|Home | About | Journals | Submit | Contact Us | Français|
Antibodies readily neutralize acute, epidemic viruses, but are less effective against more indolent pathogens such as herpesviruses. Murid herpesvirus 4 (MuHV-4) provides an accessible model for tracking the fate of antibody-exposed gammaherpesvirus virions. Glycoprotein L (gL) plays a central role in MuHV-4 entry: it allows gH to bind heparan sulfate and regulates fusion-associated conformation changes in gH and gB. However, gL is non-essential: heparan sulfate binding can also occur via gp70, and the gB–gH complex alone seems to be sufficient for membrane fusion. Here, we investigated how gL affects the susceptibility of MuHV-4 to neutralization. Immune sera neutralized gL− virions more readily than gL+ virions, chiefly because heparan sulfate binding now depended on gp70 and was therefore easier to block. However, there were also post-binding effects. First, the downstream, gL-independent conformation of gH became a neutralization target; gL normally prevents this by holding gH in an antigenically distinct heterodimer until after endocytosis. Second, gL− virions were more vulnerable to gB-directed neutralization. This covered multiple epitopes and thus seemed to reflect a general opening up of the gH–gB entry complex, which gL again normally restricts to late endosomes. gL therefore limits MuHV-4 neutralization by providing redundancy in cell binding and by keeping key elements of the virion fusion machinery hidden until after endocytosis.
Most vaccines depend on eliciting neutralizing antibodies (Zinkernagel & Hengartner, 2006). Herpesvirus carriers remain infectious despite making antibody responses. Preventing herpesvirus infections by vaccination is therefore a difficult challenge. We are using murid herpesvirus 4 (MuHV-4) to understand gammaherpesvirus neutralization. MuHV-4 binds to cells via heparan sulfate, using either gp70, a product of ORF4 (Gillet et al., 2007a), or gH–gL (Gillet et al., 2008a). Immune sera can block cell binding (Gill et al., 2006), but they block membrane fusion poorly, allowing opsonized virions to infect macrophages and dendritic cells via IgG Fc receptors (Rosa et al., 2007). Bypassing cell-binding blocks in this way is not unique to MuHV-4 (Inada et al., 1985; Maidji et al., 2006).
How might herpesvirus membrane fusion be blocked better? Answering this means understanding how fusion works. Virus-specific glycoproteins, such as herpes simplex virus gD, can modulate fusion (Avitabile et al., 2007; Atanasiu et al., 2007) and some herpesviruses can express alternative fusion complexes by using different accessory glycoproteins (Borza & Hutt-Fletcher, 2002; Wang & Shenk, 2005), but the core machinery, comprising the gH–gL heterodimer and gB (Browne et al., 2001), is conserved. MuHV-4 membrane fusion is pH-dependent and occurs in late endosomes (Gillet et al., 2008b). Fusion is associated with conformation changes in both gH and gB (Gillet et al., 2008b, c). gB probably switches between pre- and post-fusion states, like the structurally homologous vesicular stomatitis virus glycoprotein G (Roche et al., 2007), but gH is different. It switches from a gL-dependent to a gL-independent conformation in late endosomes (Gillet et al., 2008c), implying that gH and gL dissociate. Yet gL− virions, which constitutively express the downstream form of gH (‘gH-only’), remain infectious; indeed, they show premature rather than impaired membrane fusion (Gillet et al., 2008c). It therefore appears that gH changes from gH–gL to gH-only before engaging in fusion.
Not only is gH different in gL− virions: gB also shows conformational instability. This is consistent with a knock-on effect of the change in gH, as gH–gL and gB are associated in the virion membrane (Gillet & Stevenson, 2007a). A link – probably intramembrane – is maintained between gH and gB, even without gL (Gillet & Stevenson, 2007a), but any extracellular interaction must change, as the gH–gL and gH-only conformations are antigenically very different (Gill et al., 2006). The gB N terminus covers part of gH–gL, and deleting it also seems to destabilize gB (Gillet & Stevenson, 2007b). This region may therefore bridge the gB and gH–gL extracellular domains.
