Hepatitis C virus envelope glycoprotein co-evolutionary dynamics during chronic hepatitis C

Abstract

Hepatitis C virus (HCV) envelope glycoprotein co-evolution was studied in 14 genotype 1-infected and treatment naïve subjects, including 7 with mild and 7 with severe liver disease. Cassettes encoding the envelope 1 gene (E1) and hypervariable region (HVR1) of the envelope 2 gene were isolated at 38 different time points over 81 follow-up years. There were no significant differences in age, gender, alcohol use, or viral load between the mild and severe disease groups. Virus from subjects with severe disease had significantly slower evolution in HVR1, and significant divergent evolution of E1 quasispecies, characterized by a preponderance of synonymous mutations, compared to virus from subjects with mild disease. Phylogenetic comparisons indicated higher similarity between amino acid sequences of the E1 and HVR1 regions with mild disease versus severe disease (r=0.44 versus r=0.17, respectively; P=0.01). In summary, HCV envelope quasispecies co-evolution differs during mild versus severe disease.

INTRODUCTION

Hepatitis C virus (HCV), a positive-sense, single strand RNA virus of the family Flaviviridae (Lindenbach, 2001), is a major cause of chronic liver disease throughout the world (Choo et al., 1989; Di Bisceglie, 1998; Houghton et al., 1991). HCV infection accounts for at least 3 million cases of chronic hepatitis in the United States, and as many as 170 million cases worldwide (Alter et al., 1999; Kim, 2002; WHO, 1999). About 20% of infected individuals spontaneously clear the virus in the acute phase, while approximately 80% of cases result in persistent viremia that frequently develops into severe liver diseases after decades, including liver cirrhosis and hepatocellular carcinoma (Fattovich et al., 1997; Hu and Tong, 1999; Khan et al., 2000; Serfaty et al., 1998; Villano et al., 1999). Like many RNA viruses, HCV exhibits significant genetic heterogeneity as a result of high replication rates and error-prone, non-proofreading RNA polymerases (Bukh et al., 1995; Martell et al., 1992). Phylogenetic analysis has identified six genotypes or clades of HCV, each with multiple discrete subtypes (Robertson et al., 1998; Simmonds, 1995). In addition, HCV within a persistently infected individual is composed of a dynamic population of close-related, yet heterogeneous sequences, known as quasispecies (Bukh et al., 1995; Martell et al., 1992). The highest levels of HCV genetic variability are displayed in the two envelope glycoproteins E1 and E2, with maximum heterogeneity within the N-terminus of E2, commonly referred to as hypervariable region 1 (HVR1).

Although the error-prone nature of viral RNA polymerases provides the biochemical bases for rapid viral evolution (Holland et al., 1982), the quasispecies diversity is not generated by nucleotide substitutions during replication alone (Kurosaki et al., 1993; Lu et al., 2001). Immune pressure, involving neutralizing antibody, cytotoxic lymphocytes and alpha/beta interferons, is also thought to be the primary force driving HCV quasispecies evolution in vivo (Kato et al., 1993; Kurosaki et al., 1993; Neumann et al., 1998; Weiner et al., 1992; Zibert et al., 1995). HVR1, a 27 amino acid sequence located in the N-terminus of E2 protein, forms a major virus surface antigen and is the dominant epitope for strain-specific neutralizing-antibody responses (Farci et al., 1994; Farci et al., 1996; Rosa et al., 1996; Shimizu et al., 1994; Shimizu et al., 1996; Zibert et al., 1997; Zibert et al., 1995). The extensive sequence variability displayed in HVR1 correlates with the emergence of antibody escape mutants, and appears to play an essential role in maintenance of viral persistence during chronic infection (Erickson et al., 2001; Kurosaki et al., 1993; McAllister et al., 1998; Shimizu et al., 1994; van Doorn et al., 1995; Weiner et al., 1991; Weiner et al., 1992). Studies in immunosuppressed individuals have consistently reported reduced or absent variability in HVR1, presumably due to attenuation of humoral immune responses (Booth et al., 1998; Gaud et al., 2003; Gretch et al., 1996; Kumar et al., 1994; Qin et al., 2005; Sullivan et al., 1998). In spite of the high variability of HVR1, a strong negative selection is also observed against certain amino acid substitutions over the whole viral genome, which might indicate that conservative amino acids play important biological roles in the virus life-cycle (McAllister et al., 1998; Penin et al., 2001; Sobolev et al., 2000).

