Reduced SOD2 expression is associated with mortality of hepatocellular carcinoma patients in a mutant p53-dependent manner

SOD2 expression is reduced in human HCC

Altered SOD2 level has been reported in several cancer types, but the pattern is complex with down-regulation in some tumors but up-regulation in others. Thus far, SOD2 expression has not been investigated in HCC. Given the importance of SOD2 in mitochondrial integrity and liver biology, we investigated SOD2 expression in HCC. To this end, SOD2 mRNA level was analyzed in 40 pairs of human primary HCC and matching adjacent non-cancerous liver (NCL) tissues by RT-qPCR. SOD2 mRNA expression was found to be significantly reduced in 30 of 40 tumors, representing 75% of the HCC tissues compared with the NCL tissues (p = 0.001, Fig. 1a and 1b). In tumors with SOD2 down-regulation, SOD2 expression was reduced by as much as 12-fold, with the median decrease nearly 2-fold (Fig. 1b).

To verify this finding, we investigated SOD2 protein expression by immunohistochemistry (IHC) staining of a large cohort of 160 paraffin-fixed human primary HCC tumors and matching adjacent NCL tissues. Based on the study of genomic mRNA expression profiling in different mouse tissues [31], liver is one of the tissues where SOD2 is highly expressed in mice (Fig. S1). Consistently, SOD2 was found to be abundant as indicated by strong IHC staining in most of the NCL tissues (Fig. 2a). However, in tumor tissues, SOD2 protein expression showed considerably variations, ranging from negative, low, moderate to high IHC staining (Fig. 2a). Quantification of SOD2 staining IHC scores confirmed that SOD2 is indeed significantly decreased in HCC tissues as compared with their matched NCL tissues (p < 0.001, Fig. 2b). SOD2 protein expression was found to be largely reduced in 111 of 160 (69%) patients HCC tissues compared with the NCL tissues (p < 0.0001, Fig. 2b and 2c). In these 111 patients' HCC tissues, SOD2 expression was reduced by as much as 30-fold, with the median decrease 1.67-fold (Fig. 2c). Together, these results show that SOD2 expression is reduced at both mRNA and protein level in HCC.

Mechanism of SOD2 down-regulation in HCC

To understand the relationship between SOD2 mRNA and protein expression in HCC, we analyzed a panel of 10 HCC cell lines and an immortalized human hepatocyte cell line by RT-qPCR and Western blotting. Compared with the immortalized hepatocyte cell line MIHA, SOD2 mRNA was found to be lower in 7 of the 10 HCC cell lines (Fig. 3a), and protein level was lower in 8 of 10 HCC cell lines (Fig. 3b and 3c). The mRNA and protein level are largely correlated with each other (Fig. 3d), suggesting that SOD2 mRNA abundance is the primary determinant of SOD2 expression. However, there are some exceptions. Specifically, although SOD2 mRNA in HepG2 cells was higher than MIHA cells, SOD2 protein level was actually lower in HepG2 cells. QSG-7703 showed decreased SOD2 mRNA but not protein compared with MIHA cells. Thus, translational and post-translational mechanisms are likely to be involved in these cases. To understand the mechanism for altered SOD2 mRNA expression, we analyzed SOD2 copy number changes in one cohort of 97 HCC, 59 normal liver and 57 blood samples from the TCGA cancer genomic database (http://cancergenome.nih.gov). There was a pronounced decrease in SOD2 copy number in HCC versus blood and normal liver samples (Fig. 4a). Essentially the same phenomenon was observed with another cohort of 99 HCC and 86 normal liver samples obtained from the Oncomine genomic database (Fig. 4b) [32]. These observations indicate that loss of SOD2 locus is a mechanism for the decrease in SOD2 mRNA expression in HCC.