The gB and gH conformation changes present problems for antibodies that would block membrane fusion (Gill et al., 2006; Gillet et al., 2006). First, antibodies must act indirectly, either by blocking conformation changes (probably the major mechanism for gH–gL) or by causing steric hindrance (probably the major mechanism for gB) (Gillet et al., 2008b). Second, they must remain attached to their targets in late endosomes and compete with conformation changes that are energetically favourable at low pH. With glycan shielding (Gillet & Stevenson, 2007b) and poor immunogenicity (Gillet et al., 2007b) also factored in, it is perhaps unsurprising that complete MuHV-4 neutralization is so hard.
The central roles of gL in MuHV-4 cell binding and membrane fusion suggest an additional role for it in virion neutralization. Whether gL itself is a neutralization target is unknown, but gH–gL is the major mAb-defined target on wild-type virions (Gill et al., 2006). This neutralization operates downstream of cell binding, presumably by inhibiting the post-endocytic dissociation of gL from gH. Disrupting gL would remove gH–gL as a target, but could instead reveal other gH epitopes. In order to understand how gL affects neutralization, we compared the infectivity of gL+ and gL− virions after exposure to immune sera or mAbs. Our results explain some of the resistance of wild-type MuHV-4 virions to neutralization and shed new light on herpesvirus entry.
Female C57BL/6 or BALB/c mice (Harlan UK) were infected intranasally with MuHV-4 when 6–8 weeks old, in accordance with local ethics and Home Office Project Licence 80/1992. Immune sera were collected and mAbs were derived at least 3 months later. For the latter, spleen cells were taken 3 days after an intraperitoneal virus boost and fused to NS0 myeloma cells (Köhler & Milstein, 1975). Hybrids were selected with azaserine and typed for virion glycoprotein recognition (Gillet et al., 2007b). Antibodies were concentrated by ammonium sulfate precipitation, dialysed against PBS, isotyped by ELISA (Sigma) and quantified by Mancini assay (Mancini et al., 1965). They are listed in Table 1.
BHK-21 fibroblasts (ATCC CCL-10), NMuMG epithelial cells (ATCC CRL-1636), RAW-264 macrophages (ATCC TIB-71), 293T cells (ATCC CRL-11268), CHO–gH cells (Gill et al., 2006) and CHO–gp70 cells (Gillet et al., 2007a) were grown in Dulbecco's modified Eagle's medium (DMEM) with 2 mM glutamine, 100 U penicillin ml−1, 100 μg streptomycin ml−1 and 10% fetal calf serum. NS0 cells and the hybridomas derived from them were grown in RPMI medium, supplemented as for DMEM. MuHV-4 was propagated in BHK-21 cells (de Lima et al., 2004). Cell debris was removed by low-speed centrifugation (1000 g, 10 min) and virions were recovered from supernatants by high-speed centrifugation (38000 g, 90 min). gp70− (Gillet et al., 2007a), gL− (Gillet et al., 2007c, 2008c) and gM–enhanced green fluorescent protein (eGFP)-tagged (Gillet et al., 2006) MuHV-4 mutants have been described previously. The gL−STOP, gL−DEL and gL−DEL-STOP mutants used here are all phenotypically equivalent. 293T cells were transfected with expression plasmids for glycosylphosphatidylinositol (GPI)-linked forms of gH, gL and gH–gL (Gill et al., 2006; Gillet et al., 2008a) by using Fugene-6 (Roche Diagnostics).
