BEL-7404) decreased intracellular free cholesterol and cholesteryl ester (right) (n = 4, performed in triplicate). (C) Exogenous cholesterol promoted cell growth and increased intracellular cholesteryl ester concentration but had no effect on free cholesterol (n = 3, performed in triplicate). (D) Avasimibe abolished the proliferative effect of cholesterol (n = 3, performed in triplicate). (E) Cholesteryl ester directly promoted cell growth (n = 2, performed in triplicate). (F) Avasimibe abolished SQLE-induced cell growth and (G) reversed cholesteryl ester accumulation (n = 4, performed in triplicate). Data are means ± SEM. The significance of the differences in cell growth rates was determined by repeated-measures ANOVA. The significance of the difference in cholesterol concentrations was determined by Mann-Whitney U test. *P < 0.05, **P < 0.01, ***P < 0.001.

SQLE suppresses PTEN and activates the PTEN/PI3K/AKT/mTOR pathway in NAFLD-HCC

Next, we investigated the downstream molecular mechanisms underlying the oncogenic function of SQLE using cancer pathway luciferase reporter assays. Among several critical cancer-related gene reporters, SQLE significantly suppressed FOXO3 reporter [phosphatidylinositol 3-kinase (PI3K)/AKT pathway–responsive reporter] activity (P < 0.01) (Fig. 5A and fig. S7, A and B), suggesting that SQLE activated PI3K/AKT signaling. To elucidate the effect of SQLE on the PI3K/AKT pathway, we profiled gene expression of LO2-vector and LO2-SQLE cells using PI3K/AKT pathway PCR array. We found that SQLE enhanced the expression of JUN but decreased the expression of PTEN, CHUK, APC, GSK3B, and CDKN1B (Fig. 5B). Among these outlier genes, PTEN functions as an upstream regulator of the PI3K/AKT pathway, whereas other candidate genes are downstream effectors of this pathway. Hence, we hypothesized that SQLE activates PI3K/AKT pathway through silencing of PTEN. Western blot revealed that overexpression of SQLE in LO2 and HKCI10 cells silenced PTEN and induced expression of phosphorylated AKT (p-AKT) and phosphorylated mammalian target of rapamycin (p-mTOR) (Fig. 5C). Reciprocally, silencing of SQLE induced PTEN but inhibited p-AKT and p-mTOR expression in HKCl2 cells (Fig. 5C). Notably, Sqle tg mouse livers demonstrated PTEN loss and increased p-mTOR expression as compared to WT mice (Fig. 5D). To further corroborate our findings, we analyzed expression of 10 genes that are negatively regulated by PTEN. SQLE overexpression promoted expression of PTEN downstream target genes, including BCL2, SP2, NFKB, KRAS, ESR, AKT1, FGF21, FANCD, BRAF, and CD44, whereas SQLE knockdown decreased the expression of these genes (Fig. 5E). Together, our data suggested that SQLE induced PTEN silencing, resulting in the activation of PI3K/AKT/mTOR. Given that PTEN/PI3K/AKT/mTOR pathway is correlated with tumorigenesis in multiple cancers (13, 14), we postulated that SQLE promotes NAFLD-HCC by regulating this pathway.

Fig. 5 Oncogenic function of SQLE depends on PTEN/PI3K/AKT/mTOR.

(A) SQLE activated the phosphatidylinositol 3-kinase (PI3K)/AKT pathway, as indicated by FOXO3 luciferase reporter assay (n = 3, performed in triplicate). FOXO3 is a target of AKT, and PI3K/AKT pathway activation inhibited FOXO3 expression. (B) PI3K/AKT pathway PCR analysis of LO2 cells overexpressing SQLE. PTEN is one of the outlier down-regulated genes. (C and D) Representative Western blot analysis confirmed that SQLE induced PTEN silencing and activated downstream oncogenic signaling [phosphorylated AKT (p-AKT) and phosphorylated mTOR (p-mTOR)] in vitro (C) and in vivo (D). (E) mRNA expression of PTEN/PI3K/AKT/mTOR downstream effectors in SQLE-overexpressing LO2 cells and SQLE-knockdown BEL-7404 cells (n = 3, performed in triplicate). (F) SQLE downstream effects on cholesteryl ester accumulation and cell growth depend on PTEN/mTOR. Knockdown of PTEN followed by overexpression of SQLE in LO2 and HKCI10 cells prevented the effects of SQLE on cholesteryl ester accumulation (left), cell viability (middle), and PTEN/mTOR signaling (right). (G) mTOR inhibitor rapamycin suppressed the effect of SQLE on cholesteryl ester accumulation (left) (n = 4, performed in triplicate) and cell growth (right) (n = 3, performed in triplicate). Data are means ± SEM. Mann-Whitney U test was used to assess the significance of the differences in cholesterol concentrations, mRNA expression, and luciferase reporter assay. The significance of the difference between cell growth rates was determined by repeated-measures ANOVA. *P < 0.05, **P < 0.01, ***P < 0.001.