Loss of SOD2 expression is associated with advanced age and cancer progression in HCC patients

Frequent down-regulation of SOD2 suggests that it plays an important role in HCC pathogenesis. We therefore investigated the relationship between SOD2 expression and clinicopathological features of HCC. Based on the IHC scores, we divided HCC patients into SOD2 high-expression and low-expression subgroups using the median IHC score of 180 as the cutoff value. We then analyzed the correlation between SOD2 expression and 15 widely recognized clinicopathologic parameters in the cohort of 160 HCC specimens (Table 1). Consistent with the established role of SOD2 in aging, chi-square analysis shows that there is a statistically significant correlation between low expression of SOD2 and older patients (≥ 50 y, p = 0.007). Moreover, low expression of SOD2 is associated with multiple primary tumors (p = 0.012), tumor embolus (p = 0.016), tumor recurrence (p = 0.039), and advanced tumor stages as classified by the Primary Tumor, Lymph Nodes and Metastasis (TNM) (p = 0.006), and Barcelona clinic liver cancer (BCLC) (p = 0.001) staging systems. These observations suggest that low SOD2 expression is involved in the progression of HCC.

Reduced SOD2 expression is significantly correlated with poor prognosis of HCC patients

Because loss of SOD2 is associated with HCC progression, we investigated that whether differential SOD2 expression is linked to the prognosis of HCC patients. To this end, Kaplan-Meier and the log-rank test analysis were used to evaluate the survival of these HCC patients. As shown in Fig. 5a, patients with low SOD2 expression had statistically significant, shorter overall survival (OS) and relapse-free survival (RFS) than those with high SOD1 expression (p = 0.005 and p = 0.004, respectively). The postoperative median OS and RFS times of the entire cohort of 160 HCC patients were 40.87 months and 30.08 months respectively, and for the 75 patients with high SOD2 expression were 50.27 and 46.17 months, respectively. In contrast, the postoperative median OS and RFS times for the 85 cases with low SOD2 expression were 34.1 and 29.03 months, respectively. Furthermore, low SOD2 protein expression correlates with worse prognosis for patients with large tumors (tumor size > 5cm) (Fig. 5b).

Interestingly, low SOD2 expression is also associated with shorter OS and RFS in subgroups of patients characteristic of less advanced disease, including low AFP concentration (< 400 ng/μl) (Fig. 5c), no tumor embolus (Fig. 5d), no liver cirrhosis (Fig. 5e), high differentiation (Grade I+II) (Fig. 5f) and complete tumor capsule (p = 0.035) subgroup, suggesting that SOD2 is involved in the development and progression of HCC during early disease stages. Low SOD2 expression also appears to be associated with poorer prognosis in more advanced disease stages (Fig. S2). However, surgical HCC samples were usually obtained in relatively early disease stages, which limits the sample size and thus statistical significance. Further study will be necessary to elucidate the clinical significance of SOD2 expression in advanced HCC.

SOD2 is an independent prognostic predictor for HCC patients

To investigate the prognostic value of SOD2 protein expression, univariate analysis of SOD2 protein level and clinicopathological features was carried out using the Cox proportional hazard model. The result shows that low SOD2 protein expression (p = 0.005), higher serum AFP (p = 0.002), larger tumor size (p = 0.010), multiple tumors (p < 0.001), tumor embolus (p = 0.004) advanced TNM (p = 0.001) and BCLC stage (p < 0.001) are predictors for poor OS of HCC patients (Table 2). On the other hand, low SOD2, higher Serum AFP, tumor embolus, multiple tumors, advanced TNM, BCLC, and postoperative metastasis were statistically significantly correlated with shorter RFS in HCC patients (Table 2). To determine if SOD2 low expression can be an independent prognostic factor, multivariate Cox regression was further performed on those variables significantly associated with poor survival of HCC patients. The results demonstrate that low SOD2 protein expression is indeed an independent predictor for poor OS (HR = 2.310, 95% CI = 1.031-5.178, p = 0.042) and RFS (HR = 1.809, 95% CI = 1.037-3.156, p = 0.037) in patients with HCC (Table 2), suggesting that SOD2 protein is a useful prognostic biomarker for HCC patients.

