Diversity in CD8+ T Cell Function and Epitope Breadth Among Persons with Genital Herpes
- First Online:
- Cite this article as:
- Laing, K.J., Magaret, A.S., Mueller, D.E. et al. J Clin Immunol (2010) 30: 703. doi:10.1007/s10875-010-9441-2
- 134 Views
CD8+ T cells are known to be important in clearing herpes simplex virus (HSV) infections. However, investigating the specific antiviral mechanisms employed by HSV-2-specific T cell populations is limited by a lack of reagents such as CD8+ T cell epitopes and specific tetramers. Using a combination of intracellular cytokine staining flow cytometry and ELISpot methods, we functionally characterized peripheral HSV-2-specific CD8+ T cells from peripheral blood mononuclear cell (PBMC) that recognize 14 selected HSV-2 open-reading frames (ORFs) from 55 HSV-2 seropositive persons; within these ORFs, we subsequently identified more than 20 unique CD8+ T cell epitopes. CD8+ T cells to HSV-2 exhibited significant heterogeneity in their functional characteristics, proliferation, production of inflammatory cytokines, and potential to degranulate ex vivo. The diversity in T cell response in these ex vivo assessments offers the potential of defining immune correlates of HSV-2 reactivation in humans.
Intracellular cytokine staining
CMV, EBV and flu
Median fluorescence intensity
Genital herpes is a pandemic sexually transmitted infection that causes substantial morbidity and increases the risk of HIV acquisition two- to threefold [1, 2]. While antiviral therapy is available, treating infection has no effect on the epidemiology of the disease. As such, development of novel therapeutic approaches is needed. At most, there is limited information regarding the effector mechanisms of the adaptive immune system that control herpes simplex virus-2 (HSV-2) and the functional epitopes encoded by the virus that may be recognized by lymphocytes. Increased knowledge of adaptive immunity against HSV-2 is thus essential to the development of preventative and therapeutic vaccines for HSV-2.
Several lines of evidence suggest that T lymphocytes are essential to reducing the magnitude and duration of active herpetic disease and in controlling viral latency [3, 4]. In particular, cytotoxic (CD8+) T cells infiltrate and clear HSV-2 from genital herpes lesions and remain concentrated around the sensory neurons at the skin after the lesion has healed . These cells also accumulate around the sensory neurons during subclinical reactivation . Both cytolytic activity and interferon gamma (IFN-γ) secretion by T cells are required in rodents for complete clearance of HSV-2 from genital epithelium . However, the mechanisms of lymphocyte-based immune control of HSV-2 reactivation and clearance in humans are largely unknown.
Few surveys of adaptive host responses of persons with HSV-2 have been performed. However, the diversity of T cell responses to HSV-2 proteins or epitopes has been determined; the numbers of CD8+ T cells that produce IFN-γ or exhibit cytolytic activity when presented with an HSV-2-derived antigen have been estimated in HSV-2 seropositive individuals. Recent studies have identified several proteins encoded by HSV-2 that activate CD8+ T cells in a high proportion of HSV-2 seropositive subjects [7–9]. Due to the low level of circulating HSV-2-specific CD8+ T cells, previous approaches used for epitope identification have required enrichment of these T cells, either by positive selection or clonal expansion of CD8+ cells from peripheral blood mononuclear cell (PBMC). Since the enrichment steps may alter the behavior of the cell, their inclusion in functional assays is not ideal when investigating the immune function of HSV-2-specific T cells. Recently, new assays have been developed that allow the detection of low numbers of activated T cells within whole PBMC populations without enrichment. For example, an eight-color intracellular cytokine staining (ICS) assay, developed as a quantitative and qualitative high-throughput approach to evaluate T cell activity against HIV-1 , can detect and quantify co-expression of different cytokines in distinct cell populations, even when the cell of interest is infrequent.
The current study expands on our initial survey  to identify immunocompetent responses against HSV-2 in humans and reports specific CD8+ T cell responses to HSV-2-derived peptides in HSV-2 seropositive individuals using a modification of this validated ICS method. Detection of CD8+ T cell responses directly from PBMC was obtained without prior enrichment, and the polyfunctionality of these responses was determined. We observed monofunctional and polyfunctional cells in populations of HSV-2-specific CD8+ T cells that differed proportionately within individuals responding to multiple epitopes and also between different individuals. HSV-2-specific CD8+ T cells proliferate and mobilize CD107a and express lytic proteins such as perforin and granzyme. However, intra- and inter-individual variation was observed for these activities. Exploring this diversity in functional activity of CD8+ T cells in relation to clinical variability of mucosal HSV-2 infection between individuals is likely to provide novel insights on the immune control of HSV-2. Studies interrogating the epitope-specific T cell responses as they relate to disease expression may offer new insights into host–viral interaction and their interplay in containment of the virus.
Selection of Viral Open-Reading Frames and Peptides
Open-Reading Frames of HSV-2 Used During This Study
Number of amino acids
Modifications from HG52
Number of peptides in pool
T364R, C471F, E594K, G613S, P634L, R637Q, L638D, L644P, R672P, R673G, 587-590 delA
V38G, N156S, P172A, S177P
A204S, V385F, D560G, 587-589 insA
Positive control pools containing CEF and CMV peptides were obtained through the NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH: CEF control peptide pool (Cat # 9808) from DAIDS, NIAID; CMV pp65 peptide mix (Cat # 551969) from BD Biosciences. These were reconstituted, stored, and used according to the manufacturer's recommendations.
