The Highly Conserved Proline at Position 438 in Pseudorabies Virus gH Is Important for Regulation of Membrane Fusion

ABSTRACT

Membrane fusion in herpesviruses requires viral glycoproteins (g) gB and gH/gL. While gB is considered the actual fusion protein but is nonfusogenic per se, the function of gH/gL remains enigmatic. Crystal structures for different gH homologs are strikingly similar despite only moderate amino acid sequence conservation. A highly conserved sequence motif comprises the residues serine-proline-cysteine corresponding to positions 437 to 439 in pseudorabies virus (PrV) gH. The PrV-gH structure shows that proline⁴³⁸ induces bending at the end of an alpha-helix, thereby placing cysteine⁴⁰⁴ and cysteine⁴³⁹ in juxtaposition to allow formation of a strictly conserved disulfide bond. However, PrV vaccine strain Bartha unexpectedly carries a serine at this conserved position. To test the influence of this substitution, we constructed different gH chimeras carrying proline or serine at position 438 in gH derived from either PrV strain Kaplan or strain Bartha. Mutants expressing gH with serine⁴³⁸ showed reduced fusion activity in transient-fusion assays and during infection, with delayed penetration kinetics and a small-plaque phenotype which indicates that proline⁴³⁸ is important for efficient fusion. A more drastic effect was observed when disulfide bond formation was completely blocked by mutation of cysteine⁴⁰⁴ to serine. Although PrV expressing gHC⁴⁰⁴S was viable, plaque size and penetration kinetics were drastically reduced. Alteration of serine⁴³⁸ to proline in gH of strain Bartha enhanced cell-to-cell spread and penetration kinetics, but restoration of full activity required additional alteration of aspartic acid to valine at position 59.

IMPORTANCE The role of the gH/gL complex in herpesvirus membrane fusion is still unclear. Structural studies predicted a critical role for proline⁴³⁸ in PrV gH to allow the formation of a conserved disulfide bond and correct protein folding. Functional analyses within this study corroborated these structural predictions: mutation of this residue resulted in a drastic impairment of membrane fusion kinetics not only in vitro in transient transfection-fusion assays but also during virus infection. Elimination of formation of the disulfide bond yielded the same phenotype in transient assays but had a more drastic effect on virus replication. Thus, our studies add important information to structure-function analyses of herpesvirus gH.

INTRODUCTION

In contrast to many other known viral fusion systems which rely on one or two proteins, herpesvirus membrane fusion requires the concerted action of at least three viral glycoproteins (g). In the prototypical alphaherpesvirus, herpes simplex virus 1 (HSV-1), binding of gD to one of its specific host cell receptors induces a conformational change that, in turn, triggers the actual fusion process by signaling to the gH/gL heterodimer and/or gB (1–4). gB and gH/gL are essential for membrane fusion during entry as well as for direct viral cell-to-cell spread and are therefore considered the core fusion machinery of herpesviruses. While gB shows typical features of class III fusion proteins, it is not sufficient by itself to induce efficient membrane fusion but requires the presence of gH/gL. The function of this complex, whose structure does not resemble that of any known fusion protein (5–7), still remains elusive (reviewed in reference 8).

Herpesvirus gH molecules are type I transmembrane proteins with several N-glycosylation consensus sequences in the ectodomain and a short cytoplasmic tail following the C-terminal membrane anchor. Amino acid sequences are only moderately conserved across the Herpesviridae and are largely limited to several regions within the C-terminal half of the ectodomain (5, 9). One of the conserved sequence motifs corresponds to amino acids ⁴³⁷-SPCAVSLRRDL-⁴⁴⁷ in gH of the alphaherpesvirus pseudorabies virus (PrV), encompassing a highly conserved serine-proline-cysteine motif, and a second region close to the predicted transmembrane region, ⁶²⁰-LFPNGTV-⁶²⁶, comprising a conserved N-glycosylation site (9). The preservation of several other cysteine residues throughout the ectodomain already implied similarity of the tertiary structures (9), which was confirmed by comparison of the crystal structures of the herpes simplex virus 2 (HSV-2) and Epstein-Barr virus (EBV) gH/gL complexes as well as the PrV gH core domain (5–7).

All known gH homologs form a heterodimeric complex with gL, a small glycoprotein (156 amino acids [aa] in PrV) which depends on gH interaction for membrane association and virion incorporation (10–13). While in many herpesviruses gH also requires gL for correct processing and transport, PrV, bovine herpesvirus 4 (BoHV-4), and murine herpesvirus 4 (MuHV-4) gH is also incorporated into virions in the absence of gL (14–16). However, in contrast to BoHV-4 and MuHV-4, where gL is not strictly required for fusion, PrV requires gL for entry of free virions (15), as do all other herpesviruses studied so far. However, direct viral cell-to-cell transmission, which relies on a similar, but not identical, set of proteins, occurs in PrV even in the absence of gL, although at drastically decreased efficiency (15). This minimal capability for direct viral cell-to-cell spread has been used for serially passaging PrV-ΔgL in tissue culture cells, ultimately resulting in a rescuant, PrV-ΔgLPass, which efficiently replicated without gL (17). In this revertant, a hybrid protein is expressed from a fused gene consisting of the receptor binding domain of gD joined in-frame to an N-terminally truncated gH core fragment that lacks the putative gL binding domain (Fig. 1). This gDH hybrid protein is sufficient to induce membrane fusion in combination with gB and complements the defect of mutants simultaneously lacking gD, gH, and gL (17, 18), indicating that membrane fusion can be accomplished without gL and with a minimal set of two herpesviral proteins.

