Antibody limits in vivo murid herpesvirus-4 replication by IgG Fc receptor-dependent functions

Antibody is an important antiviral defence. However, it is considered to do little against human gamma-herpesviruses, which establish predominantly latent infections regulated by T cells. One limitation on analysing these infections has been that latency is already well-established at clinical presentation; early infection may still be accessible to antibody. Here, using murid herpesvirus-4 (MuHV-4), we tested the impact of adoptively transferred antibody on early gamma-herpesvirus infection. Immune sera and neutralizing and non-neutralizing monoclonal antibodies (mAbs) all reduced acute lytic MuHV-4 replication. The reductions, even by neutralizing mAbs, were largely or completely dependent on host IgG Fc receptors. Therefore, passive antibody can blunt acute gamma-herpesvirus lytic infection, and does this principally by IgG Fc-dependent functions rather than by neutralization.


INTRODUCTION
Antibody provides an important defence against many epidemic viruses. Its protective efficacy correlates with the neutralization of cell-free virions (Zinkernagel & Hengartner, 2006). This is reflected epidemiologically in virus strains that differ mainly where neutralizing antibodies bind. Whether antibody is also effective against persistent pathogens such as gamma-herpesviruses is less clear. The archetypal gamma-herpesvirus, Epstein-Barr virus (EBV), causes disease mainly when latent; inhibiting its lytic replication during acute infectious mononucleosis makes little difference to latent viral loads (Yao et al., 1989); also, long-term latent infection appears to be controlled by T cells (Rickinson & Moss, 1997). Therefore, antibody would seem to play little part in host defence. However, EBV infection takes at least 1 month to become symptomatic (Hoagland, 1964), by which time latency is already well-established. The early events in infection remain poorly defined. Substantial CD8 + T-cell responses to viral lytic antigens at clinical presentation (Callan et al., 1996) suggest that significant lytic replication occurs early on, and vaccination with a viral lytic antigen can alleviate the symptoms of primary infection (Sokal et al., 2007). Therefore, antibody might still be useful therapeutically in acute settings such as organ transplantation into EBV-naive recipients, where primary EBV infection confers a poor outcome (Walker et al., 1995).
Genetic immunodeficiencies can provide clues as to what controls human viruses (Carneiro-Sampaio & Coutinho, 2007). However, B cell-deficient humans lack the main EBV latency reservoir and consequently fail to support normal persistence (Faulkner et al., 1999). The contribution that antibody makes to host defence has therefore been hard to define. The failure of EBV to establish a significant experimental infection of non-primates further limits the opportunities for direct analysis. We and others Nash et al., 2001;Stevenson & Efstathiou, 2005) have therefore developed murid herpesvirus-4 (MuHV-4) as a small-animal model of gammaherpesvirus pathogenesis. MuHV-4 is genetically closer to Kaposi's sarcoma-associated herpesvirus than to EBV, but its tropism and pathogenesis are similar to those of EBV, and it currently provides the most tractable means to establish how antibody might interact with a gammaherpesvirus in vivo.
T-cell depletion causes disease in B cell-deficient (Stewart et al., 1998) but not immunocompetent (Stevenson et al., 1999) MuHV-4 carriers, implying that antibody helps to control established infection. In support of this idea, adoptively transferred immune sera reduced MuHV-4 replication in B cell-deficient (Gangappa et al., 2002) and T cell-depleted, CD28-deficient (Kim et al., 2002) mice, and a rabbit serum raised against the MuHV-4 gp70 complement control protein reduced MuHV-4 replication in B cell-deficient mice (Gangappa et al., 2002). However, human gamma-herpesviruses cause disease more by latencyassociated tumours rather than by chronic lytic replication, so the main opportunity for antibody to protect against them is likely to be early on. The impact of antibody on early MuHV-4 replication has not been tested.
