Androgen Receptor Enhances Hepatic Telomerase Reverse Transcriptase Gene Transcription After Hepatitis B Virus Integration or Point Mutation in Promoter Region

Study Subjects

Tumor and nontumor paired liver tissues from 70 HBV-related HCC patients (46 men and 24 women) were collected at National Taiwan University Hospital (NTUH). The liver tissues were stored in liquid nitrogen after being resected for subsequent genomic DNA (gDNA) and RNA extraction. Another 31 pairs of tumor–nontumor gDNA and RNA from 16 male and 15 female patients with HBV-related HCC were obtained from the Taiwan Liver Cancer Network (TLCN). The institutional review boards of NTUH and TLCN approved the use of these archived samples. The clinical characteristics of these 101 patients are summarized in Supporting Table S1.

Capture-NGS Analysis

gDNA was extracted from the liver tissues using a standard phenol-chloroform method, and 1 µg of the gDNA was used for bar-coded library construction with the TruSeq DNA Sample Preparation kit (Illumina, San Diego, CA). The libraries were processed for hybridization with the customized biotinylated DNA probes (IDT, Coralville, IA) together with human Cot-1 DNA (Invitrogen, Carlsbad, CA) at 65°C for 4 hours, following the manufacturer’s instructions (SeqCap Hybridization and Wash Kit; Roche, Mannheim, Germany). After hybridization, the captured DNA was purified using Dynabeads M-270 Streptavidin (Invitrogen), and unbound DNA was removed with washing buffer. The target DNA fragments were amplified by PCR with the adapter primer set (Illumina) and KAPA HiFi HotStart ReadyMix (KAPA, Wilmington, MA). The PCR products were then purified with Agencourt AMPure XP beads (Beckman Coulter, Brea, CA) and sequenced by HiSeq 2000 (Illumina).

Capture Probe Set

The capture probe set consisted of 200 biotinylated DNA oligonucleotide probes (IDT) covering the entire HBV genome and selected endogenous genetic regions (TERT promoter and coding region of tumor protein P53 [TP53]). The HBV probes were designed using HBV consensus sequences of genotypes B and C assembled from HBV whole-genome sequences that were sampled and sequenced from hepatitis B patients at NTUH. The National Center for Biotechnology Information reference sequences of selected endogenous gene regions, including –279 to +303 relative to ATG of the TERT gene (NC_000005.9) and exons 2-11 (total coding sequence) of TP53 (NM_000546.5), were used to design the endogenous gene probes.

Bioinformatics Analysis

Paired NGS data sets were treated as single reads and merged into single data sets. Quality control of the NGS data sets includes removal of low-quality reads, probe hybridization analysis, and exclusion of low-redundancy reads. Low-quality reads were defined as reads with a number of low-quality nucleotide sites (quality score <5) equal to or greater than half the read length and were excluded from further analyses. NGS data sets of high-quality/high-redundancy reads were compared against both HBV and human reference sequences by the Basic Local Alignment Search Tool (BLAST). The sequence reads containing the junction regions of HBV integration into the human genome were identified if one portion of the reads was of HBV origin and the remaining nonoverlapping portion was of human origin. This analysis was accomplished by initially comparing the sequence reads against the HBV reference sequences and then identifying partial HBV reads with a maximum HBV length of 100 bp and an e-value threshold of 1e-10. Partial HBV reads were then compared with the human hg19 reference sequences, and junction-containing reads were identified with an e-value threshold of 1e-10. Junction reads were clustered according to the pairwise BLAST results using BLAST-based assembly pipeline (BBAP)15 to identify unique junction types and to quantify the number of junction reads for each unique junction type. Junctions supported by two or more junction types and 15 or more junction reads were identified as viral/host (vh)–chimera DNA. High-quality/high-redundancy reads that were not identified as partial or full HBV reads were then assembled onto endogenous gene reference sequences consisting of the TERT promoter region and the TP53 coding sequence (CDS) with BBAP.15 Somatic mutations were identified by calculating the nucleotide frequencies according to assembly results and comparison with endogenous gene reference sequences.

The HBV genotype of a sequenced sample was determined by assembling its HBV genomic sequence from its NGS data sets using BBAP and HBV reference sequences. The HBV genotype and sequence variation were determined by comparing the assembled HBV genomic sequences to HBV reference sequences of all available genotypes (A-J). The assigned genotype and variants were analyzed for association with HBV integration or mutation at the TERT promoter region.

