In this study, we investigated the role of the E1/E2 heterodimer during the HCV life cycle and further analyzed the function of a five-amino-acid stretch in the E2 stem region that is essential for the function of the envelope proteins. By exchanging the E2 gene of a previously developed 1a (H77 core-NS2)/2a JFH1-based recombinant with the corresponding gene from genotype 1a, 1b, or 2a isolates, we found very different effects on viral viability. Although exchange of E2 with that of a different isolate from the same subtype (TN) conferred immediate viral spread without the need for compensatory mutations, no viral recovery was observed when exchanging E2 from a different genotype (J6, genotype 2a) (). Interestingly, when exchanging E2 from another subtype (J4, DH1, and DH5, genotype 1b), spread of infection was delayed. We identified single mutations within aa 706 to 710 of the E2 stem region in the recovered 1b-E2 recombinant viruses. Introduction of these mutations increased viral titers on average ~100-fold, confirming a compensatory effect.
The identified mutations, i.e., V710F or V710G in J4-E2, G706R in DH1-E2, and S707L in DH5-E2 (), were analyzed in a multiple-sequence alignment containing 513 HCV E2 sequences from genotype 1 (44
) (see Fig. S1 in the supplemental material, and data not shown). S707 was conserved in all genotype 1 sequences, and the same was observed in subtype 1a and 1b for G706, although subtype 1c had S706. At position 710, the amino acid distribution was more heterogeneous, with genotype 1 sequences containing V (~67%), A (~30%), T (~3%), or I (<1%). Interestingly, none of the sequences included in the alignment contained the residue 706R, 707L, 710F, or 710G, which we identified from the recovered 1b-E2-exchanged viruses. When analyzing the influence of the G706R, S707L, and V710F mutations on viability separately for all 1b-E2 H77C/JFH1V787A Q1247L
recombinants, we found the J4-E2 and DH1-E2 functionality to be dependent on the mutations originally identified for these particular recombinants (V710F and G706R, respectively) (). In contrast, introduction of the three E2 stem region mutations independently into DH5-E2 resulted in immediate viral spread following transfection of these mutants and production of robust infectivity titers. These observations suggested that the unique E2 stem region mutations compensated for the attenuated phenotype of the original 1b-E2 recombinants, primarily impacting the E2 from which the specific mutation was identified.
The fact that the TN-, J4-, DH1-, DH5-, and J6-E2 proteins are all functional in the JFH1-based core-NS2 recombinant from which they were derived (31
) indicates the existence of a genotype-specific interaction between E2 and core, E1, p7, or NS2. Given our findings, particularly in the pseudoparticle system, it seems likely that the functional incompatibilities of the E2-exchanged recombinants are caused by a coevolution of E1 and E2, thereby creating a genotype-specific interaction that to a lesser extent also applies to subtypes, and, furthermore, that the E2 stem region plays an important role in this interaction. Identification of single E1 mutations which in two E2-exchanged isolates fully restored viability () suggested that mutations in E1 compensated either directly for a change in the E1/E2 interaction or indirectly for that of the E1/E2 heterodimer with a third molecule such as a cellular entry factor.