The mechanism of antibody-mediated protection was not addressed in the MuHV-4 studies listed above, but was speculated to be neutralization in one study (Kim et al., 2002) and correlated with in vitro virus neutralization in the other (Gangappa et al., 2002). However, gammaherpesvirus neutralization is not well understood. For example, EBV can be neutralized in vitro (Thorley-Lawson & Geilinger, 1980), but EBV carriers continue to transmit infection with little evidence of viral antigenic variation, despite making virus-specific antibodies. Immune sera can block MuHV-4 infection of fibroblasts in vitro (Stevenson & Doherty, 1998), but they block in vivo host entry poorly  and tend to enhance MuHV-4 infection of IgG Fc receptor-positive (FcR + ) cells such as macrophages (Rosa et al., 2007). Thus, whilst immune sera can block virions binding to fibroblasts (Gill et al., 2006), virion membrane fusion seems to remain intact. Poor in vivo neutralization could consequently reflect the availability of additional binding routes that are harder for antibody to block. mAbs to gH-gL and gB can block infection post-binding and achieve a more universal neutralization (Gill et al., 2006;Gillet et al., 2006), but gH-gL and gB are poorly immunogenic, so this activity of immune sera is weak . The virion fusion machinery -gH and gB -is also protected by glycosylation  and by gB and gH-gL associating on the virion surface (Gillet & Stevenson, 2007a, b;Gillet et al., 2009). The gB-gH-gL complex appears to unfold only after a post-endocytic dissociation of gL (Gillet et al., 2008a, b), so extracellular gB and gH can remain protected. Another consideration is that in vivo gammaherpesvirus lytic spread may involve direct cell-cell contact more than cell-free virion release, and thereby resist neutralization as described for other herpesviruses (Hooks et al., 1976).
We asked here whether immune sera or monoclonal antibodies (mAbs) could reduce the primary lytic replica-tion of MuHV-4 in immunocompetent hosts and, if so, how they might act. Our aim was to define a basis for antibody-based interventions in primary gamma-herpesvirus infections.

METHODS
Mice. BALB/c, C57BL/6, 129Sv (Harlan UK) and 129Sv6B6 FcRc 2/2 FccRII 2/2 mice (Taconic Europe) were bred in the Cambridge University Department of Pathology animal unit and infected with MuHV-4 when 6-12 weeks old. Intranasal infections were performed in a volume of 30 ml, given under general anaesthesia. For luciferase imaging, mice were injected intraperitoneally with luciferin (2 mg per mouse), anaesthetized with isoflurane, then scanned with an IVIS Lumina (Caliper Life Sciences). Quantitative comparisons used the total radiance (photons s 21 cm 22 ) over each region of interest, relative to a negative-control region. All experiments conformed to local animal-ethics regulations and to Home Office Project Licence 80/1992. Sera and mAbs. Immune sera were collected at least 3 months postinfection and pooled from 10-12 mice each. Hybridomas were derived from MuHV-4-immune mice by fusion with NS0 myeloma cells (Köhler & Milstein, 1975), except for mAb DW-6F6, which was derived from mice infected with influenza A/PR/8/34 and is specific for the influenza H1 haemagglutinin. Antibodies were concentrated from hybridoma supernatants by ammonium sulfate precipitation, dialysed against PBS, isotyped by ELISA (Sigma) and quantified by Mancini assay (Mancini et al., 1965). Those used for in vivo protection are listed in Table 1.
Virus titres and neutralization assays. MuHV-4 was titrated by plaque assay on BHK-21 cells (de Lima et al., 2004). Infectious virus in ex vivo tissues was measured by plaque assay of freeze-thawed tissue homogenates. To determine virus titre in noses, we removed a block of tissue bounded by the cartilaginous tip of the nose anteriorly, the orbits posteriorly, the zygomatic arches laterally, the palate ventrally and the nasal bones dorsally. Cell monolayers were incubated with virus (2 h, 37 uC), overlaid with 0.3 % carboxymethylcellulose and, 4 days later, fixed (4 % formaldehyde) and stained (0.1 % toluidine blue) for plaque counting. Titres were compared statistically between experimental groups by unpaired, two-tailed t-tests, assuming equal variance. The calculations were performed in Excel. P values of ,0.05 were considered statistically significant.
