Point Mutations in Herpes Simplex Virus Type 1 oriL, but Not in oriS, Reduce Pathogenesis during Acute Infection of Mice and Impair Reactivation from Latency

Abstract

In vitro studies of herpes simplex virus type 1 (HSV-1) viruses containing mutations in core sequences of the viral origins of DNA replication, oriL and oriS, that eliminate the ability of these origins to initiate viral-DNA synthesis have demonstrated little or no effect on viral replication in cultured cells, leading to the conclusion that the two types of origins are functionally redundant. It remains unclear, therefore, why origins that appear to be redundant are maintained evolutionarily in HSV-1 and other neurotropic alphaherpesviruses. To test the hypothesis that oriL and oriS have distinct functions in the HSV-1 life cycle in vivo, we determined the in vivo phenotypes of two mutant viruses, DoriL-I_LR and DoriS-I, containing point mutations in oriL and oriS site I, respectively, that eliminate origin DNA initiation function. Following corneal inoculation of mice, tear film titers of DoriS-I were reduced relative to wild-type virus. In all other tests, however, DoriS-I behaved like wild-type virus. In contrast, titers of DoriL-I_LR in tear film, trigeminal ganglia (TG), and hindbrain were reduced and mice infected with DoriL-I_LR exhibited greatly reduced mortality relative to wild-type virus. In the TG explant and TG cell culture models of reactivation, DoriL-I_LR reactivated with delayed kinetics and, in the latter model, with reduced efficiency relative to wild-type virus. Rescuant viruses DoriL-I_LR-R and DoriS-I-R behaved like wild-type virus in all tests. These findings demonstrate that functional differences exist between oriL and oriS and reveal a prominent role for oriL in HSV-1 pathogenesis.

Herpesviruses establish two mutually exclusive kinds of interactions with the cells they infect: productive infection and latency (22). Productive infection of epithelial cells and fibroblasts is characterized by active expression of all viral genes in a highly ordered temporal cascade in which immediate-early (IE) and early (E) gene expression leads to viral DNA synthesis, late (L) gene expression, and the production of new infectious virus. In contrast, latent infection of sensory neurons is characterized by profoundly restricted viral-gene expression, the failure to synthesize viral DNA, and the absence of infectious virus.

Replication of the viral genome is a central event in the life cycles of herpesviruses. The initiation of viral DNA synthesis marks the commitment of the infected cell to the production of new infectious virus and, in most instances, cell death. The absence of viral DNA synthesis in neurons, however, is a hallmark of latency and neuronal viability. Viral DNA synthesis is initiated at defined sequences within viral genomes termed “origins of replication.” Viral origins are commonly composed of core palindromic sequences flanked by auxiliary sequences (3). Core sequences are essential for origin function and typically contain an origin recognition element and a DNA-unwinding element, the latter of which is often AT rich. Auxiliary sequences affect the efficiency of initiation and are composed of transcription factor binding sites, which are commonly associated with promoter/regulatory elements of genes transcribed divergently from the core origin. Specific viral and cellular proteins form complexes on origin sequences, leading to unwinding of the DNA duplex and initiation of DNA synthesis.

HSV-1 contains three origins of DNA replication of two types: one copy of oriL located at the center of the unique long (U_L) region of the genome and two copies of oriS located in the “c” repeats flanking the unique short (U_S) region of the genome (Fig. 1A). Both oriL and oriS (i) contain palindromic sequences as core elements, (ii) share greater than 90% homology in sequences common to both core elements, and (iii) reside within the promoter/enhancer regions of divergently transcribed genes important or essential for viral replication (Fig. 1B) (30). Notably, mutant viruses lacking either oriL or both copies of oriS are replication competent in vitro (11, 19), demonstrating that oriL and oriS can substitute functionally for one another during productive infection and raising the question, Why does the herpes simplex virus (HSV) genome possess two types of replication origins that appear to be functionally redundant?

FIG. 1.

Location and sequence alignment of HSV-1 oriL and oriS. (A) Diagram of the HSV-1 genome showing the U_L region flanked by inverted repeats ab and b′a′ and the U_S region flanked by inverted repeats a′c′ and ca. The locations of oriL and both copies of oriS are indicated as open ovals. Two essential E genes, UL29 and UL30, are transcribed divergently from oriL and encode the single-stranded DNA binding protein (ICP8) and DNA Pol, respectively; these transcripts are shown as black arrows beneath the genome. Three IE genes encoding viral regulatory proteins ICP4, -22, and -47 are transcribed divergently from oriS; these transcripts are shown as black arrows beneath the genome. (B) Nucleotide sequences of oriL and oriS. For simplicity and to illustrate sequence homology, the sequence of each origin is shown as a single-stranded palindrome. Black dots indicate single nucleotide differences between oriL and oriS. The designations I, II, and III indicate the locations of OBP binding sites. The two black boxes near the apexes of oriL and oriS indicate functional and degenerate GRE half-sites, respectively (9). The arrows and boxes at the bases of the two palindromes represent the two E genes (encoding ICP8 and DNA Pol) that flank oriL and the three IE genes (encoding ICP4 and ICP22/47) that flank oriS. The distances in nucleotides between the last nucleotide shown and the start of transcription of each origin-flanking gene are shown in parentheses.

Despite the apparent functional redundancy of oriL and oriS, evidence exists to support the hypothesis that they perform distinct functions in the dual life cycle of HSV type 1 (HSV-1). Despite the extensive sequence homology in core elements shared by both types of origins (Fig. 1B), the complete nucleotide sequences of the oriL and oriS core origins differ considerably. The oriL core origin consists of a perfect 144-bp palindrome and contains a 20-bp AT-rich sequence at the apex (Fig. 1B), which is thought to unwind during the initiation of DNA replication. Immediately adjacent to the AT-rich apex are consensus glucocorticoid response element (GRE) half-sites that bind the glucocorticoid receptor (9). In addition, oriL contains four high-affinity binding sites for the origin binding protein (OBP; the product of the UL9 gene): two copies of site I and two copies of site III (10) (Fig. 1B). In contrast to the oriL core origin, the oriS core origin consists of a 90-bp sequence, including an imperfect 45-bp palindrome, an 18-bp AT-rich apical sequence, degenerate GRE half-sites immediately adjacent to the apex, and three OBP binding sites: one copy each of sites I, II, and III (Fig. 1B). Although only 6 of the 71 nucleotides (nt) that are shared by oriL and oriS core sequences differ, 5 occur near the AT-rich apex, and 4 of the 5 are located within GRE half-sites (Fig. 1B).

