Herpes Simplex Virus 2 ICP0− Mutant Viruses Are Avirulent and Immunogenic: Implications for a Genital Herpes Vaccine

Herpes simplex virus 1 (HSV-1) ICP0 − mutants are interferon-sensitive, avirulent, and elicit protective immunity against HSV-1 (Virol J, 2006, 3:44). If an ICP0 − mutant of herpes simplex virus 2 (HSV-2) exhibited similar properties, such a virus might be used to vaccinate against genital herpes. The current study was initiated to explore this possibility. Several HSV-2 ICP0 − mutant viruses were constructed and evaluated in terms of three parameters: i. interferon-sensitivity; ii. virulence in mice; and iii. capacity to elicit protective immunity against HSV-2. One ICP0 − mutant virus in particular, HSV-2 0ΔNLS, achieved an optimal balance between avirulence and immunogenicity. HSV-2 0ΔNLS was interferon-sensitive in cultured cells. HSV-2 0ΔNLS replicated to low levels in the eyes of inoculated mice, but was rapidly repressed by an innate, Stat 1-dependent host immune response. HSV-2 0ΔNLS failed to spread from sites of inoculation, and hence produced only inapparent infections. Mice inoculated with HSV-2 0ΔNLS consistently mounted an HSV-specific IgG antibody response, and were consistently protected against lethal challenge with wild-type HSV-2. Based on their avirulence and immunogenicity, we propose that HSV-2 ICP0 − mutant viruses merit consideration for their potential to prevent the spread of HSV-2 and genital herpes.


Introduction
Herpes simplex virus 1 (HSV-1) or herpes simplex virus 2 (HSV-2) may cause primary genital herpes [1,2,3]. In contrast, more than 90% of cases of recurrent genital herpes are caused by HSV-2 [1,4]. Recurrent genital herpes is a physically superficial disease in most cases, but often produces emotionally debilitating effects [5,6,7]. Because HSV-2 is responsible for the bulk of disease, an effective HSV-2 vaccine is sought that may stop the spread of genital herpes.
HSV-1 and HSV-2 are similar viruses that share a nearly identical set of ,75 co-linear genes distributed across 152 and 154 kbp dsDNA genomes, respectively. HSV-1 and HSV-2 establish life-long infections in their human hosts, and both viruses persist by establishing latent infections in the human nervous system. Infections with these viruses are exceedingly common; HSV-1 infects ,4 billion people and HSV-2 infects ,1 billion people [8]. Approximately 5% of HSV-2 infected individuals live with genital herpes disease that recurs once every 3 to 12 months [9,10]. HSV-2 infections spread to new individuals at a rate of ,20 million per year. An effective HSV-2 vaccine would be useful in breaking this cycle, and protecting young adults from the 1 in 10 chance that they will acquire HSV-2 before they marry [4,11,12].
For decades, efforts to develop an HSV-2 vaccine have been predicated on the assumption that a live-attenuated HSV-2 virus would be too risky for use as a human vaccine, and safer alternatives have been sought [13,14]. The non-replicating HSV-2 vaccines that have been most seriously considered are 1. HSV-2 subunit vaccines and 2. replication-defective HSV-2 viruses. The immunodominant HSV-2 glycoprotein D (gD 2 ) antigen has been considered for its potential to serve as a subunit vaccine. After more than a decade of testing in human clinical trials, it remains unclear that vaccination with a gD 2 subunit renders human recipients immune to wild-type HSV-2 infections and/or genital herpes [15,16]. One of the limitations of an HSV-2 subunit vaccine is that vaccine recipients are only exposed to 1% of HSV-2's antigens (i.e., 1 of ,80 proteins). Replication-defective HSV-2 viruses offer the advantage that nearly all of HSV-2's antigens may be expressed at the site of inoculation and presented to CD8 + T cells in the context of the MHC class I pathway [17,18]. However, it remains unclear if a replication-defective HSV-2 virus may recapitulate the magnitude and duration to the immune response that is elicited against a live, replicating virus such as wild-type HSV-2.
An HSV-2 vaccine should not only be safe, but it should also be effective. For decades, live HSV-2 viruses have been largely excluded from consideration as a genital herpes vaccine on the grounds that a live-attenuated HSV-2 vaccine would be too dangerous. However, this safety-based rationale is incongruous with the 200-year history of viral vaccines. Approximately 75% of the vaccines that have succeeded in preventing human viral disease have contained live, replicating viruses. Pediatricians and parents have deemed the approach safe enough for the past 50 years to warrant the inoculation of hundreds of millions of children with live, replicating viruses, and millions of human lives have been spared from death or disfigurement as a result.
Historically, live-attenuated viruses have been our most effective mode of vaccination. Originally, the word 'vaccination' specifically meant to inject a person with live, replicating vaccinia virus in order to elicit a cross-protective response that provided immunity against the smallpox virus [19]. Most of our effective viral vaccines emulate the original approach, and rely on inoculation of humans with live viruses that establish mild or inapparent infections that cross-protect against their more virulent counterparts that exist in nature. Such live-attenuated viruses are the active ingredient in the oral poliovirus vaccine, the MMR (mumps, measles, rubella) vaccine, and the chickenpox and shingles vaccine [20,21,22]. Isolated reports have raised the possibility that a live-attenuated HSV-2 vaccine might be feasible [23,24,25]. However, a liveattenuated HSV-2 vaccine has not been systematically investigated due to concerns surrounding the safety of administering a live aherpesvirus to millions of people [26].
Tens of millions of children have now been inoculated with the live-attenuated Oka strain of varicella-zoster virus (VZV) [27]. Like HSV-1 and HSV-2, VZV is an a-herpesvirus that routinely establishes life-long infections in the human nervous system [28]. The live VZV vaccine has proven safe and effective in preventing the epidemic spread of chickenpox [20,27], and is now being used to prevent the age-onset disease of 'shingles' caused by reactivated VZV infections [29,30,31]. The success of the chickenpox vaccine demonstrates that a live and appropriately attenuated a-herpesvirus may be used to safely control a human disease. If this principle may be expanded to include HSV-1 and HSV-2, then genital herpes might also be prevented using a live-attenuated HSV-2 virus as a vaccine. The feasibility of this proposal remains unclear because it has not been investigated.
The current study was initiated to test this hypothesis. Five HSV-2 ICP0 2 mutant viruses were constructed and characterized using a screening procedure designed to determine if any of these mutant viruses achieved the type of balance between safety and immunogenicity that should be expected of a live-attenuated vaccine strain. We report that one HSV-2 ICP0 2 mutant virus in particular, HSV-2 0DNLS, exhibited such properties. HSV-2 0DNLS was interferon-sensitive, avirulent in immunocompetent animals, and elicited robust protection in recipients against later exposures to wild-type HSV-2. Based on these results, we propose that interferon-sensitive HSV-2 ICP0 2 mutant viruses warrant consideration for their potential to prevent the spread of HSV-2 and genital herpes in the human population.

