| Background and aims:Primary hepatocellular carcinoma (HCC) is the most common cause of cancer death in China. It is estimated that more than80%of HCC is etiologically associated with HBV in China. Chronic HBV infection results in persistent inflammatory hepatocytes damage and compensatory regeneration. Oncogenic viral proteins such as HBx and mutant large surface protein were also considered playing direct pathogenic roles. In addition, it has been proposed more than three decades ago that HBV DNA integration into the hepatocytes cellular genome played a causative role in hepatocarcinogenesis. However, this speculation was mostly based on small scale observational studies using only tumor tissue and lacked the comparative control of the non-tumor tissue counterpart. Therefore, analysis of adjacent non-tumor tissue and consequently a systematic investigation of this hypothesis are still required. Through this systematic investigation, carcinogenesis significance of HBV integration in hepatocarcinogenesis was re-evaluated in this study.Materials and method:Sixty HCC patients underwent surgical operation were recruited from He’nan Cancer Hospital from2008to2009(ranging age from30years to70years, mean age=50.7±8.46years; male:female=42:18;3of them were HBeAg-positive and57were HBeAg-negative);58of them were accompanied with liver cirrhosis.DNAs were extracted from60paired frozen HCC tissues and corresponding adjacent non-tumor liver tissues using proteinase K followed by the standard pheol/chloroform extraction and ethanol precipitation method. For aCGH study, the genomic DNAs were extracted using the Genomic DNA purification Kit (Qiagen, USA).LM-PCR was employed using cassette primers and primers specific to HBV sequences to amplify viral-host junctions. Alu-PCR was employed using specific primers to human Alu sequences and to HBV sequences to efficiently amplify viral-host junctions. The HBV specific primers were HBV X gene forward primers and HBV core gene reverse primers. The PCR products were subjected to agarose gel electrophoresis, and the bands were cut out of the gels for subsequent sequencing. We subcloned the PCR products into TA cloning vector when direct sequencing failed. HBV-captured deep DNA sequencing was performed in28paired tumor and non-tumor tissues. The viral-host sequences were analyzed by using the NCBI Blast tool and UCSC database hg19to identify viral genome sequences, and to map the integration sites in human genome separately.The sequences for the ’free’ HBx region were detected by PCR using primers acrossed from X gene to C gene. The about1100bp PCR products was directly sequenced. The integrated HBV X region sequences were acquired from the confirmed viral-host junction sequences. A total of25paired DNA samples derived from tumor tissues and the corresponding tissues were prepared. Chromosome aberration was comprehensively analyzed via aCGH. In the assay, each corresponding paired adjacent non-tumor tissue DNA was used as reference DNA.The exons2to11of TP53were amplified by4independent PCR reactions. The PCR products were directly sequenced to identify mutation. Meanwhile,9SNP sites (rs1642785, rs17878362, rs17883323, rs1042522, rs77624624, rs2909430, rs12947788, rs12951053and rs6503048) in this region were also analyzed for any potential loss of heterozygosity (LOH), in comparison with the corresponding non-tumor tissues.Through comparing the electrophography peak values of the heterozygote SNP sites, LOH was defined according to the following formula:LOH index:L=(T2/T1)/(N2:N1)(T is tumor tissue, N is the adjacent non-tumor tissue;1and2are the intensities of smaller and larger alleles.). If the LOH index was less than0.5or more than2.0, then define the case as a potential LOH site. TP53LOH was defined when there were two or more than two potential LOH sites in each tissue. This method had been validated using the25aCGH analyzed tumor and non-tumor tissues.The relative expression of genes surrounding viral insertion sites were quantitatively analyzed by qRT-PCR. The cccDNA level was also analyzed in some of the paired samples. All statistical analyses were performed using the SPSS14.0for windows.Results:HBV DNA integration showed no difference either in frequency or chromosome distribution between tumor derived and the corresponding non-tumor derived samples.A total of287different inserted sequences were identified amongst the88% (53/60) integration positive patients. Among them,84viral-host junctions being identified in68%(41/60) tumor derived samples and148in72%(43/60) non-tumor derived samples. Multiple integration events were found in50%(22/41) of the HCCs and91%(39/43) of the non-tumor derived samples. Amongst the viral-host junctions identified,233could be precisely mapped to chromosomes, of which81were from tumor derived tissues and152from non-tumor samples. The remaining54virus-host junctions could not be uniquely mapped due to repetitive or unidentified sequences. Using HBV captured deep sequencing method,292different viral-host junctions were found (such junctions were more than8reads supported). Among them,201were in tumor tissues while91in non-tumor tissues.As expected, larger chromosomes harbored more integration. However, when normalized to the number of integrations per108base pairs, no obvious preference of chromosome was observed in either tumor derived or in non-tumor samples. Amongst