Dominant Role of Host Selective Pressure in Driving Hepatitis C Virus Evolution in Perinatal Infection

Abstract

The dynamics of the genetic diversification of hepatitis C virus (HCV) populations was addressed in perinatal infection. Clonal sequences of hypervariable region 1 of the putative E2 envelope protein of HCV were obtained from four HCV-infected newborns (sequential samples spanning a period of 6 to 13 months after birth) and from their mothers (all samples collected at delivery). The data show that the variants detected between birth and the third month of life in samples from the four newborns were present in the HCV populations of their mothers at delivery. In the newborns, a unique viral variant (or a small group of closely related variants) remained stable for weeks despite active viral replication. Diversification of the intrahost HCV population was observed 6 to 13 months after birth and was substantially higher in two of the four subjects, as documented by the intersample genetic distance (GD) (P = 0.007). Importantly, a significant correlation between increasing GD and high values for the intersample K_a/K_s ratio (the ratio between antonymous and synonymous substitutions; an index of the action of selective forces) was observed, as documented by the increase of both parameters over time (P = 0.01). These data argue for a dominant role of positive selection for amino acid changes in driving the pattern of genetic diversification of HCV populations, indicate that the intrahost evolution of HCV populations is compatible with a Darwinian model system, and may have implications in the designing of future antiviral strategies.

The high rate of persistent infections distinguishes hepatitis C virus (HCV) from other members of the family Flaviviridae. Since a large proportion of subjects infected with HCV develop clinical syndromes ranging from asymptomatic infection to chronic hepatitis, liver cirrhosis, and primary hepatocellular carcinoma (1, 19, 38), virus persistence is considered to be crucial for the pathogenic potential of HCV, but the mechanisms connected with this event have not yet been completely elucidated.

In persistently infected hosts, the HCV genome is described as a dynamic population of heterogeneous, closely related variants designated quasispecies (13, 15, 17, 26), suggesting that its intrahost variability is associated with virus persistence (35). Indeed, amino acid substitutions in crucial portions of the HCV envelope proteins may allow virus variants to evade the host's immune surveillance (12, 15, 39, 42). Two hypervariable regions (HVR), designated HVR-1 and HVR-2, have been identified in the putative envelope-encoding E2 region of the HCV genome (5, 13, 17), and direct and indirect evidence suggests that the 27-amino-acid HVR-1 sequence located in the N-terminal portion of the HCV envelope protein is a dominant neutralizing epitope (8, 33, 45, 46). Thus, the presence of populations of virions heterogeneous for the HVR-1 sequence within an infected individual may be a reason for the failure of specific anti-HCV antibodies and virus-specific cytotoxic T lymphocytes to clear the virus (7).

Although the error-prone nature of viral RNA polymerases and the lack of 3′-exonuclease proofreading activity provide the biochemical bases for virus diversity, the relevant features and the determinants of the intrahost evolution of HCV populations are still unclear. Different mechanisms have been described to explain the intrahost evolution of RNA viruses, including mutation-driven (36), neutral (10), and adaptive (41) evolution. In HCV infection, adaptive evolution has recently been described by our group in primary infection in adults (25), thus indicating that the diversification is not programmed by the viral sequence (as also documented in a cohort of subjects infected from the same source who developed individually distinct HCV populations) (27) and pointing at host selective pressure as the major determinant of intrahost HCV genetic evolution.

To further address the relevant features of intrahost HCV diversity, we considered that perinatal HCV infection has specific characteristics that could contribute to elucidating some central aspects of intrahost virus evolution. Perinatal infection principally differs from infection in adults in that very limited (if any) viral diversity is observed for months (20, 30). From an evolutionary point of view, the HCV variant detected early in perinatal infection may be considered the “ancestral” sequence for that particular host, allowing easier and more correct analysis of sequence diversification over time. In this study, we analyzed the evolution of the HVR-1 sequence of HCV in sequential plasma samples from four HCV-infected newborns and studied the plasma samples collected at delivery from the HCV-transmitting mothers. While no correlation between plasma virus load or presence of specific anti-HCV antibodies and mean intra- and intersample genetic distances (GDs) of the HCV variants could be observed, the data indicate a strong correlation between the GD and the intersample K_a/K_s ratio (the ratio between the number of antonymous substitutions per antonymous site and the number of synonymous substitutions per synonymous site), thus being consistent with a crucial role of the host's selective forces in driving HCV evolution during the early phases of perinatal infection.