Results
Construction and characterization of HSV-2 ICP0 2 mutant viruses Three regions of ICP0 are conserved between HSV-1 and HSV-2; a RING finger-E3 ligase region [38], the region surrounding the nuclear localization signal (NLS) and adjacent phosphorylation sites [39,40], and a C-terminal oligomerization domain [41,42] (Fig. 1B, Fig. S1). To obtain HSV-2 viruses that varied in their level of attenuation, deletions were introduced into the ICP0 gene that removed none, one, or all of ICP0's conserved regions (Fig. 1A, 1B). All HSV-2 ICP0 2 mutant viruses bore a common deletion that replaced codons 19 to 104 of the ICP0 gene with a green fluorescent protein (GFP) coding sequence. HSV-2 0D104 contained this one mutation, and thus encoded a GFP-tagged, mutant protein that retained all of ICP0's conserved regions (Fig. 1A, 1B). A larger deletion in HSV-2 0DRING removed codons 19 to 162, and thus removed several cysteine residues necessary for ICP0's E3 ligase activity [43] (Fig. 1A, 1B). HSV-2 0DNLS contained two deletions that removed codons 19 to 104 and codons 489 to 694 of the ICP0 gene. Thus, HSV-2 0DNLS encoded a GFP-tagged protein that lacked ICP0's NLS region and a portion of the C-terminal oligomerization domain (Fig. 1A, 1B). Finally, HSV-2 0D254 and HSV-2 0D810 contained deletions in the ICP0 gene that resulted in expression of little more than GFP from the ICP0 locus (Fig. 1A, 1B).
Southern blot analysis confirmed that each HSV-2 mutant virus bore an insertion of GFP coding sequence in place of a deletion of the expected size in the ICP0 gene (Fig. 1A, 1C). Western blot analysis with a GFP-specific antibody confirmed that HSV-2 0D254 and HSV-2 0D810 encoded GFP-tagged, ICP0 peptides that were only slightly larger than native GFP expressed by an HSV-2 recombinant virus, HSV-2 MS-GFP (Fig. 1D, Fig. S2). As expected, HSV-2 0D104 and HSV-2 0DNLS encoded their respective 125 kDa and 85 kDa GFP-tagged proteins (Fig. 1D). In addition, 5 lower MW peptides of 45-80 kDa also reacted with GFP antibody (Fig. 1D). HSV-2 0DRING encoded an ,120 kDa GFP-tagged protein, and only a single additional peptide was observed (Fig. 1D). ICP0 ubiquitinates itself in a RING-fingerdependent manner [44]; hence, most of the lower MW peptides observed in cells infected with HSV-2 0D104 or 0DNLS appeared to be the result of autoubiquitination and proteolytic turnover of GFP-tagged ICP0 (Fig. 1D). Collectively, Southern and Western blot analysis indicated that the intended mutations were successfully introduced into the HSV-2 ICP0 gene.
The growth kinetics of wild-type HSV-2 MS and HSV-2 ICP0 2 mutant viruses were compared in monolayers of ICP0 + L7 cells, Vero cells, or IFN-b-treated Vero cells inoculated with 0.1 viral plaque-forming units (pfu) per cell. Wild-type HSV-2 grew to similar titers in L7 cells and Vero cells (Fig. 3A). The kinetics of HSV-2 MS replication were delayed in IFN-b-treated cells, but HSV-2 MS achieved a final titer that was only 6-fold lower than that observed in ICP0 + L7 cells (Fig. 3A). HSV-2 0D104 was modestly attenuated, and grew to 100-fold lower titers in IFN-b-treated Vero cells relative to ICP0 + L7 cells (Fig. 3B). In contrast, the other HSV-2 ICP0 2 mutant viruses 0DRING, 0D254, 0D810, and 0DNLS were acutely sensitive to IFN-a/b, and grew to 5,000-to 31,000-fold lower titers Regions of ICP0 conserved between HSV-1 and HSV-2 are color-coded above, and below is shown the effects of deletions on ICP0's conserved RING finger region, nuclear localization signal (NLS) region, and/or oligomerization domain. (C) Southern blot analysis of SacI -StuI digested DNA harvested from cells that were uninfected (UI) or were inoculated with 2.5 pfu per cell of each indicated HSV-2 virus. Replicate Southern blots were hybridized with an ICP0 exon 2-specific oligonucleotide (shown on left) or a GFP-specific oligonucleotide (shown on right). The approximate locus to which the ICP0-specific and GFP-specific probes hybridized is depicted in panel A. (D) Western blot analysis of proteins harvested from cells that were uninfected (UI) or were inoculated with 2.5 pfu per cell of each indicated HSV-2 virus. GFP or chimeric ICP0 proteins bearing a GFP tag were labeled with rabbit polyclonal GFP-specific antibody. Molecular weight markers are shown on the right. doi:10.1371/journal.pone.0012251.g001 in IFN-b-treated Vero cells relative to ICP0 + L7 cells (Fig. 3C-F). In the absence of IFN-b, HSV-2 0D254 and 0D810 replicated poorly in untreated Vero cells and achieved final titers of less than 10 4 pfu/ ml (Fig. 3D, 3E). In contrast, HSV-2 0D104, 0DRING, and 0DNLS each replicated to titers of greater than 10 5 pfu per ml in untreated Vero cells (Fig. 3B, 3C, and 3F). Thus, HSV-2 0DRING and 0DNLS were the only HSV-2 ICP0 2 mutant viruses that were IFNsensitive and retained the capacity to replicate with reasonable efficiency in cells that did not provide wild-type ICP0 in trans.
HSV-2 0DNLS is severely attenuated relative to its ICP0 + parent virus, HSV-2 MS A dose of 500 or more pfu of wild-type HSV-1 readily establishes productive infections in the eyes of mice, whereas 20fold higher doses of HSV-1 ICP0 2 mutant viruses are required to establish a productive infection in mouse eyes [46]. To determine the relative doses of wild-type HSV-2 or HSV-2 ICP0 2 mutant virus required to establish a productive infection in mouse eyes, outbred ICR mice were inoculated bilaterally with 0.8, 4, 20, or 100 thousand pfu per eye of wild-type HSV-2 MS strain or HSV-2 0DNLS (n = 8 per virus per dose). At all viral doses tested, wildtype HSV-2 MS established robust infections that led to HSV-2 shedding in mouse eyes by Day 2 post-inoculation (p.i.) and produced overt periocular disease by Day 5 p.i. (Fig. 4A, 4B). Wild-type HSV-2 MS infections were lethal in 31 of 32 ICR mice (Fig. 4C). These results were consistent with independent tests in which the 50% lethal dose (LD 50 ) of HSV-2 MS was found to be ,200 pfu per eye (not shown).
HSV-2 0DNLS required higher doses of virus to establish a productive infection in outbred ICR mice. At doses of 800 or 4,000 pfu per eye, less than 50% of mice inoculated with HSV-2 0DNLS shed virus on Day 2 p.i. (Fig. 4A). However, at doses of 20,000 or 100,000 pfu per eye, ocular shedding of HSV-2 0DNLS was observed in all mice on Day 2 p.i. (Fig. 4A). Mice inoculated with HSV-2 0DNLS remained indistinguishable from uninfected mice between Days 5 and 60 p.i. (Fig. 4B, 4D). Therefore, the LD 50 of HSV-2 0DNLS was at least 500 times greater than HSV-2 MS in an eye model of infection. Based on these results, a dose of 100,000 pfu per eye was used to inoculate mice with HSV-2 ICP0 2 mutant viruses in subsequent tests.  Interferon-sensitive, HSV-2 ICP0 2 mutant viruses establish inapparent infections The efficiency of replication of HSV-2 ICP0 2 mutant viruses was compared to wild-type HSV-2 following inoculation of the right eye with a viral dose of 100,000 pfu per eye (Fig. 5). To ensure that a subset of mice survived infection with wild-type HSV-2, one group of HSV-2 MS-infected mice was aggressively treated with acyclovir (ACV) both orally and intraperitoneally between Days 23 and +20 p.i. to limit viral spread and pathogenesis [47,48,49]. All mice inoculated with HSV-2 MS (ICP0 + ) shed more than 3,000 pfu from the right eye on Day 1 p.i., and ACV treatment only modestly reduced HSV-2 shedding on Days 2 and 3 p.i. (Fig. 5A). Each of the five mutations in HSV-2's ICP0 gene resulted in a significant reduction of HSV-2 shedding between Days 1 and 3 p.i. (Fig. 5A, 5B). Specifically, mice inoculated with HSV-2 0D254, 0D810, or 0DRING shed less than 100 pfu per eye between Days 1 and 3 p.i. (Fig. 5A, 5B). Likewise, mice inoculated in the right eye with HSV-2 0D104 or HSV-2 0DNLS shed peak titers of ,300 pfu per eye on Day 2 p.i. (Fig. 5B).
Mice inoculated with HSV-2 0DNLS are consistently protected against wild-type HSV-2 HSV-2-specific IgG levels and protective immunity were compared among the 100% of mice that survived inoculation with HSV-2 ICP0 2 mutant viruses and the 70% of ACV-treated mice that survived primary infection with wild-type HSV-2 MS (Fig. 5D). Mice were bled on Day 60 p.i. and sera were tested for the presence of IgG antibody against recombinant HSV-2 glycoprotein D (gD 2 ) [50]. Dilutions of pooled HSV-2 antiserum defined the quantitative relationship between color development and gD 2 -antibody abundance (r 2 = 1.00; Fig. S3). The serum of naïve mice defined the background of the assay, and HSV-2 MS latently infected mice served as a positive control group that should possess bona fide immunity to secondary HSV-2 infection (Fig. 6A). HSV-2 MS latently infected mice possessed levels of gD 2 -antibody that were ,1,350-fold above background (Fig. 6A, p,0.001). Only ,50% of mice inoculated with HSV-2 0DRING, 0D254, or 0D810 possessed detectable gD 2 -antibody, and thus their gD 2 -antibody responses did not significantly differ from background ( Fig. 6A; p.0.05). In contrast, the level of gD 2antibody in HSV-2 0DNLS-inoculated mice was an average 53fold above background, and thus represented 4% of that observed in HSV-2 MS latently infected mice ( Fig. 6A; p,0.001).
On Days 70 and 80 after inoculating the right eye with HSV-2 ICP0 2 mutant viruses, mice were secondarily challenged in the left eye with 500 times the LD 50 of HSV-2 MS (i.e., 100,000 pfu per eye). The summated results of these two challenge experiments are presented (Sn = 16 mice per group). As expected, 060% of naïve mice survived for more than 8 days post-challenge (Fig. 6B). Consistent with their robust gD 2antibody responses, 10060% of HSV-2 MS latently infected mice survived superinfection of the left eye with HSV-2 MS, and did not exhibit any symptoms of disease following challenge (Fig. 6B). Mice inoculated with HSV-2 0DRING, 0D254, or 0D810 were not consistently protected against wild-type HSV-2 MS challenge, and their survival rates were 70610%, 50620%, and 40620%, respectively (Fig. 6B). In contrast, 10060% of mice inoculated with HSV-2 0DNLS survived lethal challenge with HSV-2 MS, and 11 of 16 of these mice survived diseasefree for 30 days post-challenge (Fig. 6B). Therefore, HSV-2 0DNLS was the only virus tested that consistently 1. replicated at the site of inoculation, 2. established an inapparent infection in mice, and 3. elicited robust protection in mice against later exposures to wild-type HSV-2.

