INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Human herpes simplex virus (HSV) infections are common, occur at diverse sites, and have a wide range of symptoms, from inapparent to life-threatening encephalitis (2, 3). HSV infects mucosal surfaces, most commonly producing infections of the genitals, the mouth, or the eye. Greater than 90% of ocular HSV infections are due to HSV type 1 (HSV-1). HSV-1 infection of the eye can produce corneal inflammation and scarring as the result of an incompletely defined immunological response to the virus (1, 7, 14, 27, 32). This scarring is a major cause of corneal blindness (22, 35). In developed nations, HSV is the most frequent serious viral eye infection and is the most common cause of infectious blindness (22). In the United States, almost 500,000 people per year suffer primary or recurrent ocular HSV episodes that require doctor visits and medication (22). Over 1,000 corneal transplants per year are done in the United States as a direct result of HSV scarring (22, 35).

Experimental primary and recurrent ocular HSV-1 infection in the rabbit and the naturally occurring infection in humans share many characteristics. Primary infection is self-limited and is characterized by a benign transient infectious conjunctivitis. Infection of the cornea starts with epithelial keratitis, which destroys the corneal epithelium in a characteristic dendritic (neuron-like) and later geographic (amoeboid-like) pattern. As the keratitis, and its accompanying virus-induced immune-mediated inflammation, spread to the deeper part of the cornea (the stroma), the cornea becomes (temporarily) cloudy. Iritis (inflammation of the iris and anterior chamber) usually occurs only in eyes with severe keratitis. Corneal clouding and iritis follow, and are less common than, epithelial keratitis. In the rabbit, as in humans, HSV-1 epithelial keratitis with its dendritic and geographical spread is the hallmark of herpetic corneal infection. In fact, clinically, the dendritic ulcer is pathognomonic (i.e., characteristic or symptomatic of a particular disease) and requires no laboratory confirmation for the diagnosis of HSV infection. Epithelial keratitis is therefore the most relevant and important eye disease parameter in the studies described in this report.

The only Food and Drug Administration-approved treatments for primary HSV-1 ocular infection consist of the topical antivirals idoxuridine, vidarabine, and trifluorothymidine as well as oral acyclovir. Outside the United States, topical acyclovir is also used. Until recently, little attention has been given to the development of a vaccine against ocular HSV-1 infection. Virtually all work on HSV vaccine development has focused on the problem of genital HSV-2. Well-defined herpesvirus glycoprotein subunit vaccines have been developed by using recombinant DNA technology. These vaccines afford protective immunity when used prophylactically in mouse and guinea pig models of HSV-1 and HSV-2 disease (2, 8-11, 34, 36, 37).

Local immunity is likely to be especially important for mucosal surfaces such as the eye. In addition, previous HSV-1 infection at a nonocular site does not protect against nonprimary first episodes of ocular HSV-1 (4) or against recurrent ocular HSV-1 infection. Therefore, it was of interest to determine if ocular vaccine administration would be more effective than systemic vaccine administration as prophylaxis against primary infection and the resulting eye disease in rabbits following ocular challenge with a highly pathogenic HSV-1 strain (McKrae). Two different vaccines, both expected to be less than optimal, were used. Both vaccines afforded protection against lethal ocular challenge regardless of the route of administration. However, neither vaccine provided any protection against ocular disease when administered systemically. In contrast, both vaccines provided protection against ocular disease when administered periocularly, thus supporting the notion that ocular immunity is more important than systemic immunity in protecting against eye disease.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Virus. The challenge virus, HSV-1 McKrae, produces severe ocular disease in rabbits (25). The live-virus vaccine strain, HSV-1 KOS, is nonvirulent and produces no significant ocular disease in rabbits. Both viruses were triple plaque purified and prepared as previously described (25).

Rabbits. Eight to ten-week-old New Zealand White male rabbits (Irish Farms) were used for all experiments. Rabbits were housed and handled in accordance with Association for Research in Vision and Ophthalmology, American Association for Laboratory Animal Care, and National Institutes of Health guidelines.

