Dual impacts of a glycan shield on the envelope glycoprotein B of HSV-1: evasion from human antibodies in vivo and neurovirulence

ABSTRACT Identification of the mechanisms of viral evasion from human antibodies is crucial both for understanding viral pathogenesis and for designing effective vaccines. Here we show in cell cultures that an N-glycan shield on the herpes simplex virus 1 (HSV-1) envelope glycoprotein B (gB) mediated evasion from neutralization and antibody-dependent cellular cytotoxicity due to pooled γ-globulins derived from human blood. We also demonstrated that the presence of human γ-globulins in mice and immunity to HSV-1 induced by viral infection in mice significantly reduced replication in their eyes of a mutant virus lacking the glycosylation site but had little effect on the replication of its repaired virus. These results suggest that an N-glycan shield on a specific site of HSV-1 envelope gB mediated evasion from human antibodies in vivo and from HSV-1 immunity induced by viral infection in vivo. Notably, we also found that an N-glycan shield on a specific site of HSV-1 gB was significant for HSV-1 neurovirulence and replication in the central nervous system of naïve mice. Thus, we have identified a critical N-glycan shield on HSV-1 gB that has dual impacts, namely evasion from human antibodies in vivo and viral neurovirulence. IMPORTANCE Herpes simplex virus 1 (HSV-1) establishes lifelong latent and recurrent infections in humans. To produce recurrent infections that contribute to transmission of the virus to new human host(s), the virus must be able to evade the antibodies persisting in latently infected individuals. Here, we show that an N-glycan shield on the specific site of the envelope glycoprotein B (gB) of HSV-1 mediates evasion from pooled γ-globulins derived from human blood both in cell cultures and mice. Notably, the N-glycan shield on the specific site of gB was also significant for HSV-1 neurovirulence in naïve mice. Considering the clinical features of HSV-1 infection, these results suggest that the glycan shield not only facilitates recurrent HSV-1 infections in latently infected humans by evading antibodies but is also important for HSV-1 pathogenesis during the initial infection.

probably impeded the development of effective vaccines for HSV-1 and HSV-2 infections. Several decades of vaccine development have not produced a successful vaccine (4,5).
To clarify the significance of the mechanisms of immune evasion by viruses that cause diseases in humans, the mechanisms should be investigated not only in vitro but also in vivo; and research using available human samples should provide valuable information on the effective viral mechanisms in humans. However, in previous studies, human samples have generally been analyzed in vitro, and information from the in vivo evaluations of human samples on viral evasion from the immune system has been limited. To fill the gaps in our understanding of the effective mechanisms of viral immune evasion in humans, in vivo investigations of the mechanisms of immune evasion that use human samples are of crucial importance.
Although no effective vaccines for HSV-1 and HSV-2 have been developed thus far, previous clinical trials for HSV vaccines have provided important clues indicating that not only T-cell responses, but also antibody responses were important for controlling HSV infections in humans (4,5). Thus, a clinical trial of a subunit vaccine employing HSV-2 envelope glycoprotein D (gD) showed partial but consistent efficacy against the development of HSV-1 genital disease but did not offer significant protection against HSV-2 genital disease (6). Notably, antibody responses to HSV-2 gD correlated with protection against HSV-1 but not HSV-2 infections, whereas CD4 + T-cell responses did not correlate with protection against either HSV-1 or HSV-2 infections (7). In addition, a substudy of this trial, which used sera from a fraction of the vaccinated subjects, showed that neutralizing antibody titers against HSV-1 were significantly higher than the titers against HSV-2 (8). These findings were in agreement with those from another clinical study in humans that showed that the absence of HSV antibodies was associated with severe HSV infections in humans (9). It is of interest that the response to the HSV-2 gD subunit vaccine was restricted to the generation of gD neutralizing antibodies and that antibody-dependent cellular cytotoxicity (ADCC) did not develop (10).
The findings described in the previous paragraph suggesting that antibodies are important for the control of HSV-1 infections led us to attempt to identify hitherto unknown mechanisms of HSV-1 evasion from human antibodies. In this study, we focused on the glycosylation of a major envelope glycoprotein of HSV-1, gB. Glycosyla tion of a viral envelope glycoprotein sometimes acts as a glycan shield(s) that evades antibodies (11). This mechanism of antibody evasion has been documented for influenza virus, human immunodeficiency virus, Nipah virus, hepatitis C virus, Ebola virus, hepatitis B virus, lymphocytic choriomeningitis virus, and porcine reproductive and respiratory syndrome virus (12)(13)(14)(15)(16)(17)(18)(19)(20).
HSV-1 gB, which is a class III fusion glycoprotein, is a major target of antibody-medi ated immunity (21). It plays an essential role in the entry of the virus into a host cell, together with other HSV-1 envelope glycoproteins, including gD and a complex of gH and gL (gH/gL) (22). Herein, we investigated the effects of a series of N-linked glycans (N-glycans) on HSV-1 gB in the context of viral infection and identified an N-glycan that contributed to evasion from human antibodies not only in vitro but also in vivo. Notably, the N-glycan on gB was also significant for HSV-1 replication in the central nervous system (CNS) of naïve mice as well as neurovirulence, although it had no effect on viral replication in cell cultures.

Generation of recombinant viruses harboring a mutation in each of the potential N-glycosylation sites on HSV-1 gB by an improved genetic manipu lation system for HSV-1
viruses and their repaired viruses (Fig. S1). The recombinant viruses encoded mutant gBs (gB-N87Q, gB-N141Q, gB-N398Q, gB-N430Q, gB-N489Q, and gB-N674Q), in which each of the potential N-glycosylation sites was substituted with glutamine. In addition, we generated a pair of control viruses, a recombinant virus, in which Asn-888 in the cytoplasmic domain of gB was substituted with glutamine (gB-N888Q), and its repaired virus (gB-N888Q-repair) (Fig. S1).
The two-step Red-mediated recombination system consists of the first recombination for the insertion of a PCRamplified selectable marker and the second recombination for the excision of the inserted marker by a cleavage step that uses a rare-cutting endonuclease I-SceI. This system is widely used for markerless modifications of large DNA molecules such as herpesvirus genomes that are cloned into a bacterial artificial chromosome (BAC) in Escherichia coli (24).