In addition to existing structural differences, several functional differences between oriL and oriS have been reported. OBP has been shown to have a greater affinity for oriL sites I and III than for oriS sites I and III by electrophoretic mobility shift analysis (EMSA) (8). With respect to its DNA initiation function, an oriL-containing plasmid replicated with modestly reduced efficiency relative to an oriS-containing plasmid of similar size in the same vector backbone in in vitro DNA replication assays using both Vero and PC12 cells (rat pheochromocytoma cells) (29), suggesting intrinsic differences in the properties of the two origins (8, 9). Notably, in nerve growth factor (NGF)-differentiated PC12 cells, initiation of DNA synthesis from oriS was reduced nearly fivefold relative to undifferentiated PC12 or Vero cells (9). No effect of NGF differentiation of PC12 cells on the efficiency of oriL function was observed (9). In the same study, the glucocorticoid receptor was shown by EMSA to bind specifically to the consensus GRE near the apex of oriL, but not to a similar sequence present in the same position in oriS (9). In addition, in NGF-differentiated PC12 cells (but not in undifferentiated PC12 cells), the synthetic glucocorticoid dexamethasone (DEX) enhanced oriL-dependent DNA replication fivefold but repressed oriS-dependent DNA replication fourfold. The DEX-induced repression of oriS-dependent DNA replication was in addition to the repressive effect induced by NGF differentiation of PC12 cells. Mutation of the oriL GRE eliminated the DEX-induced enhancement of oriL-dependent DNA replication, indicating that the increase in oriL function was mediated through the GRE. Together, these findings demonstrate cell type-specific functional differences in the efficiencies of oriL- and oriS-dependent DNA replication and suggest that oriL might function more efficiently in neurons in vivo in response to high levels of glucocorticoids.

We have begun to test the hypothesis that oriL and oriS have distinct functional roles in the HSV-1 life cycle by generating four viruses: DoriL-I_LR (containing paired substitution mutations in both copies of oriL site I); DoriS-I (containing a single substitution mutation in single site I in both copies of oriS); and two rescuant viruses, DoriL-I_LR-R and DoriS-I-R, in which the point mutations are replaced by wild-type sequences (1). Both the paired point mutations in oriL in DoriL-I_LR and the single point mutations in the two copies of oriS in DoriS-I eliminate OBP binding to site I and, consequently, the DNA initiation function of both cloned origins (1, 2). In in vitro tests, both mutant viruses synthesized viral DNA and replicated as efficiently as wild-type virus in Vero cells and primary rat embryonic cortical neurons (1), supporting the concept that oriL and oriS can substitute for each other in vitro. The only difference observed between the two mutants was a minor alteration in the kinetics of ICP8 transcript accumulation in DoriL-I_LR-infected primary rat embryonic cortical neurons, but not in Vero cells, implying a neuron-specific role in regulating transcription of this flanking gene from oriL (1).

In this study, we examined the effects of point mutations that eliminate the oriL and oriS DNA initiation function on acute virus replication and pathogenesis in mice. Although either a single intact copy of oriL or both intact copies of oriS proved to be sufficient for acute-phase replication in mice, the paired point mutations in DoriL-I_LR (but not the single point mutations in both copies of oriS) resulted in significant alterations in viral pathogenesis following acute infection including (i) delayed and reduced morbidity, (ii) a marked reduction in mortality, and (iii) reduced efficiency of reactivation from latency. Both rescuant viruses behaved like wild-type virus in all tests ascribing the phenotypes of DoriL-I_LR to the paired point mutations in oriL. Collectively, these findings indicate that functional copies of oriS or a functional copy of oriL is not required for HSV-1 replication in vivo. They also show, however, that the two functional copies of oriS in DoriL-I_LR are insufficient to achieve wild-type parameters of pathogenesis in mice and demonstrate that the absolute redundancy in oriL and oriS function observed in vitro is not observed during the later stages of acute infection in vivo or during reactivation from latency.

MATERIALS AND METHODS

Cells and viruses.

Vero cells (ATCC CCL-81) (31) were propagated in complete Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS), penicillin (100 units/ml), streptomycin (100 μg/ml), and 2 mM l-glutamine. Vero cells stably transformed with the ICP27 gene (3-3 cells) (23) were maintained in complete DMEM containing 300 μg/ml G418 (Sigma-Aldrich Co., St. Louis, MO).

The wild-type virus used in this study was KOS at passage 11 from original isolation (25). The ICP27 mutant virus 5dl1.2, the origin mutant viruses DoriL-I_LR and DoriS-I, and rescuant viruses DoriL-I_LR-R and DoriS-I-R have been described previously (1, 17).

Inoculation of mice.

Five- to 6-week-old (∼30-g) male ICR mice (Harlan Sprague-Dawley, Indianapolis, IN) were handled in accordance with the Guide for the Care and Use of Laboratory Animals (17). Mice were distributed randomly into groups of 12 (latency experiments) or 70 (acute replication experiments). Mice were anesthetized by intraperitoneal administration of xylazine (9 mg/kg of body weight) and ketamine (100 mg/kg). Mouse corneas were scarified with a 26-gauge needle, tear film was blotted, and 2 × 10⁵ PFU of virus in 3 μl of complete DMEM were placed on each cornea as described previously (6). The titers of virus in inocula were confirmed by standard plaque assays on Vero cell monolayers and were found to vary less than twofold from one another.

Measurement of viral titers in tear film, trigeminal ganglia (TG), and hindbrain.

To determine viral titers in tear film, tear film samples were collected from both eyes using a single cotton-tipped applicator. The cotton tip was transferred to 500 μl complete DMEM and frozen at −80°C. Frozen samples were later thawed and thoroughly mixed, and infectious virus was quantified by standard plaque assays on Vero cell monolayers. Values for total PFU per cotton tip were determined and divided by 2 to calculate the approximate viral titers per eye.