HSV-2 0DNLS infections fail to spread from the site of inoculation
We were interested in clarifying how HSV-2 0DNLS consistently replicated in mouse eyes without producing disease. We considered the possibility that the 0DNLS mutation might reduce the spread of HSV-2 0DNLS infection. To test this hypothesis, we compared the spread of HSV-2 0DNLS (ICP0 2 ) infection across mouse faces relative to an ICP0 + virus that also expressed a fluorescent marker, HSV-2 MS-GFP (Fig. S2).
On Day 2 after bilateral inoculation with 100,000 pfu per eye of HSV-2 MS-GFP, GFP expression was only observed at the site of inoculation (Fig. 8A, 8B; Fig. S4). On day 4 p.i., circular foci of HSV-2 MS replication (GFP + cells) were evident on the noses of mice, and these circular lesions enlarged and merged by Day 6 p.i. (Fig. 8B). This rapid spread of HSV-2 MS-GFP infection was accompanied by intense inflammation of the epithelium, and was rapidly followed by fatal encephalitis.
On Day 2 after bilateral inoculation with 100,000 pfu/eye of HSV-2 0DNLS, GFP expression was only observed at the site of inoculation (Fig. 8A, 8B; Fig. S4). HSV-2 0DNLS did not visibly spread from the site of inoculation, and thus GFP + cells were not observed in the eyes of mice or in the periocular epithelium on Days 4, 6, or 8 p.i. (Fig. 8B, Fig. S4; Day 8 not shown). Thus, HSV-2 0DNLS replicated briefly at the site of inoculation, but failed to spread or cause visible inflammation and disease.

