Hypervariable Region 1 Sequence Stability during Hepatitis C Virus Replication in Chimpanzees

Abstract

The putative envelope 2 (E2) gene of hepatitis C virus (HCV) contains a highly variable region referred to as hypervariable region 1 (HVR1). We hypothesized that this genetic variability is driven by immune selection pressure, rather than representing the accumulation of random mutations in a region with relatively little functional constraint. To test this hypothesis, we examined the E2 sequence of a human inoculum that was passaged through eight chimpanzees, which appear to have a replicative rate (opportunity for chance mutation) similar to that of humans. Acute-phase plasma samples from a human (the inoculum) and six of eight serially infected chimpanzees were studied. For each, 33 cloned cDNAs were examined by a combined heteroduplex–single-stranded conformational polymorphism assay to assess quasispecies complexity and optimize selection of clones with unique gel shift patterns (clonotypes) for sequencing. The sequence diversity of HCV was significantly lower in the chimpanzees than in the humans, and during eight serial passages there was no change in the sequence of the majority clonotype from each animal examined. Similarly, the rates of protein sequence altering (nonsynonymous) substitution were lower in the chimpanzees than in the humans. These findings demonstrate that nonsynonymous mutations indicate selection pressure rather than being an incidental result of HCV replication.

An estimated 170 million people worldwide are infected with hepatitis C virus (HCV) (4). More than 80% of infections result in persistent viremia, which may be associated with chronic hepatitis, cirrhosis, liver failure, and hepatocellular cancer (3, 41, 44). HCV-associated disease is responsible for more than 10,000 deaths each year in the United States alone, and this mortality is expected to rise (8). Because of limitations of in vitro replication systems, studies of HCV pathogenesis have been limited to observation of natural infection of humans and experimental infection of chimpanzees.

The chimpanzee is the best experimental model of HCV infection. HCV does not grow efficiently in tissue culture, and the only other candidate model is a tree shrew that is too small to permit adequate sampling with current technology (52). In chimpanzees, HCV replication has been demonstrated within days of experimental infection, and the potential for reinfection has been demonstrated (14, 15, 37, 51). Serum HCV RNA levels and the extent of hepatic HCV involvement are similar in chimpanzees and humans, suggesting that replication rates are similar (1). The chimpanzee has also been used to test the infectivity of HCV clones and vaccine candidates. However, Bassett et al. recently demonstrated that the natural history of HCV infection in a cohort of experimentally infected chimpanzees differed from what is typically found in humans (5, 6). These chimpanzees had a higher rate of clearance of viremia, lower rate of antibody production to envelope proteins, and lower rate of envelope amino acid change. Similarly, others have reported little change in E2 amino acid sequence in chimpanzees infected with molecular clones (24). In contrast, comparisons of E2 sequences in humans (both within and between infected subjects) reveal substantial heterogeneity, especially in the 27-amino-acid region at the N terminus of E2 that has been called hypervariable region 1 (HVR1).

HCV variants coexist in each infected individual as a swarm of genetically distinct but related variants, called a quasispecies (10–12, 23, 25). HCV sequences can shift rapidly during chronic human infection (20) and interferon therapy (31), consistent with the predicted behavior of a quasispecies in a rapidly changing selective environment dominated by the immune system (40). Another prediction of the quasispecies model is that the master (or most common) sequence in the quasispecies will not change in a stable environment, while changes in minor variants will continue to occur. Evidence from immunosuppressed individuals, in whom the pace of sequence variation is reduced, suggests that this prediction applies to HCV (7, 28, 42).

If HCV persists by escaping the immune response through sequence variation, then that sequence variation will reveal important characteristics of the immune response. If, however, persistence is due to other factors and sequence variation is simply the random product of an error-prone polymerase and a rapidly replicating virus (47), then sequence variation will always occur when HCV is allowed to replicate, and sequence variation will be uninformative except to define areas tolerant of sequence variation.

