Glycoprotein B Cleavage Is Important for Murid Herpesvirus 4 To Infect Myeloid Cells

Glycoprotein B (gB) is a conserved herpesvirus virion component implicated in membrane fusion. As with many—but not all—herpesviruses, the gB of murid herpesvirus 4 (MuHV-4) is cleaved into disulfide-linked subunits, apparently by furin. Preventing gB cleavage for some herpesviruses causes minor infection deficits in vitro, but what the cleavage contributes to host colonization has been unclear. To address this, we mutated the furin cleavage site (R-R-K-R) of the MuHV-4 gB. Abolishing gB cleavage did not affect its expression levels, glycosylation, or antigenic conformation. In vitro, mutant viruses entered fibroblasts and epithelial cells normally but had a significant entry deficit in myeloid cells such as macrophages and bone marrow-derived dendritic cells. The deficit in myeloid cells was not due to reduced virion binding or endocytosis, suggesting that gB cleavage promotes infection at a postendocytic entry step, presumably viral membrane fusion. In vivo, viruses lacking gB cleavage showed reduced lytic spread in the lungs. Alveolar epithelial cell infection was normal, but alveolar macrophage infection was significantly reduced. Normal long-term latency in lymphoid tissue was established nonetheless.

G lycoprotein B (gB) is a conserved herpesvirus virion component involved in cell binding and viral membrane fusion (1). It and the structurally homologous vesicular stomatitis virus (VSV) G protein and baculovirus gp64 are considered to be class III fusion proteins (2)(3)(4)(5)(6)(7). Most herpesvirus gBs contain an R-X-K/R-R recognition motif for the ubiquitously expressed, subtilisin-like cellular endoprotease furin (8). Furin has many substrates, including cellular proteins, bacterial toxins, and glycoproteins from avian influenza virus, human immunodeficiency virus, measles virus, respiratory syncytial virus, and Ebola virus (9). It acts in the trans-Golgi network (TGN), plasma membrane, and early endosomes (9), and gB cleavage is presumed to occur in the TGN (10).
The role of gB furin cleavage has been addressed for several herpesviruses. For pseudorabies virus (PRV), cleavage of gB by furin is not required for cell entry or replication in vitro but is important for syncytium formation (11); mutation of the furin cleavage site (FCS) of bovine herpesvirus 1 does not affect viral replication but reduces plaque size, consistent with impaired cellto-cell spread (12); varicella-zoster virus FCS mutation affects neither replication kinetics nor plaque size in vitro but significantly impairs replication in human skin xenografts in SCID mice (13); human cytomegalovirus (HCMV) FCS mutation does not affect replication in vitro (14); and Epstein-Barr virus (EBV) FCS mutation reduces transfection-based cell-to-cell fusion (15). Together, these studies argue that gB cleavage is nonessential for herpesvirus replication in vitro but may have a role in gB fusion activity that manifests as reduced cell-to-cell spread. However, none of these analyses addressed directly the role of gB cleavage in host colonization, that is, entry, primary lytic replication, dissemination, and latency establishment. To do this, we employed the accessible, small-animal gammaherpesvirus infection model provided by murid herpesvirus 4 (MuHV-4).
MuHV-4 is genetically closer to Kaposi's sarcoma-associated herpesvirus (KSHV) than to EBV (16) but shares with EBV an infectious mononucleosis syndrome, exploitation of host germinal centers, and long-term latency in memory B cells (17,18).
Thus, it is likely that all these viruses colonize their hosts in fundamentally similar ways. Upon intranasal inoculation, MuHV-4 replicates in the respiratory tract and is trafficked by dendritic cells (DCs) to the draining lymph nodes, where it is transmitted to B cells (19)(20)(21). B cell activation and proliferation then drive a T cell mononucleosis (21)(22)(23)(24)(25).
The gB of MuHV-4 is predicted to have a structure very similar to those of herpes simplex virus 1 and EBV. MuHV-4 membrane fusion is blocked by gB-specific monoclonal antibodies (MAbs) that recognize an epitope close to one of the predicted fusion loops (26). Like EBV and KSHV, the MuHV-4 gB contains a furin recognition motif, and the 120-kDa precursor protein is cleaved into 65-and 55-kDa disulfide-linked N-and C-terminal fragments (27). Cell-associated gB is predominantly uncleaved while virion gB is almost entirely cleaved, suggesting that cleavage occurs during gB incorporation into virions (27). This would be consistent with the TGN being one of the main localizations of both furin activity and herpesvirus secondary envelopment (9,10,28).