Demographics of Study Subjects
HSV-2 seropositive subjects (N = 55)
HSV-2 seronegative subjects (N = 18)
Collection, Preservation, and Thawing of PBMC
Blood (200 mL) was collected from all study subjects using heparinized syringes. PBMC isolation was performed less than 8 h from the time of blood draw. Whole blood was diluted 2:1 with Hank's balanced salt solution (HBSS; Invitrogen-Gibco, USA) and loaded onto 50 mL Accuspin tubes (Sigma, USA) containing 15 mL Ficoll 1.077 (Sigma). PBMC were isolated from buffy coats, following centrifugation at 800×g for 15 min and washed in 40 mL HBSS at 1,000×g for 10 min. Cell pellets were resuspended to 2 × 107 PBMCs per milliliter in fetal calf serum (FCS; Gemini Bioproducts, CA, USA) containing 10% DMSO, and slowly frozen at −80°C in an insulated container for 1–2 days, prior to transfer to liquid nitrogen (−140°C) for long-term storage. Frozen PBMC were quickly thawed in a 37°C waterbath and washed in RPMI-1640 (Invitrogen-Gibco). They were rested at 37°C/5% CO2 overnight in RPMI-1640 containing l-glutamine (2 × 106 cells per milliliter) and supplemented with 10% FCS and penicillin (100 U/mL) and streptomycin (100 μg/mL)—referred to as R10 medium—prior to use in stimulation experiments.
Intracellular Cytokine Staining Assays
We adapted the validated ICS assays developed at the FHCRC HIV Vaccine Trials Network laboratories for these studies . Rested PBMC were washed, resuspended to 10 × 106 cells per milliliter in RPMI-1640 containing human serum (10%; Gemini Bioproducts), penicillin (100 U/mL), and streptomycin (100 μg/mL), and added to 96-well plates at 1 × 106 PBMC/well. CD107a-PeCy5 antibody was incubated, at a dilution 7:100, for 10 min at room temperature, and culture volumes made up to 200 μL. Stimulations with HSV-2 peptide pools (10 μg/mL each peptide) or single peptides (2 μg/mL) were performed at 37°C/5% CO2 for 6 h in the presence of co-stimulation using antibodies against CD28 (1 μg/mL, BD Biosciences, San Jose, CA) and CD49d (2 μg/mL, BD Biosciences) adding, after the initial hour of incubation, Brefeldin A (10 μg/mL, Sigma, St. Louis, MO) and GolgiStop (0.25%; BD Biosciences). As positive controls for T cell activation, individual responses to both CEF and CMV peptide pools, and to the superantigen staphylococcal enterotoxin B (SEB; 2.5 μg/mL), were also measured. Peptide solvent (1% DMSO) or media alone (non-stimulated) served as negative controls. Cultures were transferred to 4°C overnight at the end of the incubation period.
Following 10 min incubation at room temperature with EDTA (20 mM), cells were pelleted and stained with LiveDead Violet stain (1:1,000; Invitrogen) for 20 min. The cells were subsequently washed, lysed (FACS lyse, BD Biosciences), permeabilized (FACS perm II; BD Biosciences), and stained with a cocktail of conjugated antibodies: CD3-ECD (Beckman Coulter), CD4-FITC (BD Biosciences), CD8-PerCP/Cy5.5 (BD Biosciences), IFN-γ-APC (BD Biosciences); IL-2-PE (BD Biosciences), TNF-α-Alexa700 (BD Biosciences), and CD14-APC/Cy7 (BD Biosciences), CD19-APC/Cy7 (BD Biosciences). Stained cells were resuspended and fixed in PBS containing 1% paraformaldehyde and analyzed by flow cytometry as described below.
PBMC were thawed and rested and washed as described above. Cell pellets were resuspended in 2 mL PBS containing 1.4 μg CFSE (Vybrant CFDA SE cell tracer kit, Invitrogen) for 8 min. The staining processes was stopped by addition of cold human serum and cells washed and resuspended to 2 × 106 cells per milliliter in RPMI containing 10% human serum, penicillin (100 U/mL), and streptomycin (100 μg/mL). For each stimulation, 4 × 106 CFSE-labeled cells were added per well of a 24-well plate, in the presence of 2 μg/mL HSV-2-specific peptide or CMV peptide pool. An equivalent volume of DMSO was added to negative control wells. As a positive control and to define number of proliferation events, PBMC were incubated with anti-CD3 (OKT3; 0.1 ng/mL) and anti-CD28 (2 μg/mL; BD Biosciences) antibodies. Following incubation at 37°C for 6 days, cells were harvested, washed, and dispensed at 1 × 106 cells/well of a 96-well plate. Cells were stained with LiveDead stain and CD3-ECD, CD8-PerCP/Cy5.5, and CD4-APC/Cy7 (BD Biosciences) antibodies, washed, fixed, and analyzed by flow cytometry as described below.
Detection of Perforin and Granzyme B in T Cells
PBMC were prepared and stimulated with HSV-2-specific peptides or control reagents as described as for ICS assays. Cells were pelleted and stained using LiveDead Violet stain. CD3-ECD, CD8-PerCP/Cy5.5, IFN-γ-APC, Perforin-PE (BD Biosciences), and Granzyme B-FITC (BD Biosciences) antibodies were used to stain the cells following treatment with FACS lyse (10 min) and FACS perm II (10 min). After washing and fixing steps, the stained cells were analyzed by flow cytometry as described below.