FIG 1.

Schematic diagram of gH. The gH open reading frames are shown as rectangles. The core fragment of gH present in the gDH hybrid protein, which had been used for crystallization, is presented above. The domains are labeled with roman numbers with boundaries shown. The black diamond indicates the location of the SPC motif. The shaded boxes left and right indicate residues that were not resolved in the structure. The predicted signal sequence (SS; aa 1 to 30) is highlighted in black; the transmembrane domain (TM; aa 647 to 663) is indicated in gray. The location of the PmlI restriction site used for cloning is shown. Amino acid differences between PrV-Ba gH and PrV-Ka gH are given in one-letter code with the corresponding position.

From the crystal structures of the ectodomains of HSV-2 and EBV gH/gL (6, 7), as well as the core fragment of PrV gH, which was derived from the gDH hybrid protein (5), four distinct domains could be defined. While domain I, which is involved in binding gL, is missing in the PrV gH structure (Fig. 1), domains II to IV are largely superimposable among the three proteins. Domain II contains two conserved structures, an antiparallel beta-sheet designated the “fence” and an elongated antiparallel 3-helix bundle which shows structural similarities to a domain of cellular syntaxins and has accordingly been designated the “syntaxin-like bundle” (SLB) (5) (Fig. 2). Domain III consists mainly of alpha-helices and harbors one of the conserved sequence motifs, including residues serine⁴³⁷, proline⁴³⁸, and cysteine⁴³⁹ (numbering for PrV gH), with proline and cysteine invariably present in all herpesvirus gH homologs. Proline⁴³⁸ induces bending at the end of an alpha-helix, so that cysteine⁴³⁹ is brought into proximity to cysteine⁴⁰⁴ and is able to form a strictly conserved disulfide bond with cysteine⁴⁰⁴ (Fig. 2) (5). The membrane-proximal domain IV, which shows the highest conservation, is composed of two four-stranded beta-sheets connected by the so-called “flap” region. This hydrophilic flap, which is followed by a conserved N-glycosylation site, covers a hydrophobic patch, which was shown to play an important role during fusion. It is speculated that movement of the flap, which masks the underlying hydrophobic patch, promotes membrane fusion (5, 19).

FIG 2.

PrV gH structural features. (A) Crystal structure of the PrV gH core domain (PDB code 2XQY) with the fence, SLB, and flap region colored in light green, blue, and red, respectively. The domains are labeled with roman numbers. The N and C termini of the molecule are designated with the letters N and C. Alpha-helix 11 (α11), located in domain III and carrying the SPC motif, is highlighted in pink. (B) Zoomed-in view of the SPC motif as seen from the back of the molecule shown in panel A. Replacing Pro⁴³⁸ with any other amino acid would result in an unbent alpha-helix 11, which would project as indicated by the dotted lines. Bending of the helix is clearly important for residues Cys⁴³⁹ and Cys⁴⁰⁴ to become juxtaposed for formation of the disulfide bond.

Recently, the complete genome sequence of PrV strain Bartha (PrV-Ba), an attenuated strain widely used as a vaccine against PrV infection in pigs, was deciphered and compared to wild-type strains PrV-Kaplan (PrV-Ka) and PrV-Becker (20). While the attenuating mutations in PrV-Ba were mapped earlier by marker rescue experiments (21–24) and were shown to be a result of a large deletion within the unique short region comprising genes encoding gI, gE, pUS9, and pUS2, as well as subtle amino acid changes in pUL21 and gC, comparison of the deduced amino acid sequences revealed a total of 46 open reading frames with nonsilent mutations (20). Among these mutations, an alteration in gH within the highly conserved sequence motif ⁴³⁷serine-proline-cysteine⁴³⁹ was identified. Interestingly, PrV-Ba gH contains a serine instead of the helix-bending proline (20).

Since proline⁴³⁸ is strictly conserved throughout other herpesvirus gH homologs, as is the disulfide bridge between cysteine⁴⁰⁴ and cysteine⁴³⁹ (5, 9), we were interested in analyzing the effect of this alteration on gH function. To this end, we generated PrV gH variants expressing either serine or proline at position 438 as well as a serine mutant at position 404 and analyzed them in transient transfection-fusion assays and, after rescue of the respective recombinant viruses in PrV-Ka, in one-step growth, penetration kinetics, and cell-to-cell transmission assays.

MATERIALS AND METHODS

Cells and viruses.

