HSV type 1 (HSV-1) is a prevalent human pathogen that infects >3.72 billion individuals worldwide and can cause potentially blinding recurrent corneal herpetic disease. HSV-1 establishes latency within sensory neurons of trigeminal ganglia (TG), and TG-resident CD8+ T cells play a critical role in preventing its reactivation. The repertoire, phenotype, and function of protective CD8+ T cells are unknown. Bolstering the apparent feeble numbers of CD8+ T cells in TG remains a challenge for immunotherapeutic strategies. In this study, a comprehensive panel of 467 HLA-A*0201–restricted CD8+ T cell epitopes was predicted from the entire HSV-1 genome. CD8+ T cell responses to these genome-wide epitopes were compared in HSV-1–seropositive symptomatic individuals (with a history of numerous episodes of recurrent herpetic disease) and asymptomatic (ASYMP) individuals (who are infected but never experienced any recurrent herpetic disease). Frequent polyfunctional HSV-specific IFN-γ+CD107a/b+CD44highCD62LlowCD8+ effector memory T cells were detected in ASYMP individuals and were primarily directed against three “ASYMP” epitopes. In contrast, symptomatic individuals have more monofunctional CD44highCD62LhighCD8+ central memory T cells. Furthermore, therapeutic immunization with an innovative prime/pull vaccine, based on priming with multiple ASYMP epitopes (prime) and neurotropic TG delivery of the T cell–attracting chemokine CXCL10 (pull), boosted the number and function of CD44highCD62LlowCD8+ effector memory T cells and CD103highCD8+ tissue-resident T cells in TG of latently infected HLA-A*0201–transgenic mice and reduced recurrent ocular herpes following UV-B–induced reactivation. These findings have profound implications in the development of T cell–based immunotherapeutic strategies to treat blinding recurrent herpes infection and disease.
A staggering 3.72 billion individuals worldwide (i.e., >50% of the world population) are currently infected with HSV type 1 (HSV-1), a neurotropic member of the α herpesvirus family that develops a “steady-state” latent infection in the sensory neurons of trigeminal ganglia (TG) (1–3). The majority of infected individuals remain asymptomatic (ASYMP) (4–10). They do not experience any recurrent herpetic disease because they may have developed a “natural” protective immunity that helps to reduce/suppress virus reactivation, symptomatic (SYMP) shedding, and/or recurrent herpes disease (10–12). In contrast, a much smaller population of SYMP individuals may have lost such natural immunity, causing them to develop frequent, recurrent herpetic disease (i.e., two to five episodes of recurrent disease per year) triggered by psychological, chemical, or physical stress (13–15). Although standard antiviral drug treatments (e.g., acyclovir) can reduce recurrent SYMP disease to some extent, they do not clear the infection (16–20). An alternative immunotherapeutic herpes simplex vaccine is not available (reviewed in Ref. 12). The efforts to develop a subunit immunotherapeutic vaccine over the past 20 y have only focused on 2 of the 84+ proteins encoded by the HSV genome: glycoproteins gB and gD (21, 22). Clinical vaccine trials based on gB and/or gD have provided only moderate and transient protection, despite inducing neutralizing Abs and HSV-specific CD4+ T cells (7, 12, 21, 22). This suggests that an effective immunotherapeutic herpes vaccine must include HSV Ags other than gB and gD and would likely require induction of strong antiviral CD8+ T cell responses, in addition to neutralizing Abs and CD4+ T cell responses.