Research Article mBio
However, the second recombination step was not very efficient in our hands. Therefore replica plating, which is time-consuming and laborious, was needed to identify any E. coli recombinant harboring a BAC clone with a desired mutation. We used an improved system that was developed for this study that employed a negative selection marker, the E. coli phenylalanyl-tRNA synthetase (ePheS*), which encodes a mutant of the α-subunit of E. coli phenylalanyl-tRNA synthetase (ePheS) (25) in the presence of 4-chloro-phenylalanine (4-CP), in addition to cleavage by I-SceI for the second recom bination step (Fig. S2A). The improved system considerably increased the efficiency of the second recombination step from 17.4% to 87%, 17.4% to 94.7%, and 8.7% to 87.0% in the substitution of single amino acids, and in the deletion and insertion of a foreign gene, respectively, compared with the original system (Table S1A). Recombi nant viruses carrying an alanine substitution of Thr-190 in the HSV-1 protein UL51 (Fig.  S3A), which were generated in the original and improved systems, exhibited identical growth properties (Fig. S3B) in Vero cells and produced identical neurovirulence in mice following intracranial inoculation (Fig. S3C), suggesting that, compared with the original system, the additional negative selection in the improved system did not affect the genomic integrity of HSV-1 other than the desired mutations.

Effects of mutations at each of the potential gB N-glycosylation sites on electrophoretic mobility in the presence or absence of peptide-N-glycosidase F (PNGase)
Vero cells infected with wild-type HSV-1(F), each of the gB mutant viruses, or each of their repaired viruses were lysed, treated with or without PNGase, and analyzed by immunoblotting. As shown in Fig. 2, all of the gB mutants, except the gB-N888Q mutant, migrated faster than wild-type gB in denaturing gels. In contrast, the gB-N888Q mutant and gB from cells infected with each of the repaired viruses migrated as slowly as the wild-type gB in denaturing gels (Fig. 2). After treatment of the infected cell lysates with PNGase, all of the gB mutants migrated in denaturing gels as slowly as the wild-type gB (Fig. 2). All of the gB mutants were detected by immunoblotting at levels similar to the level of wild-type gB (Fig. 2). These results indicate that gB was N-glycosylated at each of the six potential N-linked glycosylation sites without affecting its accumulation in HSV-1-infected cells.

Effects of mutations at each of the gB N-glycosylation sites on the replication of HSV-1 in cell cultures
To investigate the effect of gB N-glycosylation on HSV-1 replication in cell cultures, Vero cells were infected with wild-type HSV-1(F), each of the gB mutant viruses, or each of their repaired viruses at a multiplicity of infection (MOI) of 5 or 0.01; and virus titers were assayed at 24 or 48 h postinfection. As shown in Fig. S4, progeny virus yields in cells infected with each of the gB mutant viruses were similar to those in cells infected with wild-type HSV-1(F) or each of their repaired viruses. These results suggest that N-glycosylation on gB did not affect HSV-1 replication in cell cultures.

Effects of mutations at each of the gB N-glycosylation sites on viral suscepti bility to neutralization by human antibodies
An estimated 67% of the global human population is infected with HSV-1 (26). Therefore, we decided to use pooled γ-globulins from human blood that contain amounts of antibodies to HSV-1 sufficient for our experiments (27). Indeed, we showed that gB antibodies in pooled human γ-globulins at a concentration of 1.3 mg/mL could still be detected at a dilution of 1:1024, whereas gD antibodies could be detected at a dilution of 1:16384 (Fig. S5). To investigate the effect of gB N-glycosylation on viral susceptibility to neutralization by human antibodies, the sensitivity to neutralization of wild-type HSV-1(F) by pooled human γ-globulins was compared to each of the recombinant gB mutant viruses. Among the gB mutant viruses tested, gB-N141Q was only the gB mutant virus that was significantly more susceptible to neutralization by pooled human γ-globulins at a concentration (dilution) of 0.041 mg/mL (1:32) compared with wild-type HSV-1(F) (Fig. S6). Therefore, we focused on N-glycosylation at gB Asn-141 and further characterized gB-N141Q in detail.
As shown in Fig. 3A and B, the growth kinetics of gB-N141Q in Vero cells at MOIs of 5 and 0.01 were almost identical to those of wild-type HSV-1(F) or gB-N141Q-repair. Similarly, gB-N141Q produced plaques in Vero cells of similar size to the sizes of plaques produced by wild-type HSV-1(F) or gB-N141Q-repair (Fig. 3C). Confocal microscopy showed that the subcellular locations of gB in cells infected with either gB-N141Q, wild-type HSV-1(F), or gB-N141Q-repair were also similar (Fig. 3D). Furthermore, Vero cells infected with gB-N141Q accumulated other envelope glycoproteins, including gD, Research Article mBio gC, gH, and gL, at levels similar to the levels in cells infected with wild-type HSV-1(F) (Fig. 3E). We observed that the amounts of gB-N141Q in the mutant virions were slightly lower than the amounts of wild-type gB in the wild-type and gB-N141Q-repair virions (Fig. 3F). When incubated in dilutions of pooled human γ-globulin ranging from 0.010 (1:128) to 0.041 mg/mL (1:32), gB-N141Q proved to be significantly more susceptible to neutralization by human γ-globulins than wild-type HSV-1(F) (Fig. 4A). Wild-type susceptibility was restored in the gB-N141Q-repair virus (Fig. 4A). Similarly, gB-N141Q of wild-type HSV-1(F), gB-N141Q, or gB-N141Q-repair was incubated with serially diluted human γ-globulins at 37°C for 1 h and then inoculated onto Vero cell monolayers for plaque assays. The percentage of neutralization was calculated from the number of plaques formed by each of the viruses that were incubated with or without human γ-globulins, as follows: 100 × [1 − (number of plaques after incubation with human γ-globulins)/(number of plaques after incubation without human γ-globulins)]. (B) 100 PFU of wild-type HSV-1(F), gB-N141Q, or gB-N141Q-repair was incubated with 1:500 diluted human serum or heat-inactivated human serum at 37℃ for 1 h and then inoculated onto Vero cell monolayers as described in A.
(C) Human γ-globulins (0.082 mg/mL) were mock depleted or depleted with gB-SE (α-gB depleted) or gD-SE (α-gD depleted) as shown in Fig. S7. Wild-type HSV-1(F), gB-N141Q, or gB-N141Q-repair was incubated with each of the depleted human γ-globulins and then inoculated onto Vero cell monolayers as described in A. Each value represents the mean ± standard error of the results of three (A and C) or nine (B) independent experiments. Statistical analysis was performed by two-way analysis of variance (ANOVA) followed by the Tukey test (A and C) or by one-way ANOVA followed by the Tukey test (B). *, P < 0.05 indicates statistically significant differences between gB-N141Q and wild-type HSV-1(F) or gB N141Q-repair; n.s., not significant; α, anti.
Research Article mBio was significantly more susceptible to neutralization by heat-inactivated pooled human sera than wild-type HSV-1(F) and gB-N141Q-repair (Fig. 4B). Heat inactivation of the pooled human sera had little effect on the degree of increased susceptibility to neutralization by the N141Q mutation in gB (Fig. 4B). Antibodies to gB or gD in the pooled human γ-globulins [0.082 mg/mL (1:16)] were then depleted by treatment of the pooled human γ-globulins with purified gB or gD fused with Strep-tag at the C-terminus (gB-SE or gD-SE, respectively) ( Fig. S7A and B). The anti-gB antibody-depleted human γ-globulins could not detect gB ectopically expressed by HEK293FT cells (Fig. S7C and D). As shown in Fig. 4C, the susceptibility of gB-N141Q to neutralization by anti-gB antibody-depleted human γ-globulins (~0.4 mg/mL) was comparable to that of wild-type HSV-1(F) and gB-N141Q-repair. In contrast, gB-N141Q was significantly more susceptible to neutralization by mock-depleted or anti-gD antibody-depleted human γ-globulins than wild-type HSV-1(F) and gB-N141Q-repair (Fig. 4C), as seen in Fig. 4A, showing its susceptibility to human γ-globulins without antibody depletion. These results suggest that the N-glycan on gB Asn-141 was required for efficient HSV-1 evasion from comple ment-independent neutralization by human antibodies that targeted gB in cell cultures.

