Hepatitis C virus quasispecies in chronically infected children subjected to interferon–ribavirin therapy

The studies included two groups of children with chronic hepatitis C (all infected with HCV-1a). The first group included 10 patients subjected to combined interferon–ribavirin therapy in 2004 (patients designated P1-01 to P1-10). The second group comprised 13 patients who were treated in the same way, but 1 year later (patients P2-02 to P2-28). The study conforms to the ethical guidelines of the World Medical Association Declaration of Helsinki. In each case, both parents and children (12 years or older) provided informed consent for treatment and participation in the study. Before the commencement of the project, an appropriate approval from the Bioethical Commission of Karol Marcinkowski University of Medical Sciences was obtained. The inclusion criteria were as follows: (1) the detection of anti-HCV antibodies and HCV RNA in the serum for at least 6 months; (2) periodically or continuously elevated ALT activity; and (3) histopathological signs of chronic hepatitis [9]. The treatment was conducted according to the following schema: interferon alpha 2b was administered in 3 million unit doses subcutaneously 3 times a week; ribavirin, in doses of 15 mg/kg per day orally [9]. The level of HCV RNA accumulation was determined for each patient, just before therapy (T0) and then after 24 (T24), 48 (T48) and 72 (T72) weeks, using an Amplicor HCV Monitor Test v. 2.0 (Roche). The patients with a positive result (a detectable amount of HCV RNA) at T24 were excluded, whereas the others were treated for 24 more weeks. After 72 weeks of treatment and follow-up, patient response to combined interferon–ribavirin therapy was evaluated. It was considered: (1) sustained (SR) if HCV RNA was not detected at T24, T48 and T72; (2) transient (TR) if HCV RNA was not detected at T24 but appeared again at T48 and was still present at T72 or HCV RNA was not detected at T24 and T48 and appeared at T72; (3) no response (NR) if HCV RNA was detected at T24, T48 and T72. Anti-HCV antibodies were detected in sera by fourth-generation ELISA (ETI-AB-HCVK-4, DiaSorin).

The structure of the HCV population was examined separately for each of 23 patients before therapy (T0) and 2 weeks after therapy had begun (T2). To this end, HCV RNA was isolated from the patients’ sera, and a 462-nt portion of the E1/E2 coding sequence (a fragment located in the HCV genome between positions 1425 and 1886) was amplified by the RT-PCR method [19] using a OneStep RT-PCR Kit (QIAGEN). All nucleotide positions refer to the standard HCV-1a genome sequence, accession number AF009606.

The following pairs of primers were used for RT-PCR:

The RT-PCR products were cloned into the pUC vector, and at least 20 randomly selected clones were sequenced per population. The nucleotide sequences obtained were translated in silico into amino acid sequences. A bioinformatic analysis was conducted separately for each population (represented by 20 randomly selected clones at each time point; over 900 sequences in total). This was done using the predicted amino acid sequence of the entire amplicon (excluding primer hybridization sites) or its fragments: hypervariable region 1 (HVR1, located in the HCV genome between positions 1491 and 1571) or the remaining portion of the amplified E1/E2 coding sequence (non-HVR1). The composition of the HCV quasispecies was established by direct comparison of the predicted amino acid sequences. As a result, the number of different variants within the population and the contribution of each individual variant to the examined population were established. Clones containing nonsense and frameshift mutations were excluded. Consequently, the number of variants analyzed was reduced for some populations. Genetic diversity was assessed based on the Hamming distance, which is defined as the number of amino acid differences between two sequences. The genetic diversity of a population was expressed as a mean Hamming distance (MHD), an average value taken for all sequence pairs. Statistical analysis was performed using the Statistica 8.0 package. The values of MHD determined for all patients were divided into two groups, (1) non-responders + transient responders and (2) sustained responders, and analyzed using the non-parametric U Mann–Whitney test (because the results of the Shapiro–Wilk test were negative). Phylogenetic trees were created using the maximum-likelihood method with 500 boostrap replicates. WAG and HKY85 substitution models were employed for amino acid and nucleotide sequences, respectively. Selection tests were made using the internal fixed effects likelihood (IFEL) and single likelihood ancestor counting (SLAC) approaches, available at www.​datamonkey.​org [23]. To count the global (SLAC) and codon-specific (IFEL) dN/dS ratios, the HKY85 model of nucleotide substitution and neighbor-joining trees [17, 28] were utilized, with a significance level value of 0.1. All available sequences of viral variants were analyzed. A consensus amino acid sequence was produced using the Consensus tool available at the Los Alamos HCV Sequence Database http://​hcv.​lanl.​gov/​content/​hcv-db/​consensus/​consensus.​html [16].

