Recently, the crystal structures of HSV-2, EBV, and PrV gH have been determined (5
). In the present study, the functional relevance of conserved structural features within the C-terminal, membrane-proximal part of the ectodomain of PrV gH (domain IV) (, ) was investigated by introduction of 16 different single and multiple nonconservative amino acid substitutions (, ). The fusogenic activity of the mutated gH proteins in the presence of PrV gB, gD, and gL was analyzed in cells cotransfected with expression plasmids, and replication properties of PrV recombinants expressing the mutated gH genes were also investigated. The major findings are the following: (i) substitution of single amino acids within the negatively charged flap by alanine had no or only little effect on protein function, whereas multiple alanine substitutions led to reduced fusion activity; (ii) multiple alanines within the hydrophobic patch, which is masked underneath the flap as revealed by the PrV gH structure, affected fusion as well as penetration and cell-to-cell spread of a corresponding virus mutant; (iii) mutation of the highly conserved N-glycosylation site (N627Q) affected protein function only moderately; (iv) mutation of single cysteines involved in disulfide bonds 3 and 4 flanking the flap () reduced (C547S) or abrogated (C571S) in vitro
fusion activity and impaired virus spread; (v) mutation of cysteine 571 as well as introduction of additional cysteine pairs capable of formation of disulfide bonds between the center of the flap and the hydrophobic patch severely affected virus penetration, cell-to-cell spread, and in vitro
fusion activity, whereas corresponding single cysteine insertions had no effects.
Surprisingly, despite the observed defects, none of the gH mutants was completely nonfunctional. Although PrV gH, like its homologs, is essential for productive virus replication (2
), PrV recombinants expressing any of the mutated proteins could be isolated and propagated in noncomplementing cells. This finding is in general agreement with previous mutational analyses of HSV-1 or varicella-zoster virus (VZV) gH (15
), which also indicated that gH tolerates more sequence alterations than, e.g., gB (41
After high MOI infection, the maximum virus titers of most of the PrV-expressing mutated gH were very similar to those of parental pPrV-ΔgGG or wild-type gH-rescued virus, and only the recombinant lacking disulfide bond 4 (C571S) exhibited an approximate 10-fold reduction in titer. The absence of significant effects on the amount of infectious progeny might be due to the fact that, after successful penetration, PrV gH is dispensable for subsequent virus replication or virion formation (37
). However, our results show that sufficient amounts of all mutated gH proteins were expressed and incorporated to generate infectious virus particles. Western blot analyses of infected cells confirmed that all mutated proteins were present at levels like wild-type gH. Only in PrV mutants carrying gH mutations C571S and E555C/V630C, increased amounts of smaller-than-mature-size gH were detected (), which may be due to slightly impaired protein processing.
Whereas the introduced gH mutations had little effect on formation of infectious virions, cell-to-cell spread of several of the PrV mutants was significantly impaired. However, while gH-deleted PrV is unable to form plaques in noncomplementing cell cultures (2
), all tested gH mutants supported at least limited virus spread in cells transfected with gH expression plasmids or infected with corresponding virus rescuants. The most pronounced defect leading to a plaque size reduction of >80% was again observed after mutating cysteine 571 to serine (C571S), resulting in loss of authentic disulfide bond 4 located C terminal of the flap. Absence of the authentic disulfide bond 3 (C547S) also resulted in a 75% reduction. Mutation of cysteine 571 caused a severe penetration defect of the corresponding virus mutant and abrogated in vitro
fusion activity of gH, whereas mutation of cysteine 547 resulted in only moderate effects. Thus, although disulfide bond 4 is not conserved in HSV (8
), it seems to be more relevant for stabilization of gH structure in PrV and VZV than the other conserved disulfide bond 3 (62
). Since disulfide bond 4 is formed at a CXXC motif, which can function as the active site of protein disulfide isomerases (7
), a general role of this sequence in protein processing appears possible. However, there is no experimental evidence for this enzymatic activity in gH, and the absence of the CXXC motif from HSV-1 and HSV-2 gH, as well as the complementation of mutations of this motif by an unrelated amino acid exchange elsewhere in domain IV as described for VZV (62
), argues against PDI activity of gH. Interestingly, a nonconserved unpaired cysteine is located at position 573 of PrV gH, and it has been speculated that formation of an alternative disulfide bond (C568
instead of C568
) might be involved in the proposed movement of the flap preceding fusion (5
). However, mutation of cysteine 573 was phenotypically inconspicuous, arguing against any functional relevance.