Rabbit kidney cells (RK13) and RK13-gHgL cells (25) were grown in Dulbecco‘s modified Eagle's minimum essential medium (MEM) supplemented with 10% fetal calf serum at 37°C in the presence of 5% CO₂. Recombinants used in this study were derived from pPrV-ΔgGG, comprising the PrV-Ka genomic DNA cloned as a bacterial artificial chromosome (BAC) with the mini-F plasmid sequence and a green fluorescent protein (GFP) expression cassette at the nonessential gG gene locus (19). In pPrV-ΔgHABF (for simplicity, here designated pPrV-ΔgH), gH codons from 51 to the end were deleted (S. Böhm, E. Eckroth, M. Backovic, B. G. Klupp, F. A. Rey, T. C. Mettenleiter, and W. Fuchs, unpublished data). pPrV-gH^Ka was generated by reinsertion of the wild-type Kaplan gH (gH^Ka) together with a downstream-located kanamycin resistance gene (kan) for selection in Escherichia coli and was used here as the wild-type control. For comparison, a BAC clone (here designated pPrV-Ba) derived from PrV-Bartha (pPrV-BaΔgGG [26]), generated after deletion and reinsertion of the Bartha gH gene (gH^Ba), including the downstream kan, was used. The gH deletion mutants were grown in RK13-gHgL cells, while all the other recombinants were propagated on RK13 cells.

Expression plasmids and gH mutagenesis.

Plasmids pcDNA-gB, pcDNA-gH, and pcDNA-gL express the full-length gB, gH, and gL of PrV-Ka, respectively, under the control of the human cytomegalovirus immediate early promoter/enhancer complex (18). pcDNA-gD expressing PrV-Ka gD was generated after PCR amplification using primers gD-fw and gD-rv (Table 1) and PrV-Ka genomic DNA as a template. The 1.2-kb PCR product was cloned into pcDNA3 (Invitrogen) after cleavage with EcoRI using the corresponding restriction sites added by the primers. pcDNA-gH^Ba was generated using primers DH-3 and DHrev1 (Table 1) on genomic PrV-Ba DNA (27). The 2.1-kb PCR product was cloned into pcDNA3 using the EcoRI and XbaI restriction enzyme sites provided by the primers. Correct amplification and insertion were verified by sequencing and indirect immunofluorescence assay after transfection (data not shown). For sequencing, a BigDye Terminator cycle sequencing kit (Applied Biosystems) was used, and products were run on an ABI 3130 genetic analyzer (Applied Biosystems).

TABLE 1.

Primers used in this study

Primer	Sequencea (5′–3′)	Locationb (nt)
gD-fw	CAC AGA ATT CAC CTG CCA GCG CCA TGC	121131–121148
gD-rv	CAC AGA ATT CCA TCG ACG CCG GTA CTG C	122370–122353
DH-3	AGA ATT CAA AGT TTG CCG TGC CCG TC	60748–60766
DHrev1	CAT CTA GAC ACG CGC ACG CAG AGA GT	62873–62856
P438S-fw	CGA CGT CCT CTC GTC GTG CGC CGT CTC	62068–62094
P438S-rv	GAG ACG GCG CAC GAC GAG AGG ACG TCG	62094–62068
S438P-fw	CGA CGT CCT CTC GCC GTG CGC CGT CTC	62068–62094
S438P-rv	GAG ACG GCG CAC GGC GAG AGG ACG TCG	62094–62068
C404S-fw	CGC ACC ACG GCC ATG TCC ACC GCG GAG CGC GCC	61964–61996
C404S-rv	GGC GCG CTC CGC GGT GGA CAT GGC CGT GGT GCG	61996–61964

^aRestriction sites are in italics; nucleotide substitutions are underlined.

^bThe locations for the primers correspond to nucleotide postitions in the sequence with GenBank accession number JQ809328 (29).

To introduce mutations into the gH sequence, a QuikChange II XL site-directed mutagenesis kit (Agilent Technologies) was used as described recently (28). Specific primers (Table 1) derived from the genomic sequence of PrV-Ka (GenBank accession no. JQ809328 [29]) or PrV-Ba (GenBank accession no. JF797217 [20]) were applied to generate pcDNA-gH^KaP⁴³⁸S, pcDNA-gH^KaC⁴⁰⁴S, pcDNA-gH^BaS⁴³⁸P, and pcDNA-gH^BaD⁵⁹V/S⁴³⁸P (Fig. 1). pcDNA-gH^Ka or pcDNA-gH^Ba was used as a template, and correct mutagenesis was verified by sequencing. To generate pcDNA-gH^KaV⁵⁹D and pcDNA-gH^BaD⁵⁹V, we cleaved pcDNA-gH^Ka and pcDNA-gH^Ba with PmlI and XbaI, and the resulting 1.4-kb fragments were ligated into the respective other plasmid. All plasmids were sequenced and tested for gH expression by indirect immunofluorescence and Western blotting (data not shown).

Fusion assays.