MATERIALS AND METHODS

Patients and samples.

Four HCV-infected newborns were included in this study. One of them was from the Department of Obstetrics and Gynecology and Pediatrics of the University of Pavia, and the other three were from the Institute of Virology, University of Milan, Italy. All the mothers tested negative for anti-human immunodeficiency virus type 1 (HIV-1) antibodies, for the hepatitis B virus (HBV) surface antigen (HBsAg), and for antibodies to other hepatitis viruses. The four newborns (all vaccinated against HBV) were followed for periods ranging from 12 to 13 months. Plasma samples were collected from the mothers at delivery and from the infants after 3, 6 to 7, 9 to 10, and 12 to 13 months (for infant f4, two earlier samples collected after 1 and 2 months were also available).

Serological assays.

Anti-HCV antibodies were assayed using an enzyme-linked immunosorbent assay method (HCV 3.0 ELISA; Ortho Diagnostic Systems, Raritan, N.J.) and a third-generation recombinant immunoblot assay (Inno-Lia HCV III; Innogenetics, Ghent, Belgium). Antibodies to hepatitis A virus, markers for HBV infection (HBsAg, anti-HBs, HBeAg, anti-HBe, and anti-HBc [immunoglobulins G and M]), and antibodies to HIV-1 were tested by routine methods (microparticle enzyme immunoassay [Abbott Laboratories, North Chicago, Ill.] and enzyme immunoassay [Sanofi Pasteur, Marnes-le-Coquette, France]).

Sucrose density gradients of plasma samples.

Sucrose density gradients of plasma samples were performed as described by Bradley et al. (2), with minor modifications. Briefly, 0.5 ml of plasma was layered on top of a continuous 20 to 60% (wt/vol) sucrose gradient prepared in 0.01 M TENB (pH 7.5) buffer (0.01 M Tris-HCl, 0.001 M EDTA, 0.15 M NaCl) and centrifuged in an SW-41 Beckman (Palo Alto, Calif.) rotor at 35,000 rpm for 18 h at 5°C using a Beckman (model Optima L-90K) ultracentrifuge. Fifteen to 19 fractions of 500 μl were collected by piercing the bottom of the tube, and density was assessed before storing the samples at −80°C. RNA was extracted from 400-μl aliquots of each fraction by the guanidinium thiocyanate method (4). The RNA pellets were dissolved in 20 μl of water, and 10 μl was quantified using competitive reverse transcription (cRT)-PCR as described elsewhere (24).

HCV genotyping and quantitation of HCV RNA molecules in plasma.

HCV genotyping was performed in all plasma samples by nested RT-PCR of the HCV core region according to the method of Okamoto et al. (32), with minor modifications (34). To determine the HCV RNA copy numbers in plasma samples, RNA was extracted from 100 μl of plasma by the guanidinium thiocyanate method (4); the RNA pellets were then dissolved in 20 μl of water, and 10 μl was quantified by cRT-PCR (24).

Amplification, cloning, and sequencing procedures.

A 612-bp sequence of the E1-E2 region encompassing HVR-1 of HCV RNA (from nucleotide 1278 to nucleotide 1889) was amplified by RT-PCR using the following primer set: sense primer, 5′-ATAAC GGGTC ACCGA TGGCA TAT; antisense primer, 5′-CACCA CGGGG CTGGG AGTGA AGCAA T. The amplified product was ligated to the pCR-Script SK(+) plasmid vector (Stratagene, La Jolla, Calif.). Plasmid DNA from single transformant colonies was extracted and purified from overnight-cultured minipreps by the Wizard DNA purification system (Promega, Madison, Wis.). To sequence the cloned DNA inserts, 10 to 20 independent clones per clinical sample were sequenced directly. The double-stranded DNA was sequenced in both forward and reverse directions by fluorescence-labeled dideoxynucleotides with an automated sequencer (model 373A; Perkin-Elmer, Norwalk, Conn.) following the sequencing conditions specified in the protocol for the ABI PRISM Dye Terminator cycle-sequencing kit and using Amply-Taq DNA polymerase FS (both from Perkin-Elmer).