To determine viral titers in TG and hindbrains (brain stem and cerebellum), acutely infected mice were euthanized by CO₂ asphyxiation and TG and hindbrains were removed, placed individually in microcentrifuge tubes containing 100 μl of 1-mm-diameter glass beads and 500 μl DMEM containing 1% FCS, homogenized with a Mini-Beadbeater (Biospec Products, Bartlesville, OK), and frozen at −80°C. Samples were later thawed, sonicated, and clarified by centrifugation (2,000 × g for 10 min). Infectious virus was quantified by standard plaque assays on Vero cell monolayers.

Measurement of viral genome loads in TG by competitive PCR.

The competitive PCR assay used in these tests has been described previously (6). Briefly, total DNA was isolated from individual latently infected TG harvested 30 to 35 days postinoculation using the Qiamp DNA minikit (QIAGEN, Hilden, Germany) as per the manufacturer's instructions. Oligonucleotide primers specific for the HSV-1 ribonucleotide reductase (RR) gene (UL39) and competitor templates were used, amplifying 243- and 322-bp fragments, respectively, from each template. Included in each assay were control samples containing known numbers of viral genomes. The PCR products were vacuum slot blotted in duplicate and hybridized to oligonucleotide probes specific for either HSV RR or competitor templates, and the relative yields of the two PCR products were determined by PhosphorImager analysis. A standard curve was plotted as the ratio of HSV RR to competitor as a function of the number of viral genomes. The viral genome load per TG was calculated as the number of genomes per 0.1 μg of TG DNA × 210, because each TG yielded approximately 21 μg of DNA.

Infectious-center assays: measurement of the number of latently infected TG cells.

On days 30 to 35 postinoculation, latently infected mice were euthanized and TG were removed. Individual TG were teased apart longitudinally using a scalpel. Each TG was placed in a tube containing 0.5 ml of dissociation buffer (0.1% collagenase [Sigma] and 0.125% trypsin [Invitrogen, Carlsbad, CA] in Hanks' balanced salt solution [HBSS]) and incubated in a shaker (200 rpm at 37°C) for 40 min. After 25 min, TG were gently triturated to mechanically dissociate individual cells from pieces of tissue. Dissociated TG cells were washed once with complete DMEM to eliminate the activities of collagenase and trypsin. Cells from each dissociated TG were resuspended in 3 ml of complete DMEM, transferred to one well of a collagen-coated six-well plate, and centrifuged at 200 × g for 5 min. At 12 to 14 h postplating (p.p.), culture medium was removed and replaced with 500 μl of complete DMEM containing 5 × 10⁶ PFU of 5dl1.2 (yielding an estimated multiplicity of infection of ≥10 PFU/cell). The actual titers of viral inocula were determined by standard plaque assays on 3-3 cell monolayers and did not vary more than twofold from calculated input titers. After 1 h of adsorption at 37°C, the inoculum was removed, cells were washed once with HBSS, and 3 ml complete DMEM containing 5 × 10⁵ Vero cells was added to each well. Three hours later, medium was removed and cells were overlaid with 1% methylcellulose. After 4 days, the methylcellulose was removed and monolayers were fixed and stained with 20% methanol-0.1% crystal violet. Plaques initiated by infectious centers (single latently infected cells) were counted.

Measurement of ex vivo reactivation efficiency.

To measure reactivation by explant cocultivation, mice were euthanized 30 to 35 days postinoculation by CO₂ asphyxiation. Latently infected TG were removed immediately, cut into eight equal-size pieces, and cocultivated with 3 × 10⁴ Vero cells in 1.5 ml of complete DMEM per well in 24-well plates. Each day postexplant (p.e.), 150 μl of medium was removed and assayed for the presence of infectious virus. Fresh complete DMEM (500 μl) was added to TG explant cultures every third day. The presence of infectious virus and, hence, reactivation, were detected by transferring the 150-μl sample of explant medium to fresh Vero cell monolayers and scoring daily for cytopathic effects.

Measurement of reactivation in primary, latently infected TG cell cultures was performed as described previously (4, 5). Briefly, on day 30 postinoculation, latently infected mice were euthanized and TG were removed (n = 6 mice or 12 TG per virus). Individual TG were teased apart longitudinally using a scalpel. All 12 teased TG latently infected with each virus were pooled and incubated in 1.2 ml of 0.1% collagenase (Sigma) in HBSS for 40 min at 37°C, and 7 ml TG growth medium (minimal essential medium supplemented with 10% FCS, 0.15% HCO₃⁻, penicillin [100 units/ml], streptomycin [100 μg/ml], 2 mM l-glutamine, and 10 ng/ml NGF) was added. Dissociated TG cells were washed three times, resuspended in TG medium, and plated in 24-well collagen-coated plates seeded the previous day with 3 × 10⁴ uninfected TG feeder cells. Twelve TG yield 24 ml of TG cell suspension, which was plated in a 24-well plate at 1 ml/well, or one-half TG/well. TG cell cultures were tested daily for the presence of infectious virus as described for explant cultures.

Southern blot analysis.

To confirm the genotypes of reactivated viruses, Vero cells were infected with medium from reactivating TG cultures. When cytopathic effects were generalized, cells were scraped into medium and cell suspensions were transferred to tubes and centrifuged (16,000 × g for 5 min). Cell pellets were resuspended in 200 μl of lysis buffer (10 mM Tris-HCl [pH 8.0], 1 mM EDTA [pH 8.0], 0.2% sodium dodecyl sulfate, and 400 μg/ml proteinase K). Cell lysis was carried out overnight at 55°C. The volume of the lysates was increased to 600 μl with TE (10 mM Tris-HCl [pH 8.0], 1 mM EDTA [pH 8.0]), and the lysate was extracted twice with an equal volume of phenol-chloroform-isoamyl alcohol (25:24:1). The DNA in the lysate was ethanol precipitated and resuspended in TE. DNA was quantitated by spectrophotometry (A₂₆₀), and 3 μg of total cellular DNA was digested with restriction enzymes. The resulting fragments were separated on 0.8% agarose gels and transferred to nylon membranes (Micron Separations Inc., Westborough, MA). DNA was UV cross-linked (200 mJ/cm²; UV Stratalinker1800; Stratagene, La Jolla, CA), prehybridized for 1 h at 68°C in ExpressHyb solution (Clontech, San Francisco, CA), and hybridized for 1 h at 68°C using ³²P-labeled randomly primed probes specific for either the BstBI fragment of HSV-1 DNA that contains oriL (nt 61191 to 65382 of the HSV-1 strain 17 genome; GenBank accession no. NC001806) or the BstEII fragment of HSV-1 DNA that contains 5′ oriS nucleotides (nt 131186 to 134063). The blots were washed as per the manufacturer's instructions, and the hybridized membrane was exposed on a PhosphorImager cassette (Molecular Dynamics, Sunnyvale, CA) overnight.