We prevented MuHV-4 gB cleavage by mutating its FCS. In vitro, the mutant viruses entered fibroblasts and epithelial cells normally but entered myeloid cells such as macrophages and DCs poorly. After intranasal infection of mice, viruses lacking gB cleavage showed significantly less lytic spread in the lungs. Histological analysis showed reduced infection of alveolar macrophages, while electrophoresis. Mutant viruses were reconstituted by transfection of mutant BACs into BHK-21 cells using FuGENE 6 transfection reagent (Roche). To ascertain that the mutant viruses carried the correct mutations, DNA was extracted from virion preparations using the Wizard genomic DNA purification kit (Promega) and analyzed with the diagnostic PCR described above and by Southern blotting. For the latter, viral DNA was digested with KpnI (Roche), separated on a 1% agarose gel, transferred and UV-cross-linked to a positively charged nylon membrane (Roche), hybridized with a P32-labeled probe consisting of the MuHV-4 genomic HindIII N fragment (genomic coordinates 16239 to 19870), and exposed to X-ray films (Fujifilm). The BAC ϩ viruses retaining the BAC cassette and the HCMV IE1 enhancer/promoter-driven enhanced green fluorescent protein (eGFP) expression cassette therein were used as eGFP ϩ reporter viruses. In order to obtain BAC Ϫ viruses for in vivo infections, the loxP-flanked BAC cassette was removed from viral genomes by passage through NIH 3T3-Cre cells (31). MuHV-4 stocks were grown in BHK-21 cells (31). The medium of infected cells was harvested, cell debris was removed by low-speed centrifugation (1,000 ϫ g, 10 min), and virions were recovered from supernatants by high-speed centrifugation (38,000 ϫ g, 90 min). After resuspension of virions in complete medium, potential remaining cell debris and virion aggregates were removed by filtration through a 0.45-m cellulose acetate filter (Whatman). To determine eGFP titers of eGFP ϩ virus stocks, viruses were added to BHK-21 cells and incubated overnight in the presence of 100 g/ml phosphonoacetic acid (PAA) to prevent secondary spread, followed by counting the proportion of eGFP ϩ cells with a FACSCalibur (BD Biosciences) or Gallios (Beckman Coulter) flow cytometer and calculation of the multiplicity of infection (MOI) using the Poisson distribution. To determine PFU titers, virus stocks were titrated by plaque assay on BHK-21 cells as described previously (31).
Western blot analysis and deglycosylation of virion gB. Equal volumes of each virus stock (1 l, corresponding to approximately 10 5 PFU) were lysed in 20 l protein loading buffer (PLB) (39) and boiled for 10 min. SDS-PAGE and Western blotting were done as described previously (36). Enzymatic deglycosylation was done using the GlycoPro enzymatic deglycosylation kit for N-linked and simple O-linked glycans (ProZyme) and recombinant endoglycosidase H (EndoH; New England BioLabs) according to the manufacturer's instructions. Briefly, virions (1 l, corresponding to approximately 10 5 PFU) were denatured in denaturation buffer for 10 min at 100°C, followed by deglycosylation with the corresponding enzymes for 3 h at 37°C. After addition of equal volumes of 2ϫ PLB, the samples were processed for SDS-PAGE and Western blotting.
Staining of infected cells for flow cytometry. BHK-21 cells were infected at an MOI of 2 PFU/cell for 18 h. After trypsinization and one wash in PBS, the cells were stained with hybridoma supernatants diluted 1:2 in PBS containing 3% bovine serum albumin (BSA) for 1 h at 4°C. After one wash in PBS, the cells were incubated with Alexa Fluor 633 goat antimouse IgG (Invitrogen) diluted 1:2,000 in PBS containing 5% normal goat serum for 1 h at 4°C. After one wash in PBS, the cells were analyzed on a FACSCanto flow cytometer (BD Biosciences).
ELISA on infected cells. NMuMG cells were allowed to adhere overnight to 96-well plates before precooling for 1 h at 4°C, followed by incubation with MuHV-4 virions diluted in ice-cold complete medium for 2 h at 4°C. The cells were then washed 3 times in ice-cold PBS to remove unbound virions and either fixed directly or first incubated at 37°C in complete medium to allow virion endocytosis. After one wash in ice-cold PBS, cells were fixed by addition of ice-cold 4% formaldehyde in PBS and being left at room temperature (RT) for 1 h. Fixation was then stopped by incubation with 0.1 M glycine in PBS (15 min, RT), followed by 3 washes in PBS. The cells were then permeabilized with 0.1% Triton X-100 in PBS (30 min, RT), blocked with 2% BSA-0.1% Tween 20 in PBS (overnight, 4°C), and then incubated with primary MAbs (hybridoma supernatants) diluted 1:2 in 2% BSA-0.1% Tween 20 in PBS (3 h, RT), followed by 3 washes in PBS-0.1% Tween 20. The plates were then incubated with alkaline phosphatase-conjugated goat anti-mouse IgG pAb (Southern Biotech) diluted 1:1,000 in 2% BSA-0.1% Tween 20 in PBS (3 h, RT), followed by 6 washes in PBS-0.1% Tween 20. Bound secondary antibodies were detected by incubation with Sigma Fast p-nitrophenyl phosphate substrate (Sigma-Aldrich) and reading the absorbance at a 405-nm wavelength on a Sunrise microplate reader (Tecan). All absorbance values were normalized to the values measured for virus bound at 4°C.
In vivo luciferase imaging. Luciferase imaging of infected mice was essentially performed as described previously (20). Briefly, 1.5 mg/mouse D-luciferin (Caliper Life Sciences) diluted in 100 l PBS was applied by intraperitoneal injection 5 min before imaging in an IVIS Lumina imager (Caliper Life Sciences) under isoflurane anesthesia. Mice were imaged from the ventral, dorsal, and left lateral sides for 1 min each. Data are shown as the total flux of photons per second in the area of the lung (sum of dorsal and ventral signals), the superficial cervical lymph nodes (SCLN) (ventral signal), or the spleen (left lateral signal).
