|Home | About | Journals | Submit | Contact Us | Français|
Cytotoxic T cells are important in controlling herpes simplex virus type 2 (HSV-2) reactivation and peripheral lesion resolution. Humans latently infected with HSV-2 have cytotoxic T cells directed against epitopes present in tegument proteins. Studies in mice of immunity to HSV have commonly focused on immunodominant responses in HSV envelope glycoproteins. These antigens have not proved to be an effective prophylactic vaccine target for most of the human population. The murine immune response against HSV tegument proteins has not been explored. We analysed cellular responses in BALB/c mice directed against the tegument proteins encoded by UL46, UL47 and UL49 and against the envelope glycoprotein gD after DNA vaccination or HSV-2 infection. After DNA vaccination, the splenocyte T-cell response to overlapping peptides from UL46 and UL47 was more than 500 gamma interferon spot-forming units per 106 responder cells. Peptide truncation studies, responder cell fractionation and major histocompatibility complex binding studies identified several CD8+ and CD4+ epitopes. Cellular responses to tegument protein epitopes were also detected after HSV-2 infection. Tegument proteins are rational candidates for further HSV-2 vaccine research.
Herpes simplex virus type 2 (HSV-2) infections are common, occurring in about 17% of adult US citizens (Xu et al., 2006). HSV-2 causes significant morbidity and can be fatal, especially after congenital transmission (Kimberlin, 2004, 2005). Vaccines targeting HSV-2 envelope glycoproteins have had limited clinical activity against infection and disease (Corey et al., 1999; Stanberry et al., 2002). Therapeutic immunization with HSV-2 glycoprotein D (gD) vaccine led to modest decreases in lesion recurrence and severity in infected individuals, suggesting that immunotherapy may be possible (Straus et al., 1994).
Studies of the immune response to HSV-2 in humans and mice suggest that T-cell responses are important in the control of disease (Koelle & Corey, 2003). HSV-specific CD8+ T cells producing gamma interferon (IFN-γ) persistently infiltrate latently infected murine ganglia (Khanna et al., 2003), and HSV-1-reactive CD8+ T cells with cytolytic and IFN-γ effector functions have been recovered from human trigeminal ganglia (Verjans et al., 2007). Infiltration of HSV-2-specific CD4+ and CD8+ cytotoxic T lymphocytes (CTLs) correlates with clearance of infectious HSV-2 from recurrent genital HSV-2 lesions in humans (Koelle et al., 1998b). HSV-2-specific CD8+ CTL specific for tegument proteins have been visualized at the dermal–epidermal junction, near sensory nerve endings, in human skin biopsies of genital HSV-2 lesions (Zhu et al., 2007). Taken together, these observations suggest that immunotherapeutic approaches to boost HSV-2-specific CD8+ T-cell responses could result in decreased HSV-2 lesion formation and shedding from the periphery.
Tegument-specific T cells are prevalent and abundant in humans (Hosken et al., 2006), suggesting their importance in control of the virus. The tegument proteins virion protein (VP)11/12, VP13/14 and VP22 are encoded by HSV-2 open reading frames (ORFs) UL46, UL47 and UL49, respectively. Epitopes have been identified in HSV-2 VP11/12 (Koelle et al., 2003), VP13/14 (Koelle et al., 2001) and VP22 (Koelle et al., 2001) that are recognized by cloned CD8+ T cells in the context of human leukocyte antigen (HLA) A*0101, A*0201 and B*0702, respectively.
Additionally, CD4+ T cells from infected humans recognize epitopes in HSV-2 VP11/12 (Posavad et al., 2003; Verjans et al., 2000), VP13/14 (Posavad et al., 2003) and VP22 (Koelle et al., 1998a, 2000a). These CD4+ T-cell responses may be important in generating and maintaining effective CD8+ T-cell memory (Northrop & Shen, 2004).