Direct evidence in animal models (6, 18, 23) and indirect observations in humans (24–26) indicate that, after the clearance of primary HSV-1 infection, a pool of memory CD8+ T cells develops and recirculates between secondary lymphoid organs and peripheral tissues, whereas a different pool of memory CD8+ T cells develops and resides within peripheral tissues (1, 5, 6, 10, 15, 17, 27–32). TG-resident CD8+ T cells play a critical role in preventing HSV-1 reactivation from latently infected sensory neurons of TG and subsequent recurrent corneal herpetic disease (1, 10). Information on the repertoire, phenotype, and function of protective CD8+ T cells that reside in latently infected TG is essential to design immunotherapeutic strategies. We hypothesize that a sufficient number of functional CD8+ T cells in TG of ASYMP individuals has likely resulted in the suppression (or abortion) of attempts of virus reactivation from latency and, hence, elimination of SYMP shedding and reduction of recurrent disease (33). In contrast, the smaller numbers of functional CD8+ T cells in TG of SYMP individuals may not be sufficient to eliminate virus reactivation and, thereby, would result in repetitive SYMP shedding and recurrent herpetic disease. Therefore, an immunotherapeutic vaccine strategy that increases the number of functional HSV-specific CD8+ T cells in TG would logically prevent or reduce virus shedding in tears and should also reduce recurrent ocular herpetic disease. Despite extensive research efforts over two decades, this goal remains unattained, mainly because during latency, the TG appear to be an immunologically restrictive compartment that is not “open” to homing and sequestering of CD8+ T cells that would migrate from the circulation (34). Although CD8+ T cells appeared to infiltrate the TG during acute infection (likely in response to high levels of T cell–attracting chemokines CXCL9, CXCL10, and CCL5), a decrease in the levels of these chemokines during latency may prevent migration of sufficient numbers of antiviral CD8+ T cells from the blood into latently infected TG. We further hypothesize that delivering the T cell–attracting chemokines CXCL9, CXCL10, and CCL5 locally in latently infected TG would increase the size of local antiviral CD8+ T cell infiltrates, which, in turn, would help to reduce/suppress virus reactivation, SYMP shedding, and/or recurrences of herpes disease.
In the current study, we used a genome-wide analysis strategy and predicted a panel of 467 HSV-1 potential HLA-A*0201–restricted epitopes. Frequent polyfunctional HSV-specific IFN-γ+CD107a/b+CD44highCD62LlowCD8+ effector memory T (TEM) cells were detected in PBMCs of ASYMP individuals and were directed primarily against three epitopes: UL9196–204, UL25572–580, and UL44400–408. In contrast, HSV-specific CD8+ T cells in PBMCs of SYMP individuals were mainly of the central memory phenotype (CD44highCD62LhighCD8+ central memory T [TCM] cells) and were directed against other nonoverlapping epitopes. Furthermore, we demonstrate that bolstering the number of functional HSV-specific CD8+ TEM cells and CD103highCD8+ tissue-resident memory T (TRM) cells in latently infected TG of “humanized” HLA-transgenic (Tg) mice through HSV-1 human epitopes/CXCL10–based prime/pull therapeutic vaccine protected against virus shedding and recurrent herpetic disease. These findings strongly suggest the feasibility of developing an effective prime/pull therapeutic herpes vaccine based on enhancing the number and function of existing HSV-specific memory CD8+ TEM and TRM cells within latently infected sensory ganglia.
Materials and Methods
Human study population
During the last 13 y (i.e., January 2003–October 2016), we have screened 803 individuals for HSV-1 and HSV type 2 (HSV-2) seropositivity (Table I). Five hundred and fifty-four individuals were white and 249 were nonwhite (African, Asian, Hispanic, and other), 410 were female, and 393 were male. Among this sample, a cohort of 293 immunocompetent individuals, ranging from 21 to 67 y old (median 39 y) and seropositive for HSV-1 and seronegative for HSV-2, were enrolled in the current study. All patients were negative for HIV and hepatitis B virus, with no history of immunodeficiency. A total of 705 patients was HSV-1, HSV-2, or HSV-1/HSV-2 seropositive, among which 642 patients were healthy and defined as ASYMP. ASYMP individuals had never had any herpes disease (ocular, genital, or dermal) based on self-reporting and clinical examination. Even a single episode of any herpetic disease would exclude the individual from this group. The remaining 63 patients were defined as HSV-seropositive SYMP individuals who suffered from frequent and severe recurrent genital, ocular, and/or orofacial lesions. Two of these SYMP patients had clinically well-documented repetitive herpes stromal keratitis (HSK), and another had 20 episodes over 20 y that required several corneal transplantations. Signs of recurrent disease in SYMP patients were defined as herpetic lid lesions, herpetic conjunctivitis, dendritic or geographic keratitis, stromal keratitis, and iritis consistent with HSK, with one or more episodes per year for the past 2 y. However, at the time of blood collection, SYMP patients had no recurrent disease (other than corneal scarring) and had experienced no recurrence during the past 30 d. They had no ocular disease other than HSK, had no history of recurrent genital herpes, and were HSV-1 seropositive and HSV-2 seronegative. Because the spectrum of recurrent ocular herpetic disease is wide, our emphasis is mainly on the number of recurrent episodes and not on the severity of the recurrent disease. No attempt was made to assign specific T cell epitopes to specific severity of recurrent lesions. Patients were also excluded if they had an active ocular (or elsewhere) herpetic lesion or had one in the past 30 d, were seropositive for HSV-2, were pregnant or breastfeeding, or were on acyclovir or other related antiviral drugs or any immunosuppressive drugs at the time of blood draw. SYMP and ASYMP groups were matched for age, gender, serological status, and race. We also collected and tested blood samples from 81 healthy control individuals who were seronegative for HSV-1 and HSV-2 and had no history of ocular herpes, genital lesions, or orofacial herpes disease. The number and characteristics of SYMP, ASYMP, and seronegative individuals enrolled in the current study is shown in Table II. All clinical investigations in this study have been conducted according to Declaration of Helsinki principles. All subjects were enrolled at the University of California Irvine under Institutional Review Board–approved protocols (IRB#2003-3111 and IRB#2009-6963). Written informed consent was received from all participants prior to inclusion in the study.
