Post-Exposure Vaccination Improves Gammaherpesvirus Neutralization

Herpesvirus carriers transmit infection despite making virus-specific antibodies. Thus, their antibody responses are not necessarily optimal. An important question for infection control is whether vaccinating carriers might improve virus neutralization. The antibody response to murine gamma-herpesvirus-68 (MHV-68) blocks cell binding, but fails to block and even enhances an IgG Fc receptor-dependent infection of myeloid cells. Viral membrane fusion therefore remains intact. Although gH/gL-specific monoclonal antibodies can block infection at a post-binding step close to membrane fusion, gH/gL is a relatively minor antibody target in virus carriers. We show here that gH/gL-specific antibodies can block both Fc receptor-independent and Fc receptor-dependent infections, and that vaccinating virus carriers with a gH/gL fusion protein improves their capacity for virus neutralization both in vitro and in vivo. This approach has the potential to reduce herpesvirus transmission.


INTRODUCTION
Persistent viruses have evolved to co-exist with established host immunity. Herpesviruses are among the most successful, and provide an archetype for many of the immune evasion mechanisms that underlie effective persistence [1][2][3]. In contrast to the plethora of information about T cell evasion, relatively little is known about how herpesviruses evade neutralization by antibody. They must do so, since they continue to transmit infection despite eliciting virus-specific antibodies; by contrast, preexisting antibodies generally block the transmission of nonpersistent viruses. Epidemiological evidence would suggest that herpesvirus antibody evasion is efficient enough even to avoid much selection of viral antigenic variants [4].
It is important when considering antibody evasion to distinguish Fab-dependent neutralization from Fc-dependent antibody functions such as cytotoxicity and opsonization. Herpesviruses transmit between hosts as cell-free virions. Here, antibody evasion must be an evasion of neutralization. In contrast, herpesviruses mainly spread within their hosts via cell/cell contacts [5,6]. These limit virion exposure to antibody [7], so antibody-dependent cytotoxicity is probably a more important host defence than neutralization [8,9]. Other Fc receptor (FcR)-dependent effector mechanisms may also operate [10,11]. Alpha-herpesviruses encode FcR homologs [12] that inhibit host FcR-dependent functions [13]. This reflects that their latency in terminally differentiated neurons makes host colonization highly dependent on lytic spread. In contrast, gamma-herpesviruses colonize their hosts mainly by latency-associated lymphoproliferation [14][15][16]. This may explain why they do not encode FcRs. Yet gamma-herpesvirus must still evade neutralization. A blockade of cell binding by immune serum neutralizes murine gamma-herpesvirus-68 (MHV-68) for infection of FcR 2 cells, but not FcR + cells [17]. Thus, opsonization can rescue the infectivity of antibody-coated virions. This implies that virion membrane fusion still operates, since it is an essential step in infection by any route. FcR-dependent infection is also described for beta-herpesviruses [18,19]. Although MHV-68 productively infects FcR + cells, latency usually pre-dominates and virus production is therefore more protracted than in epithelial cells [17,20]. Thus, with opsonization the likely effects of antibody on transmission and cell/cell spread diverge. By diverting virions into myeloid cells, antibody should damp down MHV-68 lytic infection even when it fails to achieve neutralization. It is the need for transmission rather than the capacity to cause disease that drives viral evolution. A disease readout is therefore not the best way to understand gamma-herpesvirus antibody evasion.