Oncogenic effects of SQLE depend on PTEN/PI3K/AKT/mTOR pathway

We then sought to determine whether the PTEN/PI3K/AKT/mTOR cascade is involved in mediating the oncogenic effect of SQLE. In LO2 and HKCI10 cell lines, we silenced PTEN by small interfering RNA (siRNA). Knockdown of PTEN in LO2 and HKCI10 cells induced intracellular cholesteryl ester accumulation, cell viability, and mTOR phosphorylation (Fig. 5F), thereby recapitulating the effect of SQLE. SQLE overexpression in PTEN-silenced cell lines did not further promote cell growth, intracellular cholesteryl ester accumulation, or mTOR phosphorylation (Fig. 5F), suggesting that PTEN loss plays a crucial role downstream of SQLE.

To confirm involvement of the PTEN/PI3K/AKT/mTOR cascade in SQLE-induced tumorigenesis, we treated SQLE-overexpressing cells with two PI3K inhibitors (GDC0941 and BYL719) and one mTOR inhibitor (rapamycin) (Fig. 5G and fig. S7C) in vitro. Consistent with our hypothesis, the cell viability assay demonstrated that inhibition of either PI3K or mTOR nearly abolished SQLE-mediated cell growth in both cell lines. mTOR inhibition profoundly inhibited SQLE-induced cholesterol acyltransferase 1/2 (SOAT1/2) expression (fig. S7D) and suppressed the induction of cholesteryl ester by SQLE (Fig. 5G). Collectively, our results suggest that SQLE-dependent tumor cell growth and cholesteryl ester accumulation are consequences of the activation of the PTEN/PI3K/AKT/mTOR signaling pathway.

SQLE silences PTEN via induction of ROS-DNMT3A axis

Given that SQLE functions primarily as a metabolic enzyme, we hypothesized that SQLE might silence PTEN via its metabolic products. SQLE is the second rate-limiting enzyme of intracellular cholesterol biosynthesis, and it functions as a monooxygenase that uses an oxygen atom from O2 to oxidize squalene and simultaneously reduces the other oxygen atom using reducing equivalents from NADPH (reduced form of NADP+) to generate NADP+ (nicotinamide adenine dinucleotide phosphate) (9, 15). HMGCR, the first rate-limiting enzyme of the cholesterol biosynthesis pathway, also consumes reducing equivalents in the form of NADPH (15, 16). Consistent with this notion, overexpression of SQLE significantly increased the NADP+ to NADPH ratio in vivo (P < 0.01) and in vitro (P < 0.01; Fig. 6, A and B). Blockade of HMGCR using simvastatin reversed SQLE-induced increase in NADP+/NADPH ratio, cholesterol biosynthesis, and cell viability, indicating that HMGCR- and SQLE-mediated cholesterol biosynthesis contributed to the increased NADP+/NADPH ratio and cell viability (fig. S8A). Because NADPH is essential for redox homeostasis, the exhaustion of NADPH by SQLE resulted in oxidative stress induction (Fig. 6B). To identify the specific metabolite(s) involved in the silencing of PTEN, we treated SQLE-overexpressing HCC and NAFLD-HCC cells with cholesterol, inhibitors of cholesteryl ester biosynthesis (avasimibe), and reactive oxygen species (ROS) inhibitor [glutathione (GSH)]. Exogenous cholesterol increased intracellular cholesteryl ester, and avasimibe decreased intracellular cholesteryl ester content (Fig. 4, C and G), but they both failed to inhibit PTEN expression (fig. S8, B and C). GSH markedly inhibited intracellular ROS (fig. S8D) and reversed SQLE-induced silencing of PTEN (Fig. 6C), suggesting that induction of ROS mediates the effect of SQLE on PTEN silencing. Treatment with GSH also suppressed conversion of cholesterol to cholesteryl esters, abolishing SQLE-induced accumulation of cholesteryl esters while increasing the concentration of free cholesterol (Fig. 6D). These data indicate that SQLE mediates the silencing of PTEN through induction of oxidative stress.

Fig. 6 SQLE silences PTEN through ROS-mediated DNA hypermethylation.

(A) SQLE increased NADP+/NADPH ratio in Sqle tg mice. (B) SQLE increased NADP+/NADPH ratio and ROS in LO2 and HKCI10 cells (n = 3, performed in triplicate). (C) GSH, a ROS scavenger, reversed SQLE-induced PTEN loss, as indicated by Western blot analysis. (D) GSH abolished SQLE-induced accumulation of cholesteryl ester in LO2 and HKCI10 cells (n = 4, performed in triplicate). (E to I) SQLE silenced PTEN via ROS-mediated DNMT3A expression. (E) Heat map of the Epigenetic Chromatin Modification Enzymes PCR Array using LO2 and HKCI2 cell lines overexpressing SQLE. (F) Quantitative PCR and Western blot confirmed that mRNA (top left) and protein (top right) expression of DNMT3A were positively regulated by SQLE. Nuclear DNMT3A activity was also induced by SQLE (bottom) (n = 3, performed in triplicate). (G) GSH reversed SQLE-induced DNMT3A expression. (H) Infinium HumanMethylation450 (450K) BeadChip analysis of CpG methylation in LO2-vector and LO2-SQLE cell lines revealed increased global promoter methylation in SQLE-overexpressing cells (left). Pathway analysis of hyper- and hypomethylated genes (right). (I) SQLE-induced PTEN silencing was reversed by DNMT3A knockdown (n = 3, performed in triplicate). (J) Schematic diagram showing the mechanism of action of SQLE in NAFLD-HCC. SQLE increases cholesterol biosynthesis and NADP+/NADPH-related ROS. Increased ROS induces DNMT3A expression and activation, which mediates transcriptional silencing of PTEN via promoter methylation. PTEN loss activates AKT/mTOR to promote NAFLD-HCC. AKT/mTOR activation also induces cholesteryl ester accumulation, which contributes to tumor cell growth. Data are means ± SEM. Mann-Whitney U test was used for comparing means between two groups. *P < 0.05, **P < 0.01.

