INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Following infection, herpes simplex virus (HSV) replicates in the epithelium, gains access to axonal terminae, and is retrogradely transported to sensory ganglia. A short period of viral replication within infected sensory ganglia is concomitant with the establishment of a nonproductive latent infection (22). During acute infection, HSV may reemerge from the nervous system and cause disease within the same dermatome at a location distal from the initial site of infection (4, 7, 18, 24, 32). This phenomenon, termed zosteriform spread, has been shown in mouse models to progress from flank to flank, snout to eye, mouth to eye, and neck to pinna. Spread requires an intact nerve supply and has been likened to viral reactivation in that the virus must traverse the nervous system to cause disease at a distal site (26-28). Zosteriform models have been used to determine the effects of antivirals, vaccines, or the immune response on disease (2, 8, 11, 15, 17, 21, 27, 28, 34).

The mouse eye model of HSV infection has provided much information regarding eye diseases such as stromal keratitis (13, 31). Little is known, however, about the progression to and cause of periocular diseases, such as blepharitis and conjunctivitis, despite their high prevalence in HSV-infected individuals (19). Primary corneal infection of mice with HSV-1 results in severe corneal and periocular disease (20, 32). Abundant progeny virus is observed in the eyes and periocular skin throughout acute infection, but due to the close anatomical proximity of these tissues the primary site of viral replication is unknown. Blepharoconjunctivitis, periocular hair loss, and ulcerative lesions become apparent 4 days postinfection and progress in severity until day 15, after which they resolve (29). Interestingly, recombinant viruses that replicate in the cornea but not in the nervous system fail to cause periocular disease (5). We hypothesize, therefore, that periocular disease occurs by viral spread from the cornea to the periocular tissue via the trigeminal ganglia rather than by direct spread from the cornea to the skin.

In this study, recombinant reporter viruses were constructed which express luciferase or -galactosidase under the control of immediate-early (IE) or early (E) gene promoters. These viruses were used to monitor and quantify viral gene expression and replication in the mouse ocular model. Reporter gene activity peaked in the eyes at 24 h, in trigeminal ganglia at 72 h, and in periocular tissue at 84 to 96 h. Periocular disease followed shortly thereafter. A thymidine kinase-deleted reporter virus expressed high levels of reporter activity in eyes, but no reporter gene activity was detectable in the trigeminal ganglia or periocular tissue. This demonstrates a requirement for viral replication within the nervous system for delivery to the periocular tissue and demonstrates that zosteriform spread from the cornea to the eyelids leads to periocular disease. In addition, we discuss clinical evidence that a similar progression results in human periocular disease.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cells and virus. African green monkey kidney (Vero) cells were propagated as previously described (23) and were used for the in vitro growth and reporter assays of all viruses. Growth curve experiments were performed at the appropriate multiplicity of infection (MOI) on 24-well plates seeded 18 to 20 h previously with 10⁵ Vero cells/well. All virus stocks were propagated on Vero cells as previously described (23). All viruses were constructed in the context of the KOS strain of HSV-1. The thymidine kinase null mutant, dl8.36tk, referred to in this study as KOS6tk, has been previously described (16).

Plasmids and generation of recombinant viruses. A reporter plasmid, pDlux, was generated with both firefly and Renilla luciferase genes in a divergent orientation from a single multicloning site. The plasmid was constructed by NheI/SalI digestion of pGL3-Basic and pRL-null plasmids (Promega, Madison, Wis.) followed by isolation and ligation of regions containing the firefly and Renilla genes. A BamHI/NruI 822-bp fragment encoding the origin of replication S (oriS) and flanking ICP4 and ICP22/47 promoters was isolated and blunt-end ligated into the NheI site pDlux to yield pDlux/oriS. The Dlux/oriS cassette and the previously characterized pD6p cassette (10) encoding -galactosidase under the regulation of the ICP6 promoter were cloned into the BglII site at position 106750 of plasmid pUIC to yield pUIC/Dlux/oriS and pUIC6 (D. J. Davido, D. A. Leib, and P. A. Schaffer, submitted for publication). pUIC-based clones were linearized with PstI and cotransfected with KOS infectious DNA. Resulting viruses were selected based on their ability to express either -galactosidase or luciferase. Putative recombinant viruses were plaque purified three times, and identity was confirmed via Southern blot hybridization (data not shown).

