Sirt1 deficiency upregulates glutathione metabolism to prevent hepatocellular carcinoma initiation in mice

Abstract

Sirtuin-1 (SIRT1) is involved in various metabolic pathways, including fatty acid synthesis and gluconeogenesis in the liver. However, its role in initiation and progression of liver cancer remains unclear. Studying Sirt1 liver-specific knockout (LKO) mice in combination with diethylnitrosamine (DEN) treatment, we demonstrated that loss of Sirt1 rendered mice resistant to DEN-induced hepatocellular carcinoma (HCC) development. RNA-seq revealed that livers from LKO mice exhibited an enrichment in glutathione metabolism eight months after DEN challenge. Sirt1 deficiency elevated the expression of glutathione-s-transferase family genes by increasing the level of Nrf2, a key regulator of glutathione metabolism. Hence, LKO livers displayed a reductive environment with an increased ratio of GSH to GSSG and an elevated GSH level. Furthermore, using CRISPR knockout techniques, we confirmed that the impairment of HCC formation in LKO mice is mainly dependent on NRF2 signaling. Meanwhile, HCC induced by DEN could be blocked by the administration of N-acetyl cysteine (NAC) when administered one month after DEN challenge. However, NAC treatment starting five months after DEN injection was not able to prevent tumor development. In conclusion, our findings indicate that a reductive environment orchestrated by glutathione metabolism at an early stage can prevent the initiation of HCC.

INTRODUCTION

Sirtuin-1(SIRT1) is one of seven sirtuin family members which are NAD⁺-dependent class III histone deacetylases (HDAC). SIRT1 plays a role in orchestrating energy metabolism and cellular stress [1, 2]. Initially SIRT1 was reported to exert pro-tumorigenic effects in cancer biology by regulating different factors such as p53, E2F1, and Rb [3–7]. Additional investigations have underscored a tumor suppressor role for SIRT1. For example, heterozygous loss of Sirt1 and p53 simultaneously initiates cancer formation in multiple organs, suggesting that SIRT1 is essential for maintaining genome stability and preventing tumorigenesis [8]. Sirt1 also suppresses Brca1 mutant breast tumor growth by inhibiting expression of survivin [9]. Further more, SIRT1 directly interacts with RelA/p65, a subunit of NF-κB, and reduces expression of the anti-apoptotic protein, cIAP-2 [10].

Currently, liver cancer is the fifth most diagnosed cancer, and the second leading cause of cancer deaths worldwide [11]. Hepatocellular carcinoma (HCC) accounts for 75–85% of primary liver malignancies [12]. Hepatic resection and liver transplantation are effective therapeutic approaches for early stage HCC. Multityrosine kinase inhibitors targeting vascular endothelial growth factor receptor, epidermal growth factor receptor 2, and platelet-derived growth factor are first line drugs for patients with advanced disease, improving median overall survival rates by 4–8 months [13]. Immunotherapy alone or in combination with chemotherapy in HCC has been evaluated in clinical trials for several years. The FDA has approved the use of nivolumab, a human programmed death receptor-1 (PD-1) blocking antibody, for HCC patients following prior approval of sorafenib in 2017. In 2020 FDA approved the combination of, nivolumab and the anti-CTLA-4 monoclonal antibody, ipilimumab for treatment of advanced HCC refractory to sorafenib. However, despite surgery and available drugs, the 5-year survival rate for HCC is only 18% [14].

In recent years, Western countries have started to see increasing HCC cases which are related to liver metabolism dysfunction, such as non-alcoholic or alcoholic fatty liver diseases and non-alcoholic steatohepatitis. These metabolic disorders gradually change the liver’s microenvironment and exert high risk for HCC development [15]. Preventing tumor initiation is an important issue for patients with liver diseases and is a growing study topic in the liver cancer field; however, the mechanisms mediating initiation and progression of HCC in metabolic abnormal conditions have not been fully elucidated.

Mounting evidence implicates high levels of ROS and overexpression of antioxidant enzymes in the pathogenesis of in a variety of cancers [16]. At the tumor initiation stage, antioxidants detoxify intracellular ROS to prevent damages to nucleic acids and proteins, while these can promote drug efflux from tumor cells during treatment. Reductive GSH detoxifies foreign compounds to protect DNA from damage. As a major force against ROS, glutathione and its activator NAC exhibit antitumor effects in multiple cancers [17, 18]. At the same time, glutathione and NAC were demonstrated to promote growth of other cancers [19].

Herein we demonstrate that Sirt1 liver specific knockout (LKO) inhibits HCC initiation by upregulating the glutathione metabolic pathway. Increased glutathione metabolism is supported by enhanced expression and nuclear translocation of NRF2 in the absence of Sirt1. We also demonstrate that liver specific knockout of Nrf2 increases HCC tumor formation in LKO mice exposed to DEN. Although not inducing Nrf2 expression, NAC treatment in the early stage of DEN insult inhibits HCC development; however, NAC administered at a later stage has no such effect.