Previous studies (17, 18) indicated an important role of oxidative stress in shaping the epigenetic landscape. Given that the expression of PTEN is modulated by several epigenetic regulators (19, 20), we postulated that SQLE may silence PTEN via ROS-induced epigenetic alterations. To identify epigenetic enzymes altered by SQLE, we performed Epigenetic Chromatin Modification Enzymes PCR Array on LO2 and HKCI2 cell lines with empty vector or SQLE. SQLE overexpression in LO2 and HKCI2 cells increased mRNA expression of DNA methyltransferase 3A (DNMT3A) and histone deacetylase 5 (HDAC5) but suppressed HDAC9 (Fig. 6E and table S1). DNMT3A is a de novo DNMT that initiates CpG methylation and is associated with tumorigenesis (21, 22). Quantitative PCR (qPCR) and Western blot confirmed the up-regulation of DNMT3A mRNA and protein in LO2 and HKCI10 cells overexpressing SQLE, whereas the silencing of SQLE in HKCl2 and BEL-7404 cells repressed DNMT3A expression (Fig. 6F). In parallel, nuclear DNMT3A activity was induced by SQLE overexpression in LO2 and HKCI10 cells but was reduced by SQLE knockdown in HKCI2 and BEL-7404 cells (Fig. 6F). To determine whether SQLE-associated oxidative stress mediates induction of DNMT3A, we treated SQLE-expressing cells with GSH. As shown in Fig. 6G, GSH reversed DNMT3A induction by SQLE.

Next, we documented the epigenetic alterations induced by SQLE in LO2 cells using the Illumina Infinium HumanMethylation450K BeadChip. SQLE induced global promoter methylation in LO2 cells (P < 0.0001; Fig. 6H and tables S2 and S3). KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway analysis revealed that hypermethylated genes are associated with cancer, including PI3K/PTEN/AKT and p53 signaling pathways, whereas hypomethylated genes are related to cell cycle and fatty acid metabolism (Fig. 6H). To evaluate whether PTEN was silenced by ROS-induced DNMT3A activation, we treated SQLE-overexpressing cell lines with DNMT3A-specific siRNA. We observed that knockdown of DNMT3A restored PTEN expression, suggesting that DNMT3A-mediated DNA methylation results in transcriptional silencing of PTEN (Fig. 6I). Our findings therefore unveiled an SQLE-ROS-DNMT3A-PTEN axis that contributes to HCC (Fig. 6J).

In line with our in vitro findings, we observed a positive correlation between SQLE and DNMT3A expression in our NAFLD-HCC (R = 0.275, P = 0.037), HCC (R = 0.229, P = 0.039), TCGA liver cancer (R = 0.28, P < 0.0001), and Stanford HCC (R = 0.444, P < 0.0001) cohorts (Fig. 7A), suggesting that SQLE also regulates DNMT3A expression in human HCC. SQLE protein up-regulation was correlated with DNMT3A protein up-regulation, PTEN silencing, and mTOR phosphorylation in human NAFLD-HCC and HCC tissues (Fig. 7B), which was also confirmed in Sqle tg mice (Fig. 7C). Together, these data show that enhanced SQLE expression in vivo activates a DNMT3A/PTEN/mTOR axis that drives NAFLD-HCC tumorigenesis.

Fig. 7 SQLE expression is increased in human HCC and correlates with poor survival.

(A) SQLE mRNA expression in HCC was positively correlated with DNMT3A mRNA expression. The correlation analysis between DNMT3A and SQLE mRNA expression was performed for the NAFLD-HCC cohort (n = 16) and validated in our Guangzhou (n = 69), The Cancer Genome Atlas Liver Hepatocellular Carcinoma (TCGA-LIHC) (n = 423), and Stanford (n = 72) cohorts. Pearson correlation coefficient was used. (B and C) Representative Western blot confirmed that SQLE overexpression up-regulated DNMT3A expression and mTOR phosphorylation and also silenced PTEN expression in (B) NAFLD-HCC and HCC and (C) Sqle-induced mouse HCC. (D) SQLE gene expression in HCC and adjacent normal tissue was determined in our Guangzhou HCC cohort (n = 91 pairs) and (E) validated in TCGA-LIHC (n = 50 pairs) and Stanford cohorts (n = 65 pairs). Paired two-tailed Student’s t tests were used. Data are means ± SEM. (F and G) High SQLE expression correlates with poor survival in HCC. Kaplan-Meier survival analysis and Cox regression analysis of our cohort (high, n = 43; low, n = 45) (G), and TCGA-LIHC (high, n = 155; low, n = 175) cohort based on predictive survival analysis.