Immunoblotting. N-Linked glycans were removed with protein Nglycanase F; O-linked glycans were removed with sialidase A+Oglycanase according to the manufacturer's instructions (Prozyme). Deglycosylated and intact samples were denatured (95 uC, 5 min) in Laemmli's buffer [60 mM Tris-Cl (pH 6.8) , resolved by SDS-PAGE and transferred to PVDF membranes.
The membranes were blocked with 10 % non-fat milk and probed with MuHV-4-specific mAbs plus horseradish peroxidase-conjugated rabbit anti-mouse IgG pAb (Dako Cytomation). Membranes were washed four times in PBS/0.1 % Tween 20 after each step. Bound antibody was visualized by ECL substrate development (APBiotech).

Immune sera reduce MuHV-4 replication in naive hosts
Adoptively transferred immune sera can reduce MuHV-4 replication in persistently infected mice that lack normal antibody production (Gangappa et al., 2002;Kim et al., 2002). We wanted to know whether immune sera could also reduce acute viral replication in naive, immunocompetent hosts. We tested this by giving MuHV-4-immune sera intraperitoneally to naive mice at the same time as intranasal MuHV-4 virions ( Fig. 1). Significant reductions in virus titres were observed in both C57BL/6 ( Fig. 1 a, b) and BALB/c ( Fig. 1c) mice and across multiple time points. Naive sera had no effect (Fig. 1c).
We used luciferase-positive MuHV-4 (Milho et al., 2009) as an alternative means of tracking in vivo lytic infection ( Fig. 2a, b). Immune sera again reduced significantly the extent of acute host colonization. This was evident both in lungs and in noses. Plaque assays (Fig. 2c) confirmed a . Infectious virus in lungs was measured by plaque assay 3 days later. Each point shows the titre for one mouse; ¾ shows mean values. Immune serum reduced lung titres significantly (P,0.0005 by two-tailed Student's t-test). (b) C57BL/6 mice were infected and treated or not with immune serum as in (a). The effects of immune serum on day 3 and day 6 titres were then compared. The reduction in titre was significant at day 6 (P,0.0001), but not at day 3 (P50.06). (c) BALB/c mice were infected and given immune serum as in (a), or naive serum as an additional control. Virus titres in lungs were determined after 5 days. Immune serum reduced titres significantly compared with either virus only or naive serum (P,0.01). Naive serum had no effect compared with virus only (P50.5).
reduction in nose infection by immune serum. In other experiments, we obtained similar data for noses and lungs, but nose titres were generally more variable between individual mice.
The amounts of mAb used depended on what each hybridoma produced. In general, we used the maximum amount possible, so as not to miss a therapeutic effect. We also ensured that at least 200 mg of any ineffective mAb was given, so as to be sure that ineffectiveness did not reflect Immune serum reduced luciferase expression significantly compared with naive serum (P,0.02 for noses, P,0.004 for lungs) or no-serum controls (P,0.03 for noses, P,0.006 for lungs) using Student's two-tailed t-test. (c) Noses from (a) were titrated for infectious virus by plaque assay. Each point shows the titre for one mouse. Immune serum reduced titres significantly compared with naive serum (P,0.005). Titres with immune serum were also lower than with virus only, although the wide scatter of values made this difference non-significant by t-test (P50.06).

Reduced lytic infection by antibody depends on host IgG Fc receptors
The antiviral effect of IgG occurs principally by Fabdependent virion neutralization and by Fc-dependent recruitment of other immune effectors. Recruiting cellular effectors requires signalling through the common IgG Fc receptor gamma chain (FcRc) (Nimmerjahn & Ravetch, 2008). We therefore explored the mechanism of antibodymediated protection by infecting FcRc 2/2 mice. We used mice that also lacked the FcgRII inhibitory receptor (FcRc 2/2 FccRII 2/2 ), so as to reduce any excess immunosuppressive effect of IgG binding to cells lacking activating Fc receptors. mAbs 3F7 and LT-6E8 both failed to reduce virus titres significantly in the FcR 2 mice, whereas they were effective in FcR + 129Sv controls (Fig. 5a). We obtained an equivalent result (Fig. 5b) with C57BL/6 controls (the FcR 2 mice were 129Sv6C57BL/6) with mAb 58-16D2 (Fig. 5c), and with immune sera using both 129Sv and C57BL/6 controls (Fig. 5c, d). Therefore protection by both non-neutralizing mAbs and immune sera depended on IgG Fc receptors.