RESULTS

LKO mice resists HCC formation after DEN challenge

To investigate if Sirt1 plays a role in HCC formation, we administered 25 mg/kg DEN, a well-established liver carcinogen, to WT and LKO mice on postnatal day 14 with an intraperitoneal injection (Fig. 1A). SIRT1 knockout efficiency in liver was verified by Western blot (Supplementary Fig. 1A). We analyzed liver pathology at months five and six. Five months after DEN injection, 7 out of 10 WT mice had developed hyperplasia or small nodules in their livers, but no LKO mice displayed such histologic abnormalities (N = 10) (Fig. 1B). At 6 months, we observed that 8 WT mice (N = 10) and 4 LKO mice (N = 10) had nodules on the surfaces of their livers (Fig. 1B). When analyzed at 8 months following DEN injection, LKO mice had macroscopic liver nodules that were smaller in numbers and sizes relative to WT mice (Fig. 1C). Liver indices, measured as ratios of livers to body weights, were significantly decreased in LKO mice (Fig. 1D). Consistent with the first cohort, our second experimental cohort indicated that DEN-induced tumors were significantly more numerous and larger in WT relative to LKO mice (Supplementary Fig. 1B, C). Histologically, LKO tumors resembled WT tumors (Fig. 1E).

Fig. 1. Sirt1 LKO mice resist HCC formation after DEN challenge.

A Scheme of DEN injection in mice. B Images show surface nodules in WT and LKO mice 5 and 6 months after DEN. WT, n = 10; Sirt1 LKO, n = 10 for both 5- and 6-months DEN treatment. C Sirt1 LKO mice displayed less and smaller nodules. WT and Sirt1 LKO livers after 8 months of DEN challenge. Macroscopic nodules from each group were counted for each liver. WT, n = 21; Sirt1 LKO, n = 15. Student’s t est, ***p < 0.001. D The ratio of liver weight to body weight was lower in Sirt1 LKO group compared to the WT group after 8 months of DEN treatment. Values are mean ± standard deviation, WT, n = 21; Sirt1 LKO, n = 15. Statistical analyses performed using Student’s t test. *p < 0.05. E Representative hematoxylin and eosin (H&E) image of WT and Sirt1 LKO liver with and without DEN. N = 3 for each group. Magnification scale bar, 100 μm. F, G Quantitative RT-PCR of Afp and Gpc3 comparing adjacent liver and nodule. N = 3 for each liver and nodule. Statistical analyses performed using Student’s t test. ***p < 0.001. H Immunofluorescent staining of Gpc3 in adjacent and nodule to determine cancer subtype in tumor sites. Magnification scale bar, 50 μm.

We analyzed the liver nodules for α fetoprotein (Afp), and Glypican 3 (Gpc3), which are typically elevated in HCC. Quantitative RT-PCR verified high expression of these genes in tumors from both WT and LKO mice (Fig. 1F, G). As expected, GPC3 was positively stained in the tumors from both genotypes of mice as well (Fig. 1H). Altogether, this data revealed that LKO mice exhibited resistance to the initiation of DEN-induced HCC in mice.

SIRT1 deficiency reduces cell proliferation after DEN treatment

To understand what makes LKO mice resistant to hepatocellular carcinogenesis, we examined whether SIRT1 altered hepatocyte response to DEN treatment. Under normal conditions, DEN is activated inside cells by Cyp2E1 [20, 21]. To examine whether SIRT1 deficiency affects DEN activation, we measured the expression of Cyp2e1 and found that there was no significant change in LKO liver (Fig. 2A), suggesting that the loss of SIRT1 does not affect DEN activation.

Fig. 2. Sirt1 deficiency reduces cell proliferation after DEN treatment.

A Cyp2e1 expression in WT and Sirt1 LKO mice after DEN 24 h treatment. All mice are 14 days old. 3 mice for each group. Statistical analyses performed using Student’s t test. ns no significance. B Immunoblot analysis of γH2AX expression in the indicated DEN treated time points in WT and Sirt1 LKO mice. N = 3 in each group. C γH2AX expression was higher in WT livers than in Sirt1 LKO livers at 5 days post DEN challenge, as determined by the Western blot band intensity. ns no significance; *p < 0.05; **p < 0.01. D Representative images of Phospho-Histone H3 Serine 10 (pH3) in DEN treated WT and Sirt1 LKO mice, as indicated different timepoints showed. E Quantification of pH3 positive cells was shown as mean ± SD. (n > 9 different fields from three different mice for each time point). One-way ANOVA was used, ns no significance; **p < 0.01; ***p < 0.001. F IHC analysis of γH2AX expression on liver section at the indicated month after DEN challenge in WT and Sirt1 LKO mice. N = 7 for WT and Sirt1 LKO in 5 months; N = 4 for WT and Sirt1 LKO in 6 months; N = 5 for WT and Sirt1 LKO in 8 months. One-way ANOVA was used, *p < 0.05.