We hypothesized that despite an HCV replication rate similar to that in humans (37), reduced immune pressure in chimpanzees would result in limited sequence variation of HCV. To test this hypothesis, we studied the HCV quasispecies in a cohort of animals through which HCV was serially passed and in two animals with persistent infection. The serial passage experiment was initiated using a well-characterized strain of HCV (H77), obtained via plasmapheresis from a patient during acute HCV infection (5).

Using a highly sensitive method for detection of distinct variants, we determined the complexity of the quasispecies in each specimen and selected representative cloned cDNAs for sequence analysis. Sequences from these specimens revealed minimal changes in the envelope sequence examined on transfer from human to chimpanzee, in serial passage among chimpanzees, and during chronic infection of chimpanzees. These findings suggest that passage in chimpanzees does not affect the E2 sequence and that HCV replication does not inherently result in the accumulation of substitutions in the master sequence.

MATERIALS AND METHODS

Study animals.

As part of a study of non-A, non-B hepatitis (NANBH), chimpanzees were inoculated with human serum as follows (Table 1): animal x007 (passage 1 [P1]) was inoculated with 10^2.5 chimpanzee infectious doses of serum obtained via plasmapheresis from patient H in 1977, designated H77 (2, 17). Weekly assessments for rise in alanine aminotransferase allowed identification of the acute phase of infection, and acute-phase serum was used to inoculate the next animal in the series. Animal x174 (P2) had persistent viremia, whereas the seven other animals in the passage had transient viremia (5).

TABLE 1.

Characteristics of study subjects

Subject (identifier)	Inoculum	Interval (mo)a			Log [HCV RNA]b
		Hyperacute	Acute	Chronic
x007 (P1)	H77		0.9
x174 (P2)	x007		2.6	96	6.5
x268 (P3)	x174		2.6
x186 (P4)	x268	0.5	1.8
x187 (P5)	x186	0.4	2.3
x361 (P8)	x352		2.3
x304 (x304)	NANBH		20.7c	114	6.0
Human (AB)	NKd		2.8	46	6.1
Human (AD)	NK		17.2	77	5.2
Human (AE)	NK		3.3	89	5.1
Human (AK)	NK		10.5	63	5.6
Human (AT)	NK		0e	81	5.9
Human (AZ)	NK		16.9	81	5.2

^aInterval from date of inoculation (chimpanzees) or estimated date of infection (humans). 

^bFor chimpanzees, chronic-phase specimen, measured by the COBAS method; For humans, average of available specimens (n = 5 to 12), excluding acute phase, measured by the Amplicor method. 

^cDate of inoculation for x304 estimated as described in Materials and Methods. 

^dNK, inoculum strain not known. 

^eHCV RNA was detected at first visit, prior to seroconversion. 

In a separate experiment at another facility, animal x304 was inoculated with NANBH serum and developed persistent viremia. The inoculation date for x304 was recorded as “prior to November 1986,” and so November 1, 1986, was used for calculations that involved the date of inoculation.

Human subjects.

Since 1988, a cohort of approximately 1,350 former and current injection drug users have been followed in Baltimore (46), including 43 subjects who acquired HCV infection during follow-up (45). The viral load trajectories and temporal sequence of HCV RNA and antibody detection for these subjects are described elsewhere (44). The six subjects chosen for this investigation continued to have detectable HCV RNA in the last specimen tested after at least 6 years of semiannual follow-up. The date of infection was estimated by calculating the midpoint between the date of the first specimen in which HCV RNA or antibody was detected and the date of the last prior sample in which evidence of HCV infection was not detected.

Storage of serum and testing for anti-HCV.

Serum samples were immediately centrifuged, stored for less than 1 week at −20°C, and subsequently stored at −70°C. They were tested for antibodies to HCV (HCV EIA [enzyme immunoassay] 2.0; Ortho Diagnostics, Raritan, N.J.) and, if EIA positive, by strip immunoblot assay (RIBA HCV 2.0; Chiron Corporation, Emeryville, Calif.) as previously described (45).

HCV RNA detection.