Growth curve and gene expression analysis. Vero cells were seeded at 10⁵ cells/well in 24-well plates. Eighteen to twenty hours postseeding, cells were infected with virus at an MOI of 0.1 or 5. After adsorption for 1 h at 37°C, virus was aspirated, cells were washed, and 1 ml of medium was replaced. Medium was removed at indicated hours postinfection, frozen, thawed, and titered on Vero cells. At the time of harvest, cells were lysed in 100 µl of Passive Lysis Buffer (Promega), frozen, thawed, and assayed for firefly and Renilla luciferase activity using the Dual Luciferase Assay kit (Promega) or for -galactosidase activity using previously published methods (25). Cycloheximide reversal experiments were performed as described previously (30) with minor alterations. Briefly, pretreatment of cells with 100 mg of cycloheximide per ml for 1 h was followed by infection as above at an MOI of 5 in the presence of drug. Eight hours postinfection. cells were washed and medium was replaced with 10 mg of actinomycin D per ml. Twelve hours postinfection cells were harvested in 100 µl of Passive Lysis Buffer, and luciferase of -galactosidase activity was measured.

Animal procedures. Outbred CD-1 female mice (body weight, 21 to 25 g; Charles River Breeding Laboratories, Inc., Kingston, N.Y.) were anesthetized with ketamine and xylazine, their corneas were bilaterally scarified, and they were inoculated with 2 × 10⁶ PFU of virus in a volume of 5 µl as previously described (23). Tear film material was assayed for virus as previously described (23). Whole eyeballs, trigeminal ganglia, and 6-mm biopsy punches of periocular tissue were removed, placed in dry tubes containing 1-mm diameter beads, and frozen. Upon harvest, 100 µl of reporter buffer (50 mM HEPES [pH 7.5], 1 mM EDTA, 1% gelatin, 5% glycerol, 5 mM dithiothreitol) was added, and tissue was homogenized via beadbeating (Mini-Beadbeater-8; Biospec Products, Bartlesville, Okla.) and sonication. Ten microliters of homogenate was removed, added to 990 µl of medium, frozen, and thawed, and titers were determined on Vero cells. Twenty microliters of the same lysate was assayed for firefly and Renilla luciferase activity or -galactosidase activity. Data were analyzed throughout using Student's t-test.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

In vitro replication and reporter gene expression kinetics. To validate the growth of the viruses used in this study, KOS, KOS6, KOS6tk, and KOS/Dlux/oriS were examined in Vero cells in both single- and multiple-step growth curves (Fig. 1A through D and 2A through C). Monolayers were infected at MOIs of 0.1 (data not shown) and 5 PFU/cell, and at various times postinfection supernatants were titered for infectious virus and cell monolayers were harvested to measure luciferase or -galactosidase activity. The kinetics of viral growth and egress were similar among all viruses tested (Fig. 2A through C). In addition, reporter gene activity was detectable early in infection and rose concomitantly and proportionally with the detection of viral progeny.

View larger version (15K): [in this window] [in a new window]   FIG. 1.   Maps of reporter gene viruses used in this study. (A) Prototypical arrangement of the HSV-1 genome, showing the unique long (UL) and unique short (US) segments flanked by internal (a', b', c') and terminal (a, b, c) repeats. The locations of reporter cassette insertions and the thymidine kinase deletion are shown. (B) KOS6 was constructed by insertion of a cassette containing the early ICP6 promoter regulating the expression of -galactosidase (10) into the BglII site at map position 106750. (C) KOS6tk was constructed by a 360-bp deletion of the thymidine kinase optical reading frame within the context of KOS6 (16). (D) KOS/Dlux/oriS was constructed by insertion of a cassette carrying the ICP4/22/47 and oriS regulatory region driving firefly and Renilla luciferase expression into the BglII site.

