Hepatitis B virus genotype G epidemiology and co-infection with genotype A in Canada

Footnotes

The GenBank/EMBL/DDBJ accession numbers for the sequences reported in this paper are EU833889, EU833890 and EU833892–EU833930.

A supplementary figure is available with the online version of this paper.

HBV/G-infected individuals were identified through the results of routine molecular diagnostic service at the National Microbiology Laboratory (NML) and through the Canadian enhanced hepatitis strain surveillance system (EHSSS). Clinical, demographic and risk factor data were obtained from the sending laboratory or physician. HBV DNA was extracted from clinical samples by silica gel filtration (QIAamp DNA Blood Mini kit; Qiagen) or by phenol/chloroform extraction methods to optimize sensitivity (Osiowy, 2002). Extracted DNA was amplified for genotyping, clonal and phylogenetic analysis using hepatitis B s-antigen (HBsAg)/polymerase coding region-specific primers: Spr1A (5'-GTTCAGGAACAGTAAGCCC-3') and antisense (5'-GAAAGGCCTTGTAAGTTGGCG-3'). Where necessary, the following primers were used for nested PCR: sense (5'-GGTGGACTTCTCTCAATTTTCTAGG-3') and Spr2A (5'-ACTTTCCAATCAATAGGCC-3'), using conditions described previously (Osiowy et al., 2006). Full-length genome amplification of certain samples was performed using a thermostable DNA polymerase blend (Expand High Fidelity^Plus System; Roche) and primers P1 (5'-CCGGAAAGCTTATGCTCTTCTTTTTCACCTCTGCCTAATCATC-3') and P2 (5'-CCGGAGAGCTCATGCTCTTCAAAAAGTTGCATGGTGCT GGTG-3'), as described by Günther et al. (1995). The precore region of each co-infecting HBV/A strain was also amplified using the primers and methods described by Takahashi et al. (1995), to determine the presence of a mutation at nt 1896. The HBV viral load for samples from chronically infected patients was determined by the NML or the provincial public health laboratory from which the sample originated. Viral load was determined by using either the Digene Hybrid Capture System (Digene) in copies ml^–1, the Versant HBV bDNA Assay (Siemens Healthcare Diagnostics) in copies ml^–1 or the Cobas TaqMan Assay (Roche Molecular Systems) in IU ml^–1. The approximate conversion factor for copies ml^–1 to IU ml^–1 is 5 (Saldanha et al., 2001).

Amplicons were gel-purified prior to being sequenced or cloned using a TA-based ligation method (pDrive cloning vector; Qiagen). The directly amplified product and approximately 20–30 clonal inserts were sequenced to detect majority and minority genotype populations and to estimate their ratio. The genotype of each directly amplified and clonal sequence was initially determined by the NCBI HBV Genotyping Tool () (Rozanov et al., 2004). In order to detect minority populations in directly amplified samples in a more sensitive manner to sequencing, reverse hybridization probe analysis using the line probe assay (LiPA) kit (INNO-LiPA HBV Genotyping; Innogenetics) was also performed on all directly amplified sequences, according to the manufacturer's instructions.

Directly amplified sequences were assembled and analysed using DNA sequence analysis software (Lasergene software suite v7.1.0, DNASTAR). Sequence alignments were performed using CLUSTAL_X v1.8 (Thompson et al., 1997) and trimming of alignment ends to obtain the equivalent length sequences for each sample (411 bp; nt 261–671 of the HBsAg gene) was performed using BioEdit v5.0.9 (Hall, 1999). Sequence divergence was calculated based on the number of nucleotide changes per total number of nucleotides analysed (411 bp) in the alignment using MEGALIGN software (Lasergene v7.1.0). Pairwise genetic distances were estimated by Kimura two-parameter analysis and a phylogenetic tree was constructed by the neighbour-joining method with 1000 bootstrap replicates using MEGA software v4 (Tamura et al., 2007). Bootscan analysis software (SimPlot v3.5.1) was used to identify intergenotypic recombination (Lole et al., 1999). The full-length HBV genome sequences of patients 5 and 6 were queried against six GenBank reference sequences for each of genotypes A and G (potential parentals) and six GenBank reference sequences for genotype D (outgroup control). One or two GenBank reference sequences for all other HBV genotypes were also included to assess informative sites for evidence of recombination with genotypes other than the suspected parentals.

Thirteen patients infected with HBV/G were identified; three acutely infected patients were identified through the EHSSS, while 10 chronic HBV patients were identified through diagnostic HBV genotype testing. The HBV/G strains represented approximately 1.2 % (3/244) of all acute cases and 2.3 % (11/482) of all chronic HBV cases genotyped at the NML over a 6 year period, from 2001 to 2007. However, as initial detection of cases was by amplicon direct sequencing, samples with levels of HBV/G below the limit of sequence detection may have been missed. A total of 39 patient samples were collected during this period; 6 month consecutive samples were available for seven patients (33 of 39 samples).

HBV/G was observed in patients from three different provinces in Canada (Table 1). The majority of infections occurred in males (11/12; gender, age and risk factor was unavailable for one patient) having a median age of 46.5 years. The single female patient was 25-years-old and acutely infected. At the time of testing, seven of the 10 chronic patients were HBeAg-positive and eight patients were receiving adefovir therapy added to lamivudine therapy (all were genotypically and phenotypically resistant to lamivudine). Most HBV/G-infected chronic patients had significant viral load, which appeared to decrease over time in those patients on therapy (Table 1). Male homosexual sex (MSM) was the predominant risk factor associated with HBV/G infection (8/12; 67 %).