Flow Cytometric Analysis
Stained cells were analyzed using an LSRII (BD Biosciences). The flow cytometer was calibrated at the beginning of each run using SPHERO™ UltraRainbow Fluorescent Particles (Spherotech, Inc.), ensuring the positions of bead histograms, for each appropriate color, did not vary more than 1% (with a CV of less than 10) between different days. Single antibody-stained cells were used for compensation under the same conditions. All events were collected without compensation.
Flow cytometry data were analyzed and compensated using FlowJo Software (Version 8; TreeStar, Inc.). The gating strategy allowed for selection of single lymphocytes (using FSC-A versus FSC-H parameters) and removal of dead cells from all analyses. Differences in gating strategies for distinct analyses were as follows:
Cells positive for CD14 (monocytes) and CD19 (B cells) were excluded from viable cells. The remaining CD3 positive (T) cells were separated into CD4+ and CD8+ cells and gated for cytokine (IFN-γ, IL-2, or TNF-α) or CD107a expression as described previously ; CD4+ T cells were excluded from the analyses. Multiple functional marker expression events were observed using the Boolean gate feature in FlowJo.
CD3+CD8+ cells were selected from live and single cells as described above. Gate locations of proliferated cells were established using CD3/CD28 stimulated cells; proliferated (CFSElo) cells were defined as those with half or less CFSE staining as parental cells.
Granzyme and Perforin Assay
Selection of CD3+CD8+ T cells from live and single cells was performed as described above. Granzyme B+, Perforin+, or IFN-γ+ events were independently gated from CD8+ T cells. Multiple functional marker expression events were observed using Boolean gates.
Scoring and Comparison of Antigen-Specific Responses
ICS Assay—Single Cytokine Positivity
Positive or negative responses were evaluated separately for each cytokine and for each antigen within each subject. Individual cytokine responses were scored on total cell frequencies expressing a given cytokine regardless of whether or not other cytokines were co-expressed. The Fisher's exact test (one-sided) was used to determine whether the proportion of cells expressing a given cytokine was higher in the antigen-exposed wells versus the appropriate negative control. Each SEB-treated positive control was determined to be valid when it contained a significantly greater proportion of cells expressing cytokine than non-stimulated control wells. All other experimental assays (cells exposed to HSV-2 pools 1–29, CEF pools, or CMV pools) were determined to be positive when the experimental well had a significantly greater proportion of cells expressing cytokine than solvent-exposed wells (DMSO). A p value <0.02 was used as a cutoff for all positive subject responses; this was determined to provide positivity in 5% of seronegative subjects through trial and error experimentation during validation, thus allowed for a standard 5% type I (false positive) error rate. SEB well positivity was used to confirm assay validity, and all were found valid.
To evaluate immunodominance of HSV-2 ORFs, the frequency of subjects with positive responses to a given ORF were compared for each cytokine. To assess if timing of ORF expression impacted response frequencies, ORFs were classified into groups depending on their expression kinetics (Table I) and the response rate per ORF group determined for each cytokine. Statistically significant differences in response rate were tested using Poisson regression for correlated outcomes (generalized estimating equations) . Similar analyses were performed to examine length-adjusted response rates, in which the response rate was divided by the number of amino acids in the region and offset by the log of that length.
ICS Assay—Multifunctional Analysis
Activated T cells were further assessed for co-expression patterns of IFN-γ, IL-2, TNF-α, and/or CD107a to determine if HSV-2-specific T cells produce monofunctional or polyfunctional responses; this was performed using an approach described previously . Positive CD8+ T cell responses in subjects were divided according to HSV-2 ORF, and proportions of all responding cells displaying each possible functional phenotype were calculated; there were 15 possible combinations of the four functional markers, and the number of cells within each combination was observed as the percent of total (summed) responding cells for all 15 combinations. Responses detected in cells exposed to DMSO (background) were subtracted from responses in antigen-stimulated samples for each functional combination; all negative values were set to zero.
Distinct monofunctional or polyfunctional populations were also assessed for levels of each cytokine or CD107a, estimated from differences in the median fluorescence intensity (ΔMFI) relative to cells negative for a given protein. Differences in ΔMFI within the antigen-specific T cells of the tested subjects were compared using a non-parametric Friedman's (Conover pairs) test and the ΔMFI of the unique cell populations were considered significantly different if p < 0.05.
Cell division indices (CDI) were calculated by dividing the percent of CD8+ cells that were CFSElo in antigen-exposed cells by the percent of CD8+ cells that were CFSElo following exposure to DMSO; proliferation was considered positive if CDI ≥ 2 .
Granzyme and Perforin Assay
The number of HSV-2-specific cells expressing lytic granules was assessed in two ways. First, CD8+ T cells expressing IFN-γ, following antigen exposure, were selected and the proportions of IFN-γ+ cells co-expressing granzyme B and/or perforin determined. Second, granzyme B or perforin positive cells were selected and IFN-γ positive cells, relative to each of these lytic granules, were visualized and counted; differences in the numbers of IFN-γ+ T cells, in HSV-2 peptide exposed versus control (DMSO-treated) cultures in cells, expressing or lacking granzyme B or perforin, were compared using the Wilcoxon test for matched pairs and considered significantly different if p < 0.05.
Using Boolean gates, cell populations positive for different combinations of granzyme B, perforin, and/or IFN-γ were identified; levels of each granule protein or cytokine were estimated in each distinct population by observing the ΔMFI relative to cells negative for a given protein. Differences in ΔMFI within the antigen-specific T cells of the individuals tested, were compared using a non-parametric Friedman's (Conover pairs) test and the ΔMFI of the unique cell populations were considered significantly different if p < 0.05.