Sequence analysis.

Sequence editing and assembling were performed with the Sequence Navigator program included in the AB373 software package. The alignments of both nucleotide and amino acid sequences were performed with the Clustal W program version 1.7. Simple sequence similarity comparisons were performed with the Megalign program (DNAstar Inc., Madison, Wis.). Phylogenetic reconstructions were generated by using programs from version 3.572 of the Phylogeny Inference Package (PHYLIP) (9). The DNADIST (with Kimura's two-parameter method) and the DNAPROT (with Kimura's formula) programs (18) were applied to generate a pairwise matrix of evolutionary distances of nucleotide and amino acid sequences, respectively. Phylogenetic trees were constructed from the same distance matrices with the NEIGHBOR program (neighbor-joining algorithm). Bootstrap analysis was performed with SEQBOOT (100 resamplings), followed by the DNADIST or DNAPROT, NEIGHBOR, and CONSENSE programs. The rates of synonymous nucleotide substitutions per synonymous site (K_s) and antonymous substitutions per antonymous site (K_a) were estimated by the method of Nei and Gojobori (28) by using the Jukes-Cantor correction for multiple substitutions as implemented in the MEGA program package (version 1.02, 1993).

Statistical analysis.

All of the analyses were performed with StatView version 4.5 (Abacus Concepts, Berkeley, Calif.). The unpaired t test was used to compare group means. The Friedman test was used to analyze variations of evolutionary parameters with time. Two-way analysis of variance was used to compare group means for evolutionary parameters at the different time points.

Nucleotide sequence accession numbers.

The sequences described here have been submitted to GenBank and assigned accession numbers AF192415 to AF192461.

RESULTS

Relevant features of the cell-free virus in transmitter and nontransmitter mothers.

Cell-free HCV RNA molecules, quantified by cRT-PCR in plasma samples from the four transmitter mothers, ranged from 1.38 × 10⁶ to 6.64 × 10⁶ copies per ml of plasma (Table 1). These values substantially overlapped with those obtained in 23 HCV RNA-positive mothers analyzed in a preliminary phase of this study who did not transmit the virus to their newborns (mean, 3.02 × 10⁶ copies per ml of plasma). Moreover, since an intriguing hypothesis for the relatively low rate of perinatal HCV transmission in anti-HIV-1-negative women is neutralization of circulating virus by maternal antibodies, and since these antibodies (principally those directed against the putative envelope protein 2 [E2]) are also believed to play a role in the intrahost selection of HCV variants (16), we studied plasma samples from two mothers who transmitted the virus (m3 and m4) and from two nontransmitter mothers (all collected at delivery) by sucrose density gradients (Fig. 1); the samples had similar levels of plasma viremia (1.07 × 10⁶ and 1.84 × 10⁶ copies per ml, [HCV genotypes 1b and 2c], respectively). The comparative analysis of HCV RNA copy numbers in the fractionated plasma from both transmitter and nontransmitter mothers documented the fact that in the former, HCV RNA molecules were detected almost exclusively in the light-density fractions (from 1.13 to 1.09 g/cm³), while in the latter, viral genome molecules were detected in the heavy-density fractions of the gradient (from 1.23 to 1.16 g/cm³) and only partially in the light fractions (from 1.10 to 1.07 g/cm³). These results, in line with previous reports (14, 20), are consistent with the hypothesis that the presence of HCV in dense plasma fractions (probably containing antibody-bound HCV virions) indicates a relative inefficiency in transmitting the virus during pregnancy or at delivery; the presence of virions in the light fractions (probably containing virions bound to low-density lipoproteins) may indicate higher infectivity.