In all HCV seroconverters, the presence of HCV RNA was evaluated in sera collected 6 months before seroconversion, at seroconversion, and during subsequent semiannual visits (median of 8) (44). HCV RNA was initially detected with a quantitative reverse transcriptase PCR assay (AMPLICOR HCV MONITOR; Roche Diagnostic Systems, Branchburg, N.J.), the linear range of which was 500 to 500,000 copies per ml of serum in this and other laboratories (21, 34). For chimpanzees with persistent viremia, HCV RNA was quantitated using the COBAS AMPLICOR 2.0 detection system (Roche Diagnostic Systems). In a comparison of these two methods using paired patient specimens and dilutions of H77, the COBAS method generally gave slightly higher values, but the difference was consistently less than a factor of 10 (A. Valsamakis, personal communication).

Envelope region amplification.

HCV RNA characterization was based on examination of 33 cloned cDNAs spanning the 1,026-nucleotide (nt) region thought to encode envelope protein E1 and a segment of E2, including HVR1 (33). RNA was extracted from 100 μl of plasma or serum by using a QIAamp viral RNA mini kit as specified by the manufacturer (Qiagen, Valencia, Calif.). One-fifth of the extract was used to generate cDNA in a 20-μl reaction at 37°C for 1 h with 20 U of Moloney murine leukemia virus reverse transcriptase (Perkin-Elmer, Foster City, Calif.) and first-round PCR reverse primer. The entire 20-μl cDNA synthesis reaction was used for first-round PCR in a 25-μl reaction containing 0.625 U Expand HF polymerase mixture (Boehringer Mannheim, Indianapolis, Ind.), 1.5 mM MgCl₂, 0.2 mM deoxynucleoside triphosphates and 0.4 μM primers. The primers (and positions relative to the HCV-1 genome polyprotein [9]) were outer forward (493 to 518; 5′-GCAACAGGGAACCTTCCTGGTTGCTC-3′), outer reverse (1745-1723; 5′-GGGCAGDBCARRGTGTTGTTGCC-3′), inner forward (502-527; 5′-AACCTTCCTGGTTGCTCTTTCTCTAT-3′), and inner reverse (1527 to 1507; 5′-GAAGCAATAYACYGGRCCACA-3′). Degenerate bases are indicated with standard International Union of Pure and Applied Chemistry codes. Ten microliters of the first reaction product was used as template for the inner nested PCR. Thermal-cycling conditions for both the inner and outer reactions were denaturation for 120 s at 94°C, followed by 35 cycles of 15 s at 94°C, 30 s at 65°C, and 60 s at 72°C (during the last 25 cycles, the elongation time was increased by 20 s per cycle).

Cloning of cDNA and complexity analysis of 33 cloned cDNAs by gel shift.

The 1-kb HCV cDNA product was ligated into vector pCR 2.1 and used to transform INVαF′ cells (TA cloning kit; Invitrogen, Carlsbad, Calif.). Transformants were detected per manufacturer's protocol, and cloning efficiency was >90%.

For each subject, the gel shift patterns of 33 cloned cDNAs were examined by amplifying a 452-bp region spanning HVR1 and by using a nonradioactive method that detects distinct variants within a sample by using a combination of heteroduplex analysis (HDA) and single-stranded conformational polymorphism (SSCP) on a single gel (HDA+SSCP) (48). Sequences obtained from the serial passage were analyzed by using a divergent variant from the chronic-phase specimen from P2 as the driver. A clonotype is defined as two or more cloned cDNAs that have indistinguishable patterns of electrophoretic migration by HDA+SSCP. In our earlier study, the mean (± standard deviation) genetic diversity of cloned cDNAs belonging to the same clonotype (intraclonotype diversity) was 0.6% (±0.9%), with 98.7% differing by less than 2% (48). The complexity of the quasispecies was characterized by the clonotype ratio, calculated as the number of clonotypes divided by 33, the number of cloned cDNAs examined. The clonotype ratio therefore varies from 0.03 (homogenous) to 1 (highly complex).

Sequencing of representative cloned cDNAs.