Epitope Identification and Confirmation
Peptide pools that were determined to give positive responses in CD3+CD8+ lymphocytes were subjected to further analysis in order to identify the responsible peptide. Subpools of no more than ten peptides were generated such that each peptide was represented once in two subpools. PBMC were thawed and rested as described above and resuspended at 2 × 106 cells per milliliter in R10 medium. R10 medium (100 μL) was added to each well of an ELISpot plate precoated with anti-IFN-γ (Human IFN-γ ELISpot PLUS kit (ALP); Mabtech, Ohio, USA). Peptide pools were added to give a final concentration 0.2 μg/mL and 2 × 105 cells added per well. Following overnight incubation of cultures at 37°C/5% CO2, plates were washed with PBS and incubated with detection antibody (1:200) in PBS, containing 1% FCS, for 2 h. Bound antibody was detected using filtered NBT substrate following the manufacturer's instructions, the plates dried overnight, and spots counted using Bioreader® 3000 ELISpot reader (BioSys) within the Immune Monitoring Laboratory at the Fred Hutchinson Cancer Research Center. Responses with ≥11 spots per well and ≥3× the number of spots in control (DMSO) wells were considered positive.
Single 15-mers predicted to give responses in CD8+ T cells, using this method, were verified by further ELISpot analysis. T cell (9-mer) epitopes embedded within a given 15-mer were predicted with the assistance of epitope prediction software at the Immune Epitope Database (beta.immuneepitope.org). A common 9 amino acid epitope (FLWEDQTLL) predicted within UL25 was synthesized at >90–95% purity (New England Peptide) and verified by serial (log) dilution using ELISpot, and sent to John Sidney at La Jolla Institute for Allergy and Immunology (La Jolla, California), to perform peptide-binding assays to MHC class I, as described previously . HLA-A*0201 tetramers binding FLWEDQTLL or a human CMV pp65 HLA-A*02 specific epitope (NVLPMVATV) were synthesized at the Immune Monitoring Laboratory at the Fred Hutchinson Cancer Research Center, conjugated with Phycoerythrin (PE), and used to recognize HSV-2-specific CD3+CD8+ cells in responding individuals by flow cytometry.
HSV-2 ORF-Specific Responses by Cytotoxic T Cells
The number and length of proteins within each ORF class varies. To exclude the effect of ORF size on response rate, we compared response rates by classification of ORFs after adjustment for number of amino acids. Responses to each cytokine were compared using a similar analysis to that described above. For IFN-γ, the amino acid-adjusted late response remained less common than the early response (RR = 0.53, 95% CI 0.28–0.99, p = 0.049), thus ORF length appears not to be responsible for the observed difference in response frequency, although this is not significant with Bonferroni correction. No differences were observed for ORF length-adjusted responses for IL-2 or TNF-α. Thus IFN-γ responses to HSV proteins appeared to be more common to the early proteins, which included UL11, UL29, UL39, and UL49.
Cytotoxic T Cells Responding to HSV-2 Peptide Pools Make IFN-γ and are Often Monofunctional
Analysis of T Cell Populations
HSV-2-specific CD8+ T cells produced multiple cytokines, but were overall, more likely to express IFN-γ than IL-2 or TNF-α following exposure to HSV-2 peptide pools. Poisson regression for correlated data confirmed that the differences in ORF-specific response rates between IFN-γ and both IL-2 (p = 0.009) and TNF-α (p = 0.010) were statistically significant. Considering all positive responses identified in CD8+ T cells of seropositive subjects, regardless of ORF, 80% were positive for IFN-γ (median value; range 0–100%), 50% were positive for IL-2 (median value; range 25–100%), and 50% were positive for TNF-α (median value; range 0–100%) (Supplementary Table S1). Relative to all seropositive subjects, the highest proportion of subjects with IFN-γ+CD8+ T cell responses was identified for both UL39 and ICP0 (27% each), followed by UL49 (22%), UL25 (20%), and UL46 (16%) (Fig. 1). UL19-specific IFN-γ+CD8+ cells were observed in only 14% subjects, while CD8+ T cells of 11% subjects produced IFN-γ in response to UL29 peptides. IFN-γ+ responses against other ORFs were identified in <10% subjects each.
Although the frequency of subjects with detectable responses for other cytokines was lower than for IFN-γ, the same general ORF immunodominance was reflected. IL-2+CD8+ T cells were most frequently identified in response to ICP0 and UL49 peptides (18% of subjects each), followed by UL46 (15%), UL19, and UL39 (12% each). Each of the other ORFs stimulated IL-2 production in <10% subjects. The most frequent TNF-α+CD8+ T cell responses were observed following stimulation with UL39 (20%) and ICP0 (14%), followed by UL49 and UL46 (11% each) similar to the pattern observed for IFN-γ, although at lower frequencies. Peptides to other ORFs each produced TNF-α+CD8+ T cells in <10% subjects. Only two ORFs, with responses to HSV-2 peptides, gave no detectable IFN-γ+CD8+ T cells; four responses to ICP27 were IL-2+ only and one response each to US6 was singly positive for IL-2 and TNF-α. All responses to both ORFs comprise ≤0.05% CD8+ T cells above background levels. No significant differences in the type of cytokine produced by HSV-2-specific CD8+ T cells were observed when comparing responses between different ORFs.