The MHV-68 glycoprotein-specific antibody response predominantly targets gp150, and gp150-specific antibodies account for most of the FcR-dependent infection that is driven by immune sera [21]. Monoclonal antibodies (mAbs) directed against gH/gL block infection at a post-binding step close to membrane fusion [22]. However, gH/gL appears to be poorly immunogenic, as gH/ gL-specific antibodies are only a minor component of the total response [21]. Viral membrane fusion remains obligatory whatever the route of infection. So if gH/gL-specific antibodies can also block FcR-dependent infection, a weak gH/gL-specific response might be crucial to viral evasion of neutralization, and inducing stronger gH/gL-specific immunity might be a means of reducing viral spread. The steady state CD8 + T cell response of MHV-68 carrier mice is not fixed, but can be altered by postexposure vaccination [23]. Such an approach also has the potential to alter antibody responses. We have tested here whether the presentation of gH/gL alone, without its usual accompaniment of more immunogenic MHV-68 virion glycoproteins, can shift the antibody response of virus carriers towards better neutralization.

RESULTS gH/gL-specific antibodies inhibit FcR-dependent MHV-68 infection
We first sought to establish whether gH/gL-specific mAbs could block FcR-dependent infection (Fig. 1). We have shown previously [17] that RAW264.7 macrophages provide a reasonable model of the interaction between antibody-coated virions and FcR + cells, and have validated eGFP expression from the MHV-68 genome as a comparative measure of infection between cell populations [17]. We therefore used this system for in vitro analysis. In contrast to the non-neutralizing gp70-specific mAb 6H10, the neutralizing gH/ gL-specific mAb 7D6 failed to enhance the infection of RAW264.7 macrophages by wild-type MHV-68 and inhibited their infection by gp150-deficient MHV-68 (Fig. 1A). (Gp150-deficient MHV-68 shows enhanced infection of cells such as macrophages that have low glycosaminoglycan expression [24,25].) MAb 7D6 also blocked BHK-21 cell infection by both viruses. Baseline RAW264.7 cell infection by wild-type MHV-68 was too low to identify clear reductions by fluorescence microscopy, but an inhibition of infection by various gH/gL-specific mAbs was evident on flow cytometry (Fig. 1B). A gH/gL-specific mAb was also able to reverse the FcR-dependent infection driven by a nonneutralizing gp150-specific mAb (Fig. 1C).

The gH/gL-specific component of immune serum limits FcR-dependent MHV-68 infection
Although immune sera mainly promote FcR-dependent MHV-68 infection [17], very high doses of some sera can be inhibitory. To test what gH/gL-specific antibodies normally contribute to the effect of whole serum on FcR-dependent infection, we compared sera from mice infected with wild-type or gL-deficient [26] MHV-68 (Fig. 2). gL-deficient MHV-68 mutants colonize mice much like the wild-type [26], and are just equally immunogenic as measured by ELISA for total MHV-68-specific serum antibody (data not shown). However, they elicit no gH/gL-specific antibodies, because they express no gL-dependent epitopes [26].
Wild-type immune serum neutralized both wild-type and gLknockout virions for BHK-21 cell infection. gL knockout-immune serum neutralized wild-type MHV-68 relatively poorly, consistent with gH/gL being a major neutralization target [22]. But much more striking was its very strong enhancement of FcR-dependent infection by wild-type virions. In contrast, gL knockout-immune serum inhibited FcR-dependent infection by gL-knockout virions. This may reflect that gH alone is more readily neutralized than gH/gL. Note that gH-specific antibodies are also present in wildtype immune sera [26]. Thus, it appeared that both monoclonal and polyclonal gH/gL-specific antibodies can inhibit FcR-dependent MHV-68 infection. This argued that the level of gH/gLspecific immunity critically determines the fate of antibodyexposed virions.