The genetic heterogeneity of HCV quasispecies has been shown to correlate with the evolution of chronic infection, outcome of antiviral therapy and progression of liver disease (Arenas et al., 2004; Farci et al., 2000; Farci et al., 2002; Pelletier et al., 2000; Ray et al., 1999; Sanchez-Fueyo et al., 2001). Dynamic evolution of HVR1 correlates with outcome of acute infection: subjects harboring stable viral populations frequently clear acute infection, while those with diversifying quasispecies progress to chronic infection (Farci et al., 2000; Ray et al., 1999). Higher quasispecies diversity in baseline specimens has been linked to a poor response to antiviral therapy (Farci et al., 2002; Morishima et al., 2006; Polyak et al., 1998). Finally, longitudinal studies of HCV infections after orthotopic liver transplantation (OLT) indicated that the complexity and diversity of HCV quasispecies is predictive of the histological severity of hepatitis C disease recurrence in liver allografts (Arenas et al., 2004; Gretch et al., 1996; Pelletier et al., 2000; Sanchez-Fueyo et al., 2001; Sullivan et al., 1998). Low diversity quasispecies are associated with worse disease in the post-transplant setting.

Previous studies of HCV quasispecies during natural infection have been conflicting. A missing element in all previous reports was a long-term longitudinal cohort of treatment naïve subjects with well-characterized patterns of disease progression. In a recent study, we recruited fifty-two treatment-naïve subjects with chronic HCV genotype 1 infection over multiple decades and known disease status based on liver biopsy results (Sullivan et al., 2007). The long-term dynamics of HCV infection and their associations with hepatitis C disease were analyzed. Subjects that developed mild disease had higher serum RNA titers after 2 decades of infection, greater fluctuations in serum HCV titers over time, and higher rates of HVR1 diversification over time compared to subjects that developed severe disease. In the present study, we analyzed the evolutionary dynamics of the E1 glycoprotein gene in addition to HVR1 of the E2 glycoprotein gene in 7 subjects with severe disease, and 7 non-diseased subjects randomly selected from the above cohort. Genetic parameters and their behaviors over time were carefully compared between viral genomes isolated from the two subject groups, and molecular evolutionary patterns associated with disease progression were identified for the E1 and HVR1 regions during chronic infection.

RESULTS

Patient characteristics

The fourteen subjects included in this study all underwent at least one liver biopsy on, or in the case of some subjects with mild disease, after the time point of the last serum collection. Seven subjects were defined as non-progressors with mild disease (fibrosis score of 0 or 1), and seven were defined as progressors with severe disease (bridging fibrosis or cirrhosis, fibrosis score of 3 or 4). Five of the subjects had more than 1 liver biopsy result available. Subjects in the two disease groups were comparable in most clinical and virological features, including gender (3 females in the mild group vs. 2 females in the severe group, p = 1.00), genotype (four genotype 1a and three genotype 1b in the mild group vs. five genotype 1a and two genotype 1b in the severe group, p = 1.00), mean viral load (5.2 log IU/ml vs. 5.5 log IU/ml for the mild and severe groups, respectively, p = 0.22), estimated duration of HCV infection (mean 18 years in the mild group vs. 21 years in the severe group, p = 0.56), average number of specimens (2.6 in the mild group vs. 2.8 in the severe group, p = 0.74), and years of follow-up (mean 7.4 years in the mild group vs. 7.3 years in the severe group, p = 0.93). However, subjects in the two groups differed with respect to age (48 years in the mild group vs. 36 years in the severe group, p = 0.06), and the severe disease group had a lower mean age at infection than the mild disease group (30 years in the mild group vs. 15 years in the severe group, p < 0.01). Since older age normally associates with more severe hepatitis C (Poynard et al., 1997), the fact that our mild disease subjects were older and were infected at an older age only strengthens confidence that disease classifications were accurate.

Intra-specimen genetic parameters

To characterize genetic parameters of the E1 gene and HVR1 of the E2 gene of HCV, extensive sequence analysis of quasispecies present in the serial serum samples was performed. 362 nucleotide sequences encompassing the region from nucleotide 817 at the end of the HCV core gene to nucleotide 1626 in E2 were obtained, with 169 sequences from 18 serial specimens of the individuals with mild disease and 193 sequences from 20 serial specimens of the individuals with severe disease. Intra-specimen genetic heterogeneity parameters of the E1 and HVR1 quasispecies are summarized in Table 1. In the E1 region, nonsynonymous substitutions (dN) were significantly less frequent than synonymous substitutions (dS) in both the mild and severe disease groups, with mean dN/dS ratios of 0.2627 and 0.1896, respectively. In contrast, nonsynonymous substitutions predominated in HVR1 in both disease groups. The dN/dS ratios were above 1 at 7 of 12 time points in the subjects with mild disease and 7 of 16 time points in the subjects with severe disease. The mean dN/dS values of E1 and HVR1 were not significantly different when comparing sequences from the two disease groups, although a trend towards differences between groups was evident in the E1 gene (mean dN/dS of 0.2627 versus 0.1896 for the mild versus severe disease group; P = 0.08). Genetic distance was also compared between the two disease groups with respect to both nucleotide and deduced amino acid sequences. There was no significant difference in genetic distance in either E1 or HVR1 between the subjects with mild disease and those with severe disease.

Table 1.

Comparison of intra-specimen genetic parameters between the subjects with mild disease and the subjects with severe disease.