Analysis of Single Cells
Next, we assessed the frequency of responding cells with distinct phenotypes (15 were possible) based upon different combinations of IFN-γ, IL-2, TNF-α, and CD107a expression. Most HSV-2 ORFs induced polyfunctional cells that were largely IFN-γ+; polyfunctional CD8+ T cells predominantly co-expressed IFN-γ, IL-2, and TNF-α (highest mean response 22% of total; UL49), or IFN-γ and TNF-α (highest mean response 29% of total; UL35). Cytokine-secreting cells did degranulate; there were CD107a+ polyfunctional cells. The co-expression of IL-2 and CD107a, particularly in the absence of IFN-γ expression, was infrequent (mean 0%; range 0–8%); typically, of total responding CD8+ T cells, 1–3% of co-expressed all four markers, 1–4% co-expressed IFN-γ, IL-2 and CD107a, 1% co-expressed IL-2, TNF-α, and CD107a, and 1% co-expressed IL-2 and CD107a (UL27 only).
Monofunctional T cells were more likely IFN-γ+ (highest mean response 49%; UL11) or CD107a+ (highest mean response 63%; UL19) than IL-2+ (highest mean response 11%; ICP0) or TNF-α+ (highest mean response 32%; US6). Their proportions were highest (92–98% of responsive CD8+ T cells) during reactions to ICP27 (IL-2 only), UL11, and US6 (IFN-γ only) and lowest (35–53% of responsive CD8+ T cells) during reactions to UL25, UL35, UL46, and UL49. ORFs inducing the highest proportions of monofunctional cells also induced the lowest frequency of cytokine-positive CD8+ T cells, suggesting that monofunctional phenotype and low frequency of activated cells are related (Fig. 1). Only 2 (UL35 and UL46) of the 14 ORFs studied produced responses in which >50% of the cells were polyfunctional. In contrast, <50% of the CD8+ T cells activated by CMV or CEF control peptide pools were monofunctional.
In summary, most CD8+ T cells to HSV-2 were monofunctional, with IFN-γ being the most commonly expressed cytokine and many HSV-2-specific T cells able to degranulate. No differences in the functional profiles were noted between the prevalent ORFs of HSV-2.
Identification of Specific Epitopes Within HSV-2 ORFs
Peptides that Induce Responses in CD8+ T Cells of HSV-2 Seropositive Subjects
Location in ORF
HLA-A and -B typesb
Epitope IC50 (μM)d
A01, A03, B08, B63
A02, A11, B08, B15
A02, A34, B35, B62
A01, A02, B08, B44
A02, A24, B35, B44
A02, A03, B07, B44
A02, A32, B14, B44
A03, B07, B47
A01, A03, B07, B08
A03, A68, B35, B53
A02, A30, B07, B35
A01, A31, B07, B60
A02, A74, B07, B14
A01, A03, B07, B08
A24, A31, B07, B62
A02, A68, B07, B14
A01, A03, B07, B08
A24, A30, B18, B40
A01, A03, B07, B08
A03, B07, B47
A02, A03, B07, B44
A33, A74, B18, B35
A02, A24, B35, B44
A03, A11, B39, B51
B39 allele not in IEDB
A29, A68, B71, B37
250. B37/71 not in IEDB
A02, A03, B07, B44
A24, A31, B07, B62
A02, B50, B62
A03, A11, B07, B51
A02, A03, B49, B62
A01, A03, B08, B63
A01, A03, B07, B08
A01, A31, B07, B60
A24, A34, B14, B50
Alleles not in IEDB
Of particular interest, we identified a previously uncharacterized peptide embedded in the HSV-2 capsid-associated protein UL25 that is antigenic in multiple HLA-A02 positive persons. This peptide (UL25369–383) elicited T cell responses in at least 5 (17%) of the 29 (53%) subjects that possessed an HLA-A02 allele. A listing of the HLA alleles in our population is shown in Supplementary Table S2. Analysis of the peptide sequence by epitope prediction programs identified a high putative affinity HLA-A02 binding epitope (FLWEDQTLL) embedded in the middle of this 15-mer. Likely, proteasomal processing of this epitope was also predicted using NETCHOP (not shown).
Qualitative Analysis of CD8+ T Cells to Single Epitopes
In other viral infections, polyfunctional T cells have higher functional capacity than monofunctional T cells [17–19]. To determine whether this is the case in HSV-2-specific T cells, we assessed the relative levels of each functional marker in responding cells expressing different cytokine and degranulation markers. To determine per-cell differences in cytokine or degranulation levels in polyfunctional and monofunctional HSV-2-specific T cells, the relative intensities of cytokine and degranulation markers were compared between cells expressing one or more markers. IFN-γ and TNF-α levels were higher in polyfunctional cells than in monofunctional cells (Fig. 5b). Significant increases in median fluorescence intensity were observed for IFN-γ (8-, 11-, and 13-fold) in polyfunctional HSV-2-specific cells with two, three, and four detectable markers, confirming that polyfunctional cells are making large amounts of this cytokine; this is consistent with increases in IFN-γ observed following SEB stimulation. Increases in TNF-α levels were also observed (1.4 and 1.6-fold) in polyfunctional cells with three and four detectable markers, relative to monofunctional cells following exposure to HSV-2 peptides, but not to SEB. IL-2 levels were increased in cells with two or three functions (1.5- and 1.6-fold, respectively), but not in those with four functions, relative to monofunctional cells; cells that mobilized CD107a did not have a higher ΔMFI for IL-2 (not shown), suggesting that IL-2 production is less likely to occur in cells that degranulate. The quantity of CD107a mobilization was not significantly different between monofunctional or polyfunctional HSV-2-specific T cells, although increases were observed for SEB-challenged cells.
