In Vivo Replication of an ICP34.5 Second-Site Suppressor Mutant following Corneal Infection Correlates with In Vitro Regulation of eIF2α Phosphorylation

Abstract

In animal models of herpes simplex virus type 1 (HSV-1) infection, ICP34.5-null viruses are avirulent and also fail to grow in a variety of cultured cells due to their inability to prevent RNA-dependent protein kinase (PKR)-mediated inhibition of protein synthesis. We show here that the inability of ICP34.5 mutants to grow in vitro is due specifically to the accumulation of phosphorylated eIF2α. Mutations suppressing the in vitro phenotype of ICP34.5-null mutants have been described which map to the unique short region of the HSV-1 genome, resulting in dysregulated expression of the US11 gene. Despite the inability of the suppressor mutation to suppress the avirulent phenotype of the ICP34.5-null parental virus following intracranial inoculation, the suppressor mutation enhanced virus growth in the cornea, trigeminal ganglia, and periocular skin following corneal infection compared to that with the ICP34.5-null virus. The phosphorylation state of eIF2α following in vitro infection with the suppressor virus was examined to determine if in vivo differences could be attributed to differential regulation of eIF2α phosphorylation. The suppressor virus prevented accumulation of phosphorylated eIF2α, while the wild-type virus substantially reduced eIF2α phosphorylation levels. These data suggest that US11 functions as a PKR antagonist in vivo, although its activity may be modulated by tissue-specific differences in translation regulation.

The ability of a virus to replicate within specific cell types often depends on viral interactions with, and alterations of, well-regulated cellular pathways. In the case of herpes simplex virus type 1 (HSV-1), a crucial factor is its ability to regulate the phosphorylation state of the translation initiation factor eIF2α. The activation of the interferon-inducible double-stranded RNA-dependent protein kinase (PKR) during the normal course of infection leads to the phosphorylation of eIF2α and the inhibition of protein synthesis (13, 34). One of the protein products of HSV-1, ICP34.5, functions as an antagonist of the PKR response by redirecting the host protein phosphatase 1α (PP1α) to dephosphorylate eIF2α (8, 17). ICP34.5 has homology to the mammalian DNA damage response protein GADD34, which also binds PP1α and regulates eIF2α phosphorylation (30). Viruses carrying mutations in the ICP34.5 gene have profound growth deficits in vivo (3, 9) as well as in certain cell types in vitro (3, 10, 19). In nonpermissive cell types, including some of neuronal origin, the cells undergo an inhibition of protein synthesis following infection with ICP34.5-null viruses, due to PKR-mediated phosphorylation of eIF2α. These in vitro and in vivo growth deficits are specifically due to the mutant virus' inability to counter the antiviral effects of PKR, since in both cells and mice deficient for PKR viruses lacking ICP34.5 replicate comparably to wild-type HSV-1 (7, 24).

Recently, a second PKR antagonist encoded by HSV-1 has been described, the product of the US11 gene. This function was initially described in the context of a spontaneous extragenic suppressor of an ICP34.5 mutation (25). In these spontaneous viral mutants, the US11 gene, normally expressed with late kinetics, is expressed under the control of the α47 promoter due to deletions spanning the α47 open reading frame (ORF) and the US11 promoter region. As a result, the US11 protein accumulates with immediate-early kinetics (IE-US11), which is sufficient to restore protein synthesis in neuroblastoma cells despite the lack of functional ICP34.5 (5, 16, 27). Subsequent biochemical studies have shown that US11 can inhibit PKR-mediated phosphorylation of eIF2α in vitro and that US11 physically interacts with PKR, possibly acting as a pseudosubstrate (4). US11 has also been shown to be capable of blocking PKR activation by PACT (31). Interestingly, the carboxy-terminal RNA-binding domain of US11 is sufficient for the suppression of the ICP34.5-null phenotype (32), and recent data suggest that US11 may interact with PKR via an RNA bridge (4). In light of biochemical data validating the function of US11 as a PKR antagonist, we have been interested in assessing the ability of these suppressor mutants to rescue the severe attenuation seen in vivo with ICP34.5 mutants.