To examine each specimen's quasispecies for trends in sequence variation, a subset of cloned cDNAs was identified for sequencing. For each subject, at least three cloned cDNAs were selected for sequencing based on gel shift patterns: two from the majority clonotype, one from each clonotype consisting of more than 10% of the 33 cloned cDNAs examined, and the cloned cDNA with the largest heteroduplex gel shift. Plasmid DNA was isolated from a 3.5-ml broth culture (High Pure plasmid isolation kit; Boehringer Mannheim) according to the manufacturer's protocol. Sequences were determined from this DNA by using universal reverse primers with a PRISM version 2.1.1 automated sequencer (ABI, Foster City, Calif.). Sequences were assembled and edited in Sequencher (Gene Codes, Ann Arbor, Mich.) by a technician who was blinded to our hypotheses. Primer sequences were removed prior to analysis.

Phylogenetic analysis.

Sequence alignments were randomly permuted 100 times by using the SEQBOOT program from the PHYLIP package version 3.572c (18, 19). DNA distance matrices were calculated by using the DNADIST program, maximum likelihood model, with a transition-to-transversion ratio of 4.25 (39). Permuted trees were generated using the NEIGHBOR program with random addition, and bootstrap values were obtained by using CONSENSE. Subtype reference sequences used for phylogenetic analysis had the following accession numbers (subtype designations in quotes remain controversial): 1a, AF009606 and M62321; 1b, D90208; 1c, D14853; 2a, D00944; 2b, D10988; 3a, D17763; 4a, Y11604; 5a, Y13184; 6a, Y12083; “7a,” D84263; “8a,” D84264; “9a,” D84265; “10a,” D63821; and “11a,” D63822. Nonsynonymous and synonymous substitution frequencies were calculated by the method of Nei and Gojobori (27).

Statistical analysis.

After examination of the distribution of data, statistical inference was made by using the nonparametric Mann-Whitney test of medians.

Nucleotide sequence accession numbers.

The sequences obtained from chimpanzees were submitted to GenBank and assigned accession no. AF230416 through AF230459.

RESULTS

Amplification and HDA+SSCP analysis.

From the six animals (P1, P2, P3, P4, P5, and P8) that comprised the serial eight passage lineage, the acute-phase plasma (collected 4 to 11 weeks after inoculation and representing the inoculum for the next numbered animal) was studied. Specimens collected within 2 weeks of inoculation (designated hyperacute) were also available for study from P4 and P5 (Table 1). Chronic-phase specimens from animals P2 and x304 were studied. Animal P2 sustained chronic viremia 8 years after inoculation with x7 (H77) serum, and x304 sustained chronic viremia 10 years after inoculation with NANBH (HCV-1). The quasispecies complexity was examined by assessing 33 cDNA clones from each specimen using HDA+SSCP. The median clonotype ratio (range) of the acute phase of chimps, 0.27 (0.18 to 0.39), was similar (P > 0.5) to that found in the first RNA-positive specimen from acutely infected humans, 0.39 (0.09 to 0.67).

Detection of persistent clonotypes through passages.

Use of a common cDNA clone to drive the HDA+SSCP gels permitted comparison of clonotypes among specimens. Among 198 cloned cDNAs from acute-phase specimens, 49 distinct patterns (clonotypes) were identified (Fig. 1). A subset of clonotypes persisted during multiple passages in chimpanzees (Fig. 2). Clonotype a was the most frequently detected clonotype in animals P1, P3, P4, P5, and P8, representing 24 to 76% of cloned cDNAs. Clonotype b was detected in animals P1, P3, P4, and P5. Four clonotypes were observed in three animals each, and nine clonotypes were detected in two animals. For animals P1, P3, P4, P5, and P8, 91 to 94% of cloned cDNAs were observed in at least one other animal. In contrast, the six clonotypes observed for P2 were distinct from all others. Animals P6 and P7 were not examined. Therefore, HDA+SSCP analysis indicated the persistence of clonotype a through chimpanzee passages P1 to P8, with the exception of animal P2, which was the animal that developed persistent viremia. In contrast to the persistence of clonotypes during serial passage, acute and chronic specimens from animals P2 and x304 shared no clonotype (Fig. 1). All of these observations were supported by subsequent sequence analysis (see below).