Titration of virus in infected lungs. The lungs of infected mice were harvested and stored in complete medium at Ϫ70°C until analysis. The thawed tissues were then homogenized in complete medium with a blender, added to BHK-21 monolayers, and incubated for 2 h at 37°C before being overlaid with complete medium containing 0.3% carboxymethyl cellulose. After 4 days of culture, the cells were washed once in PBS, fixed with 4% formaldehyde in PBS, and stained with 0.1% toluidine blue.
Infectious center assay. Spleens from infected mice were homogenized in Griffiths tubes, filtered through a 70-m strainer, added to BHK-21 monolayers, and incubated for 2 h at 37°C before being overlaid with complete medium containing 0.3% carboxymethyl cellulose. After 4 days of culture, the cells were washed once in PBS, fixed with 4% formaldehyde in PBS, and stained with 0.1% toluidine blue.
qPCR for MuHV-4 DNA. DNA was extracted from lymph node or spleen suspensions with the Wizard genomic DNA purification kit (Promega). Eighty nanograms of cellular DNA was then analyzed by quantitative PCR (qPCR) on a Rotor-Gene 3000 thermal cycler (Corbett Re-search) using HotStarTaq DNA polymerase (Qiagen). To quantify MuHV-4 DNA, part of the M2 open reading frame (ORF) (genomic coordinates 4166 to 4252) was amplified with primers M2 S (GTCAGTCG AGCCAGAGTCCAAC) and M2 R (ATCTATGAAACTGCTAACAGTG AACC). The PCR product was quantified by hybridization with probe M2 TM (TCCAGCCAATCTCTACGAGGTCCTTAATGA) labeled with 6-carboxyfluorescein (FAM) (5=) and 6-carboxytetramethylrhodamine (TAMRA) (3=). Amplification of serial dilutions of the MuHV-4 HindIII E fragment (genomic coordinates 107 to 6263) cloned into plasmid pBR322 (40) served as a standard curve. To quantify the cellular DNA, part of the adenosine phosphoribosyl transferase (APRT) gene was amplified with primers APRT F (GGGGCAAAACCAAAAAAGGA) and APRT A (TGTGTGTGGGGCCTGAGTC) and quantified with probe APRT TM (TGCCTAAACACAAGCATCCCTACCTCAA) labeled with Cy5 (5=) and BlackBerry quencher (BBQ) (3=). Amplification of serial dilutions of the PCR product cloned into plasmid pGEM-T Easy (Promega) served as a standard curve. The MuHV-4 genome copy numbers were then normalized to the APRT copy numbers.
Immunohistofluorescence. Lungs were fixed in PBS containing 1% formaldehyde, 10 mM sodium periodate, and 75 mM L-lysine (24 h, 4°C); equilibrated in 30% sucrose (18 h, 4°C); and then frozen in OCT and sectioned (7 m) on a cryostat. Sections were air dried (2 h, RT) and blocked with PBS containing 2% BSA and 2% normal goat serum (1 h, RT). Sections were then incubated with primary antibodies diluted in PBS containing 2% BSA and 2% normal goat serum for 18 h at RT. Virusexpressed eGFP was detected with a rabbit pAb (Abcam), CD68 was detected with a rat MAb (clone FA-11; BioLegend), and podoplanin was detected with a goat pAb (R&D Systems). After incubation, sections were washed 3 times in PBS and then incubated (1 h, RT) with Alexa Fluor 568 donkey anti-goat IgG (Invitrogen). After 3 washes in PBS, potential free binding sites on the donkey anti-goat IgG conjugate were blocked by incubation (30 min, RT) with 2% normal goat serum. After 3 washes in PBS, the sections were incubated (1 h, RT) with a combination of Alexa Fluor 488 goat anti-rabbit IgG and Alexa Fluor 633 goat anti-rat IgG (Invitrogen). After 3 further washes in PBS, the sections were mounted in Prolong Gold containing 4=,6-diamidino-2-phenylindole (DAPI; Invitrogen). Fluorescence was visualized using a Leica SP5 confocal laser scanning microscope, and images were analyzed with ImageJ.
Statistical analysis. Unless indicated otherwise, the pooled values of the gB FCS Ϫ viruses (⌬FCSv1, ⌬FCSv2, and ⌬FCSv3) were compared to the pooled values of the gB FCS ϩ viruses (WT, ⌬FCSv1 R, ⌬FCSv2 R, and ⌬FCSv3 R) by Student's t test on Microsoft Excel 2010 using the following settings: two-tailed distribution and two-sample unequal variance. P values of Ͻ0.05 were considered statistically significant and are mentioned in the figure legends.

Mutation of the furin cleavage site of MuHV-4 gB.