In silico prediction of potential HLA-A*0201–restricted CD8+ T cell epitopes from HSV-1 genome
The HSV-1 open reading frames (ORFs) used in this study were from strain 17 (National Center for Biotechnology Information accession number NC001806). We first searched the deduced amino acid sequence encoded by the entire HSV-1 genome (strain 17) for potential HLA-A*0201–binding regions. Candidate HLA-A*0201–restricted epitopes were identified from all 84+ HSV-1 ORFs using BIMAS software (Bioinformatics and Molecular Analysis Section, National Institutes of Health; http://bimas.dcrt.nih.gov/molbio/hla_bind/) and the SYFPEITHI algorithm (http://www.syfpeithi.de/) (6). Potential cleavage sites for human proteasome were identified using NetChop 3.0 (http://www.cbs.dtu.dk/services/NetChop/) and MHC Pathway (http://www.mhc-pathway.net) for screening (6). These algorithms identified a total of 467 potential regions (i.e., CD8+ T cell epitopes) with high predicted affinity to the HLA-A*0201 molecule (Supplemental Table I). All epitopes are restricted to HLA-A*0201, a haplotype chosen for this study because it is present in >50% of the world human population, irrespective of gender and ethnicity (5, 18).
Based on the bioinformatics analysis, 467 putative HLA-A*0201 binding peptides (9 mer long) from 84+ ORFs with high estimated half-times of dissociation (Supplemental Table I) were synthesized by Magenex (San Diego, CA) on a 9050 Peptide Synthesizer Instrument using solid-phase peptide synthesis and standard 9-fluorenylmethoxycarbonyl technology (PE Applied Biosystems, Foster City, CA). The purity of peptides was between 75 and 96%, as determined by reversed-phase HPLC (Vydac C18) and mass spectroscopy (Voyager MALDI-TOF System). Stock solutions were made at 1 mg/ml in 10% DMSO in PBS. All peptides were aliquoted and stored at −20°C until assayed. Synthetic peptides corresponding to the 467 potential CD8+ T cell epitopes, which belong to a total of 84+ ORFs encoded by the HSV-1 genome, were synthesized and divided into 22 groups of peptides (grp.1 to grp.22) (Supplemental Table I). Each group of peptides contained an average of 20 peptides (range 12–27 peptides) derived from three to five ORF proteins.
Binding with soluble HLA-A*0201 molecules
Quantitative assays to measure binding of each of 467 peptides to soluble HLA-A*0201 molecules are based on inhibition of binding of a radiolabeled standard peptide, as recently described (35). Radiolabeled peptide (1–10 nM) was coincubated with 1 μM–1 nM purified MHC and 1–3 μM human β2-microglobulin. After 2 d, binding of radiolabeled peptide to MHC class I molecules was determined by capturing MHC–peptide complexes on Greiner LUMITRAC 600 microplates coated with W6/32 Ab and measuring bond cpm using a TopCount microscintillation counter. Concentration of peptide yielding 50% inhibition of binding of radiolabeled probe peptide (IC50) was calculated (Supplemental Table I).