Boosting gH/gL-specific immunity reduces IgG FcRdependent and FcR-independent infections
We next tested whether neutralizing antibodies can be boosted by post-exposure vaccination of MHV-68 carrier mice. Our first task was to express a suitable form of gH. Although gH alone can reach the cell surface, it does so in a conformation not recognized by neutralizing mAbs [22]. To ensure that gH would adopt a suitable conformation, we fused it to gL. Since a glycosyl-phosphatidylinositol (GPI)-linked form of gL can fold gH in transfected cells [22], gL is probably sited close to the membrane in the mature gH/gL heterodimer. We therefore fused gL to the C-terminus of the gH extracellular domain (gHL), again with a GPI anchor. Cells transfected with this fusion protein were recognized by all our gH/ gL-specific neutralizing mAbs (n.30, data not shown). For gene delivery, we transferred this construct into vaccinia virus. Cells infected with vaccinia virus expressing the gHL fusion protein (VAC-gHL) displayed gH/gL neutralization epitopes (Fig. 3A). We also generated a vaccinia virus recombinant (VAC-gB) expressing the gB extracellular domain with a GPI anchor, a construct equivalent to one we have used before to identify gBspecific mAbs [27]. A third vaccinia virus, expressing the Nterminal third of gp150 with a GPI anchor, has been described [21]. Infecting MHV-68 carrier mice with each vaccinia recombinant boosted antibody against the cognate MHV-68 glycoprotein, as evident by flow cytometric staining of glycoprotein-transfected cells with immune sera (Fig. 3B).
This boosting correlated with a marked change in the impact of immune sera on MHV-68 infectivity: VAC-150 increased RAW264.7 cell infection, consistent with gp150 driving this process [21], but boosting with VAC-gB or VACgHL reduced it (Fig. 4A). Boosting with VAC-gB or VAC-gHL also improved the serum-mediated neutralization of BHK-21 cell infection (Fig. 4B).
Five months after boosting, the neutralization titres of sera pooled from VAC-gB-boosted or VAC-gHL-boosted mice had declined somewhat, but they still remained significantly above those of control mice for both fibroblast (Fig. 5A) and RAW264.7 cell (Fig. 5B) infections.
gB-specific neutralizing mAbs are only detectable in a minority of MHV-68-infected mice, possibly because significant neutralization requires IgM isotype antibodies [28] and most MHV-68specific B cells produce IgG [29,30]. We therefore also tested individual sera for neutralization (Fig. 5C). The titers of VAC-gBboosted mice were highly variable. One serum neutralized strongly-and probably dominated the pooled sample-but others overlapped with the controls. In contrast, sera from VAC-gHLboosted mice all neutralized MHV-68 better than the controls. This was consistent with gH/gL-specific neutralizing mAbs being more reliably recovered from MHV-68 carriers [22]. Thus, gHL was a more uniformly effective post-exposure vaccine for boosting neutralizing antibodies.

In vivo neutralization
Extrapolating in vitro neutralization to an in vivo setting is not straightforward. Herpesvirus entry routes into naive hosts are not well characterized and are probably multiple. We have seen already with IgG Fc receptors that accessory uptake pathways can bypass apparent blocks to infection [17]. Lectin-mediated uptake may do the same [31]. An in vivo test of neutralization is therefore desirable to confirm that in vitro assays are realistic. One approach to has been to inject mice with congenic [32] or xenogenic [33] immune sera. However, MHV-68 infection is simply too complicated to interpret the effects of such treatments as neutralization without more direct evidence (see Introduction). Indeed, it is unlikely that neutralizing antibodies stop cell/cell viral spread. The major opportunity for neutralizing antibodies to act comes when cell-free virions pass from an infected to a naive host. Therefore one major area of complexity to encompass in neutralization assays is the uptake of antibody-exposed virions at a mucosal surface. With this in mind, we tested the infectivity of virions that had been exposed to immune sera by inoculating them   intranasally into naive mice (Fig. 6A). The readout was simply infection or not, based on virus titers in lung homogenates at 7 days post-inoculation.
In the absence of immune serum, in vivo infectivity was comparable to in vitro: 1 p.f.u. infected 4/6 mice, consistent with a Poisson distribution. Serum-mediated neutralization was much more effective in vitro than in vivo. Thus, 10% immune serum reduced in vitro infectivity at least 1000-fold (1000 p.f.u. was reduced to 0 p.f.u.), but reduced in vivo infectivity no more than 100-fold (100 p.f.u. mixed with immune serum still infected 10/14 mice). We then tested in vivo neutralization by the boosted sera (Fig. 6B). All MHV-68-immune sera neutralized 10 p.f.u. for intranasal infection, but only the gHL-boosted and gB-boosted sera completely neutralized 100 p.f.u. and only the gHL-boosted sera were able to stop at all (3/6 mice) infection by 1000 p.f.u.. In vivo neutralization was therefore improved significantly by postexposure vaccination.