To investigate the role of gB cleavage in MuHV-4 infection, we mutated its canonical FCS, 431 R-R-K-R 434 , located in a short loop within domain II (Fig. 1A). Three different mutations were introduced: first, deletion of the entire motif (⌬FCSv1); second, replacement by a 431 G-T-G-G 434 motif (⌬FCSv2); and third, replacement by a 431 G-T 432 motif (⌬FCSv3) (Fig. 1A). The mutations were introduced into the background of a recombinant MuHV-4 strain 68 expressing firefly luciferase from an early lytic cycle MuHV-4 promoter (M3-LUC), which allows in vivo imaging of infected mice without detectable virus attenuation (20). Mutagenesis of the bacterial artificial chromosome (BAC)-cloned M3-LUC MuHV-4 genome was performed by RecA-mediated homologous recombination in E. coli as previously described (29,30). The recombinant BACs containing the ⌬FCSv2 and ⌬FCSv3 mutations, including the diagnostic KpnI site, were identified by KpnI digestion and agarose gel electrophoresis (not shown). Since the ⌬FCSv1 muta- To this end, we analyzed viral DNA by KpnI digestion and Southern blot analysis to identify the ⌬FCSv2 and ⌬FCSv3 mutations and by the diagnostic PCR spanning the FCS to identify the ⌬FCSv1 mutation. Southern blot analysis confirmed that the ⌬FCSv2 and ⌬FCSv3 mutant viruses showed two KpnI fragments, 3.2 and 2.4 kb, as opposed to the wild-type (WT), ⌬FCSv1, and revertant viruses, which showed a single 5.6-kb KpnI fragment (Fig. 1D). The diagnostic PCR confirmed that the ⌬FCSv1 and the ⌬FCSv3 mutants showed shorter PCR products than did the WT, ⌬FCSv2, and revertant viruses, consistent with the deletion of 12 bp in the ⌬FCSv1 mutant and 6 bp in the ⌬FCSv3 mutant (Fig. 1E).
Deletion of the furin cleavage site abolishes cleavage of gB. The effect of the FCS mutations on gB cleavage was assessed by Western blotting for virion gB. A MAb (MG-2C10) recognizing an epitope N terminal of the FCS detected in WT and revertant viruses a weak 120-kDa band-full-length gB-and a strong 65-kDa band-its N-terminal gB cleavage product ( Fig. 2A). In contrast, only full-length gB was detected in the mutant viruses ( Fig.  2A), demonstrating that the FCS mutations completely abolished gB cleavage. The 65-kDa band appears as a doublet because it comigrates with the bovine serum albumin (66 kDa) present in virion preparations. A MAb (MG-4D11) recognizing an epitope C terminal of the FCS similarly recognized mainly cleaved gB (55 kDa) with WT and revertant viruses and only the full-length form in the FCS mutants (Fig. 2B).
The weak 60-kDa band preserved in the mutant viruses (open arrowhead in Fig. 2B) appeared not to be gB because it differs in size from the 55-kDa C-terminal cleavage product. We confirmed this for the ⌬FCSv1 mutant by enzymatic deglycosylation prior to Western blotting ( Fig. 2C and D). Full-length gB showed similar endoglycosidase H (EndoH) sensitivity and similar extents of Nlinked (N-glycanase/peptide-N-glycosidase F [PNGase F] treatment) and O-linked (sialidase A plus O-glycanase treatment) glycosylation between WT and ⌬FCSv1 mutant viruses. Enzymatic removal of N-linked glycans made clear that the 60-kDa band observed for the mutant viruses in Fig. 2B (open arrowhead) was also present in the WT virus but hidden by the strong gB-specific 55-kDa band (Fig. 2D, open arrowheads). We presume that this band reflected weak cross-reactivity of MAb MG-4D11 with an unrelated glycoprotein.
Uncleaved gB shows no obvious change in antigenicity. gB is antigenically different between extracellular virions (i.e., prefusion) and after capsid and tegument release (i.e., postfusion). Prefusion gB is recognized by MAbs BN-1A7 and SC-9E8 and not by MAb MG-1A12; postfusion gB is recognized by MAb MG-1A12 and not by MAbs BN-1A7 and SC-9E8 (26,33). To determine whether FCS mutation affected the stability of prefusion gB, we stained the surface of infected BHK-21 cells with MAbs BN-1A7, SC-9E8, and MG-1A12, followed by flow cytometry. Staining with MAbs T2C12 recognizing gH/gL and BN-3A4 recognizing gp150 controlled for equivalent infection of cells. As shown in Fig. 3A, the ⌬FCSv1, ⌬FCSv2, and ⌬FCSv3 mutants displayed the same staining as did WT and the revertant viruses.
To determine if FCS mutation affected the antigenic transition of prefusion into postfusion gB during cell entry, we performed a kinetic analysis of virion entry by ELISA on cell monolayers (36). A sensitive assessment of the changes in virion antigenicity during cell entry requires a synchronous entry of the whole virion popu-lation. These experiments were conducted in NMuMG epithelial cells, because virion entry is most efficient and synchronous in this cell line. To this end, virions were bound to cells at 4°C, unbound virions were removed by washing, and the cells were fixed either directly or after further incubations at 37°C to allow virion endocytosis. The gB on virions bound to cells at 4°C is in the prefusion conformation (BN-1A7 ϩ MG-1A12 Ϫ ), while after incubation for 2 h at 37°C, capsid and tegument have been released and gB is in the postfusion conformation (BN-1A7 Ϫ MG-1A12 ϩ ) (33). As shown in Fig. 3B, the ⌬FCSv1 and ⌬FCSv3 mutant viruses displayed similar kinetics of BN-1A7 epitope loss and MG-1A12 epitope gain as did the WT and ⌬FCSv1 revertant viruses. To- gether, these data demonstrate that FCS mutation does not affect the stability of prefusion gB and its transition into postfusion gB during cell entry.