SQLE expression is associated with poor survival of HCC patients

We analyzed mRNA expression of SQLE in three independent HCC cohorts (our Guangzhou cohort, TCGA, and Stanford). SQLE was highly up-regulated in primary HCC as compared with adjacent normal tissues (n = 91, P < 0.0001) as determined by qPCR (Fig. 7D), and its overexpression was validated in TCGA (n = 50, P < 0.0001) and Stanford cohorts (n = 65, P < 0.0001) (Fig. 7E). We then assessed the clinical relevance of SQLE in human HCC. Multivariate Cox proportional hazards regression analysis revealed that high SQLE expression was an independent prognostic factor associated with poor disease-specific survival [P < 0.0001; hazard ratio, 4.31; 95% confidence interval (CI), 1.87 to 8.72] (Fig. 7F and table S4). We validated the prognostic relevance of SQLE in TCGA cohort (n = 330). Kaplan-Meier curve showed that increased expression of SQLE mRNA was associated with poor survival in HCC patients (P = 0.02) and was an independent prognostic factor (P = 0.02; hazard ratio, 1.553; 95% CI, 1.042 to 2.314) (Fig. 7G and table S5). These data indicate that SQLE expression is associated with poor prognosis in HCC.

Pharmacological inhibition of SQLE suppressed NAFLD-HCC growth

Given the important oncogenic role of SQLE in NAFLD-HCC, we evaluated whether a specific SQLE inhibitor, terbinafine (used widely to treat fungal infections in humans), can be repositioned for prevention or treatment of NAFLD-HCC. We treated HKCI2 and HKCI10 cells with different doses of terbinafine (25 nM to 50 μM; fig. S9, A to C). At 25 to 50 μM, terbinafine markedly suppressed the viability of HKCI2, HKCI10, and HepG2 cell lines, as determined by cell growth and colony formation assays (Fig. 8A and fig. S9, A and B). Western blot indicated that terbinafine suppressed SQLE and PCNA but restored PTEN expression (Fig. 8B and fig. S9C). Because terbinafine had no effect on mRNA expression of SQLE (fig. S9D), we examined whether terbinafine might affect SQLE protein degradation through ubiquitination or autophagy. HKCI2 and HKCI10 cells were treated with terbinafine plus proteasome inhibitor (MG132) or autophagy inhibitors (chloroquine or bafilomycin A1). Chloroquine (fig. S9E) or bafilomycin A1 (fig. S9F) restored SQLE protein expression, whereas MG132 had no such effect (fig. S9G). In keeping with this, terbinafine induced autophagy activation, as evidenced by reduced p62 protein expression (fig. S9, E and F). Collectively, these data suggest that terbinafine induces SQLE protein degradation by inducing autophagy activity. Furthermore, terbinafine reduced free cholesterol and cholesteryl ester (Fig. 8C). Terbinafine also suppressed the expression and activity of DNMT3A (fig. S9H), in line with the results of SQLE knockdown by RNA interference (Fig. 6F).

Fig. 8 SQLE inhibitor terbinafine suppresses NAFLD-HCC growth in vitro and in vivo.

(A) Terbinafine treatment suppressed cell growth (left) and colony formation (right) in NAFLD-HCC (HKCI2, HKCI10) and HepG2 cell lines (n = 3, performed in triplicate). (B) Terbinafine suppressed SQLE expression and reversed the silencing of PTEN, as determined by Western blot (n = 3, performed in triplicate). (C) Terbinafine reduced the amounts of free cholesterol and cholesteryl ester in HCC cell lines (n = 4, performed in triplicate). (D) Terbinafine (80 mg/kg per day, oral) inhibited growth of subcutaneous HepG2 xenografts, as evidenced by reductions in tumor volume and weight. (E) Terbinafine increased the survival of mice harboring HepG2 xenografts. Kaplan-Meier analysis and log-rank test were used. (F) Terbinafine (80 mg/kg per day, oral) attenuated the growth of orthotopic HKCI2 xenografts. Both tumor volume and weight were reduced. (G) Terbinafine decreased the amounts of free cholesterol and cholesteryl ester in HepG2 xenografts and HKCI2 orthotopic nude mouse model. (H) Terbinafine (80 mg/kg per day, oral) suppressed tumorigenesis in DEN-injected and HFHC diet–treated Sqle tg mice (left), in terms of both tumor incidence and tumor number (right). (I) H&E and Ki-67 staining of vehicle and terbinafine-treated livers. The red arrows show the positive cells. Scale bars, 100 μm (H&E) and 50 μm (Ki-67). (J) Terbinafine treatment decreased liver/body weight ratio (left), liver and serum cholesterol concentrations (middle), and NADP+/NADPH ratio (right). (K) Representative Western blot analysis showed that terbinafine suppressed Sqle expression and reversed the effect of SQLE on downstream factors DNMT3A and PTEN. Data are means ± SEM. The significance of the difference between cell growth rates and tumor growth rates in nude mice was determined by repeated-measures ANOVA. Mann-Whitney U test was used for comparing means between two groups. *P < 0.05, **P < 0.01, ***P < 0.001.

