Effects of WT-HBx and the HBx mutants on hepatocarcinogenesis in SB mice
Empty SB vector and those carrying WT-HBx and the HBx mutants were successfully delivered into 42 mice. None of the empty vector-injected mice (n=9) developed tumors. Tumors and shrunken and cirrhotic livers were evident in M2-HBx-, M3-HBx-, and Ct-HBx-injected mice (shown in Fig 1). The incidence of tumors was 14.3% (1/7), 50% (2/4), 42.8% (3/7), 87.5% (7/8), and 57.1% (4/7) in the SB models whose hepatic genomic DNA was integrated with WT-HBx, M1-HBx, M2-HBx, M3-HBx, and Ct-HBx, respectively. Compared to WT-HBx-injected mice, all HBx mutants-injected mice displayed a trend toward higher tumor burden, especially the M3-HBx and Ct-HBx groups. No apparent pathological change was observed in the livers of the vector-injected mice. Inflammatory cell infiltration and cancer nests were observed in the HBx mutants-injected mice (shown in Fig 2A), which were similar to the histopathological changes of mouse hepatitis-induced HCC . Histopathologic analyses revealed that inflammatory cell infiltration was more severe in livers injected with HBx mutants than in those with WT-HBx. Our IHC analysis confirmed that the tumors were positive for HBx protein as well as cytokeratin 18 (CK18) and α-fetoprotein (AFP), the classic biomarkers of HCC (shown in Fig 1).
The insertion sites of HBx fragments
The integration sites of WT-HBx and the HBx mutants in the SB mice were examined via HBV-capture sequencing. In total, 1750 integrations were identified (Table S4). HBx fragments were mostly integrated in the intergenic and intron regions, with only 3.71% located in the exon regions. There were nine genes with the insertion sites detected in more than three samples (Table S5). The first three most frequently integrated genes were Fah (42.8%), catenin alpha 3 (Ctnna3) (28.6%), and fragile histidine triad gene (Fhit) (28.6%). Gene encoding telomerase reverse transcriptase (TERT) was not an integration size. The nine frequently integrated genes have low ratios of the reads with HBx integration to the normal reads, ranging from 0.09% to 0.25%. The difference in the levels of integration sites was not significant between liver tissues and tumor tissues (shown in Fig 2B). The number of integration sites was not significantly different among WT-HBx-, M3-HBx-, and Ct-HBx-injected SB mice (shown in Fig 2C). These data suggest that HCCs developed in the SB mice were induced by the HBx proteins, rather than insertional mutations.
Serum levels of proinflammatory cytokine in WT-HBx and the HBx mutants SB mouse models
The levels of Th1 cytokines (IFNγ, TNFα, IL-1β, and IL-12), Th2 cytokines (IL-4 and IL-5), IL-6, IFNα, TGFβ, VEGF, and GM-CSF were detected in the sera of the SB mice (shown in Fig 3 and Fig S3). Among the Th1 cytokines, IFNγ and TNFα were positive in all groups (shown in Fig 3A, 3B). The level of IFNγ was significantly higher in the M3-HBx-injected mice than in the controls. The positivity rate of IL-1β was significantly higher in M2-HBx- and M3-HBx-injected mice than in the empty-vector-injected mice (shown in Fig 3C). IL-5 and IL-6 were measurable in all samples. The levels of IL-5 and IL-6 were significantly higher in M3-HBx mice than in the empty vector- and WT-HBx-injected mice (shown in Fig 3E, 3F). The levels of IFNγ, TNFα, IL-5, and IL-6 were significantly higher in the mice with tumors than in the tumor-free mice (shown in Fig 3G). The positive rate of IL-1β was also significantly higher in the mice with tumors (shown in Fig 3H). IFNα was only positive in two M3-HBx-injected mice and one Ct-HBx-injected mouse (shown in Fig S3). IL-4 was only positive in one Ct-HBx-injected mouse. The HBx mutants did not upregulate the expression of TGF-β, VEGF, and GM-CSF (shown in Fig S3).