SIRT1 has been shown to relate directly to DNA damage response [22] and can promote double-stranded break repair by homologous recombination and non-homologous end joining [23–25]. Because DEN treatment leads to damage within the genome [20], we hypothesized that hepatocytes lacking the SIRT1 protein may experience more DNA damage than their WT counterpart. We sought to analyze the extent of DNA damage in hepatocytes of WT and LKO mice as evidenced by H2AX phosphorylation. To our surprise, while induction of H2AX phosphorylation was not significantly different at 3 days post DEN injection, WT mice exhibited significantly more γH2AX protein compared to their LKO counterparts (Fig. 2B, C) by day 5, suggesting LKO mice harbor less DNA damage induced by DEN in hepatocytes as time goes on.

It has been reported that hepatocyte injury or death triggers compensatory proliferation, contributing to liver tumorigenesis [26]. In the absence of DEN treatment, there was no significant difference in mitotic index (proliferation status marker) in WT and LKO mice (Fig. 2D, E). However, mitotic index was markedly lower in the livers of LKO mice compared to WT mice following DEN exposure (Fig. 2E), raising the possibility that differences in proliferation rates could account at least in part for subsequent HCC development. In vivo γH2AX level was also studied along a timeline post DEN treatment (Fig. 2F). IHC staining for γH2AX revealed that WT mice livers contained higher γH2AX phosphorylation level than LKO mice liver at 5, 6, and 8 months post DEN injection (Fig. 2F).

It has been reported that inflammation playes a key role in HCC formation. We speculated that the absence of Sirt1 might reduce hepatic inflammation and expression of pro-inflammatory cytokines, such as IL1β, IL-6, and TNFα, that have been shown to promote hepatocellular carcinogenesis [27, 28]. We examined mRNA levels of F4/80, Il1β and Tnfα and protein levels of IL-6 in the livers of WT and LKO mice, and found that F4/80 and Tnfα mRNA levels as well as IL-6 protein levels were significantly higher in LKO mice livers compared with WT mice livers (Supplementary Fig. 2A, B). Increased hepatic pro-tumorigenic inflammatory signals appeared inconsistent with decreased liver tumor burden in LKO mice.

DEN-induced IL-6 secretion by Kupffer cells promotes abnormal compensatory proliferation of surviving hepatocytes that initiate HCC [27]. To examine inflammatory responses of LKO and WT mice in the presence of DEN treatment, we examined hepatic tissues at 24 and 48 h post DEN treatment. DEN induced inflammatory cytokine expression in WT livers at 24 h (Supplementary Fig. 2A). There was no change in F4/80 and Il1β mRNA in both genotypes (Supplementary Fig. 2A). No differences in IL-6 protein levels were discernable between WT and LKO mice livers at 24 or 48 h post DEN treatment (Supplementary Fig. 2B). At one, four, and eight months after DEN treatment, there were no significant alterations of F4/80, Il1β,Tnfα and Il6 mRNA levels between WT and LKO mice (Supplementary Fig. 2C). These results suggest that LKO and WT mice have similar inflammatory responses to DEN; hence, inflammation responses caused by DEN cannot account for the different tumor burdens in WT and LKO mice.

SIRT1 deficiency enhances glutathione metabolism in the liver with DEN treatment

To further identify signaling pathways responsible for delayed tumor occurrence in LKO mice, we performed RNA-Seq on tumors and adjacent liver tissues from mice treated with DEN for eight months as well as hepatocyte tissues from age-matched mock treated WT and LKO mice. Consistent with our previous study [29], metabolic pathways including lipogenesis, were upregulated in untreated LKO mice (Fig. 3A). In addition, LKO mice exhibited increased expression of genes regulating metabolism of xenobiotics and drugs, indicating a higher detoxification effect in the absence of hepatic Sirt1. Glutathione metabolism, which mediates the degradation of xenobiotics and maintains an extracellular redox environment, was also upregulated in LKO mice. Expression levels of genes regulating glutathione metabolism, degradation of xenobiotics and drugs, and cytochrome p450 activity were higher in normal liver tissues from LKO relative to WT mice (Fig. 3A). However, once the RNA-Seq data from LKO and WT tumors were compared, differential modulation of these pathways was no longer evident (Fig. 3A, B).

Fig. 3. Sirt1 deficiency enhances glutathione metabolism in liver with DEN treatment.

A Pathway enrichment analysis showed significantly upregulated (in red box) or downregulated (in blue box) pathways. Control liver means Sirt1 LKO versus WT liver without DEN. DEN liver means Sirt1 LKO versus WT adjacent liver with DEN treatment. DEN tumor means Sirt1 LKO versus WT tumor with DEN treatment. N = 3 mice for each treatment group. B Left, GSEA analysis showed that glutathione metabolism was upregulated in Sirt1 LKO and DEN treated liver but not in tumor, comparing WT counterparts. Right, Gene expression patterns of glutathione metabolism related genes. The comparisons are same as Fig. 3A. C Validation of Gsta1, Gsta2, and Gstm4 expression in Sirt1 LKO liver. *p < 0.05. D Determination of the GSH/GSSG ratio (left) and the reduced GSH (right) in WT and Sirt1 LKO liver. Values are mean ± SD. *p < 0.05. E Genes of Gstm family are upregulated in Sirt1 LKO adjacent liver with DEN treatment. *p < 0.05. F Detection of glutathione de novo synthesis, Gclc, expression in Sirt1 LKO adjacent liver with DEN treatment. ***p < 0.001. G Determination of GSTA1 protein level in WT and Sirt1 LKO liver. N = 3 mice for each group. H Western blot analysis of GSTA1, GSTM1, GSTM2, and GSTM3 in WT and Sirt1 LKO adjacent liver and tumor. L, adjacent liver; T, tumor. I Determination of the GSH/GSSG ratio (up) and the reduced GSH (down) in DEN treated WT and Sirt1 LKO livers and tumors. Values are mean ± SD. *p < 0.05; ns no significance.