Effects of the N-glycan on gB Asn-141 on human ADCC
It has been reported that gB on the surface of infected cells mediates ADCC (28); therefore, we examined the effect of the N-glycan on gB Asn-141 on ADCC induced by human γ-globulins. Vero cells infected with wild-type HSV-1(F), gB-N141Q, or gB-N141Qrepair at an MOI of 1 for 24 h were subjected to an activating Fcγ receptor (R) IIIA ADCC assay in the presence or absence of human γ-globulins. As shown in Fig. 5A, human γ-globulins at concentrations (dilutions) of 0.33 mg/mL (1:3.9) and 1.0 mg/mL (1:1.3) induced significantly higher FcγRIIIA activation in cells infected with gB-N141Q than in cells infected with wild-type HSV-1(F) or gB-N141Q-repair. The gB and gD antibodies present in samples of human γ-globulins at a concentration of 3.04 mg/mL were then depleted by treatment with purified gB-SE or gD-SE (Fig. S7). The anti-gB antibody-depleted human γ-globulins could barely detect gB ectopically expressed by HEK293FT cells (Fig. S7E). In this case, the anti-gB antibody-depleted human γ-globulins (~1.0 mg/mL) induced slightly increased FcγRIIIA-mediated activation in cells infected with gB-N141Q than in cells infected with wild-type HSV-1(F) or gB-N141Q-repair ( Fig.  5B), because we used the depleted human γ-globulins at a much higher concentration than the concentration of the depleted human γ-globulins used in the neutralizing assay described in the previous section. However, the degrees of differences between the levels of FcγRIIIA-mediated activation in cells infected with gB-N141Q and the levels of FcγRIIIA-mediated activation in cells infected with wild-type HSV-1(F) or gB-N141Q-repair in cultures containing anti-gB antibody-depleted human γ-globulins were lower than the degrees of differences between the levels of FcγRIIIA-mediated activation in cells infected with those viruses in cultures containing mock-depleted or anti-gD antibodydepleted human γ-globulins (Fig. 5B).
To eliminate the possibility that the higher level of FcγRIIIA-mediated activation in gB-N141Q-infected cells was due to increased expression of mutated gB on the surface of the infected cells, we investigated the effect of the N-glycan at gB Asn-141 on the expression of gB in the infected cells. Vero cells were infected with wild-type HSV-1(F), gB-N141Q, gB-N141Q-repair, ΔgB, or ΔgB-repair as described in the experi ments reported in the previous paragraph and depicted in Fig. 5 and used flow cytometry to show the level of gB expression on the surface of infected cells or the total accumulation of gB in the infected cells. As shown in Fig. 6, the levels of expression of gB on the surface of cells infected with gB-N141Q were significantly lower than those levels on cells infected with wild-type HSV-1(F) or gB-N141Q-repair. In contrast, the total level of gB in cells infected with gB-N141Q was similar to the total levels in cells infected with wild-type HSV-1(F) or gB-N141Q-repair (Fig. 6). These results suggest that the N-glycan at gB Asn-141 was required for the efficient expression of gB on the surface of HSV-1-infected cells. Thus, although cells infected with gB-N141Q expressed lower levels of mutated gB on their surface membranes than the levels of gB expressed by cells infected with wild-type HSV-1(F) or gB-N141Q-repair, human γ-globulins resulted in increased FcγRIIIA-mediated activation of cells infected with gB-N141Q than seen for cells infected with wild-type HSV-1(F) or gB-N141Q-repair. These results eliminated the possibility that the higher level of FcγRIIIA-mediated activation in gB-N141Q-infected cells was due to the increased expression of mutated gB on the surface of the infected cells. Altogether, the results suggest that the N-glycan at gB Asn-141 was required for the efficient evasion of ADCC induced by human antibodies to gB in infected cell cultures. (A) Vero cells were infected with wild-type HSV-1(F), gB-N141Q, or gB-N141Q-repair at an MOI of 1 for 24 h and co-cultured with ADCC effector cells in the presence or absence of serially diluted human γ-globulins for 6 h. A luciferase assay was then performed. (B) Human γ-globulins (3.04 mg/mL) were mock depleted or depleted with gB-SE (α-gB depleted) or gD-SE (α-gD depleted) as depicted in Fig. S7 and used as described in 4A. Values are fold induction relative to controls without antibody. Each value is the mean ± standard error of the results of three biologically independent samples. The statistical analysis was performed by two-way ANOVA followed by the Tukey test. *, P < 0.05 indicates statistically significant differences between gB-N141Q and wild-type HSV-1(F) or gB-N141Q-repair; n.s., not significant; α, anti; ANOVA, analysis of variance.