View larger version (20K): [in this window] [in a new window]   FIG. 2.   Single-step growth kinetics and reporter gene activity for KOS (A), KOS6 and KOS6tk (B), and KOS/Dlux/oriS (C). Vero cells were infected at an MOI of 5. Data represent standard errors of the means of three independent experiments. The limit of detection is 10 PFU/ml. RLU, relative light units.

View larger version (28K): [in this window] [in a new window]   FIG. 3.   Reporter gene activity under cycloheximide reversal or control conditions. Vero cells were pretreated for 1 h and then were infected at an MOI of 5 in the presence or absence of cycloheximide (100 µg/ml) for 8 h, after which time cells were washed and plated in the presence of actinomycin D (10 µg/ml) for 4 h. Monolayers were harvested and assayed for reporter activity. RLU, relative light units.

Clinical disease in infected animals. Mice were infected with KOS/Dlux/oriS, KOS6, or KOS6tk and scored for periocular disease using a masked, semiquantitative scale (Fig. 4). Little or no clinical disease was observed until day 5 despite detectable viral titers within the periocular tissue from 24 to 96 h postinfection. Clinical scores of mice infected with KOS/Dlux/oriS and KOS6 increased from days 5 to 8 and in accordance with previously published studies involving wild-type virus (29). KOS6tk-infected mice, however, displayed no clinical disease at any time scored (Fig. 4).

View larger version (21K): [in this window] [in a new window]   FIG. 4.   Periocular disease scores in mice following corneal scarification and infection with 2 × 106 PFU per eye of KOS/Dlux/oriS, KOS6, and KOS6tk. Animals were scored as follows: 0, no lesions; 1, minimal eyelid swelling; 2, moderate swelling and crusty ocular discharge; 3, severe swelling, moderate periocular hair loss, and skin lesions; 4, severe swelling with eyes crusted shut, severe periocular hair loss, and skin lesions. Data represent combined averages of at least 10 mice per time point.

In vivo growth kinetics and gene expression. To determine viral IE and E gene expression patterns during acute infection, mice were infected via the scarified cornea with KOS/Dlux/oriS and KOS6 (Fig. 5 and 6). At various times postinfection mice were sacrificed and titers were measured in whole globes, trigeminal ganglia, periocular skin, and the tear film via corneal swab. Reporter gene expression was also measured in the same samples harvested for viral titration in order to correlate the presence of virus with infectious centers within adjacent tissues.

View larger version (30K): [in this window] [in a new window]   FIG. 5.   In vivo growth and reporter gene expression of KOS/Dlux/oriS after ocular infection. Mice were infected via corneal scarification and inoculation of 2 × 106 PFU per eye. At various times postinfection tissues were harvested and assayed for infectious virus and luciferase activity. Data points represent 12 tissues from two independent experiments with three mice. In periocular skin, both luciferase activity and titer were significantly increased (P < .05) at 84 h relative to earlier time points. RLU, relative light units.

View larger version (34K): [in this window] [in a new window]   FIG. 6.   In vivo growth and reporter gene expression of KOS6 and KOS6tk after ocular infection. Mice were infected via corneal scarification and inoculation of 2 × 106 PFU of KOS6 or 8.5 × 107 PFU of KOS6tk per eye. At various times postinfection tissues were harvested and assayed for infectious virus and -galactosidase activity. Data points represent 12 tissues from two independent experiments with three mice. In KOS6 infections, -galactosidase activity and titer were significantly increased (P < 0.05) at 84 h relative to earlier time points and significantly higher than those of KOS6tk (P < 0.05) from 72 h onward.