Growth properties of gB FCS ؊ viruses. To address the effect of the gB FCS mutations on viral replication, we performed growth curves in BHK-21 fibroblasts. Upon low-MOI infection (0.05 enhanced green fluorescent protein [eGFP] units/cell), all gB FCS Ϫ viruses (⌬FCSv1, ⌬FCSv2, and ⌬FCSv3) showed moderately but significantly reduced growth compared to the gB FCS ϩ viruses (WT, ⌬FCSv1 R, ⌬FCSv2 R, and ⌬FCSv3 R) (Fig. 4A). Upon high-MOI infection (15 eGFP units/cell), the ⌬FCSv1 and ⌬FCSv3 mutants grew only marginally less well than the gB FCS ϩ viruses, while the ⌬FCSv2 mutant showed a more substantial growth defect (Fig. 4B). Since the growth defect was significantly greater for the ⌬FCSv2 than for the ⌬FCSv1 and ⌬FCSv3 mutants, we concluded that the observed phenotype of the ⌬FCSv2 mutant is due not to lack of gB cleavage but perhaps to an unspecific effect of its introduced G-T-G-G sequence. The reason for this remains unclear, especially since ⌬FCSv2 is the only mutation to retain the correct spacing within the gB molecule (Fig. 1A). This mutant was therefore excluded from further analysis. Together, these findings show that lack of gB cleavage moderately reduces multistep virus growth, while it only marginally affects single-step growth.
gB FCS ؊ viruses show impaired myeloid cell infection. MuHV-4 host colonization involves at least four cell types: epithelial cells for lytic replication in the lung (21), DCs and macrophages for transport to lymph nodes and transmission to B cells (19,41), and B cells for systemic dissemination and persistence (42,43). We tested the capacity of gB FCS Ϫ viruses to infect each of these cell types in vitro. Infection was assessed by eGFP expression from eGFP ϩ versions of WT, ⌬FCSv1 and ⌬FCSv3 mutant, and ⌬FCSv1 revertant viruses. We included phosphonoacetic acid (PAA) in the cultures to prevent viral DNA replication and thus any secondary spread. After addition of viruses, the cells were incubated overnight and the proportion of eGFP ϩ cells was then determined by flow cytometry. Input virus titers were determined on BHK-21 fibroblasts; their infection therefore controlled for equivalent virus doses. The gB FCS ϩ and FCS Ϫ virus stocks had similar titers (not shown) and protein contents (Fig. 2); infection at identical MOIs therefore involved the addition of similar virion amounts. Figure 5A and B show that NMuMG epithelial cell infection exactly mirrored BHK-21 cell infection. Thus, the mutant viruses had no obvious defect in epithelial infection. They did show a defect in the infection of the NS0 myeloma cell line (Fig.  5C). However, in vitro B cell infection by MuHV-4 is poorly efficient, yielding here only 0.4% eGFP ϩ cells at 100 fibroblast eGFP units/cell. Thus, although the significance of the gB FCS Ϫ defect was hard to judge, it nonetheless indicates that cleavage of gB may have some role in B cell infection.
gB FCS Ϫ viruses showed a marked infection defect for RAW 264.7 monocytes/macrophages (Fig. 5D). This was confirmed for ex vivo peritoneal macrophages, which are infected better than RAW 264.7 cells (Fig. 5E), and for bone marrow-derived DCs, which are infected somewhat less well (Fig. 5F). Thus, gB FCS Ϫ MuHV-4 appears to have a general defect in myeloid infection.
The main barrier to myeloid cell infection by MuHV-4 is poor binding, and infection is markedly increased by opsonization with viral glycoprotein-specific antibodies that allow binding to endocytic myeloid immunoglobulin Fc receptors (FcRs) (44). To test if the gB FCS Ϫ viruses show a similar deficit for entry by this route, we preincubated virions with a nonneutralizing MAb directed against gp150 (T1A1). This enhanced macrophage and DC infection, but a substantial, although somewhat less pronounced, gB FCS Ϫ infection defect remained (Fig. 5D to F).