Statistics.

The statistical tests used in this study are those recommended by Richardson and Overbaugh (21). With the exception of Fig. 6, data are presented as the means ± standard errors of the means. The significance of differences in numbers of infectious centers and mean time to reactivation was determined by two-tailed t test. The significance of differences in mortality rates and reactivation efficiencies was determined by Fisher's exact test. Two-way analysis of variance (ANOVA) was used to determine the significance of differences among experimental groups at multiple time points. Linear regression was used to evaluate the reliability of the standard curves for competitive PCR.

FIG. 6.

Reactivation efficiencies of origin mutants in explant and TG cell cultures. (A) Efficiencies of reactivation from explant cultures of TG latently infected with wild type, DoriL-I_LR, DoriL-I_LR-R, DoriS-I, or DoriS-I-R. TG from mice infected 30 to 34 days p.i. were cut into six to eight pieces and cocultured on Vero cell monolayers. Culture medium was assayed daily for the presence of infectious virus as indicated by cytopathic effects in indicator plates of Vero cells. (B) Efficiencies of reactivation from TG cell cultures. TG cell cultures were prepared by pooling 12 to 16 TG per virus group, gently teasing them apart, treating with collagenase to generate single-cell suspensions, and distributing the resulting cell suspensions into 24-well plates. Culture medium was assayed daily for the presence of infectious virus as indicated by cytopathic effects in indicator plates of Vero cells. Each point in panels A and B represents the cumulative percentage of reactivating TG and TG cell cultures from two to four experiments. The number (n) of TG (A) and the numbers of TG cell culture wells (B) are indicated.

RESULTS

Effects of site I origin mutations on acute viral replication and pathogenesis in mice.

To test the effects of site I point mutations in oriL and oriS on acute viral replication and pathogenesis in vivo, mice were infected via the ocular route following corneal scarification with 2 × 10⁵ PFU/eye of wild-type virus, origin mutant viruses DoriL-I_LR and DoriS-I, or rescuant viruses DoriL-I_LR-R and DoriS-I-R. Although the acute-replication experiments (see Fig. 2 through 4) were performed by comparing all five viruses simultaneously, for ease of interpretation, results are presented as two graphs (wild type), DoriL-I_LR, and DoriL-I_LR-R in panel A in each figure and wild type, DoriS-I, and DoriS-I-R in panel B (the wild type is the same in both panels).

FIG. 2.

Tear film titers of wild-type, origin mutant, and rescuant viruses during acute infection of mice. Twenty-six mice per group of 30-g (5- to 6-week-old) male ICR mice were infected in both eyes with 2 × 10⁵ PFU/eye of the indicated virus following corneal scarification. On the indicated days, eyes were swabbed as described in Materials and Methods. Viral titers were determined by standard plaque assay on Vero cell monolayers. The results shown are the means ± standard errors of the means (error bars). Although this experiment was performed comparing all five viruses simultaneously, the results are separated into two graphs for ease of interpretation. (A) Comparison of DoriL-I_LR and DoriL-I_LR-R to wild type. (B) Comparison of DoriS-I and DoriS-I-R to wild type. (Note that wild type is the same in panels A and B.)

FIG. 4.

Titers of wild-type, origin mutants, and rescuant viruses from hindbrain during acute infection of mice. Seven mice per group of 30-g (5- to 6-week-old) male ICR mice were infected in both eyes with 2 × 10⁵ PFU/eye of the indicated virus following corneal scarification. Hindbrains were removed and processed on the indicated days as described in Materials and Methods. Viral titers were determined by standard plaque assay on Vero cell monolayers. The results shown are the means ± standard errors of the means (error bars). Although this experiment was performed comparing all five viruses simultaneously, the results are separated into two graphs for ease of interpretation. (A) Comparison of DoriL-I_LR and DoriL-I_LR-R to wild type. (B) Comparison of DoriS-I and DoriS-I-R to wild type. (Note that wild type is the same in panels A and B.)

(i) Viral titers in tear film.

To examine the efficiency of acute viral replication at the site of inoculation in mice, viral titers in tear film were determined daily for 9 days following corneal inoculation (Fig. 2). Both DoriL-I_LR (Fig. 2A) and DoriS-I (Fig. 2B) replicated as efficiently as the wild-type and rescuant viruses for the first 3 days postinfection (p.i.). From day 4 through day 8 p.i., DoriL-I_LR replicated slightly but significantly less efficiently, with levels two- to threefold lower than wild type (P = 0.001, two-way ANOVA), whereas DoriL-I_LR-R replicated with wild-type efficiency (P = 0.67). Replication of DoriS-I between days 4 and 8 p.i. was more significantly impaired (P = 4.3 × 10⁻¹¹), with titers two- to eightfold lower than wild-type titers. DoriS-I-R replicated with wild-type efficiency (P = 0.64). These observations demonstrate that the point mutations in DoriS-I, and to a lesser extent those in DoriL-I_LR, slightly impaired the ability of both viruses to achieve wild-type levels of virus in tear film during acute infection.

(ii) Viral titers in TG.