			E1					HVR1
	No. of subjects	No. of specimen	dS	dN	dN/dS	Nucleotide distance	Amino acid distance	dS	dN	dN/dS	Nucleotide distance	Amino acid distance
Mild	7	18	0.0232	0.0044	0.2627	0.0092	0.0091	0.0376	0.0586	1.4240	0.0576	0.1393
			(0.0040)	(0.0007)	(0.0331)	(0.0013)	(0.0010)	(0.0113)	(0.0211)	(0.3944)	(0.0179)	(0.0462)
Severe	7	20	0.0348	0.0050	0.1896	0.0124	0.0115	0.0459	0.0527	1.2283	0.0592	0.1324
			(0.0048)	(0.0011)	(0.0265)	(0.0018)	(0.0025)	(0.0139)	(0.0166)	(0.2864)	(0.0154)	(0.0349)
P-value			0.10	0.92	0.08	0.18	0.86	0.54	0.72	1.00	0.59	0.46

Next, intra-specimen dS, dN, dN/dS ratio, genetic distance and Shannon entropy were analyzed as a function of duration of infection for the time periods of 5–10 years, 11–19 years and 20–25 years after infection (Figure 1 and Figure 2). Wide fluctuations in HCV genetic parameters were observed over time in both groups of subjects. However, as shown in Figure 1A and Figure 1B, viral populations in the individuals with severe disease accumulated significantly more synonymous and nonsynonymous substitutions in the E1 region during the three decades of chronic HCV infection compared to the individuals with mild disease (P = 0.03 for dS, and P = 0.02 for dN). For HVR1, there was no consistent trend in accumulation of synonymous and nonsynonymous substitutions in either disease group (Figures 2D and 2E), although the mean dN/dS value of HVR1 exhibited a decreasing trend over time in the subjects with severe disease (Figure 2F). A detailed illustration of the intra-specimen genetic distance and Shannon entropy of the E1 region is shown in Figure 2. Comparison of the nucleotide and amino acid distances showed no significant changes over time in subjects with mild disease, but revealed an increase of genetic distances in those with severe disease, which was statistically significant between the first and third decades (P = 0.01 for both nucleotide and amino acid distances, respectively) (Figures 2A and 2C). At the time period 20–25 years after infection, the mean E1 genetic distances in the subjects with severe disease were higher than those in subjects with mild disease, for both nucleotide and amino acid populations (P = 0.02 and 0.06 for nucleotide and amino acid distances, respectively). For HVR1, the genetic distances were relatively stable over decades in both disease groups (data not shown). Shannon entropy was also plotted to determine whether the viral population variability differed according to disease progression (Figures 2B and 2D). In the mild disease group, nucleotide and amino acid entropy exhibited only moderate variation in the E1 region during HCV infection, whereas both indices increased in the severe group in a pattern similar to that observed for the E1 genetic distances. The differences between the first and third decades were statistically significant for the nucleotide as well as the amino acid population (P = 0.01 and P = 0.02 for nucleotide and amino acid distances, respectively). Again, the Shannon entropy did not exhibit significant variation in HVR1 over time in either disease group (data not shown). Taken together, the data indicate that higher synonymous and nonsynonymous substitution rates in E1, along with a greater increase of quasispecies diversity and complexity in E1, are closely associated with a higher likelihood of progression to severe liver disease during long-term chronic HCV infection. The relationship between E1 evolution and hepatitis C disease severity is completely opposite from the relationship between HVR1 evolution and hepatitis C disease severity we recently described (higher HVR1 genetic diversification was associated with mild disease progression) (Sullivan et al., 2007).

Figure 1.

Variation of intra-specimen dS, dN and dN/dS ratio of the E1 and HVR regions according to years of disease progression in the mild and severe disease groups. Panel A, synonymous substitution rate of the E1 region in the mild and severe disease groups. Panel B, non-synonymous substitution rate of the E1 region in the mild and severe disease groups. Panel C, non-synonymous to synonymous (dN/dS) ratio of the E1 region in the mild and severe disease groups. Panel D, synonymous substitution rate of the HVR1 region in the mild and severe disease groups. Panel E, non-synonymous substitution rate of the HVR1 region in the mild and severe disease groups. Panel F, non-synonymous to synonymous (dN/dS) ratio of the HVR1 region in the mild and severe disease groups.

Figure 2.

Variation of intra-specimen genetic distance and Shannon entropy of the E1 region according to years of disease progression in the mild and severe disease groups. Panel A, intra-specimen nucleotide distances of the E1 region in the mild and severe disease groups. Panel B, intra-specimen nucleotide entropy of the E1 region in the mild and severe disease groups. Panel C, intra-specimen amino acid distances of the E1 region in the mild and severe disease groups. Panel D, intra-specimen amino acid entropy of the E1 region in the mild and severe disease groups.