TABLE 1.

Virological follow-up of mother-to-infant infections

Mother	HIV (status)	Delivery	Breast feeding	History of hepatitis	HCV genotype	HCV-RNAa (copies/ml)	Infant	Time (mo) (sample)	HCV genotype	HCV-RNA (copies/ml)	Inno-Lia HCV IIIb
											Core 1 + 2	Core 3 + 4	NS1	NS3	NS4	NS5
m1	Neg	Vaginal	No	No	3a	6.64 × 105	f1	3 (f1-3)	3a	2.60 × 106	±	−	−	−	−	−
								6 (f1-6)		6.53 × 105	+++	±	++	−	−	++++
								10 (f1-10)		9.37 × 105	+++	++	++	++	±	+++
								13 (f1-13)		6.64 × 105	+++	+++	++	+	++	+++
m2	Neg	Vaginal	Yes	No	1b	1.50 × 106	f2	3 (f2-3)	1b	7.85 × 106	+++	+	+	+++	−	−
								6 (f2-6)		1.47 × 105	+++	++	±	++	++	−
								9		NDa	+++	+++	++	+	+++	−
								13		ND	+++	+++	++	++	++	−
m3	Neg	Vaginal	No	?	1a	1.38 × 106	f3	3 (f3-3)	1a	1.76 × 106	+++	+++	±	±	++	−
								6 (f3-6)		1.08 × 106	++	−	−	++	−	−
								9 (f3-9)		9.26 × 106	+++	−	−	+++	±	−
								12		ND	+++	+	−	+++	++	−
m4	Neg	Vaginal	No	No	2c	2.30 × 106	f4	1 (f4-1)	2c	4.38 × 106	+++	+++	±	++	−	−
								2 (f4-2)		3.08 × 106	+++	+++	++	++	±	−
								3 (f4-3)		1.26 × 106	+++	+++	++	++	++	−
								7 (f4-7)		3.04 × 106	+++	+++	++	+	++	−
								10 (f4-10)		1.06 × 106	+++	+++	++	+	++	−
								12 (f4-12)		8.65 × 105	+++	+++	++	+	++	−

^aQuantitation of HCV-RNA molecules in plasma samples was performed with a cRT-PCR assay (25). 

^bInno-Lia-HCV III, third-generation recombinant immunoblot assay detecting antibodies to structural (core 1 + 2 and core 3 + 4) and nonstructural (N1, NS3, NS4, and NS5) HCV proteins. +, ++, and +++, intensity of positive bands; ± and −, indeterminate and negative reactivity, respectively. 

^cND, not done. 

FIG. 1.

HCV RNA copy numbers in plasma fractions obtained by sucrose density gradient. Plasma samples from two transmitter (C and D; subjects m3 and m4 of this work) and two nontransmitter (A and B) mothers were collected at delivery. After collection of the fractions, HCV RNA molecules were quantified in each fraction by quantitative cRT-PCR.

Sequences of HVR-1 of HCV and phylogenetic analysis.