DoriL-I_LR, DoriS-I, and the rescuant viruses were compared to the wild type with regard to levels of infectious virus in TG during the acute phase of replication (Fig. 3). Levels of DoriL-I_LR-R (Fig. 3A), DoriS-I, and DoriS-I-R (Fig. 3B) were similar to wild type (P = 0.73, 0.80, and 0.35, respectively; two-way ANOVA) in TG in that infectious virus was detected on day 1 p.i., peaked with similar titers between days 3 and 4, and decreased progressively through day 9 p.i. In contrast, levels of DoriL-I_LR were significantly reduced (P = 1.7 × 10⁻⁵) in TG, most notably on days 2 through 5 p.i., relative to wild type (Fig. 3A). As with the other four viruses, DoriL-I_LR was first detected on day 1 p.i.; however, on 7 of the 9 days tested, the levels of DoriL-I_LR were reduced by as much as 30-fold (days 3 and 5 p.i.) relative to wild type.

FIG. 3.

Titers of wild-type, origin mutant, and rescuant viruses from trigeminal ganglia during acute infection of mice. Fourteen mice per group of 30-g (5- to 6-week-old) male ICR mice were infected in both eyes with 2 × 10⁵ PFU/eye of the indicated virus following corneal scarification. TG were removed and processed on the indicated days as described in Materials and Methods. Viral titers were determined by standard plaque assay on Vero cell monolayers. The results shown are the means ± standard errors of the means (error bars). Although this experiment was performed comparing all five viruses simultaneously, the results are separated into two graphs for ease of interpretation. (A) Comparison of DoriL-I_LR and DoriL-I_LR-R to wild type. (B) Comparison of DoriS-I and DoriS-I-R to wild type. (Note that wild type is the same in panels A and B.)

(iii) Viral titers in hindbrain.

We also examined the levels of the wild-type, mutant, and rescuant viruses in the hindbrain (i.e., brain stem and cerebellum) during acute infection as an indicator of the relative ability of these viruses to (i) spread from the peripheral nervous system (PNS) to the central nervous system (CNS) and (ii) replicate in the CNS (Fig. 4). The wild type, DoriS-I (Fig. 4B), and the rescuants (Fig. 4A and B) were detected in hindbrain on day 1 p.i.; however, the detection of DoriL-I_LR (Fig. 4A) in the hindbrain was delayed by 1 day relative to the other four viruses tested. Although levels of DoriS-I, DoriS-I-R, and DoriL-I_LR-R were similar to levels of the wild type (P = 0.75, 0.24, and 0.88, respectively; two-way ANOVA), the levels of DoriL-I_LR were significantly reduced (P = 1.8 × 10⁻⁴), as shown by reduced titers throughout the course of the 9-day assay. This reduction was especially apparent on days 3 through 5 p.i., when DoriL-I_LR titers were 28-, 10-, and 10-fold lower than the wild type, respectively.

Collectively, these findings demonstrate that a functional copy of oriL or two functional copies of oriS are not required for acute replication of HSV-1 in mice; however, both oriL and oriS are required to achieve wild-type levels of virus in tear film, whereas there is a greater requirement for functional oriL for efficient spread to and/or replication in TG and hindbrain.

(iv) Gross pathology.

Relative to mock-infected mice, physical examination of mice acutely infected with the wild type, DoriS-I, or the rescuants revealed little difference in gross pathology throughout the 30-day test period. Mice infected with each of the four viruses exhibited moderate-to-severe symptoms of blepharoconjuctivitis, ulcerative lesions, periocular hair loss, and encephalitis. Many were lethargic, disoriented, or hyperreactive beginning on day 7 p.i., with symptoms resolving by day 14 p.i. In contrast, an extension in the time of onset and a marked reduction in the severity of gross pathological changes were noted in DoriL-I_LR-infected mice relative to wild-type virus-infected mice. Only a subset of mice acutely infected with DoriL-I_LR exhibited mild symptoms of blepharoconjuctivitis, periocular hair loss, and encephalitis. Moreover, while cumulative mortality rates from four independent experiments were similar among mice infected with the wild type (26%), DoriL-I_LR-R (34%), DoriS-I (19%), and DoriS-I-R (21%), the mortality rate was significantly reduced in mice infected with DoriL-I_LR (4.2%; P = 0.006, Fisher's exact test; Table 1). Based on these findings we conclude that the two functional copies of oriS in DoriL-I_LR are insufficient to produce wild-type levels of morbidity and mortality, but one functional copy of oriL in DoriS-I is sufficient.

TABLE 1.

Mortality rates for male ICR mice infected with wild-type, mutant, and rescuant viruses

Virus	Mortality ratea in expt:				Σ Mortalityd	% Mortality	Pc
	1	2	3	4
Wild type	2/16	6/13	4/16	4/16	16/61	26
DoriL-ILR	1/16	0/12	1/10	0/10	2/48	4.2	0.006
DoriL-ILR-R	NDb	6/12	2/13	5/13	13/38	34	0.34
DoriS-I	1/16	1/12	3/10	4/10	9/48	19	0.31
DoriS-I-R	ND	2/12	2/11	3/11	7/34	21	0.41

^aNumber of deaths/total number of mice.

^bND, not determined.

^cThe significance of differences in levels of mortality during acute replication was determined by Fisher's exact test.

^dΣ Mortality, cumulative mortality rate from experiments 1 to 4.

Effects of origin mutations on the establishment and reactivation of latency.

To determine the effects of the oriL and oriS site I mutations on establishment and reactivation of latency, mice latently infected with wild-type, mutant, and rescuant viruses were euthanized between days 30 and 35 p.i. and TG were removed to determine (i) the efficiency of establishment of latency as assayed by viral-genome loads in TG (Fig. 5), (ii) the relative number of latently infected ganglionic cells that support reactivation in infectious-center assays (Table 2), and (iii) the kinetics and efficiency of reactivation (Fig. 6).

FIG. 5.

Viral-genome loads in latently infected TG. The number of viral genomes in individual TG of mice latently infected with the indicated virus was determined 30 to 34 days p.i. by competitive PCR. The numbers (n) of TG analyzed are shown below the graph. The results shown are the means ± standard errors of the means (error bars). (Inset) Representative standard curve comparing the relationship between numbers of viral genomes and ratio of HSV RR to competitor. Linear regression was used to evaluate the reliability of the standard curve.

TABLE 2.