Since genes that are involved in glutathione metabolism also participate in the metabolism of xenobiotics, drugs, and activity of cytochrome p450, we decided to further investigate expression patterns of genes regulating glutathione metabolism. Many glutathione-s-transferases were elevated in LKO livers and LKO/DEN treated adjacent liver tissues relative to WT controls, but this phenotype disappeared in tumors caused by DEN (Fig. 3B). Increased gene expression levels of glutathione-s-transferase (Gst) a1, a2, and m4 were validated by qRT-PCR (Fig. 3C). Consequently, livers from LKO mice had higher reduced glutathione (GSH)/oxidized glutathione (GSSG) ratios as well as higher levels of reduced GSH than livers from WT mice (Fig. 3D). These observations indicate that LKO livers remain in a more reductive status compared to WT livers.

We next sought to investigate the impact of DEN on this reductive environment. As shown in Fig. 3E, in the presence of DEN for eight months, “normal” liver tissues of LKO mice still exhibited significantly increased expression of Gstm1, 2, 3, 4, and 5 relative to WT mice (Fig. 3E). Thus, we postulated that GSH synthesis might be augmented in DEN-treated LKO mice. Accordingly, we investigated expression of Gclc-the gene responsible for GSH synthesis. Following DEN treatment, Gclc was upregulated in LKO mice (Fig. 3F), suggesting that livers from LKO mice have a greater ability to maintain a cellular redox environment. We also found that Gsta1 as well as Gstm1, 2, and 3 expression levels were elevated in LKO livers 3 days after DEN challenge (Supplementary Fig. 3A). Sirt1 loss increased intrahepatic GSH/GSSG levels (Supplementary Fig. 3B). One month after DEN exposure, Gsta1 and Gstm1, 2, and 3 expression levels were still significantly higher in livers from LKO mice (Supplementary Fig. 3C). We also examined protein expression of these family members in tumors and adjacent liver tissues. GSTA1 protein levels were significantly higher in livers from LKO mice regardless of DEN exposure (Fig. 3G); however, GSTA1, GSTM1, GSTM2, and GSTM3 protein levels were significantly reduced in both WT and LKO tumors (Fig. 3H). Finally, GSH/GSSG ratios were also measured in DEN treated livers and tumors. As shown in Fig. 3I, GSH/GSSG levels and total GSH were higher in DEN treated livers from LKO mice compared to WT mice. But there was no significant difference between LKO and WT tumors. Based on these findings, we hypothesized that GST family proteins play a role in DEN induced HCC initiation.

Increased NRF2 activity in LKO regulates glutathione genes in DEN treatment

Nuclear factor-erythroid 2-related factor 2 (NRF2, also known as NFE2L2) is the master regulator of oxidative stress. NRF2 heterodimerizes with MafG and binds to antioxidant response elements (AREs) to initiate the transcription of genes in the glutathione metabolism pathway [30]. Since we observed elevated expression of genes regulating the glutathione antioxidant system in LKO liver tissues, we examined the levels of NRF2 expression in WT and LKO primary hepatocytes. We found elevated Nrf2 and Nqo1 mRNA levels in LKO livers 3 days after DEN treatment (Fig. 4A). Primary hepatocytes were isolated from Sirt1 flox/flox male mice and then infected with Ad-Cre to knockout Sirt1 in cell culture. Knock-out of Sirt1 increased Nrf2 and Nqo-1 mRNA levels (Fig. 4B). Accordingly, total NRF2 protein was also elevated in SIRT1 deficient hepatocytes (Fig. 4C). To examine the cellular localization of NRF2, we isolated hepatocytes from WT and LKO mice and detected protein levels by Western blot in either the cytoplasmic or the nuclear fractions. Whereas cytoplasmic levels of NRF2 were comparable in WT and LKO hepatocytes, nuclear NRF2 protein levels were significantly increased in Sirt1 depleted hepatocytes (Fig. 4D). This effect of SIRT1 was also analyzed in the immortalized human hepatocyte cell line, MIHA cell using shRNA to knockdown SIRT1. Immunofluorescent staining showed that nuclear levels of NRF2 were elevated upon SIRT1 knockdown in MIHA cells (Fig. 4E). Western blot analysis demonstrated that knocking down SIRT1 increased NRF2 protein levels in total cell lysates and the nuclear compartment, without changing cytoplasmic NRF2 levels (Supplementary Fig. 4A).