We next evaluated the efficacy of terbinafine in vivo. Terbinafine significantly suppressed the growth of subcutaneous HepG2 xenografts (77.8%, P < 0.01; Fig. 8D). We also assessed the survival of mice harboring HepG2 xenografts (tumor size 400 mm3 as cutoff) and found that terbinafine significantly prolonged the overall survival (P < 0.01; Fig. 8E). Terbinafine also suppressed the growth of orthotopic HKCI2 xenografts (>85%, P < 0.01; Fig. 8F), in terms of both tumor size and tumor weight. In these xenograft models, tumor free cholesterol and cholesteryl ester were suppressed (Fig. 8G). We further validated the efficacy of terbinafine in Sqle tg mice injected with DEN and fed with an HFHC diet (fig. S9I). Terbinafine treatment significantly reduced tumor incidence [four of nine mice in terbinafine group versus eight of nine mice in phosphate-buffered saline (PBS); P < 0.05] and tumor number (P < 0.01; Fig. 8H). H&E staining of livers from the vehicle and terbinafine-treated mice confirmed a reduction in HCC tumorigenesis and cell proliferation by terbinafine (Fig. 8I). In Sqle tg mice, terbinafine decreased liver/body weight ratio and liver and serum cholesterol concentrations (Fig. 8J). In parallel, terbinafine also inhibited NADPH oxidation, thereby reducing the NADP+/NADPH ratio (Fig. 8J). Moreover, terbinafine inhibited SQLE and DNMT3A protein expression but restored PTEN expression in the livers of Sqle tg mice (Fig. 8K). Collectively, these data indicate that terbinafine, by inhibiting SQLE, suppressed the accumulation of liver cholesterol/cholesteryl ester and blocked the SQLE-ROS-DNMT3A-PTEN oncogenic axis, ultimately resulting in inhibition of hepatocarcinogenesis. Moreover, no significant change was found in serum ALT and AST after terbinafine treatment, suggesting that terbinafine did not cause any liver injury or toxicity (fig. S9J). Pharmacological inhibition of SQLE is hence a promising approach that should be safe and effective for the prevention and treatment of NAFLD-HCC.

DISCUSSION

Here, we established SQLE as an oncogenic factor amplified in NAFLD-HCC. SQLE exerts its effect through the action of two key downstream metabolites, cholesteryl ester and NADP+, resulting in epigenetic reprogramming and activating the PTEN/PI3K/AKT/mTOR signaling cascade to drive carcinogenesis in NAFLD-HCC cell lines and in hepatocyte-specific Sqle tg mice. Treatment with terbinafine conferred a therapeutic benefit in NAFLD-HCC, including cell culture and animal models, corroborating SQLE as a therapeutic target in this subset of HCC.

RNA-seq revealed frequent overexpression of SQLE in NAFLD-HCC. SQLE is located on chromosome 8q24.13, a genomic region that is frequently amplified in multiple cancers (2325). We observed that SQLE up-regulation is associated with its gene amplification in NAFLD-HCC. Moreover, in silico and in vitro analyses demonstrated that MEIS1 is a transcription factor that drives SQLE overexpression in NAFLD-HCC. In keeping with our findings (26, 27), SQLE gene amplification and overexpression were observed in multiple NAFLD-HCC and HCC patient cohorts, consistent with an oncogenic role of SQLE in hepatocarcinogenesis.

Our findings reinforced the concept that cholesterol, especially cholesteryl esters, has oncogenic properties (28, 29). Although SQLE is a rate-limiting enzyme for cholesterol biosynthesis, SQLE overexpression caused a more notable rise in cholesteryl esters. Cholesteryl esters were able to induce NAFLD-HCC cell growth (30). Inhibition of cholesterol esterification using avasimibe abolished the growth-promoting effect of SQLE or free cholesterol. Consistent with our data, intracellular cholesteryl esters are associated with increased cell growth and tumor aggressiveness in breast, pancreatic, and prostate cancers (29, 31, 32). Our data thus identify SQLE as a mediator of increased biosynthesis of cholesteryl esters, which is essential for NAFLD-HCC cell growth.

In addition to cholesterol and cholesteryl ester biosynthesis, we found that SQLE consumes reducing equivalents in the form of NADPH, inducing oxidative stress. SQLE-mediated oxidative stress then triggers a sequence of events that profoundly alters the epigenetic landscape. First, oxidative stress stimulates DNMT3A expression and activity, silencing PTEN via promoter methylation. PTEN loss, in turn, activates the PI3K/AKT/mTOR pathway. PI3K/AKT/mTOR promotes the biosynthesis of cholesteryl esters, thereby forming a positive feedback mechanism. Our data imply that the mechanism of action of SQLE is a direct consequence of its catalysis metabolites, which then activate downstream oncogenic pathways.

We validated the oncogenic potential of SQLE in the context of NAFLD-HCC in a model of hepatocyte-specific Sqle tg mice. Sqle tg expression fully recapitulated the signaling cascades induced by ectopic SQLE expression in vitro. Moreover, Sqle tg together with an HFHC diet exacerbated the accumulation of liver cholesterol/cholesteryl ester and induction of oxidative stress, which, in turn, further triggers the downstream oncogenic signaling cascades. As a consequence, Sqle tg expression in mice accelerated HCC development in an experimental model of NAFLD-HCC, concomitant with increased cell growth and the inhibition of apoptosis. Our results provide evidence that Sqle functions as an oncogene in NAFLD-HCC.