Single peptides were able to recall polyfunctional cells regardless of their specific ORF or subject variability. However, while cytokine and degranulation marker expression patterns for SEB-stimulated cells between individuals were similar, responses to HSV-2-specific peptides and ORFs varied by individual. Consistent with studies using peptide pools, IFN-γ was the most abundant cytokine produced by HSV-2-specific CD8+ T cells; the majority of HSV-2-specific CD8+ T cells were positive for IFN-γ, regardless of whether these cells were mono- or polyfunctional, reflecting rises in IFN-γ production with higher functional complexity, as well as the high frequency of IFN-γ+ monofunctional cells (Fig. 5c).
The most variable functional markers, in our study, were CD107a and IL-2. Although the polyfunctional cells in two individuals (Pt 03 and Pt 07) were still IFN-γ positive, these subjects had a larger percentage of monofunctional cells that, respectively, stained for IL-2 (51% of monofunctional T cells) or CD107a (58–61% of monofunctional T cells). Higher levels of CD107a+ CD8+ T cells, in general, were detected in Pt 07 than in the other subjects. For all subjects, very few HSV-2-specific T cells were observed that both produced IL-2 and exposed CD107a, although Pt 07 showed the highest cellular frequency (∼30% of trifunctional T cells) with detectable expression of both IL-2 and CD107a; these were observed following stimulation with peptide ICP0209–755. Of particular interest, all three individuals tested with UL49-derived peptides had an unusual pattern in their “triple positive” T cells relative to responses to the other ORFs; most (∼80–100%) CD8+ T cells, responding to UL49 peptides and positive for three activation markers, produced IFN-γ, IL-2, and TNF-α but were negative for CD107a staining. In fact, for these persons, only a small percentage of T cells with recalled responses were CD107a positive, suggesting the UL49-specific T cells of these individuals degranulate less than cells that recognize the other ORFs.
Thus, although HSV-2-specific cells predominantly expressed IFN-γ in all subjects, there were qualitative differences in the proportions of cells expressing IL-2 and CD107a, potentially reflecting differential proliferation or cytotoxic capability of epitope-specific cells in distinct subjects. This is interpreted with a caveat; since T cells were activated with 15-mer peptides in these studies rather than exact epitopes (8–10-mer), it is plausible that suboptimal T cell activation generated some of these differences.
Epitope-Specific T Cell Responses Vary Within an Individual
Variations in functional phenotypes were observed between T cells activated by different peptides (Fig. 6c). Peptide UL4945–59 produced the largest proportion (46%) of CD8+ T cells that were positive for three functional markers; of these, >80% produced IFN-γ, IL-2, and TNF-α, but did not show detectable degranulation (CD107a negative). Cells specific for this peptide also showed the fewest monofunctional cells (11%) and gave bifunctional cells that were predominantly IFN-γ+TNF-α+. Although T cell responses to peptides UL4993–108 and UL49129–143 showed similarities in their polyfunctional profiles to UL4945–59 specific T cells, a larger percentage of IFN-γ+TNF-α+CD107a+IL-2− cells were observed for these stimuli, and greater variation in cytokine combinations were observed for bifunctional cells. Both peptides generated monofunctional responses in a significant proportion of CD8+ T cells; 42% and 54% of responding cells were monofunctional following exposure to UL4993–108 and UL49129–143, respectively. Polyfunctional T cells for these latter peptides produced mainly IFN-γ, while contrasting UL4945—59 monofunctional cells were frequently positive for CD107a. The relative levels of IFN-γ also appeared higher for UL4945–59 than the other peptides (Fig. 6d); this was not observed for other functional markers.
Thus, both immunodominance and functional differences between CD8+ T cells to the same ORF within one subject were apparent. These cannot be related to genetic differences or HLA restriction, and are presumably influenced in some way by the biochemical properties of different peptide epitopes and/or the T cell receptors that bind them.
Detection of Granule Proteins in HSV-2-Specific CD8+ T Cells
During ICS analyses, we used mobilization of CD107a as an indicator of degranulation following T cell exposure to antigen; this protein is exposed on the inner membrane of a granule as it opens to release its contents, but provides no precise information on the stored constituents.
Likewise, increases in IFN-γ+ T cells were detected in GrzB+, GrzB−, and Perf− T cell populations, but not Perf+ cells, following activation with HSV-2 peptides (Fig. 7c, d). Comparing averaged HSV-2 responses within subjects, a 0.14% (median; range 0.04–0.28%) increase in IFN–γ+GrzB+ and 0.18% (median; range 0.11–0.41%) increase in IFN–γ+GrzB− cells above baseline (DMSO control) values was observed. The frequency of IFN-γ+ T cells increased by 0.17% (median; range 0.09-0.24%) relative to controls in Perf negative cells from cultures exposed to HSV-2 peptides. Although, this increase was not observed in Perf+ CD8+ T cells (median increase 0%; range 0–0.45%), the analysis was limited by low numbers of cells with this phenotype.