Fusion activity was measured after transient expression in RK13 cells. For each assay, approximately 1.8 × 10⁵ RK13 cells per well in 24-well culture dishes were seeded the day before transfection. Cells were then transfected with 125 ng of each plasmid expressing one of the gH variants, full-length gB, gD, and gL, as well as pEGFP-N1 (Clontech) as a marker in 50 μl of serum-free MEM using 1 μl polyethylenimine (PEI). After incubation for 15 min at room temperature, the mixture was added to the cells. pcDNA3 was used as a negative control, substituting for the gH expression plasmid in the transfection assays. Forty-eight hours posttransfection, cells were washed with phosphate-buffered saline (PBS), fixed with 3% paraformaldehyde for 20 min, and washed twice with PBS. To quantify the fusion activity, we evaluated 10 randomly selected fields of view in each well by counting the number of green-fluorescing syncytia and measuring the polykaryocyte area using a Nikon Eclipse Ti-S fluorescence microscope and Nikon NIS-Elements imaging software (Nikon). Syncytia were defined as cells with three or more nuclei. Total fusion activity was calculated by multiplication of the number of syncytia by the mean value of the syncytium area in each assay. The experiment was repeated three times, and the average values (values for assays with pcDNA-gH^Ka set as 100%) of the results with the corresponding standard deviations are given.

Generation of gH recombinants.

The transfer plasmids used for reinsertion of authentic, heterologous, or mutated gH genes into PrV-Ka were based on the expression plasmid pcDNA-gHKDE (Böhm et al., unpublished), comprising the full-length gH^Ka followed by kan and a DrdI restriction site. A 1.7-kbp KpnI-BamHI fragment covering most of the gH gene (corresponding to codons 88 to 662) was replaced with the corresponding fragments of pcDNA-gH^Ba, pcDNA-gH^KaP⁴³⁸S, and pcDNA-gH^KaC⁴⁰⁴S to generate pPrV-gH^BaD⁵⁹V, pPrV-gH^KaP⁴³⁸S, and pPrV-gH^KaC⁴⁰⁴S. The transfer plasmid for pPrV-gH^KaV⁵⁹D was generated after substitution of a 0.5-kbp SanDI-BsrGI fragment (comprising codons 42 to 216) in pcDNA-gHKDE by the corresponding fragment from pcDNA-gH^Ba. The triplet encoding serine 438 in gH^Ba was mutated to a proline codon by site-directed mutagenesis with primers S⁴³⁸P-fw and S⁴³⁸P-rv (Table 1), resulting in the transfer plasmid for pPrV-gH^BaS⁴³⁸P. By recloning of the 1.7-kb KpnI/BamHI fragment from this plasmid into pcDNA-gHKDE, the transfer plasmid for pPrV-gH^BaD⁵⁹V/S⁴³⁸P was obtained. All plasmids were digested with DrdI, and the 3.5-kbp fragments containing the complete gH gene followed by kan were isolated and subsequently used for Red-mediated mutagenesis of pPrV-ΔgH (Böhm et al., unpublished). PrV recombinants were characterized by restriction analyses and Southern blot hybridization of genomic DNA. Correct mutagenesis was verified by PCR amplification and sequencing of the complete gH open reading frame (data not shown).

Western blotting.

For immunoblotting, RK13 cells were infected at a multiplicity of infection (MOI) of 3. Twenty-four hours after infection, cells and supernatants were harvested, pelleted by centrifugation, and washed twice with PBS. The pellet was lysed in sodium dodecyl sulfate (SDS) sample buffer, and the proteins were separated in SDS–10% polyacrylamide gels. After blotting onto nitrocellulose and blocking with 5% skimmed milk, membranes were incubated with rabbit sera specific for gH (17), capsid triplex component pUL38 (unpublished data), or a monoclonal mouse anti-gE antibody (17). Bound antibodies were detected by incubation with peroxidase-conjugated goat anti-rabbit or goat anti-mouse secondary antibodies (Dianova). Blots were analyzed using an imager (VersaDoc; Bio-Rad) after incubation with SuperSignal West Pico chemiluminescent substrate (Thermo Scientific).

One-step growth kinetics.

For analysis of replication properties, RK13 cells were infected at an MOI of 3 and incubated on ice for 1 h. Then the inoculum was replaced by prewarmed medium and the cells were further incubated at 37°C. After 1 h of incubation at 37°C, nonpenetrated extracellular virus was inactivated by low-pH treatment (30). The cells were harvested at different times thereafter by scraping into the supernatant and were stored at −80°C. After thawing, the supernatant was titrated on RK13 cells. This assay was repeated two times, and mean values with the corresponding standard deviations were calculated.

Plaque assays.

To determine the plaque size, RK13 cells grown in 12-well culture dishes were infected with the different recombinants with approximately 100 PFU under plaque assay conditions and fixed with 3% paraformaldehyde 2 days postinfection (p.i.). The fluorescent area of 30 plaques was measured microscopically with a Nikon Eclipse Ti-S fluorescence microscope using Nikon NIS-Elements imaging software and was compared to the area measured for pPrV-gH^Ka, which was set as 100%. Mean values with standard deviations from three independent experiments are shown.

Penetration kinetics.