Mouse models have long been used to study cellular and humoral immune responses after HSV-2 infection. However, detailed understanding of the murine cellular immune response after HSV-2 infection is limited. Strain-specific immunodominant CD8+ T-cell epitopes have been described in C57BL/6 (Stock et al., 2006; Wallace et al., 1999) and BALB/c (Banks et al., 1993; Haynes et al., 2006) mice. C57BL/6 mice direct over 70% of their cellular immune response against a single, type-common HSV epitope in the gB envelope glycoprotein (Stock et al., 2006; Wallace et al., 1999) and adaptive immunity against this single epitope can confer protection against subsequent HSV challenge by various routes (Blaney et al., 1998; Gierynska et al., 2002; Kumaraguru et al., 2002; Orr et al., 2007). BALB/c mice are more susceptible to HSV infection than C57BL/6 mice (Lopez, 1975), but HSV-specific cellular immune responses in BALB/c mice are less well-characterized. Little data are available on subdominant cellular immune responses to HSV in mice (Stock et al., 2006). The ability of tegument proteins to prime T-cell immunity in mice has not been studied. We included studies of the more established immunogen, gD (encoded by ORF US6), as a comparison.
The current study uses plasmid DNA constructs and an attenuated HSV-2 virus to elucidate tegument-specific cellular immune responses in BALB/c mice. The tegument-specific cellular response after DNA immunization is shown to include both CD4+ and CD8+ epitopes. Some epitopes have functional avidities in the nanomolar range. Several of these epitopes are shown to be primed in the context of infection with an attenuated strain of HSV-2. These data provide the initial description of the tegument-specific cellular immune response in BALB/c mice, add additional tools to the study of cellular immunity to HSV-2 and suggest that further studies to evaluate therapeutic vaccines directed against tegument protein epitopes are warranted.
The sequences of the plasmid DNA constructs of tegument genes were based on clinical HSV isolates. HSV-2 was isolated from swab samples of subjects with serologically documented symptomatic primary genital herpes (Ashley et al., 1988). Virus was isolated using mink lung cells and one cycle of serial dilutions in Vero cells; a well at the highest dilution exhibiting cytopathic effect was chosen to approximate one plaque purification. Virus was expanded on Vero cells and DNA was prepared using a DNA extraction kit (Qiagen).
A thymidine kinase-deficient HSV-2 strain 333 (tk−-HSV-2) (Stanberry et al., 1985) was used to infect groups of mice. Virus was grown and titrated on mycoplasma-free Vero cells using standard methods.
The UL49 and gD (or US6) ORFs were amplified by PCR as single amplicons. For UL46 and UL47, three overlapping regions were amplified using Pfu DNA polymerase (Stratagene). PCR products were gel-purified (Qiagen) and sequenced (BigDye 3.0; Applied Biosystems) (amplification and sequencing primers are available on request from the authors). Bidirectional coverage was achieved and ambiguities were resolved with Seqman II (Lasergene). Sequences were aligned with clustal w (megalign; Lasergene).
Synthetic HSV-2 genes were codon-optimized using proprietary algorithms developed at Vical. DNA was synthesized by GeneArt and sequenced to threefold redundancy. HSV-2 genes were subcloned into expression plasmid VR1012 (Hartikka et al., 1996). pDNA vaccine test lots (Endo-Free Gigapreps; Qiagen) were negative for endotoxin and their identity was confirmed by sequencing to twofold redundancy. Full-length HSV-2 genes were used for UL46, UL47 and UL49. For gD, amino acids (aa) 1–340 were used, corresponding to the extracellular domain but deleting the transmembrane and cytoplasmic regions. In contrast with some studies with gD variants (Higgins et al., 2000), we retained the N-terminal leader and signal domains.
Vaccines were tested in either PBS or with 1.5 mg ml−1 poloxamer-based non-ionic copolymer adjuvant CRL1005 (Todd et al., 1997). Frozen vaccines were thawed at ambient temperature immediately prior to injection.
All protocols were approved by the University of Washington Institutional Animal Care and Use Committee. Female BALB/c mice, aged 4–8 weeks (Charles River) were housed five to a cage in a modified specific pathogen-free environment. Mice were inoculated with virus or DNA vaccine by 14 weeks of age.
For vaccination using pDNA, mice were injected intramuscularly into each rectus femoris with a total of 100 μg of vaccine (50 μg pDNA in 50 μl injection per side) using a three dose regimen, with doses delivered 2 weeks apart. Negative control groups received empty vector VR1012 administered at the same dose and to the same schedule. In some experiments, mice who completed immunization using the three-dose regimen more than a month previously received an additional 100 μg dose at least 2 weeks prior to evaluation of their immunological response.