The complexity of HCV quasispecies was assessed twice for all 23 patients: just before treatment (T0) and 2 weeks later (T2). The number of different variants within each population and each population’s genetic diversity were determined. The analysis was carried out for the entire amplified E1/E2 region, and then separately for its two fragments: HVR1 and the remaining portion of E1/E2 (non-HVR1). Finally, three sets of data describing the complexity of HCV quasispecies (Online Resources 1, 2, 3, 4) were examined in relation to parameters characterizing the patients’ response to anti-HCV therapy, the level of HCV RNA accumulation in their blood and the level of ALT activity (Table 1). As a result, a significant correlation was only found between the outcome of antiviral therapy and HCV quasispecies complexity if the latter was determined at T0 and based solely on HVR1. In this case, HCV populations isolated from non-responders and transient responders were composed of a markedly lower number of variants (mean: 6,462; SD: 3,406) than populations derived from sustained responders (mean: 11; SD: 3,496; Fig. 1a; Online Resource 3). The level of genetic diversity (expressed by the mean Hamming distance: MHD) differed significantly (p < 0.05) between sustained responders and children who failed to clear the virus (non-responders plus transient responders). The obtained p values were equal to 0.0011 and 0.0039 for T0 and T2, respectively. The MHD was lower for HCV populations derived from non-responders and transient responders (mean at T0: 1,899; SD: 2,043) and higher for quasispecies coming from sustained responders (mean at T0: 6,152; SD: 2,486; Fig. 1b). Only for one population isolated from a child classified as a non-responder did MHD reach a relatively high value of 4.93. For two quasispecies derived from sustained responders, the MHD value was considerably below average: 0.79 and 3.14. The analysis performed for the whole amplified E1/E2 region revealed tendencies similar to those observed for HVR1. However, they were much less pronounced (Online Resource 4).

The 2-week interferon–ribavirin therapy differentially influenced the composition of the viral populations. At T2, an increase in the number of variants was observed in the majority of populations present in non-responders. A decrease was observed in those from sustained responders. A closer analysis of individual variants revealed diverse patterns of quasispecies evolution over the 2 weeks of treatment. Only in some populations was the contribution of the dominating variant increased. For others, we observed one of the following: (1) a lack of any substantial changes in the population structure; (2) the formation of new co-dominants; (3) the extinction of the dominant variant or (4) the development of new dominating variants that had not been identified before treatment (Online Resource 3). The influence of the applied therapy on MHD was quite diverse. It was not related to the patients’ response to antiviral treatment.

To determine not only the diversity of variants constituting each individual HCV population but also the relationships between them, phylogenetic trees were constructed for all of the quasispecies (Fig. 2; Online Resources 5, 6, 7). Again, the most significant correlation between the outcome of antiviral therapy and the character of HCV phylogeny was found for trees produced on the basis of HVR1 identified at T0. Trees depicting populations derived from non-responding children were compact, almost monophyletic. Interestingly, the only quasispecies from this group for which a visibly higher MHD value was obtained (P2-23) was found to be composed of two intrinsically homogeneous subpopulations. In contrast, trees based on populations isolated from sustained responders were very extended, without any clear clusters.

After the 2 weeks of treatment (T2), the topology of the phylogenetic trees remained mostly unaffected. Some changes could be observed almost exclusively for populations isolated from sustained responders. In these cases, a reduction or an increase in the degree of tree branching was observed. An analogous analysis based on the entire E1/E2 region or its non-HVR1 fragment did not yield conclusive results (Online Resources 5 and 6). This again confirmed that HVR1 best reflects differences between quasispecies obtained from children with different responses to anti-HCV therapy.