Effects of the N-glycan on gB Asn-141 on the replication of HSV-1 in the eyes of mice in the presence of human antibodies
To examine the effects of human antibodies on HSV-1 replication in vivo in the pres ence or absence of the N-glycan at gB Asn-141, mice were mock injected or injected intraperitoneally with pooled human γ-globulins and were then ocularly infected with gB-N141Q or gB-N141Q-repair 1 day after injection (Fig. 7A). Samples of tear films were collected at the indicated times (Fig. 7A), and viral titers in the tear films were measured. As shown in Fig. 7B, the presence of human γ-globulins did not affect the viral titers of the tear films in mice infected with gB-N141Q-repair at 1, 3, and 5 days postinfection. In contrast, the presence of human γ-globulins significantly reduced viral titers of the tear films of mice infected with gB-N141Q at 1 and 5 days postinfection (Fig. 7B). Thus, the ratios of gB-N141Q titers in the absence of human γ-globulins to those in the presence of human γ-globulins were higher than the ratios of gB-N141Q-repair titers in the absence of human γ-globulins to those in the presence of the human γ-globulins (Fig. 7C).
Furthermore, viral titers of the tear films of mice infected with gB-N141Q in the presence of human γ-globulins at 1, 3, and 5 days postinfection were significantly lower than the titers of the tear films of mice infected with gB-N141Q-repair (Fig. 7B). In contrast, the viral titers of the tear films of mice infected with gB-N141Q in the absence of human γ-globulins at 1, 3, and 5 days postinfection were comparable to those of the tear films of mice infected with gB-N141Q-repair; although, as the infection progressed, the viral titers of the tear films of mice infected with gB-N141Q in the absence of human γ-globulins tended to be lower than those titers in mice infected with gB-N141Q-repair. These results indicate that the presence of human antibodies inhibited the replication of gB-N141Q in the peripheral organs of mice more efficiently than it inhibited the replication of Research Article mBio gB-N141Q-repair and also suggest that the N-glycan at gB Asn-141 was required for the efficient evasion of HSV-1 from human antibodies in vivo. It has been reported that human IgG subclasses could bind to murine Fcγ Rs and efficiently activate murine cells to enable ADCC (29,30), suggesting that the effects of pooled human γ-globulins in the experiments described in this section might result from both neutralization and ADCC.

Effects of the N-glycan on gB Asn-141 on the replication of HSV-1 in the eyes of mice immunized against HSV-1
To examine the effects of the N-glycan at gB Asn-141 on HSV-1 replication in vivo in the presence of physiologically induced immunity against HSV-1, mice were subcutaneously mock immunized or immunized with wild-type HSV-1(F). At 9 weeks after inoculation, the immunized mice were ocularly infected with gB-N141Q or gB-N141Q-repair (Fig.  8A). Samples of tear films were collected at the indicated times, and viral titers of the Research Article mBio tear films were determined (Fig. 8B). As shown in Fig. 8B, whereas the viral titers of the tear films of immunized mice infected with gB-N141Q at 1 and 2 days postinfection were comparable to those in mock-immunized mice, the viral titers of the tear films of immunized mice infected with gB-N141Q at 3 days postinfection were significantly lower than those in the mock-immunized mice. In contrast, the viral titers of the tear films of immunized mice infected with gB-N141Q-repair at 1, 2, and 3 days postinfection were comparable to those in mock-immunized mice infected with gB-N141Q-repair (Fig. 8B). Thus, the ratios of gB-N141Q titers in mock-immunized mice to those in immunized mice at 3 days postinfection were higher than the ratios of gB-N141Q-repair titers in mockimmunized mice to those in immunized mice (Fig. 8C). Furthermore, the viral titers of the tear films in mock-immunized or immunized mice infected with gB-N141Q at 1 and 2 days postinfection were comparable to those in mock-immunized or immunized mice infected with gB-N141Q-repair (Fig. 8C). In contrast, the viral titers of the tear films in immunized mice infected with gB-N141Q at 3 days postinfection were significantly lower Research Article mBio than those in immunized mice infected with gB-N141Q-repair, although viral titers of the tear films in mock-immunized mice infected with gB-N141Q at 3 days postinfection were comparable to those in mock-immunized mice infected with gB-N141Q-repair (Fig.  8B). These results indicate that the presence of immunity against HSV-1 in mice inhibited replication of gB-N141Q more efficiently than the replication of gB-N141Q-repair and suggest that the N-glycan at gB Asn-141 was required for the efficient evasion of HSV-1 from immunity induced in mice previously immunized against HSV-1.

