Patients and blood samples

Serological HBV marker detection

Serological markers were detected by electrochemiluminescence immunoassay on a Roche E170 modular immunoassay analyzer following the manufacturer's protocols (Roche Diagnostics, Germany).

HBV DNA quantification

Real time PCR was performed to determine viral DNA titers using an FQ-PCR Kit for HBV (DaAn Gene Co., China) performed in a GeneAmp 5700 Sequence Detection System (PE Applied Biosystems, USA).

Viral DNA extraction

Viral DNA was extracted from 200 μL sera per sample using an AccuPrep Genomic DNA Extraction Kit (Bioneer, Korea) or QIAamp MinElute Virus Spin (Qiagen, Germany) as instructed in the manufacturer's manuals. All DNA samples were stored at -20°C before thawing for PCR.

Fragment amplification

All PCRs were performed in 20 μL or 50 μL reaction mixtures containing specimen DNA. PCR for amplification of target regions was performed with a hot start at 95°C for 150 s, followed by 35 cycles of denaturation at 94°C for 1 min, annealing at 58°C for 90 s, and elongation at 72°C for 3 min. All reactions were performed on a PTC-200 Peltier Thermal Cycler (MJ Research, USA). The primers employed were CP5: 5'-CTTCGTCTGCGAGGCGAGGG-3' (nt 2381-2400) and CP22: 5'-GAGACCACCGTGAACGCCCA-3' (nt 1611-1630).

PCR and clone sequencing

PCR products were purified with a Montage PCR96 column (Millipore, USA). The final DNA concentration in the sequencing reaction was 10 ng/μL. DNA sequencing was carried out on a Prism 3730 (ABI, USA). Contigs were assembled using SeqMan (DNASTAR, USA) and all the sequences were aligned by ClustalW for further analysis (all sequences analyzed in this paper have been submitted to GenBank) (Additional file 1, Table S3).

The PCR products were first purified with a Takara gel purification kit and were then ligated into pMD18 T vector (Takara Bio, Japan). Vectors were subsequently transfected into DH5α cells and white/blue colony selection was used to detect recombinant vectors. Inserts from positive clones were PCR amplified with primers RV-M/M13-47 according to the manufacturer's instructions to verify the target fragments.

Viral genotyping

The phylogenetic tree (Additional file 1, Figure S1) was built for all 157 sequences and 858 strains from 8 genotypes retrieved from the NCBI database, using Mega 4 [23] and annotated by TreeDyn [24]. All the fragments of BCP and preC were from whole genomes in the NCBI database with genotypes annotated.

Statistical Analysis

Statistical analysis was performed using SPSS software (version 13.0; SPSS Inc, USA). Logistic regression was used for evaluating the roles of candidate clinical factors and viral mutations in liver disease progression and HBeAg negativity.

High prevalent mutations in BCP and preC regions

A total of 300 viral fragments from nt 1725 to 1900, covering the BCP and preC regions in the HBV genome, were analyzed. As shown in Table 1, direct sequencing resulted 157 fragments from patients of 13 asymptomatic carrier (ASC), 75 chronic hepatitis (CH), 44 liver failure (LF), and 11 hepatocellular carcinoma (HCC). LF, LC, and HCC were defined as advanced liver diseases (ALD). The rest 143 sequences came from clone sequencing in which 95 clones were from 17 samples of LC patients and 48 from 18 samples of HCC patients. Figure 1 illustrates the nucleotide substitution patterns identified by PCR sequencing. Similar mutation profiles were shown in samples collected in Beijing (136) and in a relatively isolated town 600 km south from Beijing (21). Mutations were rarely seen in gene overlapping regions. About 82% (128/157) sequences were genotype C and the rest were type B by clustering with reported sequences in NCBI database (Additional file 1, Figure S1). There were 12 nucleotide substitutions with their mutation rates over 5% (Additional file 1, Table S2). Further comparison of these sites with 233 genotype B and 311 genotype C of HBV sequences from NCBI database demonstrated that five were at the genotype specific positions and were therefore precluded from further analysis (Table 2). Interestingly, in three previously reported common mutations G1764A, A1762T, and G1896A in this region, the first two types were observed with very high prevalence in these samples (70% and 67%, respectively) (Additional file 1, Table S2).