Nucleotide sequences of HCV HVR-1 were obtained after RNA purification, amplification, cloning, and sequencing of the plasma samples collected at delivery from the four transmitter mothers and of a set of two to four plasma samples from each infected infant. Sampling spanned a period from 6 to 13 months after birth, and a total of 255 viral sequences were analyzed. Figure 2 shows the nucleotide and deduced amino acid sequence alignments of HVR-1 from the newborns and their mothers. In the first sample available for newborn f1 (collected at 3 months), a single virus variant (designated f1-3a [Fig. 2]) was observed, representing 100% of the 20 clones tested; this variant was also present in the mother's (m1a) plasma at delivery. In the subsequent samples from newborn f1 (collected at 6, 10, and 13 months), sequence diversity was limited, as was also documented by the phylogenetic reconstruction (Fig. 3; f1-m1). In the sample collected at 3 months from newborn f2, two very close HVR-1 variants were revealed (f2-3a and f2-3b), representing 95 and 5%, respectively, of the clones tested; the f2-3a variant was also present in the mother's plasma at delivery (m2a). Interestingly, generation of a heterogeneous virus population was observed at 6 months, when 11 viral HVR-1 variants were documented (Fig. 2 and 3). Newborn f3 showed three very close variants at 3 months (f3-3a, f3-3b, and f3-3c, representing 90, 5, and 5%, respectively, of the 20 clones tested; these sequences are very similar, albeit not identical, to the mother's variant, m3a); in this subject, limited diversity was observed in the samples collected at 6 and 9 months. Finally, a single HVR-1 variant (100% of the clones tested) was documented in the samples collected at 1, 2, and 3 months from newborn f4 (f4-1a, -2a, and -3a, similar to two variants, m4a and m4b, present in the mother's sample collected at delivery). In this newborn, sequence diversification could be detected starting at 7 months (four variants, from f4-7a to f4-7d) and in subsequent samples (Fig. 2 and 3).

FIG. 2.

Deduced amino acid sequence alignments of HVR-1 from four newborns (f1 to f4) and their mothers (m1 to m4; samples were collected at delivery). The initial nucleotide (∗) and amino acid (°) sequences for each mother-infant pair represent the reference sequences. The serial time points are indicated by a number following the subject's identification corresponding to the month after birth; diverging clonal sequences at each time point are indicated by final letters. The sequences from each infant and from the corresponding mother are aligned with the sequence from the first sample. Dashes indicate identity with the reference sequence. Shading indicates amino acids that differ from the reference sequence.

FIG. 3.

Phylogenetic reconstruction of the evolutionary relationships within the four HCV-infected infants and their mothers. The deduced amino acid sequences of all clonal sequences were analyzed using the Kimura's formula distance matrix fed into a neighbor-joining tree construction algorithm. Branch lengths are drawn to scale. Different samples are indicated by numbers (as in Fig. 2) and colors: red, mothers' sequences (delivery); black, green, blue, and pink, sequential samples from infants.

Analysis of the intra- and intersample GDs of HCV variants and of host selective pressure on the HVR-1 sequence in infected newborns.

In order to evaluate whether and to what extent nucleotide sequence variability and accumulation rates of synonymous and antonymous substitutions varied with sampling time, pairwise comparisons of sequences were performed within (intrasample) and between (intersample) the time points (Table 2). For the intersample analysis, the sequence (or sequences) of the first sample was the term used for comparison of all subsequent sequences for each patient. Intertime GD documented two distinct profiles among the four HCV-infected newborns under study: two newborns (f2 and f4) showed increasing intra- and intertime GDs, and two (f1 and f3) had low, stable GDs (GD f1-f3 mean, 1.168 ± 0.711; GD f2-f4 mean, 6.190 ± 2.920; t test, 3.7689; P = 0.00699). Accumulation of synonymous (K_s) and antonymous (K_a) substitutions and K_a/K_s ratios were analyzed to screen for positive selection for amino acid changes in the HVR-1 sequence; a significant difference in K_a/K_s ratios between the two groups of subjects was also observed (K_a/K_s f1-f3 mean, 0.0644 ± 0.254; K_a/K_s f2-f4 mean, 8.07 ± 2.256; t test, 6.6754; P = 0.00028). Notably, the increase in intertime K_a/K_s ratios to very high values (documenting strong host selective pressure) observed in newborns f2 and f4 (but not in f1 and f3) paralleled significantly the increase in GD over time (Friedman test; P = 0.01). Two-way analysis of variance (analysis of variance table for repeated measures) also yielded a significant difference between the two groups (P < 0.05).

TABLE 2.