Lack of gB cleavage does not affect cell binding and virion endocytosis. In Fig. 5, overnight incubation with virus allowed ample time for binding and endocytosis. To investigate more specifically whether a lack of gB cleavage affected these processes, The titers of the ⌬FCSv1 and ⌬FCSv3 mutants were marginally but significantly reduced compared to the gB FCS ϩ viruses at 12 (P Ͻ 10 Ϫ5 by Student's t test), 24 (P Ͻ 10 Ϫ4 ), 36 (P Ͻ 10 Ϫ4 ), and 48 (P Ͻ 0.001) h p.i. The titers of the ⌬FCSv2 mutant were more substantially reduced compared to the gB FCS ϩ viruses at 6 (P Ͻ 0.001), 12 (P Ͻ 10 Ϫ10 ), 24 (P Ͻ 10 Ϫ6 ), 36 (P Ͻ 10 Ϫ10 ), and 48 (P Ͻ 0.05) h p.i. The titers of the ⌬FCSv2 mutant were also significantly lower than those of the ⌬FCSv1 and ⌬FCSv3 mutants at 6 (P Ͻ 0.002), 12 (P Ͻ 10 Ϫ6 ), 24 (P Ͻ 10 Ϫ4 ), and 36 (P Ͻ 10 Ϫ6 ) h p.i. Equivalent data were obtained in a repeat experiment. BHK-21 fibroblasts (Fig. 6A), NMuMG epithelial cells (Fig. 6B), and RAW 264.7 monocytes/macrophages (Fig. 6C) were exposed for increasing times to eGFP ϩ WT, ⌬FCSv1 and ⌬FCSv3 mutant, or ⌬FCSv1 revertant viruses before being washed with PBS to remove unbound virions or with phosphate-citrate buffer (pH 3) (acid wash) to inactivate all extracellular virions. After washing, all the cells were cultured overnight to allow eGFP expression and analyzed by flow cytometry.
In BHK-21 (Fig. 6A) and NMuMG cells (Fig. 6B), the entry kinetics looked similar for the ⌬FCSv1 and ⌬FCSv3 mutants and the WT and ⌬FCSv1 revertant viruses. However, when the exposure times required for half-maximal infection levels (t 50% ) were compared between the gB FCS Ϫ (⌬FCSv1 and ⌬FCSv3) and gB FCS ϩ (WT and ⌬FCSv1 R) viruses, the mean values were consistently moderately higher for the gB FCS Ϫ than for the gB FCS ϩ viruses. However, in BHK-21 cells the difference was statistically significant only after acid wash, and in NMuMG cells, it was significant only after PBS wash. We therefore concluded that the lack The same appeared to be true also for RAW 264.7 monocyte/ macrophage infections: the relative deficits of the gB FCS Ϫ viruses compared to the gB FCS ϩ viruses were similar irrespective of how long the cells were exposed to virus before the PBS or acid wash (Fig. 6C). This was confirmed by the observation that the t 50% values were not significantly different between the gB FCS Ϫ and For PBS wash, the difference in t 50% values was statistically significant (P Ͻ 0.02 by Student's t test). Equivalent data were obtained in a repeat experiment. (C) RAW 264.7 monocytes/macrophages were exposed at 37°C to eGFP ϩ WT, ⌬FCSv1 and ⌬FCSv3 mutant, and ⌬FCSv1 revertant viruses at an MOI of 20 eGFP units/cell in the presence of 100 g/ml PAA for the times indicated on the x axis. The cells were then washed with PBS or phosphate-citrate buffer (pH 3) (acid wash), followed by incubation at 37°C in the presence of 100 g/ml PAA. At 24 h p.i., the cells were treated with 1 g/ml lipopolysaccharide for 6 h to activate viral eGFP expression. The proportion of eGFP ϩ cells was then determined by flow cytometry. Each point shows the mean Ϯ standard error of the mean from 3 wells. The t 50% values were as follows (mean Ϯ standard error of the mean): PBS wash, 6.9 Ϯ 0.02 h for gB FCS ϩ and 6.5 Ϯ 0.28 h for gB FCS Ϫ viruses; acid wash, 8.3 Ϯ 0.47 h for gB FCS ϩ and 8.3 Ϯ 0.5 h for gB FCS Ϫ viruses. Equivalent data were obtained in a repeat experiment. FCS ϩ viruses both after PBS wash and after acid wash. Thus, PBS or acid wash seems not to accentuate the deficit of the gB FCS Ϫ viruses, suggesting that they bind normally to cells and are normally endocytosed from the cell surface. Their myeloid infection defect therefore seems to occur at a postendocytic entry step. As PRV and EBV gB cleavages are important for cell-to-cell fusion (11,15), it seemed most likely that mutant MuHV-4 lacking gB cleavage was impaired in viral membrane fusion and thus in escape from the endosomes. We have previously studied these processes extensively in epithelial cells by immunofluorescence (26,36,45). However, the relative inefficiency of myeloid cell infection prevented equivalent analysis giving clear answers (data not shown). Thus, we could conclude only that gB FCS mutation impaired infection after virion binding and endocytosis.
gB FCS disruption alters the virion neutralization profile. The MuHV-4 gB associates with gH/gL (46), and a loss of gL increases virion susceptibility to gB-directed neutralization (47). To test whether gB FCS disruption might also affect virion neutralization, we incubated eGFP ϩ WT, ⌬FCSv1 and ⌬FCSv3 mutant, and ⌬FCSv1 revertant viruses with increasing concentrations of the gB-specific neutralizing MAbs SC-9E8 and SC-9A5 (26), then added them to BHK-21 cells, and, after overnight incubation in the presence of PAA, enumerated eGFP ϩ cells by flow cytometry. Incubation with the gH/gL-specific neutralizing MAb 8F10 was included as a control. Figure 7 shows that the ⌬FCSv1 and ⌬FCSv3 mutant viruses were significantly more susceptible to neutralization by MAbs SC-9E8 and SC-9A5 than were WT and ⌬FCSv1 revertant viruses and significantly less susceptible to neutralization by MAb 8F10. This altered neutralization profile suggested a subtle change in the fusion complex, consistent with the idea that virion membrane fusion was impaired.