Immunization of mice with live tk−-HSV-2 virus used intravaginal inoculation (Parr et al., 1994). Six days prior to intravaginal instillation of virus, each mouse received 2 mg subcutaneous medroxyprogesterone acetate (Depo-Provera) delivered in 50 μl. Vaginas of anaesthetized mice were swabbed with sterile polyester swabs (Copan) prior to infection, and 10 μl virus diluted in 0.9% (w/v) sodium chloride was delivered via micropipette.
Mice were anaesthetized with ketamine–xylazine and blood was obtained by retro-orbital bleed 1 day prior to each DNA immunization and sacrifice. Serum was stored at −20 °C. Mice were sacrificed by CO2 inhalation. Spleens were homogenized in RPMI 1640 (Invitrogen) with 2% fetal bovine serum (FBS) (Intergen). Red blood cells were lysed with ACK buffer (Invitrogen). Splenocytes were washed, counted and maintained for enzyme-linked immunospot (ELISPOT) assay in complete medium [RPMI 1640 with 2 mM l-glutamine, 5% FBS, 1 mM sodium pyruvate, 0.1 mM non-essential amino acids, 1% penicillin–streptomycin (Invitrogen) and 20 mΜ 2-ME (Sigma)]. Splenocytes were enriched for CD4+ or CD8+ T cells using negative selection (Miltenyi) and were found to be >90% pure by flow cytometry.
Polypeptides (13 aa long, overlapping by 9 aa) were synthesized by Synpep (UL49) or Chiron Mimotopes (UL46 and UL47). Shorter peptides were synthesized by CPC Scientific . The UL47 and UL49 13 aa polypeptides matched strain HG52 (Dolan et al., 1998), while the UL46 peptides matched a consensus of wild-type sequences (see Results). Each peptide was analysed by mass spectroscopy. Peptides that failed synthesis were not included in T-cell tests; these included one peptide in UL47 and two in UL49 (data available on request from the authors). Similar peptides were prepared covering the gD sequence. Peptides were stored at 10 mg ml−1 at −20 °C in DMSO until use.
HSV-2 tegument proteins were expressed by transiently transfecting VM92 cells with candidate vaccine plasmids and collecting supernatants. Protein was stored at −80 °C until use. Seroconversion was tested by ELISA using a method similar to that described by Milligan & Bernstein (1995). Briefly, 96-well flat-bottom plates (Costar) were coated with a 1:5 dilution of protein in NaHCO3 buffer. Three to five serial threefold dilutions of serum (beginning at 1:100) were added overnight at 4 °C. Immunoglobulin (Ig) G antibodies were detected by addition of biotinylated goat anti-mouse Fc (Biodesign), followed by streptavidin-horseradish peroxidase (HRP) (Pierce) and TMB peroxidase substrate (KPL). Plates were read at OD450 using a Fusion or Victor3 1420 plate reader (Perkin Elmer).
Pooled sera from human subjects who were HSV-2-infected or HSV-1/HSV-2-dually seronegative (Ashley et al., 1988) were obtained from Dr R. Morrow (University of Washington). Subjects gave informed consent in a University of Washington Institutional Review Board-approved protocol. Testing used a similar ELISA protocol using serial dilutions in BTBST [Tris-buffered saline (50 mM Tris-HCl, 0.2 M NaCl, 3 mM KCl, pH 9) with 0.05% Tween-20 and 0.1% BSA] and goat anti-human-IgG-HRP (Chemicon) for detection.
Transformed monkey kidney cells (COS-7) were cultured in DMEM (Invitrogen) containing 1% penicillin-streptomycin, 2 mM l-glutamine and 10% heat-inactivated FBS (Hyclone). Cells were seeded at 9000 cells per well in 96-well flat-bottom plates and transfected the next day using 50 ng per well full-length cDNA encoding HLA A*0201, A*0101 or B*0702 with Fugene 6 (Roche) (Koelle, 2003). Cells were co-transfected with increasing concentrations of vaccine pDNA or full-length HSV-2 strain HG52-origin UL46 or UL47 in pCDNA3.1-HIS vectors (Invitrogen) or full-length HSV-2 UL49 fused to the C-terminal of enhanced green fluorescent protein (eGFP) in pEGFP-C1 (Clontech) (Koelle et al., 2000b). After 2 days, cloned CD8+ T cells specific for known epitopes in UL46, UL47 or UL49 (Koelle et al., 2001, 2003) were thawed, washed and added at 5×104 cells per well in 120 μl T-cell medium (Koelle et al., 1993) containing 2 U ml−1 recombinant human IL-2 (Chiron). After 24 h, supernatants were removed and assayed for IFN-γ by ELISA (Koelle et al., 2001).