To evaluate the relationships between HVR1 variants present in different quasispecies at T0, an integrated phylogenetic tree was constructed. It was obtained based on all (445) variants identified in 23 HCV populations. While variants from sustained responders were noticeably dispersed, those derived from patients responding transiently and from non-responders tended to concentrate in clusters (Online Resource 8). Astonishingly, the variants isolated from ten non-responding children clustered into only seven groups. A more detailed analysis revealed that one of them, cluster 1, encompassed variants from five patients. Together with the closely related cluster 2, they constituted a pair grouping variants derived from six epidemiologically distinct viral populations (P1-01, P1-02, P1-03, P1-05, P2-05 and P2-19) (Online Resource 8; Fig. 3a). This observation suggested that HVR1 variants formed in different non-responding patients were often identical or quite similar. All of the analyses presented above were based on comparisons of amino acid sequences. To verify the results concerning inter-quasispecies conservation of HVR1, nucleotide sequences of variants grouping in clusters 1 and 2 were compared. Both phylogenetic analysis (Fig. 3b) and sequence alignment revealed population-specific differences occurring at the nucleotide level. Consequently, consensus sequences unique to each quasispecies could be generated (Fig. 4). This provided additional evidence that the similarity/identity between HVR1 variants was not caused by cross-contamination. These results suggest that the similar/identical HVR1 variants that were identified were already optimal: their appearance was strictly connected with the lack of patients’ response to the antiviral therapy. None of them were found in populations isolated from patients with transient and sustained response (Online Resource 8). These variants were present in viral populations before treatment and persisted at least through the first 2 weeks of its duration (Online Resource 3: identity of dominating variants between T0 and T2).

A viral quasispecies is subjected to natural selection throughout its existence within the host. Positive selection acts towards introducing nonsynonymous mutations and increasing the diversity of the population. In contrast, negative selection favors synonymous changes and restricts the process of diversification, keeping the population more homogenous. In order to determine what kind of selection pressure shaped the viral population in the group of patients studied, the nucleotide sequences of all HVR1 variants identified at T0 and T2 were compared. The analysis included only the HVR1 region because it best reflected differences between populations derived from patients with different responses to antiviral therapy. Initially, the nucleotide sequences of 904 HVR1 variants derived from 23 patients were divided (according to the sampling time and type of response to the antiviral therapy) into six groups: NR-T0, NR-T2, TR-T0, TR-T2, SR-T0, SR-T2 (NR—no response, TR—transient response, SR—sustained response). First, the number of synonymous and nonsynonymous substitutions per site (dS and dN, respectively) was established. Then, the global and codon-specific dN/dS ratio (parameter characterizing selection pressure) was determined (Table 2). Values lower than 1 are characteristic of negative selection; values higher than 1, of positive selection. Under positive selection, only 2 and 4 codons were identified within the sequences of variants forming the NR-T0 and NR-T2 groups, respectively; under negative selection, 4 and 5 codons were identified in the same sequences. A similar trend was detected among HVR1 sequences from the TR-T0 and TR-T2 groups. In contrast, as many as 10 codons were subjected to positive selection, and only 2 were subjected to negative selection in variants belonging to the SR-T0 group. In SR-T2, 9 codons were found to be under positive selection and 4 codons were under negative selection. The global character of selection within the entire HVR1 was negative in non-responders and transient responders. It was strongly positive in sustained responders. After 2 weeks of therapy, the number of selected sites increased in all three groups, but the character of the pressure operating globally in HVR1 remained unchanged. This observation clearly indicates that selection pressure is the key modulator of HCV quasispecies diversity. Strikingly, the commonly described process of continuous selection and adaptation was apparently ongoing only in sustained responders. In contrast, the evolution of the quasispecies in non-responding patients seemed to be rigorously confined, most probably by the necessity for maintaining defined HVR1 properties.