Multiple Determinants Contribute to the Virulence of HSV Ocular and CNS Infection and Identification of Serine 34 of the US1 Gene as an Ocular Disease Determinant

Abstract

purpose. The virulence of any given strain of herpes simplex virus (HSV) is probably due to the effects of the constellation of genes in that strain and how they act in concert to promote disease. The goal of this work was to develop a system to identify and study the role of multiple genes in HSV disease.

methods. Mixed ocular infection with HSV-1 strains CJ394 and OD4 yield recombinants with increased ocular and central nervous system (CNS) virulence. Clones and subclones of the CJ394 genome were cotransfected with intact OD4 DNA into Vero cells, the transfection pools were inoculated into BALB/c mouse eyes, and disease severity was scored. Fragments transferring increased ocular or CNS disease were sequenced. Site-directed mutagenesis was used to revert one mutation to wild type.

results. Five of the determinants (UL9, -33, -41, and -42 and US1) increased ocular disease when transferred singly. Transfer of the UL36/37 determinant increased both ocular and CNS disease. Transfer of the UL41 and -42 genes increased mortality and a combination of the UL36/37, -41, and -42 determinants increased virulence further. Reversion of the S34A change in the OD4 US1 gene to wild type restored ocular virulence.

conclusions. Multiple HSV genes can operate to increase virulence. The UL9, -33, -36/37, and -42 genes have not previously been identified as virulence determinants. The UL41 and US1 genes are known to affect disease, but the changes identified had not been described. Multiple novel mutations were found in the OD4, UL9, UL36, and US1 genes, and we showed that S34 in the US1 gene is essential in ocular disease.

Herpes simplex virus (HSV) is a significant human pathogen causing mucocutaneous lesions that are self-limiting in immunocompetent individuals. ¹ In developed countries, HSV is the leading cause of sporadic viral encephalitis and infectious blindness and also causes disseminated lethal infections in neonates. ¹ Numerous factors contribute to the outcome of HSV infection, including innate resistance and the host immune response. Several laboratories have documented virulence differences between viral strains, indicating that the genetic composition of the virus is also critical in determining the severity of an infection. ^{2 3 4 5 6}  

Two basic strategies have been used to identify genes contributing to HSV disease. Before sequencing of the HSV genome, spontaneously arising, induced, or recombinant avirulent mutants were isolated, and the virulence determinants were mapped. The availability of the genome sequence ⁷ facilitated mapping efforts and also made it possible to engineer null or other mutants followed by in vivo testing for virulence. With the former method, attenuating mutations in presumably any gene could be obtained, whereas the latter method is restricted to genes not necessary for replication in vitro. Genes not necessary for in vitro replication, referred to as “supplemental,” are likely to be important in virus–host interactions. ^{8 9} Most virulence studies to date have focused on the identification of genes that promote the ability to cause CNS disease. ^{10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27} The analysis of CNS disease is a relatively insensitive marker for comparing the effects of genes or mutations, because of the all-or-none nature of the disease. Quantitation of neurologic symptoms is difficult, and although these symptoms are often mentioned, LD₅₀ values are primarily used for comparisons. 

Most, if not all, the previous studies on the role of genes in HSV disease have focused on the role of single genes. In addition, large effects such as million-fold changes in the LD₅₀ are usually studied. Such dramatic alterations are rarely, if ever, seen in natural HSV infections. In reality, the virulence of a given HSV strain is determined by the constellation of genes carried by that strain, with each of several genes adding its own contribution to the total effect. How each gene functions in the context of the other genes is critical to the overall level of virulence of a given strain. 

Several studies have shown that mixed infections with two avirulent strains of HSV can result in synergistic increases in the severity of encephalitis, primarily through the generation of recombinant viruses. ^{15 17 28 29} Single genes were subsequently implicated in the transfer of neurovirulence between the strains in these studies. ^{18 20} We have reported that mixed ocular infection in vivo with HSV-1 strains OD4 and CJ394 results in more severe ocular disease and increased mortality from encephalitis than does single infection with either parental virus. ³⁰ We have also shown that the mixed-infection–generated recombinant viruses are highly virulent in both ocular and CNS disease. In addition, we have shown that strain OD4 replicated poorly in mice, and although OD4 could transfer to the trigeminal ganglia (TG), no virus was detected in the brain. The results of restriction fragment length polymorphism (RFLP) mapping indicate the presence of multiple crossovers in each recombinant, suggesting that multiple genes may be involved in the increase in virulence. ³⁰ Because neither of the parents was neurovirulent, the mixed infections generated new combinations of genes that increased the disease’s severity. It is therefore likely, that the OD4 and CJ394 viruses could serve as the basis for a system allowing us to examine the contribution of multiple genes in both peripheral and CNS virulence. 