DISCUSSION
Immune sera block MHV-68 infection of fibroblasts, but fail to block and even enhance its infection of FcR + cells such as macrophages and dendritic cells. At mucosal surfaces, FcR + and DC-SIGN + [31] dendritic cell processes [34]; mucosal epithelial FcRn [35]; and M cell transcytosis [36] all provide potential uptake routes for antibody-coated virions. Merely blocking cell binding may therefore not suffice to block viral transmission. In contrast, membrane fusion is essential for enveloped virions to infect regardless of the uptake route. It should therefore be a universally effective neutralization target. gH/gL-specific antibodies block MHV-68 infection at a point between cell binding and membrane fusion [22]. We showed here that they can inhibit the infection of both FcR 2 and FcR + cells, and can therefore potentially block all routes into the naive host.
MHV-68 infection elicits gH/gL-specific antibodies, but quantitative analysis of the glycoprotein-specific antibody reponse indicates that gH/gL is quite poorly immunogenic [21]. gH/gL neutralizing epitopes are lost if gL and gH dissociate [22], and B cells specific for other, more immunogenic virion glycoproteins probably reduce the gH/gL antigen load [21]. The effectiveness of gH/gL mAbs and polyclonal sera in blocking MHV-68 infection argued that poor gH/gL immunogenicity is a crucial factor in allowing antibody-exposed virions to remain infectious in vivo. We showed that the steady state antibody response of virus carriers is not fixed, but can be modified by post-exposure vaccination. Subunit vaccination avoided the problem with whole MHV-68 of eliciting antibodies mainly to gp150, which promote FcRdependent infection [21]. By selectively boosting gH/gL-specific antibody, the balance of immune serum was tipped from promoting FcR-dependent infection to inhibiting it. In vivo neutralization-which was notably more difficult for immune sera to achieve than in vitro neutralization-was also improved. Immunization with a gH-gL fusion protein therefore has the potential to improve unfavourable antibody responses in herpesvirus carriers, and thereby to reduce their infectivity.
The limitation of this approach is likely to be that herpes virions are intrinsically hard to neutralize. The relatively low efficiency with which gH/gL-specific mAbs neutralize MHV-68 [22] implies that the fusion machinery is quite well protected. Even saturating levels of a MHV-68 gH/gL-specific neutralizing mAb fail to protect against in vivo infection with 10 p.f.u. MHV-68 (P.G. Stevenson, unpublished data). Fortunately, polyclonal immune sera generally neutralize more effectively than monoclonal antibodies because different antibody specificities can have synergistic inhibitory effects. It evidently remained feasible for a boosted, polyclonal gH/gL-specific response to block infection in vivo. As 1 virion in principle infects no less well than 1000, the neutralization of 100 p.f.u. by gHL-boosted sera implied that the average infectivity of each virion could be reduced to less than 1% of normal for all infection routes. Even if in vivo infectivity is not abolished completely, the limited transmissibility of persistent viruses suggests that modest infectivity reductions might suffice to reduce significantly viral prevalence.