The kinetics of penetration of the different viruses was analyzed in RK13 cells. Cells grown in 12-well culture dishes were infected on ice with approximately 150 PFU of the different recombinants. After 1 h, the inoculum was replaced by prewarmed medium and the cells were incubated at 37°C; after 0, 5, 10, 20, and 40 min, the remaining extracellular virus was inactivated by low-pH treatment in one well, while a parallel well was used as a 100% penetration control and washed with PBS only. Two days p.i., cells were fixed and stained with crystal violet, and plaques were counted. To calculate the penetration rate, we compared the number of PFU surviving the low-pH inactivation to the number of plaques formed in the control. Mean values with standard deviations from three independent assays are given.

RESULTS

Sequencing of PrV-Bartha gH.

Since the published genome sequences of PrV-Ka generated by Szpara et al. (20) and our laboratory (29) differ at several positions in the coding regions, we first verified whether the Bartha strain used by us (Bartha K61) (27) comprises the same mutations in the gH open reading frame as those previously reported (20). To this end, we amplified the complete gH gene of PrV-Ba genomic DNA and sequenced it after cloning. Comparison of the deduced amino acid sequences revealed changes in PrV-Ba gH identical to those previously reported (20), namely, V⁵⁹D, P⁴³⁸S, EE^481,482G, and I⁵³⁹L (Fig. 1), thus verifying the proline-to-serine mutation at position 438 within the highly conserved sequence motif (5).

Serine at position 438 in PrV gH reduces fusion activity in transient transfection-fusion assays.

To analyze the influence of the mutation in gH in our transient transfection-fusion assay, RK13 cells were cotransfected with plasmids expressing full-length gB, gD, gL of PrV-Ka, and the different gH plasmids pcDNA-gH^Ka, pcDNA-gH^Ba, pcDNA-gH^KaP⁴³⁸S, and pcDNA-gH^BaS⁴³⁸P. pEGFP-N1 was included for easy visualization of transfected cells. Two days after transfection, syncytium formation was analyzed under a fluorescence microscope. Figure 3 shows the mean values for three independent assays with the corresponding standard deviations. Cotransfection of the plasmids expressing the native PrV-Ka glycoproteins resulted in approximately 60 syncytia in the 10 randomly selected fields of view, which was set as 100%. Substitution of proline⁴³⁸ with serine (gH^KaP⁴³⁸S) led to a ca. 90% reduction of fusion activity resulting in activity similar to that observed in assays with gH^Ba and just slightly higher than in assays without any gH expression plasmid (pcDNA3), showing that proline at position 438 plays an important role for gH function during membrane fusion in these transient assays.

FIG 3.

Transient transfection-fusion assays. To test for fusion activity, RK13 cells were cotransfected with plasmids coding for gH^Ka, gH^Ba, or the respective gH mutants and plasmids encoding the full-length forms of gB, gD, and gL, as well as enhanced green fluorescent protein (EGFP) as a marker. The empty vector pcDNA3 served as a negative control, substituting for the gH expression plasmid. Total fusion activity (number of syncytia × syncytium area) in assays containing gH^Ka was set as 100%, and the corresponding standard deviations for three independent assays were calculated.

Bending of alpha-helix 11 in gH induced by proline⁴³⁸ was suggested to be important for formation of a highly conserved disulfide bond connecting cysteine⁴⁰⁴ and cysteine⁴³⁹ in PrV gH (Fig. 2) (5). To substantiate this assumption, we generated a Ka-gH expression plasmid with cysteine⁴⁰⁴ mutated to serine (gH^KaC⁴⁰⁴S). In transient-fusion assays, gH^KaC⁴⁰⁴S yielded reduced fusion activity similar to that of gH^KaP⁴³⁸S (Fig. 3), suggesting that proline⁴³⁸ is indeed involved in mediating bridging between cysteine⁴⁰⁴ and cysteine⁴³⁹.

Restoring proline⁴³⁸ in gH^Ba (gH^BaS⁴³⁸P) resulted in slightly enhanced fusion activity, which, however, did not reach the levels of fusion obtained with gH^Ka (Fig. 3). Thus, to test which other mutation(s) present in gH^Ba may influence its fusion activity, we altered other amino acids which differed between PrV-gH^Ba and PrV-gH^Ka (Fig. 1). The V⁵⁹D substitution in gH^Ka (gH^KaV⁵⁹D) resulted in a drastic drop in activity, while restoring V⁵⁹ in gH^Ba (gH^BaD⁵⁹V) showed only a slight increase in syncytium formation. However, when both positions in gH^Ba were altered to the wild-type PrV-Ka residues (gH^BaD⁵⁹V/S⁴³⁸P), fusion activity reached a level similar to that of native gH^Ka, indicating that besides proline⁴³⁸, valine⁵⁹ plays an important role in the fusion process. Apparently, the other alterations (EE^481,482G, I⁵³⁹L), which remained unchanged, had no significant effect on fusion (data not shown).

The glycosylation pattern is altered in gH mutants carrying serine⁴³⁸ or serine⁴⁰⁴.