Fig. 4. Increased Nrf2 activity in LKO regulates glutathione genes in DEN treatment.

A Quantification of Nrf2 and Nqo1 mRNA expression 3 days after DEN challenge in 14 days old mice. *p < 0.05; **p < 0.01. B In vitro Sirt1 knockout in primary hepatocytes increased Nrf2 and Nqo1 mRNA and protein level. **p < 0.01. C Sirt1 LKO liver has elevated total NRF2 and; D nucleus protein level. E Left, immunofluorescence analysis of NRF2 in MIHA cells with silencing of SIRT1 expression. (magnification scale bar, 10 μm). Right, quantification of NRF2 immunofluorescent intensity in nucleus. **p < 0.01. F SIRT1 overexpression reduced HO-1 ARE luciferase activity in L02 cells. *p < 0.05; ***p < 0.001. G Left, SIRT1 knockdown enhanced ARE luciferase activity in L02 cells with and without 48 h DEN treatment. **p < 0.01. Right, western blot analysis showed SIRT1 knockdown efficiency in L02 cells. H Detection of glutathione gene expression in Sirt1 Flox and Cre. *p < 0.05; **p < 0.01. I Validation of Nrf2 knockdown in Sirt1 knockout hepatocytes. J Decreased Gsta1, Gstm1, Gstm2, and Gstm4 mRNA after Nrf2 shRNA infection in Sirt1 null hepatocytes. **p < 0.01; ***p < 0.001. K Protein level of GSTM2, NRF2, and SIRT1 in Sirt1 knockout and Nrf2 knockdown hepatocytes. L Downregulation of Nrf2 increased γH2AX in Sirt1 null hepatocytes with and without 100 μM DEN treatment for 24 h. M Nrf2 protein level in WT and Sirt1 LKO liver and tumor at DEN 5 m, 6 m, and 8 m. 3 mice for each indicated groups.

To examine if SIRT1 impacts downstream transcriptional activities of NRF2, we measured the activation of the HO-1 ARE promoter, a well-known target of NRF2 [31]. Briefly, NRF2 transcriptional activity was measured using a HO-1 ARE promoter luciferase vector in the context of SIRT1 overexpression and knockdown. Cells with SIRT1 overexpression exhibited decreased luciferase activity (Fig. 4F), whereas cells with SIRT1 knockdown displayed a greater HO-1 ARE luciferase activity (Fig. 4G), indicating that SIRT1 reduces NRF2 regulated transcription. Nevertheless, Nqo1 mRNA levels were elevated upon Sirt1 knockout in primary hepatocytes (Fig. 4B). Because loss of SIRT1 altered expression of NRF2 target genes, we wanted to know specifically whether glutathione pathway genes were regulated by Nrf2 in the absence of Sirt1. We examined expression of Gstm1, Gstm2, Gstm3, Gstm4, and Gsta1 following Sirt1 Cre knockout (Fig. 4H). Expression levels of all these genes were significantly elevated in Sirt1 knockout cells relative to WT cells. Given the increased levels of Nrf2 protein upon Sirt1 depletion, we postulated that the glutathione pathway genes were induced by NRF2 promotion in SIRT1 deficient hepatocytes. shRNA knockdown of Nrf2 in Sirt1 knockout cells reduced Gsta1, Gstm1 and Gstm2 mRNA and protein levels (Fig. 4I–K). Altogether, these findings indicate that elevated levels of NRF2 in SIRT1 knockout cells increases transcription of glutathione pathway genes.

Previously, it was shown that NRF2 may facilitate the repair of radiation induced DNA damage [32]. Loss of NRF2 increases DNA damage by impairing the ATM and ATR DNA repair pathways [33, 34]. Thus, we reasoned that perhaps elevated levels of NRF2 in SIRT1 knockout cells might have a protective effect against DNA damage. To examine this issue, we performed Western blot analysis of γH2AX in lysates from primary hepatocytes cultured in the presence or absence of DEN (Fig. 4L). While Sirt1 knockout appeared to have little effect, Nrf2 knockdown in Sirt1 deficient hepatocytes significantly increased γH2AX levels (Fig. 4L). DEN challenge further increased γH2AX levels in both WT and Sirt1 knockout hepatocytes. However, in the context of DEN treatment, Sirt1 knockout cells displayed relatively lower levels of γH2AX, indicating that SIRT1 loss itself reduces DNA damage upon DEN treatment (Fig. 4L). Further knockdown of Nrf2 in DEN-treated Sirt1 deficient hepatocytes increased γH2AX levels (Fig. 4L), suggesting that the lower γH2AX levels in LKO cells upon DEN insult were mediated by increased NRF2 expression. Sirt1 and Nrf2 relationship in terms of HCC formation upon DEN challenge was also investigated in vivo (Fig. 4M). At 5 months post DEN treatment, LKO liver displayed higher Nrf2 protein level than WT liver and WT liver tumor; at 6 and 8 months, LKO liver showed elevated Nrf2 protein level than WT liver, WT liver tumor and LKO liver tumor, suggesting maintaining higher Nrf2 level is a key for preventing HCC, and Sirt1 directly affects Nrf2 level by a negative regulation. Elevated NRF2 expression increased the reductive environment by upregulating glutathione metabolism to protect DNA damage in SIRT1 negative hepatocytes. Thus, Sirt1 deficiency induced Nrf2 expression and translocation of NRF2 into the nucleus. This effect induces expression of glutathione pathway genes, so that Sirt1 knockout liver/hepatocytes are maintained under reductive status in the presence or absence of DEN.