HCC patients suffer from poor survival because of a lack of effective treatment options. Identification of additional treatment strategies for HCC is therefore urgent and important. Our work here revealed that terbinafine (SQLE inhibitor) has promising efficacy in inhibiting NAFLD-HCC. Terbinafine is a U.S. Food and Drug Administration–approved oral drug commonly used for the treatment of fungal infection. We found that terbinafine suppressed NAFLD-HCC cell viability in vitro. Using orthotopic NAFLD-HCC xenografts and Sqle tg mice, we showed that terbinafine was efficacious and safe in inhibiting NAFLD-HCC growth in vivo. Terbinafine has an excellent safety profile and relatively few adverse drug-drug interactions (33). Besides blocking cholesterol biosynthesis, targeting of SQLE by terbinafine is known to cause accumulation of squalene (34), which exhibits both antioxidant and anticancer properties in numerous animal models (3537). Nevertheless, clinical testing in humans will be required to validate the beneficial effect of terbinafine in human NAFLD-HCC. Additional work will also be needed to define the mechanism of terbinafine-induced autophagy and its role in hepatocarcinogenesis. Work by us and others (3840) indicated an association of cholesterol with the inhibition of autophagy, suggesting that terbinafine might activate autophagy by suppressing intracellular cholesterol accumulation. The detailed mechanism and implication of terbinafine-induced autophagy will need to be elucidated in future studies.

The impact of our findings was strengthened by the observation that SQLE was consistently and extensively overexpressed in HCC compared with adjacent normal tissues in independent HCC patient cohorts. SQLE expression is an independent prognostic factor associated with poor survival in HCC in our cohort and TCGA data set. SQLE may therefore be useful as a prognostic biomarker for differentiating patients according to their survival outcomes in human HCC.

In conclusion, our discovery of SQLE, an oncogene in NAFLD-HCC, has unraveled links between cholesterol metabolism, epigenetic dysfunction, and hepatocarcinogenesis. Inhibition of SQLE may be a promising therapy for the prevention and treatment of NAFLD-associated HCC.

MATERIALS AND METHODS

Study design

This study was designed to investigate the biological function of SQLE in NAFLD-HCC and assess the therapeutic efficacy of targeting SQLE in NAFLD-HCC. Clinical impact of SQLE was determined in human NAFLD-HCC and HCC cohorts. All human sample collection and study protocol were approved by the Clinical Research Ethics Committee of the Chinese University of Hong Kong (CUHK) or the Sun Yat-sen University of Medical Sciences. Biological function of SQLE was assessed in vitro using NAFLD-HCC cell lines. For in vivo experiments, WT mice and hepatocyte-specific Sqle tg mice were injected with a single dose of DEN and fed an HFHC diet to investigate the oncogenic function of SQLE in NAFLD-HCC development. Therapeutic efficacy of terbinafine, an SQLE inhibitor, was determined in NAFLD-HCC cell lines in vitro and in nude mice and Sqle tg mice in vivo. For in vitro studies, three or four independent experiments were performed in triplicate. For tg mice models, at least 6 to 10 male mice were randomly allocated into different treatment groups. For the nude mouse model, at least five male mice were randomly allocated per group. In vitro experiments were not blinded. In vivo experiments were blinded.

Human samples

Human NAFLD-HCC tumor tissues and adjacent normal tissues were collected from patients with biopsy-proven NAFLD-HCC in Prince of Wales Hospital, CUHK. Human HCC tumors and adjacent normal tissues were collected during operations before any therapeutic intervention at the Third Affiliated Hospital of Sun Yat-sen University (Guangzhou, China). Written informed consent was obtained from all subjects, and the study protocol was approved by the Clinical Research Ethics Committee of the Sun Yat-sen University of Medical Sciences.

Animal model and treatment

Sqle tg mice (pCAG-loxp-stop-loxp-Rosa26-Sqle) were generated by BIOCYTOGEN Company. Sqle-IRES (internal ribosomal entry site)–eGFP (enhanced green fluorescent protein) was cloned into Rosa26 WT allele to generate a gene-targeting vector. Then, the Rosa26-Sqle-IRES-eGFP vectors were transfected into embryonic stem cells with C57BL/6 background. After selection and identification by PCR and Southern blot, positive clones were injected into mouse blastocysts to generate chimeric mice. Chimeric mice were mated with WT C57BL/6 mice to obtain the Rosa26-Sqle mice. To drive the hepatocyte-specific expression of Sqle, Rosa26-Sqle mice were crossed to B6.Cg-Tg (Alb-Cre) 21Mgn/JNju mice (Nanjing University). Sqle tg/Alb-Cre mice were confirmed by PCR genotyping.

An orthotopic human NAFLD-HCC mouse model was established using HKCI2 cells (11, 12). HKCI2 cells (1 × 107 cells in 0.1 ml of PBS) were injected subcutaneously into the left dorsal flank of 4-week-old male Balb/c nude mice. Subcutaneous tumors were harvested once they reached about 10 mm3 and cut into 1.0 mm3 pieces. One piece of a tumor was implanted into the left liver lobe in a separate group of nude mice (4 weeks old). Four weeks after implantation, these mice were divided into vehicle group (PBS, oral) and terbinafine group (80 mg/kg, oral). Eight weeks after implantation, the mice were sacrificed and examined. All animal studies were performed in accordance with the guidelines approved by the Animal Experimentation Ethics Committee of CUHK.

