High-resolution phylogenetic analysis of hepatitis C virus adaptation and its relationship to disease progression

Abstract

Hepatitis C virus (HCV) persists in the majority of those infected despite host immune responses. Evidence has accrued that selectively fixed mutations in the envelope genes (E1 and E2) are associated with viral persistence, particularly those that occur within the first hypervariable region of E2 (HVR1). However, the individual amino acid residues under selection have not been identified, nor have their selection pressures been measured, despite the importance of this information for understanding disease pathogenesis and for vaccine design. We performed a high-resolution analysis of published gene sequence data from individuals undergoing acute HCV infection, employing two phylogenetic methods to determine site-specific selection pressures. Strikingly, we found a statistically significant association between the number of sites selected and disease outcome, with the fewest selected sites in fulminant HCV cases and the greatest number of selected sites in rapid progressors, reflecting the duration and intensity of the arms race between host and virus. Moreover, sites outside the HVR1 appear to play a major role in viral evolution and pathogenesis, although there was no association between viral persistence and specific mutations in E1 and E2. Our analysis therefore allows fine dissection of immune selection pressures, which may be more diverse than previously thought. Such analyses could play a similarly informative role in studies of other persistent virus infections, such as human immunodeficiency virus.

FIG. 1.

Summary of the estimated number of selected sites in each data set. (a) Ray et al. (19) data set. The figure shows the number of sites with a >90 or >95% posterior probability of being selected, estimated by using CODEML. More selected sites were found in sequences from patients with progressing HCV infection. (b and c) Farci et al. (9) data set. (b) Proportion of codons under selection in at least one patient. (c) Proportion of instances (number of patients × number of codons) under selection. There is greater evidence for positive selection in HVR1 than in the surrounding regions. Selected sites were identified by using both MacClade and CODEML.

FIG. 2.

Site-by-site analysis of positive selection in the Farci et al. (9) data set. Selected sites were identified by using both MacClade and CODEML. (a) Number of patients that showed positive selection at each codon position. The shaded area represents HVR1. E1 is to the left of this region, and E2 is to the right. (b) A close-up of the results for HVR1, revealing heterogeneous selection among sites within the HVR.

FIG. 3.

Relationship between disease progression and the number of putative selected sites identified in each patient for the Farci et al. (9) data set. Sites were identified by using both MacClade and CODEML. (a) Number of selected sites per patient in the whole region sequenced. (b) Number of selected sites per patient in E1 and E2, excluding HVR1.

FIG. 4.

Viral adaptation in patient 10 of the Farci et al. (9) data set. Samples from this patient, who rapidly progressed to chronic hepatitis, were obtained at 5 time points during acute infection, indicated by the letters A to E, which correspond to weeks 3, 8, 13, 16, and 21 after infection, respectively. (a) Parsimony reconstruction of amino acid changes at codon 348, superimposed upon the ML phylogeny. Arrows indicate amino acid changes. Three mutations from isoleucine (black) to valine (red) occur between time points C and D, of which one spreads to fixation at time point E. One other equally parsimonious reconstruction is possible (data not shown). (b) Location of the 33 selected sites in patient 10. MacClade identified all 33 sites, whereas CODEML detected only 18 of these. Codon 348 is marked by an asterisk.