Plasmid Constructs

The reporter plasmid wild-type (WT)–TERT–luciferase (Luc) was constructed by inserting –47 to –209 of the TERT promoter fragment in front of the luciferase gene in the promoterless pGL4.10 vector. The G(–124)A-TERT-Luc and G(–146)A-TERT-Luc constructs contained the G to A mutation at the –124 and –146 position (relative to ATG) in WT-TERT-Luc. These constructs were kindly provided by Professor Costello at the University of California, San Francisco.16 The HBV-TERT-Luc reporter plasmids were constructed by insertion of the DNA fragments PCR-amplified from HCC of patients M15, M33, M52, M57, and F36, which contained HBV integrations within the TERT promoter region, into the NheI/SacI and HindIII restriction enzyme sites of the pGL4.10 vector. The PCR products amplified by the same primer sets with the DNA from normal liver were constructed as control plasmids in the transfection analysis. The primer sets used for the PCR are listed in Supporting Table S2, and the inserted sequence was confirmed by Sanger sequence analysis.

The details for the plasmids expressing the androgen receptor (pSG5-AR), estrogen receptor alpha (pCMV5-ESR1), and the reporter constructs of 5XARE-Luc and estrogen response element (ERE)–E1b-Luc have been described.9, 17

Validation of the Capture-NGS Platform for Identifying HBV Integration in HCC: Results Comparable to Genomic Sequencing

To examine whether the HBV integration patterns in HCC show any gender differences, the integration sites were identified and compared between male and female patients with HCC. HCCs from 62 male and 39 female patients with HBV-related HCC were included in our analysis, with the clinical information summarized in Supporting Table S1.

Instead of genomic sequencing, we developed a capture-NGS platform to identify the viral integration sites in these HCCs following the flowchart outlined in Fig. 1A. In brief, the tumor gDNA was extracted for library construction, in which the DNA fragments containing HBV were enriched by hybridization with the probes covering the whole HBV genome. In addition to the pure HBV DNA fragments, the DNA fragments containing the HBV-host gene junctions at viral integration sites, namely, the vh-chimera DNA, were enriched by hybridization. Moreover, we included the probes covering the TERT promoter and CDS of TP53 to analyze the two common somatic mutations in the HCC samples. The captured DNA fragments were then processed for NGS and bioinformatics analysis as summarized in Fig. 1A.

HBV integrations were identified in 94 out of the 101 HCCs in our platform (93.1% of samples containing HBV integration). The number of integrations in these HCCs is summarized in Fig. 1B, ranging from 1 to 15 integrations in individual HCCs. All of the major HBV integrations, including those close to the hotspot genes of TERT, MLL4, and CCNE1 and some others, have been confirmed by junction-specific PCR, with the representative results shown in Supporting Fig. S1. Each chimera DNA was only detected in one HCC as a unique signature for the individual HCC. The integration breakpoints in HBV genome are schematically summarized in Fig. 1C, showing hotspot clustering approximately in the HBV DRI and DRII regions, consistent with previous studies.10 The distribution of HBV integration sites in the HCC genome, exon, promoter, intragenic (intron), and intergenic regions is summarized in Fig. 1D. The integration sites close to the specific cellular genes were plotted (Fig. 1E), with details summarized in Supporting Table S3. The results showed that the integrations located across all of the chromosomes but with hotspots adjacent to the TERT, MLL4, and CCNE1 genes in 28, 14, and 6 out of 101 HCC cases (Fig. 1E). The average number of HBV integrations for one HCC (3.8 integrations/HCC) and the integration patterns were largely consistent with those identified by previous whole-genome sequencing analysis,10 thus validating our capture-NGS platform for identifying HBV integrations in the human HCC genome.

In addition to the HBV integrations, point mutations in the TERT promoter and the CDS of the TP53 gene were also detected using our capture-NGS analysis. Twenty out of 101 HCCs contained TERT promoter mutations (19.8%), and 38 out of 101 HCCs contained TP53 mutations (37.6%). These somatic mutations were validated by Sanger sequencing analysis and summarized in Supporting Table S4. Of note, the TERT promoter mutation and HBV integrations in the TERT promoter region occurred in a mutually exclusive manner, accounting for 47.5% (48 out of 101) of the HCC cases in combination.