For IFN-γ ELISPOT assays, plates (Millipore) were coated with 10 μg ml−1 rat IgG1 anti-murine IFN-γ monoclonal antibody (mAb) clone R4-6A2 (BD Pharmingen) in 50 μl per well of 0.2 M carbonate/bicarbonate buffer, pH 9.4 (Pierce), and held at 4 °C for use 12–48 h later. Plates were washed three times in PBS and blocked for 2 h at 37 °C with RPMI 1640 with 2% heat-inactivated FBS. Splenocytes (5×105 or 1×106 cells per well) and stimulator peptides were added separately in 100 μl complete medium. Concanavalin A (Sigma) at 2.5 μg ml−1 was the positive control; medium was the negative control. After 18–24 h at 37 °C in a humidified 5% CO2 incubator, plates were washed once with water and three times with PBS. Detection used 1 μg ml−1 biotinylated rat anti-mouse IFN-γ (BD Pharmingen) in 50 μl per well of PBS, incubated at 4 °C overnight or at room temperature for 2 h. After 5 washes with PBS, alkaline phosphatase-streptavidin (Bio-Rad), diluted 1:1000 in PBS, was added in 50 μl for 60 min at room temperature. After five washes, substrate (Bio-Rad) was added at 50 μl per well for 3–5 min. Reactions were stopped with water and plates were air-dried and counted using a Biosys Bioreader 3000 ELISPOT reader (Karben) or an ImmunoSpot Series 1 reader (Cellular Technology). All plates within single experiments were read with the same plate reader. ELISPOT data are reported as means of duplicate measurements. Responses to DMSO control antigen were less than 12 spot-forming units (s.f.u.) per 106 splenocytes; values below 10 s.f.u. per well were considered negative.
Splenocytes from individual immunized animals were tested against pooled peptides (18–24 peptides per pool) at final concentrations of 10 μg ml−1 peptide. DMSO concentrations were held below a final concentration of 0.2%. Responses to single peptides were measured in duplicate using pooled splenocytes at 1×106 cells per well and peptides in serial 10-fold dilutions. The effective concentration of peptide that stimulates half the maximal number of IFN-γ s.f.u. per 106 responder cells (EC50) was estimated graphically from s.f.u. versus peptide concentration curves. Single peptides were tested with CD4+ or CD8+ splenocyte responders from individual immunized animals at 1 μg ml−1. For these assays, splenocytes from immunized animals were enriched by negative selection for either CD4+ or CD8+ cells, and used at 0.3×105–5×105 cells per well. These cells were stimulated with individual peptides mixed with 0–9×105 splenocytes per well from naïve adult female mice used for antigen presentation. ELISPOT for tk−-HSV-2-infected animals used splenocytes from individual animals and single peptides at 1 μg ml−1. For some ELISPOT assays, a BALB/c CD8+ T-cell epitope in HSV-2 ICP27 (Haynes et al., 2006) was used at 1 μg ml−1 as a positive control and comparison.
Selected peptides were evaluated for binding to H-2Kd, using an inhibition assay (Sidney et al., 1998; Tscharke et al., 2005). Briefly, radiolabelled, tightly binding control peptide and test peptide were co-incubated at room temperature with H-2 molecules in the presence of 1 μM human β2-microglobulin (Scripps Laboratories). After 2 days, binding of radiolabelled peptide to H-2Kd was determined by capturing MHC–peptide complexes on Greiner Lumitrac 600 microplates (Greiner Bio-One) coated with mAb SF1-1.1.1 and measuring bound radioactivity (Topcount). The concentration of peptide that inhibited binding of the radiolabelled peptide by half was calculated (IC50).
Comparison of s.f.u. values between vaccinated and unvaccinated groups of mice was done using unpaired, two-tailed t tests.