MuHV-4 was titrated by plaque assay (de Lima et al., 2004). After incubation with virus (2 h, 37 °C), BHK-21 cell monolayers were overlaid with 0.3% carboxymethylcellulose and 4 days later fixed and stained in 4% formaldehyde/0.1% toluidine blue. Viruses with an eGFP expression cassette (Adler et al., 2000) were alternatively titrated by eGFP expression. Cells were exposed to virus overnight in phosphonoacetic acid (100 μg ml−1) to prevent infection spreading. RAW-264 cells were further treated with LPS (6 h, 100 ng ml−1) to activate the eGFP expression cassette maximally (Rosa et al., 2007). The proportion of infected cells in each culture was determined by flow cytometry. BHK-21 or NMuMG cells were infected at 0.1–0.3 p.f.u. per cell and RAW-264 cells at 1–3 p.f.u. per cell to give 20–60% eGFP+ cells, eGFP titres being typically 2-fold higher than plaque titres for BHK-21 cells and 10-fold higher for BHK-21 than for RAW-264 cells. Virus titres were calculated by assuming each eGFP+ cell to be a single hit. For neutralization, viruses were incubated with dilutions of serum or mAb (2 h, 37 °C) before being added to the cells and assayed for infectivity as above. All sera were pooled from at least three mice. The sera within each experiment were equivalent in ELISA titre for IgG binding to Triton X-100-disrupted virions (Stevenson & Doherty, 1999).
MuHV-4 virions (3 p.f.u. per cell) were exposed or not to antibody (2 h, 37 °C) then bound to cells on glass coverslips (2 h, 4 °C). The cells were then washed three times in PBS to remove unbound virions, and shifted to 37 °C to allow endocytosis. After incubation at 37 °C, the cells were fixed in 4% paraformaldehyde (30 min), permeabilized with 0.1% Triton X-100 (15 min) and stained with virus-specific mAbs plus Alexa 488-conjugated goat anti-mouse IgG1 (Invitrogen) and Alexa 568-conjugated goat anti-mouse IgG2a. Nuclei were counterstained with DAPI (4,6-diamidino-2-phenylindole). Fluorescence was visualized with a Leica SP2 confocal microscope and analysed with ImageJ. None of the mAbs stained uninfected cells detectably.
Transfected or MuHV-4-infected cells (2 p.f.u. per cell, 18 h) were trypsinized, washed in PBS and incubated (1 h, 4 °C) with MuHV-4-glycoprotein-specific mAbs, followed by fluorescein-conjugated rabbit anti-mouse IgG pAb (Dako Cytomation) or Alexa 633-conjugated goat anti-mouse IgG (Invitrogen). The cells were washed in PBS after each incubation and analysed on a FACScalibur (BD Biosciences).
Virions were lysed and denatured by heating (95 °C, 5 min) in Laemmli's buffer, resolved by SDS-PAGE and transferred to PVDF membranes. The membranes were probed with the ORF17 capsid antigen-specific mAb 150-7D1 plus horseradish peroxidase-conjugated rabbit anti-mouse IgG pAb (Dako Cytomation), followed by ECL substrate development (Amersham Pharmacia Biotech).
We first compared the susceptibility of gL+ and gL− virions to neutralization by sera from mice infected with wild-type MuHV-4 (Fig. 1a). gL− virions were consistently more susceptible than gL+ virions. This applied to independently derived mutants and not to a gL+ revertant (Fig. 1b). gL− mutants were also more susceptible to neutralization by sera from mice infected with gL− MuHV-4 (Fig. 1c).
MuHV-4 binds to heparan sulfate via gH–gL or gp70. gL− virions can only bind via gp70, so a possible explanation for their greater susceptibility to neutralization was that immune sera more readily blocked cell binding. We tested this by neutralizing wild-type, gL− or gp70− virions with sera from wild-type-immune, gL−-immune or gp70−-immune mice (Fig. 2). Sera from mice infected with knockout viruses should selectively lack antibodies against the knocked out gene product. All sera were pooled from at least three mice and had equivalent ELISA titres for detergent-disrupted wild-type virions.
We know from hybridoma analysis that even pooled mice show quite different neutralizing-antibody responses. However, the basic hierarchy between different viruses and sera was very reproducible: the neutralizations of gL− virions by gp70−-immune sera and of gp70− virions by gL−-immune sera were consistently poor, and eight out of eight wild-type-immune serum pools neutralized gL− virions better than wild-type virions. In Fig. 2(a), whilst wild-type-immune serum neutralized gL− virions only marginally better than wild-type virions, the result was clearly different from that for gp70−-immune serum, which neutralized gL− virions less well than wild-type virions. The difference was more obvious in Fig. 2(b), where the ratios of serum to virus were lower. gp70-specific antibodies therefore appeared to be a major reason for gL− virions being more susceptible to neutralization.