View larger version (15K): [in this window] [in a new window]   FIG. 7.   Growth of KOS6 and KOS6tk in periocular tissue explants after ocular infection. Mice were infected via corneal scarification and inoculation of 2 × 106 PFU of KOS6 or 8.5 × 107 PFU of KOS6tk per eye. Twelve hours postinfection tissue was harvested and explanted in medium. Supernatants were titers at various times postinfection. Data represent the mean of data from four mice from two independent experiments.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Reporter gene viruses have been previously utilized to monitor and elucidate viral promoter regulation (3, 14, 33). Recombinant viruses containing promoter-reporter cassette insertions into genes dispensable for growth (e.g., tk and gC) have been useful for elucidating promoter elements during the lytic life cycle in vitro. Such viruses, however, face limitations with respect to pathogenesis in vivo due to insertions into loci which may be virulence determinants. To address these limitations, we utilized a locus whose disruption did not significantly affect growth in vitro and pathogenesis in vivo (this study and data not shown). This locus was a BglII site (map position 106750) which interrupts the C-terminal 12 amino acids of the UL49.5 gene (1). Insertion at this site has no effect on protein synthesis, signal sequence cleavage, protein targeting of UL49.5, or viral growth in vitro. Insertions at this site have only a limited effect on in vivo parameters, including acute replication within the eyes, trigeminal ganglia, and periocular tissue, disease kinetics and severity, as well as reactivation from explanted trigeminal ganglia (this study and Davido et al., submitted). These viruses have clear utility for the study of viral spread and gene expression during acute infection as well as for future experiments examining the efficacy of antivirals and immunity in limiting viral gene expression.

Utilizing these reporter gene viruses, we have demonstrated that HSV-1 progresses from the corneal surface to the periocular tissue via the trigeminal ganglion. This progression was observed irrespective of reporter gene measured or the kinetic class of promoter regulating its expression. The onset of periocular disease correlated with increased viral titer and reporter gene activity within this tissue. The pattern of viral titer, reporter gene activity, and disease development is consistent with periocular disease being a manifestation of zosteriform spread through the nervous system rather than direct spread from cornea to skin.

Although zosteriform spread appeared to be the primary mode of dissemination, the close anatomical proximity of the eye and skin does allow limited direct infection of the periocular tissue. Indeed, small amounts of reporter activity and viral titer are detectable within the periocular tissue at 24 h. In addition, mice develop a mild, transient blepharoconjunctivitis 24 to 36 h postinfection. Infection with a thymidine kinase-deleted virus, however, did not cause disease despite its demonstrated ability to directly infect and replicate within the periocular tissue. In agreement with these results, ribonucleotide reductase mutants are also unable to cause appreciable periocular disease after corneal infection (5). We therefore conclude that limited direct spread does occur, but viral replication within the sensory nervous system and centrifugal spread are necessary for the development of periocular disease.

The experiments above have detailed a requirement for zosteriform spread of HSV from the cornea to the periocular tissue via the sensory nerves to cause disease in mice. Despite the obvious contribution of the mouse model to our understanding of ocular disease, the relevance of such observations to human herpetic eye disease should be considered. For humans, a number of reports have detailed periocular skin involvement after ocular infection with HSV (9, 12 and S. Yamamoto, Y. Shimomura, S. Kinoshita, and Y. Tano, Letter, Arch. Ophthalmol. 112:1515-1516, 1994). In a case study of 122 HSV ocularly infected individuals, 54% exhibited eyelid and conjunctival disease during their first episode and 31% during recurrent episodes (19). This previous study suggests that periocular involvement following ocular infection is highly prevalent in HSV-infected individuals. We detail here a case of human ocular HSV disease in which the timing of periocular involvement relative to corneal infection strongly agrees with our observations in the mouse ocular model. An 18-year-old male presented to the San Francisco General Hospital Ophthalmology Clinic with an ocular infection which was characterized by a geographic corneal ulcer. A diagnosis of primary HSV geographic epithelial keratitis was made, and he was prescribed topical trifluoridine. He was seen again 5 days later, and upon examination his corneal ulcer had reduced in size by 50%, but he now had periorbital swelling and redness with discrete vesicles, characteristic of HSV, on the lid margin and brow (Fig. 8). The patient was treated with oral acyclovir and had an uneventful recovery. The timing and progression of symptoms in this case of primary ocular infection strongly correlate with the mouse data presented in this study, suggesting a similar mode of viral spread. This suggests parallel patterns of zosteriform spread within mice and humans and that treatment of primary infection with systemic, rather than topical, antivirals may serve to better control the spread of infection.

View larger version (113K): [in this window] [in a new window]   FIG. 8.   Human clinical correlate to periocular disease in mice. Five days following initial presentation with primary corneal disease, the subject shown had periorbital swelling and redness. Discrete vesicles and lesions, characteristic of HSV, are visible on the lid margin and brow.