Using a previously validated marker transfer and infection method ³¹ coupled with sequencing, we now report the mapping of at least six different virulence determinants in this system. We show that when transferred alone, five of the determinants (UL9, -33, -41, and -42 and US1) increased peripheral disease severity only, whereas one (UL36/37) increased both peripheral and CNS disease, confirming our earlier suggestion that the two traits are separable. ⁵ The data also show that a combination of the UL41, -42, and -36/37 genes promotes increased CNS disease. We have shown that the UL9, -33, and -36/37 genes contain virulence determinants. These genes have not been identified as having a role in virulence. Two of the genes in which we identified virulence mutations, had previously been associated with virulence (US1 and UL41), but the mutations we found had not been reported. All the nucleotide sequence changes we have identified are novel, and further study of these will provide greater insights into the function of the proteins. Finally, using site-directed mutagenesis, we showed that serine 34 of the US1 gene is required for ocular virulence. These results demonstrate the power of this system for the identification and analysis of HSV virulence determinants and show that multiple virulence genes can cooperate to increase disease severity within a given strain of virus. 

Materials and Methods

Cell Culture and Virus

Vero cells and BALB/c 3T3 fibroblasts (BALB/c cells) were cultured in Dulbecco’s modified Eagle’s medium (DMEM) with 5% serum (1:1 mixture of fetal bovine serum and defined supplemented calf serum; Hyclone, Ogden, UT), 100 U/mL penicillin, 100 μg/mL streptomycin sulfate, and HEPES buffer, as we described previously. ⁵ Murine C1300 neuroblastoma cells (clone Neuro2A; Neuro2A cells) were grown in modified Eagle’s medium (MEM) supplemented with 20% serum and antibiotics. High-titer stocks of HSV-1 strains OD4 and CJ394 were prepared in Vero cells grown in 2% serum, as we described previously. ⁵ The virulence properties of OD4 and CJ394 have been described previously. ^{5 30} Briefly, HSV strain OD4 is avirulent, causing neither keratitis, nor neurologic disease. Strain CJ394 causes moderate keratitis with mean peak disease scores (MPDSs) of 1.8 and 2.1 for vascularization and keratitis, respectively, but does not cause lethal encephalitis in mice older than 4 weeks. 

Isolation of Viral DNA

Viral DNA for cloning and transfections was purified as we have described, using the modifications for large-scale preparations. ³² Briefly, 20 confluent Vero plates (10 cm) were infected at a multiplicity of infection (MOI) of 1.0. Twenty-four hours after the plates reached 100% cytopathic effect (CPE), they were scraped and the infected cells were isolated by centrifugation at 2000g for 10 minutes at 4°C. The medium was removed and saved. The cell pellets were resuspended in 10 mL of the saved medium, frozen, and thawed three times, and the cell debris was removed by centrifugation, as before. The supernatant was removed and combined with the previously saved medium. This material was then layered on a cushion of 36% sucrose in RSB buffer (10 mM Tris-HCl [pH 7.4]; 10 mM NaCl, and 3 mM MgCl₂) and virions were pelleted through the cushion by centrifugation at 13,500 rpm for 80 minutes in a rotor (SW28; Beckman, Fullerton, CA). ³² The virions were resuspended in 3 mL of TE buffer (10 mm Tris [pH 7.4] and 1 mm EDTA) with 0.15 M sodium acetate and 50 μg/mL RNase A. After incubation for 30 minutes at 37°C, the viral DNA was purified by phenol-chloroform extraction and ethanol precipitation. The DNA was stored in purified water at −20°C. 

Cloning of the CJ394 Genome

Purified CJ394 DNA was digested with EcoRI; purified by phenol-chloroform extraction; ligated into EcoRI-digested pBR325, pUC19, or pZErO-1; and electroporated into either Escherichia coli HB101 or TOP10F′. Clones containing inserts of CJ394 DNA were identified by colony screening with probes prepared from an EcoRI library of HSV-1 KOS. ³³ Fragments for probing were isolated and labeled as we described previously. ³⁰ The hybridization conditions have also been described. ³⁰ Subclones for further mapping were prepared by isolating the appropriate EcoRI fragment from the CJ394 library, by using gel electrophoresis and a kit (Geneclean III; Bio101, Vista, CA). Plasmid DNA was purified by the alkaline lysis method followed either by CsCl₂ gradient centrifugation and phenol-chloroform extraction or separation on affinity columns (Endofree Plasmid Kit; Qiagen, Inc., Valencia, CA). DNA concentrations were determined by the optical density at 260 nm. 