We have recently shown that, despite the ability of the suppressor viruses to grow and maintain protein synthesis in neuroblastoma cells, these viruses remain avirulent (26). A virus expressing IE-US11, but with ICP34.5 restored, displayed wild-type virulence, showing a requirement for functional ICP34.5 in the pathogenesis of HSV-1 encephalitis. In order to address more fully the impact of these suppressor mutations on the pathogenesis of HSV-1, we have tested them in the mouse ocular model of HSV-1 infection. Here we report that an ICP34.5-null recombinant virus expressing IE-US11 replicated to higher levels than its ICP34.5-null parental virus at early times postinfection. The suppressor virus did not, however, replicate equivalently to a wild-type strain of HSV-1. This intermediate restoration of growth mediated by the suppressor mutation was mirrored in the levels of phosphorylated eIF2α seen in a mouse embryo fibroblast (MEF) cell line following infection. These data suggest that maintaining a low level of phosphorylated eIF2α is crucial in determining the outcome of infection.

MATERIALS AND METHODS

Cells and viruses.

African green monkey kidney (Vero) cells were propagated as previously described (33). Wild-type, Ser51Ala (39), and PKR^−/− (45) MEFs were maintained at 5% CO₂ in a humidified incubator and propagated in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 1× essential amino acids (GibcoBRL, Gaithersburg, Md.), 1× nonessential amino acids (Washington University Tissue Culture Support Center), and antibiotics (250 U of penicillin/ml, 250 μg of streptomycin/ml, and 250 ng of amphotericin B [Fungizone]/ml). Ser51Ala and wild-type control MEFs were on a C57BL/6J background, while PKR^−/− and wild-type control MEFs were on a 129 Ev/Sv background.

The Patton strain of HSV-1 is the wild-type background of SPBg5e, XN1, and 34.5RΔSUP (Fig. 1). SPBg5e is an HSV-1 recombinant lacking both copies of the ICP34.5 gene, which have been replaced by β-glucuronidase cassettes. The XN1 suppressor virus is isogenic to SPBg5e but also carries a 595-bp deletion in the unique short region between the endogenous NruI site and a synthetic XbaI site introduced at nucleotide (nt) 145415 (27) upstream of the US11 ORF. The recombinant virus was isolated following selection for suppressors of the SPBg5e phenotype in U373 cells as described elsewhere (27). Following two rounds of plaque purification, stocks were prepared, and the physical structure of the mutant US11 locus was confirmed by Southern analysis (data not shown). 34.5RΔSUP contains the suppressor deletion in the unique short region but carries wild-type copies of the ICP34.5 gene at both loci. As specificity controls, two additional viruses were used: 17/tBTK(−) (here termed 17Δtk), which carries a deletion in the thymidine kinase gene, and its marker-rescued virus, 17/tBRTK(+) (here termed 17tkR). The background strain of these viruses is 17Syn+. All recombinant viruses used in this study have been described in detail elsewhere (26, 27, 43). Southern blot analysis of viral DNA was performed as previously described (33). A 2.2-kb XhoI fragment corresponding to the unique short portion of the BamHI N fragment was labeled with ³²P by random priming and used as a probe to examine the integrity of the junction between U_S and IR_S.

FIG. 1.

Structures of the recombinant viruses used in this study. All of the pictured viruses were generated in the background of the wild-type HSV-1 Patton strain. All nucleotide numbers refer to the HSV-1 strain 17 sequence (GenBank accession no. X14112). (A) The structure of the γ₁34.5 loci in SPBg5e and XN1. In SPBg5e, the γ₁34.5 gene (small arrows) was replaced with a gene encoding β-glucuronidase (large arrows) at both loci (25). XN1 was derived from SPBg5e and contains identical mutations at the γ₁34.5 loci. In 34.5RΔSUP, the γ₁34.5 loci have been restored to wild type. (B) The suppressor mutation in XN1 and 34.5RΔSUP. XN1 contains a 595-bp deletion between a synthetic XbaI site at nt 145415 and an endogenous NruI site at nt 146010 (27). The deletion in 34.5RΔSUP (583 bp) is indicated by the small arrowheads and is based on a spontaneous suppressor deletion between nt 145415 and 145999 occurring during serial passage of SPBg5e in nonpermissive cells (26). The TR_S begins at nt 145584. US10, US11, and US12 ORFs are shown in boxes. Stars represent promoter elements, and arrows indicate transcripts generated from this mutant locus. These suppressor mutations remove most of the US12 ORF and all of the US11 promoter, allowing transcripts initiating from the immediate-early US12 promoter to direct the synthesis of US11. (C) The structure of the IR_s and U_s regions in viruses carrying suppressor mutations. Viral DNA was isolated from infected Vero cells, cut with BamHI, and fractionated by agarose gel electrophoresis. DNA was transferred to a nitrocellulose membrane and hybridized to a ³²P-labeled 2.2-kb XhoI fragment corresponding to the unique short portion of the BamHI N fragment.