To test the influence of the amino acid substitutions during infection, we generated pPrV-Ka recombinants expressing the different gH proteins by rescuing pPrV-ΔgH. Expression of gH was tested by immunoblotting using monospecific anti-gH serum. As shown in Fig. 4, at 24 h after infection, in pPrV-gH^Ka-infected cells gH is detectable as a single protein species, indicating efficient processing by transfer through the Golgi apparatus, while in pPrV-Ba as well as in pPrV-gH^Ba, the PrV-Ka-derived mutant expressing gH^Ba, two gH species most likely representing immature, partially glycosylated and mature, completely glycosylated forms were detected. The same pattern was observed for all PrV-Ka mutants expressing gH with serine⁴³⁸ (gH^KaP⁴³⁸S, gH^BaD⁵⁹V), while mutants with proline⁴³⁸ (gH^KaV⁵⁹D, gH^BaS⁴³⁸P, and gH^BaD⁵⁹V/S⁴³⁸P) showed the wild-type pattern, indicating that serine⁴³⁸ influences protein processing. In pPrV-gH^KaC⁴⁰⁴S-infected cells, only immature gH could be demonstrated, and no gH-specific signal was detected in pPrV-ΔgH-infected cells. It is notable that the mutations did not introduce an additional N-linked glycosylation site. The capsid protein pUL38, which served as a loading control, was present in similar amounts in all infected cell lysates, while gE was absent from the Bartha-derived recombinant pPrV-Ba.

FIG 4.

Western blotting. RK13 cells were infected with the indicated viruses for 24 h, harvested, and lysed. Proteins were separated on SDS–10% polyacrylamide gels, blotted, and incubated with monospecific antisera or monoclonal antibodies as indicated on the right. The locations and molecular masses (in kDa) of marker proteins are shown on the left.

The proline⁴³⁸serine substitution in PrV gH has no influence on final viral titers.

To test for viral replication, RK13 cells were infected at an MOI of 3 with the different viruses. Cells and supernatants were harvested 0, 4, 8, 12, 24, and 32 h after infection, and titers were determined on RK13 cells. The mean values from two independent experiments are shown in Fig. 5. While the gH deletion mutant pPrV-ΔgH does not productively replicate on RK13 cells, titers of viruses expressing the different gH variants were comparable, demonstrating that the proline⁴³⁸serine substitution has no effect on production of infectious progeny. pPrV-gH^KaC⁴⁰⁴S proved to be replication competent, but titers were approximately 10-fold lower, indicating that a complete block of formation of the disulfide bond has a greater effect than impairment by mutation of the neighboring proline.

FIG 5.

One-step growth kinetics. RK13 cells were infected with parental or mutant viruses at an MOI of 3. At the indicated times, cells and supernatants were harvested, and virus progeny titers on RK13 cells were determined. The mean values for two independent assays and the corresponding standard deviations are shown.

The proline⁴³⁸serine substitution in PrV gH affects direct viral cell-to-cell transmission.

Since gH is essential for direct viral cell-to-cell spread, plaque areas were measured 2 days after infection. PrV-Ba was already known to form small plaques, primarily due to a large deletion in the unique short region comprising, among others, the genes encoding gI and gE (22), which is similar for the BAC-derived mutant pPrV-Ba (Fig. 6). Expression of gH^Ba by PrV-Ka also resulted in a small-plaque phenotype. This was also found for all mutants expressing gH variants with serine⁴³⁸. In contrast, recombinants with proline⁴³⁸ formed plaques which were comparable to those of pPrV-gH^Ka, indicating that serine⁴³⁸ impairs, while proline at this position promotes, direct viral cell-to-cell spread. In contrast to the transient-fusion assays where valine at position 59 enhanced fusogenicity, no further increase could be observed in pPrV-gH^BaD⁵⁹V/S⁴³⁸P. pPrV-gH^KaC⁴⁰⁴S produced only tiny foci of infected cells, congruent with reduced final viral titers after one-step replication. Infection of RK13 cells with pPrV-ΔgH resulted in only single fluorescent cells (Fig. 6B).

FIG 6.

Plaque formation of virus mutants. (A) Cell-to-cell spread was tested after infection of RK13 cells with the different viruses under plaque assay conditions. The plaque areas of 30 plaques for each virus were measured microscopically. The plaque size of PrV-Ka was set as 100%. Mean values and the corresponding standard deviations from three independent assays were plotted. (B) RK13 cells were infected with the indicated viruses. After 2 days, cells were fixed and fluorescent images were acquired with a fluorescence microscope (Nikon Eclipse Ti-S). In the negative control (pPrV-ΔgH), only single infected cells were observed. Bar = 400 μm.

The gH proline⁴³⁸serine substitution delays penetration kinetics.

Penetration kinetics was analyzed to assess the influence of the proline⁴³⁸serine substitution on the rate of fusion during entry. While pPrV-gH^Ka entered cells extremely fast, with 80% of input virus protected from low-pH treatment after approximately 5 min, pPrV-gH^Ba reached this level only after approximately 20 min. The penetration kinetics of pPrV-gH^KaP⁴³⁸S was as slow as that of pPrV-gH^Ba, indicating a significant influence of the proline⁴³⁸serine substitution on fusion kinetics during virus entry (Fig. 7). In contrast, the introduction of proline⁴³⁸ into gH^Ba restored penetration kinetics to rates comparable to those for pPrV-gH^Ka (Fig. 7). Penetration of pPrV-gH^KaC⁴⁰⁴S was the slowest, but the virus was still able to enter cells and to initiate productive infection.