Neutralization has at most a minor role in antibody-mediated MuHV-4 control As mAb SC-9A5 is an IgG 3 , it should not engage strongly with IgG Fc receptors (Nimmerjahn & Ravetch, 2008). Therefore, as a neutralizing mAb (Fig. 3), it might be expected to reduce virus titres equally in FcR 2 and FcR + mice. In one experiment (Fig. 6a), SC-9A5 reduced the virus titres of FcR 2 mice, but not to a significant level; in a repeat experiment (Fig. 6b), the reduction was significant, although there was also a significant reduction by mAb 3F7 (P,0.02), something not seen in other experiments (e.g. Figs 5a, 6a). In both experiments, SC-9A5 gave a substantially greater reduction in FcR + than FcR 2 mice (4.6-fold rather than 1.9-fold in Fig. 6a; 17-fold rather than 3.7-fold in Fig. 6b).
We examined further a possible contribution of neutralization to in vivo titre reductions with the gH-gL-specific neutralizing mAbs T2C12 and 230-5B2 (Fig. 6c). Again, there were substantial reductions in FcR + mouse titres, and only small reductions that did not reach statistical significance in FcR 2 mouse titres. Therefore, even neutralizing mAbs acted in vivo largely through IgG Fc receptor engagement.
recognizes an epitope in the region 151-270, also gave no reduction (data not shown).
gp150 recognition was not reduced uniformly: recognition by T1A1 was relatively preserved, whereas recognition by T7F5 was reduced markedly. Therefore, it seemed more likely that gp150 recognition was being masked.
gp150 has at least some some O-linked glycans, because it is immunoprecipitated readily by jacalin (May et al., 2005b). Removing O-linked glycans did not increase convincingly the detection of NMuMG-cell-derived gp150 (Fig. 7e). However, the detection of BHK-21 cellderived gp150 decreased markedly, presumably because gp150 became unstable (or was otherwise lost) when its Olinked glycans were removed. (Note that the actual molecular mass of gp150 is very difficult to infer from SDS-PAGE: O-glycan digestion retards its migration even further.) The greater equivalence of NMuMG and BHK-21 gp150 signals after O-linked glycan removal suggested that glycosylation substantially modified recognition by gp150specific mAbs. Therefore, differential glycosylation could explain the inefficacy of gp150-specific mAbs in reducing in vivo virus titres.

DISCUSSION
Passive antibody, given promptly, attenuates primary alphaherpesvirus infections (Whitley, 1994;Ogilvie, 1998). Gamma-herpesviruses are generally viewed as less accessible to antibody-based interventions, as they colonize their hosts largely by latency-associated lymphoproliferation. We show here that immune sera and mAbs can nonetheless reduce MuHV-4 acute lytic replication and that they act predominantly by IgG Fc receptor engagement.
Antibody can protect mice against herpes simplex virus (HSV) both by neutralization and by IgG Fc receptordependent functions (McKendall, 1985;Ishizaka et al., 1995). Early studies mainly compared neutralizing and non-neutralizing mAbs. Whilst antibodies capable of neutralization need not act by neutralization, the finding that glycoprotein C was a viable target for antibody in vivo and not for neutralization (Balachandran et al., 1982;Kümel et al., 1985) argued that, here at least, protection involved host Fc receptors. Protection by immune serum was subsequently shown to be IgG Fc receptor-dependent (Chu et al., 2008). The difficulty with interpreting such results is that the HSV gE-gI binds human but not mouse IgG Fc (Johansson et al., 1985). In contrast, MuHV-4 naturally infects mice (Kozuch et al., 1993) and gammaherpesviruses are not known to express IgG Fc receptors. Our findings with MuHV-4 -i.e. that non-neutralization epitopes provided effective antibody targets in vivo and that protection by immune serum was largely IgG Fc receptor-dependent -therefore suggested that antibodies generally act via Fc receptors in host defence against herpesviruses.