FIG. 1.

Clonotype distributions and alignment of predicted amino acid sequences from chimpanzees in the passage experiment (A) and for animal x304 (B). Clonotype designations were assigned sequentially from a to z and then aa to ww for the passage and from A to U for x304. In the upper portion of each panel, the numbers below the clonotype designations indicate the number of cloned cDNAs (out of 33 assessed for the animal in that row) with HDA+SSCP pattern consistent with the clonotype indicated in that column. Periods indicate identity to the reference sequence in the first row. The asterisk in sequence P4_b indicates a nonsense mutation.

FIG. 2.

Polyacrylamide gel electrophoresis of HDA+SSCP assay as described in Materials and Methods. This photograph of a single gel shows the SSCP bands (upper panel) and HDA bands (lower panel) of cloned cDNAs chosen to illustrate the persistence of electrophoretic patterns. Clonotypes (groups of electrophoretically indistinguishable cloned cDNAs) were assigned sequential letter designations (a through z, then aa, bb, etc.). Chimpanzee identifiers reflect the passage number (e.g., P2) and whether the specimen was hyperacute (H), acute (A), or chronic (C).

Increase in complexity during infection of chimps and humans.

In each pair of specimens from the same chimpanzee, the complexity of the quasispecies assessed using the clonotype ratio increased with time from inoculation or seroconversion, both during the acute phase (0.08 and 0.05 per month in animals P4 and P5) and also during the chronic phase (0.004 and 0.002 per month in animals P2 and x304). This was also true of four of the six humans who had clonotype ratio changes of +0.003, +0.005, +0.006, and +0.007. However, decreasing complexity was observed in subjects AT and AZ, who had per-month clonotype ratio changes of −0.007 and −0.004 (P > 0.1 for comparisons of clonotype ratios and rates of change). Of note, the infecting subtype of subjects AT and AZ (1b) differed from that of the other hosts (1a), and subject AT seroconverted for human immunodeficiency virus during the period of observation. Having found that human and chimpanzee quasispecies were similar in the number of distinct variants in both acute and chronic infection, we compared the substitution rates.

Initial sequence analysis.

By using HDA+SSCP to select representative cloned cDNAs, 72 distinct cloned cDNAs were identified for sequencing. To determine the genetic identity of cDNA clones of the same clonotype, two representative sequences representing the majority clonotype were compared for each specimen. There were nine differences among 11 pairs of 390-nt sequences (99.8% identity), and all but one difference represented sporadic substitutions (occurring only once in a set of related sequences) as defined by Smith et al. (38). The eight sporadic substitutions were probably artifactual, occurring at a frequency of 3.1 × 10⁻⁵ sporadic substitutions per site per PCR cycle, consistent with the misincorporation frequency of the thermostable polymerase and sequencing reactions (38) and similar to the rate others and we have observed (26, 33). For each majority clonotype, one of the two sequences was free of sporadic substitutions and was used in all subsequent analyses to represent that clonotype. In addition, no two cloned cDNAs identified as being distinct by HDA+SSCP analysis had identical sequences (data not shown). Therefore, the HDA+SSCP method is both highly sensitive and specific in detecting differences among cDNA clones, as previously reported (48). Since the cDNA sequence of one member of a clonotype represents the other members of that clonotype, the 72 cDNA sequences that we obtained represented 200 (72.2%) of the 363 cDNA clones examined.

The most striking aspect of the chimpanzee HCV sequences was the low frequency of substitutions. A phylogenetic tree (Fig. 3A) shows that the acute-phase sequences form a cluster without monophyletic grouping of sequences from individual specimens representing eight sequential passages. Acute-phase sequences from P2 and inoculum strain H77 appear in this cluster. In contrast to the indistinct acute-phase sequences, chronic-phase sequences formed monophyletic clusters. Sequences from animal x304 clustered with reference sequence HCV-1, agreeing with previous studies that suggested a close relationship between the inoculum of x304 and HCV-1 (6). This similarity is also apparent in Fig. 1.