To construct the tegument-based DNA plasmids used in this study, we needed to know whether there were any polymorphisms in the tegument genes in different strains of HSV-2. There is a single full-length genomic HSV-2 sequence in GenBank (strain HG52; Dolan et al., 1998). It is not known how tegument genes of clinical strains differ from those in this sequence and from those in another common HSV-2 laboratory strain, 186. We sequenced the UL46, UL47, UL49 and US6 genes from strain 186 and from six to eight wild-type HSV-2 strains obtained from subjects with primary genital HSV-2 infections. Coding differences from the laboratory reference strain HG52 (Dolan et al., 1998) were identified at several loci, principally in UL46 and UL47. Using these data, we derived consensus amino acid vaccine sequences for UL46 and UL47 plasmid vaccines (Table 1). For US6 (gD), only the previously reported polymorphism V353A (Terhune et al., 1998), and new polymorphisms V169A and L375P, were detected among six wild-type genes. The detected coding polymorphisms in UL49 were sporadic and variable. Therefore, plasmid vaccine coding sequences for UL49 and US6 were identical to the predicted polypeptide for strain HG52. The tegument plasmids encoded predicted full-length ORFs, while the US6 plasmid was designed to encode aa 1–340 of gD, omitting the transmembrane and signal peptide domains of gD (Watson, 1983). Non-coding nucleotide modifications were also made with the goal of optimizing protein expression within eukaryotic cells. HSV-2 strain 186 differed from the vaccine-encoded protein at four loci in UL46 (aa 587–588, 594, 613 and 634), at four loci in UL47 (aa 38, 156, 266 and 697–698) and at one locus in UL49 (insertion of two amino acids after aa 72).
HSV-2 genes were cloned into VR1012, a standard vaccine vector featuring a cytomegalovirus promoter and intron, a polyadenylation signal and kanamycin resistance (Hartikka et al., 1996). To confirm expression of HSV-2 tegument epitopes, COS-7 cells were co-transfected with plasmids expressing HSV-2 genes and cDNA encoding human HLA class I heavy chain. Human CD8+ T-cell clones known to respond to epitopes within the relevant tegument proteins were added (Koelle et al., 2001, 2002). We detected specific IFN-γ secretion in response to candidate synthetic vaccine constructs, as well as to positive control wild-type tegument protein genes PCR-cloned from strain HG52 into alternative eukaryotic expression vectors (Fig. 1a–c). The T cell clones did not recognize COS-7 cells transfected with HLA class I cDNA alone, pDNA alone or co-transfected with pDNA and the incorrect HLA class I cDNA in any cases (data not shown). Similarly, sera from HSV-2 seropositive individuals bound recombinant tegument proteins (Fig. 1d). Sera from HSV-2 seronegative individuals did not react with the recombinant HSV-2 proteins. Expression of gD was confirmed by transfection of COS-7 cells and immunoblot with a gD-specific mAb (data not shown).
We immunized mice with the tegument plasmids using a three-dose regimen. Every mouse responded with production of antibodies against the relevant tegument protein by the third immunization (data not shown). Overlapping peptides were used to measure BALB/c cellular immune responses to the vaccines. Peptides were 13 aa long, covered the full-length of each tegument protein or gD and were offset by 4 aa. Initial pooled peptide assays used splenocytes from single animals, allowing demonstration of individual animal responses but not providing epitope specificity. Follow-up single-peptide breakdowns of positive pools used pooled splenocytes from groups of immunized animals. For peptide pools, BALB/c responses to all tegument proteins were statistically significantly different from naïve mice. Mice immunized with UL46 or UL47 plasmid had an overall mean response of >500 s.f.u. per 106 splenocytes, compared with <10–20 s.f.u. per 106 splenocytes in naïve mice (P<1×10−14), while the overall mean response of mice immunized with UL49 plasmid was 231 s.f.u. per 106 splenocytes compared with 11 s.f.u. per 106 splenocytes in naïve mice (P=0.003).
Positive pools were used to select individual peptides for further evaluation of tegument epitopes. For gD, the initial round of evaluation of cellular responses used individual peptides covering the full sequence of the immunogen, rather than pools. ELISPOT assays with the constituent peptides within the tegument protein pools disclosed numerous individual 13 aa peptides that stimulated IFN-γ responses in immunized BALB/c mice. Overall, reactive single peptides were identified in each tegument protein and in gD (Table 2).
The 13-mer peptide data and epitope prediction algorithms (Bui et al., 2005; Parker et al., 1994; Rammensee et al., 1999) were used to select shorter peptides for further evaluation in BALB/c mice. If responses were noted in whole splenocytes, the phenotype of responding cells was determined using CD4+- or CD8+-enriched splenocytes isolated from mice immunized with plasmid constructs (Table 3). Strong H-2d CD8+ epitopes were identified within each tegument protein and in gD.