Glycoproteins and adjuvant for subunit vaccine. HSV-2 glycoproteins B (gB2) and D (gD2) were prepared by expression of the modified genes in Chinese hamster ovary cells followed by purification to near homogeneity, using a series of traditional chromatographic steps as previously described and as previously used by Chiron Corp. in human clinical trials of a vaccine for genital herpes (20). The adjuvant MF59 was prepared as previously described (23). The vaccine was prepared by mixing 1 volume of gB2 plus gD2 in 2× phosphate-buffered saline with 1 volume of MF59.

Systemic vaccination. Rabbits received three inoculations at 3-week intervals. Each inoculation with gB2/gD2/MF59 was delivered by a single intramuscular (i.m.) injection on one side of the lower back. Each dose contained 25 µg of each glycoprotein in a total volume of 0.1 ml. Three systemic vaccinations with live HSV-1 KOS were given intradermally 3 weeks apart. Each intradermal vaccine dose was divided into four or five 0.1-ml aliquots and was injected into separate sites on the back for a total dose of 2 × 10⁷ PFU of live HSV-1 KOS.

Control vaccine. The control vaccine was the adjuvant MF59 without glycoprotein (1 volume of MF59 plus 1 volume of 2× PBS) and was delivered identically and on the same schedule as the gB2/gD2/MF59 systemic vaccine.

Subconjunctival vaccinations with the gB2/gD2 vaccine. Inoculations were given as previously described (23), with each eye receiving three inoculations at 3-week intervals. Each inoculation contained 7.5 µg of each glycoprotein in 0.1 ml. Vaccination resulted in approximately 25% of the eyes showing mild to moderate conjunctival inflammation for up to 7 days.

Topical ocular vaccination with HSV-1 KOS. Eyes were vaccinated twice at a 3-week interval with 2 × 10⁵ PFU of live KOS per eye as described below for HSV-1 challenge with McKrae. Since KOS produces no stromal keratitis and requires corneal scarification to produce epithelial keratitis (17, 26a, 38), the method used for inoculating with KOS resulted in no clinically recognizable disease.

Ocular challenge of rabbit eyes with HSV-1 McKrae. Vaccinated and mock-vaccinated rabbits were bilaterally infected without scarification or anesthesia by placing 2 × 10⁵ PFU (HSV-1 McKrae), in a total volume of 0.1 ml, into the conjunctival cul-de-sac, closing the eye, and rubbing the lid gently against the eye for 30 s (31). In naive rabbits, this dose of virus infects all eyes and produces moderate to severe ocular disease in about 90% of eyes. It results in the death of approximately 30 to 50% of the rabbits within 18 days (23, 25, 29, 30). Animals were challenged 3 weeks following the final dose of vaccine.

Measurement of titers. HSV serum neutralizing antibody titers were measured as previously described (21), using an HSV-2 plaque reduction neutralization assay in the presence of added complement with twofold serum dilutions. The reported titer is the reciprocal of the serum dilution required to inhibit the cytolysis of a confluent monolayer of Vero cells by 50%.

Determination of clinical eye disease. Clinical eye disease patterns were determined by examining the rabbit eyes in a masked fashion on days 3, 5, 7, 10, and 14 postinfection for scoring the incidence and severity of conjunctivitis, iritis, dendritic and geographic ulcers characteristic of HSV (epithelial keratitis), and acute transient stromal keratitis and edema (corneal clouding). Epithelial keratitis was evaluated by slit lamp biomicroscopy using 0.75% fluorescein stain (26). Conjunctivitis was determined by direct visual observation. The magnitude of epithelial disease was scored as 0.25, 0.5, 0.75, 1, 1.5, 2, 2.5, 3, 3.5, or 4, with 0, 1, 2, 3, and 4 representing no disease and disease involving 25, 50, 75, and 100% of the corneal surface, respectively. The levels of inflammatory severity of conjunctivitis, iritis, and stromal keratitis were assessed by using the same scale, with 0, 1, 2, 3, and 4 representing no inflammation, mild but recognizable inflammation, moderate easily recognizable inflammation, moderately severe inflammation, and very severe inflammation, respectively. To eliminate bias, these gradings were done in a masked fashion by readers highly experienced in this system.