Marker Transfer

Purified cloned CJ394 restriction fragments were mixed with intact purified OD4 DNA and cotransfected into Vero cells, using a modification of the CaPO₄ transfection procedure we described previously. ³⁴ Three modifications were made in the procedure to increase transfection efficiency. First, the CJ394 fragments for transfection were modified by filling in the ends with dideoxyguanosine and Klenow enzyme to prevent circularization and exonuclease degradation. ³⁵ Second, CaPO₄ precipitates were allowed to form for only 60 seconds at room temperature to reduce the size of the particles. ³⁶ Finally, the cells were split 1:1 12 to 18 hours before transfection. They were then trypsinized again at the time of transfection, mixed with the DNA precipitate, and replated. Four to six hours after addition of the precipitate, the cells were shocked with 10% dimethyl sulfoxide (DMSO) in HEBES buffer (140 μM NaCl, 5 mM KCl, 0.7 mM Na₂HPO₄, 6 mM glucose, and 20 mM HEPES [pH 6.9]), rinsed twice with medium, and incubated overnight. They were then refed with DMEM (2% serum) and incubated at 37°C. When the transfected cultures reached 100% CPE, the cells were collected by centrifugation at 2000g for 10 minutes, resuspended in 250 μL of medium with 2% serum, and subjected to three freeze–thaw cycles. After centrifugation at 2000g for 10 minutes to remove debris, the samples were stored at −80°C. The titers of the marker transfer stocks ranged from 2.1 × 10⁸ to 4.0 × 10⁸, with an average of 3.3 × 10⁸ plaque-forming units (PFU)/mL. 

Animal Infection and Disease Scoring

Mice (4- to 6-week-old BALB/c females) were purchased from Harlan Sprague-Dawley (Indianapolis, IN). Infections and scoring of the severity of corneal neovascularization and stromal keratitis have been described previously. ³⁷ Briefly, the mice were anesthetized with halothane, the corneas were scarified with a 30-gauge sterile needle, and 5 μL of the inoculum was applied, resulting in infection with an average inoculum of 1.6 × 10⁶ PFU/eye. The mice were examined microscopically at various times after infection. Vascularization was scored as follows: 1+, less than 25% of the cornea involved; 2+, 25% to 50% involved; and 3+, 75% involved. Stromal disease was scored: 1+, cloudiness, some iris detail visible; 2+, iris detail obscured; 3+, cornea totally opaque; and 4+, cornea perforated. CNS disease was scored as a percentage of deaths (encephalitis), and thus we use this term to represent both the ability to spread from the periphery to the central nervous system (CNS) and the ability to replicate in the CNS. MPDSs for the mice showing disease were calculated, as we described previously. ^{5 30} All mice showing disease were used to calculate the MPDS. It was not necessary to isolate the recombinant viruses before infection, because of the low virulence strain of OD4, as previously validated by Thompson et al. ³¹ The use of animals in this study was approved by the Research Animal Resources Center at the University of Wisconsin and conforms to the guidelines of the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Criteria for Scoring Fragments Positive for Transfer of Virulence

We have shown that strain OD4 causes only mild blepharitis (scores of 1 or less) and fails to cause vascularization or stromal keratitis. ⁵ Strain OD4 does not cause CNS disease ⁵ even when injected intracerebrally in nude mice (Brandt C, unpublished data, 1993). Based on these observations, we adopted the following criteria for counting a transfection positive. A given mouse had to display a minimum score of 1 for vascularization or 2 for stromal keratitis or had to die of encephalitis to be counted as positive. Blepharitis was not counted, but its presence indicated that the mice were infected. To score a fragment as positive for transferring virulence, a minimum of 20% of the mice had to score positive for virulence (P < 0.05), and mice in at least two of five independent transfections, five mice per transfection for each fragment, had to score positive for transfer of virulence. It should be noted that the use of these criteria probably resulted in underestimating the number of potential virulence determinants. 

In Vitro Growth Curves

One-step growth curves were performed in Vero, BALB/c, and Neuro2A cells. Cells were infected at an MOI of 2, and the virus was allowed to adsorb at 37°C for 1 hour. The infected cells were rinsed with medium, fresh medium containing 2% serum was added, and the infected cells were incubated at 37°C. At 0, 3, 6, 12, 24, and 48 hours after infection, the infected cells were harvested, and titers of infectious virus were measured on Vero cell monolayers. Two replicates of peak titers were compared by analysis of variance, and significant differences were determined at the 95% confidence level with the Fisher protected least significant difference (PLSD) test. 