Animal procedures.

Strain 129 Ev/Sv mice were housed and bred in the Washington University School of Medicine biosafety level 2 animal facility, where sentinel mice were screened every 3 to 6 months for mouse pathogens and determined to be negative. This strain of mouse is susceptible to the strains of HSV-1 used in this study and has been used previously in similar studies (24). Female mice were anesthetized intraperitoneally with ketamine (87 mg/kg of body weight) and xylazine (13 mg/kg). Corneas were bilaterally scarified with a 25-gauge needle and inoculated with 2 × 10⁶ PFU of virus in a volume of 5 μl as previously described (33). Eye swab material was collected and assayed for virus by standard plaque assay as previously described (33). Trigeminal ganglia and 6-mm biopsy punches of periocular skin were removed and placed in preweighed tubes containing 1-mm glass beads and 1 ml of medium. Trigeminal ganglia and periocular skin homogenates were prepared by freezing and thawing the samples, mechanically disrupting in a Mini-Beadbeater-8 (Biospec Products, Bartlesville, Okla.), and sonicating. Homogenates were assayed for virus by standard plaque assay, and the amount of virus was expressed as PFU per milliliter of tissue homogenate. All mice were housed at the Washington University School of Medicine biohazard facility and euthanized when necessary in accordance with all federal and university policies.

Assay for eIF2α phosphorylation.

MEFs were seeded at 2 × 10⁵ cells/well in six-well plates. Upon contact inhibition of growth (24 to 48 h after plating), cells were infected with wild-type or recombinant viruses at a multiplicity of 5 PFU/cell or mock infected. After adsorption for 1 h at 37°C, the inoculum was aspirated and replaced with 1 ml of medium. To harvest protein, medium was aspirated and 500 μl of protein loading dye (50 mM Tris-HCl [pH 6.8], 10% glycerol, 2% sodium dodecyl sulfate [SDS], 100 mM dithiothreitol, 0.1% bromophenol blue) was added. Cells were scraped and collected into sterile 1.5-ml tubes. Lysates were then boiled for 10 min to denature the solubilized proteins. Proteins were electrophoretically separated by SDS-polyacrylamide gel electrophoresis, transferred electrically to polyvinylidene difluoride membrane, and reacted with either a monoclonal anti-eIF2α antibody or a peptide antibody specific for the Ser51-phosphorylated form of eIF2α (Cell Signaling Technology, Beverly, Mass.). Proteins were detected using an ECL-Plus kit (Amersham Pharmacia Biotech, Piscataway, N.J.) essentially as recommended by the manufacturer. To visualize proteins and perform quantification, developed blots were scanned on a Molecular Dynamics STORM 860 PhosphorImager. Signals were quantified using Molecular Dynamics ImageQuant 5.2 software. Antibodies were then removed from the membranes by incubation in stripping buffer (62.5 mM Tris-HCl [pH 6.8], 2% SDS, 100 mM 2-mercaptoethanol) for 30 min at 50°C with shaking. Following stripping, membranes were probed with a monoclonal anti-β-actin antibody (Sigma Inc., St. Louis, Mo.), and protein was detected as described above.

To assess relative levels of phosphorylation, phosphorylated eIF2α signals were normalized to background and to actin as a loading control. The phospho-specific eIF2α signal intensity (normalized to actin) for mock-infected cells was set to 100%, and all other phospho-specific signals were expressed as a percentage of the normalized signal from mock-infected cells.

In vitro growth analysis.

In vitro growth analyses of these viruses were carried out essentially as described previously (3). Either wild-type or Ser51Ala MEFs were seeded into six-well plates at a density of 2.5 × 10⁵ cells/well. These cells were allowed to undergo contact inhibition and were then infected with either wild-type or recombinant viruses at a multiplicity of 0.01 PFU/cell. After 1 h of adsorption, the inoculum was removed and replaced with 1 ml of medium. At various times postinfection, cells were scraped into medium and collected into 1.5-ml tubes. Following disruption of cells by sonication, titers of samples were determined on Vero cells.

RESULTS

The suppressor virus shows intermediate growth in mice.