G1896A mutation in liver disease progression

G1776A statistically correlated to HBeAg negativity

Common multi-mutations in BCP and preC regions

Next we analyzed the effect of multi-mutations in target region. First we defined multi-mutations detected in a single fragment as the mutation combination or the combination in short. Five point mutations (1753, 1762, 1764, 1846, and 1896) with their rates larger than 10% were included in combination analysis. Nineteen categories containing wild type, single base mutations and observed combinations were resulted as illustrated in Figure 2. The top three combinations in all 192 patients were the double mutation A1762T/G1764A (36%), the triple mutation A1762T/G1764A/G1896A (11%), and the quadruple mutation T1753(A/C)/A1762T/G1764A/G1896A (8%).

While comparing the results of PCR sequencing and clone sequencing, we noticed an interesting phenomenon. All the mutation combinations observed in PCR sequencing were also detected in single strains obtained by clone sequencing, suggesting that point mutations tended to coexist in single genomes rather than to occur in various fragments among viral quasispecies (Figure 2). In addition, these combinations were also seen in NCBI sequences of both genotype B and C (Figure 3).

Multi-mutations in ALD patients

As shown in Figure 2, the distribution of above multi-mutations was different between ALD and other patients (ASC and CH) (FET, P < 0.0001). Patients with the combinations of more than 3 mutations were more likely to have ALD (OR = 3.1, 95% CI: 1.6-6.0, P = 0.001), although only one combination (T1753(A/C)/A1762T/G1764A) was dominant in these patients (FET, P = 0.025). Furthermore, ALD patients (mean age 46 ± 11) were older than others (mean age 33 ± 12; T-Test, P < 0.0001) and the patients with no less than 3 viral mutations (mean age 42 ± 12) were also older than the rest (mean age 36 ± 13) (P = 0.009), suggesting that the age, which usually represent the infection history appears to be a correlating factor in liver disease progression.

Candidate mutations responsible for the loss of e antigen

Previous studies have suggested high mutations in BCP and preC regions are major cause for the loss of HBeAg but without true disappearance or reduction of viral load. Thus far only G1896A, which leads a 2/3 truncated e antigen by a stop codon caused by mutation, was verified to be responsible for HBeAg negativity. In this study, we showed that G1776A is a new candidate mutation correlating to HBeAg negativity. Since nt1776 localizes at the binding site to transcription factor FXRalpha, which promotes the expression of e antigen [25], one possible mechanism for G1776A mutation is to interrupt normal transcription of preC region. As this is the first time proposing possible role of G1776A, further verification is needed in larger sample size.

Viral mutation accumulation in patients with long history of infection

Evidenced by the appearance of some mutations earlier than others during the infection, the accumulation of mutations in viral genome had been observed in previous studies [26]. Our results suggested these accumulated mutations fell into two categories. Mutations in first type were adaptive to host system after long infection but had little effect in the disease progression. A feature of those mutations, including T1753A/C, G1764A and A1762T, is their significance as risk factors in disease progression only shown by univariate analysis but not in the multivairate model when patient age was considered. The second type, such as G1896A, has high occurrence in ALD patients and is independent from infection history, suggesting mutants carrying this substitution may lead to worsening symptoms once they appear.

The occurrence rates of T1764 and A1762 were very high in our study (70% and 67%, respectively). This is perhaps because the major path of HBV infection is vertical transmission in China and patients in their ages of 30 s are usually already with long infection history [9]. The effects of G1762A and A1764T in disease progression had attracted attentions especially in genotypes B and C [27, 28], however our study suggests that the appearance of these mutations may just be a sign of long infection history and may not be very important in disease prognosis.

Common combinations of prevalent substitutions in BCP and preC

Few studies have been conducted about combination of common nucleotide substitutions, and here we present multi-mutations with high frequencies in patients with various diagnoses. Mutation combination appeared as a common type of variation in HBV genome regardless of genotypes. Despite many possible patterns, only 13 combinations were observed based on 5 common substitutions in our study. These limited multi-mutation patterns indicate only certain combinations may have selective advantage for viral functions which makes it possible to interrogate their effects in disease progression. However, limited samples for each mutation combination reduce the statistic power to dissect their distribution in patients with different diagnoses. Therefore, larger sample size should be considered in further investigation