FIG 7.

Penetration kinetics. To analyze the rate of entry, RK13 cells were infected with approximately 150 PFU per well of the different viruses on ice. At 1 h p.i., the inoculum was replaced with prewarmed medium and the cells were incubated for the time periods indicated before inactivation of the remaining extracellular virus by low-pH treatment. Two days after infection, cells were fixed and stained with crystal violet. The rate of penetration was calculated as the number of PFU in low-pH-inactivated wells compared to that in control wells, which were treated with PBS only and set as 100%. Mean values with standard deviations from three independent assays are shown.

DISCUSSION

The gH/gL complex is an essential component of the herpesviral fusion machinery, but its role is not well understood. Although an active role in membrane fusion has been suggested for gH (31–33), analysis of the available crystal structures showed no signatures of known fusion proteins (5–7). More recent data indicate that the gH/gL complex plays a regulatory role, mediating the timely and efficient transition of gB into its fusion-active state (6). However, how this is accomplished remains unknown.

Amino acid sequence conservation is only moderate between the different herpesvirus gH molecules. However, the “SPC” motif located at positions 437 to 439 in PrV gH, with proline and cysteine invariably present in gH homologs of all herpesvirus subfamilies, was thought to be indispensable (5). It was, therefore, surprising that PrV strain Bartha, an attenuated vaccine strain, carries a serine at position 438 (20). To test the influence of this unexpected alteration on gH function, we mutated proline⁴³⁸ to serine in PrV-Ka gH and changed serine⁴³⁸ to proline in PrV-Ba gH.

In transient transfection-fusion assays, a drastic impairment in syncytium formation was evident when RK13 cells were cotransfected with expression plasmids for full-length gB, gD, and gL and plasmids expressing gH with the serine⁴³⁸ substitution. This indicates that proline⁴³⁸ plays a major role in supporting efficient cell-cell fusion. Since it was speculated that proline⁴³⁸ is necessary for the bending of alpha-helix 11 as a prerequisite for cysteine⁴³⁹ to come into the vicinity of and form a disulfide bridge with the conserved cysteine⁴⁰⁴ (Fig. 2) (5), we tested whether the replacement of cysteine⁴⁰⁴ with serine resulted in the same effect. Fusion assays with gH^KaC⁴⁰⁴S also demonstrated a drastic drop in syncytium formation, similar to that for gH^KaP⁴³⁸S, emphasizing the importance of these residues for protein function. While mutational analyses of the conserved proline have not been performed for any herpesvirus gH so far, mutation of the corresponding cysteines involved in disulfide bond formation in herpes simplex virus (HSV) (C5/C6 = cysteine⁵⁵⁴/cysteine⁵⁸⁹) and varicella-zoster virus (VZV) (cysteine⁵⁴⁰/cysteine⁵⁷⁵) rendered the proteins nonfunctional, pointing to an essential role for the disulfide bond (34, 35).

The proline⁴³⁸serine substitution also affected gH maturation, as demonstrated by the increased presence of immature gH (Fig. 4). It can be speculated that the presence of serine instead of proline in alpha-helix 11 (5) influences the folding of gH and that incorrectly folded gH molecules are retained in the endoplasmic reticulum (ER) and are not efficiently transported through the Golgi apparatus. Alternatively, improper folding may impair access to glycosylation sites during the transit of gH through the secretory pathway. Nevertheless, PrV-Ba gH, which contains this substitution, is functionally active to support viral replication, as is gH^KaP⁴³⁸S, at least in cell culture. In cells expressing gH^KaC⁴⁰⁴S, only immature gH was detected, in line with the importance of this disulfide bond for gH function in HSV and VZV.

Whether the proline⁴³⁸serine substitution affects virulence in vivo remains to be determined. However, it is important to note that full restoration of the virulence of PrV-Ba was achieved after repair of the deletion in the unique short region and point mutations in pUL21 (21, 23, 36). Thus, proline⁴³⁸ in gH is apparently not essential for PrV virulence.

Repair of proline⁴³⁸ in gH^Ba enhanced syncytium formation, but not to the same levels as gH^Ka, indicating that the other alterations present in gH^Ba also have a negative effect on membrane fusion. Mutation of aspartate⁵⁹ to valine only slightly increased the fusion activity of gH^Ba, while introduction of aspartate⁵⁹ into gH^Ka instead of valine drastically reduced fusogenicity, indicating that this amino acid position is also important for gH function. Since domain I, which also contains the gL binding site, was not part of the gH construct that was crystallized (Fig. 2), it is unclear whether this mutation might affect domain III directly or whether it influences fusogenicity by affecting gL binding. Unfortunately, the region encompassing V59 cannot be aligned with the structure of the other gH proteins. However, gH^Ba is functional in vitro and in vivo, indicating that this alteration does not abolish gL binding and gH/gL function. Restoration of both D⁵⁹ and S⁴³⁸ in gH^Ba (gH^BaD⁵⁹V/S⁴³⁸P) resulted in syncytium formation comparable to that with native gH^Ka, demonstrating that the other differences between gH^Ba and gH^Ka do not play a role in membrane fusion.