Intra- and intersample variability of HVR-1 in mother-to-infant HCV transmission

Intrasample nucleotide divergence					Intersample nucleotide divergence
Samplea	Kab	Ksb	Ka/Ks	GDc (mean ± SD)	Samples	Ka	Ks	Ka/Ks	GD (mean ± SD)
f1-3	0.00	0.0083	NVd	0.21 ± 0.35
f1-6	0.0086	0.00114	0.75	0.91 ± 0.65	f1-3–f1-6	0.0036	0.0096	0.37	0.50 ± 0.62
f1-10	0.0077	0.00	NV	0.53 ± 0.62	f1-3–f1-10	0.0035	0.0045	0.77	0.36 ± 0.52
f1-13	0.0107	0.0207	0.51	1.41 ± 1.26	f1-3–f1-13	0.0065	0.0168	0.38	1.36 ± 1.22
f2-3	0.003	0.0099	0.30	0.40 ± 0.54
f2-6	0.0048		NV	3.53 ± 1.53	f2-3–f2-6	0.0366	0.0038	9.63	2.74 ± 1.16
f3-3	0.0126	0.0251	0.50	1.51 ± 0.91
f3-6	0.0129	0.0262	0.49	1.54 ± 1.02	f3-3–f3-6	0.0195	0.0252	0.77	1.99 ± 1.17
f3-9	0.0092	0.0146	0.63	1.10 ± 0.81	f3-3–f3-9	0.0167	0.0179	0.93	1.63 ± 1.06
f4-1	0.00	0.0078	NV	NV
f4-2	0.00	0.0078	NV	NV	f4-1–f4-2	0.00	0.00	NV	NV
f4-3	0.00	0.00	NV	NV	f4-1–f4-3	0.00	0.0077	NV	NV
f4-7	0.0187	0.0078	NV	2.78 ± 1.97	f4-1–f4-7	0.0244	0.0037	6.59	4.90 ± 3.20
f4-10	0.0235	0.00	NV	4.43 ± 4.26	f4-1–f4-10	0.0343	0.0032	10.71	7.93 ± 3.39
f4-12	0.0274	0.0114	2.40	3.66 ± 3.70	f4-1–f4-12	0.0525	0.0098	5.35	9.19 ± 6.56

^aSamples were designated according to the newborns' code (f1 to f4) and to the time of plasma collection (1 to 13 months after birth). 

^bK_a, antonymous nucleotide substitutions per antonymous site; K_s, synonymous nucleotide substitutions per synonymous site. 

^cThe intra- and intersample GDs were calculated using DNADIST software (PHYLIP version 3.5c). 

^dNV, not valuable. 

DISCUSSION

The genetic diversification of HCV, its dynamics, and the effect of the host's selective forces on a hypervariable domain of the HCV envelope protein were addressed in this study of perinatal infection. HCV transmission from persistently infected mothers to their newborns has been documented (37) and is estimated to occur in 5% of cases (6, 40), indicating a risk lower than that reported for mother-to-infant transmission of HBV and HIV-1. Although the mechanism for neonatal HCV infection has not been precisely clarified, previous studies have suggested that the presence of chronic hepatitis, high levels of HCV RNA in plasma, and coinfection with HIV-1 in mothers are associated with more efficient HCV transmission (3, 23, 43, 44).

In this study, we chose to evaluate virus evolution in mother-to-infant infection. In most perinatal infections, a single incoming variant (or a small group of very close variants) is stable for weeks or months (20, 30), while in acute infection of adults, a rapidly evolving population of related variants is generally observed (25). Indeed, we considered that, in perinatal HCV infection, the incoming variant may be regarded as the ancestral sequence for that particular host. This is of crucial importance in tracking intrahost virological evolution, as the ancestral sequence (the fittest for that particular host at that time) can be used as a reference for estimating the rate and dynamics of subsequent nucleotide substitutions in each infected subject. Furthermore, since it has recently been suggested that adaptation of HCV for persistence is driven by selective forces of the host, including specific immune response (22, 31), the evolutionary analysis of virus populations in HCV-infected newborns (i.e., in a host with an immature humoral and cytotoxic immune response) could allow us to gain insights into the natural history of HCV infection, the viral pathogenic potential, and the virus-host interplay, despite the infrequency of HCV transmission from HIV-1-negative and anti-HCV-positive mothers. Recently, the quasispecies nature of HCV populations and viral genetic evolution have been addressed in some infected mothers and their infants. These studies have documented the fact that the initial HCV variant infecting the newborn is closely related to the mother's quasispecies, that perinatal infection is not influenced by the dominant virus variants present in the mother's plasma at delivery, that the viral HVR-1 variants infecting the infant are genetically stable for several weeks after birth, and, finally, that HCV evolution in neonates is different from that observed in their mothers (20, 30).