Knockout of Nrf2 in Sirt1 LKO mice augments DEN induced HCC formation

To investigate NRF2 function in Sirt1 LKO under DEN treatment, we knocked out liver Nrf2 using a sgNrf2 hydrodynamic tail vein injection. SgNrf2 Cas9 injection was administered twice at the 4th and 5th week after DEN treatment (Fig. 5A). SgNrf2 injection in LKO mice decreased the expression of Nrf2 compared to WT mice. Knockout efficiency was confirmed through the absence of NRF2 protein level as determined by Western blots (Fig. 5B). Accordingly, Nqo1, a direct target of Nrf2, also showed a reduction in mRNA levels in the sgNrf2 injection group (Fig. 5C). Cohorts of WT DEN, LKO DEN, and LKO sgNrf2 DEN were followed until 8 months. Again, LKO DEN mice displayed significantly reduced tumor burden when compared to WT/DEN mice; however, when Nrf2 was depleted in LKO mice, i.e., in LKO sgNrf2 DEN mice, there was a significant increase in tumor burden compared to Sirt1 knockout alone (Fig. 5D). The double knockout mice displayed the highest number of liver tumors, even surpassing that of WT (Fig. 5D). When examining tumor burden by size, we found that LKO sgNrf2 DEN mice had fewer small sized tumors (diameters < 2 mm); however, they had significantly greater numbers of tumors with diameters > 5 mm (Fig. 5E). This indicates that knocking out Nrf2 in LKO promotes the growth of larger sized tumors. Accordingly, liver to body mass ratios were significantly higher in LKO sgNrf2 DEN mice relative to WT DEN or LKO DEN mice (Fig. 5F).

Fig. 5. Knockout Nrf2 in Sirt1 LKO mice augments DEN induced HCC growth.

A Scheme of hydrodynamic tail vein injection of LentiV2 sgNrf2 in Sirt1 LKO DEN mice. B Immunoblot analysis of decreased NRF2 protein in sgNrf2 mice compared to Sirt1 LKO DEN mice. C Quantification of Nrf2 and Nqo1 mRNA expression in sgNrf2 mice. *p < 0.05; **p < 0.01. D Left, macroscopic photo showed increased tumor number and size after sgNrf2 injection. Right, tumor number in each mouse was counted. WT DEN, n = 14; Sirt1 LKO DEN, n = 14; Sirt1 LKO DEN + sgNrf2, n = 14. *p < 0.05; **p < 0.01; ***p < 0.001. E Display of tumor size and their proportions in the group. *p < 0.05. F Determination of the ratio of liver weight to body weight in sgNrf2 mice. *p < 0.05; ***p < 0.001. G Decreased expression of Gsta1, Gstm1, Gstm2, and Gstm3 in sgNrf2 mice compared to Sirt1 LKO DEN mice. *p < 0.05; **p < 0.01. H Nrf2 knockout in Sirt1 LKO DEN liver reduced GSH/GSSG ratio. N = 6 for each group. Values are mean ± SD. **p < 0.01. I Western blot showed that Nrf2 knockout in Sirt1 LKO DEN liver increased Cyclin B1 (CCNB1) and Phospho-Histone H3 Serine 10 level.

Previous studies have demonstrated that NRF2 orchestrates glutathione pathway gene expression and glutathione metabolism. Thus, we sought to get a better understanding of the regulation of the GST gene family in LKO sgNrf2 DEN mice by measuring expression of Gsta1, Gstm1, and Gstm2. Whereas transcription levels of these genes were increased in LKO mice, their expression levels were comparable to WT mice after sgNrf2 injection (Fig. 5G). In addition, we measured the ratio of GSH to GSSG after Nrf2 sgRNA injection. GSH/GSSG ratios were significantly lower in LKO sgNrf2 DEN mice compared to LKO DEN mice (Fig. 5H), demonstrating that knocking out Nrf2 reshaped LKO mice from a strong reductive environment to a relative oxidative environment.

To understand how this oxidative environment affects tumor growth, cell proliferation status was examined. By Western blot analysis, we found that LKO tumors had lower CCNB1 (Cyclin B1) while LKO sgNrf2 tumors had CCNB1 levels that were comparable WT tumors (Fig. 5I). In addition, histone H3 Serine 10 phosphorylation levels were enhanced in LKO sgNrf2 mice tumors (Fig. 5I). These findings demonstrated that NRF2 exerts control over the glutathione regulated redox environment in the liver. The reduction of DEN-induced HCC formation in LKO mice is due to elevated glutathione metabolism via Nrf2 activation.