Statistical analyses. Statistical analyses were performed with Instat, a personal computer software program. Tests included the Student t test and the Fisher exact test.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Groups of naive rabbits were vaccinated either systemically or periocularly with either a gB2/gD2 subunit vaccine or live KOS, a nonvirulent HSV-1 strain that produces no eye disease in rabbits. The five experimental groups were as follows: KOS ocular vaccine, 11 rabbits; KOS systemic vaccine, 15 rabbits; gB2/gD2 ocular vaccine, 15 rabbits; gB2/gD2 systemic vaccine, 15 rabbits; and control (systemic adjuvant), 16 rabbits.

Protection against mortality. Three weeks after the final vaccination, rabbits were ocularly challenged with HSV-1 McKrae. Protection against mortality was similar for each vaccine, regardless of the vaccination route (Fig. 1). The KOS vaccines were significantly more efficacious than the control (P = 0.008 [ocular] and 0.002 [systemic]; Fisher exact test). Survival in the individual gB2/gD2 vaccine groups was not significantly more than for controls (P = 0.14 [systemic] and 0.054 [ocular]; Fisher exact test). However, since the two gB2/gD2 vaccine groups were similar, the data could be combined to increase the power of the analysis. This resulted in significant protection compared to mock vaccination (P = 0.036; Fisher exact test).

View larger version (23K): [in this window] [in a new window]   FIG. 1.   Survival of vaccinated rabbits following ocular challenge with HSV-1 McKrae. Rabbits were vaccinated either i.m. or ocularly with either gB2/gD2 or HSV-1 KOS as described in Materials and Methods. The mock-vaccinated control (Cont) group was vaccinated i.m. with adjuvant alone. Three weeks after the final vaccination, all rabbits were challenged with 2 × 105 PFU of HSV-1 McKrae in both eyes as described in Materials and Methods. Survival was determined 3 weeks after challenge. The ratio above each bar shows the number of surviving rabbits/the number of challenged rabbits. Comparisons were done as follows: periocular HSV-1 KOS group versus mock control group (P = 0.008 [Fisher exact test]; systemic HSV-1 KOS group versus mock control group (P = 0.002); periocular gB2/gD2 versus mock (P = 0.054); systemic gB2/gD2 versus mock (P = 0.14); combined gB2/gD2 groups versus mock (P = 0.036 [double-sided Fisher exact test]; P = 0.02 [single-sided Fisher exact test]).

Protection against conjunctivitis. No differences in the average severity of conjunctivitis were observed between mock-vaccinated rabbits and rabbits vaccinated systemically with gB2/gD2 or KOS on any of the days examined (Fig. 2A) (P > 0.05; Student t test). In contrast to systemic vaccination, when the route of vaccination was periocular, both the subunit gD2/gB2 and the live KOS vaccines appeared to provide protection against conjunctivitis. In rabbits vaccinated ocularly with KOS, the average peak severity of conjunctivitis (day 5 following ocular challenge) was significantly less than in the control rabbits and in rabbits systemically vaccinated with KOS (Fig. 2B; P = 0.048 and 0.004, respectively; Student t test). Rabbits vaccinated ocularly with gB2/gD2 also appeared to have less conjunctivitis than control rabbits, but the difference did not quite reach statistical significance (P = 0.08; one sided). However, the gB2/gD2 ocularly vaccinated rabbits had significantly less conjunctivitis than the gB2/gD2 systemically vaccinated rabbits (Fig. 2B; P = 0.008). Thus, for both vaccines, compared to systemic vaccination, ocular vaccination provided significantly more protection against conjunctivitis.