HSV-2 ICP0 2 mutant viruses that are interferon-sensitive are severely attenuated in vivo
Prior studies have established that HSV-1 ICP0 2 mutant viruses are interferon-sensitive, avirulent, and elicit protective immunity against HSV-1 [37]. Based on these observations, we have inferred that HSV-2 ICP0 2 mutant viruses might be useful as a liveattenuated vaccine to prevent genital herpes [13,37]. Understandably, enthusiasm for this proposal has been limited by the absence of data describing HSV-2 ICP0 2 mutant viruses. The results of the current study address this gap in knowledge.
HSV-2 MS infection of the eye is almost invariably lethal in naïve ICR mice. In contrast, recombinant HSV-2 ICP0 2 mutant viruses that were IFN-sensitive (Fig. 2, 3) were also avirulent in ICR mice (Fig. 4, 5). Hence, mutations in the ICP0 gene are sufficient to attenuate even a highly virulent HSV-2 strain such as MS. One HSV-2 MS-derived ICP0 2 mutant, HSV-2 0D104, was not attenuated and produced lethal disease in 24 of 25 mice (Fig. 5D). In hindsight, this outcome was predictable based on HSV-2 0D104's resistance to IFN-inducible repression in cultured cells (Fig. 2, 3). The correlation between IFN-sensitivity and avirulence is unlikely to be a coincidence. HSV-1 ICP0 2 mutant viruses are avirulent in immunocompetent mice, but acquire the ability to replicate efficiently and cause disease when host IFN signaling pathways are compromised [35,37]. Based on this precedent, we were able to correctly predict that the avirulent phenotype of HSV-2 0DNLS would be dependent on the IFN-signal transducing factor, Stat 1 (Fig. 7). Thus consistent with the established properties of HSV-1 ICP0 2 viruses, the IFN-induced antiviral state appears to be essential to the avirulent phenotype of IFN-sensitive HSV-2 ICP0 2 mutant viruses in immunocompetent animals.
Are the phenotypes of mutant HSV-2 viruses a consequence of mutations in the ICP0 gene?
Four lines of evidence indicate that the phenotypes of HSV-2 0DNLS are a consequence of the intended mutation in the ICP0 gene, and this data is considered, as follows. First, Southern blot and Western blot analysis verify that the predicted genetic change was introduced into the ICP0 locus of HSV-2 0D104, 0DRING, 0DNLS, 0D254, and 0D810 (Fig. 1). Second, the phenotypes of IFNsensitivity and avirulence are shared by four independent HSV-2 ICP0 2 mutant viruses; HSV-2 0DRING, 0DNLS, 0D254, and 0D810. It is improbable that four independent HSV-2 ICP0 2 mutant viruses would each acquire the expected phenotypes of IFNsensitivity and avirulence based on secondary mutations randomly scattered throughout portions of the HSV-2 genome outside of the ICP0 locus. Third, it was previously predicted that HSV-2 ICP0 2 mutant viruses should be IFN-sensitive and avirulent [13,37] based on the fact that multiple HSV-1 ICP0 2 mutant viruses exhibit these phenotypes [35,36,45]. Finally, whatever genetic defect impairs the replication of HSV-2 0DRING, 0DNLS, 0D254, and 0D810 in Vero cells is functionally complemented when wild-type ICP0 is provided in trans from the ICP0 + Vero-derived cell line, L7 cells (Fig. 3). Therefore, mutations within the ICP0 gene of HSV-2 0DNLS appear to be responsible for the observed phenotypes of IFN-sensitivity and avirulence. i. from the eyes of (A) wild-type strain 129 mice, (B) lymphocyte-deficient rag2 2/2 mice, or (C) IFN-signaling-deficient stat1 2/2 mice inoculated bilaterally with 100,000 pfu/eye of wild-type HSV-2 MS, HSV-2 0DNLS, or HSV-2 0DRING (n = 10 mice per group). A single asterisk (*) denotes p,0.05 and a double asterisk (**) denotes p,0.001 regarding the probability, p, that viral shedding was equivalent to HSV-2 MS shedding in the same strain of mice. (D-F) Duration of survival of (D) strain 129 mice, (E) rag2 2/2 mice, or (F) stat1 2/2 mice following inoculation with wild-type HSV-2 MS, HSV-2 0DNLS, or HSV-2 0DRING (n = 10 mice per group). A double asterisk (**) denotes p,0.001 regarding the probability, p, that the duration of survival was equivalent to HSV-2 MS-infected mice of the same mouse strain. doi:10.1371/journal.pone.0012251.g007 HSV-2 ICP0 2 mutant viruses that lack ICP0's RING finger region are overattenuated HSV-2 ICP0 2 mutant viruses that included deletions of ICP0's RING finger domain were avirulent, but did not elicit consistent protection against HSV-2 (Fig. 6B). Thus, deletions in ICP0's RING finger region produced HSV-2 viruses that appeared to be overattenuated, and not particularly likely to succeed as a liveattenuated HSV-2 vaccine strain. In contrast, HSV-2 0DNLS exhibited the properties of a desirable live-attenuated HSV-2 vaccine strain. HSV-2 0DNLS consistently replicated at sites of ocular inoculation, and was avirulent in immunocompetent mice. Due to its consistent replication, HSV-2 0DNLS consistently stimulated an immune response in recipient mice, as measured by IgG antibodies against HSV-2's immunodominant gD antigen, and protective immunity against wild-type HSV-2 (Fig. 6).
The results suggest that HSV-2 replication at the site of immunization may be proportional to the magnitude of the resulting immune response. For example, mice inoculated with HSV-2 0DRING shed peak titers of virus that were ,1% of the peak titers shed by HSV-2 MS-infected mice and mounted gD 2antibody responses that were ,0.5% of that in HSV-2 MS-infected mice (Fig. 5, 6). Likewise, mice inoculated with HSV-2 0DNLS shed peak titers of virus that were 10-fold higher than the peak titer of virus shed by HSV-2 0DRING-infected mice, and mounted a gD 2 -antibody response that was ,10-fold greater than HSV-2 0DRING and which was 1/25 th of that elicited by HSV-2 MS (Fig. 5, 6). Therefore, the average peak titer of virus shed following inoculation (Fig. 5A, 5B) was roughly predictive of the average gD 2 -antibody response observed on Day 60 p.i. (Fig. 6A). Importantly, serum levels of gD 2 -antibody correlated with functional protection against HSV-2 MS challenge (Fig. 6A vs 6B).