Site-Directed Mutagenesis

The reversion of the OD4 US1 gene at position 34 was performed using two consecutive PCR reactions. The first PCR step introduced the desired mutation (A→S) and generated the megaprimer for the next step. The forward and reverse primers were 5′-AAG CCA CTC TGG TCC TCC GA-3′ and 5′-GAG AGG GGT CGC GAA GGG C-3′, respectively. The first reaction included 200 ng pOD4HE containing the OD4 US1 gene, 300 ng of each primer, 100 mM of each dNTP (400 mM total), 1% DMSO, and 2.5 units of polymerase (pfu; Promega, Madison, WI) in a volume of 100 μL. The cycling conditions were as follows: 1 cycle 95°C for 5 minutes; 35 cycles of 95°C for 1 minute, 47°C for 1 minute, and 72°C for 17 seconds and 1 cycle at 72°C for 7 minutes. The product was electrophoresed in a 1.5% agarose gel and the 145-bp megaprimer was isolated with a commercial kit (QIAaquick Gel Extraction; Qiagen, Inc.) In the second step, the megaprimer and reverse primer (5′-AAA CAA GGA AGC TTG CAC ACG-3′) were used. The reaction mix included 200 ng pOD4HE containing the OD4 US1 gene, 300 ng of each primer, 100 mM of each dNTP (400 mM of dDNTPs total), 1% DMSO, and 2.5 units of polymerase (pfu; Promega) in a volume of 100 μL. The cycling conditions were as follows: 1 cycle 95°C for 5 minutes; 30 cycles of 95°C for 1 minute, 47°C for 2 minutes, and 72°C for 1 minute 43 seconds and 1 cycle at 72°C for 7 minutes. The product was gel purified and cloned into pZErO-1 (Invitrogen, Carlsbad, CA). Note that an NruI site restriction site was engineered into the mutagenic primer to identify the nucleotide change. The resultant construct was sequenced to confirm that the desired changes had been introduced and that there were no other alterations in the OD4 sequence (data not shown). 

Results

To screen for virulence determinants, we initially constructed a library of CJ394 EcoRI fragments, cotransfected these fragments with purified OD4 genomic DNA, and screened the resultant progeny virions for ocular and CNS disease in mice. As shown in Figure 1 , the EcoRI D, O, and A fragments met the criteria we defined for transfer of virulence. Several fragments, including J, N, M, and L were completely negative for transfer of virulence. Six of the EcoRI fragments (JK, G, F, I, EK, and H), however, were intermediate for transfer of virulence. Although we did not score these as positive, we cannot rule out the presence of one or more virulence determinants in these fragments. Because the EcoRI O fragment was only 1.3 kb, we sequenced the fragment directly and did not attempt further mapping using marker transfer (described later). We also noted that of the CJ394 EcoRI fragments tested, only EcoRI A transferred increased CNS disease (Table 1) . 

Mapping the Virulence Determinant in the EcoRI D Fragment to the UL9 Gene

The EcoRI D fragment is approximately 16 kb in length and encodes for all or part of 11 genes extending from UL5 to -15. To localize the virulence determinant, we digested the CJ394 EcoRI D fragment with BamHI and subcloned the restriction fragments. BamHI cut CJ394 EcoRI D into three fragments designated DA, DB, and DC in order of their size. As shown in Figure 2 , only the DC fragment transferred increased ocular virulence, and none of the fragments transferred increased CNS disease. Because the DC fragment is wholly contained in the UL9 gene, this result placed the virulence determinant in the gene encoding the origin-binding protein (OBP). 

Mapping Multiple Virulence Determinants in the EcoRI A Fragment

The EcoRI A fragment is 21.5-kb and encodes all or part of 10 genes, extending from UL36 (part) to -44 (part). To further map the virulence determinants in EcoRI A, we first digested purified EcoRI A with BglII, yielding two fragments denoted AA and AB in order of decreasing size. The transfer of the AA fragment resulted in 70% of the mice showing increased ocular disease (Fig. 3) and 40% mortality from encephalitis (Table 1) . Transfer of the AB fragment resulted in severe ocular disease in more than 90% of the mice (Fig. 3) , with 60% of the mice dying of encephalitis (Table 1) . 

To map the CJ394 AA fragment more finely, we digested purified AA with SpeI and tested the resultant AAA and AAB fragments. Neither the AAA, nor AAB fragments transferred increased virulence, suggesting that the SpeI site was located at or very near the virulence determinant. To confirm the location of the determinant in the AA fragment, we isolated the BamHI–ScaI fragment spanning the SpeI site (AAC fragment) and tested it for virulence. As shown in Figure 3 , the AAC fragment transferred increased ocular virulence, with 100% of the mice showing increased ocular disease. The AAC fragment also transferred increased CNS disease, with 36% of the mice dying from encephalitis (Table 1) . 

To map the AB fragment, we first digested AB with SphI. This generated three fragments (ABA, ABB, and ABC). As shown in Figure 3 , both the ABA and ABC fragments scored positive for transfer of virulence, with 20% and 50% of the mice showing increased ocular disease, respectively. Because the ABC fragment was small, we proceeded directly with sequencing, and because this fragment contained only the UL42 gene, the virulence determinant was mapped to the UL42 gene. The ABA fragment at 3.5 kb was considered large enough for further subcloning and was divided by digesting with EcoRV (Fig. 3) . The smaller fragment containing UL40 was designated ABAB, and the fragment containing the 3′ end (UL41) was designated ABAA. Infections in mice resulted in neither fragment meeting the virulence criteria, suggesting that the virulence determinant was located near the EcoRV site. We, therefore, sequenced the entire ABA fragment (described later). In summary, marker transfer studies identified at least three determinants in the EcoRI A fragment localized to UL42 and -41 and the amino-terminal region of UL36 or the carboxyl-terminal region of the UL37 genes (UL36/37). 