HBV Integrations and Point Mutations in the Tert Promoter Region Preferentially Occur in Male Patients with HCC

To examine any gender differences in HBV integration patterns in these HCC cases, the NGS results were stratified by gender. No significant difference was identified between male and female patients with HCC regarding the integration frequency, the integration number/tumor (Fig. 2A), or the breakpoints in the HBV genome (Fig. 2B). For the distribution pattern of integrations in exon/promoter/intron/intergenic regions in the host genome, we observed a significantly higher frequency of HBV integrated in gene promoter regions in male patients with HCC (Fig. 2C). Interestingly, a markedly higher frequency of integrations was observed in the TERT promoter region in male patients with HCC (Figs. 2D and 3) (22/58 male patients with HCC, 6/36 female patients with HCC, P = 0.0285). Such a gender difference, however, did not exist for integrations in the MLL4 and CCNE1 genes (Figs. 2D and 3).

The details for the HBV integrations close to the TERT, MLL4, and CCNE1 genes and the point mutations in the TERT promoter and TP53 CDS regions for these 101 HCC cases are schematically summarized in Fig. 3A (upper panel, male patients with HCC; lower panel, female patients with HCC). Notably, not only the HBV integrations in the TERT promoter region but also the point mutation in the TERT promoter, at –124 relative to the ATG (–124G>A), occurred more frequently in male patients with HCC compared to female patients with HCC (17/62 male, 3/39 female, P = 0.0201). In contrast, no gender difference was found for mutations in the TP53 CDS (22/62 male, 16/39 female, P = 0.5777) (Fig. 3B). As these two types of genetic aberrations at the TERT promoter occur in a mutually exclusive manner, comparison of these two types of genetic changes in combination between male versus female HCCs showed a marked gender difference, which accounted for 39/62 (62.9%) male patients with HCC but only for 9/39 (23.1%) female patients with HCC (P < 0.0001) (Fig. 3B).

The position of the HBV integration sites in the TERT, MLL4, and CCNE1 gene regions are illustrated in Fig. 4A-C. The integrations were clustered in the promoter region of TERT and CCNE1 but clustered at introns 3-5 of MLL4, which was also consistent with the pattern identified in a previous NGS analysis.10 Moreover, we examined the mRNA expression of TERT (stratified by the promoter integration and promoter point mutation), MLL4, and CCNE1 in male and female HCCs by quantitative RT-PCR analysis. The results indicated that both HBV integrations and point mutations in the TERT promoter region showed elevated TERT expression but that the HBV integrations conferred an even higher TERT expression level than the TERT promoter mutations in male patients with HCC (Fig. 4D). However, the integrations close to introns 3-5 of MLL4 did not increase its expression in HCC (Fig. 4E). For CCNE1, although the frequency was low, we observed that HBV integrations in the promoter region also elevated the gene expression level (Fig. 4F). These results suggested that the HBV integrations adjacent to the TERT and CCNE1 promoter likely activated gene expression but that integration in MLL4 might exert a different effect.

In summary, comparison of the capture-NGS results in male and female patients with HCC indicated that both HBV integration and point mutations in the TERT promoter region, which are associated with higher TERT expression, occurred more commonly in male than in female patients with HCC (P < 0.0001, in combination). Therefore, some gender-specific factors might contribute to the elevation of TERT in male patients with HCC, through either HBV integrations or specific TERT mutations in the TERT promoter region. It may provide a growth advantage for the clonal expansion and carcinogenesis of these hepatocytes, which are thus selected and enriched in the resulting male patients with HCC.

Inverse Effects of Androgen and Estrogen Pathways in Regulating Tert Transcription Through the Integrated HBV in the Tert Promoter Region

We addressed whether the sex hormones affect the transcription of TERT through integrated HBV or point mutations in the TERT promoter using a reporter assay. The reporter plasmid WT-TERT-Luc containing the luciferase gene under the control of the TERT promoter region, covering –209 to –47 relative to ATG, was used to introduce either the HBV integration or point mutations.16 The effects of the androgen and estrogen pathways on the mutant reporter constructs were then evaluated compared with those of the WT construct.