Pretreatment of antioxidant reduces DEN induced HCC growth

N-acetylcysteine (NAC) is the acetylated precursor of reduced glutathione (GSH). NAC has been shown to prevent tumor initiation and act synergistically with other drugs in treating cancers [35, 36]. To further examine the anti-tumor effects of glutathione activation in HCC induced by DEN in WT and LKO mice, we treated DEN challenged WT mice with 1 g/L NAC in drinking water at 1 month or 5 months of age (Fig. 6A). WT mice that received NAC treatment at one month of age exhibited a reduction in tumor nodules. However, if WT mice were administered NAC at month 5, the treatment was insufficient to decrease the number of tumor nodules (Fig. 6B, C). Failure of the tumor inhibition effect of NAC treatment at this stage suggests that NAC can prevent cancer initiation but cannot inhibit cancer growth. Liver versus body mass ratios were similar in control and treatment groups (Fig. 6D).

Fig. 6. Pretreatment of antioxidant reduces DEN induced tumor growth.

A Protocols for 1 g/L NAC treatment in WT DEN mice. B Representative images of gross liver of H₂O, 1 g/L NAC drinking water starting from 4 weeks (NAC 1 m) and 5 m (NAC 5 m). C Quantification of macroscopic tumor number in WT and NAC treatment mice. H₂O mice, n = 17; NAC 1 m mice, n = 11; NAC 5 m mice, n = 16. *p < 0.05. D No significance of the ratio of liver weight to body weight between H₂O, NAC 1 m and NAC 5 m mice. E NAC treatment starting from 1 month elevated GSH/GSSG level. Values are mean ± SD. *p < 0.05; ***p < 0.001. F Quantitative RT-PCR analysis of Gsta1, Gstm1, Gstm3, and Nrf2 mRNA level in H₂O, NAC 1 m and NAC 5 m liver. *p < 0.05. G Left, representative CC3 IHC staining in tumors. Right, IHC staining of CC3 in tumor was quantified as mean ± SD. More than 7 fields were included for each group. ***p < 0.001.

To investigate if NAC treatment increases glutathione levels in the liver, we measured reduced glutathione level and oxidative glutathione. When NAC was given to DEN-injected WT mice, they also displayed improved antioxidant status, as indicated by an increase in GSH to GSSG ratios (Fig. 6E). At the same time, we also detected elevated expression of glutathione genes, Gsta1, Gstm1, and Gstm3 upon NAC treatment (Fig. 6F). However, we did not observe a significant upregulation of Nrf2 in NAC treatment (Fig. 6F), suggesting that NAC supplied a reduced form of glutathione without affecting Nrf2 expression. Simultaneously, strong CC3 staining was detected in mice treated with NAC at one-month-old compared to the no treatment group (Fig. 6G). Even though the mice treated with NAC at month 5 did not exhibit significant tumor remissions, they did display cell death (Fig. 6G). Altogether, these results indicate that NAC treatment maintains a reductive environment in the liver which may help in prevent HCC initiation.

DISCUSSION

Sirt1’s function in cancer formation and progression has been controversial. While there are reports that Sirt1 is a tumor suppressor, there are mounting evidence demonstrating it could function as an oncogene [7]. Several studies have suggested that Sirt1 is negatively correlated with HCC growth, proliferation, survival and chemoresistance [37, 38]. However, what role(s) Sirt1 plays in HCC formation progress remains unclear.

Hepatic deletion of Sirt1 reduced HCC incidences upon DEN challenge. The RNA-Seq analysis on normal livers, tumors, and tumor adjacent liver tissues from WT and LKO mice exhibited a glutathione and redox signaling imbalance in the LKO mice. Since it is well known that oxidative stress induced by DEN insult is a major contributing factor for the pathogenesis of experimental hepatocarcinogenesis [39], we went on to investigate how Sirt1 affects glutathione and redox pathways, and whether creating a reductive environment in liver can help prevent DEN-induced HCC.

Mounting evidence has shown that oxidative DNA damage plays a pivotal role in human hepatocarcinogenesis associated with chronic viral infection and liver cirrhosis [40–42]. However, whether DNA damage repair genes relates with SIRT1 expression in HCC human patients is not clear. Searching for human database, we found that several genes in DNA repair pathways, such as ATM, GADD45a, MDC1, MRE11A, RAD50, and TOPBP1 is significantly lower in SIRT1 low HCC patient than in SIRT1 high patients, suggesting SIRT1 is involved in DNA damage repair. In our study, we found DNA damage, as reflected by γH2AX level, was significantly reduced in mutant liver than in WT liver (Fig. 2B, DEN 5d). These analysis results line together with our in vivo observation that hepatic Sirt1 deletion rendered the LKO hepatocytes to lower DNA damage sensing status, and it could be an additional reason for reduced HCC induced by DEN. This in vivo finding could be highly relevant to human HCC formation.