Table 1. Demographic, clinical and risk factor association of Canadian HBV/G Patient numbers in bold refer to those patients for which consecutive 6-month samples were available. NA, Data not available.

Phylogenetic analysis was performed on all 39 directly amplified HBsAg sequences (GenBank accession nos EU833892[GenBank] –EU833930[GenBank] ) to determine interrelatedness within Canadian HBV/G isolates and among HBV/G strains throughout the world. Canadian HBV/G strains all clustered within the HBV/G clade (Fig. 2), although for several patient time point samples, only the HBV/A sequence was directly amplified (10c–e and 11i). These sequences clustered with GenBank-derived HBV/A strains, although HBV/G was detectable in each sample by LiPA analysis but not by sequencing of the directly amplified product. Phylogenetic analysis of the HBsAg-coding region of co-infecting clonal sequences determined that all HBV/A co-infecting strains were subgenotype A2 (data not shown).

Despite the apparent low prevalence of HBV/G throughout the world, the viral genotype appears to circulate globally, as it has been reported in several European countries (De Maddalena et al., 2007; Jardi et al., 2008; Lacombe et al., 2006; Tran et al., 1991; Vieth et al., 2002), North and South America (Alvarado-Esquivel et al., 2006; Bottecchia et al., 2008; Kato et al., 2004; Osiowy & Giles., 2003; Sánchez et al., 2007), Nigeria (Olinger et al., 2006) and Asia (Shibayama et al., 2005; Suwannakarn et al., 2005; Toan et al., 2006). Canadian HBV/G was highly associated with MSM as a risk factor for infection, corresponding to previous studies (Bottecchia et al., 2008; Shibayama et al., 2005; Sánchez et al., 2007). When information is available regarding HBV/G co-infection, the co-infecting genotype is normally that which is highly prevalent within the geographical region, i.e. genotype A in Europe and North America, genotype H in Mexico and genotype C in Vietnam, thus suggesting that HBV/G is capable of co-infecting with any other HBV genotype. Results of the present study show that Canadian HBV/G infection occurs exclusively in the context of HBV/A co-infection, although genotypic dominance and the genotype population ratio appear to fluctuate over time. It is important to note that the patients, for whom longitudinal samples were available, were all being treated with adefovir and lamivudine. Antiviral therapy may have influenced the quasispecies population to initiate a switch in genotypic dominance. This has been suggested following the selection of specific strains during treatment of HBV/A/D co-infected patients (Hannoun et al., 2002; Jardi et al., 2008).

Clinical monoinfection with HBV/G has been suggested following LiPA analysis of an acutely infected blood donor and the corresponding platelet recipients (Chudy et al., 2006), and by LiPA analysis of Mexican (Alvarado-Esquivel et al., 2006) and European (Pas et al., 2008) chronic carriers. HBV/G has also been shown to be capable of replication in the absence of a co-infecting genotype following in vitro transfection of Huh7 cells (Li et al., 2007) or uPA/SCID mouse infection (Sugiyama et al., 2007; Tanaka et al., 2008), albeit at close to undetectable levels. However, results of the present study suggest that the observation of monoinfection by LiPA or clonal analysis is due to the co-infecting genotype falling below the level of detection sensitivity. In both cases of apparent monoinfection, the patient was determined to be co-infected at an earlier and later time point during infection. This observation, together with the findings from uPA/SCID mouse studies, suggests that chronic infection with HBV/G can only occur in the context of co-infection with another HBV genotype, and thus would be transmitted as such.

Recombination between the co-infecting genotypes was frequently observed among the genotypic population of each patient, and HBV/G/A was present as the dominant strain in two patients. Recombinant genomes demonstrated variable breakpoints within the polymerase, preS1 and HBsAg coding regions, as well as within the X gene. Several of these regions have been described as putative hot spot sites of recombination (Simmonds & Midgley, 2005; Yang et al., 2006), although a breakpoint within the X coding region has not yet been described.

HBV genotype G has been described as being genetically homologous due to the observed low nucleotide divergence among globally distributed isolates (Lindh, 2005; Tanaka et al., 2008). However, the Canadian HBV/G HBsAg-coding sequences (411 bp) examined in this study were found to have a mean genetic distance over twofold greater than HBV/G sequences described to date (0.017 versus 0.007). Increased sequence divergence may be the result of antiviral therapy, as an increased mutation rate within the HBsAg/polymerase overlap region is known to occur during treatment with lamivudine (Conzelmann et al., 2005). This selection pressure likely contributes to increased spontaneous mutation as well as quasispecies selection of more fit minor strains. However, phylogenetic analysis that included only the first available sample from each of the 13 patients (data not shown) still resulted in a mean intragroup distance of 0.015. Therefore, global intragenotypic HBV/G homology, at least within the HBsAg-coding region, may be closer to 98.6–98.9 %. This remains a very high sequence identity among such disparate strains, with respect to geography and co-infecting genotypes, and thus raises the interesting question as to the mechanisms controlling HBV/G evolution.