Effects of the N-glycan at gB Asn-141 on HSV-1 neurovirulence and replica tion in the CNS of naïve mice
To investigate the effects of the N-glycan at gB Asn-141 on the neurovirulence and replication of HSV-1 in the CNS of naïve mice, mice were infected intracranially with gB-N141Q or gB-N141Q-repair, and the mortality rates of these injected mice were monitored for 14 days. As shown in Fig. 9A, the mortality rate of mice infected with gB-N141Q was significantly lower than the rate of mice infected with its repaired virus (gB-N141Q-repair). We also harvested the brains of mice infected with gB-N141Q or gB-N141Q-repair at 1, 3, and 5 days postinfection and measured viral titers in their brains. As shown in Fig. 9B, the viral titers in the brains of mice infected with gB-N141Q at 1 day postinfection were comparable to those of mice infected with gB-N141Q-repair.
In contrast, at later time points (3 and 5 days postinfection), viral titers in the brains of mice infected with gB-N141Q were significantly lower than the titers in the brains of mice infected with gB-N141Q-repair. These results suggest that the N-glycan at gB Asn-141 was required for efficient HSV-1 neurovirulence and replication in the CNS of naïve mice. The results also led us to investigate whether the N-glycan at gB Asn-141 acted specifically in neural cells. As shown in Fig. 9C, progeny virus yields in human neuroblastoma SK-N-SH cells infected with gB-N141Q were similar to those in the cells infected with wild-type HSV-1(F) or gB-N141Q-repair. These results further supported our observation from the results of in vitro experiments described previously that N-glycosy lation on gB does not appear to play a role in the replication of HSV-1 in cell cultures.

DISCUSSION
It is unquestionable that studies using human samples to analyze the mechanisms of infection utilized by human pathogenic viruses are crucial for understanding the mechanisms effective in humans. Considering that the gap between what can be observed from in vitro and in vivo viral infections is significant, evaluations of human samples in vivo should provide more valuable information on the mechanisms of infection than evaluations of the samples in vitro. However, there has been a lack of in vivo information because human samples that could be used for in vivo analyses as well as in vivo models that could represent the pathogenesis of viral infections in humans are limited.
In this study, we clarified a novel immune evasion mechanism used by HSV-1, namely, that an N-glycan shield at Asn 141 of the HSV-1 gB mediated evasion from the deleteri ous effects of human antibodies such as in vitro neutralization and ADCC. A molecular model of N-glycosylation at gB Asn-141, which is based on the prefusion structure of HSV-1 gB (23), predicts that N-glycosylation at gB Asn-141 masks 27 amino acids in the gB molecule (Fig. S8). Of these amino acids, 19 were mapped to the functional region (FR)2 and FR3 of gB (Fig. S8), both of which were previously defined based on the epitopes seen for a panel of neutralizing monoclonal antibodies (31,32). Notably, the N-glycosylation at gB Asn-141 was predicted to mask Asp-419, a residue critical for the binding of gB to the C226 antibody, which shows high neutralizing activity in preventing the association of gB with a complex of gH and gL and fusion (32). Furthermore, the amino acids predicted to be masked by N-glycosylation at Asn-141 included or were positioned near those (Pro-361, Asp-408, Asp-419, Asn-430, Asn-458, Arg-470, Pro-481, Ile-495, and Thr-497) previously shown to be critical for gB receptor-or gD receptormediated fusion (33-35), introducing the possibility that antibodies target these amino Each value is the mean ± standard error of the results of three independent experiments. Statistical analysis was performed by the log-rank test (A), the two-tailed Student t test (B), or one-way ANOVA followed by the Tukey test (C). n.s., not significant; ANOVA, analysis of variance.
Research Article mBio acid residues. These results suggest that the N-glycan at gB Asn-141 prevented the binding of antibodies to gB epitopes and further support our conclusion that the Nglycan at gB Asn-141 was required for the efficient evasion of HSV-1 from human antibodies. Similarly, HSV-1 envelope glycoproteins, including gC and a complex of gI and gE, have been reported to mediate evasion from human antibodies in vivo (36)(37)(38). The gI/gE complex acts as an Fc receptor that captures human IgG (39,40). The gC binds to the complement component C3b (41), inhibits complement activation that enhances antibody neutralization (42,43), and shields gB from complement-independent antibody neutralization (44). Thus, HSV appears to have evolved multiple mechanisms to mediate evasion from human antibodies, highlighting the importance of antibodies for control ling HSV infections in humans as described above.
Based on the structure of gB shown in Fig. 1A, although N-glycans at gB Asn-398, Asn-430, and Asn-489 appear to be adjacent to the N-glycan at gB Asn-141, only this specific N-glycan was shown to shield gB from human γ-globulins. Similar observations were reported in other enveloped viruses (15,16,45). We speculate that the N-glycan at gB Asn-141 might cover an area different from the area that the N-glycans at gB Asn-489, Asn-398, and Asn-430 cover; the orientation of the N-glycan at gB Asn-141 might vary or the N-glycan at gB Asn-141 might exhibit different flexibility or different macro-and microheterogeneity, making it unique in its ability to modulate the antigenic effects of gB.
One may argue that the difference between the sensitivities of wild-type HSV-1 and gB-N141Q to neutralization by human γ-globulins was due to slightly less incorporation of the mutant gB into virions. However, it seems unlikely, based on earlier studies of various viruses showing that decreased amounts of expression of a viral antigen by various virions are associated with decreased sensitivities to neutralization (46)(47)(48)(49)(50).