We first focused on the HBV integration type of mutation in the TERT promoter region. In the 28 HCCs with HBV integrations at the TERT promoter region, 27 cases could contain the responsive elements for AR and ERα (Supporting Fig. S2). The responsive elements were highly conserved among most of the integrated HBV, regardless of virus genotype B or C (Supporting Table S5). Among them, a representative HBV-TERT-Luc reporter plasmid containing HBV integrated in the TERT promoter (identified in 1 HCC patient, no. M15) was constructed (Fig. 5A). We observed significantly higher Luc activity in HBV-TERT-Luc(M15) than in WT-TERT-Luc transfected HepG2 cells (~100 times) (Fig. 5B). The results indicated that the integrated HBV induced a strong activation effect on TERT transcription. The AR and ERα expression plasmids were then individually transfected into the cells to examine the effect on WT-TERT-Luc and HBV-TERT-Luc(M15) activity. As control sets, the androgen and estrogen pathways stimulated ARE-Luc and estrogen response element (ERE)–Luc reporter activity in a ligand-dependent manner (Fig. 5C). However, neither pathway had a significant effect on WT-TERT-Luc activity (Fig. 5B), suggesting that the WT-TERT promoter is not under the regulation of sex hormones. For this HBV-TERT-Luc construct, the AR pathway further increased Luc activity in a ligand-dependent manner (Fig. 5B, middle panel). In contrast, the Luc activity of this HBV-TERT-Luc construct was suppressed by the ERα pathway, which was also dependent on the ligand (Fig. 5B, right panel). The results have been validated in another hepatoma cell line, Huh-7 (Supporting Fig. S3). In addition, the HBV-TERT-Luc reporter activity driven by another four HBV integrations at different positions in the TERT promoter region, cloned from HCC of 3 male patients (M33, M52, M57) and 1 female patient (F36), also showed similar hormone responses in both HepG2 and Huh-7 cell lines (Supporting Fig. S4). Finally, a similar result was also confirmed in a female HCC cell line, SNU387 (Supporting Fig. S5). Therefore, the HBV integrations rendered the TERT promoter responsive to sex hormone regulation.

In our previous study, chelation of HNF4α, a major positive transcription factor targeting the core domain of HBV EnhI, was the mechanism responsible for ERα suppression of HBV transcription.9 Thus, we examined the role of HNF4α in these reporter assays. Depletion of HNF4α by lenti-small interfering (si) HNF4α alone dramatically decreased the increment of TERT expression induced by HBV integration (Fig. 5D). Moreover, both the AR and ERα effects on the HBV-TERT-Luc activity were diminished by HNF4α depletion (Fig. 5D). Therefore, HNF4α functions as one fundamental transcription factor responsible for TERT activation through integrated HBV; both androgen-modulating and estrogen-modulating effects are dependent on HNF4α.

The Androgen Pathway Increases Tert Transcription Through the Common Point Mutation in the Tert Promoter, Which is Dependent on GA Binding Protein Transcription Factor Subunit Alpha

Next, we examined whether the TERT promoter mutations –124G>A and –146G>A were also responsive to sex hormone regulation, though the –146G>A mutation was not identified in all 101 HCCs but has been reported in HCC from other cohorts and thus was also included in our analysis.18⁾ The –124G>A and –146G>A mutations were introduced into the WT-TERT-Luc construct, namely, G(–124)A-TERT-Luc and G(–146)A-TERT-Luc, respectively (Fig. 6A). In Huh-7 cells, a significant elevation in the reporter activity in both G(–124)A and G(–146)A-Luc-transfected cells was identified relative to that of the WT-TERT-Luc-transfected cells (Fig. 6B, left panel). The results confirmed the effects of both promoter mutations on TERT transcription in hepatocytes. In the presence of AR, the reporter activity of G(–124)A and G(–146)A-TERT-Luc was further elevated (~3 times) in a ligand-independent manner (Fig. 6B, right panel). In contrast, the reporter activity was not affected by the presence of ERα (Fig. 6C). These findings have been validated in HepG2 cells (Supporting Fig. S6), indicating that the AR, but not the ERα, pathway might also contribute to the up-regulation of the transcription of hepatic TERT through the –124G>A and –146G>A mutations.

It has been well documented that both –124G>A and –146G>A TERT promoter mutations create new binding sites for GA binding protein transcription factor subunit alpha (GABPA), a key molecule for increasing TERT transcription.16 To evaluate the role of GABPA in mediating AR function to regulate TERT transcription, we depleted GABPA using lenti-si-GABPA and examined the effect on AR function. The results first showed that GABPA knockdown abolished the increase of Luc activity induced by both mutations (Fig. 6D), supporting the fundamental role of GABPA in the activation of the mutated TERT promoter. Interestingly, the increase in Luc activity by AR was also diminished by GABPA depletion (Fig. 6D; Supporting Fig. S6C), indicating that the AR effect on mutant TERT promoter activity was dependent on GABPA. Collectively, the results identified GABPA as the key mediator of AR function to stimulate TERT expression through the common point mutations in the TERT promoter region.