Mapping a Virulence Determinant in the Unique Short Region

As noted, several CJ394 EcoRI fragments were able to transfer virulence greater than background, but did not reach our stringent criteria for scoring positive for virulence. Because a virulence determinant may be partially suppressed or masked or may just be weaker than those mapping in the D, O, or A fragments, it was of interest to map one of these fragments further. We chose to focus on EcoRI H, which encodes the US1 through -12 genes and portions of the inverted repeats flanking the US region. Purified EcoRI H was digested with BglII, and the resultant HA and HB fragments were tested for transfer of virulence. Although the entire H fragment resulted in less than 10% of the mice showing disease, transfer of the HA fragment resulted in significant ocular disease in 60% of the mice. Mortality was not increased (0% for H and HA). The HB fragment was negative, indicating the virulence determinant did not map in the US10 through -12 genes. Because the portion of the short repeat sequence in the HB fragment is also contained in HA, the determinant in HA also did not map in the terminal repeat sequence of the short region. 

The HA fragment encodes the US1 gene through part of the US9 gene. To map this region more precisely, we cloned a HindIII–BamHI fragment containing part or all of the US5 to -12 genes (Fig. 4 , fragment HC) and tested it for transfer of virulence. As shown in Figure 4 , the HC fragment was negative for transfer of virulence. We also tested the EcoRI–HindIII fragment (denoted HE), which encodes the 5′ 60% of the US1 gene. Transfer of the HE fragment (1.8 kb) resulted in significant ocular disease in 30% of the mice, but did not transfer increased CNS disease. These results identified the US1 gene as a virulence determinant. Because the entire H fragment failed to transfer virulence, whereas HA and HE did, our results suggest that a function mapping in the HB region of CJ394 (US7–12) has a negative influence on the US1 determinant. 

Mean Peak Disease Scores

To determine the severity of ocular disease transferred by the various CJ394 clones, we calculated the MPDSs for these fragments (Fig. 5) . It should be noted that this analysis was not restricted to those mice with disease scores that met the criteria for scoring the fragments as positive and that all mice showing disease were included (see Materials and Methods). The most severe ocular disease was associated with fragments that also transferred CNS disease (A, AA, AB, AAC, and ABC). This is consistent with our previous studies of OD4/CJ394 recombinants. ³⁰ The HE fragment (US1 gene) induced severe ocular disease (MPDS ≥ 2.0) although it did not transfer increased CNS disease. The D, DC, and O determinants transferred less severe disease than the fragments from the EcoRI A region. The stromal disease was more severe than vascularization with the D, DC, and O fragments. 

Sequence Analysis of Virulence Regions

Once the virulence determinants were localized to short regions, sequencing was used to identify potential virulence mutations. The corresponding regions of both CJ394 and OD4 were sequenced and compared with each other. Strain 17 is the only HSV-1 strain sequenced in its entirety, thus we compared the sequences OD4 and CJ394 to strain 17. ⁷ The amino acids identified in the CJ394 sequences were identical with those of strain 17, but several differences were noted between the CJ394 and OD4. The results are summarized in Figure 6 . Three amino acid changes were identified in the UL9 gene, all localized in the first 50 amino acids of the protein. Sequencing of the EcoRI O fragment revealed a single change, A109D, in the UL33 gene. In the AAC fragment, we found sequence changes in both the UL36 and -37 genes; thus, either or both of these genes could contain a virulence determinant. A single difference, R18H, was found in the UL41 gene of OD4. We also identified an additional change (P386R) in the CJ394 UL41 protein, but because the OD4 sequence (avirulent) was wild type at this position, the altered amino acid position 386 in CJ394 is not related to virulence (Table 2) . We also found a single change in the OD4 UL42 gene, D146N, suggesting that this is a virulence mutation. 

Sequencing of the US1 region revealed multiple differences between CJ394 and OD4. One change at position 281 did not alter the amino acid sequence. Two changes, D46E and D68E, resulted in conservative mutations, exchanging aspartic acid for glutamic acid. The two other mutations resulted in nonconservative changes, exchanging alanine for serine at position 34 and cysteine for tyrosine at position 116. Thus, one or more of these mutations could be responsible for the difference in ocular virulence between OD4 and CJ394. Table 2 lists other nucleotide sequence changes that were identified, but comparisons between the strain 17, CJ394, and OD4 sequences revealed they were not related to virulence. The changes that were shared between CJ394 and OD4 but differed in strain 17, or where the OD4 sequence matched the strain 17 sequence at the position, were considered to not be related to virulence. 