Correlation between gD 2 -antibody titers and protective immunity against HSV-2?
When HSV-2 gD has been used as a subunit vaccine in human clinical trials, the potency of the resulting gD 2 -antibody response does not correlate with protective immunity against HSV-2 [55,56,57]. This observation appears to contradict our finding that mice immunized with live HSV-2 viruses generate a protective immune response that varies in proportion to serum levels of gD 2 -antibody (Fig. 6A, 6B). Two explanations may resolve this discrepancy. First, mouse IgG antibodies against HSV-2 may be more protective than human IgG antibodies against HSV-2 [58]. Thus, the correlation between gD 2 -antibody titers and protection against HSV-2 challenge may be an artefact of the mouse model used in this study. A second possibility is that HSV-2 gD subunit vaccines may be only weakly protective against HSV-2 infections regardless of the magnitude of the antibody response elicited against the gD 2 subunit. We favor the latter explanation for two reasons.
First, immunization with an HSV-2 gD subunit vaccine is the equivalent of vaccinating an animal with ,1% of HSV-2 (i.e., 1 of 80 HSV-2 proteins). In contrast, immunization with a live virus such as HSV-2 0DNLS may drive the clonal expansion of B and T lymphocytes specific for nearly the entire proteome of HSV-2, which consists of thousands of combinations of 6 to 10 amino-acid epitopes distributed across HSV-2's 80 proteins. The increased breadth of the immune response against a live-attenuated virus may explain why this mode of vaccination has been so effective in preventing severe viral diseases such as smallpox and measles.
Second, gD 2 -antibody responses provide a robust index of protective immunity against HSV-2 when mice are immunized with a live HSV-2 virus, but this is not the case when mice are immunized with an HSV-2 gD subunit vaccine (unpublished data of W. Halford). For example, mice immunized with an HSV-2 gD subunit vaccine mount a gD 2 -antibody response that is 20-fold greater than that elicited by immunization of mice with HSV-2 0DNLS (unpublished data of W. Halford). However, only 3 of 45 gD-immunized mice survive ocular or vaginal challenge with wildtype HSV-2, whereas 119 of 120 0DNLS-immunized mice survive ocular or vaginal challenge with wild-type HSV-2 (unpublished data of W. Halford). Based on our experience in mice, we infer that when a live HSV-2 virus is used as the immunogen, then serum levels of gD 2 -antibody provide a reliable index of protective immunity against HSV-2. It remains to be determined if this inference holds true in other species.

HSV-2 latently infected mice as a positive control for vaccine-challenge studies
Most HSV-2 vaccine-challenge study designs only compare immunity to HSV-2 in vaccinated animals versus naïve controls [16,18,59,60,61]. Such study designs establish that an HSV-2 vaccine candidate elicits more than ''0% immunity'' against HSV-2. However, the efficacy of HSV-2 vaccine candidates might be more clearly defined on a scale of 0 to 100% immunity against HSV-2. Given that animals latently infected with wild-type HSV are resistant to superinfection [62,63,64], the efficacy of HSV-2 vaccine candidates could be described in terms of ''% immunity'' relative to such a positive control. In the current study, we describe the efficacy of the HSV-2 0DNLS vaccine candidate in such terms (Fig. 6A, 6B). Specifically, a single exposure to HSV-2 0DNLS elicited a gD 2 -antibody response that was 4% of that observed in wild-type HSV-2 MS latently infected mice (Fig. 6A). Likewise, HSV-2 0DNLS showed promise as a vaccine candidate because recipient mice survived lethal HSV-2 MS challenge at the same 100% frequency as HSV-2 MS latently infected mice (Fig. 6B). We propose that future vaccine studies might benefit from the inclusion of such a positive control that helps define the ''100% target value'' of a bona fide protective immune response against HSV-2.

Safety concerns surrounding a vaccine based on a live-attenuated a-herpesvirus?
Many questions remain to be addressed about a live-attenuated HSV-2 vaccine strain. Is it possible that a live-attenuated virus such as HSV-2 0DNLS may recombine with endogenous or superinfecting HSV-2 viruses? Would HSV-2 0DNLS establish latent infections in vaccine recipients, or reactivate from the latent state? These are important questions that merit further study. However, an initial risk-benefit analysis may be performed in light of clinical experience with the live VZV vaccine. Like HSV-2, VZV is an a-herpesvirus that establishes life-long infections in human neurons. Due to the extensive similarity between these viruses, the safety concerns surrounding a live HSV-2 vaccine strain are similar to the Oka vaccine strain of VZV.
In the .20-year history of vaccinating against chickenpox, recombination between the VZV Oka strain and wild-type VZV has not been reported as a clinical problem. Rather, the major problem is that the VZV Oka strain may produce disease in recipients that are severely immunocompromised [27]. While the VZV Oka strain may establish latent infections in human neurons and reactivate from the latent state [28], clinical experience suggests that the risks associated with this live VZV vaccine are outweighed by the benefits of not leaving a population susceptible to the .90% risk of being infected with wild-type VZV [65]. If clinical experience with the VZV Oka strain is any indication, then the risks associated with a live-attenuated HSV-2 vaccine strain would likely be preferable to the current situation in which wild-type HSV-2 is carried by ,1 billion people, and ,20 million people are newly infected with these disease-causing strains of HSV-2 each year.

Conclusion
Most vaccines that have succeeded in preventing viral disease have been based upon live, replicating viruses. Millions of children receive vaccines every year that contain live-attenuated variants of poliovirus, mumps virus, measles virus, rubella virus, and VZV. Despite these successes, efforts to develop a genital herpes vaccine have primarily focused on non-replicating HSV-2 vaccine candidates such as gD 2 -protein subunits [15,16,66,67,68] and replication-defective HSV-2 viruses [17,18,61,69]. These approaches are extraordinarily safe, but have not slowed the spread of genital herpes. Perhaps it is time to consider the possibility that a live-attenuated HSV-2 vaccine strain might be more effective.
Deletion of conserved regions of ICP0 is a feasible approach to obtain a new class of live-attenuated HSV-2 vaccine strains that is safe in theory and practice due to their exquisite sensitivity to repression by the innate IFN system of the animal host. Therefore, we conclude that HSV-2 ICP0 2 mutant viruses merit further consideration for their potential to prevent the spread of HSV-2 and genital herpes in the human population.

Ethics Statement
Mice were handled in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. This study was approved by the Southern Illinois University School of Medicine Laboratory Animal Care and Use Committee in August 2008, and was assigned Protocol Numbers #205-08-018 and #205-08-019. These protocols remain active and are associated with an NIH-funded grant for the ''Development of an Effective Genital Herpes Vaccine (R21 AI081072).