HSV-2-specific T cells lacking lytic proteins may have released their granule contents in response to peptide activation prior to intracellular staining. To assess this, we investigated whether GrzB or Perf levels decreased in IFN-γ+ (activated) T cells. Mean fluorescent intensities (ΔMFI; see “Methods”) were used for estimating the relative levels of GrzB or Perf in cells co-expressing either granule protein and/or IFN-γ (Fig. 7e, f). ΔMFI for GrzB were approximately twofold higher in HSV-2 peptide exposed CD8+ T cells that co-expressed Perf (median ΔMFI = 782; p < 0.001) or Perf and IFN-γ (median ΔMFI = 707; p < 0.001) than in cells that expressed IFN-γ but lacked Perf (median ΔMFI = 398), similar to the relative ΔMFI of GrzB in SEB-challenged T cells. The GrzB+IFN-γ+ cells that lack Perf have likely released Perf in response to T cell activation by their specific antigen; reduced GrzB levels in these cells is consistent with co-release of GrzB during degranulation. Similarly, Perf levels were higher in CD8+ T cells that co-expressed GrzB (median ΔMFI = 600; p < 0.001) or GrzB and IFN-γ (median ΔMFI = 820; p < 0.001) than in GrzB−IFN-γ+ CD8+ T cells (median ΔMFI = 384), consistent with release of Perf and GrzB following activation of the antigen-specific T cells.
The same HSV-2 peptides were able to increase CD107a mobilization, as well as release cytokines, in all but one of the subjects tested (Fig. 7g), relating evidence that HSV-2-specific T cells can degranulate with observed decreases in lytic proteins.
Proliferation of HSV-2-Specific CD8+ T Cells
Our studies provide several new insights about the CD8+ T cell response to HSV-2 in immunocompetent persons with genital herpes. This investigation is the first to use a newly optimized high-throughput flow cytometry technique to detect HSV-2-specific T cell responses. Given the low level of HSV-2-specific CD8+ T cells relative to T cells that respond to other viruses, our ability to detect these cells and monitor differential cytokine production represents a significant methodological advance in HSV immunology. Using this novel method, we confirm that HSV-2-specific CD8+ T cells, although varied between individuals, are often directed at the immediate-early, capsid and tegument proteins and can be detected without prior enrichment from PBMC. Interestingly, at the single cell level, many of the CD8+ T cells appear to be of the effector phenotype and secrete IFN-γ alone. Polyfunctional responses correlate with a high magnitude of T cell response and are often found at higher levels when induced with single peptides than in assays induced with peptide pools. The latter phenomenon likely reflects methodological issues, related to higher levels of DMSO added to cultured cells when using peptide pools (0.5% v/v) than single peptides (0.005%), since DMSO has inhibitory effects on T cells . HSV-2-specific CD8+ T cells store lytic proteins and are able to degranulate and proliferate but do show considerable inter- and intra-individual variation. Even within an individual and with responses directed at a single ORF, there is considerable variation in the magnitude and function of T cell responses.
Proteins encoded by immediate-early or early genes were the predominant targets for HSV-2-specific CD8+ T cells and induced IFN-γ, IL-2, and TNF-α production. IFN-γ responses were identified for a higher proportion of subjects, consistent with HSV-2-specific T cells having a predominantly effector phenotype with lower levels of IL-2 and TNF-α production. The highest frequency of CD8+ T cell responses were detected for the immediate-early protein, ICP0, whose gene is one of the first expressed following HSV reactivation  and UL39 (ICP10), which is transcribed with early kinetics . In contrast, late genes, such as the glycoproteins gB, gD, and gJ, were rarely recognized by CD8+ T cells; IFN-γ producing CD8+ T cells specific for these ORFs were found in significantly fewer persons than those recognizing early ORFs. Although not thought to occur during HSV infections , a similar distinction was previously observed for virus-specific T cell responses, with up to threefold more responses observed against immediate-early proteins than late proteins from CMV [24, 25]. This may occur because viral gene products synthesized with earlier kinetics become processed and presented by the host cell presentation machinery, to control viral activity, before late gene products are produced. Alternatively, the virus may employ a mechanism such as inhibition of TAP activity by HSV ICP47 [26, 27], encoded by immediate-early gene US12, to reduce the likelihood of peptides from proteins synthesized later during the viral lytic phase being presented to T cells of the host. As we were primarily concerned with investigating the functional characteristics of HSV-2-specific T cells, the present investigation concentrated on screening prevalent HSV-2 ORF responses in a generalizable population of HSV-2 infected patients. Study subjects were, hence, distinct to our original studies, were newly enrolled and assessed, and we were required to identify HSV-2-specific responses in our cohort de novo. Whether investigation of less prevalent ORFs would provide different insights is of course unknown.
For many HSV-2-specific CD8+ T cell responses, we succeeded at identifying the single peptide antigen and could predict the probable HLA-restricted epitopes embedded within. Epitopes derived from ORFs expressed after immediate-early stages lacked diversity and variety in HLA restriction, contrasting the high diversity of identified epitopes, and associated HLA restrictions were observed for ICP0. This is consistent with the hypothesis that ICP47 inhibits TAP and, therefore, TAP-dependent HLAs during later stages of viral activity. Although we should bear in mind that our methods may favor activation of cells restricted by HLAs that bind exogenous peptide from endosomal compartments rather than T cells that are specific for TAP-dependent epitopes, TAP-independent epitope presentation pathways or proteins made early during the lytic phase of viral activity should be considered during the design of therapeutic or preventative interventions that invoke T cell responses against HSV. TAP-independent presentation of HSV-derived peptides in mice has been suggested previously . Further confirmation of TAP-dependency on HSV-2 epitope processing and loading in humans is required.