In this study, we found that Sirt1 is involved in regulating glutathione metabolism at least partially via regulating Nrf2 mRNA levels and intracellular translocation of NRF2 protein. NRF2 has been shown to play a dual role in cancer progression [43]. Its absence promoted urethane-induced lung cancer initiation, while in the late stage Nrf2 loss inhibited cancer cell growth [44]. SIRT1 decreases NRF2 acetylation at lysine 588 and 591 [45]. Deacetylated NRF2 translocates to the cytoplasm so that its transcriptional activity is reduced. Our data revealed that Sirt1 deficient primary hepatocytes and Sirt1 shRNA knockdown cancer cells displayed increased Nrf2 transcription and nuclear translocation, which is accompanied with elevated expression of glutathione pathway genes and increased GSH/GSSG ratios, indicating that loss of Sirt1 pushed hepatocytes into a reductive state. We found that knocking out Nrf2 in DEN-treated LKO mice accelerated tumor growth (Fig. 5D). Upon Nrf2 deletion, intracellular expression of glutathione pathway genes was deeply affected so that an oxidative environment (reduced GSH/GSSG ratios) were generated in the hepatocytes. Reductions in intracellular GSH/GSSG ratios following Nrf2 knockout reflected the importance of glutathione in preventing tumor growth in a Sirt1 deficient background. This may also indicate that NRF2 elicits differential functions in different tissues.

The loss of Sirt1 predisposed hepatocytes/liver in a reduced environment. In LKO livers, higher GSH/GSSG ratios protected genomic damage caused by DEN, and at the same time, it also pushed cells that were exposed to DEN to apoptosis. We verified this hypothesis by feeding WT mice with daily NAC after DEN challenge. NAC treatment that started two weeks after the DEN injection time point was able to reduce tumor incidence in wildtype mice. However, if NAC was given at the time that liver lesions have appeared (i.e., 5 months) (Fig. 1B), it was not able to decrease the tumor incidence, indicating that an earlier reductive environment in liver can prevent HCC initiation upon carcinogen stimulation. Our results raise the possibility that glutathione and reductive agents such as NAC may function differently in tumor initiation and tumor progression, i.e., they display anti-initiation function at an early stage, but cannot inhibit growth of established tumors. In LKO livers, Nrf2 activation induced the expression of glutathione genes and GSH/GSSG level elevation, consequently preventing HCC initiation upon DEN challenge.

GSH-GSSG imbalance triggers cell survival or death in a cell type specific manner and threshold dependent phenotype [46]. Intracellular high GSH prevents cell death; this process is associated with protecting mitochondrial function and inhibiting caspase-3 dependent downstream signaling [47]. But low GSH or depletion of GSH with BSO exacerbates cell death by proapoptotic stimuli [48]. GSH depletion in a genetic knockout of Gclccauses apoptosis and is embryonic lethal due to differentiation failure [49].

In summary, our data demonstrate that Sirt1 plays an oncogenic role in HCC initiation induced by DEN. NRF2 activation in HCC patients with low level of SIRT1 may achieve clinical therapeutic efficacy.

MATERIALS AND METHODS

Mice

All mice were housed at the University of Macau Animal Facility with approved animal protocol (UMARE-AMEND-100). Sirt1^{flox5−6/flox5−6}; Alb-cre mice (LKO) and control cohorts were established by breeding Sirt1 ^{flox 5−6/flox 5−6} females with Sirt1 ^{flox5−6/flox5−6}; Alb-cre males [29]. To induce HCC, 14-day-old males were intraperitoneally injected with a single shot of 25 mg/kg DEN (Sigma N0756). All livers were collected during 9–11 a.m.

RNA-seq

Normal liver and tumor specimens were stored in RNA later. Total RNA was extracted with TRIzol (Thermo, cat# 15596026), and mRNA was enriched using oligo(dT) beads. RNA-Seq was performed by Novogene Bioinformatics Technology Co., Ltd on Illumina HiSeq PE150.

GSH/GSSG measurement

Total glutathione/Oxidized glutathione was assessed by kit (cat# A061–1) from Nanjing Jiancheng Bioengineering Institute (Nanjing, China). Briefly, 100 mg snap frozen liver was homogenized, then centrifuged of 3000 rpm for 10 min at 4 °C; total GSH and GSSG in the supernatents were quantitated according to the manufacturer’s protocol.

Histology

Liver samples were fixed with 10% neutralized formalin and embedded with paraffin. Four um thick sections were prepared for immunohistochemistry, immunofluorescence or hematoxylin & eosin scored as described before [50].

NAC treatment

Starting from 1 or 5 months after the DEN injection of 14-day-old WT male mice, 1 g/L NAC was used in drinking water until liver collecting. NAC water was refreshed every 2 days.

Statistical analysis

Data are expressed as mean ± standard deviation (SD). Statistical significance was analyzed by two-tailed Student’s t test or One-way ANOVA. Significant differences are shown by asterisks as: *p < 0.05; **p < 0.01; ***p < 0.001.

DATA AVAILABILITY

All RNA-sequencing data generated in this study have been deposited in GEO (https://www.ncbi.nlm.nih.gov/geo/) under the accession number PRJNA728801.