Cells and viruses
Vero cells and U2OS cells were obtained from the American Type Culture Collection (Manassas, VA), and High Five TM insect cells were obtained from Invitrogen Corporation (Carlsbad, CA). The ICP0-complementing L7 cell line [70] was kindly provided by Neal Deluca (University of Pittsburgh). Cell lines were propagated in Dulbecco's Modified Eagle's medium supplemented with 5% fetal bovine serum and antibiotics. The HSV-2 recombinant viruses used in this study were derivative of HSV-2 MS (American Type Culture Collection). Most HSV-2 lab strains and HSV-2 clinical isolates grow to ,100-fold lower titers than HSV-1. To circumvent this problem, all HSV-2 viruses were propagated in U2OS cells at 34uC following inoculation with a multiplicity of infection of 0.01 pfu per cell, which generally resulted in an ,10fold increase in HSV-2 titers. For both wild-type HSV-2 and HSV-2 ICP0 2 mutant viruses, viral stocks were generated for animal experiments that were concentrated 10-fold by ultracentrifugation to achieve a minimum titer of 3610 7 pfu/ml. An HSV-2 glycoprotein D-expressing baculovirus was used to purify the gD-2 306t protein [50], and was generously provided by Dr. Gary Cohen and Dr. Roslyn Eisenberg (University of Pennsylvania).

Plasmid precursors of HSV-2 recombinant viruses
A plasmid template for mutagenesis of the HSV-2 ICP0 gene was derived, as follows. An 11.5 kb HindIII -KpnI DNA fragment was subcloned from wild-type HSV-2 MS genomic DNA into a pUC18 plasmid vector. This DNA fragment encompassed the long-internal repeat (IR L ) of the HSV-2 genome, and spanned bases 117,071-128,610 (Genbank #NC_001798). The resulting plasmid, pHSV2-R L -UL56, contained the HSV-2 ICP0 gene but was too large for mutagenesis of the ICP0 gene. Excess DNA sequence was removed via a HindIII -MluI deletion (D117,071-120,806), and a subsequent AgeI -KpnI deletion (D126,058-128,610) to obtain pUC-HSV-2-ICP0, which contained bases 120,807-126,057 of the HSV-2 genome. Each mutant allele of the ICP0 gene was constructed in this background, as follows.
ii. p0DRING: The plasmid p0DRING was derived from p0D104 by deleting the DNA sequence between a BamHI and PmlI restriction site that spanned codons 105 to 162 (bases 124,055-124,229). The resulting plasmid p0DRING possessed an open-reading-frame that encoded amino acids 1 to 18 of HSV-2 ICP0, GFP, and amino acids 163 to 825 of ICP0 (Fig. 1A, 1B).
iii. p0D254: The plasmid p0D254 was created by a BamHI to PstI deletion of codons 105 to 254 in p0D104. This deletion resulted in a translational frameshift, such that codons 255 to 825 of ICP0 were not in the correct open-reading frame. Consequently, the plasmid p0D254 encoded amino acids 1 to 18 of HSV-2 ICP0 and GFP (Fig. 1A, 1B).
iv. p0D810: The plasmid p0D810 was created by replacing a NotI to AscI fragment that spans codons 19 to 810 of the HSV-2 ICP0 gene (bases 121,927-124, 893) with a GFP coding sequence flanked by matching NotI and AscI restriction sites. This GFP coding sequence was generated by PCR amplification off of the template peGFP-N1 (Clontech Laboratories) using the following oligonucleotide primers: NotI-GFP-a primer: 59 -ccgagcggccgctgagcaagggcgaggagctgt -39 and AscI-GFP-b primer: 59-gcgcg-ggcgcgcccagctcgtccatgccgag-39. Upon sub-cloning of the PCRamplified GFP coding sequence, the resulting plasmid p0D810 possessed an open-reading-frame that encoded amino acids 1 to 18 of HSV-2 ICP0, GFP, and amino acids 811 to 825 of ICP0 (Fig. 1A, 1B).
v. p0DNLS: The plasmid p0DNLS was derived from p0D104 by deleting DNA sequence between PpuMI and XhoI restriction sites that spanned codons 489 to 694. The plasmid p0DNLS possessed an open-reading-frame that encoded amino acids 1 to 18 of HSV-2 ICP0, GFP, amino acids 105 to 488 and amino acids 695 to 825 of ICP0 (Fig. 1A, 1B).
vi. Plasmid precursor of HSV-2 MS-GFP: The plasmid pUC-DLAT-GFP was derived from the parent plasmid pHSV2-R L -UL56, as follows. Excess DNA sequence was removed via a SapI -XhoI deletion to derive a plasmid, pUC-HSV-2 LAT, that contained bases 117,071-122,280 of the HSV-2 genome, which spanned most of HSV-2's LAT gene. A CMV-GFP expression cassette was subcloned from another plasmid into PvuII and BspEI restriction sites that flanked the transcriptional start site of HSV-2's LAT gene. Hence, a CMV-GFP expression cassette was placed in the same orientation of the LAT gene (i.e., antisense to the ICP0 gene) and replaced the equivalent of bases 119,359-119,530 of the internal R L region of the HSV-2 genome (Fig. S3).