Inter-specimen genetic parameters

Although intra-specimen genetic parameters serve as a good indication of diversity and complexity of viral quasispecies at any one point in time, they only provide limited insight of accumulation of nucleotide substitutions and changes of viral population over time. To further investigate the latter aspect of HCV quasispecies biology, measurements of nucleotide substitutions and genetic distances were compared between envelope genes present in any paired specimens obtained from the same individual at different time points. Table 2 compares inter-specimen genetic data from HCV-infected mild (20 specimens pairs) and severe (27 specimen pairs) disease groups. Similar to the intra-specimen analysis, synonymous substitutions predominated in the E1 region when data from both disease groups were combined (dN/dS < 1). Although the two disease groups exhibited similar rates of synonymous nucleotide substitutions in the E1 region (P = 0.60), the rates of non-synonymous substitutions were significantly higher in E1 genes from the mild disease group (P = 0.01), resulting in higher dN/dS ratios in the individuals with mild disease (0.2213 in the mild disease group versus 0.1332 in the severe disease group; p < 0.01). In HVR1, the mild disease group exhibited significantly higher values of both synonymous substitutions (0.1034 in the mild group versus 0.0704 in the severe group, P = 0.02) and non-synonymous substitutions (0.1912 in the mild group versus 0.1312 in the severe group, P = 0.03), while the overall dN/dS ratios remained similar for both disease groups (2.570 in the mild group versus 2.407 in the severe group, P = 0.26). Genetic distances between the paired specimens were also compared between the two disease groups. HVR1 quasispecies present in the paired samples showed higher changes in diversity and complexity over time in the individuals with mild disease than in those with severe disease with respect to both nucleotide distances (0.1826 in the mild group versus 0.1360 in the severe group, P = 0.03) and amino acid distances (0.4405 in the mild group versus 0.3404 in the severe group, P = 0.07). E1 quasispecies in the mild disease group also exhibited higher amino acid distance compared to the quasispecies in the severe group (0.0212 in the mild group versus 0.0155 in the severe group, P = 0.04).

Table 2.

Comparison of inter-specimen genetic parameters between the subjects with mild disease and the subjects with severe disease.

			E1					HVR1
	No. of subjects	No. of pairwise specimen	dS	dN	dN/dS	Nucleotide distance	Amino acid distance	dS	dN	dN/dS	Nucleotide distance	Amino acid distance
Mild	7	20	0.0581	0.0110	0.2213	0.0222	0.0212	0.1034	0.1912	2.570	0.1826	0.4405
			(0.0072)	(0.0014)	(0.0190)	(0.0027)	(0.0026)	(0.0137)	(0.0232)	(0.4680)	(0.0179)	(0.0454)
Severe	7	27	0.0606	0.0071	0.1332	0.0198	0.0155	0.0704	0.1312	2.407	0.1360	0.3404
			(0.0041)	(0.0010)	(0.0186)	(0.0014)	(0.0021)	(0.0083)	(0.0143)	(0.1514)	(0.0112)	(0.0280)
P-value			0.60	0.01	< 0.01	0.50	0.04	0.02	0.03	0.26	0.03	0.07

Next, the inter-specimen genetic parameters were analyzed relative to the more recent time point of the paired samples for the time periods of 5–10 years, 11–19 years and 20–25 years after infection, respectively (Figure 3 and Figure 4). No significant differences were observed according to decade of infection for HVR1 between the two disease groups (data not shown). For the E1 region, while the synonymous substitution rate remained relatively stable in the mild group, there was a significant increase in the dS rate in the individuals with severe disease between the first and the third decades (P = 0.04) (Figure 3A). At the same time, however, little fluctuation was observed for the nonsynonymous substitution rate or the dN/dS ratio in those subjects (Figures 3B and 3C). E1 quasispecies present in the paired samples from subjects with severe disease showed a progressive increase in nucleotide distances over time between the first and the third decades of infection (P = 0.02), whereas E1 quasispecies from subjects with mild disease did not show this evolutionary pattern (Figure 4A).

Figure 3.

Variation of inter-specimen dS, dN and dN/dS ratio of the E1 region according to years of disease progression in the mild and severe disease groups. Panel A, synonymous substitution rate of the E1 region in the mild and severe disease groups. Panel B, non-synonymous substitution rate of the E1 region in the mild and severe disease groups. Panel C, non-synonymous to synonymous (dN/dS) ratio of the E1 region in the mild and severe disease groups.

Figure 4.

Variation of inter-specimen genetic distance of the E1 region according to years of disease progression in the mild and severe disease groups. Panel A, inter-specimen nucleotide distances of the E1 region in the mild and severe disease groups. Panel B, inter-specimen amino acid distances of the E1 region in the mild and severe disease groups.