OD4 Virus Replication In Vitro

We have shown that OD4 replicated poorly in mice. ³⁰ Thus, it was important to determine whether OD4 had a general replication defect. If so, the determinants we identified would be classified as “supplemental essential replication genes.” We performed single-step growth curves at 37°C in Vero cells, BALB/c 3T3 fibroblasts (BALB/c cells), and mouse 3100 Neuro2A cells. As shown in Figure 7A , the growth curves in Vero cells were essentially identical. In the BALB/c cells (Fig. 7B) , the peak titers of CJ394 were 0.5 log higher than OD4, suggesting a slight replication disadvantage for OD4, but the differences were not significant (P > 0.05). For the Neuro2A cells (Fig. 7C) , the peak titers of CJ394 were 1 log higher than OD4 at 24 hours, but the titers were essentially identical at 12 and 48 hours (P > 0.05). The data thus suggest that OD4 may replicate slightly less well in mouse cells than in monkey cells, but clearly show that OD4 does not have a generalized replication defect. 

Serine 34 of the US1 Gene

The identification of a single-sequence change in a gene suggests that the alteration is involved in the virulence differences. We found single changes in the UL33, -41, and -42 genes. For the other determinants, we found multiple changes, and it is possible that only some of these are related to virulence. To extend our results and confirm that our identification of virulence mutations, we chose to examine further the mutations identified in the US1 gene. Site-directed mutagenesis was used to revert the S34A mutation in the OD4 US1 gene. We then used the marker transfer and infection assay to test the ability of the gene to transfer increase virulence. As shown in Figure 8A , correction of the OD4 mutation to wild type at position 34 resulted in significantly increased ocular disease in 41% of the mice. The MPDSs (Fig. 8B) for all mice showing vascularization and stromal disease were 1.1 and 1.5, respectively, compared with none for the OD4 transfection. Thus, the serine residue at position 34 of the US1 gene is an ocular disease determinant. 

Discussion

Most studies of virulence genes in HSV-1 infection have primarily focused on the effects of a single gene and have used relatively large changes, such as a million-fold difference in the LD₅₀ values as the criteria for classifying a gene as having a role in virulence. However, in nature, the virulence of any given strain is most likely the result of the effects of the constellation of genes carried by that strain. How these genes and their products interact with each other and interact with host proteins determines the outcome of infection. Mutations that dramatically affect virulence are likely to be detrimental for the virus, and such viruses would not be maintained in nature. As our data show, when considered alone, mutations in five of the six genes (UL9, -33, -41, and -42 and US1) increased ocular disease only. Transfection of the determinant overlapping the UL36/37 genes increased both peripheral disease and mortality (AAC fragment, Table 1 ). Transfer of the UL41 and -42 determinants together increased the severity of both ocular disease and CNS disease (AB fragment, Table 1 ), indicating that these two determinants act cooperatively. Finally, transfer of the UL36/37, -41, and -42 determinants together increased ocular disease and CNS disease even further (EcoRI A fragment, Fig. 5 , Table 1 ). Thus, our results confirm that multiple virulence determinants interact to increase the severity of HSV-1 disease in a particular strain of virus. Our data also show that the use of ocular infection allows for the detection of determinants involved in both peripheral (keratitis) and CNS disease and that scoring for keratitis allows for more subtle distinctions in disease severity than encephalitis, skin infection, or genital infection models. Table 3 summarizes the current knowledge concerning the virulence genes we have identified. 

Several points concerning our findings are relevant: (1) Previous studies on the genetics of HSV virulence have revealed that the UL41 (vhs) and US1 genes are virulence determinants. ^{88 89} That we also identified UL41 and US1 as virulence determinants validates our system for the identification of HSV virulence genes. (2) The advantage of our system over other models is demonstrated by our identification of additional virulence determinants in the UL9, -33, -36/37, and -42 genes, which represent the first association of these genes with a role in virulence. (3) Our results showing that the HE fragment (US1 gene) transferred virulence, whereas the entire EcoRI H fragment was negative, suggest that viral genes can antagonize virulence properties and that hidden determinants also exist. This observation has not been noted previously. (4) A number of functional domains or motifs have been identified in several of the proteins. Some of our mutations mapped in large domains with identified functions, but none of the mutations mapped to known functional motifs in these domains. Thus, studies of the mutations we have identified will provide important information about the function of the proteins. (5) The ability of the CJ394 UL41, -42, and -36/37 genes to transfer neurovirulence to OD4 suggests that one or more OD4 genes complement the CJ394 genes, because CJ394 does not cause encephalitis. Alternatively, CJ394 could contain a function that suppresses neurovirulence. (6) A number of fragments appeared to transfer virulence, but did not reach the level of significance. Given that other fragments were completely negative (e.g., EcoRI N), it is possible that additional virulence determinants remain to be mapped. For example, transfer of the ER and JK fragments resulted in 18% and 9%, respectively, of the mice showing significant disease. The γ₁34.5 gene, a known neurovirulence determinant, ^{23 31} maps to EK and JK, and the virulence associated with these fragments could be due to the γ₁34.5 gene. Further studies are needed to ascertain whether γ₁34.5 is involved in the virulence difference between OD4 and CJ394. (7) It is notable that two of the mutations we identified in the US1 gene would remove residues that could be phosphorylated (S34 and Y116), and we have shown that S34 is required for peripheral virulence. To date, the phosphorylation sites in the US1 protein remain unknown. It would be intriguing if the S34 mutation resulted in the loss of a phosphorylation site in the US1 gene that correlated with the loss of specific isoforms and virulence changes. (8) The UL41 and US1 genes can be deleted from the virus while allowing for replication in cell culture, which is an advantage for studying the role of such genes in virulence. Such a strategy, however, is not feasible for UL9, -33, -36/37, or the -42 which are essential for replication. ⁸ Thus, our system has the advantage of allowing for the identification of novel virulence mutations in essential genes. (9) All the mutations we have identified are novel. That these mutations were first identified as being associated with virulence changes increases the importance of further studies to ascertain how the mutations affect the function of each of the proteins. 