One complication of post-exposure vaccination is that several herpesviruses assemble alternative fusion complexes [37][38][39]. The best vaccine would presumably be the predominant fusion complex on shed virions-with EBV that incorporating gp42 [40]. It may prove sufficient to target just the fusion complex associated with the major route into naive hosts, or it may prove necessary to target more than one fusion complex. Another key task is to test post-exposure vaccination in an in vivo transmission model. One has not yet been established for MHV-68. Nevertheless, our study of MHV-68 shows that while the difficulty of in vivo neutralization should not be under-estimated, it can in principle be achieved by boosting antibodies against appropriate virion glycoproteins. Cells BHK-21 cells, RAW264.7 cells, CHO and CHO-gB cells [27] cells, L929-gp150 cells [21], NIH-3T3-CRE cells [41] and TK -143 cells were grown in Dulbecco's modified Eagle medium (Invitrogen, Paisley, U.K.) supplemented with 2 mM glutamine, 100 U/ml penicillin, 100 mg/ml streptomycin and 10% fetal calf serum (PAA laboratories, Linz, Austria). Where indicated, cells Wild-type MHV-68 virions were filtered to remove any aggregates, then diluted as shown (nominal PFU) and incubated or not with 10% serum pooled from mice infected 3 months previously with wild-type MHV-68. Each sample was then titrated on BHK-21 cells (p.f.u.) and used to infect naive mice intranasally (mice infected). Infection of mice was determined by plaque assay of lungs harvested 7 days later, and were scored simply as positive or negative depending on whether any virus was present. The in vivo data are pooled from 3 different experiments that all gave consistent results. B. Wild-type MHV-68 virions were incubated with pooled serum samples from vaccinia virus boosted MHV-68 carrier mice as in A, then given intranasally to naive mice (5 ml of serum per mouse with the virus dose shown, using 6 mice per group). Lungs were titered for infectious virus at 7 days post-infection and scored as either infected or not. The data are pooled from 2 equivalent experiments (3 mice per group each). doi:10.1371/journal.pone.0000899.g006

Viruses
MHV-68 was derived from a genomic BAC, which also transcribes eGFP from a human cytomegalovirus IE-1 promoter [42]. gL-deficient [26] and gp150-deficient [24] derivatives have been described. For all in vivo infections, the loxP flanked eGFP and BAC sequences were removed by passaging viruses through NIH-3T3-CRE cells. Viruses were grown in BHK-21 cells. Infected cultures were cleared of infected cell debris by low-speed centrifugation (10006g, 3 min). Virions were then concentrated by high speed centrifugation (380006g, 90 min). Virus stocks for in vivo neutralization assays were also filtered (0.45 mm) to remove any aggregates. Virus titers of stocks, antibody-treated samples and lung homogenates were determined by plaque assay on BHK-21 cells [15]. To make gHL-GPI, the gL coding sequence lacking its stop codon and the first 13 amino acid residues of its signal peptide [22] was amplified by PCR (Phusion DNA polymerase, New England Biolabs, Hitchin, U.K.) with NotI-restricted primers and cloned into the NotI site between gH and its GPI anchor in pBRAD-gH [22] to make pBRAD-gHL. Residues AFVSLSTC of the predicted gL signal peptide were retained as a linker between gL and gH. CHO cells were transfected with pBRAD-gHL to make CHO-gHL cells. The correct expression of virion gH/gL epitopes by this cell line was confirmed with.30 gH/gL-specific mAbs (data not shown). To make VAC-gHL, the gHL-GPI coding sequence was amplified from pBRAD-gHL using 59 AvrIIrestricted and 39 HinDIII-restricted primers and cloned into the NheI/HinDIII sites of pMJ601 [43]. To make VAC-gB, the gB-GPI coding sequence was amplified from pBRAD-gB [27], again with 59 AvrII-restricted and 39 HinDIII-restricted primers, and cloned into pMJ601 as for VAC-gHL. pMJ-601-gHL-GPI and pMJ601-gB-GPI were each transfected into vaccinia virus WRinfected TK -143 cells. Thymidine kinase-deficient recombinants were selected by passage in 25 mg/ml 59-bromo-29-deoxyuridine (Sigma Chemical Co, Poole, U.K.) and identified by beta galactosidase expression using X-gal substrate. They were purified to homogeneity by limiting dilution cloning. VAC-gp150 has been described [21].