Our observations that the N-glycan shield on the specific site of HSV-1 gB seemed to significantly increase viral replication in the eyes of mice not only in the presence of HSV-1 immunity but also in the presence of human antibodies, support our predic tion from the clarified in vitro effects of the glycan shield on HSV-1 that the glycan shield would be effective in the presence of human antibodies in vivo and probably in human beings. We should note that the immune response to HSV-1 exhibited by mice immunized by wild-type HSV-1 in this study included not only antibody responses but also a T-cell-mediated immune response and innate immune responses. Interestingly, we also found that the N-glycan shield at a specific site on HSV-1 gB was required for viral neurovirulence and efficient replication in the CNS of naïve mice. Thus, we have identified an important glycan shield on the HSV-1 gB that appears to have two affects: in vivo evasion from human antibodies and neurovirulence in naïve hosts. Considering the clinical features of HSV-1 infection, these results suggest that the glycan shield not only facilitates recurrent HSV-1 infections in latently infected humans by evading antibodies but also is important for HSV-1 pathogenesis during the initial infection.
At present, the mechanisms involved in viral neurovirulence and in vivo replication that are induced by N-glycosylation at gB Asn-141 remain unclear. It has been reported that gB is a determinant of viral neurovirulence in mice (51) and that the appropriate expression of gB on the cell surface was required for HSV-1 neurovirulence and efficient replication in the CNS of mice (52). Together with the observations in this study that N-glycosylation at gB Asn-141 was required for the efficient expression of gB on the cell surface, the appropriate expression of gB on the cell surface that is regulated by the N-glycosylation of gB at a specific site might be involved in the in vivo neurovirulence and replication of HSV-1.
In this study, we used pooled human γ-globulins from human blood as human antibodies. HSV-1 is a ubiquitous human pathogen; approximately 70% of the global human population is infected with HSV-1, and most HSV-1-infected humans have been reported to be latently infected with the virus (1-3, 26). Therefore, it is conceivable that the effects of pooled human γ-globulins represent the effects of antibodies in humans latently infected with HSV-1. Thus, the mouse model with passive transfer of pooled human γ-globulins used in this study potentially mimicked the in vivo effects of human antibodies in humans latently infected with HSV-1. HSV-1 frequently reacti vates from latent infections and is transmitted to new human hosts. Therefore, the host's immune responses to HSV-1 persist in latently infected humans because of the repeated stimulation of the immune system, resulting in the progressive enhancement of long-term immunity (1-3). The viral strategy of using the N-glycan shield at a specific site of HSV-1 gB to evade antibodies, which was clarified in this study, may protect reactiva ted viruses from existing antibodies to HSV-1 in latently infected humans and thereby facilitate their transmission to new human hosts. Notably, the N-glycosylation site on HSV-1 gB is widely conserved in viruses subclassified in the alphaherpesvirus subfamily of herpesviruses (53), suggesting that this is a general viral mechanism of evasion from the immune system. Moreover, clarification of the HSV-1 mechanism of evasion from antibodies supports earlier conclusions (5-9) that were based on previous clinical trials of HSV vaccines; namely, that antibodies are essential to the control of HSV-1 infections in humans. Additional studies to identify other glycan shields against human antibodies on HSV envelope glycoproteins are important and should be of interest. Those studies and this present study may provide insights into the design of effective therapeutic HSV vaccines against frequent recurrences of herpes virus infections such as genital herpes.
Previous studies have characterized N-glycosylation at Asn-133 of HSV-2 gB and at Asn-154 of the pseudorabies virus (PRV) gB, which correspond to the N-glycosylation at Asn-141 of HSV-1 gB (54,55). None of those studies addressed the effects of the N-glycosylation of those viruses' gB on in vivo evasion from antibodies and pathogenesis. In agreement with our observation in this study that the N141Q mutation in HSV-1 gB did not affect viral replication in Vero and SK-N-SH cells, the ectopic expression of the PRV gB-N154Q mutant rescued the entry deficiency of a gBdeficient PRV so that its entry would be at a level similar to that of PRVs with wild-type gB (55). In contrast, the ectopic expression of the HSV-2 gB-N133Q mutant barely rescued the entry deficiency of a gBdeficient HSV-2 (54). As observed with the N141Q mutation in HSV-1 gB in the context of viral infection, the ectopic expression of both the HSV-2 gB-N133Q and PRV gB-N154Q mutants showed impaired cell surface expression of the mutants. These observations point out both the similarities in and differences between the roles of the N-glycosylation of gB in viruses.
To generate a fusion protein of maltose binding protein (MBP) and a domain of HSV-2 VP5, pMAL-VP5o-P3 was constructed by amplifying the domains encoding the codon-optimized HSV-2 VP5(VP5o) codons 401 to 600 by PCR from pEGFP-VP5o, which was engineered according to the GenScript's OptimumGene algorithm, synthesized (Genscript, Piscataway, NJ, USA) using the primers listed in Table S1B, which was followed by cloning the DNA fragments into pMAL-c (New England BioLabs, Ipswich, MA, USA) in frame with MBP. The VP5o sequence is shown in Table S1C.