We demonstrated that chimeric H77-E1/1b-E2 HCVpp, although able to enter Huh7.5 cells, have impaired infectivity compared to H77C E1/E2. Others have reported functional chimeric H77C-E1/1b-E2 (Con1 isolate) HCVpp being able to reach infectivity levels comparable to those for the H77 E1/E2 control (47
). These differences in infectivity between H77C-E1/1b-E2 chimeras might be explained by the entry efficiency differences in the genotype 1b isolates used, even though we observed decreased entry for all three 1b isolates. It should be noted that in contrast to our HCVpp infection assays, which were performed with Huh7.5 target cells to better correlate results with the HCVcc data, the previously published study used the original Huh7 target cells (47
While we were not able to adapt the intergenotypic 1a/2a HCVcc construct with E2 from J6, two previous studies report intergenotypic H77C-E1/E2 combinations to be functional for entry, concluding that H77C-E1 is compatible with E2 from the same and different genotypes in the HCVpp system (25
). We speculate that in the HCVcc context, other factors in the HCV life cycle besides entry play an important role in the intergenotypic incompatibility observed here. We cannot rule out, however, that this is a result of the recognized differences between the HCVpp and HCVcc systems with regard to entry (22
We used the CD81-deficient S29 cell single-cycle assay to investigate the impact of the E2 stem region mutations on replication and assembly/release. Introduction of compensatory mutations did not influence replication (). This was expected, since only NS3-NS5B is required for replication (14
). In contrast, compared to that for H77C/JFH1V787A Q1247L
, exchange of the E2 gene resulted in a significant decrease of core release for all 1b-E2 H77C/JFH1V787A Q1247L
recombinants, with this decrease being different in the distinct isolates tested (). The different levels of decrease in core release could be explained by the heterogeneity within the E2 ectodomain, given the importance of the E2 ectodomain for viral assembly (48
). Following introduction of compensatory E2 stem region mutations in 1b-E2-exchanged recombinants, we observed a further reduction in the release of core protein compared to that for the original 1b-E2 H77C/JFH1V787A Q1247L
. Thus, the single compensatory E2 mutations seemed to increase the infectivity of viral particles at the expense of core release, thereby increasing the specific infectivity of the virus.
We observed variations in the levels of incorporation of E1 in both HCVpp and HCVcc between isolates when exchanging the E2 gene. Reduced E1 incorporation levels were most pronounced when exchanging E2 with J4 or DH5 isolates ( and ). Interestingly, the degree of envelope protein integration did not correlate with differences in infectivity following introduction of the E2 compensatory mutations, as peak infectivity titers between isolates were comparable (). Thus, the amount of envelope protein on the virus particle did not seem to be the limiting factor for production of infectious virus from the original E2 exchange recombinants. In the HCVpp system, a previous study reported increased particle infectivity concomitantly with increasing amounts of incorporated E1 and E2 (50
), although this finding cannot be directly compared to our E2-exchaged recombinants with and without compensatory mutation, as other factors, e.g., protein heterodimerization or proper protein folding, could be the limiting factor for infectivity. However, the co-IP CD81-LEL analysis on all H77-E1/1b-E2 recombinants indicated that heterodimerization was maintained following E2 exchange. In agreement with the HCVcc and HCVpp results, the DH1-E2 recombinant with or without G706R had increased envelope protein amounts compared to those of J4-E2 and DH5-E2 (, , and ). This points toward increased heterodimerization ability within H77-E1/DH1-E2 resulting in increased glycoprotein incorporated in the particle or improved CD81 affinity of the DH1-E2 protein. Interestingly, we furthermore found no influence of E2 stem region mutations on the degree of heterodimerization within isolates (). This preserved heterodimerization ability was supported by a previous study investigating specific E2 domains (25
As introduction of the E2 stem region mutations did not enhance intracellular heterodimerization or the release of virus particles as measured by the levels of core, E1, and E2, despite an ~2-log increase in infectivity, we decided to investigate the impact of the mutations on viral particle stability, host cell binding, and CD81 receptor usage. However, none of these were increased by introduction of E2 stem region mutations. The results resembled those from previous reports on the particle thermostability degradation rate (51
) and CD81 receptor dependency for efficient infection (31
). Our combined data suggest that the E2 stem region mutations primarily influence steps downstream of the CD81 receptor interaction in the HCV entry pathway.