While multiple epitopes were evident across several HSV-2 ORFs, common epitopes were identified and constituted the majority of responses for UL49—the most abundant tegument protein, VP22, that is expressed early —and UL25—an abundant C-capsid-associated protein with leaky-late expression that assists in maturation, packaging, and retention of viral DNA, and increases capsid stability [30–34]. A high frequency of subjects responded to UL49, confirming the findings of previous investigations [7, 35]. In UL49, a previously confirmed common and potent HLA-B07 restricted epitope (RPRGEVRFL) [35, 36] was responsible for CD8+ T cell activation in most responding subjects. However, we also identified four novel UL49 epitopes; three were predicted to be restricted by HLA-B07 with one (UL49249–263) restricted by HLA-B08. The common epitope (FLWEDQTLL) found in UL25 is the first common CD8+ T cell epitope identified for HSV-2 that is restricted by HLA-A*0201; the same epitope may also be presented by HLA-A*0202, *0203, and *0206. Accordingly, all subjects identified with CD8+ T cell responses to UL25 possessed an HLA-A2 allele. FLWEDQTLL is conserved in UL25 of HSV-1 and is thus potentially type common; responses were detected in HSV-1−/HSV-2+ and HSV-1+/HSV-2+ individuals. Since the epitopes identified in UL49 and UL25 were recognized by T cells of multiple individuals and utilize HLA-types that are common in the population, development of T cell-inducing HSV vaccines with a wide population coverage may benefit from inclusion of these targets .
Screening for HSV-2-specific CD8+ T cells facilitated investigating our major interest—to determine if functional effector CD8+ T cells could be identified in subjects with chronic HSV-2 infection. Chronic viral infections often cause T cell exhaustion that culminates in loss of cytotoxic function, proliferation, and cytokine production , and we predicted this would be the case for HSV-2 infection. However, HSV-2-specific CD8+ T cells did not appear dysfunctional. Polyfunctionality within HSV-2 epitope-specific CD8+ T cell populations was confirmed for the first time in our study. Since polyfunctionality is associated with better viral control [17–19], HSV-2-specific T cells appear to have retained some high level effector function to control HSV-2 reactivations even after long-term infection. CD8+ T cell populations that were activated by HSV-2 peptides comprised a mixture of monofunctional and polyfunctional cells with differing ability to produce IFN-γ, IL-2, and TNF-α or to degranulate. IFN-γ levels were higher in polyfunctional cells than in monofunctional T cells, similar to vaccinia virus-specific T cells identified in a previous study . Also consistent with other studies, IL-2 was typically lower in cells that mobilize CD107a, suggesting that cells that degranulate have less potential to induce proliferation, or vice versa. IL-2 was also the least abundant cytokine—in terms of detectable cellular frequency—in our assay; higher frequencies of HSV-2-specific CD8+ T cells made TNF-α and IFN-γ. CD8+ T cells are known to produce less IL-2 than CD4+ T cells , thus observing that lower frequency of IL-2+ than IFN-γ+ CD8+ T cells is typical and not an indicator of dysfunction [40, 41], thus HSV-2-specific T cells do not appear to be visibly impaired.
In addition to producing multiple cytokines, a high frequency of CD8+ T cells mobilized CD107a, an indicator of degranulation, and had decreased granzyme B and perforin levels following exposure to HSV-2-derived peptides; this implies that HSV-2-specific CD8+ T cells can kill infected cells by releasing cytotoxic mediators from granules and supports the hypothesis that these T cells can control virally infected cells by means of degranulation. In a recent study in mice, granzyme B has been shown to control HSV reactivation by directly damaging ICP4, an essential HSV protein with a role in maintaining viral latency . It would be of use to know if viral control of HSV-2 by granzyme B also occurs in humans.
We also confirmed that HSV-2-specific CD8+ T cells retain the ability to proliferate, although there are clear differences in the magnitude and proliferative ability of distinct T cell clonotypes that appear related to the level of activation achieved by those cells and opposed to a preponderance to degranulate. Our observation was verified during a recent study that compared proliferation and cytotoxicity of CD8+ T cells specific for several different viruses; markers associated with proliferation were identified on distinct T cell populations from markers associated with degranulation and cytotoxicity, whereby central memory T cells had high proliferative capacity but low cytotoxicity, and both effector memory and terminal effector memory subsets displayed high cytotoxic activity but low potential to proliferate .
In summary, our study demonstrates the use of ICS for confirming polyfunctionality of HSV-2-specific CD8+ T cells, and for identifying numerous novel epitopes responsible for activating these T cells from selected HSV-2 ORFs. Ex vivo analysis of the HSV-2-specific T cells revealed a variety of effector functions, raising the question whether CD8+ T cell function is associated with the frequency and severity of HSV-2 reactivation in the genital mucosa. The relationship between disease severity and the diversity of cytokine proteins, kinetics of proliferation or degranulation are questions raised by our study, such that a future investigation into the ability of cells with different functional profiles to control viral replication is on our agenda. The epitopes discovered during this investigation and the flow cytometric techniques we have optimized will be highly beneficial tools that can be utilized for answering such questions.
The authors gratefully thank Selin Caka and Silvia Altmann for assistance with recruitment and sampling of study subjects, Mike Remington for assistance with sample collection, Stacy Selke for assisting with subject selection, and Helen Horton for assistance with ICS panel design. Additional thanks go to John Sidney for performing peptide-binding assays and to Rachel Tompa for technical editing of the manuscript. This work was supported by NIH P01-AI030731, NIH P30 AI207757, NIH AI 50132, and NIH R37-AI042528.