The converse was true, to a lesser degree, of gp70− virions: they were neutralized poorly by gL−-immune sera, which lack antibodies to gH–gL. As MuHV-4 neutralization by immune sera correlates with a block to cell binding (Gill et al., 2006) and this requires either gp70 or gH–gL, but not both (Gillet et al., 2008a), it was surprising that gL−-immune and gp70−-immune sera still neutralized wild-type virions quite well (Fig. 2a, bb).). This presumably reflects that immune sera can also neutralize in other ways. For example, some mice mount gB-specific neutralizing responses (Gillet et al., 2006). Antibodies specific for abundant virion glycoproteins could also hinder infection sterically. This would explain why gL−-immune sera neutralized wild-type virions better than they neutralized gp70− virions: gp70 is normally highly abundant and immunogenic (Gillet et al., 2007b), but would be missing from gp70− virions.
Although immune sera block MuHV-4 fibroblast and epithelial-cell infections quite well, they tend to enhance dendritic-cell and macrophage infections via IgG Fc receptors (Rosa et al., 2007). This reflects that Fc receptor binding allows opsonized virions to bypass blocks to conventional cell binding. The infection enhancement depends mainly on antibodies to gp150 (Gillet et al., 2007b); the chief inhibitory antibodies recognize gH–gL (Gillet et al., 2007d).
The greater gL− virus neutralization observed for BHK-21 fibroblasts did not apply to RAW-264 macrophages (Fig. 2b). Very large amounts of immune serum reduced wild-type infection, lower amounts increased it, gp70− virions were similar, and gL− virions showed marked infection enhancement even at serum doses virtually abolishing BHK-21 cell infection. The neutralization differences between gL−-immune and gp70−-immune sera for different virions were also greatly reduced with RAW-264 cells. These data indicated further that the strong neutralization of gL− virions for BHK-21 cell infection by wild-type-immune and gL−-immune sera reflected mainly a better block of cell binding.
The striking enhancement of gL− RAW-264 cell infection by immune sera should not be overinterpreted. gL− viruses showed capsid protein:p.f.u. ratios approximately 3-fold higher than for gL+ (Fig. 2c), i.e. gL− viruses were 3-fold less infectious by plaque assay. gL− and gL+ virus eGFP+ titres were proportionate to their plaque titres. Therefore, gL− virions showed approximately 3-fold less infectivity for RAW-264 cells without antibody than did gL+ virions. The main gL-dependent infection deficits are reduced binding to BHK-21 cells (Gillet et al., 2007c) and premature membrane fusion in NMuMG cells (Gillet et al., 2008c). Opsonization would overcome any macrophage-binding deficit and could alleviate any premature membrane fusion by diverting virions into different endosomes. Such effects would restore gL− infectivity back towards wild-type levels. Thus, it is difficult to compare degrees of gL+ and gL− infection enhancement. Our conclusion from RAW-264 cell infections was simply that gL+ and gL− virions showed much less difference in neutralization when their dependence on heparan sulfate for cell binding was reduced.
Immune sera contain complex mixtures of immunoglobulin specificities and isotypes. We therefore used mAbs to define gp70-directed gL− virion neutralization more precisely (Fig. 3a; Table 1). gp70 comprises four short consensus repeats (SCRs 1–4) and an S/T-rich cytoplasmic domain (Kapadia et al., 1999). As with the homologous protein of the Kaposi's sarcoma-associated herpesvirus (Mark et al., 2006), heparan sulfate binding maps to gp70 SCRs 1–2 (Gillet et al., 2007a). We identified mAb-recognition sites on gp70 by staining cells transfected with membrane-anchored gp70 C-terminal truncation mutants, as described previously (Gillet et al., 2007a). The most effective gL−-virus-neutralizing mAb, LT-6E8, recognized SCR2; T2B11 and T1G10, which also neutralized, recognized SCR1; mAbs specific for SCRs 3–4 did not neutralize (Fig. 3a). gp70-directed neutralization therefore mapped to the same domains as heparan sulfate binding.