ChIP assay

A total of 1 × 107 LO2 cells stably transfected with Meis1-DYK (OHu19435) were cross-linked with 1% formaldehyde for 10 min at room temperature and quenched with 125 mM glycine. After sonication, protein-DNA complexes were immunoprecipitated (IP) by 2 μg of anti-DYK–tag antibody (no. 14793, Cell Signaling Technology) or anti-rabbit immunoglobulin G antibody (Abcam) overnight with rotation at 4°C, followed by the addition of 20 μl of Dynabeads magnetic beads (Millipore) and incubation for 2 hours. Next, we washed the antibody/chromatin complex by resuspending the beads in four immune complex wash buffers (Millipore) with different salt concentrations (low salt, high salt, LiCl, and tris-EDTA buffer) step by step. Then, we added 8 μl of 5 M NaCl buffer to each IP and input sample and incubated at 65°C overnight to reverse the DNA protein cross-links. After elution, all samples were treated with 1 μl of ribonuclease A at 37°C for 30 min and incubated with 4 μl of 0.5 M EDTA, 8 μl of 1 M tris-HCl, and 1 μl of proteinase K at 45°C for 1 to 2 hours. Finally, the IP and input DNA were purified using spin columns (Millipore). For target gene validation, PCR primers targeting a region of the putative binding site were designed to detect IP and input DNA. The sequences of primers used are listed in table S6.

Plasmids

pRL-cyto-megalovirus (pCMV)–SQLE (RC202008) and pCMV-entry control plasmids, SQLE shRNA (TL309122V), and control shRNA (pGFP-C-shlenti) plasmids were all ordered from OriGene. SQLE promoter reporter clone (HPRM22529-OG04) was ordered from GeneCopoeia. Meis1 (OHu19435) and control plasmids (pcDNA3.1-DYK) were ordered from GenScript.

RNA interference

PTEN siRNA (siPTEN), DNMT3A siRNA (siDNMT3A), and negative control (siControl) were ordered form Santa Cruz Company. siPTEN, siDNMT3A, or siControl (50 nmol) was transfected into cells using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions.

RNA extraction, semiquantitative reverse transcription PCR, and real-time PCR analyses

Total RNA was extracted from cells and tissues using TRIzol Reagent (Molecular Research Center Inc.). Complementary DNA (cDNA) was synthesized from 1 μg of total RNA using Transcriptor Reverse Transcriptase (Roche). Real-time PCR was performed using an SYBR Green master mixture (Roche) on LightCycler 480 Instrument. Each sample was tested in triplicate. ΔΔCt method was applied to determine the fold change in gene expression. ΔCt method was applied to determine the relative expression of corresponding genes.

Western blot analysis

Total protein was separated by SDS–polyacrylamide gel electrophoresis (SDS-PAGE). The proteins in SDS-PAGE were transferred onto nitrocellulose membranes (GE Healthcare). The membrane was incubated with primary antibodies overnight at 4°C (table S7) and then with secondary antibody at room temperature for 1 hour. Proteins of interest were visualized using ECL Plus Western blotting Detection Reagents (GE Healthcare).

Colony formation assay

For the cell colony formation assay, stably transfected cells (1000 per well) were plated in six-well plates. After culturing for 5 to 7 days, cells were fixed with 70% ethanol and stained with 0.5% crystal violet solution. Colonies with more than 50 cells per colony were counted. All experiments were conducted three times in triplicate.

Apoptosis and cell cycle analyses

Apoptosis was assessed using an annexin-phycoerythrin/7-aminoactinomycin D staining kit (BD Biosciences). For cell cycle analysis, cells were serum-starved overnight and stimulated with complete medium for 4 to 8 hours. Cells were fixed in 70% ethanol, stained with propidium iodide, and analyzed by flow cytometry.

Luciferase reporter assay

To check the interaction between MEIS1 and SQLE, SQLE promoter reporter clone (SQLE 1200-GLuc, GeneCopoeia), which contains the promoter of human SQLE 1.2 kb upstream of the transcriptional start site, was transfected into MEIS1-transfected LO2 and HKCI2 cells. To investigate the signaling pathways modulated by SQLE, a series of signaling pathway luciferase reporters were examined in SQLE-transfected LO2 and HKCI10 cells, including p53-luc, AP1-luc, WNT-luc, and FOXO3-luc.

The cell lines (LO2 and HKCI10) were stably transfected with pCMV-SQLE or pCMV-vector (1 × 105 cells per well) in 24-well plates and cotransfected with luciferase reporter plasmid (0.2 μg per well) and pCMV-vector (5 ng per well) using Lipofectamine 2000 (Life Technologies). Cells were harvested 48 hours after transfection, and luciferase activity was analyzed by the Dual Luciferase Reporter Assay System (Promega).

Cholesterol/cholesteryl ester concentrations

Cells (106) or tissues (2 mg) were harvested, and cholesterol/cholesteryl ester concentrations were detected by Cholesterol/Cholesteryl Ester Quantification kit (ab65359, Abcam) according to the manufacturer’s instructions. All experiments were conducted three times in triplicate. Results were shown as means ± SEM.

NADP+/NADPH ratio

Cells (2 × 106) or tissues (50 mg) were harvested, and NADPH/NADP+ ratio was quantified by the NADP/NADPH Assay Kit (ab65349, Abcam) according to the manufacturer’s instructions. All experiments were conducted three times in triplicate. Results were shown as means ± SEM.