Construction and isolation of HSV-2 recombinant viruses
Infectious HSV-2 DNA was prepared by a protocol that relies upon dialysis to minimize shearing of genome-length HSV-2 DNA; this is a modification of a protocol that was generously provided by Karen Mossman (McMaster University, Hamilton, Ontario). Five 100 mm dishes of Vero cells (3610 7 cells) were inoculated with 2 pfu per cell of HSV-2 strain MS and were incubated overnight at 34uC. After 24 hours, cells were scraped, centrifuged, rinsed with PBS, resuspended in 7.0 ml of 200 mM EDTA pH 8.0, and transferred into a 15 ml conical. Proteinase K (75 ml of 10 mg/ml) and 375 ml of 10% SDS were added to virusinfected cells, and the tube was incubated in a rotisserie (hybridization) oven with slow rotation at 50uC for 16 hours. Proteins were removed by phenol : chloroform extraction, DNA was transferred into a 0.5-3.0 mL Slide-a-lyzer cassette (10,000 MW cutoff; Pierce Chemical Co., Rockford, IL), dialyzed against 0.16 standard saline citrate for 24 hours, aliquoted and frozen at 280uC until use.
Recombinant HSV-2 viruses were generated by co-transfecting a 60 mm dish containing 8610 5 ICP0-complementing L7 cells with 1. 2 mg infectious HSV-2 MS DNA and and 2. 1 mg of each plasmid bearing a GFP + mutant allele of the HSV-2 ICP0 gene. After 6 hours, co-transfection medium was replaced with complete DMEM containing 1% methylcellulose and GFP + plaques were selected on the stage of a TE2000 fluorescent microscope (Nikon Instruments, Lewisville, TX). GFP + recombinant viruses were repeatedly passed in ICP0-complementing L7 cells until a uniform population of viruses was obtained that produced 100% GFP + plaques, at which time Southern blot analysis was used to confirm that the anticipated ICP0 2 mutant allele was transferred into HSV-2.
Western blot analysis to characterize GFP-tagged, mutant ICP0 proteins Vero cell cultures were established at a density of 3610 5 cells per well in 12-well plates, and were infected at an MOI of 2.5 pfu per cell. After 18 hours incubation at 34uC, proteins were harvested using mammalian protein extraction reagent (Pierce Chemical Co., Rockford, IL) supplemented with 1 mM dithiothreitol and protease inhibitor cocktail set I (Calbiochem, La Jolla, CA). After heat denaturation, 20 mg of each protein was resolved in a 10% polyacrylamide gel with a 4% stacking gel, and were transferred to nitrocellulose membranes. Protein blots were blocked in phosphate-buffered saline (PBS) containing 5% nonfat dry milk, and were incubated overnight at 4uC in PBS+0.1% Tween-20+5% nonfat dry milk containing a 1:1000 dilution of a rabbit polyclonal anti-GFP antibody (Clontech Laboratories Inc.). Following incubation with primary antibody, membranes were washed four times with PBS+0.1% Tween-20 (PBS-T), and were then incubated for 1 hour with a 1:20,000 dilution of goat antirabbit IgG conjugated to the infrared fluorescent dye IRDyeH 680 (LI-COR Bioscience, Lincoln, NE). Protein blots were washed three times in PBS-T, rinsed in PBS (to remove Tween-20), and were scanned for two-color fluorescence using the Odyssey Infrared imaging system, and data were analyzed using Odyssey application software version 3.0.16 (LI-COR Bioscience).

Southern blot analysis of ICP0 locus in HSV-2 ICP0 2 viruses
Cultures of L7 or Vero cells were established at a density of 1.5610 6 cells per plate in 60 mm dishes, and inoculated with MOIs of 2.5 pfu per cell for 24 hours at a temperature of 34uC. DNA was isolated, digested with the restriction enzymes Sac I and Stu I, and was separated on 1.2% agarose gels, blotted onto Zeta Probe GT nylon membranes (Biorad Laboratories, Hercules, CA), and hybridized with radiolabeled oligonucleotides specific for exon 2 of the HSV-2 ICP0 gene 59 -tgaagg tcgtcgtcagagattcccacctcggtctcctcct-39 or the GFP coding sequence (59-atagacgttgtggctgttgtagttgtactccagcttgtgc-39). Oligonucleotides were end-labeled with [a-32 P] dATP using terminal deoxynucleotidyl transferase (Promega Corporation, Madison, WI) and were hybridized to their target sequence via 16 hours of hybridization at 37uC in a solution containing 5 ng/ml labeled probe, 7% SDS, 120mM NaH 2 PO 4 , and 250mM NaCl. Excess probe was removed from membranes by sequential rinses in 0.16 standard saline citrate containing 0.1% SDS. Blots were exposed to phosphor screens, which were scanned and analyzed with a Cyclone PhosphorImager and OptiQuant software (Perkin Elmer, Boston, MA).
Measurements of interferon sensitivity of wild-type HSV-2 and HSV-2 ICP0 2 viruses i. Plaque assays. Cultures of ICP0-complementing L7 cells or Vero cells were established in 6-well plates at a density of 3.5610 5 cells per well in 2.0 ml complete DMEM in the morning. Six hours later, one-half of Vero cell cultures were treated by the addition of recombinant IFN-b to achieve a concentration of 200 U/ml, or 0.15 nM (PBL Biomedical Laboratories, Piscataway, NJ). Sixteen hours later, Vero cells were inoculated with log-dilutions of HSV-2 MS or each of the HSV-2 ICP0 2 mutant viruses. After allowing 45 minutes for adsorption, the viral inoculum was replaced with complete DMEM containing 0.5% methylcellulose. The subset of Vero cell cultures that were pretreated with IFN-b were overlaid with complete DMEM containing 0.5% methylcellulose and 200 U/ml IFN-b. Cultures were incubated for 72 hours to allow plaques to develop. Cell monolayers were fixed and stained with a solution of 20% methanol and 0.1% crystal violet, and plaques were counted.
ii. Growth curves. Cultures of ICP0-complementing L7 cells or Vero cells were established in 24-well plates at a density of 2610 5 cells per well in 0.5 ml complete DMEM in the morning. Six hours later, one-half of Vero cell cultures were treated by the addition of recombinant IFN-b to achieve a final concentration of 200 U/ml (PBL Biomedical Laboratories). Sixteen hours later, cultures of L7 cells or Vero cells were inoculated with 200 ml of a 100,000 pfu/ml inoculum of each virus to achieve an MOI of 0.1 pfu per cell. After allowing 45 minutes for adsorption, the viral inoculum was aspirated, cell monolayers were rinsed with 0.5 ml of complete DMEM, and each culture received a final volume of 0.5 ml complete DMEM. Vero cell cultures that were pre-treated with IFN-b received complete DMEM containing 200 U/ml IFNb. Cultures were incubated at 37uC and were transferred to a 280uC freezer at 3, 6, 12, 18, 24, 36, or 48 hours p.i. Upon thawing, viral titers were determined by a microtiter plaque assay on monolayers of freshly seeded L7 cells.

Inoculation of mice with HSV-2 viruses
Mice were first inoculated with HSV-2 at 6-to 10-weeks of age, and were handled in accordance with the NIH Guide for the Care and Use of Laboratory Animals. Female ICR mice were obtained from Harlan Sprague Dawley (Indianapolis, IN). Female strain 129 mice, rag2 2/2 mice, and stat1 2/2 mice were obtained from Taconic Farms (Germantown, NY).
Prior to viral inoculation, mice were anesthetized by i.p. administration of xylazine (7 mg/kg) and ketamine (100 mg/kg). Ocular inoculation of female mice with HSV-2 viruses was performed by scarifying the left and/or right corneas with a 26gauge needle and by placing 4 ml complete DMEM containing the stated amount of virus on each scarified eye. Viral titers in the ocular tear film of mice were determined at times after inoculation by swabbing the ocular surface of eyes with a cotton-tipped applicator, and transferring the tip into 0.4 ml complete DMEM. Viral titers were determined by a 96-well plate plaque assay on ICP0-complementing L7 cells cultured in complete DMEM containing 0.5% methlycellulose. After two to three days, cell monolayers were stained with a solution of 20% methanol and 0.1% crystal violet and plaques were counted.