mAb T2B11 blocks cell binding by IgG Fc fusions that contain the gp70 heparan sulfate-binding site (Gillet et al., 2007a). LT-6E8 was also effective (Fig. 3b). A fusion of SCR domains 1 and 2 was used here. LT-6E8 also blocked cell binding by SCR1–2–3–Fc and SCR1–2–3–4–Fc (data not shown). Inhibition of SCR1–2–Fc binding by soluble heparin is shown for comparison. Immune sera caused some non-specific inhibition of cell binding, but wild-type-immune and gL−-immune sera were clearly more effective than gp70−-immune or naive sera. Surprisingly, mAb T1G10 failed to block SCR1–2–Fc cell binding even though it neutralized gL− virions. Its epitope may have a different relationship to the heparan sulfate-binding site between the Fc-linked and virion forms of gp70. Overall, gp70-directed neutralization appeared to target either heparan sulfate binding or a very closely linked function.
No gp70-specific IgG mAb blocked RAW-264 cell infection by gL− virions (Fig. 3a). This was again consistent with gp70-directed neutralization blocking cell binding, and therefore being unable to block infection when IgG Fc receptors provided an alternative binding route. In contrast, the gp70-specific IgM mAb T2B11 inhibits RAW-264 infection by wild-type MuHV-4 moderately (Rosa et al., 2007), presumably because its bulk causes steric hindrance and RAW-264 cells lack high-affinity Fcμ binding. gp70-specific IgGs generally enhanced gL+ RAW-264 cell infection better than gL− RAW-264 cell infection. Interestingly, SCR1/2-specific IgGs gave the best enhancement. Such antibodies may mimic the orientation of normal ligand binding and so optimally recruit the virion fusion machinery.
We then tested neutralization by mAb LT-6E8 in combination with mAb 230-4A2 (Gillet et al., 2008a), which blocks heparan sulfate binding by gH–gL–Fc (Fig. 3c). LT-6E8 only inhibited wild-type MuHV-4 infection of BHK-21 cells when combined with 230-4A2. 230-4A2 alone inhibited moderately, presumably because it also stabilizes gH–gL to inhibit membrane fusion; it was much more inhibitory when combined with LT-6E8 to block heparan sulfate binding by both gH–gL and gp70. RAW-264 cell infection resisted this inhibition. Thus, mAbs LT-6E8 and 230-4A2 recapitulated the hierarchical effects of immune sera (Fig. 2): 230-4A2 was analogous to gp70−-immune sera (no LT-6E8-type response), LT-6E8 to gL−-immune sera (no 230-4A2-type response) and both mAbs together to wild-type-immune sera. These data further supported the idea that immune sera inhibit fibroblast infection mainly by blocking heparan sulfate binding.
Although the major neutralization difference between gL+ and gL− virions mapped to heparan sulfate binding by gp70, this did not rule out other additional effects. In particular, the vulnerability of fibroblast infection to cell-binding blocks and the complexities of RAW-264 cell infection by opsonized gL+ and gL− virions would have made post-binding inhibitions by immune sera hard to identify. We therefore explored gL-dependent neutralization further by testing mAbs from MuHV-4-infected mice for preferential neutralization of gL− virions. We identified five mAbs: four were equivalent to LT-6E8, recognizing gp70, and were therefore not analysed further. However, LT-5D3 (Fig. 4a, bb)) recognized gH-only, the gH antigenic form expressed by gL− virions (Gillet et al., 2007c). Other gH-only specific mAbs also neutralized gL− virions (Fig. 4c). They enhanced gL− virus infection of RAW-264 cells at low doses and inhibited it, albeit weakly, at high doses (Fig. 4d), a pattern similar to that of gH–gL-specific mAbs with wild-type virions (Gillet et al., 2007d).