Number of latently infected cells per TG

Virus	Expt 1 (n = 10a)		Expt 2 (n = 5)
	Mean ± SEMb	Pc	Mean ± SEM	P
Wild type	48 ± 8.8		46 ± 6.5
DoriL-ILR	46 ± 6.8	0.82	39 ± 18	0.75
DoriL-ILR-R	50 ± 4.7	0.84	48 ± 12	0.88
DoriS-I	53 ± 6.9	0.69	39 ± 6.3	0.47
DoriS-I-R	69 ± 14	0.32	39 ± 12	0.66

^an, number of TG per group.

^bMean number of infectious centers per TG.

^cThe significance of differences in the number of infectious centers was determined by two-sided t test.

(i) Viral-genome loads.

The efficiency of establishment of latency was determined by measuring viral-genome loads in individual TG by competitive PCR (6). As shown in Fig. 5, both DoriL-I_LR and DoriS-I established latency as efficiently as wild-type and rescuant viruses (1.0 × 10⁴ to 1.8 × 10⁴ genomes/TG), indicating that none of the site I point mutations in either origin significantly affected the efficiency of establishment of latency.

(ii) Number of reactivatable latently infected ganglion cells.

The relative number of latently infected ganglion cells containing reactivatable viral genomes was determined by infectious-center assays (16). Individual TG from latently infected mice were removed on days 30 to 35 days p.i., ganglia were dissociated and treated with collagenase and trypsin, and the resulting cell suspension was plated. Twelve to 14 h p.p., dissociated ganglion cells were superinfected at an estimated multiplicity of infection of ≥10 PFU per cell of 5dl1.2 (a replication-incompetent, ICP27⁻ virus) and cocultivated with Vero cells, and the resulting monolayers were overlaid with methylcellulose. Because 5dl1.2 is replication incompetent, any plaques that form are initiated by complementation or recombination between 5dl1.2 and replication-competent latent viral genomes. In two independent experiments, we observed no significant difference (P = 0.32 to 0.88; two-sided t test) in the ability of wild-type, mutant, or rescuant viruses to produce infectious centers (Table 2). Specifically, ganglion cultures from mice latently infected with wild-type virus produced averages of 46 and 48 infectious centers in two independent experiments. Cultures derived from mice latently infected with mutant virus DoriL-I_LR or DoriS-I and rescuant virus DoriL-I_LR-R or DoriS-I-R produced similar numbers of infectious centers, with averages ranging from 39 to 69 infectious centers. Thus, there was no evidence to suggest that the point mutations in DoriL-I_LR and DoriS-I affect either the relative numbers of viral genomes in mouse TG or the biological activity of latent viral genomes as measured by infectious-center assays.

(iii) Reactivation by explant cocultivation.

The kinetics and efficiency of reactivation of the five viruses were first assayed by explant cocultivation of latently infected TG (Fig. 6A). The mean times required to detect reactivated virus in culture medium were similar among the wild type (4.2 ± 0.2 days p.e.), DoriL-I_LR-R (3.8 ± 0.1 days p.e.), DoriS-I (3.9 ± 0.1 days p.e.), and DoriS-I-R (4.1 ± 0.2 days p.e.). Reactivation of DoriL-I_LR was delayed, however, with respect to the wild type, with a mean reactivation time of 4.8 ± 0.2 days p.e. (P = 0.04; two-sided t test). Relative to the wild type, DoriL-I_LR-R, DoriS-I, and DoriS-I-R, reactivation of DoriL-I_LR was also less efficient at early times p.e. in that the numbers of reactivated cultures were lower than wild type, especially between days 3 and 5 p.e., resulting in delayed reactivation kinetics; DoriL-I_LR ultimately achieved wild-type levels of reactivation, however, in that 97% of the cultures had reactivated by 7 days p.e.

(iv) Reactivation in TG cell cultures.

Reactivation frequencies were also assessed in TG cell cultures. An interesting characteristic of this assay is that differences in the kinetics and efficiency of reactivation observed by the TG explant assay are often enhanced in the TG cell culture assay (5). This is likely due to the facts that (i) fewer latently infected neurons are present per well (an estimated 1/40 the number of neurons are present per well in TG cell cultures relative to explanted TG; this number is based on the fact that each well is seeded with a volume of cell suspension equivalent to one-half a TG and some cell death occurs during preparation of the TG cell suspensions) and (ii) the TG microenvironment is disrupted in the latter assay. These differences notwithstanding, the times to the first appearance of reactivated virus were similar in TG cell cultures and explant cultures for each virus tested (compare Fig. 6A with B). In TG cell cultures, the wild type, DoriL-I_LR-R, DoriS-I, and DoriS-I-R reactivated with similar efficiencies in that 91 to 100% of the cultures produced new infectious virus by 10 days p.p. with mean times to reactivation of 4.2 ± 0.2, 4.6 ± 0.2, 4.3 ± 0.2, and 4.0 ± 0.2 days p.p., respectively (Fig. 6B). In contrast, DoriL-I_LR reactivated with a significant delay and reduced efficiency. Thus, with a mean time to reactivation of 5.1 ± 0.3 days p.p. (P = 0.005; two-sided t test), only 78% of DoriL-I_LR cultures had reactivated (P < 0.001; Fisher's exact test) by 10 days p.p.

Collectively, using two ex vivo models of reactivation, these findings indicate that neither type of origin is essential for viral reactivation from latency. To achieve wild-type kinetics and efficiency of reactivation, however, the single functional copy of oriL in DoriS-I was sufficient whereas two functional copies of oriS in DoriL-I_LR were not.

(v) Genotypes of reactivated viruses.

Because it was theoretically possible that the phenotypes of the viruses just described might have been influenced by the presence of revertants generated during acute infection in mice (particularly in the case of DoriS-I infections where little or no phenotype was observed), viral DNA from individual reactivated TG of mice infected with wild-type virus, DoriL-I_LR, or DoriS-I 30 to 34 days previously was examined by Southern blot analysis to determine whether the site I mutations (which produced novel BstBI sites) were present (Fig. 7). Indeed, viral DNA from nine of nine DoriL-I_LR-infected TG and nine of nine DoriS-I-infected TG remained sensitive to BstBI cleavage, demonstrating that the point mutations introduced into oriL or both copies of oriS were stable during replication in mice.

FIG. 7.