gB FCS ؊ viruses show reduced lytic virus spread in the lungs. Having established the effects of gB FCS mutation in vitro, we next assessed its consequences for host colonization. To this end, BALB/c mice were infected with BAC Ϫ WT, ⌬FCSv1 and ⌬FCSv3 mutant, and ⌬FCSv1 revertant viruses by intranasal inoculation. The viruses were constructed with a firefly luciferase reporter, so infection could be tracked by live imaging (20). After low-dose infection (10 PFU/mouse), luciferase signals from the thorax at days 5, 7, 10, and 14 postinfection (p.i.) were significantly lower for the gB FCS Ϫ (⌬FCSv1 and ⌬FCSv3) than for the gB FCS ϩ (WT and ⌬FCSv1 R) viruses ( Fig. 8A and B). Plaque assays at day 7 p.i. confirmed a significant replication defect of the gB FCS Ϫ viruses in lungs (Fig. 8C). While the virus loads in mediastinal lymph nodes (MLN) at day 28 p.i. were marginally higher for the gB FCS Ϫ than for the gB FCS ϩ viruses (Fig. 8D), the latency levels in spleens were not significantly different between gB FCS Ϫ and gB FCS ϩ viruses (Fig. 8E). After high-dose infection (5,000 PFU/ mouse), the gB FCS Ϫ viruses showed a moderate, but again significant, lytic replication defect in lungs at days 7 and 10 p.i. (Fig.  8F and G). While the virus loads in MLNs on day 28 p.i. were not significantly different between gB FCS Ϫ and gB FCS ϩ viruses (Fig.  8H), the latency levels in spleens were subtly but significantly lower for the gB FCS Ϫ than for the gB FCS ϩ viruses (Fig. 8I). However, because the gB FCS Ϫ viruses did not show any latency deficit upon low-dose infection (Fig. 8E), we concluded that they were not obviously impaired in establishment of normal longterm latency levels. In contrast, a lack of gB cleavage resulted in significantly reduced lytic replication in the lower respiratory tract.
gB cleavage is important for infection of alveolar macrophages. To understand better the in vivo infection defect, we analyzed lungs by immunostaining 1 day after virus inoculation ( Fig.  9). We hypothesized that while later time points might show quantitative defects, reasons for such defects might be more apparent early in infection. To detect both lytic and latent infection, we stained for virus-expressed eGFP. MuHV-4 antigens have previously been identified in alveolar epithelial cells and large mononuclear cells 5 days after intranasal inoculation (21). We similarly detected virus-expressed eGFP in 2 main cell populations: single, round mononuclear cells and much larger, elongated cells-or clusters of cells-that lined the alveoli. The mononuclear cells were CD68 ϩ , F4/80 ϩ , and CD11c ϩ ( Fig. 9 and data not shown), consistent with them being alveolar macrophages (48,49), while the cells lining the alveoli stained positive for podoplanin (Fig. 9), consistent with them being alveolar type I epithelial cells (50). The  numbers of infected alveolar epithelial cells were not significantly different between gB FCS ϩ (WT and ⌬FCSv1 revertant) and gB FCS Ϫ (⌬FCSv1 and ⌬FCSv3) viruses, but gB FCS Ϫ viruses infected significantly fewer alveolar macrophages (Table 1). Thus, gB cleavage was important for myeloid cell infection both in vitro and in vivo.
gB FCS ؊ viruses show subtly reduced transport to lymphoid organs. Given the role of myeloid cells in trafficking virus from the site of lytic replication to lymphoid organs (19,41), impaired myeloid infection might be expected to result in an impaired colonization of the latency reservoir. The finding that the gB FCS Ϫ viruses establish normal long-term latent loads does not preclude a trafficking defect, because long-term latent loads are relatively insensitive to variation in primary lytic infection and the amount of virus seeded to the draining lymph nodes (51,52). We therefore aimed to find out if the gB FCS Ϫ viruses have a deficit in reaching lymphoid organs upon intranasal and intraperitoneal infection.
The observation that the gB FCS Ϫ viruses have a lytic replication deficit in the lung makes analysis of virus transport from the lung to the draining MLN difficult to interpret. In contrast, the gB FCS Ϫ viruses show normal levels of lytic replication in the nose upon high-dose intranasal infection (10 4 PFU/mouse) (Fig. 10A), rendering this site more suitable for investigating virus transport to the draining lymph nodes. While on day 9 p.i. the luciferase signals from the draining superficial cervical lymph nodes (SCLN) were slightly but significantly, lower for the gB FCS Ϫ viruses than for the gB FCS ϩ viruses (Fig. 10B), no such difference was observed on day 13 p.i. (Fig. 10C). Similarly, the signals from the spleens on day 13 p.i. were not significantly different between gB FCS Ϫ and gB FCS ϩ viruses (Fig. 10D). Second, we investigated the colonization of the spleen upon intraperitoneal infection, a route by which incoming virus can reach the spleen directly (53). As shown in Fig. 10E and F, the splenic loads of gB FCS Ϫ viruses were significantly lower than those of gB FCS ϩ viruses as assayed by infectious center assay and quantitative PCR (qPCR) at day 3 p.i. Together, these findings suggest that the gB FCS Ϫ viruses have a subtle deficit in reaching lymphoid organs upon intranasal and intraperitoneal infection.