View larger version (32K): [in this window] [in a new window]   FIG. 2.   Protection against eye disease. Rabbits were vaccinated three times either systemically or periocularly, and ocularly challenged, and eye disease was monitored on days 3, 5, 7, 10, and 14 as described in Materials and Methods. Each time point represents the average readings of both eyes from 11 to 16 rabbits from vaccinated or mock-vaccinated rabbits (ocular KOS vaccine, 11 rabbits; systemic KOS vaccine, 15 rabbits; ocular gB2/gD2 vaccine, 15 rabbits; systemic gB2/gD2 vaccine, 15 rabbits; adjuvant control, 16 rabbits). Single asterisks indicate points that are significantly less than for the mock-vaccinated control group at the 95% level, using the Student t test; double asterisks indicate values for the periocular vaccine group that are significantly less than values for the corresponding systemic vaccine group. Statistical analyses were done on a per-rabbit basis. Thus, the readings from both eyes were averaged for each rabbit, and the resulting value for each rabbit was used for the analyses (i.e., n for each group was the number of rabbits, not the number of eyes, in the group). The same groups showed statistical significance when analyses were done on a per-eye basis (not shown).

Protection against acute herpetic iritis. Iritis is characterized by anterior chamber inflammation (cells and fibrin) and redness of the normally pink iris. Prophylactic systemic vaccination with gB2/gD2 or KOS did not reduce the peak severity of iritis (day 7 after ocular challenge) (Fig. 2C; P > 0.05). In contrast, prior periocular vaccination with gB2/gD2 resulted in significant reductions in average iritis severity compared to control and systemic gB2/gD2 vaccination on day 7 postchallenge (Fig. 2D; P = 0.02 and 0.03, respectively). Ocular vaccination with KOS also significantly reduced the peak average iritis severity compared to control and the corresponding systemic vaccination (Fig. 2D; P = 0.0008 and 0.03, respectively). Thus, as with conjunctivitis, periocular vaccination provided much more protection against manifestations of herpetic iritis than did systemic vaccination.

Protection against corneal clouding. Corneal clouding (a transient corneal edema with corneal inflammation) is a measure of stromal keratitis. Systemic vaccination with gB2/gD2 or KOS did not affect the severity of corneal clouding (Fig. 2E). Periocular vaccination with gD2/gB2 also did not lessen stromal keratitis (Fig. 2F; P > 0.05). In contrast, periocular vaccination with KOS completely eliminated corneal clouding. Because of the low overall levels of corneal clouding in all groups, protection against the severity of corneal clouding by ocular KOS vaccination on the day of peak disease reached statistical significance only by a single-sided analysis (Fig. 2F; P = 0.04).

Protection against epithelial keratitis. HSV-1 produces corneal epithelial cell loss in a characteristic dendritic pattern. As the dendrite expands, it develops smoother edges and takes on a geographic pattern. Epithelial keratitis is the combination of these dendritic and geographic lesions. Epithelial keratitis is the key pathologic finding indicative of ocular HSV-1 infection and is the clinical hallmark of acute herpetic corneal disease. Iritis and clouding are not present without epithelial keratitis.

Vaccine immunogenicity. All rabbits were bled 3 weeks after the final vaccination, just prior to challenge with HSV-1. Neutralization and ELISAs (against HSV-2 and HSV-2 glycoproteins) were done individually on all sera (Tables 1 and 2). Both vaccines induced HSV-2 neutralizing antibody titers significantly greater than those for the adjuvant control, regardless of the route of vaccination (Tables 1 and 2). With both vaccines, the systemic vaccination showed a tendency to induce a higher neutralizing antibody titer than ocular vaccination, but the differences were not statistically significant.