FIG. 3.

Neighbor-joining phylogenetic trees constructed from 372-nt envelope gene sequences of chimpanzees and human subjects infected with HCV subtype 1a (A) and 1b (B). The scale in genetic distance is indicated, and the number of 100 permuted trees supporting a clade is indicated when that proportion was greater than 70%. Identifiers correspond to those in Table 1, followed by the timing of the specimen (acute [A] or chronic [C], as defined in Materials and Methods). Triangles indicate sequences from subject H, and circles indicate sequences from animal P2 (open and filled symbols indicate acute and chronic specimens, respectively). Unlabeled taxa in the enlargement (inset) included sequences from P1, P3, P4, P5, and P8. All sequences from the six chimpanzees in the acute-phase passage series were included in the small encircled clades, without significant clustering. Trees that included 14 HCV subtype reference sequences confirmed clustering of sequences represented in panels A and B with subtypes 1a and 1b, respectively (data not shown).

Comparison of acute- and chronic-phase sequences.

The mean pairwise genetic distance between acute- and chronic-phase sequences was consistently lower in chimpanzees than in humans. This is illustrated by phylogenetic analysis (Fig. 3), which shows shorter branch lengths (lower genetic distance) between acute- and chronic phase sequences from chimpanzees than those from humans. For both chimpanzees with persistent viremia, the mean genetic distance between acute and chronic sequences (interval of 8 to 10 years) was 0.013. In contrast, the mean genetic distance between acute- and chronic-phase sequences for human subject H (interval of 22 years) was 0.093, and in the other six humans (interval of 4 to 7 years) it was 0.044 (range, 0.028 to 0.054). Substitution rates calculated from these data were lower for chimpanzees P2 and x304 (0.0018 and 0.0015 substitutions per site per year) than for human subject H (0.0042) and the other humans (mean, 0.0083; range, 0.0075 to 0.010) (P = 0.055 by Mann-Whitney U test).

In humans, HVR1 is unlike the rest of the HCV genome because nonsynonymous substitutions are as frequent as synonymous substitutions (26, 39), but this was not the case in the chimpanzees in this study. The substitutions observed during chronic infection of chimpanzees were predominantly synonymous, resulting in no change in the amino acid sequence, whereas they were predominantly nonsynonymous (resulting in changes in amino acid sequence) in humans. For chimpanzees P2 and x304, the nonsynonymous substitution frequencies (d_N) were 0.011 and 0.012, the synonymous substitution frequencies (d_S) were 0.025 and 0.023, and d_N/d_S ratios were 0.44 and 0.52, respectively. For human subject H, d_N was 0.082, d_S was 0.13, and d_N/d_S was 0.65, whereas for the other six humans, mean (range) d_N was 0.047 (0.024 to 0.060), mean (range) d_S was 0.034 (0.017 to 0.057), and mean (range) d_N/d_S was 1.7 (0.9 to 3.5). Rates derived from these data (Fig. 4) show that the nonsynonymous rate was higher for humans than for chimpanzees (P = 0.055 by Mann-Whitney U test), synonymous rates varied widely (P > 0.10), and the ratio of these rates was uniformly higher for humans than for chimpanzees (P = 0.055 by Mann-Whitney U test).

FIG. 4.

Nonsynonymous rate, synonymous rate, and nonsynonymous/synonymous (Nonsyn/Syn) rate ratio for seven humans and two chimpanzees with chronic HCV viremia. In each panel, the humans and chimpanzees are presented in the order AB, AD, AE, AK, AT, AZ, H, P2, and x304.

Detailed analysis of sequences from serial passages.