Phylogenetic analysis and the co-evolution coefficients between E1 and HVR1 of E2

Phylogenetic reconstruction was performed on nucleotide sequence alignments to assess the specific evolutionary patterns of E1 and HVR1 according to disease progression. Phylogenetic trees for representative subjects from both disease groups are shown in Figure 5. As seen in the left part of the figure, the E1 quasispecies sequences diverged away from the initial viral population in both disease groups, with similar patterns of evolutionary drift. Interestingly, the sequence cluster of the third time point (1996) from the subject with severe disease drifted toward the initial time point (1989) instead of diverging away as a discrete lineage, suggesting that these variants were selectively advantageous between the second and third time points and were fixed as the major viral population. Divergent evolution was also evident in HVR1 from subjects in the mild disease group, with three distinct evolutionary clusters corresponding to the three time points, which exactly corresponded to the clusters in the E1 phylogenetic tree (Figure 5A, the left panel). However, HVR1 sequences from the representative subject with severe disease were closely related to each other, and showed little evidence of divergent evolution when variants at different time points were compared (Figure 5B, right panel).

Figure 5.

Phylogenetic trees constructed according to the amino acid sequences of the E1 and HVR regions obtained from representative individuals within the mild and severe disease groups. The trees were constructed with the neighbor-joining method, and Poisson-correction distances were shown with each branch. Panel A, phylogenetic trees based on the E1 (on the left) and HVR (on the right) amino acid sequences from a representative individual with mild disease. Sequences at various time points are indicated by various symbols (□, 1987; ■, 1988; ○, 1996). Panel B, phylogenetic trees based on the E1 (on the left) and HVR (on the right) amino acid sequences from a representative individual with severe disease. Sequences at various time points are indicated by various symbols (□, 1989; ■, 1994; ○, 1996; ●, 1997).

To evaluate E1 and HVR1 co-evolution and its association with the disease progression, we compared interrelatedness between the E1 and HVR1 phylogenetic trees from each study subject. Similarity between the phylogenetic trees was evaluated for statistical association of correlation coefficients, and interpreted as an indication of coordinated evolution under a similar evolutionary pressure (Pazos et al., 2005; Pazos and Valencia, 2001). Genetic distance was calculated for each individual E1 or HVR1 sequence by comparing the sequence to all the other E1 or HVR1 sequences obtained from the same subject. Then, a linear correlation coefficient was evaluated between the data set of E1 and HVR1 genetic distances for each subject. All of the correlation coefficients are plotted according to disease groups in Figure 6. The correlation of amino acid distances between E1 and HVR1 was significantly higher among the subjects with mild disease compared to those with severe disease (r = 0.44 versus r = 0.17, P = 0.01). These data indicate the E1 glycoprotein and HVR1 of the E2 glycoprotein evolve at a higher coordinate level in the subjects with mild disease compared to the subjects with severe disease.

Figure 6.

Linear correlation coefficients between genetic distances of the E1 and HVR regions in the mild and severe disease groups. Panel A, linear correlation coefficients between nucleotide distances of the E1 and HVR regions in the mild and severe disease groups. Panel B, linear correlation coefficients between amino acid distances of the E1 and HVR regions in the mild and severe disease groups.

Amino acid comparisons and signature sequence analysis of E1

To determine whether specific amino acid usage patterns were associated with disease outcomes, the deduced consensus amino acid sequences of E1 from each subject were assessed and compared between the mild and severe disease groups (Figure 7). HVR1 amino acid comparisons between disease groups was previously reported (Sullivan et al., 2007). The 14 consensus E1 sequences exhibited complete conservation at 126 of the 192 (66%) amino acid positions, suggesting that the E1 region is under a high degree of purifying selection. There were 5 putative N-linked glycosylation sites (NXT or NXS) (Goffard and Dubuisson, 2003; Meunier et al., 1999) and 8 cysteine residues present throughout the E1 region, and all of them were highly conserved among the 14 sequences except one potential glycosylation site which presented an N to S mutation in 58% of sequences from a subject with severe disease (S030).

Figure 7.

Alignment of the concensus amino acid sequences of the E1 region from each individual subject with mild (M) and severe (S) disease. *, putative N-linked glycosylation site (NXT or NXS); #, cysteine residue.

VESPA program was used to search for signature motifs between the mild and severe disease groups. Three amino acid positions were identified at which the majority amino acid differed according to disease groups: position 219 with 2T 2A 2M 1V in the mild group and 4A 2M 1T in the severe group, position 235 with 3S 2A 1I 1T in the mild group and 2V 2T 1A 1I 1S in the severe group, and position 359 with 4L 3I in the mild group and 4I 2L 1V in the severe group. However, none of the majority amino acids at these positions was unique to either group of subjects. The pI values were evaluated for the consensus E1 sequences from each individual subject; sequences from the subjects with severe disease were more positively charged than those from subjects with mild disease (pI = 8.16 versus pI = 7.32, P = 0.14).