Increased mortality of HCC patients with low SOD2 expression is p53-dependent

The tumor-suppressor p53 plays an important role in oxidative stress response. It has been shown that SOD2 is a p53 target gene at the transcriptional level: p53 normally binds to SOD2 promoter and represses SOD2 transcription [33], but activates SOD2 expression during oxidative stress [34]. Because TP53 is frequently mutated in liver cancer, we investigated the relationship between TP53 mutations and SOD2 expression in the survival of HCC patients using genomic and clinical data from cBioPortal for Cancer Genomics (http://www.cbioportal.org). HCC patients were divided into the SOD2 high expression and low expression subgroups using the median SOD2 mRNA value of 12645.314 as the cutoff. Kaplan-Meier and the log-rank test analysis showed that differential SOD2 mRNA expression is not statistically significantly correlated with patient survival in this cohort of HCC patients (Fig. 6a). We therefore further divided HCC patients into wild type and mutant TP53 subgroups. Strikingly, low SOD2 mRNA expression is associated with much shorter OS (p < 0.034) and RFS (p < 0.028) in HCC patients carrying the mutant TP53, but not the wild type TP53 (Fig. 6b). These results suggest that low SOD2 expression alters the mortality of this HCC patient population in a p53-dependent manner.

Clinical specimens and cell lines

Formalin-fied paraffin-embedded (FFPE) HCC and matchingnon-cancerous liver (NCL)tissues from 160 patients with primary HCC, who underwent curative hepatectomy from January 2004 to August 2012, were collected from the Department of pathology of the Sun Yat-sen University Cancer Center, Guangzhou, China. The detailed clinicopathological features of patients are summarized in Table 1 and Table S1. The criteria of the 7th Edition tumor-node-metastasis (TNM) classification of the American Joint Committee on Cancer Staging [44] and the Barcelona Clinic Liver Cancer (BCLC) staging system [45] were used to determine the clinical stages. The cases were followed up regularly since the date of primary liver resection and the median follow-up time was 40.7 months. Overall survival (OS) was defined as the interval elapsed between the date of primary liver resection and the date of HCC-associated death or last follow-up. Recurrence-free survival (RFS) was calculated from the date of primary liver resection to the date of HCC-associated death, relapse, metastasis, or last follow-up. In addition, another 40 pairs of frozen HCC and matched NCL tissues were used for mRNA expression analysis. All the paired tissues were confirmed histologically by two pathologists of this cancer center. Written informed consent was obtained from all patients for this research. This study was approved by the Institutional Review Board and Human Ethics Committee of Sun Yat-Sen University Cancer Center prior to the study.

Human HCC cell line BEL-7402 is maintained in RPMI-1640 medium (Gibco, Carlsbad, CA) supplemented with 10% FBS. Human HCC cell lines CRL-8024, HEP-3B, HEP-G2, HUH-7, MHCC-97H, MHCC-97L, QSG-7703, SK-HEP-1 and SMMC-7721, and the immortalized human hepatocyte cell line MIHA, were cultured in DMEM (Life Technologies, Inc.) with supplemental 10% FBS. All cell lines were incubated at 37°C with 5% CO₂.

RNA extraction and RT-qPCR

Total RNA was extracted from the aforementioned cell lines and snap-frozen HCC/NCL tissues using the TRIzol reagent (Invitrogen, 15596-018, USA) according to the manufacturer's instruction. Two-step RT-qPCR was performed to assess the mRNA level of SOD2 as previously described [46]. First strand cDNA was synthesized by GoScript™ reverse transcriptas and random primers (Promega, A5001, Madison, WI, USA). cDNA was quantified in triplicate by GoTaq® qPCR Master Mix (Promega, A6001, Madison, WI, USA) on Roche LightCycler480(Roche, Basel, Switzerland). GAPDH was used as an internal control. Relative SOD2 mRNA expression was presented by 2−ΔΔCT method. Paired primer sequences used for SOD2 are: 5′-GACAAACCTCAGCCCTAACG-3′ (forward) and 5′-GAAACCAAGCCAACCCCAAC-3′ (reverse); for GAPDH: 5′-TGCACCACCAACTGCTTAGC-3′ (forward) and 5′-GGCATGGACTGTGGTCATGAG-3′ (reverse).

