Acceleration of hepatitis C virus envelope evolution in humans is consistent with progressive humoral immune selection during the transition from acute to chronic infection

Abstract

During the transition from acute to chronic infection in individuals persistently infected with hepatitis C virus (HCV), cellular responses initiate within the first 6 months of primary infection and collapse thereafter, whereas humoral responses activate later during the chronic phase. Whether and how this deviation of immune responses specifically influences HCV evolution are unknown. To determine the pattern of HCV evolution during this critical period, we conducted extensive sequence analysis on annual clonal hemigenomic sequences for up to 3 years in six well-characterized subjects, using statistical methods that accounted for repeated measures. Significantly different evolutionary rates were observed in envelope versus nonenvelope genes, with an increasing rate of nonsynonymous change (dN) in envelope genes and a stable dN in nonenvelope genes (P = 0.006). The ratio of the envelope to nonenvelope nonsynonymous rate increased from 2 in year 1 to 5 in years 2 and 3. Centripetal changes (reversions toward matching of the worldwide subtype 1a consensus sequence) were frequently observed during the 3-year transition from acute infection to chronicity, even in the presence of neutralizing antibody (NAb) pressure. Remarkably, sequences of hypervariable region 1 (HVR1) remained stable for up to 21 months in the absence of NAb pressure in one subject, followed by rapid changes that were temporally associated with the detection of NAb responses, which strongly suggests that HVR1 evolution is shaped by NAb pressure. These data provide the first systematic estimates of HCV evolutionary rates in multiple genes during early infection in vivo and provide additional evidence for deterministic, rather than random, evolution of HCV.

FIG. 1.

HCV RNA and ALT levels of six subjects during the 3-year transition from acute to chronic infection. The time of initial viremia (time zero) was calculated as the midpoint between HCV RNA-negative and -positive visits. ALT levels prior to initial viremia were available for all but subject 56 (S56). For subject 65, no serum samples were available for more frequent ALT testing between initial viremia and the year 1 visit.

FIG. 2.

Phylogenetic trees of hemigenomic and partial E1E2 sequences. The center unrooted maximum-likelihood tree demonstrates the phylogenetic relationship among the hemigenomic sequences from different subjects. All sequences are represented by circles except for those of subject 30, whose sequences are represented by triangles to differentiate them from epidemiologically linked subject 29. Subtype 1a, 1b, and 1c sequences obtained from GenBank are included as references, and accession numbers are given, except for H77, for which the accession number is AF009606. For each study subject, a maximum-likelihood tree was constructed based on partial E1E2 sequences from approximately 20 clones for each visit, with 103 reference sequences and genotype 7 used as an outgroup. For clarity, only the clade containing the subject-specific sequences and nearest reference sequence neighbors is shown.

FIG. 3.

Rate of nonsynonymous evolution during 3 years of follow-up after primary HCV infection. Plotted are the rates of nonsynonymous change in each gene of the HCV hemigenome, represented by a median value from six individual subjects (A), and in envelope (E1 and E2) versus nonenvelope (core, p7, NS2, and NS3) genes for each individual, with medians represented by horizontal lines (B). Rates are plotted along the horizontal axis at the midpoint between the two annual visits being compared. The difference between E1E2 and non-E1E2 rates was significant (P = 0.006) based on a GEE model that corrects for multiple measurements per individual.

FIG. 4.

Forward (centrifugal) and reverse (centripetal) amino acid substitutions during the transition from acute to chronic HCV infection. HCV viral hemigenomic sequence comparisons are depicted as horizontal bars, with the time interval shown on the left. Amino acid substitutions are indicated as vertical lines such that the height and direction are proportional to the change in amino acid frequency among 388 subtype 1a reference sequences, with relative change away from consensus shown above and change toward consensus shown below the horizontal bar, on a base 2 logarithmic vertical scale.

FIG. 5.

Nonsynonymous change during 3 years of follow-up, showing median values of the six subjects calculated in a sliding window 50 codons wide, with one-codon increments, and comparing sequential annual visit sequences using the Nei-Gojobori model. The 5′ hemigenomic map is depicted at the top for orientation.

FIG. 6.

HVR1 evolution correlated with neutralizing antibody responses in subject 29. Autologous neutralization during 3 years of follow-up from acute to chronic HCV infection is depicted on the right, with the corresponding amino acid sequence evolution on the left. Type 1 sequence logos were used to demonstrate the variability among 388 reference sequences (1a Ref) as well as the initial viral quasispecies from subject 29 (month 0). Amino acid positions are depicted above according to H77 sequence (GenBank accession number AF009606). For months 2 through 41, type 2 logos (20) were used to compare amino acid sequences to month 0 sequence, with the height of each amino acid determined by the log₂ unlikelihood of an amino acid at a given position relative to the initial sequence. To determine NAb ID₅₀ titers, HCVpp expressing E1E2 from month 2 and month 25 visits were incubated with serial 2-fold dilutions of autologous plasma. When 50% neutralization was not detected at the starting plasma dilution of 1:50 (dashed line), the result was recorded as one-half this value, a titer of 1:25. Neutralization results represent a single experiment in which both autologous HCVpp samples were assayed concurrently with the full array of plasma samples.