Establishment of stable Vero cells with tetracycline-inducible codon-opti mized gB and ICP4 (gBo and ICP4o) expression
Vero cells were transduced with supernatants of Plat-GP cells cotransfected with pMDG (63) and pRetroX-Tet3G (TaKaRa), selected with 1 mg/mL G418 solution (Wako, Osaka, Japan) to generate Tet3G-Vero cells. The cells were further transduced with a mixture of supernatants of Plat-GP cells cotransfected with pMDG and pRetroX-TRE3G-gBo, and supernatants of Plat-GP cells co-transfected with pMDG and pRetroX-TRE3G-ICP4o to establish gBo/ICP4o-TetON-Vero cells. After double selection with 1 mg/mL of G418 solution and 5 µg/mL of puromycin, a single clone in which expression of gBo and ICP4o was induced by DOX was selected.

Two-step Red-mediated recombination using the KanR/ePheS* cassette
The two-step Red-mediated mutagenesis procedure used in this study was performed as described previously (24,64). Briefly, linear DNA fragments containing an I-SceI recognition sequence, KanR and ePheS*cassettes, and target homologous sequences were amplified by PCR from pBS-KanR-ePheS* using the primers listed in Table S1D. The linear fragments were electroporated into the electrocompetent E. coli strain GS1783 containing the pYEbac102Cre (58,65). The transformed bacteria were then incubated at 32℃ for 40 to 60 min and plated on LB agar plates containing 20 µg/mL of chloram phenicol and 40 µg/mL of kanamycin to select E. coli clones harboring pYEbac102Cre containing the KanR and ePheS* cassettes (KanR/ePheS* cassettes). Kanamycin-resistant colonies were screened by PCR with the appropriate primers. Next, the KanR/ePheS* cassettes were excised by expressing the I-SceI homing enzyme in GS1783 through induction with arabinose, followed by induction of the Red recombination machinery by raising the temperature. Briefly, 100 µL of an overnight culture of kanamycin-resistant E. coli clones grown in LB medium containing chloramphenicol and kanamycin was inoculated into 2 mL of LB medium containing chloramphenicol only. Bacteria were incubated at 32°C for 2-4 h with shaking, followed by addition of 10% (wt/vol) L-arabi nose (Wako) to the culture at a 1:5 ratio, and incubated for another 1 h at 32°C. Finally, the E. coli culture was incubated at 42°C for 30 min. It was then shaken at 32°C for another 1-2 h, and 50 µL of 10 −3 to 10 −4 dilutions of the culture was plated onto LB agar plates containing 20 µg/mL of chloramphenicol and 1 mM of 4-CP to select E. coli clones harboring the pYEbac102Cre, from which the KanR/ePheS* cassette was excised. Chloramphenicol-and 4-CP-resistant colonies were screened by PCR with appropriate primers, which was followed by nucleotide sequencing for confirmation of the desired mutation.
The recombinant virus YK696 (ΔgB-repair), in which the deletion mutation in gB was repaired, was generated by cotransfection with pYEbac102Cre carrying the gB-deletion mutation and pCRxgB (66) into Vero cells. Plaques were isolated and purified on Vero cells. Restoration was confirmed by nucleotide sequencing.
The recombinant virus YK717 (gB-SE), which expresses gB fused to a TEV protease cleavage site and a Strep-tag; and recombinant virus YK718 (gD-SE), which expresses gD fused to a TEV protease cleavage site and a Strep-tag, were generated by the two-step Red-mediated mutagenesis procedure using E. coli GS1783 containing pYEbac102Cre, as described previously (24,64), with the exception that the primers used instead of those described previously are listed in Table S1D.
In experiments in which YK695 (HSV-1 ΔgB) was used, viruses were propagated and assayed in HSV-1 gBo/ICP4o-TetON-Vero cells in the presence of doxycycline (DOX) (1 mg/mL). Other viruses used in this study were propagated and titrated in Vero cells.

Production and purification of MBP in E. coli
The MBP fusion protein MBP-VP5o-P3 was expressed in E. coli BL21 Star (DE3) (Thermo Fisher Scientific) that had been transformed with pMAL-VP5o-P3 and purified using amylose resin (New England BioLabs) as described previously (67).
To generate rabbit polyclonal antibodies to VP5, rabbits were immunized with purified MBP-VP5o-P3, following the standard protocol of Scrum (Tokyo, Japan).

PNGase F digestion and immunoblotting
Vero cells were infected with each of the indicated viruses at an MOI of 5 for 21 h and lysed with T-PER Tissue Protein Extraction Reagent (Thermo Scientific, Waltham, MA, USA). The lysates were sonicated and denatured with Glycoprotein Denaturing Buffer (NEB) by heating them at 100°C for 10 min. Aliquots of the lysates were incubated with 2,500 units of PNGase F (NEB) at 37°C for 1 h. Aliquots of the lysates incubated under the same conditions without PNGase F were used as controls. The incubated mixtures were subjected to immunoblotting as described previously (68).

Detection of gB-or gD-specific antibodies in pooled human γ-globulins by flow cytometry
Pei Max (Polyscience, Inc., Warrington, PA, USA) was used to transfect HEK293FT cells with selected plasmids. At 48 h posttransfection, the transfected cells were detached from their culture plates and washed once with phosphatebuffered saline (PBS) supplemen ted with 2% fetal calf serum (FCS) (washing buffer). Cells were fixed and permeabilized with Cytofix/Cytoperm (Becton Dickinson, Franklin Lakes, NJ, USA) and incubated with diluted human γ-globulins (G4386; Sigma-Aldrich) on ice for 30 min. After the cells were washed with washing buffer, they were further incubated with anti-human IgG conjugated to Alexa Flour 647 (Invitrogen) on ice for 30 min. After the cells were washed again, they were analyzed with a CytoFLEX S flow cytometer (Beckman Coulter, Brea, CA, USA). The data were analyzed with FlowJo 10.8.1 software (Becton Dickinson).
For the ADCC assay, human γ-globulins were incubated twice with gB-SE-beads, gD-SE-beads, or control beads, as follows: samples of gB-SE or gD-SE immobilized on Strep-Tactin Sepharose beads were incubated with 1 mL of diluted human γ-globulins (G4386; Sigma-Aldrich) (3.04 mg/mL in ADCC assay buffer) at 4°C overnight, and after centrifugation, the supernatants were further incubated with gB-SE or gD-SE immobi lized on Strep-Tactin Sepharose beads. After centrifugation, the supernatants containing gB-or gD-antibody depleted human γ-globulins were filtered.
Similarly, HSV-1(F)-infected Vero cell lysates prepared as described for the deple ted human γ-globulins was incubated with Strep-Tactin Sepharose beads, and after centrifugation and washing, the beads (control beads) were incubated once or twice with human γ-globulins diluted as described, and then centrifuged, followed by filtration of the supernatants to produce samples of mock-depleted human γ-globulins.

Neutralization assay
Pooled human γ-globulins (G4386; Sigma-Aldrich) or human serum (H4522; Sigma-Aldrich) without heat treatment at 56°C were serially diluted in medium 199 containing 1% FCS. The serial dilutions were mixed 1:1 with 100 PFU of each selected virus in medium 199 containing 1% FCS, incubated at 37°C for 1 h and then inoculated onto Vero cell monolayers to perform plaque assays. At 2 days postinfection, the plaques were counted. The percentage of neutralization was determined as follows: the numbers of plaques formed by the virus samples that had been incubated with or without human γ-globulins as a value with the following formula: 100 × [1− (numbers of plaques produced after incubation of viral samples with human γ-globulins)/(numbers of plaques produced after incubation of viral samples without human γ-globulins)]. Human serum which had been preincubated at 56°C for 30 min to inactivate complement was also used.