gH–gL-directed neutralization of wild-type virions occurs after binding (Gill et al., 2006). This was apparent from staining cells for ORF25 major capsid antigen with mAb BH-6D3. Fig. 5(a) shows NMuMG cell entry by non-neutralized gL+ and gL− virions, with gN co-staining for comparison. Incoming gL+ virion capsids migrated to the nuclear margin, whereas gL− capsids remained scattered in the cytoplasm (Gillet et al., 2008c). Note that BH-6D3 stains both perinuclear capsids and those still in intact virions at the cell surface. Pre-treating gL+ virions with mAb T2C12 (gH–gL-directed neutralization) blocked the accumulation of perinuclear capsids (Fig. 5b). Virions were still endocytosed, but membrane fusion was presumably blocked (Gill et al., 2006). Neither 8C1 (pan-gH, non-neutralizing) nor MG-4A12 (gH-only) inhibited gL+ capsid transport. gL− neutralization was more difficult to analyse because gL− capsids do not migrate to the nuclear margin. However, the MG-4A12 infection block clearly occurred after virion binding and endocytosis. Such a block was consistent with gH-only normally appearing only after endocytosis, with gH–Fc showing no detectable cell binding (Gillet et al., 2008a) and with gH-only-specific mAbs inhibiting RAW-264 cell as well as BHK-21 cell infections (Fig. 4c).
Although none of the mAbs selected for preferential gL− virion neutralization recognized gB, such a specificity could easily have been missed, as few BALB/c mice make good gB-directed neutralizing responses (Gillet et al., 2006). Several established gB-specific mAbs showed gL-dependent neutralization (Fig. 6a, bb).). Indeed, only MG-2C10, an IgM that binds to the gB N terminus (Gillet et al., 2006), neutralized gL− and gL+ virions similarly. BN-6B5, which recognizes a distinct epitope (Gillet et al., 2008b), was substantially more effective against gL− virions (Fig. 6a). BN-1A7, an IgG2a mAb whose epitope overlaps that of BN-6B5, did not neutralize wild-type virions, but had some effect against gL−. SC-9E8, which recognizes a different epitope in the N-terminal half of gB, was 10-fold more effective against gL− virions than against gL+. mAb T7H9 (Fig. 6b) recognizes an epitope just C-terminal to that of MG-2C10; BH-8F4 is similar to BN-6B5; GB-7D2 has not been mapped; MG-4D11 recognizes the C-terminal half of gB (Gillet et al., 2006). All of these mAbs, as well as 10 others with unmapped gB epitopes, neutralized gL− virions better than gL+. Thus, gL− virions were vulnerable to gB-directed neutralization across multiple sites. This was due neither to gL− virions containing less gB, nor to a conformational difference in gB (Gillet et al., 2007c).
The MuHV-4 gL is small – approximately 19 kDa – and probably lies close to the virion membrane (Gill et al., 2006). It is therefore unlikely to be easily accessible. We derived two gL-specific mAbs. Both stained 293T cells transfected with a membrane-anchored form of gL (Fig. 7a) and stained BHK-21 cells infected with wild-type but not gL− MuHV-4 (Fig. 7b). Immunofluorescence of infected NMuMG cells showed gL expression in a distribution consistent with the endoplasmic reticulum (Fig. 7c). Our gL-specific mAbs neutralized neither wild-type nor gp70− virions. Fig. 7(d) shows representative data. Also, they did not block cell binding by gH–gL–Fc (data not shown). These results were consistent with gL contributing to cell binding and membrane fusion only via gH.
gL is a small but important component of the herpesvirus entry machinery (Roop et al., 1993). The MuHV-4 gL folds gH for heparan sulfate binding (Gill et al., 2006), then dissociates from gH after endocytosis to allow membrane fusion (Gillet et al., 2008c). Both cell binding and membrane fusion are potential neutralization targets, and gL substantially influenced the fate of antibody-exposed virions. The major quantitative effect was on cell binding: blocking the heparan sulfate interaction of gp70 blocked the binding of gL− but not gL+ virions. Therefore, a possible explanation for the redundancy of MuHV-4 heparan sulfate binding – something common to many herpesviruses – is antibody evasion.