the233precisely mapped viral insertion sites,64were found to lie within a known fragile site. Amongst the233precisely mapped viral insertion sites,64were found to lie within a known fragile site. The following fragile sites were found being hit more than once:FRA19B in5cases, FRA1A, FRA5A, and FRA7J each in4cases, FRA7B and FRAIL in3cases, FRA1F, FRA12A, FRA3D, FRA6D and FRA19A in2cases. Interestingly, a significant disparity in the frequency with which fragile sites were mapped occurred between the tumor derived and non-tumor samples was observed (31/81sites in tumor vs.33/152sites in non-tumor, P=0.0077).Cellular gene containing areas in the human genome were the favored target site for HBV integration.Alignment analysis using the UCSC blat revealed that47%(110/233) of the viral insertion sites mapped were in introns (35/81in tumor vs.75/152in non-tumor) and4%(10/233) fell in exons (4/81in tumor vs.6/152in non-tumor). The remaining48% (113/233) were mapped to non-coding regions of the human genome (43/81in tumor vs.70/152in non-tumor). In addition,11of82integration sites mapped from tumor derived samples fell within+5kb of transcription start sites whilst the same was true for only15of the152integration sites mapped for non-tumor derived samples. These data indicate that, promoter, exon and intron areas in the genome are the favored target sites for HBV integration. Of course, it is possible that direct inserted into a gene area could affect the function of the targeted gene. In the HBV captured deep sequencing technique group,27%(78/292) of the viral insertion sites mapped were in introns (58/201in tumor vs.31/91in non-tumor) and4%(11/292) fell in exons (7/201in tumor vs.4/91in non-tumor). The remaining48%(113/233) were mapped to non-coding regions of the human genome (43/81in tumor vs.70/152in non-tumor). However, no significant differences of integration preference were observed between the tumor derived and the non-tumor derived samples.In the present study, insertion in or around the hTERT gene was found in3tumor derived samples. This observation provides additional evidence that hTERT is frequently hit by HBV integration. HBV insertion directly into the FN1gene was also found in12cases. However, only2of these were found in the tumor derived samples.No difference in the viral break point pattern was found between tumor derived and non-tumor derived samples.Sequence analysis of the287inserted viral fragments showed that163of them harbored partial X gene sequences and121harbored preC/C sequences, leaving3that contained neither the X gene nor preC/C sequences. As previously reported, most of the break points occurred around the DR1site. This is the first study in which large numbers of inserted preC/C sequences have been found in HCC patients.About75%of the break-points mapped between nt1601and nt1834of the viral genome, with24%(68/287) of them being located in the5’CTTTTT-3’ topoisomerase motif (1820to1825nt) and the DR1region (1824-1834nt). The DR2(1590-1600nt) region was rarely found as the break point. Therefore, overall the data indicated that the topoisomerase I motif and the DR1region of the viral genome were the preferred HBV genome break-points in the mapped integration sites, but failed to reveal any difference between those from tumor derived and non tumor derived samples.Comparative analysis of mutations in the inserted viral DNA failed to reveal any difference between tumor derived and non-tumor derived samples.Meta-analysis of our own previously published data and that of others has shown that the number of mutations of the HBV genome (C1653T, T1753V and A1762T/G1764A) gradually increased with disease progression and correlated with hepatocarcinogenesis.Direct sequencing of PCR amplicons generated from both tumor derived and non-tumor derived integrated viral sequences (integrated group) revealed that they were carrying fewer point mutations than were found in ’free’ non integrated HBV genomes. The frequencies of C1653T, T1753V and A1762T/G1764A mutations found were at the same level as that seen in the serum derived samples of the CHB group, and were significantly lower than that in the LC and HCC group. A1.1kb amplicon (from nt1264-2362) from ’free’ non integrated HBV DNA was successfully amplified and sequenced from a total of30tumor derived and34non-tumor derived samples (Table2). As expected, the frequencies of C1653T, T1753V, A1762T/G1764A mutations in these ’free’ HBV DNAs from both tumor derived and non-tumor derived samples were at the same level as that for the serum derived samples of the LC group and HCC group. It is worthy to note that the frequency of either C1653T or A1762T/G1764A mutations seen was significantly higher than that found for the integrated group (C1653T,9%vs.22%, P=0.0110; T1762/A1764,78% vs.54%, P=0.0060). This significant difference remained when the frequencies of the T1762or A1764point mutation (P<0.05) were separately compared.None of134confirmed X region insertions sequenced carried a whole X gene, with all of the inserts having a31terminal truncation. No significant difference was found for these or the other mutations described above between integrated viral genomes in tumor derived and non-tumor derived samples.Rearrangements of host DNA surrounding integration sites was a rare event in both tumor derived and non-tumor derived samples.The position of the sequencing primer