Determination of plaque size
Vero cells were infected with 200 PFU of wild-type HSV-1(F) or each of the recombinant viruses. After adsorption for 1 h, the inoculum was removed, and the cell monolay ers were overlaid with medium 199 containing 1% FCS and 1% agarose. At 2 days postinfection, the agarose-containing medium was removed. Cells were then fixed in methanol, permeabilized with 0.1% Triton X-100 in PBS, blocked with 10% human serum (Sigma-Aldrich) in PBS, reacted with anti-ICP4 antibodies, reacted with goat anti-mouse IgG-Alexa Fluor 488 (Invitrogen). Plaques in each culture were examined under a fluorescence microscope equipped with a BZ-X800 (KEYENCE, Osaka, Japan) and BZ-X800 Analyzer (KEYENCE).

ADCC reporter assay
The extent of ADCC activation induced by human γ-globulins was evaluated with the use of an ADCC Reporter Bioassay (Core Kit; Promega, G7010) and the EnSpireMultimode Plate Reader (PerkinElmer, Waltham, MA, USA). The assay was performed according to the manufacturer's instructions. Briefly, Vero cells were infected at an MOI of 1 with each selected virus in medium 199 containing 1% FCS. After adsorption for 1 h, the inoculum was removed, and the cell monolayers were overlaid with medium 199 containing 10% FCS. At 24 h postinfection, the culture medium was replaced with ADCC assay buffer containing effector cells and diluted human γ-globulins at a 2:1 ratio and incubated at 37°C for 6 h. Bio-Glo luciferase reagent was then added, and the luciferase signals were quantitated as relative light units (RLUs) on an EnSpire reader. Extent of induction was calculated as follows: Fold induction = (RLUs with antibody − RLUs background )/(RLUs no antibody − RLUs background ).

Determination of gB expression on the surfaces of HSV-1-infected cells
The expression of HSV-1 glycoproteins on the surfaces of infected cells was analyzed as described previously (52). Briefly, Vero cells were infected at an MOI of 1 with each selected virus in medium 199 containing 1% FCS. After adsorption for 1 h, the inoculum was removed, and the cell monolayers were overlaid with medium 199 containing 10% FCS. At 24 h postinfection, cell monolayers were detached with PBS containing 0.02% EDTA and were washed one time with PBS supplemented with 2% FCS (washing buffer). To analyze the total expression of gB, infected Vero cells were detached as described, and fixed and permeabilized with Cytofix/Cytoperm Fixation/Permeabilization Solution (Becton Dickinson). Treated and untreated cells were then incubated with mouse anti-gB monoclonal antibody in washing buffer on ice for 30 min. After the cells were washed with washing buffer, they were further incubated with anti-mouse IgG conjugated to Alexa Flour 647 dye (Invitrogen) on ice for 30 min. After the cells were washed again, they were analyzed with a CytoFLEX S flow cytometer (Beckman Coulter). The data were analyzed by FlowJo 10.8.1 software (Becton Dickinson).

Immunofluorescence assays
Immunofluorescence assays were performed as described previously (69).

Purification of virions
Virions were purified as described previously (70).

Animal studies
Female ICR mice were purchased from Charles River Laboratories. For ocular infections by each selected virus in mice in the presence of pooled human γ-globulins, 4-week-old mice were injected intraperitoneally with 1,250 mg/kg of human γ-globulins or PBS. One day after administration, the mice were ocularly infected with 3 × 10 6 PFU/eye of each selected virus, as described previously (57). For mice immunized with HSV-1 before ocular infection, 3-week-old mice were injected subcutaneously in the neck with 5 × 10 5 PFU of HSV-1(F). The immunized mice were then infected ocularly 9 weeks after immunization with 3 × 10 6 PFU/eye of each selected virus as described previously (57). Virus titers in the tear films of mice were determined as described previously (71).
For intracranial infections, 3-week-old mice were inoculated intracranially with each selected virus as described previously (57). Mice were monitored daily, and mortality occurring from 1 to 14 days postinfection was attributed to the infecting virus. To measure viral titers in the brains of infected mice, 3-week-old female ICR mice were each inoculated intracranially with 1 × 10 3 PFU of each selected virus. At 1, 3, and 5 days postinfection, the brains of the mice were harvested, and virus titers were determined on Vero cells. All animal experiments were carried out in accordance with the Guidelines for Proper Conduct of Animal Experiments, Science Council of Japan. The protocol was approved by the Institutional Animal Care and Use Committee (IACUC) of the Institute of Medical Science, the University of Tokyo (IACUC protocol approval number: A21-55).

Modeling of the N-glycosylated gB protein
The N-glycan core (Man 3 GlcNAc 2 ) was modeled according to the prefusion structure of HSV-1 (23) gB (PDB ID: 6Z9M) using the Glycan Reader and Modeler (72) and the CHARMM-GUI program (73,74). Discovery Studio 2021 software (Dassault Systèmes) was used to change the χ1 angle (N-Cα-Cβ-Cγ) of the Asn141 B chain (PDB ID: 6Z9M) from −178° to −66° to avoid steric clash of the N-glycan with the neighboring polypeptide. Visualization of the protein 3D structure was performed in the PyMol Molecular Graphics System, version 2.5 (Schrödinger, LLC, New York, NY, USA).

Analysis of protein surface
The AREAIMOL program (CCP4 package, version 6.2) was used to determine the accessible surface area (ASA) (75). A spherical probe with a radius of 10 Å, which is similar to the dimension of the antigen-binding fragments (single-chain variable fragment) of the antibodies, was used in the estimation of the ASA (76,77). The extent of glycan shielding (ΔASA) was estimated for each amino acid residue by calculating the difference between the ASAs of N-glycosylated and nonglycosylated gB structures [ΔASA = ASA (nonglycosylated gB) − ASA (N-glycosylated gB)].

Statistical analysis
The unpaired t test was used to compare two groups. One-way or two-way ANOVA followed by the Tukey or Dunnett multiple comparisons tests was used for multiple comparisons. A P value <0.05 was considered significant. For the statistical analysis of viral titers, data were converted to common logarithms (log 10 ). For values below the detection limit, statistical processing was performed assuming that the values were those of the detection limit. GraphPad Prism 8 (GraphPad Software) was used to perform statistical analysis.