One difference between MuHV-4 and HSV is that the HSV gD provides a good neutralization target (Minson et al., 1986); the analogous (although non-homologous) MuHV-4 gp150 does not, and seems to function instead in antibody evasion. gp150 is abundant on infected-cell surfaces, but gp150-specific mAbs failed to reduce in vivo virus replication. Some uncertainty remains over gp150 processing: the equivalent gene of most other gammaherpesviruses can be spliced (Beisel et al., 1985) to exclude some of the region equivalent to the immunogenic 'stalk' of gp150 . However, we have found no evidence of gp150 splicing by RT-PCR, either in vitro or in vivo (data not shown). A mAb recognizing the C-terminal part of the gp150 extracellular domain identifies additional, faster-migrating forms on SDS-PAGE, suggesting the possibility of post-translational cleavage (data not shown), but we have not detected corresponding N-terminal fragments in infected-cell supernatants, so these may simply be different glycoforms. Also, gp150 cleavage alone would not explain gp150-specific mAbs being ineffective, as gp70 is shed abundantly into infected-cell supernatants (Kapadia et al., 1999;Gillet et al., 2007c) yet remained a good antibody target. Differential gp150 glycosylation seems more likely to limit the impact of gp150-specific mAbs.
The importance of IgG Fc receptors in limiting in vivo MuHV-4 replication was consistent with MuHV-4 neutralization being much more difficult than in vitro assays of fibroblast infection would suggest . One consideration is virions blocked for fibroblast infection by immune sera using IgG Fc receptors to productively infect myeloid cells . However, there was no evidence for neutralization being more effective in FcRc 2/2 than in FcRc +/+ mice. Therefore, a more important consideration is probably the in vivo accessibility of virions to antibody. gp48, which promotes direct MuHV-4 cellcell spread via actin polymerization (Gill et al., 2008), contributes more to in vivo virus replication (May et al., 2005c) than does gp150, which functions in virion release (de Lima et al., 2004), or gL, which functions in virion cell binding (Gillet et al., 2007d). Similarly, HSV cell-cell spread via gE contributes significantly to dissemination in vivo but not in vitro (Dingwell et al., 1994), whilst cell binding by gp50 is important for pseudorabies virus dissemination in vitro, yet is largely dispensable in vivo (Peeters et al., 1993). Thus, herpesviruses seem generally to remain cell-associated in vivo, affording limited opportunities for neutralization. As the main function of cell-free virions is genome transmission between hosts, their release need not extend beyond apical epithelial cells. The difficulty of herpesvirus neutralization may limit what passive antibody can achieve -certainly the inhibition of MuHV-4 replication achieved here was relatively modest. Further study is required to determine whether this reflects a specific mechanism of viral evasion.
Which IgG Fc receptor-dependent effector mechanism limits MuHV-4 replication remains unclear. Antibodydependent cytotoxicity seems the most likely. Depleting natural killer (NK) cells with mAb PK136 (anti-NK1.1) had no effect on antibody-mediated protection (data not shown). However, macrophages, monocytes or granulocytes may be more important mediators of such cytotoxicity in mice (Nimmerjahn & Ravetch, 2008). Another (non-exclusive) possibility is enhanced virion uptake by dendritic cells improving T-cell priming. Thus, the failure of gp150-specific antibodies to reduce viral titres could have reflected that their opsonization allows particularly efficient dendritic-cell infection after uptake, and therefore disruption of antigen presentation , as opposed to virion uptake without infection, which should allow efficient antigen presentation. The in vivo fate of opsonized virions and infected cells clearly requires further study. The present data indicate that it is this, rather than neutralization, that chiefly determines the therapeutic effect of antiviral antibodies.
Might passive antibody also be effective against EBV? Whilst unlikely to prevent infection, it could potentially be helpful acutely in immunosuppressed patients exposed to EBV by reducing the extent of lytic viral spread. Antibody might also be a useful adjunct to latency-directed therapies such as adoptive T-cell transfer. Most considerations of anti-EBV antibody have focused on gp350 because anti-gp350 mAbs can block B-cell infection in vitro (Thorley-Lawson & Geilinger, 1980). The ineffectiveness of mAbs against the MuHV-4 gp150 suggested that other targets might be better in vivo.