Genotypes of reactivated viruses. Total DNA was from Vero cell cultures infected with supernatant fluids from individual wells containing reactivated virus from TG of mice infected 30 to 34 days previously with the wild type, DoriL-I_LR, or DoriS-I. Fragments from restriction digestion of total cellular DNA were separated by agarose gel electrophoresis and transferred to a nylon membrane. Southern blotting was performed using a probe specific for either oriL (A) or for oriS (B). (A) Diagram of oriL and the expected fragment sizes after digestion with BstBI. The horizontal line represents the central portion of the U_L segment of the genome. The vertical black line indicates the location of oriL. Black arrows above the genome represent genes transcribed divergently from oriL. Numbers above the ends of the line indicate map locations in nucleotides. Locations of BstBI sites relative to oriL are indicated. Numbers below enzymes indicate nucleotide positions in the viral genome. Numbers beneath fragments indicate their sizes in kb. Letters in parentheses correspond to the fragments shown in the Southern blot to the right of the diagram. (Note that the 42-bp BstBI fragment “d,” containing oriL apical sequences, was not resolved in these tests). (B) Diagram of both copies of oriS and the expected fragment sizes after digestion with the indicated enzymes. Boxes indicate portions of the internal (IR_S) and terminal repeats (TR_S) flanking the U_S region of the genome. Numbers above boxes indicate map locations in nucleotides. Vertical black lines indicate the locations of oriS. Locations of BstEII and BstBI sites relative to oriS are indicated. Numbers below enzymes indicate nucleotide positions in the viral genome. Numbers beneath fragments indicate their sizes in kb. Letters in parentheses correspond to the fragments shown in the Southern blot to the right of the diagram.

DISCUSSION

Although most DNA-containing viruses utilize a single origin to initiate DNA replication, the genomes of several herpesviruses including HSV-1, HSV-2, pseudorabies virus, Kaposi's sarcoma-associated herpesvirus, and Epstein-Barr virus (EBV) contain two types of origins. Of these, the best characterized are the origins of EBV, which includes two copies of oriLyt and one copy of oriP. Each type of origin functions at a distinct stage of the EBV life cycle, oriLyt during lytic infection (7) and oriP during latency (20, 32). To date, however, the specific roles of the two types of HSV-1 origins, oriL and oriS, remain elusive. Thus, relative to wild-type virus, no in vitro replication phenotype was observed with deletion mutant viruses lacking either oriL (19) or both copies of oriS (11). Although the properties of deletion mutants were the first to demonstrate functional redundancy during HSV-1 replication in vitro (11, 19), these in vitro results have now been confirmed by the point mutation-containing viruses used in this study and extended to include tests of primary rat embryonic cortical neurons (1). The question that prompted this study is, If the two types of HSV-1 origins are not functionally distinct in vitro, are they functionally distinct in vivo?

To investigate the requirement for oriL and oriS during viral infection in vivo, two viruses that eliminate the initiation function of oriL and oriS by virtue of point mutations in site I sequences, DoriL-I_LR and DoriS-I, respectively (1), were compared to wild-type and rescuant viruses in in vivo tests in mice. DoriL-I_LR (i) exhibited a slight, but statistically significant, reduction in tear film titers when compared to wild-type and rescuant virus DoriL-I_LR-R, (ii) replicated less efficiently in TG and hindbrain during the acute phase of replication, and (iii) reactivated from latency with delayed kinetics and reduced efficiency even though it established latency with wild-type efficiency. In contrast, relative to wild-type and rescuant virus DoriS-I-R, DoriS-I (i) exhibited a modest, but statistically significant, reduction in tear film titers, (ii) replicated with wild-type efficiency in TG and hindbrain, and (iii) was indistinguishable from wild-type virus with regard to the efficiency of establishment and reactivation of latency. Most notably, the pathogenesis and mortality rate for DoriL-I_LR were markedly reduced relative to the other four viruses. Collectively, the results of these tests demonstrate that the two functional copies of oriS in DoriL-I_LR are insufficient to produce wild-type levels of acute-phase replication, symptomatology, mortality, or reactivation from latency in mice, whereas a single functional copy of oriL in DoriS-I is sufficient to support wild-type levels of each of these parameters of HSV-1 infection. These results also demonstrate that in vivo tests are more informative than in vitro tests in elucidating the functions of HSV-1 oriL and oriS.

Effects of oriL and oriS site I mutations on HSV-1 pathogenesis.

In addition to its impaired replication phenotype in vivo, DoriL-I_LR produced reduced pathological changes and a markedly reduced mortality rate during the acute phase of infection. The notion that oriL is required for wild-type-like pathogenesis and mortality is supported by recent results of studies using male ICR mice with the deletion mutant ts+7, which lacks oriL core sequences (19). In seven independent tests (n = 116 mice) conducted in parallel with the studies described in this paper, no gross pathology or mortality was observed in ts+7-infected ICR mice whereas wild-type-infected mice exhibited the gross pathological changes observed in this study (unpublished observations). It should be noted that our recent observations differ from our previous study in which ts+7 achieved a level of mortality similar to that of wild-type virus (19). Polvino-Bodnar et al. (19) also saw little effect of the deletion on tear film titers and titers in TG on day 3 and a small reduction in reactivation efficiency. In that early study, a 10-fold-greater viral inoculum was used and only one experiment utilizing a smaller sample size (10 mice) of CD-1 mice was performed with ts+7. A comparison of these two in vivo studies serves to emphasize the significance of mouse type, sample size, and dose in designing and interpreting experiments in vivo.

Evidence from the present study suggests that the attenuated phenotype of DoriL-I_LR is likely due to inefficient viral replication and/or spread within the PNS and CNS. Measurements of DoriL-I_LR titers in TG and hindbrain demonstrated a significant reduction in the levels of infectious virus relative to wild-type and rescuant viruses. Because the virus detected in both TG and hindbrain is likely generated by replication within these tissues as well as by spread from upstream sites (the eye in the case of TG and TG in the case of hindbrain), the reduced levels of infectious virus in TG and hindbrain may be due to inefficient viral replication within these tissues and/or to inefficient spread of the virus to these sites. Indeed, the likelihood that spread is at least partly responsible for the reduced titers of DoriL-I_LR in hindbrain is supported by the following observations: (i) DoriL-I_LR was not detected at this site until 2 to 3 days p.i. in contrast to the wild type, which was detected in the hindbrain on day 1 p.i., and (ii) the 50% lethal dose of ts+7 following intracranial inoculation is equivalent to wild type (unpublished observations), suggesting that oriL is not required for replication and spread within the CNS. Additionally, inefficient spread of DoriL-I_LR from the TG to peripheral sites via the maxillary nerve would also explain the lack of significant clinical pathology characteristic of this virus. An immunohistochemical study to address these points is under way.