DISCUSSION

In the present study, the evolution of the HCV envelope glycoproteins was analyzed in patients selected from a cohort with long-term chronic HCV infection followed prospectively to determine viral factors associated with the progression to cirrhosis. Subjects with HCV genotype 1 infection and who had not received any antiviral therapy were recruited and divided into two groups, one with severe disease and one with mild disease, based on histological findings on liver biopsy. 362 nucleotide sequences encompassing the E1 and HVR1 regions were generated from sequential serum specimens, and subjected to various phylogenetic analyses. The evolutionary trend of the E1 and HVR1 regions differed markedly according to disease status. As previously reported in the HCV-infected liver transplant model (Arenas et al., 2004; Gretch et al., 1996; Lyra et al., 2002; Sullivan et al., 1998), the HCV and HIV coinfection model (Qin et al., 2005), and during natural infection (Curran et al., 2002; Sullivan et al., 2007), reduced genetic diversification of HVR1 over time correlated with severe disease. In contrast, the present results demonstrate that during natural infection, severe disease was associated with increased diversification of the E1 glycoprotein gene over time, with progressive accumulation of both synonymous and nonsynonymous mutations. Comparison of E1 and HVR1 phylogenetic trees demonstrated that the two regions were less evolutionarily correlated in the individuals with severe disease.

A major limitation of the study is the relatively small sample size, making it premature to draw conclusions on the association between HCV glycoprotein evolution and disease progression in hepatitis C. For example, dN/dS ratios in HVR1 appeared to decline over time during severe disease, yet increase over time during mild disease (Figure 1, panel F). Such data might imply that immune pressure on HVR1 decreased over time during disease progression. However, the data did not reach statistical significance, possibly because of the limited sample size. Therefore, larger studies are needed to confirm the present findings. A major strength of this study is that the two disease groups were very well defined and matched (e.g., all therapy naïve, all infected with genotype 1, concomitant infections rigorously excluded, equivalent alcohol use, all were indigenous American Peoples), disease endpoints were well defined by liver biopsy, sampling was extensive, and quality genetic data were obtained. Verification of these findings is warranted by the potential clinical significance of HCV quasispecies biology, including the association between low HVR1 diversity and more severe disease outcome (Sullivan et al., 2007), and the association between high HVR1 diversity and poor response to interferon-based therapies (Koizumi et al., 1995; Morishima et al., 2006; Pawlotsky et al., 1998; Polyak et al., 1997).

Nucleotide substitutions of HVR1

HVR1 of E2, with the highest nucleotide substitution rate of the HCV genome, has been used as the most popular target in molecular studies of HCV quasispecies heterogeneity and phylogenies (Alfonso et al., 2004; Allain et al., 2000; Curran et al., 2002). Host immune pressure is considered to be the primary force driving quasispecies evolution in this region (Booth et al., 1998; Farci et al., 1994; Farci et al., 1996; Ray et al., 1999; Shimizu et al., 1994; Shimizu et al., 1996). With nonsynonymous substitutions predominant, the evolution of HVR1 occurs under immune pressure, either by fixation of spontaneous mutations at specific sites, or by shifts of major variants in a quasispecies (Lu et al., 2001). Variations are unequally distributed within HVR1, with the presence of both highly variable and highly conserved positions. It has been suggested that the positively selected sites are most likely to be surface-exposed, and therefore prime targets for host antibody responses (Wack et al., 2001; Yagnik et al., 2000). At the same time, a negative pressure is probably exerted against conserved positions essential for conformation conservation and/or host cell attachment (Penin et al., 2001). Previous investigators have proposed that HVR1 acts as an immunologic decoy during HCV infection, stimulating a strong yet ineffective immune response that results in a genetically unstable population without producing significant growth disadvantages or effective antiviral control (Ray et al., 1999; Sheridan et al., 2004). The present study could not determine the extent that quasispecies homogenization resulted from reduced immune pressure, versus accumulation of variants that successfully escaped antibody responses, versus formation of an equilibrium between the viral population and host immune responses.

Nucleotide substitutions of E1

E1 is a type-I transmembrane glycoprotein with a short C-terminal transmembrane domain of about 30 amino acids (Deleersnyder et al., 1997; Op De Beeck et al., 2000). E1 shares sequence similarity to the scaffold of class II membrane-fusion protein as observed in other members of the Flaviviridae family (Garry and Dash, 2003). By forming noncovalent heterodimers with E2, E1 plays pivotal roles in binding to receptors during host cell recognition and inducing fusion with the cell membrane (Bartosch et al., 2003). In contrast to HVR1 of E2, the E1 glycoprotein is reported to have a high degree of sequence conservation, appears to be less immunogenic in vivo, and by inference, is less likely to be surface-exposed (Allison et al., 2001; Fournillier et al., 2001). Our present results confirm that the dN/dS ratio of the E1 region is significantly lower than 1, regardless of patient disease status, suggesting that the majority of positions within E1 are under strong negative selection with amino acid constraints. The data may indicate a need to conserve E1 protein modification and/or folding processes that determine the glycoprotein confirmation and/or specific virus-cell receptor interactions during viral recognition and attachment to the host cells.