Since the suppressor deletions present in XN1 and 34.5RΔSUP extend into the TR_s, we wished to ensure that this deletion had not been recombined into the IR_s. Such a recombination event would likely affect the ICP22 and possibly US1 genes. Southern blot analysis on DNA from XN1, 34.5RΔSUP, and Patton viruses revealed no alterations in the BamHI N fragment, which spans the junction between the IR_s and US regions (Fig. 1C). As expected, the Patton lane contained a 5-kb fragment (BamHI N) and showed the natural expansion of the repeats in 500-bp increments. The 5-kb band was also present in the XN1 and 34.5RΔSUP lanes. No fragments smaller than 5 kb, however, were present in the XN1 or 34.5RΔSUP lanes. This demonstrated that the deletion had not been incorporated into the internal repeat region and that effects of the suppressor mutation could not be attributed to recombination at a distal site.

In order to ascertain whether the extragenic suppressor mutations rescued the deficits of an ICP34.5-null virus in the mouse ocular model, mice were infected with 2 × 10⁶ PFU of either wild-type HSV-1 (Patton), an ICP34.5-null virus (SPBg5e), an ICP34.5 second-site suppressor virus (XN1), or a suppressor virus with repaired ICP34.5 genes (34.5RΔSUP) on bilaterally scarified corneas. Growth in the corneas was assessed during the first 5 days of acute infection (Fig. 2). The virus detected 24 h postinfection and later represents newly formed virus, as input virus is not detectable at these times (22, 23). Titers of SPBg5e were consistently 100- to 1,000-fold lower than those obtained with Patton (P < 0.0001 by Student's t test), and during days 2 to 5 SPBg5e was undetectable in greater than 70% of the samples. The suppressor mutant XN1, however, grew to levels intermediate to those of SPBg5e and Patton. XN1 appeared to exhibit early sustained growth during the first 3 days of infection, replicating 5- to 70-fold better than SPBg5e (P < 0.015) but 3- to 30-fold less well than Patton (P ≤ 0.0007). At days 4 and 5 postinfection, however, XN1 titers had fallen to levels indistinguishable from those of SPBg5e. At all time points, 34.5RΔSUP grew equivalently to Patton, indicating that the presence of wild-type ICP34.5 is dominant to any effects of the suppressor mutation.

FIG. 2.

The growth of the suppressor virus (XN1) in corneas is intermediate compared to that of the ICP34.5 mutant virus (SPBg5e) and those viruses encoding wild-type ICP34.5 (Patton and 34.5RΔSUP). Mice were infected on bilaterally scarified corneas with 2 × 10⁶ PFU per eye. Back-titers of inocula were determined following infection to ensure mice had received equivalent doses of virus. Back-titers for SPBg5e, XN1, and Patton were 2.2 × 10⁸ PFU/ml, and for 34.5RΔSUP it was 1.2 × 10⁸ PFU/ml. Data points represent the geometric mean PFU per milliliter of eye swab material ± the standard error of the mean for 12 samples per virus per time point. The limit of detection of this assay, as indicated on the y axis, is 10 PFU/ml.

In order to address spread to and replication in the sensory nervous system, titers in trigeminal ganglia homogenates of infected mice were determined (Fig. 3). Similar to the data for corneal replication, on day 3 (Fig. 3A) XN1 was present at levels intermediate to those with SPBg5e and Patton, growing 100-fold better than SPBg5e and 15-fold less well than Patton (P < 0.0001 for both). On day 5 (Fig. 3B), however, XN1 and SPBg5e titers were equivalent and 100-fold lower than Patton titers (P < 0.0001). Since it has been shown that most of the virus found in periocular skin at day 3 postinfection and beyond is the result of zosteriform spread from the trigeminal ganglia (41), we were interested to determine if differences in viral growth in the ganglia were reflected in replication in the skin (Fig. 4). At 3 days postinfection (Fig. 4A), SPBg5e was undetectable in the skin, likely due to the combinatorial effects of its growth attenuation in both the corneas and the trigeminal ganglia. At this time point, all other viruses tested (XN1, Patton, and 34.5RΔSUP) were replicating to significantly higher levels (P < 0.022). Growth of XN1 at day 3 was just outside statistical significance compared to that of Patton (P = 0.051), though XN1 was not equivalent to 34.5RΔSUP (P = 0.043), suggesting a reproducible difference between XN1 and the ICP34.5-expressing viruses. At day 5 (Fig. 4B), SPBg5e and XN1 were growing equivalently in the skin, suggesting that the attenuation of the ICP34.5 mutant in the trigeminal ganglia provided only a slowing of zosteriform spread and that the skin may be a more permissive tissue for this virus than the corneas or ganglia. Growth of XN1 and SPBg5e at this time point was reduced compared to that of either Patton or 34.5RΔSUP (P < 0.013).