We have shown that transfection procedures can induce mutations, ^{88 89} raising the possibility that a mutation that increases the virulence of the parental OD4 virus could account for the data. This is highly unlikely for several reasons. To date, we have performed transfection and infection assays with OD4 DNA more than 20 times and have never detected a transfection pool with increased virulence. We have also performed ocular infections with OD4 in several hundred mice and have yet to isolate a variant with increased virulence. Finally, we have also inoculated OD4 intracerebrally into nude mice (20 animals) and found no evidence of a reversion to virulence (Brandt C, unpublished data, 1993). When considered together, these observations indicate that reversion to virulence of the OD4 strain is not a concern. Another potential criticism of the marker transfer–infection strategy is that each transfection could generate differing amounts of recombinant virus within the pool. However, as we have documented, ⁹⁰ ocular disease is a threshold phenomenon and is not dependent on the inoculum over a wide range. ⁵ Once the inoculum reaches a given amount, stromal disease ensues. This is true over a wide range of inocula, and at more than 1 × 10⁶ PFU/eye we were clearly above threshold. ^{5 90} Given that disease requires a threshold inoculum, variations in the percentage of recombinant virus in the transfection pools would affect the results only if the recombinant viruses were below the threshold, in which case the transfection would score as negative, and this system thus would be more likely to miss virulence determinants. 

Our earlier findings ³⁰ have clearly indicated that OD4 replicates poorly in vivo. Thus, it is possible that the genes we have identified merely assist in increasing viral replication. Although we argue that any gene that increases replication qualifies as a virulence determinant, our data clearly show that OD4 does not have a generalized replication defect, because in Vero cells, mouse fibroblasts, and mouse neuroblastoma cells, titers of OD4 and CJ394 were not significantly different (Fig. 7) . When considered together, the data clearly show that OD4 does not have a generalized replication defect. Thus, the mutations we have identified could be considered “host range” determinants. This is also consistent with data showing that deletion of the entire US1 gene reduces virulence in mice and negatively affects replication specifically in mouse cells, qualifying the US1 gene as a “host range” determinant. ⁷⁴  

In conclusion, we report a number of novel findings. First, we describe a system that allows for the study of the effect of mutations in multiple HSV-1 genes on virulence, and we show that multiple genes are involved in determining the outcome of infection. To our knowledge, this represents the first description of such a system for a large DNA virus. We identified novel mutations in seven different HSV genes. We showed for the first time that mutations in three and possibly four genes, UL9, -33, -36/37, and -42, are involved in peripheral and/or CNS disease and coincidentally identified novel mutations in two genes (UL41 and US1), previously identified as affecting virulence. Finally, we showed that a hidden determinant, S34, in the US1 gene is involved in peripheral virulence. Further studies using this system and analysis of the effects of the mutations we have identified will increase our understanding of the function of the proteins and their role in HSV-induced disease. 

 Supported by National Eye Institute Grant EY07336 (CRB); The Retina Research Foundation, Alice McPherson Chair (CRB); Robert McCormick Special Scholar Award from Research to Prevent Blindness (CRB); an unrestricted grant from Research to Prevent Blindness to the Department of Ophthalmology and Visual Sciences; and Biotechnology Training Grant T326M08349.

 Submitted for publication October 30, 2002; revised December 12, 2002; accepted January 6, 2003.

 Disclosure: C.R. Brandt, None; A.W. Kolb, None; D.D. Shah, None; A.M. Pumfery, None; R.L. Kintner, None; E. Jaehnig, None; J.J. Van Gompel, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Transfer of increased virulence to strain OD4 by EcoRI fragments of strain CJ394. Purified cloned EcoRI fragments of strain CJ394 were cotransfected into Vero cells with purified OD4 DNA and the resultant viral stock (5 μL) used to inoculate mouse corneas. The severity of disease was scored at various times, and the percentage of mice with significantly increased ocular disease was determined. Five separate transfections were used to inoculate five mice each. P ≤ 0.05 is at 20% or above. Top: EcoRI map of CJ394; bottom: percentage of mice showing increased disease.

Table 1.