Peptide titration generally showed that each protein studied contained at least one CD8+ T-cell epitope in BALB/c mice with an EC50 value in the picomolar to nanomolar range (Table 3). Some CD8+ T-cell epitopes showed biological activity at concentrations as low as 1×10−12 M (Fig. 2). Two peptides predicted to bind H-2Kd were tested in H-2Kd binding assays. UL46 peptide 183–191 (KYAAAVAGL) had an IC50 of 9.91 nM and UL49 peptide 200–208 (VFCAAVGRL) had an IC50 of 371 nM.
To determine whether immune responses to tegument proteins occur in the context of HSV-2 infection, BALB/c mice were inoculated intravaginally with an attenuated HSV-2 strain. Infection with tk−-HSV-2 generally led to only mild and transient symptoms (McDermott et al., 1984; Milligan & Bernstein, 1995), though at higher doses, chronic-appearing lesions were sometimes observed. All mice survived infection and showed seroconversion by ELISA using either whole HSV-2 lysate or tegument proteins encoded by the UL46 or UL49 plasmids as the test antigens (not shown), while six of ten mice seroconverted to the UL47 gene product. Splenocyte IFN-γ ELISPOT assays were performed with a subset of the optimized peptides identified using DNA-immunized mice (above). Infected mice showed splenocyte IFN-γ responses to tegument and gD peptides (Table 4). Responses were detected to CD4+ epitopes in UL49 and gD, and to CD8+ epitopes in UL46, UL47 and UL49. The magnitude of the responses were lower than for a known BALB/c CD8+ epitope in HSV-2 ICP27, aa 318–326 (Haynes et al., 2006) (HGPSLYRTF).
The HSV-2 tegument proteins VP11/12, VP13/14 and VP22 encoded by UL46, UL47 and UL49, respectively, stimulate relatively immunodominant HSV-specific cellular immune responses in humans. The human population prevalence of CD8+ T-cell responses to these ORFs is more than 70% (Hosken et al., 2006) and the numerical level of tegument-specific CD8+ T cells within latently infected individual human subjects is up to 0.6% of CD8+ T cells (D. M. Koelle and others, unpublished). The abundance of HSV-2-specific CD8+ CTLs is inversely correlated with HSV-2 symptoms (Posavad et al., 1997). Recurrent HSV-2 lesions are infiltrated by tegument-specific CD8+ T cells, which persist at the dermal–epidermal junction and increase during subclinical recurrences (Zhu et al., 2007). Keratinocytes, the predominant cell type infected with HSV-2 in mucocutaneous lesions, are recognized by tegument-specific CTLs (Koelle et al., 2001). Together, these observations suggest the importance of tegument-specific CTLs in the control of HSV-2 recurrences.
We showed in this study that BALB/c mice, like humans, generate T-cell responses directed against tegument protein epitopes. We characterized in detail the relative strength of responses to different epitopes after immunization using tegument-based DNA vaccines (Tables 2 and 33 and Fig. 2). We also showed that tegument-directed responses are relevant after intravaginal infection with an attenuated HSV-2 strain (Table 4), though we did not test all peptide epitopes characterized in vaccinated mice in the challenge model. T cells were identified on the basis of IFN-γ production, suggesting that functional antiviral responses against tegument proteins are generated in mice after infection and immunization. Additional favourable T-cell characteristics may include magnitude, diversity and avidity. Overall, the magnitude of T-cell responses to UL46 and UL47 were higher than responses to UL49. The within-ORF epitope diversity in BALB/c mice was largest in UL47. In this study, CD8+ T cells stimulated by DNA vaccination were found to have low EC50 in many cases, and we measured relatively strong peptide binding to H-2Kd for UL46 183–191 and UL49 200–208 epitopes. We hope to define the MHC binding molecules of additional peptide epitopes to permit testing of peptide–MHC tetramers.