The lack of change in the E2 sequence during eight sequential chimpanzee passages was striking. The majority acute-phase nucleotide sequence for each animal was identical to the majority sequence for the inoculum (H77) strain, except for a single nucleotide change in the third HVR1 codon in animal P2. When all sequenced variants were included in the analysis, there was less than 0.009 weighted genetic sequence distance between any two sequential animals, and from P1 to P8 the weighted genetic distance was 0.0028. This limited variation was not focused in HVR1 as occurs in humans but was distributed evenly across the region sequenced (data not shown).

Figure 5 shows the HVR1 amino acid sequence of the predominant clonotype from each chimpanzee and human specimen. It is notable for the persistence of the majority HVR1 sequence from H77 through all passages (except P2) and the persistence of the majority clonotype sequence in x304. The arginine substitution at residue 3 in all P2 acute-phase sequences was not seen in any other chimpanzee specimen in this study (Fig. 1) but has been described for other chimpanzees infected with H77 (29).

FIG. 5.

Alignment of deduced amino acid sequences of HVR1, one sequence for each of the majority clonotypes detected in each specimen from chimpanzees and humans infected with HCV. Periods indicate identity at that position with the reference sequence above. To the left of each sequence, the host is identified, followed by the timing of the specimen (acute [A] or chronic [C], as defined in Materials and Methods), followed by numbers indicating the number of cloned cDNAs (out of 33 total) represented by each clonotype. The x304 sequences are compared to HCV-1 because the inoculum for x304 was probably related to HCV-1.

The substitutions observed during serial passage in chimpanzees were predominantly synonymous (resulting in no change in the amino acid sequence). Between P1 and P8, d_N was 0.0022 and d_S was 0.0050. For the intervening sequential pairs of animals, d_N ranged from 0.0020 to 0.0072 and d_S ranged from 0.010 to 0.0180, except for the second passage (P1 to P2), for which d_S was 0.0016. The ratio d_N/d_S, an indicator of immune selection pressure, ranged from 0.16 to 0.54 for all passages except the second, for which it was 2.4. Of note, animal P2 was the only animal in the passage series to develop persistent viremia.

DISCUSSION

In this investigation we demonstrated that HVR1 of two strains of HCV subtype 1a underwent very little sequence variation in chimpanzees, whether assessed during serial passage or between acute and chronic phases. This is a striking finding since substantially greater sequence variation was noted in humans with similar HCV RNA levels, indicating the laboratory methods and number of replication cycles were sufficient to detect change. These results suggest that these HCV-infected chimpanzees exerted more negative selection (suppressing changes in the amino acid sequence of HCV) than humans or less positive selection (driving changes in the amino acid sequence), or both.

Strong negative selection is not a likely explanation for the limited change in E2 protein sequences from chimpanzees, since this would imply that there are greater functional constraints on the E2 protein in chimpanzees than in humans. However, a wide spectrum of E2 sequences can be detected in chimpanzees, suggesting that the observed stability of E2 sequences is not due to restrictions imposed by the host.

Weaker positive selection is a more likely explanation for the observed sequence stability. A number of studies indicate that positive selective forces affect the HCV sequence. In particular, HCV envelope sequence evolution has been correlated with antibody production (22, 35, 49), and lower rates of evolution have been observed in agammaglobulinemic subjects. In addition, HVR-specific antibodies may neutralize HCV infectivity (16, 36, 53). van Doorn and coworkers also demonstrated the temporal correlation of anti-HVR antibody production and amino acid changes in chimpanzees (43). In this regard, it is important to note that the sequence stability observed in this study through the lineage of chimpanzees represented samples taken before the measured humoral and cellular immune response.

The supposition that positive selection drives sequence evolution for HCV is supported by studies of other RNA viruses. There have been a number of investigations correlating cytolytic T-lymphocyte responses with human immunodeficiency virus sequence variation (32, 50). In addition, recent data correlate nonsynonymous sequence changes with escape from dominant cytolytic T-lymphocyte responses in chimpanzees experimentally infected with simian immunodeficiency virus (13). These data indicate that replication alone does not drive amino acid mutation in a quasispecies.