The results reported here extend the study of perinatal HCV infection to the evaluation of the impact of the host's selective pressure on HVR-1 evolution. First, these data confirm that the incoming HVR-1 variant (or the group of closely related variants) remains unmodified for several weeks despite highly active virus replication. A recent study has evaluated the dynamic features of cell-free HCV virions in plasma, documenting a high turnover (the half-life of plasma virions is approximately 2.7 h) (29). Under these conditions, and owing to error-prone viral RNA polymerase, the detection of sequences stable for several weeks in a viral HVR suggests that this variant is the fittest for that particular host (environment) and that HCV replication per se can determine only minimal changes in the HCV population in vivo. Second, the genetic diversification of HVR-1 of the putative HCV envelope gene (documented by increasing intersample GD) started 6 to 7 months after birth in two of the four newborns under study. Indeed, while in two subjects a single incoming virus variant was observed up to 6 to 7 months after birth, in the other two subjects (f2 and f3), two and three variants, respectively, were detected in samples collected 3 months after delivery. Due to the very close phylogenetic distance of these early virus variants (Fig. 3), this result (probably as a consequence of the high number of clones assayed in this study) has not substantially modified the analysis carried out in this study. Third, to further determine whether the results of HVR-1 diversity are a consequence of the high rate of virus replication and of the error-prone nature of viral RNA polymerase or of the different levels of selective constraints, the rate of antonymous over synonymous substitutions was evaluated. An excess of antonymous over synonymous substitutions is an unambiguous index of positive selection at the molecular level. Several methods have been proposed to estimate the antonymous and synonymous substitution rates, including an explicit codon substitution model (11) and comparison between two sequences (21, 28). In the present study, we used the analysis of intertime K_a/K_s ratios to evaluate the level of host selective forces on HVR-1 of the putative HCV envelope. Importantly, the study documents the fact that, in subjects f2 and f4, the increasing GD is strongly correlated with very high levels of host selective pressure.

Although the data shown here indicate that the host's selective pressure is a major determinant of intrahost HCV evolution in perinatal infection, this study has not addressed the nature of the selective constraints of the host. However, considering the time course of HCV genetic diversification in our subjects and the possible interaction of viral envelope domains with structures of the cell surface, it is likely that two main aspects of the virus-host relationships are involved: (i) the humoral and cytotoxic immune responses to HCV domains and (ii) the host cell range. Specific analysis of these aspects (including the influence of the host's genetic features on the immune response) and long-term follow-up of HCV-infected newborns are clearly necessary to evaluate precisely the real impact of these aspects over time as selective forces for HCV envelope domains.

Although the present study cannot explain whether the different features observed in subjects f1 and f3 and subjects f2 and f4 are related to a delayed immune response, to a different host cell range of HCV variants, and/or to a different outcome of the infection, these results, documenting a significant direct association between GD and levels of selective constraints, extend our understanding of HCV evolution in infected hosts. A possible interpretation of the data presented here is that the incoming viral variant (or group of closely related variants) was not neutralized by circulating immunoglobulins in the mother's blood at delivery (as suggested by the sucrose gradient analyses); this variant (the fittest, for some months, for that particular host) does not change despite active virus replication, unless strong and specific selective forces emerge. From this point of view, our data indicate that the intrahost evolution of HCV populations is perfectly compatible with an ideal Darwinian model system. Overall, these data have implications for the understanding of HCV pathogenesis and virus-host relationships and for the designing of effective anti-HCV strategies.