The diversity of tegument-specific T-cell responses in mice may be even greater than we measured. Overly long or incorrectly synthesized peptides could lead to missed epitopes. It is uncommon for peptide-based T-cell surveys to document peptide identity by mass spectroscopy for every peptide. We took this measure and confirmed synthesis for all peptides in the library. Also, because 13-mers may be less efficient agonists for CD8+ than for CD4+ responses (Kiecker et al., 2004; Maecker et al., 2001), CD8+ responses may have been disproportionately ‘missed’ in our assays. Our peptide libraries were based on the HG52 sequence which may have led to missed epitopes in UL47, as the UL47 vaccine was based on the consensus sequence from clinical strains (Table 1). Although the IFN-γ ELISPOT assay format is sensitive, some CD4+ T-cell responses may have gone undetected because some virus-specific T cells produce IL-2 in response to antigen but are IFN-γ-negative (Betts et al., 2004; Waldrop et al., 1997). Finally, ELISPOT detection of CD4+ T-cell responses using the overlapping peptide approach often employs peptides larger than the 13 aa length we used (Klinman, 2008; Mashishi & Gray, 2002), which may increase the ability to detect some CD4+ epitopes.
CD8+ CTL responses to complex viral pathogens can display remarkable immunodominance; for example, in C57BL/6 mice, more than 70% of the CD8+ CTL response to HSV-1 is directed against a single epitope in gB (Stock et al., 2006). While we have discovered new CD8+ HSV-2 epitopes in H-2Kd haplotype mice, further research will be required to define dominance hierarchies. Our ELISPOT data indicate that an epitope in ICP27 may stimulate more abundant responder cells in BALB/c mice than epitopes in UL46, UL47 or UL49. This finding provides a system for detailed investigation of the genesis of CD8+ T-cell memory responses in the setting of vaccination. Overall, our data provide additional details on the complex cellular response generated against HSV-2 proteins.
In addition to our focus on tegument proteins, which have never been evaluated in mice before, we evaluated responses to gD. At least two novel epitopes were identified. The first is a CD4+ epitope in the hydrophobic transmembrane domain that is part of the signal sequence, at aa 13–25. The high avidity epitope at aa 157–169 is, we believe, the first CD8+ epitope in HSV-2 gD to be identified in mice. This latter epitope may be useful because an HSV-2 gD vaccine with CD4+ T-cell and antibody stimulatory activity has partial activity in humans (Stanberry et al., 2002). We can now test CD8+ T-cell responses to this antigen in mice using vaccine platforms that have the potential to prime or boost this type of response in addition to antibodies and CD4+ T cells.
There is an unmet medical need for immunological approaches to prevent or modify HSV infections. Goals for therapeutic vaccination include reduction of symptomatic recurrent lesions and symptoms, and also reduction of HSV-2 shedding and optimal transmission. Therapeutic vaccines designed to yield CD8+ T cells, either by boosting pre-existing virus-induced responses or eliciting new primary responses, may be a rational approach to immunotherapy. DNA vaccines encoding HSV-2 tegument proteins VP11/12 (UL46), VP13/14 (UL47) and VP22 (UL49) are immunogenic for T-cell responses in mice, and encode many defined CD4+ and CD8+ epitopes. This indicates that HSV-2 tegument proteins access antigen processing pathways in both mice and humans as a class. Although we have so far been unable to demonstrate convincing prophylactic protection using these vaccine constructs (data not shown), the intended primary application for CD8+ T-cell-directed vaccines, i.e. modification of existing disease, will be difficult to evaluate without phase I clinical trials in humans.
We thank the University of Washington Virology Research Clinic, Dr Anna Wald, Dr Rhoda Morrow and Anne Cent for providing clinical viral isolates and sera. We are grateful to Misty Saracino of the Fred Hutchinson Research Center for assistance with statistical analysis. We thank Dr Gregg Milligan (University of Texas Medical Branch, Galveston, TX) for HSV-2 strains 186 and thymidine kinase-deficient strain 333 (tk−-HSV-2). We thank Jane Morrow, Denis Rusalov and Andrew Geall for their assistance with animal procedures and formulation of DNA vaccines. This work was sponsored by NIH grants R41 AI065015 and AI50132 (to D.M.K.). W.J.M. was supported through different parts of this project by NIH grant T32 AI007411 and by a Child Health Research Career Development Award through the Department of Pediatrics, Feinberg School of Medicine, and Children's Memorial Research Center at Northwestern University.
GenBank/EMBL/DDBJ accession numbers for HSV-2 gene sequences are EU029137–EU029158, EU035980, EU281624–EU281626 and AY779750–AY779754.