Disrupting gL also made the gH-only conformation of gH a neutralization target. This supported the idea that gH-only, although downstream of gH–gL, is still pre-fusion (Gillet et al., 2008c). Thus, gL limits gH-directed neutralization to inhibiting the gH–gL to gH-only transition. Antibody binding to gL-dependent gH epitopes or compound gL–gH epitopes would stabilize gH–gL. Antibodies specific for gL alone are unlikely to neutralize unless gL also changes its conformation significantly when it dissociates from gH. The recognition of both virus-infected cells and recombinant gL by our gL-specific mAbs implied that it does not. Accordingly, they did not neutralize.
Finally, gL disruption made multiple gB epitopes better neutralization targets. This presumably reflected the same molecular events as the previously observed gL-dependent conformational instability of gB (Gillet et al., 2008c). However, the gB on extracellular gL− virions is conformationally normal – its instability manifests only after endocytosis (Gillet et al., 2008c), and gL− virions were susceptible not to new gB-specific antibodies, but to those also recognizing gL+ virions. Therefore the greater vulnerability of gL− virions to gB-directed neutralization appeared to reflect a greater exposure of its normal, pre-fusion form.
Several lines of evidence indicate that disrupting gL destabilizes gB by abolishing an extracellular interaction between gB and gH–gL. First, the gH–gL and gB extracellular domains are both unstable when expressed alone, gH–gL becoming gH-only (Gillet et al., 2008a) and gB adopting mainly its post-fusion form (Gillet et al., 2008b). Second, gH–gL associates with gB and, although their strongest link is probably intra-membrane (Gillet & Stevenson, 2007a), it is hard to envisage how their extracellular domains could avoid being associated too. Third, the gB N terminus hides and so presumably contacts part of gH–gL (Gillet & Stevenson, 2007b). As gB is trimeric (Heldwein et al., 2006), each gB spike could contact three copies of gH–gL; if gH–gL were also multimeric, a two-dimensional lattice could form, and such clustering would allow gH–gL to hide an appreciable portion of gB. gL dissociation would then prime virions for fusion by both changing gH and revealing gB. Normally this would happen in late endosomes. A lack of gL would trigger it sooner, making pre-fusion gB accessible to antibody. Such a model would explain why most gB-specific mAbs that neutralize wild-type virions are IgMs (Gillet et al., 2008b): wild-type virions express some gH-only (Gillet et al., 2007c), implying that some gB is accessible; IgMs could bind one or two vulnerable gBs extracellularly, then use their remaining arms to bind newly revealed gBs in late endosomes.
A complete definition of MuHV-4 entry awaits gB–gH–gL and gB–gH crystal structures, but already some points seem clear. First, the gH–gL–gB composite presents epitopes from both gB and gH–gL. The gH–gL heparan sulfate-binding site defined by mAb 230-4A2 is not made more accessible by deleting the gB N terminus (Gillet & Stevenson, 2007b), and the N-terminal gB epitope defined by MG-2C10 was not revealed better by deleting gL. These must therefore be surface features. Second, when gL is lost, multiple epitopes on both gB and gH–gL are revealed, indicating a large-scale change. Third, the partial availability of these epitopes on wild-type virions implies that not all entry complexes are the same, perhaps because not all gH is bound to gL.
What are the implications for neutralization-based vaccines? As immunodominant viral antigens work against neutralization (Gillet et al., 2007b), one priority is to identify minimal expression systems for key epitopes. An example is gH and gL fused together, which stably present native gH–gL epitopes (Gillet et al., 2007d). gB-specific neutralizing IgMs are probably an unrealistic vaccine goal. However, reconstitution of the epitope of SC-9E8, a neutralizing IgG, may allow a similar approach with gB. Our increasing understanding of MuHV-4 entry makes it a suitable model to test the value of neutralization in a persistent infection.
L.G. is a Postdoctoral Researcher of the Fonds National Belge de la Recherche Scientifique. M.A. is supported by the Portuguese Foundation for Science and Technology. D.L.G. is supported by the Swiss National Science Foundation. P.G.S. is a Wellcome Trust Senior Clinical Fellow (GR076956MA). This work was also supported by Medical Research Council grants G0701185 and G9800903 and CR-UK grant C19612/A6189.