View larger version (22K): [in this window] [in a new window]   FIG. 3.   HSV-specific tear sIgA. Rabbits were vaccinated once, either systemically or periocularly, with gB2/gD2 and ocularly challenged 21 days later as described in Materials and Methods. Tears were collected at the times indicated, and the relative amount of HSV-1 gD2-specific sIgA in individual tear samples was determined by ELISA. The arrows indicate vaccination (day 21) and ocular challenge (day 0) as labeled. The asterisk indicates that both vaccine groups had significantly higher ELISA titers than the mock vaccine group (P < 0.05).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

We previously showed that periocular vaccination of latently infected rabbits with gB2/gD2 in the adjuvant MF59/MTP-PE provided therapeutic protection against recurrent HSV-1, as judged by a statistically significant decrease in ocular spontaneous recurrent virus shedding (23) and spontaneous recurrent corneal disease (24). The present study, comparing the abilities of periocular and systemic vaccinations to protect against primary ocular HSV-1 challenge, was undertaken as part of our inquiries into the role and efficacy of periocular vaccination against primary and recurrent ocular herpes infection. The subconjunctival route used for periocular vaccination allows the use of adjuvant and ensures that the vaccine is delivered and retained at the local site. Subconjunctival injection is routinely used in clinical ophthalmology.

In this study, two vaccines that were expected to be less than completely effective were chosen. This was done because the goal was to determine if local periocular vaccination provided better ocular protection than systemic vaccination. Obviously, a vaccine that gave 100% protection against eye disease regardless of the vaccine route would not allow us to easily determine which route was more effective. A live and a subunit vaccine were both used to determine if any route specificity found would be consistent for different types of vaccines.

gB2/gD2 with MF59 was chosen as the suboptimal subunit vaccine because both glycoproteins are type 2 rather than type 1 (the McKrae challenge virus is type 1) and because the adjuvant lacks the added immunostimulant MTP-PE that was included in our recent successful therapeutic vaccine studies (23, 24). The avirulent HSV-1 KOS strain was chosen as the less than optimal live vaccine for this study because compared to other available wild-type HSV-1 strains (such as the virulent McKrae strain used as the challenge virus), its ability to replicate and spread in the rabbit is low. Thus, it was expected that HSV-1 KOS would be a suboptimal live HSV-1 vaccine. This hypothesis was supported by a small pilot study which showed that rabbits vaccinated systemically with HSV-1 McKrae, but not with HSV-1 KOS, were completely protected against eye disease (not shown).

The increased vaccine efficacy of periocular compared to systemic vaccines against HSV-1-induced ocular disease (summarized in Table 3) was even more impressive, since to strengthen the likelihood that any increased protection observed with ocular vaccination would be meaningful, the vaccinations were biased toward systemic vaccinations. Thus, a lower dose of gB/gD was used periocularly (7.5 µg of each glycoprotein per eye versus 25 µg of each glycoprotein given systemically) and a lower dose of virus and fewer periocular vaccinations were given with HSV-1 KOS (2 × 10⁵ PFU/eye versus 2 × 10⁷ PFU systemically; two periocular vaccinations versus three systemic vaccinations). Because suboptimal vaccines were specifically chosen for these studies, even ocular vaccination did not always provide significant protection against eye disease. However, in all situations in which differences in ocular protection were detected between the systemic and periocular routes of vaccine, the local periocular vaccine was superior.

As judged by neutralization and ELISA titers, both vaccines used in this study induced stronger humoral immune responses when the vaccine was delivered via a systemic route (i.m. or subcutaneous) than via an ocular route (topically on the cornea or subconjunctival injection). This finding suggests that in the rabbit ocular model of primary HSV-1 infection, the ability of a vaccine to induce serum neutralization titers and serum ELISA titers is not predictive of vaccine efficacy against eye disease and supports the hypothesis that vaccine efficacy against ocular disease is due to local/mucosal immunity and not systemic immunity.