FIG. 3.

The growth of the suppressor virus (XN1) in trigeminal ganglia is enhanced compared to that of the parental ICP34.5-null virus (SPBg5e). At days 3 (A) and 5 (B) postinfection, trigeminal ganglia were removed from mice corneally infected with 2 × 10⁶ PFU per eye. Data sets represent the geometric mean PFU per milliliter of sample material ± the standard error of the mean for at least 12 samples per virus per time point. The mean titer of samples (in PFU per milliliter) is reported above the bars. The limit of detection of this assay, as indicated on the y axis, is 10 PFU/ml.

FIG. 4.

The growth of the suppressor virus (XN1) in periocular skin is enhanced compared to that of the parental ICP34.5-null virus (SPBg5e). At days 3 (A) and 5 (B) postinfection, periocular tissue biopsies were taken from mice corneally infected with 2 × 10⁶ PFU per eye. Data sets represent the geometric mean PFU per milliliter of sample material ± the standard error of the mean for at least four samples per virus per time point. The mean titer of samples (in PFU per milliliter) is reported above the bars. The limit of detection of this assay, as indicated on the y axis, is 10 PFU/ml. *, no virus was detected in the periocular skin of SPBg5e-infected mice at 3 days postinfection.

The suppressor virus shows restored growth in wild-type MEFs.

Viruses lacking ICP34.5 are impaired for growth in MEF cells (3). The initial demonstration of the IE-US11-mediated suppression of an ICP34.5 mutant, however, was performed in SK-N-SH neuroblastoma cells (25). In order to rule out a cell-type-specific suppression of the ICP34.5-null phenotype, we sought to examine the ability of the suppressor virus to grow in a primary cell type normally nonpermissive for ICP34.5 mutants. MEFs were infected at a multiplicity of infection (MOI) of 0.01 PFU/cell and harvested at various points during a 72-h course of infection (Fig. 5A). As expected, SPBg5e replicated to titers 100-fold lower than those obtained using the wild-type virus (P < 0.005). XN1, on the other hand, replicated to levels indistinguishable from those with wild-type virus. This is significant in that it both confirms the ability of the suppressor virus to restore growth to an ICP34.5 mutant in another cell line and helps validate the use of and comparison between these viruses in a mouse pathogenesis model of HSV-1 infection by reproducing in primary murine cells the in vitro suppression phenotype previously seen in immortalized human cells (25, 26, 42).

FIG. 5.

Multistep growth kinetics of SPBg5e, XN1, and Patton in wild-type (A) or Ser51Ala (B) MEFs infected at an MOI of 0.01. Cells were allowed to grow until contact inhibited, and then the cells in one well were trypsinized and counted to determine cell number. Cells and supernatants were collected in 1 ml of medium and sonicated, and virus titer was determined on Vero cells.

Expression of a Ser51Ala mutant of eIF2α restores growth to an ICP34.5 mutant.

In order to determine if the growth phenotypes of the various viruses tested in MEF cells were directly due to their differing abilities to maintain pools of nonphosphorylated eIF2α, we made use of a mutant form of this translation factor which replaces the serine at position 51 with an alanine (Ser51Ala eIF2α). This mutation has been shown to eliminate the phosphorylation site used by PKR, preventing translation inhibition normally caused by active PKR (28, 40). MEFs from mouse embryos carrying a homozygous knock-in Ser51Ala mutation in eIF2α (Ser51Ala MEF cells [isogenic to the wild-type MEF cells used in this study]) were infected at an MOI of 0.01 PFU/cell and harvested over a 72-h time course (Fig. 5B). Interestingly, this host cell mutation allowed all the viruses tested (SPBg5e, XN1, and Patton) to replicate equivalently. All viruses replicated in these cells to levels that were identical to levels seen with wild-type virus in wild-type MEFs, indicating that there is no inherent difference between these two cell types in their ability to support replication of a wild-type virus. Significantly, these data provide direct evidence that the growth attenuation of ICP34.5 mutants in vitro is due entirely to their inability to prevent the accumulation of phosphorylated eIF2α following viral infection and not their inability to counter another activity of PKR.

eIF2α phosphorylation is intermediate in cells infected with the suppressor virus.