Fluorescent microscopy of mouse eyes and mouse faces
Fluorescent photographs of the eyes and faces of mice infected with HSV-2 MS-GFP or HSV-2 ICP0 2 mutant viruses were obtained using a Nikon TE2000 inverted fluorescent microscope (Nikon Instruments, Lewisville, Tex.) fitted with an Olympus DP72 digital camera (Olympus America, Center Valley, PA) controlled by Olympus DP2-BSW microscope digital camera software. Mice were anesthetized by i.p. administration of xylazine (6.6 mg/kg) and ketamine (100 mg/kg) and placed face down on a clear petri dish. Photographs of eyes were captured with a 46 objective using a constant exposure and lighting conditions.

HSV-2 gD-antibody capture ELISA
Enzyme-linked immunosorbent assay (ELISA) was used to measure HSV-2 gD (gD 2 )-specific antibody titers in the serum of mice, as follows. Mice were bled on Day 60 post-inoculation by collecting blood from the right retroorbital sinus with heparinized, Natelson blood collecting tubes. The serum fraction was collected from centrifuged blood at 18 hours post-collection, and stored at 280uC until used in ELISA.
The coating antigen used in antibody-capture ELISA was a recombinant protein isolated from High Five TM insect cells infected with a gD 2 -306t-expressing baculovirus; this reagent was generously provided by Dr. Gary Cohen and Dr. Roslyn Eisenberg (University of Pennsylvania, Philadelphia, PA). The gD 2 -306t protein was engineered by Nicola, et al. (1996) to possess an N-terminal honeybee melittin secretion signal in place of gD 2 's native leader peptide, followed by amino acids 25-306 of HSV-2 gD and a C-terminal His 6 affinity-purification tag [50]. The gD 2 -306t protein was isolated as follows. A flask containing 2610 8 High Five TM insect cells was inoculated with 2 pfu per cell of HSV-2 gD 2 -306t-expressing baculovirus, and incubated while shaking at 27uC for 48 hours. Baculovirus-infected cells were removed by centrifugation, and secreted gD 2 -306t protein was purifed from supernatants, as follows. Supernatants were dialyzed against an excess of 20 mM Tris pH 8.0, 300 mM NaCl, and 10% glycerol overnight, and 10 mM imidazole was added to the dialyzed supernatant prior to affinity purification on a HisTrap TM HP column (GE Healthcare Biosciences, Piscataway, NJ) using an Ä KTApurifier TM fast-performance liquid chromatography system (GE Healthcare Biosciences). The gD 2 -306t protein was eluted from the column with 300 mM imidazole, and purity was verified at .90% by SDS-PAGE and Coomassie blue staining. Aliquots of gD 2 -306t were stored at -80uC.
High-binding EIA 96-well plates (Costar, Corning, NY) were coated overnight at 4uC with 100 ml per well of sodium carbonate buffer (pH 9.6) containing 1.5 mg per ml gD 2 -306t protein [50]. Wells were blocked for 2 hours with 400 ml of 2% dry milk dissolved in PBS+0.02% Tween-20 (polyoxyethylene-20-sorbitan monolaurate), hereafter referred to as PBS-T buffer. Mouse serum was diluted 1:100 in PBS+1% fetal bovine serum+0.02% Tween-20. After discarding blocking buffer from ELISA plates, duplicate 100-ml samples of 1:100 diluted mouse serum were added to gD 2 -306t-coated wells and were incubated for 2 hours. ELISA plates were rinsed seven times with an excess of PBS-T buffer prior to the addition of 100 ml secondary antibody diluted 1:2500 in PBS-T buffer; the secondary antibody was alkaline phosphatase-conjugated rabbit anti-mouse c chain (Rockland Immunochemicals, Gilbertsville, PA). After allowing 1 hour, secondary antibody was rinsed from plates seven times with PBS-T buffer, and 200 m1 of pnitrophenyl phosphate substrate (Sigma Chemical Co., St. Louis, MO) was added to each well, and colorimetric development (OD 405 ) was measured after a 30 minute incubation in an ELISA plate reader (Bio-tek Instruments, Inc., Winooski, VT).

Mathematical and statistical analysis of results
Viral titers were transformed by adding a value of 1 such that all data could be plotted and analyzed on a logarithmic scale. The significance of differences in log (IFN sensitivity), log (HSV-2 shedding per eye), log (gD 2 -antibody), and duration of survival was compared by one-way analysis of variance followed by Tukey's post hoc t-test. The significance of differences in survival frequency was determined by Fisher's Exact Test. All statistical analysis was performed using Instat v3.0 software (Graphpad Software, La Jolla, CA).
The quantitative relationship between color development in ELISA and abundance of anti-gD 2 antibody was defined, as follows. A standard curve was developed based on dilutions of pooled antisera from HSV-2 MS latently infected mice that spanned dilutions of 1:46 to 1:215,000. A hyperbolic tangentbased standard curve [71] of the form x = x 50 +DX N arctan OD 405 {y 50 DY was used to calculate gD 2 -antibody abundance (Fig. S3). Serum samples of naïve mice were used to define the background of the gD 2 -antibody capture assay, and the abundance of gD 2 -antibody in all serum samples was normalized to this background value. Figure S1 Amino acids in HSV-1 ICP0 and HSV-2 ICP0 are aligned to demonstrate the basis for concluding that the RING finger region is conserved at a level of 87% amino acid homology (shown in blue), the nuclear localization signal (NLS) region which includes flanking phosphorylation sites is conserved at a level of 65% amino acid homology (shown in brown), and the C-terminal oligomerization domain is conserved at a level of 86% amino acid homology (shown in red). The dilution factor of serum is shown above (red text), and the relative abundance of gD2antibody is expressed below in terms of the logarithmic increase over the background of the assay (black text). (B) The relationship between the logarithm (gD2-antibody abundance) and OD405 color development was described using a hyperbolic tangent-based standard curve. The equation, shown in panel B, relies on four constants defined by the standard curve, and which are briefly explained as follows: the data point x50, y50 represents the midpoint of the hyperbolic tangent (S-shaped curve); 2deltaX equals the range of gD2-antibody abundance over which 76% of the change in OD405 occurs; and deltaY equals one-half of the total change in OD405 that occurs. The curve-fitting methods used to derive hyperbolic tangent-based standard curves are described elsewhere by Halford,et al. [72]. Closed yellow circles indicate the actual OD405 (yellow color) observed, and the '+' symbols represent the values of OD405 predicted by the hyperbolic tangent equation over a 10,000-fold range of gD2antibody concentrations. This standard curve was used to calculate gD2-antibody abundance in the serum samples shown in Figure 6A.