To analyze the influence of the proline/serine variation during virus infection, we generated PrV-Ka recombinants expressing the different gH molecules. Despite the obvious impairment of syncytium formation in the transfection-fusion assays, the PrV recombinants expressing gH with the serine substitution at position 438 replicated to titers comparable to those of pPrV-gH^Ka and pPrV-gH^Ba, indicating that the mutation has no effect on virus production after high-MOI infection, which is in line with the notion that once inside the cell, gH is not required for continuing viral replication (25). In contrast, pPrV-gH^KaC⁴⁰⁴S replicated to ca. 10-fold-reduced viral titers. Although this represents a significant defect, it clearly shows that, in contrast to HSV and VZV, PrV does not require the disulfide bond involving C⁴⁰⁴ for gH function. Our data suggest that the P⁴³⁸S mutation may impair protein folding such that only a fraction of the molecules folds correctly by a spontaneous tilt of alpha-helix 11, is able to leave the ER, and is properly processed. In contrast, another fraction is retained in the ER and remains in an immature state, as indicated by our protein analyses. Since correctly folded and properly processed gH will be incorporated into progeny virions, infectious particles will form but may contain less gH than the wild type. Maturation of gH is even more impaired by the complete inactivation of the disulfide bond in the C⁴⁰⁴S mutant, which correlates with the more severe phenotypic defects. However, even in the absence of the disulfide bond, PrV gH retains a residual capacity to function during entry.

Direct viral cell-to-cell spread as measured by plaque size was significantly reduced in RK13 cells infected with the serine⁴³⁸ variants, but plaque size could be restored to wild-type levels by serine⁴³⁸proline reversions. The strongest phenotype is shown by pPrV-gH^KaC⁴⁰⁴S, which produces only tiny foci of infected cells, yet there is noticeable viral spread compared to infection of only single cells with the gH deletion mutant.

The molecular mechanism of direct cell-to-cell transmission of herpesviruses has remained unknown. While it relies on the same set of proteins as fusion of free virions in many herpesviruses, PrV can spread via direct cell-cell transmission in the absence of the receptor binding gD (37, 38) and also, partly, in the absence of gL (15), while gB and gH are absolutely required. This demonstrates that both processes depend on similar but not identical mechanisms. In addition, other proteins such as the gE/gI complex facilitate the direct transit of infectivity from infected to neighboring uninfected cells. It was suggested that gE is targeted to cell junctions, thereby helping virus spread (39). In consequence, restoration of the deleted unique short region to PrV-Ba resulted in significantly larger plaque sizes (36). In our experiments, plaque size could also be increased by just altering serine⁴³⁸ in PrV-Ba gH expressed in a PrV-Ka background. We are currently analyzing whether in a PrV-Ba background in the presence of the deletion in the unique short region, eliminating expression of gE and gI, alteration of these amino acids in gH has a similar effect.

Penetration kinetics showed that the velocity of virus entry is delayed in viruses expressing serine⁴³⁸ instead of proline. While 80% of pPrV-gH^Ka entered cells within approximately 5 min after temperature shift, all mutants expressing gH with serine⁴³⁸ required longer to reach the same value, indicating that either the fusion at the plasma membrane is less efficient or uptake might take place by a different mechanism, e.g., endocytosis. Thus, the transient transfection-fusion assay, which shows a distinct impairment of gH activity by the proline⁴³⁸serine alteration, correlates with the results from penetration kinetics and plaque size determination, but not one-step replication after high MOI infection and final virus titers. Penetration of pPrV-gH^KaC⁴⁰⁴S is slowest, with only ca. 50% of the virus protected from acid inactivation after 40 min. Thus, inability to form this disulfide bond, which is essential in HSV and VZV, strongly impairs gH function in PrV but does not disable it completely. Therefore, despite the significant effect of the C⁴⁰⁴S mutation on PrV gH maturation as well as on viral cell-cell spread, penetration, and one-step replication, the mutated gH is still able to function, although at drastically lower efficiency. This unexpected result demonstrates that it is imperative to perform multiple assays with multiple viruses to unravel the functional consequences of specific alterations within this conserved protein.

In summary, alteration of the conserved proline at position 438 of PrV gH to serine influences the fusion activity of the protein in transient assays as well as cell-cell spread and penetration kinetics during infection. It also affects gH maturation. All these effects are mimicked, but increased, by abolition of the disulfide bond involving cysteine⁴⁰⁴. Thus, our data support the assumption that proline⁴³⁸ assists in, but is not essential for, formation of the cysteine⁴⁰⁴-cysteine⁴³⁹ disulfide bond by bending alpha-helix 11 and that formation of this disulfide bond is impaired but not eliminated by the amino acid exchange. Therefore, PrV-Ba retains replication competence despite the presence of the proline⁴³⁸serine mutation, which is a prime prerequisite for its use as a safe and efficacious PrV vaccine.