To characterize further the relationship between the viral infection and eIF2α phosphorylation levels in a quantitative manner, immunoblot analyses were performed on cells which were mock infected or infected with either SPBg5e, XN1, 34.5RΔSUP, or Patton (Fig. 6). While ICP34.5 mutants display growth defects at low multiplicities in MEFs, these defects are compensated by infecting at a high multiplicity. To ensure synchronous and equivalent infection, and to ensure that effects on eIF2α phosphorylation were not attributable to differences in growth, wild-type MEFs were infected at an MOI of 5 PFU/cell. At 6 h postinfection, Western blotting was performed using either an eIF2α-specific monoclonal antibody or a polyclonal antibody specific for Ser51-phosphorylated eIF2α (Fig. 6A, lanes 1 to 5). To control for loading, the blots were later stripped and reprobed with an anti-β-actin monoclonal antibody and all eIF2α quantification was normalized to actin levels (data not shown).

FIG. 6.

The suppressor virus (XN1) prevents accumulation of PKR-phosphorylated eIF2α. (A) Western blot of phosphorylated eIF2α from wild-type (lanes 1 to 5) or PKR^−/− (lanes 6 to 10) MEFs infected with SPBg5e, XN1, Patton, or 34.5RΔSUP or mock infected. (B) Quantified intensities of duplicate phospho-specific eIF2α bands, normalized to that of actin and compared to the normalized, phospho-specific eIF2α signal from mock-infected cells (set at 100%). Cells were infected at an MOI of 5, and proteins were harvested 6 h postinfection.

By comparing the signal intensity of phosphorylated eIF2α bands (normalized to actin levels) from infected cells to that of mock-infected cells, the effect of viral infection on eIF2α phosphorylation can be quantitatively determined (Fig. 6B, left). Cells infected with SPBg5e displayed an increase in phosphorylated eIF2α, as expected due to this virus' inability to cycle eIF2α back to its nonphosphorylated form. In cells infected with Patton or 34.5RΔSUP, the situation was reversed, showing a decrease in eIF2α phosphorylation over time, as expected for viruses carrying wild-type copies of ICP34.5. The XN1-infected cells, however, displayed levels of phosphorylation similar to those seen in mock-infected cells. This suggests that IE-US11 may effectively inhibit PKR, but there remains a pool of eIF2α that is either stably phosphorylated or is maintained by the action of other eIF2α kinases. Over the course of infection, the levels of total eIF2α did not significantly vary compared to mock-infected cells (data not shown), indicating that changes seen in phosphorylated eIF2α represent changes not only in the total amount of phosphorylation but also in the proportion of total eIF2α that was phosphorylated. In a similar experiment, no phosphorylated eIF2α was detected in Ser51Ala MEFs under any conditions (data not shown).

To determine if the pool of phosphorylated eIF2α seen in the XN1-infected cells is dependent on PKR function, MEFs prepared from PKR-deficient mice were infected and Western blotting was performed as described above. In the absence of PKR, no accumulation of phosphorylated eIF2α was seen in the SPBg5e-infected cells compared to mock-infected cells (Fig. 6A, lanes 6 to 10, and B, right). The level of phosphorylated eIF2α decreases in cells infected with viruses carrying wild-type copies of ICP34.5 (Patton and 34.5RΔSUP), consistent with ICP34.5 acting directly on eIF2α. In XN1-infected cells, levels of phosphorylated eIF2α remained comparable to that in mock-infected cells. These data suggest that IE-US11 is a robust inhibitor of PKR and that the phosphorylated eIF2α seen in XN1-infected cells represents the activity of other eIF2α kinases. Interestingly, these data also show that other eIF2α kinases are either not induced following HSV-1 infection or are inhibited by the virus in an ICP34.5-independent manner.

Specificity controls.