Effects of site I mutations on the establishment and reactivation of latency.

Paradoxically, DoriL-I_LR established latency as efficiently as wild-type virus although viral titers were significantly reduced during acute-phase replication in TG. In a previous study, we showed that infection of mice by the ocular route with 10³ to 10⁶ PFU/eye resulted in equivalent genome loads in the TG on day 30 p.i. (6). A similar observation has been reported with other replication-impaired mutant viruses (27). Collectively, these observations support the notion that only a threshold level of viral replication in the eye and TG is required to establish latency efficiently as reported previously (6) and that amplification of viral genomes occurs in neurons without the production of new infectious virus. Hence, the phenotypes of viral genome amplification and the production of new infectious virus in TG are likely separable. Consistent with this concept is the report that a DNA sequence located in the UL39 gene facilitates autonomous replication when cells are transiently transfected with plasmids containing this sequence (24). Thus, it is possible that this sequence, termed oriH, may mediate amplification of viral genomes utilizing cellular DNA replication proteins, independent of their viral counterparts.

Examination of reactivation frequencies of DoriL-I_LR and DoriS-I in two different ex vivo models revealed significantly impaired kinetics and reduced efficiency of reactivation of DoriL-I_LR, but not DoriS-I, relative to wild-type and rescuant viruses. That viral-genome loads and numbers of latently infected neurons per TG were similar in DoriL-I_LR-, DoriS-I-, and wild-type-infected TG indicates that the impaired-reactivation phenotype of DoriL-I_LR was not a result of the inefficient establishment or maintenance of latency. Rather, these findings demonstrate that intact copies of oriL site I, but not oriS site I, are necessary and sufficient for wild-type levels of reactivation.

Absence of DoriS-I phenotype.

The absence of a significant in vivo phenotype for DoriS-I was unexpected and implies that the oriS initiation function is dispensable for HSV-1 replication in mice. The fact that all neurotropic alphaherpesviruses contain two copies of oriS (but not all contain oriL, e.g., varicella-zoster virus [26]) suggests that oriS is a critical element in alphaherpesvirus replication and pathogenesis. One possible explanation for the absence of a significant in vivo phenotype for DoriS-I is that the point mutation in each copy of oriS, which inhibits oriS-dependent DNA replication in Vero cells (2), does not inhibit initiation from this origin in neurons in vivo. Unfortunately, it is currently not possible to test the initiation function of either HSV-1 origin during viral replication in vivo. Evidence exists, however, to support the concept that the point mutation in the oriS site does inhibit the initiation function of oriS during virus replication in cells of neural lineage. Thus, DoriS-I exhibits reduced efficiency of viral replication and viral DNA synthesis, but not reduced transcription of oriS flanking genes, in NGF-differentiated PC12 cells, suggesting that the mutations in this virus do affect the oriS initiation function in cells of neural lineage (our unpublished observations). Furthermore, in the same tests, restoration of the wild-type sequence in both copies of oriS in DoriS-I (i.e., DoriS-I-R) restored viral replication and viral DNA synthesis to wild-type levels in NGF-differentiated PC12 cells (unpublished observations). Both the difficulties in generating a virus containing deletions in both copies of oriS (11) and the observation that elimination of the initiation function of oriS did not produce a major phenotype in mice suggest that oriS may perform a function in addition to the initiation of viral DNA synthesis. Given the importance of testing origin mutant viruses in vivo as demonstrated in this study and the fact that mice are not humans, the functional roles of HSV-1 oriS may prove difficult to resolve outside the natural human host of HSV-1.

Mechanistic implications.

The impaired in vivo replication and reactivation phenotypes of DoriL-I_LR in TG and hindbrain together with the enhanced initiation activity of oriL relative to oriS in NGF-differentiated PC12 cells (9) are consistent with a model in which oriL is the preferred site of initiation of viral DNA synthesis in neurons. Thus, the site I mutations that eliminate initiation from oriL (i.e., DoriL-I_LR) in neurons in vivo may result in reduced efficiency of initiation of viral DNA synthesis from the two intact copies of oriS, affect the structure of replicative intermediates, or affect the efficiency of recombination or processing of the viral DNA, resulting in reduced numbers of progeny viral genomes. Moreover, reduced efficiency of DNA replication may also result in reduced levels of transcription of IE and E genes as there is a documented relationship between HSV-1 DNA synthesis and viral gene IE and E expression in neurons of the PNS (13-15, 18) and CNS (our unpublished observations). In this regard, we have recently shown that the kinetics of accumulation of UL29 (ICP8) transcripts, but not DNA Pol transcripts, are altered in DoriL-I_LR-infected primary rat embryonic cortical neurons relative to wild-type-infected neurons (1). Although this alteration did not affect the efficiency of DoriL-I_LR replication in cortical neurons in single-cycle growth assays, modest changes in the levels of ICP8 expression may result in altered viral replication and spread in vivo given the multiple interactions between ICP8 and cellular and viral proteins (28).

Although this study was designed to test the DNA initiation function of oriL and oriS in vivo, the possibility exists that sequences not associated with either type of origin may support initiation of viral DNA synthesis (e.g., oriH [24]). This possibility will be tested as KOS genome-containing bacterial artificial chromosome technology advances. In the absence of such tests, the results of this study demonstrate that the initiation function of oriS contributes minimally to the morbidity and mortality of HSV-1 infection in mice and is not required for efficient viral replication or reactivation from latency in PNS neurons. In contrast, the initiation function of oriL contributes significantly to the morbidity and mortality of HSV-1 infection in mice and to efficient viral replication and reactivation from latency in PNS neurons. Thus, distinct functional properties can now be attributed to oriL and oriS.