The ability of these vaccines to induce tear sIgA specific for HSV-1 also did not appear to be predictive of vaccine efficacy against eye disease. In fact, there was an opposite tendency. Systemic vaccination tended to produce higher peak tear sIgA titers than did the more efficacious ocular vaccination. Furthermore, the peak tear sIgA titers produced following ocular challenge of mock-vaccinated rabbits tended to be higher than in either vaccine group. This type of result, in which following ocular challenge, the highest local ocular immune response occurs in the least well protected group, suggests that the higher immune response is the result of more virus replication in the eye due to the poorer efficacy of the vaccine. This, of course, can complicate the correlation of specific immune responses with vaccine efficacy and was similar to some of our previous findings in mice, in which following ocular HSV-1 challenge, fewer infiltrating immune cells were detected in the corneas of mice vaccinated with the more efficacious vaccines (14).

Interestingly, vaccine efficacy against mortality was similar regardless of vaccine route. Thus, vaccine efficacy against eye disease appeared to require local or mucosal immune responses at the eye, while vaccine efficacy against mortality could be obtained by systemic immunity. Systemic immunity could protect against mortality (due to viral encephalitis) without reducing eye disease by reducing viral replication in the trigeminal ganglia or the brain, or by reducing transit of virus between the eye and the trigeminal ganglia or between the trigeminal ganglia and the brain.

Various different immune responses have been implicated as being most important in protecting the mouse eye against HSV-1 infection. CD4⁺ T cells and CD8⁺ T cells have alternatively each been reported to protect against ocular HSV-1 and to be responsible for HSV-1 ocular disease (6, 13, 15, 16, 19, 27, 28, 33). In the mouse model, there is a very strong correlation between the ability of a vaccine to induce anti-HSV-1 serum antibody titers and vaccine efficacy against HSV-1 ocular infection (5, 12). Even intraperitoneal administration of neutralizing antibody can completely block HSV-1-induced ocular disease (18, 39). In contrast, in this report we did not find any correlation between serum antibody and protection against ocular disease in the rabbit. Serum antibody in humans also does not appear to protect against ocular HSV-1, since individuals with high rates of recurrent ocular HSV-1 often develop very high HSV-1 neutralizing antibody titers yet continue to have recurrent episodes. Thus, it appears that the immune responses (and vaccine efficacy) involved in protecting the mouse eye against HSV-1 may not be predictive of vaccine efficacy in humans. In particular, since serum antibody alone can protect the mouse eye against ocular HSV-1, using the mouse model as the sole basis of understanding vaccine efficacy against ocular HSV-1 may cause us to underestimate the importance of vaccine-induced local/mucosal immunity in humans. Thus, although the state of the art in rabbit immunology still lags significantly behind that of the mouse, the rabbit may be a more useful model for studying vaccine efficacy against primary and recurrent ocular HSV-1.

Why does humoral immunity protect the mouse eye and not the rabbit or human eye against ocular HSV-1? One possibility is the smaller size of the mouse eye. In rabbits and humans, capillaries are seen only in the outer 1 mm of the cornea, effectively isolating the central cornea from circulating immune factors. In the mouse, capillaries are also confined to the outer 1 mm of the cornea. However, because the mouse cornea is smaller than the corneas of rabbits and humans, circulating immune factors can rapidly diffuse from these peripheral capillaries into the central cornea, thus allowing serum antibody to protect the mouse cornea.

The ability to analyze cell-mediated immunity and ocular mucosal immune factors (other than tear sIgA) in the rabbit is still in its infancy compared to the situation for mice. Thus, the local/mucosal immune factors responsible for the vaccine efficacy against HSV-1-induced ocular disease in this study have not yet been determined. Although the sIgA analyses reported here suggest that tear sIgA is not a key protective immune response against primary HSV-1 ocular challenge, preliminary studies using a recurrent ocular HSV-1 vaccine model suggest that tear sIgA may play a role in protection against recurrent ocular HSV-1.