The HDA+SSCP analysis used in this investigation reveals classical quasispecies characteristics of HCV throughout the chimpanzee passage experiment. Each specimen contained a swarm of distinct but related variants. The number of variants (complexity) was similar to that found in acute human HCV infection. In each animal there was a master (most commonly observed) sequence representing 24 to 85% of the 33 variants examined, accompanied by minor variants which were nearly always (>90%) found in another animal, usually the prior or subsequent animal in the passage (Fig. 2), except for animal P2. The master sequence (represented by clonotype a in Fig. 2) did not change during the eight passages, except for a single nonsynonymous substitution in the one animal (P2) that developed persistent viremia. The inoculum for P3 was obtained from P2 18 days prior to the P2 specimen available for this study. Therefore, explanations for the reappearance of clonotype a in P3 include (i) the mutation occurred or became dominant during the 18-day interval, (ii) a reversion mutation occurred in P3, and (iii) clonotype a was present in P2 below the level of detection but was preferentially expanded in P3.

Our findings are limited to the envelope gene region examined. It is possible that significant changes occurred in other regions during the passage experiments or during chronic infection. Nonetheless, full-genome sequence data from 15-month follow-up of two chimpanzees infected with a molecular HCV clone (constructed from H77 sequence) also found little variability in the N terminus of E2 (24). In that study, one animal had a single amino acid substitution in HVR1 at week 51, despite readily detectable anti-HVR1 in both animals.

Although this study involves a relatively large number of chimpanzees, only two strains of HCV were assessed: H77 and NANBH (similar to HCV-1). In addition, both animals with chronic viremia (P2 and x304) received serum from an infected chimpanzee rather than a (possibly more complex) human inoculum. Other studies of chimpanzees infected with different HCV strains or human serum have shown higher rates of substitution (30, 43), while 1-year follow-ups of chimpanzees infected with HCV molecular clones have demonstrated lower substitution rates in E2 (24; J. Bukh, M. Yanagi, S. U. Emerson, and R. H. Purcell, Abstr. Fifth International Meeting on Hepatitis C Virus and Related Viruses: Molecular Virology and Pathogenesis, abstr. 28, 1998). These different rates could be due to a variety of factors, including differences in the complexity of the inoculum, duration of follow-up, or possibly the subtype. In the two studies with higher rates, the infecting subtype was 1b, while our study and the studies of molecular clones involved subtype 1a. Although too little is understood about the factors that affect sequence evolution, our data demonstrate that chronic HCV infection need not result in the accumulation of nonsynonymous mutations. This finding implies that immunodominant domains can be identified by examination of nonsynonymous change in a quasispecies, an important adjunct to existing measurements of the host-virus interaction.

As shown in Fig. 4, the nonsynonymous substitution rate observed for human subject H was intermediate between the rates of the other humans and those of the chimpanzees. Because the H77 strain was the inoculum for animal P1, the low rate of substitution for animal P2 might be attributed to some characteristic of the H77 strain. However, this would not explain the similar rate found in animal x304. In addition, the rate obtained in this study from patient H is based on a 22-year sampling interval, and saturation at nonsynonymous sites may have occurred (26).

The possibility of PCR contamination was carefully evaluated. Negative control samples were carried through every reaction and were consistently negative, and the highly diverse human samples were processed concurrently with the much less diverse chimpanzee samples. Also, while at least one of the majority clonotype sequences from each chimpanzee was consistently identical to H77, the minority clonotype sequences consistently revealed differences, in accordance with a quasispecies rather than H77 contamination. In addition, the same minor variants were frequently observed in sequential animals in the passage (Fig. 3B). If these differences had been due to misincorporations during amplification and cloning, a more random distribution of variants would have been expected.

Because we found that d_N/d_S is low in chimpanzees and strong negative selection is unlikely, these results indicate that positive selection pressure on the HCV quasispecies is weaker in chimpanzees than in humans. These findings have important implications for interpretation of data obtained using the chimpanzee model of HCV infection, and they support the prediction that a quasispecies under reduced selective pressure will undergo reduced change in the master sequence.