In order to ensure that the restored growth of SPBg5e seen in Ser51Ala MEFs was specific to an ICP34.5 mutant and not a general phenomenon with this cell type and any growth-impaired virus, in vitro growth assays were performed using an HSV-1 mutant deficient in thymidine kinase (17Δtk). This virus displays growth deficits in MEFs and in vivo (43). Wild-type or Ser51Ala MEFs were infected with either 17Δtk or its marker-rescued virus, 17tkR, and growth assays were performed as described in Materials and Methods (Fig. 7A and B). In wild-type MEFs (Fig. 7A), 17Δtk was impaired relative to 17tkR (P < 0.0001). In two independent experiments, 17Δtk grew significantly less well than 17tkR (P ≤ 0.005) in Ser51Ala MEFs (Fig. 7B). Examination of the pooled data suggested that the Ser51Ala MEFs represent a slightly more permissive host cell for all viruses than the wild-type cells (P = 0.07). This was expected, given the effect of eIF2α phosphorylation on multiple cellular pathways as well as protein synthesis regulation. These data, however, show that loss of eIF2α phosphorylation does not nonspecifically restore wild-type growth to growth-defective and attenuated viruses.

FIG. 7.

Specificity controls. (A and B) Multistep growth kinetics of 17Δtk and 17tkR in wild-type (A) or Ser51Ala (B) MEFs infected at an MOI of 0.01. Cells were allowed to grow until contact inhibited, and then the cells in one well were trypsinized and counted to determine cell number. Cells and supernatants were collected in 1 ml of medium and sonicated, and virus titer was determined on Vero cells. (C) eIF2α phosphorylation assay. Wild-type MEFs were infected with 17Δtk or 17tkR at an MOI of 5 or mock infected. Six hours postinfection proteins were harvested and Western blotting was performed. Band intensities of duplicate phospho-specific eIF2α bands were normalized to actin and compared to the normalized, phospho-specific eIF2α signal from mock-infected cells (set at 100%).

Since XN1 and Patton grow to similar levels in wild-type MEFs and neither of these viruses cause an increase in phosphorylated eIF2α, we wanted to ensure that the accumulation of phosphorylated eIF2α seen with SPBg5e was due specifically to a lack of ICP34.5 and not a general effect of infection with an attenuated virus. To address this, eIF2α phosphorylation assays were performed on wild-type MEFs infected with either 17Δtk or 17tkR (Fig. 7C). A decrease in phosphorylated eIF2α, relative to that in mock-infected cells, was observed in cells infected with either virus, though the decrease in phosphorylation seen following 17Δtk infection was not as robust as the decrease induced by 17tkR. As HSV-1 genes expressed with late kinetics (including ICP34.5) depend on viral DNA replication for optimal expression and thymidine kinase is involved in DNA replication, the difference between 17Δtk and 17tkR in this assay may be due to suboptimal expression of ICP34.5. Despite this difference, the accumulation of phosphorylated eIF2α seen in SPBg5e-infected cells can be specifically attributed to a lack of ICP34.5.

DISCUSSION

The ability of HSV-1 to inhibit the PKR response plays a critical role in determining the outcome of viral infection in the host. This is made clear by work previously done in our lab using ocular and intracranial infection models. While an ICP34.5-deficient virus remained completely avirulent in these models, the same virus caused 100% mortality and grew to wild-type levels following infection of PKR-deficient mice (24). These in vivo data are consistent with in vitro observations recently published indicating that the growth of an ICP34.5 mutant in wild-type MEFs is restored in PKR^−/− MEFs (7). In this regard, it was surprising to discover that the extragenic mutation present in the suppressor viruses, while restoring to an ICP34.5 mutant the ability to grow in nonpermissive cells, did not restore virulence following intracranial infection (26).

The mechanism by which ICP34.5 counters the PKR response may render it intrinsically more potent than US11. The phosphorylation of eIF2α is a key checkpoint in several stress signaling pathways, and four mammalian kinases (PKR, PERK, GCN2, and HRI) have been described which may phosphorylate Ser51 of eIF2α in response to a variety of stimuli (2, 6, 15, 18, 44). The interaction of ICP34.5 with the host PP1α results in direct dephosphorylation of eIF2α, circumventing several pathways that may be active in an infected cell. The prevailing model for US11-mediated PKR inhibition, however, is that it binds to and inhibits PKR by either acting as a pseudosubstrate or by blocking activation by PACT or double-stranded RNA. US11, therefore, is only able to act through inhibition of a single pathway, and so the effect of US11 on eIF2α phosphorylation would likely be less robust than the redirected enzymatic activity of the ICP34.5/PP1 complex. Secondly, other eIF2α kinases may be active during infection, although in vitro data presented here suggest that these other kinases are either not activated above basal levels by viral infection or are inhibited by the virus.