on the X gene side of the break-point in the viral genome has a sufficient distance to the viral-host junction, so that viral genome rearrangements including deletions, inversions and duplications of viral sequence could be observed in the integrants. These rearrangements were found in15of58insertions in tumor and13of106insertions in non-tumor respectively (x2=4.8958, P=0.0269)(Figure3). In contrast the position of the preC/C sequencing primer was much closer to the viral-host junction which meant that the integrated viral sequences were too short for a similar analysis for viral genome rearrangements to be carried out Sequencing of the host cell genome close to the viral-host junction revealed only a few micro-deletions, micro-insertions, point mutations and translocations, with no significant difference being found between tumor derived and non-tumor derived samples.We found that HBV inserted in a specific area more than two times in one sample by using the HBV captured deep sequencing method. However, it showed no difference between tumor group and non-tumor group.The number of chromosome aberrations found in tumor derived samples did not correlated with HBV integration.A comprehensive aCGH assay was used to analyze host cell chromosomal abnormalities in25individuals from the60recruited patient cohorts, selected to allow any effect of the number of HBV insertion events on chromosomal aberration to be examined. In each assay, material obtained from corresponding adjacent non-tumor tissue was used as the reference control. The number of HBV integration events detected in across25tumor samples analyzed ranged from0to11and the total number of genome aberrations (gain and loss) detected in the assay ranged from11to537. No correlation was found between the number of genome aberrations identified and the number of HBV integration in the tumor samples analyzed (P=0.6520).The TP53status in all60patients was then analyzed. Twenty-one TP53point mutations were found in20(33%) of the tumor derived tissues, including18single nucleotide missense mutations and1single nucleotide synonymous mutation. One sample contained2point mutations at TP53191eu and241ys. Amongst the point mutations found10of them were AGG to AGT transversion at codon249of TP53. The majority of the point mutations were located in exon7(62%,13/21), with2mutations in exon2,2in exon4,2in exon5, and1in exon8, respectively. In contrast, no mutational changes in TP53were found in the paired adjacent non-tumor derived samples.LOH of TP53was successfully assayed in51of the60samples. The remaining9samples had no informatic SNP data and therefore could not be assayed for LOH. Twenty-two of the51(41%) patients successfully assayed showed LOH for TP53. TP53point mutation and LOH concurrence was present in20%(12/60) of the patients. Overall mutational change (point mutation and/or LOH) causing loss of TP53function was detected in48%(29/60) of the recruited patient cohort. No correlation was found between mutational change in TP53and the number of HBV integration events.Interestingly, among the25patients whose samples were subjected to aCGH analysis9of the11tumor derived samples harboring TP53mutations were found to have higher numbers of chromosomal aberrations. In contrast,7of the14tumor derived samples that had no mutations in TP53was found lie amongst those with a higher number of chromosomal aberrations.The mutational status of various tumor suppressor genes such as retinoblastoma1(RBI), TP73, cyclin-dependent kinase inhibitor2A (CDKN2A), breast cancer1, early onset (BRCA1), BRCA2, TP53and TP53BP2(tumor protein p53binding protein,2) in the patient samples subjected to aCGH analysis was also examined (Table3). As expected, a positive correlation was found between the number of specific tumor suppressor gene mutations identified and the number of whole chromosomal aberrations observed (r=0.6625, P=0.0003).The relative expression levels of genes successive to HBV integrationExcept hTERT gene,9other genes successive to the integration sites were randomly selected to determine if their expression was affected by HBV DNA insertion. Using C-terminal binding protein1(CTBP1) gene as an internal control, in comparison with the adjacent non-tumor tissue, the mRNA levels of DNLZ, SNAPCA4, and ATP8B2in tumor tissues with HBV integration nearby were detected with more than2fold up-regulation. The remained6genes showed no changes at the mRNA levels.The hTERT expression level was associated with the distance between viral insertion and the start site of gene. The hTERT up-regulation was found in2tumor tissues in which the HBV sequence inserted into the hTERT promoter regions. No difference of hTERT expression was observed in the third tumor tissues when compared with its paired non-tumor tissue, in which the HBV inserted into55Kb downstream of the hTERT with opposite orientation.Conclusion HBV integration events occurred at the early stage of disease progression. Instead of HBV integration, the status of several tumor suppressor genes might play a crucial role in the number of chromosome aberration. With exception of a significant high frequency of fragile sites hit in tumor group, all other properties of HBV insertion were highly similar between tumor and non-tumor. Our data failed to demonstrate a strong co-relationship between HBV integration and hepatocarcinogenesis. |
PDF Full Text Request