 

Summary of Mortality Due to Encephalitis after Transfer of Specific CJ394 Fragments to OD4

Fragment	% Mortality	Determinants
A	100	UL36/37, UL41, UL42
AA	40	UL36/37
AB	60	UL41, UL42
AAC	36	UL36/37

Figure 2.

 

Mapping of a virulence determinant in the EcoRI D fragment of CJ394. The EcoRI D fragment was digested with BamHI and the resultant three fragments (DA, DB, DC) cloned. Virulence was tested by marker transfer to OD4 (see Fig. 1 ), followed by corneal infection and scoring of disease. Top: map of EcoRI D and the encoded genes; bottom: percentage of mice showing increased disease.

Figure 3.

 

Mapping of multiple virulence determinants in the CJ394 EcoRI A fragment. The EcoRI A fragment of CJ394 was digested with various enzymes, and the resultant subclones generated are shown (top), along with the percentage of mice showing increased disease (bottom). The cloned DNA was purified, and virulence was tested after marker transfer (Fig. 1) .

Figure 4.

 

Mapping of virulence in the EcoRI H fragment of CJ394. The EcoRI H fragment of CJ394 was digested with various enzymes, and the resultant clones were generated (top). Purified fragments were tested for virulence using marker transfer and corneal inoculation. Top: map of the region, the clones tested, and the genes mapping to the EcoRI fragment; bottom: percentage of mice showing increased disease.

Figure 5.

 

MPDSs of all mice showing evidence of infection. The mean severe vascularization and stromal keratitis scores in each mouse showing disease were averaged and plotted. This includes disease scores that did not reach the level required to deem transfection positive. (▪) Vascularization; ( ) stromal keratitis.

Figure 6.

 

Schematic diagram showing the potential location of the virulence changes identified in this study.

Table 2.

 

Sequence Changes Not Associated with Virulence Differences

Gene	Position	Strain 17*	CJ394	OD4
UL36	204	I	T	T
	280	E	D	D
UL37	1106	I	S	S
UL40	17	D	G	G
UL41	386	P	R	P
UL42	13	P	H	H
USI	281	T	A	A

Figure 7.

 

One-step growth curves in Vero cells (A), BALB/c 3T3 fibroblasts (B), and Neuro2A (C) cells at 37°C. The cells were infected at an MOI of 2.0 and, at the indicated times, titers were determined on Vero cell monolayers.

Figure 8.

 

Serine 34 is a virulence determinant. Site-directed mutagenesis was used to revert the S34A mutation in the OD4 US1 gene to wild type. The revertant was tested for virulence by using the marker rescue–infection procedure, and ocular disease was scored. (A) Percentage of mice showing ocular disease above the cutoffs. (B) MPDSs for all mice showing disease. None of the mice showed development of encephalitis.

Table 3.

 

Characterization of Virulence Determinants

Gene	Kinetic Class	Functions	Identification as a Virulence Determinant	Previous Known Mutations	Comments	References
UL9	β	Binds origin of replication; helicase activity; dimerization domain; attracting polymerase complex ICP8 binding; binding cellular proteins	Current study	None	Mutations lie in the helicase domain, but not in conserved motifs. Mutations identified in current study are novel.	38 39 40 41 42 43 44 45
UL33	γ	Packaging of DNA; localization to nuclear replication compartments; binding UL28 and UL15 proteins	Current study	ts1233 (I17N)	The A104R mutation in OD4 is novel.	46 47 48
UL36	γ	Tegument protein; ATP binding motif; leucine zippers; binds “a” sequence; binds to a 140-kDa viral protein	Possible, current study	None	Serine phosphorylated. Currently identified mutations are novel.	49 50
UL37	γ	Tegument protein; binds ICP8; ATP binding motif	Possible, current study	None	Phosphorylated. Currently identified mutations do not map to known functional regions.	51–53
UL41	γ	mRNA degradation; endonuclease; host shutoff	Previously identified null mutants, HSV-1/HSV-2 chimeras, vhs-1 mutant	T214I (vhs1 1)	Currently identified our mutation is novel and lies near a putative Mg++ binding site in the nuclease domain.	54 55 56 57 58 59 60 61 62 63 64 65
UL42	β	Processivity factor for pol; binds HSV DNA pol; oligomerization interfaces	Current study	Linker insertions, deletion mutants, fusion proteins	The D146N change in OD4 is in an exposed loop.	66 67 68 69 70 71 72 73
US1	α	IE regulatory protein; regulates subset of late genes; alters phosphorylation of host RNA pol II; binds to host cell proteins; colocalizes with UL4 and UL3 in nucleus; modulates antigen presentation	Known host range gene	Null mutants	Extensively modified. Multiple isoforms in cells, phosophorylated by UL13 kinase and host casein kinase II.	74 75 76 77 78 79 80 81 82 83 84 85 86 87

The authors thank Teresa Compton and Donna Peters for critical reading of the manuscript.