Viral envelope glycoproteins play an important role at different steps of the viral life cycle. During virion morphogenesis, they take part in the assembly process, whereas in the early steps of the viral life cycle, they are involved in receptor binding and in fusion between the viral envelope and a cellular membrane. To fulfill these functions, viral envelope glycoproteins have to adopt dramatically different conformations at these different steps of the infectious cycle. Importantly, these conformational changes have to occur at a precise time and thus have to be tightly controlled. Despite extensive research on HCV envelope glycoproteins, the structures and functions of these proteins remain poorly understood. Here, we used a genetic approach to identify functional determinants in HCV glycoprotein E2. By generating HCVpp containing E1 and E2 from different genotypes, we identified intergenotypic incompatibilities between these two proteins. By using a nonfunctional E1E2 complex, we identified new functional regions in E2 by exchanging protein regions between two incompatible genotypes. This led to the identification of several determinants of the E2 ectodomain that play a role in E1E2 assembly (Table ). Furthermore, we also characterized the structural and lipid binding features of the C-terminal part of the ST segment by CD and NMR using a synthetic peptide denoted E2-SC. This segment, which is involved in HCV entry and located close to the TM domain, includes a central amphipathic helix, which folds upon binding to lipid mimetics. Its features suggest that this helix could easily switch from helical to random conformation, depending on its microenvironment and/or binding partners. Together, these data highlight new functional regions in HCV envelope glycoprotein E2.
Intergenotypic incompatibilities exist between HCV glycoproteins E1 and E2 from different genotypes. Although the overall structure of HCV proteins is not expected to differ significantly between HCV genotypes, coevolution within a genotype or subtype can potentially lead to functional incompatibilities between partner proteins. Such genetic incompatibilities have indeed already been reported for HCV, highlighting potential protein-protein interactions between viral polypeptides (2
). In the case of HCV envelope glycoproteins, biochemical analyses have shown that they assemble in the cell as noncovalent heterodimers (11
). Furthermore, these interactions are important for the cooperative folding of these two proteins (reviewed in reference 38
). It is therefore not surprising that HCV envelope glycoproteins have coevolved in the different genotypes and that this coevolution can lead to functional intergenotypic incompatibilities between E1 and E2. It is however more puzzling that in the case of genotype 1a, E1 was compatible with the E2 of any genotype tested. A potential explanation for this phenotype is that the E1 protein of genotype 1a has more flexibility to accommodate changes in conformation or oligomerization during the fusion process. However, this needs further investigation.
Residues of the ST region close to the TM domain play a major role in HCV entry. Indeed, HCVcc infectivity was reduced by almost 3 logs when segment 705-715 of genotype 1a was introduced in the context of E1E2 of genotype 2a. We have previously shown that the interactions between TM domains play a major role in E1E2 heterodimerization (8
) as well as in virus entry (5
). Since segment 705-715 corresponds to the upstream sequence of the E2 transmembrane domain, these two regions might be functionally connected for the assembly of the E1E2 heterodimer. However, heterodimerization was not affected for the 705-715 chimera, as shown by coprecipitation in a CD81 pull-down assay. Furthermore, particle assembly and release was not affected by this mutation. The incompatibility between genotypes 1a and 2a in the ST region therefore highlights a role for the 705-715 segment in the entry process. The ST region is relatively flexible and is supposed to play a major role in the reorganization of the envelope glycoproteins during the fusion process (34
). Amino acid residues 705 to 715 are therefore likely involved in interactions with E1 region at some stage of the fusion process.
The aa sequence analyses and the structural investigations of the corresponding synthetic peptide E2-SC by CD in various media clearly show that this region exhibits potential lipid binding properties and could fold into α helices upon binding. The three-dimensional structure analysis of this peptide in SDS or 50% TFE used to probe the peptide conformational preferences as well as to mimic the membrane environment revealed that the major structural elements consist of a central amphipathic helix (689 to 703, but it could extend to aa 684 as revealed by SDS analysis) and a C-terminal fraying helix (706 to 714 or so) connected by a short flexible segment, including a glycine residue (704). Both helices require a hydrophobic environment for folding, indicating that lipid interactions and/or protein interactions could contribute to their structural stability. In addition, the fraying of the C-terminal helix could be due to the absence of the downstream transmembrane sequence, which would likely stabilize this helix. Its relative amphiphilicity and its connection with the TM of E2 suggest that this helix should be located at the membrane interface, maybe with an in-plane topology. The amphipathic nature of the highly conserved central helix 689-703 together with the helix folding upon binding to lipid mimetics suggest that this helix could bind in-plane to a membrane interface. However, the relatively low free energy of membrane association, as calculated using the MPEX program (74
), suggests that this helix could be easily released from the membrane interface. This is consistent with the limited stability of this amphipathic helix in SDS. One can thus hypothesize that this helix is able to ensure a conformational transition between helix folding when bound to the membrane interface and a coil state upon membrane release. According to the conformational change of the stem region proposed in the model of fusion events for class II glycoproteins (22
), it is tempting to speculate that this amphipathic helix could ultimately fold again upon binding to the trimeric postfusion complex of glycoproteins.