DISCUSSION
We investigated the role of MuHV-4 gB cleavage in host colonization by mutating its FCS. Each of three independent mutations prevented gB cleavage without affecting its expression levels, glycosylation, or antigenic conformation. The gB FCS Ϫ viruses showed normal entry in BHK-21 fibroblasts and NMuMG epithelial cells but significantly reduced entry in RAW 264.7 monocytes/macrophages, ex vivo peritoneal macrophages, and bone marrow-derived DCs. The myeloid infection defect appeared to be postendocytic and was presumably in membrane fusion. After intranasal inoculation of mice, gB FCS Ϫ viruses infected significantly fewer alveolar macrophages,  While the inefficiency of myeloid cell infection precluded a direct analysis of fusion kinetics by immunofluorescence, gB FCS Ϫ viruses showed an increased susceptibility to gB-specific neutralizing antibodies blocking viral membrane fusion, consistent with a change in the fusion complex that might compromise its function. We have identified 3 gB antigenic forms: extracellular virions exclusively show one and virions after capsid and tegument release exclusively show another; these MAb-defined forms presumably correspond to pre-and postfusion gB conformations (33). MuHV-4 fuses only when it reaches late endosomes, and postendocytic, prefusion virions show a third, intermediate antigenic form of gB, which we interpret, based on analyses of the VSV glycoprotein G (6, 7), as a dynamic equilibrium between the preand postfusion conformations, before membrane interactions make it irreversibly postfusion (36). Normally, the MuHV-4 virion gB is bound to gH/gL (46). When gB changes from its prefusion to its intermediate form after virion endocytosis, gH/gL epitopes are also lost (36), and the gB of genetically gL Ϫ virions more readily adopts the intermediate form (54), suggesting that the gB prefusion conformation is stabilized by association with gH/gL (and vice versa). gL Ϫ virions also show greater susceptibility to gB-directed neutralization (47). Thus, although it was not possible to track gB conformation changes in myeloid cells, the altered neutralization profile of gB FCS Ϫ mutants suggested that the normal extracellular interaction of gB with gH/gL was compromised, causing a subtle impairment of membrane fusion.
The infection defect of gB FCS Ϫ MuHV-4 was consistent with that observed for similar mutants of other herpesviruses. However, cell-type-specific defects have not previously been noted. Why the requirement of MuHV-4 for gB cleavage differed between fibroblast/epithelial and myeloid infections remains unclear. It is possible that the relatively low infectivity of free MuHV-4 virions for myeloid cells-up to 100-fold less than its infectivity for fibroblast/epithelial cells-increased the impact of poor membrane fusion. This would be consistent with enhanced myeloid cell infection by bound IgG reducing the relative deficit of gB FCS Ϫ viruses. Alternatively, qualitative differences that impinge on gB function may exist between fibroblast/epithelial and myeloid cell endosomes.
A significant advance here on previous data was the identification of an in vivo host colonization phenotype for gB FCS Ϫ virus mutants. After intranasal inoculation, they showed significantly less lung infection than did WT virus at both low (10-PFU) and high (5,000-PFU) doses, with the defect being more pronounced at a low dose. There was reduced infection of alveolar macrophages, a major early target of MuHV-4 in the lung, with apparently normal alveolar epithelial cell infection. Thus, myeloid infection was selectively impaired both in vitro and in vivo by a lack of gB cleavage. Consistent with myeloid cells being important for trafficking virus from the site of lytic replication to the lymphoid The signals in the SCLN on day 9 p.i. (B) were subtly, but significantly, lower for the gB FCS Ϫ viruses (⌬FCSv1 and ⌬FCSv3) than for the gB FCS ϩ viruses (WT and ⌬FCSv1 R) (P Ͻ 0.03 by Student's t test). (E and F) Intraperitoneal infection. BALB/c mice were infected intraperitoneally with 10 5 eGFP units/mouse of eGFP ϩ WT, ⌬FCSv1 and ⌬FCSv3 mutant, and ⌬FCSv1 revertant viruses. On day 3 p.i., virus loads in spleens were determined by infectious center assay (E) and quantitative PCR (F). The splenic loads of the gB FCS Ϫ viruses were significantly lower than those of the gB FCS ϩ viruses by infectious center assay (P Ͻ 10 Ϫ7 by Student's t test) and quantitative PCR (P Ͻ 0.008).
organs (19,41), the gB FCS Ϫ viruses showed a subtle deficit in reaching lymphoid organs early in infection. However, from day 13 p.i. onwards, the virus loads in lymph nodes and spleens were similar for gB FCS Ϫ and FCS ϩ viruses, suggesting that the onset of latency amplification leads to an equilibration of virus loads. The main in vivo manifestation of a lack of gB cleavage therefore seems to be reduced infection of alveolar macrophages and reduced lytic virus spread in the lung, the former presumably being responsible for the latter. As such, our in vivo studies demonstrated a subtle and quite specific role for gB cleavage in host colonization.