Several HCVpp studies have shown that heterodimerization of E1 and E2 is a requirement to obtain infectious HCVpp (26
). Thus, the co-IP results correlated with the observation that infectious HCVpp was produced for all H77-E1/1b-E2 chimeras with or without E2 stem region mutations. Interestingly, introduction of V710F, G706R, and S707L into the corresponding chimeric pseudoparticles H77C-E1/J4-E2, H77C-E1/DH1-E2, and H77C-E1/DH5-E2, respectively, resulted in an average >3-fold increase in infectivity ( and ). Our results are supported by a previous study highlighting functional regions of the E2 glycoprotein, including the importance of the stem region for HCV entry (25
). In the previous study, a 2.5-log titer decrease was observed in the JFH1 genotype 2a infectious cell culture system following exchange of aa 705 to 715 in the E2 stem region with the corresponding sequence of a genotype 1a isolate. In addition, attempts to identify specific stem region residues important for infectivity in a wild-type 2a virus were made. However, none of the tested mutations, including mutations at aa 707 and aa 710, influenced virus viability (25
). In immunoblot detection of E2 in both co-IP and HCVpp assays, we found that intracellular E2 from DH1 and DH5 migrated at a smaller protein size than J4 E2 ( and ). Through amino acid sequence analysis, we identified the presence of 11 previously described N-linked glycosylation sites (54
) in J4-E2, whereas only 10 sites were found in DH1 and DH5 (i.e., glycosylation site 5 [aa 478] was absent). Thus, the difference in protein size could potentially be caused by the absence of this N-linked glycosylation site. However, considering the location of the glycosylation difference and the fact that no differences in infectivity between isolates were observed, we find this unlikely to have an influence on the function of the E2 stem region mutations. The lack of influence on infectivity and heterodimerization of E2 glycosylation site 5 was confirmed in a previous study (55
Based on the heterodimerization and HCVpp results, chimeric E1 and E2 form functional heterodimers able to bind to CD81, which is not increased by E2 stem region mutations. The mutations identified in our study could acts at a later step in entry; one possibility is that they induce conformational changes in the stem region, thereby increasing the functionality of the E1/E2 complex and possibly resulting in an improved unshielding of host cell receptor binding sites or fusogenic domains in late steps of the HCV entry pathway. A previous study has proposed that the stem region has an influence on envelope protein reorganization during the fusion process (56
). Thus, E2 stem region mutations could favor a direct reestablishment of correct envelope protein interaction.
Interestingly, when performing equilibrium gradient density centrifugation for physiochemical analysis of the produced particles, we observed a tendency for peak infectivity titers and peak core levels at slightly higher densities for the original E2-exchanged recombinants in comparison with the positive control and the mutant 1b-E2 recombinants ( and ). This could be due to the high production of noninfectious particles, which might fail to interact with a sufficient amount of lipids, thus increasing their density. Previous studies have elucidated the importance of lipoprotein association to HCV particles in relation to an increased specific infectivity (45
); hence, a decrease in this association could give rise to the low specific infectivity in the original 1b-E2-exchanged recombinants (). We furthermore observed a low specific infectivity in all E2-exchanged recombinants at the densities with peak infectivity titers, indicating the presence of large amounts of noninfectious particles in these fractions (). This was in agreement with findings from other groups, where densities with peak RNA titers correlated to low specific infectivity (58
). In general, the highest specific infectivity was found in the low-density fractions, in agreement with previous findings (22
). Therefore, it cannot be excluded that the viability increase for the mutated 1b-E2 recombinants could be caused indirectly through increased interaction of virion-associated lipid with host cell receptors.
In future studies, it would be of interest to further study host receptor binding, especially at late steps of the HCV entry process, to investigate where the identified mutations in the E2 stem region have specific influence on host cell receptor interaction and thus on virus particle uptake. In addition, it would be of relevance to examine the importance of the E2 stem region using similar approaches for other HCV genotypes.
In summary, we have used a genetic and biochemical strategy to characterize the HCV E1 and E2 glycoprotein function in context of the complete virus life cycle, aiming at a better understanding of their role in HCV infection. To our knowledge, this study is the first to identify single amino acid residues in the E2 stem region with functional importance for viral entry. Further knowledge about residues important for HCV entry could be important for HCV disease and control, as the envelope proteins might constitute a valuable target for future therapy. Furthermore, a better understanding of the HCV envelope protein functionality could have impact on the design of antibody-based HCV vaccine or prophylaxis.