HCV glycoprotein dynamics and disease progression

Absolute nucleotide substitution rates can provide great insight into factors involved in a change of the evolutionary process (Seo et al., 2004). Synonymous substitution rates can be affected by differences in viral generation time or reproductive rate, while nonsynonymous substitution rates can be altered by either viral generation time, changes in patterns of selection, or effective viral population size. In the present study, inter-specimen synonymous and nonsynonymous substitution rates of the HVR1 region were consistently higher in subjects with mild disease, and viral quasispecies accumulated more genetic heterogeneity between different time points compared to virus infecting subjects with severe disease. In the post-transplant setting, subjects with severe disease had significantly reduced CD4+ T cell responses against HCV antigens compared to subjects with mild disease (Rosen et al., 1999), suggesting that deficient host immune responses may contribute to quasispecies homogenization in HVR1 during severe disease. Unfortunately, immunology data are not yet available for the present cohort.

In contrast to HVR1, E1 glycoprotein genetics accelerated during severe disease, with increasing dS and dN substitution rates resulting in increasing genetic heterogeneity at both the nucleotide and amino acid levels. Increases in both dS and dN within a poorly immunogenic region (such as found in E1) suggest increased replication rate. One hypothesis concerning HCV replication dynamics is that variants with increased replication fitness acquire survival advantages, and become the dominant population during progression of severe disease, as has been proposed for HIV disease (Lemey et al., 2007; Nowak and May, 2000). If true, viral emergence might be associated with reduced host pressure (i.e., immune exhaustion), viral escape, virus-immune equilibrium, or viral knockout of host control mechanisms. Alternatively, the data may be interpreted as resulting from higher turnover of infected cells, as proposed by Wang et al. (Wang et al., 2007).

Previous studies have defined important interactions between the E1 and E2 glycoproteins of HCV. Therefore, we analyzed amino acid relationships between the two glycoproteins, according to disease status. In subjects with mild disease, primary amino acid sequences in E1 and HVR1 remained more closely related over time compared to the same regions in subjects with severe disease. The data suggest fundamental differences in co-evolutionary dynamics between the HCV E1 and HVR1 glycoprotein domains in subjects with mild versus severe disease, perhaps indicating differences in the ability of virus to escape immunity, and/or differences with respect to efficiency and specificity of cell receptor usage.

In a recent study of HCV replicative intermediate RNA using strand-specific in situ hybridization, we reported a highly significant association between higher indices of intrahepatic HCV replication and more severe liver disease (Pal et al., 2006). Diseased livers also have higher levels of HCV non-structural antigens than non-diseased livers (Gretch et al., 1995; Hiramatsu et al., 1992). However, HCV serum viral loads do not increase during progression to severe disease, at least during natural infection (Sullivan et al., 2007). The most logical explanation is hepatic exhaustion, i.e., that more extensive liver damage results in fewer liver cells that are capable of supporting HCV replication. Alternatively, viral variants may become cell-associated, due to decreased efficiency of virogenesis as a result of intracellular congestion and/or toxicity associated with accumulation of viral replication products, even though viral replication is more robust. Thus, the relationship between viral population size in circulation and viral replication rate in liver may be different in subjects with mild versus severe disease, a possibility that deserves further study.

N-linked glycosylation

There are five potential N-linked glycosylation sites over the HCV E1 glycoprotein (Goffard and Dubuisson, 2003; Meunier et al., 1999). N-linked glycosylation sites, necessary for correct glycoprotein confirmation and cellular entry, can be sites of occurrence of adaptive mutations (Goffard et al., 2005; Huang et al., 1997; Li et al., 1993; Wu et al., 1995). Changes in N-linked glycosylation of the HIV envelope gene product, gp120, have recently been proposed as important in neutralization escape and viral persistence (Choisy et al., 2004; Wei et al., 2003). In our dataset, only one glycosylation site was detected missing, and this mutation occurred in only one subject. These data suggest that in contrast to the glycan shield of HIV, which can be rearranged to provide protection against neutralizing antibodies, the major function of glycosylation of HCV E1 glycoprotein is to ensure correct glycoprotein processing, folding and cellular entry, rather than immune shielding.

In summary, the data indicate that divergent evolution of E1 and decreasing quasispecies heterogeneity in the E2 hypervariable region strongly associated with progression to severe chronic hepatitis C disease. Accumulation of synonymous mutations in E1 accompanied by homogenization of sequences in HVR1 implicate viral outgrowth as secondary to fitness advantage. The collective data can be interpreted to indicate that during severe disease, variants with increased replication fitness (based on E1 genetics) successfully escaped from or equilibrated with the neutralizing antibody responses (based on HVR1 genetics). We hypothesize that HCV can adapt to non-sterilizing, disease-arresting antiviral pressures of the infected host, and that such adaptation somehow triggers increasing severity of liver disease. To prove the hypothesis requires systems to discriminate whether viral genetic changes trigger disease, or whether they are simply a consequence of other mechanisms driving disease progression.