Reduction of the hydrophilicity of the flap, or of the hydrophobicity of the patch, by multiple alanine substitutions also significantly affected fusogenic activity of gH and cell-to-cell spread of the respective PrV mutants, whereas most single amino acid exchanges were fully tolerated. This indicated that not precise sequence motifs but the overall hydrophilicity of these regions is relevant for function. The alanine substitutions in the hydrophobic patch, but not in the flap, also led to significantly delayed penetration kinetics. Such differences between the phenotypes of several of the investigated gH mutants in plaque and penetration assays might reflect different protein requirements for direct cell-to-cell spread than for infection by free virus particles. This is highlighted, e.g., by the fact that PrV gD is required for infection by virions but dispensable for direct viral spread at synapses and other cell junctions (44
). Thus, since gD has been shown to interact with gH during the fusion process in HSV-1 (20
), it is conceivable that different structural features of gH are required during different ways of virus entry.
Interestingly, the exchange of glutamic acid at position 555 by alanine, but not by cysteine, led to significantly reduced plaque sizes of the corresponding PrV rescuant without exhibiting any adverse effect on syncytium formation in in vitro
fusion assays. The observed defect was unlikely to be caused by a mutation elsewhere in the PrV genome since, like the defects of all other PrV mutants analyzed in this study, it was complemented on gH-expressing cells. However, a mutation in gL cannot be excluded, since wild-type gL was coexpressed in RK13-gH/gL cells (37
). In contrast, unwanted mutations at other than the desired sites in gH were excluded by amplification and sequencing of the gH genes of all generated virus mutants. Thus, the structural requirements for gH function in fusion of two adjacent plasma membranes might also be different from those for fusion of the plasma membrane with the envelopes of attached virions during cell-to-cell spread, which occurs at tight junctions (27
). Such mechanistic differences might also explain that removal of cysteine 571-abrogated fusogenic activity of gH but did not block either direct spread or infectivity of released virus particles.
Like the C571S mutation, the introduction of two pairs of cysteines at adjacent positions in the gH structure predicted to form bridges between central parts of the flap and the hydrophobic patch (E555C/V630C, S556C/V631C) caused severe defects in all types of assays performed, whereas neither of the corresponding single cysteine substitutions significantly affected fusogenic activity of gH or in vitro
replication properties of the virus mutants. This finding strongly indicated that the additional disulfide bonds were formed () (5
) and impeded structural changes within domain IV during the fusion process.
As previously demonstrated for HSV-1 (15
), removal of the highly conserved, functional N-glycosylation site at position 627 of PrV gH had no effect on replication kinetics and maximum virus titers. Nevertheless, the mutation resulted in a moderate reduction of plaque size and of syncytia in fusion assays, indicating that glycosylation at this site is beneficial, although masking of the hydrophobic part of domain IV by the flap is obviously sufficient to keep gH largely functional.
Based on the conserved structures of HSV-2, EBV, and PrV gH, targeted point mutations have been also introduced into the gH of VZV (62
). In these studies, most mutations in the α-helical central part of gH (domains II and III) abolished protein function, whereas mutations introduced into the N-terminal, gL-binding part affected skin pathogenesis in SCID mice only. Disruption of the disulfide bonds flanking the flap in domain IV (in VZV gH designated domain III) by Cys-to-Ala mutations also severely affected fusogenic activity in vitro
. Mutation of the first cysteine in the CXXC motif was lethal for VZV replication, but the other cysteine mutants were still replication competent, although they exhibited reduced virulence. These results parallel our findings demonstrating that disulfide bonds 3 and 4 are important for gH function. Interestingly, the defects associated with mutation of either cysteine residue in the CXXC motif of disulfide bond 4 of VZV gH were rescued by a compensatory mutation, S694F, located at an edge of the hydrophobic patch underneath the flap, basically expanding the hydrophobic patch (62
), which we did not observe in our PrV studies. Based on the effects of mutations of the flanking disulfide bonds and the compensatory S694F mutation, which was proposed to rigidify the domain by providing a bulkier hydrophobic side chain, the VZV gH data had been interpreted as suggesting that domain IV rigidity is a major requirement for gH function. Our finding that disulfides SS3 and SS4 of PrV gH domain IV are functionally important could be explained by their involvement in the maintenance of the domain IV fold, consistent with the observation in VZV. Our observation that newly introduced cysteine pairs predicted to form bridges between the flap and the hydrophobic patch impair gH function supports the interpretation that displacement of the flap is involved in gH function. The extended polypeptide chain forming the flap, located at the surface of domain IV, could thus move independently of the rest of domain IV, thereby not compromising its rigidity. However, this remains to be tested experimentally.
In summary, our results define a functionally important region in domain IV of PrV gH and demonstrate that structural conservation of herpesvirus gH homologs correlates with functional conservation of individual structural elements within domain IV. In particular, the results confirm the relevance of the hydrophobic region in the membrane proximal part of the gH ectodomain for fusion, which had been indicated by earlier studies on HSV-1 (15
). They also demonstrate the importance of hydrophilic residues in the flap region overlaying the hydrophobic patch. Both regions contain residues which are important for in vitro
fusion-promoting activity of gH and, although nonessential, also for gH function during viral replication.