Western blotting

Total proteins were extracted from the above-mentioned cell lines and fresh tissues by using a whole cell lysis kit (Keygen, KGP2100, China). The Western blot procedure is as previously described [47]. Briefly, 30 μg of total protein from each sample was separated in 12% SDS-PAGE gel and transferred to PVDF membrane (Millipore, Billerica, MA). After blocking in PBS buffer containing 6% nonfat milk at room temperature for 1 hr, the membrane was incubated with rabbit anti-SOD2 (1:5000, Cat No. ab13533, Abcam, USA) and anti-GAPDH (1:5000, Cat No. 5174, Cell Signaling Technology, USA) primary antibodies at 4°C overnight. After washing with TBS buffer containing 0.1% Triton X-100 for three times, the membrane was then probed with anti-rabbit HRP-conjugated secondary antibodies (dilution for 1:5000, Jackson Immunoresearch Inc, PA, USA) and detected by using the enhanced chemiluminescence substrates (PerkinElmer, MA, USA). GAPDH was used as a loading control. The intensity of immunoblot signals was determined using the Bio-Rad software Quantity One (Bio-Rad Laboratories Inc., Hercules, California, USA).

Immunohistochemistry

Immunohistochemistry (IHC) was performed on 4 μM thick sections that were cut from the FFPE tissue blocks. The sections first were dried in an oven at 60°C for 2 hr, deparaffinized three times with xylene for 10 min each, rehydrated in a graded series of decreased alcohol solutions, washed three times with PBS for 10 min to remove the residual alcohol, immersed in 0.3% H₂O₂ in methanol to block endogenous peroxidase activity, washed three times with PBS for 10 min each to remove the residual H₂O₂ and methanol, boiled in 0.01 M sodium citrate buffer (pH 6.0) using a microwave oven for 15 min at low heat to retrieve the antigens, and cooled to room temperature by running water. After incubating in 10% FBS for 30 min to prevent nonspecific staining, the sections were incubated with rabbit anti-SOD2 (1:400, Cat No 13533, Abcam, USA) antibodies at 4°C overnight. In the second day, the tissue sections were rinsed three times with PBST (0.1% Triton-X100 in PBS), labeled with secondary antibody conjugated with horseradish peroxidase (Dako Corp., Carpinteria, CA) for 30 min at 37°C, washed three times with PBS for 10 min each to remove the remaining secondary antibodies, developed with 3, 3′-diaminoben-benzidine (Dako Corp., Carpinteria, CA), counterstained with hematoxylin (Sigma), dehydrated in increased concentration of alcohol, and dried on the air and mounted with neutral resins. All tissue slides were evaluated using the semi-quantitative immunostaining score (ISS) method by two pathologists who were blind to all clinical and biological variables. The score of immunostaining was defined as 0 – 3 (0, negative; 1, weak; 2, moderate; 3, strong). The percentage of positive immunostaining was scored as 0 to 100%, yielding an overall score ranging from 0 to 300 points. The median score was used as a cutoff for classification of patients into high and low expression groups.

Statistical analysis

The SPSS software package for windows (version 17.0; Chicago, IL, USA) was used for all statistical analysis. Paired Student's t test was used to analyze the difference of mRNA and protein expression levels between the paired HCC and NCL tissues. The chi-square test was performed to determine the correlation of SOD2 expression with the patients' clinicopathological features. The Kaplan-Meier method and log-rank test were used to analyze survival of HCC patients. To determine if SOD2 is an independent factor of survival and recurrence, univariate and multivariate survival analyses were performed using the Cox proportional hazards regression model. The two-sided p value less than 0.05 was defined as statistically significant for all statistical analyses.

Analysis of TCGA and Oncomine genomic data

We downloaded and analyzed SOD2 mRNA expression (RNA Seq V2 RSEM) data, the p53 mutant data and the clinical data of the Hepatocellular Carcinoma (TCGA, provisional) from the website cBioPortal (http://www.cbioportal.org/data_sets.jsp) [48, 49]. The SOD2 copy number data of TCGA (http://cancergenome.nih.gov/) and Guichard Liver [32] were downloaded and analyzed from the OncoMine cancer genomic database (https://www.oncomine.org/).