MATERIALS AND METHODS

Subjects and samples

The 14 subjects included in the present study were recruited from the cohort of Alaska Natives and American Indians (AN/AI) cohort, and were included in previously published studies of the long term patterns of HCV genotype 1 infection in humans (McMahon et al., 2004; Sullivan et al., 2007). All subjects were negative of HIV and HBV coinfections, and all were interferon treatment-naïve. Disease status was inferred from current liver biopsies that were systematically evaluated by a single study pathologist (HD) according to the Knodell scoring system (Ishak et al., 1995; Knodell et al., 1981; Scheuer et al., 2002). Seven subjects were classified with severe disease (bridging fibrosis or cirrhosis) and seven were classified as with mild, non-progressive disease. The date of HCV infection was estimated based on risk factor data from patient interviews (the date of first intravenous drug use or pre-1992 blood transfusion) and was correlated with antibody seroconversion dates (Bruden et al., 2004). A total of 38 serial serum samples were collected from the 14 individuals over 81 follow-up years, and stored at −80°C for future RNA extraction.

Cloning and sequencing of E1 and HVR1

Reverse-transcription PCR (RT-PCR), cloning and sequencing was performed as previously described by our laboratory (Wilson et al., 1995). RT-PCR was performed in a 25 µL volume in the presence of anti-sense primer JBE1-1as (1639–1600, AGGCTTTCATTGCAGTTCAAGGCCGTGCTATTGATGTGCC) at 37°C for 1 h using the M-MLV reverse transcriptase kit (Invitrogen, Carlsbad, CA). The resulting cDNA was used as template in nested PCR to amplify a 0.84 kb fragment encompassing the E1 glycoprotein gene and HVR1 of the E2 glycoprotein gene. Aliquots (10 µL) of cDNA were mixed with 0.5 µM anti-sense primer JBE1-1as and sense primer JBE1-1s (802–841, GCGTCCGGGTTCTGG-AAGACGGCGTGAACTATGCAACAGG), 1 mM dNTPs and 0.5 U Platinum Taq Polymerase High Fidelity (Invitrogen, Carlsbad, CA) in a 25 µL reaction volume with 1 × PCR buffer. After an initial denaturation at 95°C for 2 minutes, PCR was carried out for 35 cycles with 30 seconds of denaturation at 95°C, 30 seconds of annealing at 55°C, and 1 minute of extension at 72°C, followed by a final extension at 72°C for 5 minutes. Then, one microliter of the first-round PCR product was used as template in second-round reactions with primers JBE1-2s (817–856, AAGACGGCGTGAACTATGCAACAGGG-AACCTTCCTGGTTG) and JBE1-2as (1626–1587, AGTTCAAGGCCGTGCTATTGATGTGCCAACTGCCGTTGGT). The second-round amplification was carried out using identical conditions to the first-round PCR.

PCR fragments were analyzed by agarose gel electrophoresis and extracted from the gel using Qiagen DNA purification kit (QIAGEN Inc., Valencia, CA). The fragments were ligated into the TA cloning vector (Invitrogen, Carlsbad, CA) and transformed into Escherichia coli TOP-10 competent cells for recovery of plasmid clones. Eight to twelve clones representative of each sequential time point for each subject were selected, amplified with primers JBE1-2s and JBE1-2as, and sequenced at both directions using an ABI 3730XL high-throughout capillary DNA Analyser.

Phylogenetic analysis

Nucleotide sequences were aligned using the CLUSTAL W program (Thompson et al., 1994) included in MacVector version 9.0 (Symantec Corp., Cupentino, CA) with manual adjustment. The rates of inter and intra-sample non-synonymous substitutions (dN) and synonymous substitutions (dS) were estimated using the method of Nei and Gojobori (Nei and Gojobori, 1986) and the Synonymous/Non-synonymous Analysis Program on Los Alamos National Security website (http://hcv.lanl.gov/content/hcv-db/SNAP/SNAP.html) (Korber, 2000). The resulting matrix was used to estimate mean inter- and intra-sample dS, dN and dN/dS ratios. Protein sequences were deduced from the nucleotide sequences using MacVector version 9.0. The mean genetic distance was calculated using the Kimura two-parameter method and the PHYLIP version 3.5 package (Felsenstein, 1993). Shannon entropy was determine using the Entropy program on the Los Alamos National Security website (http://hcv.lanl.gov/content/hcv-db/ENTROPY/entropy.html) (Korber et al., 1994). Phylogenetic trees were constructed using the neighbor-joining method (Saitou and Nei, 1987) and MacVector version 9.0. Signature pattern analysis was performed using the Viral Envelope Signature Pattern Analysis (VESPA) program (Korber and Myers, 1992).

Statistical analysis

All statistical calculations were performed using SPSS Base 11 (SPSS Inc., Chicago, IL). To determine the associations between patient characteristics and disease progression, independent-sample Mann-Whitney tests were used to compare age, viral load, age at HCV infection and length of HCV infection between the mild and severe disease groups, while Fisher exact tests were used for evaluating gender and genotype. Independent-sample Mann-Whitney tests were also used to analyze differences between the genetic parameters of mild and severe subjects as a whole group or at various time points. The linear correlation coefficient (r) was calculated to determine the relationship between genetic distances of the E1 and HVR1 regions. P-values of less than 0.05 were considered to be statistically significant.