Effects of the HBx mutations on malignant phenotypes of cancer cells
To investigate whether the function of the HBx mutation was hepatocyte-specific, we evaluated the oncogenic effects of the HBx mutants in HepG2 and HeLa cells. The expression of WT-HBx and HBx mutants (M1-HBx, M2-HBx, M3-HBx, and Ct-HBx) in HepG2 and HeLa cells via the lentiviral infection was confirmed by Western blot (shown in Fig S4A). Ectopic expression of M1-HBx, M2-HBx, M3-HBx, and Ct-HBx significantly increased cell proliferation and the S phase proportion of HepG2 and HeLa, compared to WT-HBx (shown in Fig 4A, 4B). No significant difference in migration and invasion was observed between HepG2 cells expressing WT-HBx and those expressing HBx mutants; the same was true for HeLa cells (shown in Fig 4C and shown in Fig S4B–D). The effects of HBx mutants on cell proliferation, cell cycle, and migration were repeated in HeLa cells (shown in Fig 4D–F). As M3-HBx and Ct-HBx displayed the strongest tumorigenic capability in the SB mouse models, the effects of M3-HBx and Ct-HBx on the growth of cancer cells were further investigated using the xenograft of HeLa cells in nude mice (HepG2 is not tumorigenic in nude mice). It was found that the tumor weight was significantly higher in HeLa cells with M3-HBx than in those with WT-HBx (shown in Fig 4G). Thus, the HBx mutants, especially M3-HBx and Ct-HBx, promote cancer development, and this effect is not hepatocyte-specific.
Gene expression profiles affected by the HBx mutants
Based on cDNA microarray assay of HeLa cells, we identified the differentially expressed genes in each of the mutant HBx groups relative to WT-HBx group. The PPI network for each HBx mutant group was generated using STRING software. Six hub genes were selected because they had strong PPI function (the highest STRING score) in at least one PPI network (shown in Fig S5A). The six hubs are E3 ubiquitin protein ligase (also named MDM2), replication factor C subunit 3 (RFC3), polo-like kinase 1 (PLK1), cell division cycle 6 (CDC6), early growth response 1(EGR1), and PAI1. The effects of HBx mutants on the expression of these hub genes were also investigated with the microarray data of the SB mouse models with the highest tumor burden, M3-HBx and Ct-HBx. WT-HBx mice served as control. The expression levels of PAI1 were significantly higher in M3-HBx mice (fold change (FC) = 16.93, P < 0.001) and Ct-HBx mice (FC = 3.36, P = 0.001) than in WT-HBx control (Table S6).
The differentially expressed genes identified in HeLa cells and those identified in M3-HBx- and Ct-HBx-injected mice vs. WT-HBx-injected mice were subjected to GSEA analysis. Four cancer-related gene sets [19-22] were identified in both microarray assays: BERENJENO_TRANSFORMED_BY_RHOA_UP, MARKEY_RB1_CHRONIC_LOF_UP, VERNELL_RETINOBLASTOMA_PATHWAY_UP, and CHIANG_LIVER_CANCER_SUBCLASS_PROLIFERATION_UP. The first three gene sets were significantly enriched by M3-HBx and Ct-HBx in the cell and tissue microarray data. The remaining one was significantly enriched by M3-HBx in both cell and tissue microarray data. The four cancer-related gene sets included PAI1, P21, SKP2, and CDC20, respectively (shown in Fig 5A–D). PAI1 is the hub gene of the PPI network; P21 and CDC20 were also identified as functional molecules in the PPI network. P21 was significantly upregulated in both M3-HBx-injected mice (FC = 3.26, P = 0.006) and Ct-HBx-injected mice (FC = 1.77, P = 0.009). CDC20 was significantly upregulated in M3-HBx-injected mice (FC = 2.64, P = 0.005). SKP2, a molecule previously identified to be involved in HBx-induced carcinogenesis , was not identified in the PPI analysis. M3-HBx- and Ct-HBx-injected mice had a trend toward a higher level of SKP2 (M3-HBx, FC = 1.66, P = 0.08; Ct-HBx, FC = 1.67, P = 0.09), compared to WT-HBx-injected control. We also identified four inflammation-related gene sets [23-27] in the SB models. IL-6-related gene set, CROONQUIST_IL6_DEPRIVATION_DN, was significantly enriched in the M3-HBx- and Ct-HBx-injected mice (shown in Fig S5B, S5C). The IFN-γ receptor 1-related gene set, MATSUDA_NATURAL_KILLER_DIFFERENTIATION, and an inflammatory response-related gene set, NEMETH_INFLAMMATORY_RESPONSE_LPS_DN, were significantly enriched in the M3-HBx-injected mice (shown in Fig S5D, S5E). In total, 317 genes were involved in the four cancer-related gene sets, while 197 genes were involved in the four inflammation-related ones. Overall, 77 genes overlapped in the cancer-related gene sets and the inflammation-related gene sets. In total, 72 genes were involved in the two IL-6-related gene sets enriched in M3-HBx- and Ct-HBx-injected mice. Interestingly, of the 72 genes, 55 (76.3%), including P21, SKP2, and CDC20, were also included in the cancer-related gene sets.
Effects of the HBx mutants on the expression of functional molecules
We then investigated the expression levels of the functional molecules identified in the above bioinformatics in HepG2 and HeLa cells stably expressing WT-HBx or HBx mutants. The transcription levels of MDM2, RCF3, PLK1, CDC6, and EGR1 were not significantly affected by the four HBx mutants (shown in Fig S6). Interestingly, the mRNA and protein levels of PAI1 and CDC20 were significantly upregulated by M3-HBx and Ct-HBx in HepG2 and HeLa cells (shown in Fig 6A–D). HBx combo mutations (M1-HBx, M2-HBx, and M3-HBx) significantly downregulated the mRNA level and protein level of p21 in HepG2 rather than in HeLa cells (shown in Fig 6B, 6D). However, Ct-HBx significantly upregulated the expression of p21 in both HepG2 and HeLa cells. At the protein level, Ct-HBx significantly upregulated SKP2 in HeLa and HepG2 cells. The HBx mutants showed consistent effects on the RNA and protein expression of PAI1, CDC20, and p21, respectively. However, our luciferase assays showed that M3-HBx and Ct-HBx had no direct effects on the promoter activities of the three molecules, compared with WT-HBx in HepG2 and HeLa cell lines (shown in Fig S7). We then investigated the expression of PAI1, CDC20, and p21, the newly identified molecules participating in HBx-induced carcinogenesis, in the tumor tissue of the SB mice. The protein levels of CDC20 and PAI1 were significantly higher in the tumors from M3-HBx and Ct-HBx-injected mice, compared to their expression in the tumors from WT-HBx-injected mice (shown in Fig 7A, 7B). The protein level of p21 in the tumors was significantly lower in the M2-HBx group than in the WT-HBx group.
Reversal of HBx mutations’ effects by silencing functional molecules
To figure out if the HBx mutations regulate the biological activity of HepG2 and HeLa cells via upregulating PAI1, CDC20, and p21, we knocked-down the three genes and then examined the effect of stably expressed WT-HBx and HBx mutants on the proliferation of HepG2 and HeLa cells. PAI1 silencing significantly neutralized the effects of M3-HBx and Ct-HBx on the proliferation and cell cycle of HepG2 and HeLa cells. CDC20 silencing significantly reversed the effects of M3-HBx on the proliferation and cell cycle in HeLa cells. P21 silencing reversed the Ct-HBx-induced increase in the proportion of S phase cells, rather than cell proliferation in HeLa and HepG2 cells (shown in Fig 8).