Reactive oxygen species

Intracellular ROS were quantified by the DCFDA Cellular ROS Detection Assay Kit (microplate) (ab113851, Abcam) according to the manufacturer’s instructions. All experiments were conducted three times in triplicate. Results were shown as means ± SEM.

Serum cholesterol, ALT, and AST

The serum cholesterol, ALT, and AST concentrations were detected by the Catalyst One Chemistry Analyzer according to the manufacturer’s instructions (IDEXX). Thirty microliters of serum from WT or Sqle tg mice was diluted to 90 μl by physiological saline buffer. The diluted samples and specific slides (cholesterol, ALT, and AST) were then loaded into the analyzer for automatic analysis.

Serum AFP

Mouse serum AFP was detected by the mouse AFP/AFP ELISA Kit according to the manufacturer’s instructions (MAFO00, R&D Systems). Serum (10 μl) from WT or Sqle tg mice was diluted to 200 μl with Calibrator Diluent RD5-26 buffer (diluted 1:4). We then added 50 μl of diluted samples, standard, and control to each microplate well for further analysis.

Ki-67 staining

Paraffin slides from DEN-injected, HFHC-fed WT and Sqle tg mice were used. Ki-67 signal was assessed by an anti–Ki-67 antibody (ab833, Abcam). The proliferation index was determined by counting the numbers of positive staining cells as percentages of the total number of liver cells. At least 1000 cells were counted each time.

TUNEL staining

Paraffin slides from DEN-injected, HFHC-fed WT and Sqle tg mice were used. TUNEL signal was assessed using the DeadEnd Colorimetric TUNEL System (Promega). Briefly, we removed paraffin by xylene and fixed slides in 4% paraformaldehyde in PBS. We then added 100 μl of TdT reaction mix to the tissue sections on the slides and incubated the slides for 60 min at 37°C in a humidified chamber. After incubation with 100 μl of streptavidin horseradish peroxidase (diluted 1:500 in PBS), 100 μl of 3,3′-diaminobenzidine was added to the slides for detection.

Statistical analysis

All statistical tests were performed using SPSS or GraphPad Software. Data are presented as means ± SEM. The Pearson correlation coefficient was used to evaluate the correlation between SQLE gene amplification and expression in the clinical samples. Multiple group comparisons were analyzed by one-way ANOVA. Overall survival in relation to expression was evaluated by the Kaplan-Meier survival curve and the log-rank test. Mann-Whitney U test or Student’s t test was performed to compare the variables in two groups. The difference in cell viability and tumor growth rate between the two groups of nude mice was determined by repeated-measures ANOVA. P values of <0.05 were taken as being statistically significant.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/10/437/eaap9840/DC1

Fig. S1. SQLE expression in NAFLD-HCC is controlled by transcription factors.

Fig. S2. Transcription factors SREBP2 and MEIS1 bind to SQLE promoter region and activate SQLE gene expression.

Fig. S3. SQLE promotes HCC cell growth.

Fig. S4. SQLE in HCC cell lines promoted cell cycle progression at G1-S phase and inhibited apoptosis induction.

Fig. S5. Sqle tg expression in mice induced apoptosis and the expression of proinflammatory cytokines and chemokines.

Fig. S6. Cholesteryl ester and cholesterol concentrations are increased in NAFLD-HCC.

Fig. S7. Oncogenic function of SQLE is dependent on the PTEN/PI3K/AKT/mTOR pathway.

Fig. S8. SQLE silences PTEN expression through ROS-mediated DNA hypermethylation.

Fig. S9. SQLE inhibitor terbinafine suppressed NAFLD-HCC growth in vitro and in vivo.

Table S1. Epigenetic Chromatin Modification Enzymes PCR Array data.

Table S2. Hypermethylated genes in LO2-SQLE cells.

Table S3. Hypomethylated genes in LO2-SQLE cells.

Table S4. Clinicopathological features of SQLE mRNA expression in our HCC cohort.

Table S5. Clinicopathological features of SQLE mRNA expression in TCGA HCC cohort.

Table S6. Primers used in this study.

Table S7. Antibodies used in this study.

REFERENCES AND NOTES

Funding: This study was supported by the Research Grants Council (RGC)–General Research Fund Hong Kong (14106415, 14101917, 14111216, and 14114615), Shenzhen Virtual University Park Support Scheme to CUHK Shenzhen Research Institute, grant from Faculty of Medicine CUHK, and direct grant from CUHK. Author contributions: D.L. was involved in the study design, conducted the experiments, and drafted the paper; H.C., L.Z., C.L., and Y. Zhou performed animal experiments; Y. Zhang and W.X. performed the experiments; Y.Y., B.W., G.C., and P.B.-S.L. collected human samples; N.W. performed human NAFLD-HCC genome sequencing and provided NAFLD-HCC cell lines; L.F. and J.J.Y.S. designed and commented on the study; C.C.W. drafted the paper and commented on the study; J.Y. designed and supervised the study and critically revised the paper. Competing interests: The authors declare that they have no competing interests. Data and materials availability: The RNA-seq data have been deposited into the European Genome-Phenome Archive and are available at accession number EGAD00001004326.
View Abstract

Navigate This Article