The observations from this study, however, support the conclusion that even relatively small differences in the balance of phosphorylated and unphosphorylated eIF2α may have important consequences for the pathogenesis of HSV-1, shifting the balance of infection towards clearance by the immune system. The suppressor mutants are less able to maintain large pools of active eIF2α compared to wild-type virus, although they also block the overaccumulation of inactive eIF2α seen in cells infected with ICP34.5 mutants. Whether the phosphorylated eIF2α seen in XN1-infected cells represents a stable pool present before infection or the activity of other eIF2α kinases is unclear at this time. Similarly, in a mouse model, the replication of XN1 was intermediate between SPBg5e and wild-type levels following corneal infection, while remaining nonneurovirulent following intracranial infection. As PKR plays a role in regulating growth and differentiation, the incongruity between growth and virulence results may reflect differential regulation of translation in these tissues. It is important to note that in the initial characterization of the suppressor virus, protein synthesis was not restored to wild-type levels as measured by [³⁵S]methionine incorporation, although replication was restored to near wild-type levels (25). These data, combined with the data presented here, suggest that the threshold levels of eIF2α phosphorylation that prevent robust replication of HSV-1 may be different in vivo than in continuous cell lines in vitro. This may be true of different tissues in vivo as well, as the ICP34.5-deficient SPBg5e displayed a large increase in viral titers between days 3 and 5 in the periocular skin, while over the same period in the trigeminal ganglia there was essentially no change in titers. Factors influencing tissue permissivity may include differentiation state and cellular division rate. The skin represents a more mitotically active tissue than the trigeminal ganglia and may less tightly regulate eIF2α phosphorylation. This is clearly not the only factor in determining tissue permissivity, however, as SPBg5e is severely impaired in the infected corneal epithelium, which also represents a rapidly regenerating, mitotically active tissue.

One potential caveat to these studies is that the suppressor deletion affects multiple genes. While the expression of US11 as an immediate-early gene is sufficient for the in vitro suppression of an ICP34.5 mutation, the associated loss of ICP47 expression from the US12 gene may result in increased antigen presentation (16, 27). While ICP47 appears to be important for neurovirulence, it does not, however, affect replication of HSV-1 in the cornea (14). Based on these data and the data provided in this study using 34.5RΔSUP, which encodes wild-type ICP34.5 but also contains a suppressor deletion, we believe that it is unlikely that US12 deletion contributes to the in vivo phenotype of XN1. It remains possible, however, that the restoration of growth seen with XN1 in vivo would be more complete in the presence of ICP47 function.

An important outcome of this study was the demonstration that the Ser51Ala eIF2α cells restored the growth of an ICP34.5 mutant. While ICP34.5 has been shown previously to circumvent the effects of activated PKR, PKR has multiple roles in infected cells. Biochemical studies have shown that ICP34.5 can prevent accumulation of phosphorylated eIF2α, but the relevance of this interaction to replication has not been assessed. This use of a double-genetic approach with mutant virus and mutant cells directly demonstrates that successful wild-type replication requires down-regulation of eIF2α phosphorylation, mediated by ICP34.5 or US11. Using this approach in vivo is unfortunately impossible, since homozygous mice carrying the Ser51Ala eIF2α allele do not survive beyond 12 h postpartum, which is insufficient for pathogenesis studies (39).

Taken together, the data presented here provide evidence that US11 may act as a PKR antagonist in vivo. It is unclear, however, if this role is important in the context of a virus that expresses US11 under the control of its own promoter. It is important to note that the role of US11 is unclear and several functions have been ascribed to it, including RNA binding (20, 35-37), intracellular trafficking (12, 21), and the regulation of gene expression (1, 11, 38). Some of these activities may be related; it appears that the RNA-binding domain is important for inhibiting PKR (4, 20, 32). The effect of US11 on pathogenesis in an animal model of HSV-1 infection has not been thoroughly examined. In one study (29), a virus that carried a spontaneous deletion affecting multiple genes including US11 was found to be as virulent as wild-type virus following intracranial infection but had slightly higher 50% lethal dose values when inoculated at peripheral sites. These data may represent the same types of tissue- and route-dependent effects displayed by the suppressor virus. It also appears that US11 is dispensable for latency (29). It has been suggested that the PKR-antagonizing function of US11 may represent a role that was at least partially supplanted following the acquisition by the herpes simplex viruses of ICP34.5, a GADD34 homologue (5). In this light, it is possible that the presence of ICP34.5 may mask any effect on the virus due to loss of US11 function. It will be of interest, therefore, to examine the pathogenesis of an HSV-1 mutant lacking any PKR inhibitory function.