HVR2 and IgVR are essential determinants for HCV particle assembly. Based on results with HVR1 mutants, it was thought that the most variable regions on of E2 would be dispensable for HCV infectivity. Indeed, although it reduces infectivity, the deletion of HVR1 in the context of an infectious virus is not lethal (21
). Furthermore, deletion of the three variable regions HVR1, HVR2, and IgVR in a truncated form of E2 does not seem to affect its folding as measured by binding of conformation-dependent MAbs and CD81 pull-down (44
). However, very recent data indicate that deletion of HVR2 or IgVR is lethal in the HCVcc system, suggesting that the presence of these regions in E2 can play a functional role in virus assembly (45
). In our case, we used a less drastic approach consisting of replacing these regions with the corresponding segment from another genotype. In such chimeric viruses, introducing HVR2 or IgVR from genotype 1a in the context of an infectious clone of genotype 2a was also lethal for the production of infectious virus, which was due to an alteration in the production of viral particles as measured by a core release assay.
Exchanging HVR2 or IgVR affects the incorporation of E1 into HCVpp. In the context of the HCVpp system, the recognition of the chimeric E2 by CD81 suggests that at least domain DI of this protein is properly folded after replacing HVR2 or IgVR by the corresponding region from another genotype. However, the level of incorporation of the E1 into HCVpp was barely detectable for these chimeras. This lack of E1 incorporation into HCVpp is in contrast with the interactions between E1 and E2 as observed in the CD81 pull-down assay. However, further analyses of the intracellular E1E2 complexes recognized by CD81 indicated that E1 formed disulfide bond-linked high-molecular-weight complexes with these chimeric E2 proteins (data not shown). Interestingly, we have recently shown that in the context of the HCVcc system, virion-associated E1 and E2 envelope glycoproteins form large covalent complexes stabilized by disulfide bridges, whereas the intracellular forms of these proteins assemble as noncovalent heterodimers (70
). The presence of disulfide bridges between HCV envelope glycoproteins suggests that lateral protein-protein interactions assisted by disulfide-bond formation might play an active role in the budding process of HCV particles. Therefore, we cannot exclude the possibility that in the context of our chimeric viruses, the introduction of HVR2 or IgVR from another genotype leads to the formation of premature intermolecular disulfide bonds within infected cells, which is no longer coordinated with the other steps of the assembly process. It is worth noting that, when the whole ectodomain of E2 was replaced by the corresponding sequence from genotype 1a, E1 was correctly incorporated into HCVpp, suggesting that chimeras containing HVR2 or IgVR induce an additional defect, which is likely due to interdomain incompatibilities within E2. There might indeed be molecular cross-talk between HVR2 or IgVR and domain DIII and/or the ST region, since a defect in assembly was also observed when we dissociated the DI-DII region from the DIII-TM region (Fig. ).
In addition to interdomain incompatibilities within E2, we also identified a genetic incompatibility within a single E2 domain. Indeed, exchanging sequences between genotypes 1a and 2a within domain DI can lead to protein misfolding. as shown by replacing DIb of genotype 2a with the equivalent sequence from genotype 1a, which led to the absence of recognition of E2 by CD81. Importantly, infectivity and CD81 binding were restored when both DIa and DIb of genotype 1a were introduced in the context of E1E2 of genotype 2a. Therefore, our data indicate that the two parts of the DI domain interact together to form the major CD81 binding determinant, which is in agreement with the recently proposed E2 model (34
). It is worth noting that the chimeric E2 protein containing DI domain from genotype 1a was still less efficiently pulled down by CD81, suggesting that in addition to the major CD81 determinants identified in DI (13
), other residues within DIII might also modulate CD81 binding, as suggested previously (34
). However, we cannot exclude the possibility that the involvement of domain DIII is indirect since mutations in this domain, which affect CD81 binding, can affect DIII folding (31
In this study, we cannot exclude the possibility of some effect of the changes introduced in E1 in reconstructing the A4 epitope. Indeed, we observed some decrease in HCVpp infectivity when the A4 epitope was engineered with E1 of genotype 2a, suggesting some modulation of E1E2 interaction mediated by this region. However, JFH1 infectivity was very similar in the presence or absence of the modified epitope, suggesting that the change in the A4 epitope has only minor effects in the HCVcc system.
In conclusion, we identified several important determinants of E2 ectodomain that are involved in virion assembly or in HCV entry. We also revealed the presence of a well conserved amphipathic helix in the stem region, which likely undergoes conformational changes during the fusion process. Together, these data highlight the complexity of the intermolecular interplay between E1 and E2.