Stabilization of HLA-A*0201 on class I HLA–transfected B × T hybrid cell lines (T2 cell line)
To determine whether synthetic peptides could stabilize HLA-A*0201 molecule expression on the T2 cell surface, peptide-inducing HLA-A*0201 upregulation on T2 cells was examined according to a previously described protocol (6). T2 cells (3 × 105 per well) were incubated with different concentrations of individual peptides in 48-well plates for 18 h at 26°C. Cells were then incubated at 37°C for 3 h in the presence of human β-2 microglobulin (1 μg/ml) and BD GolgiStop (5 μg/ml) to block cell surface expression of newly synthesized HLA-A*0201 molecules. The cells were washed with FACS buffer (1% BSA and 0.1% sodium azide in 1× PBS) and stained with anti-HLA-A2.1–specific mAb BB7.2 (BD Pharmingen) at 4°C for 30 min. After incubation, the cells were washed with FACS buffer, fixed with 2% PFA in 1× PBS, and analyzed by flow cytometry using a BD LSR II (Becton Dickinson, Mountain View, CA). The acquired data, including mean fluorescence intensity (MFI), were analyzed using FlowJo software version 9.9.4 (TreeStar, Ashland, OR). Percentage MFI increase was calculated as follows: (MFI with the given peptide − MFI without peptide)/(MFI without peptide) × 100. Each experiment was performed three times, and mean ± SD was calculated.
HLA-A2 subtyping was performed using a commercial Sequence-Specific Primer kit (SSPR1-A2), following the manufacturer’s instructions (One Lambda, Canoga Park, CA). Genomic DNA extracted from PBMCs of HSV-seropositive SYMP and ASYMP individuals was analyzed using a TECAN DNA workstation from a 96-well plate with 2 μl volume per well, as we described previously (29). The yield and purity of each DNA sample were tested using a UV spectrophotometer. The integrity of DNA samples was ascertained by electrophoresis on agarose gel. Each DNA sample was then subjected to multiple small-volume PCR reactions using primers specific to areas of the genome surrounding the single-point mutations associated with each allele. Only primers that matched the specific sequence of a particular allele would amplify a product. The PCR products were subsequently electrophoresed on a 2.5% agarose gel with ethidium bromide, and the pattern of amplicon generation was analyzed using HLA Fusion Software (One Lambda). Additionally, the HLA-A2 status was confirmed by staining PBMCs with anti–HLA-A2 mAb (BB7.2; BD Pharmingen) at 4°C for 30 min. The cells were washed, acquired on a BD LSR II, and analyzed using FlowJo software version 9.9.4 (TreeStar).
Individuals (negative for HIV and hepatitis B virus and with or without any HSV infection history) were recruited at the University of California Irvine’s Institute for Clinical and Translational Science. Between 40 and 100 ml of blood was drawn into yellow-top Vacutainer Tubes (Becton Dickinson). The serum was isolated and stored at −80°C for detection of anti–HSV-1 and HSV-2 Abs, as we described previously (36). PBMCs were isolated by gradient centrifugation using leukocyte separation medium (Cellgro). The cells were washed in PBS and resuspended in complete culture medium consisting of RPMI 1640, 10% FBS (Gemini Bio-Products, Woodland, CA) supplemented with 1× penicillin/streptomycin/l-glutamine, 1× sodium pyruvate, 1× nonessential amino acids, and 50 μM 2-ME (Life Technologies, Rockville, MD). Freshly isolated PBMCs were also cryopreserved in 90% FCS and 10% DMSO in liquid nitrogen for future testing.
For each stimulation condition, ≥500,000 total events were acquired on a BD LSR II, and data analysis was performed using FlowJo version 9.9.4 (TreeStar). PBMCs were analyzed by flow cytometry after staining with fluorochrome-conjugated human specific mAbs. The following anti-human Abs were used: CD3 (clone SK7) PE-Cy7, CD44 (clone G44-26) A700, CD8 (clone SK1) allophycocyanin-Cy7, CCR7 (clone 150503) Alexa Fluor 700, CD45RA FITC, CD62L allophycocyanin, IFN-γ Alexa Fluor 647, CD107a (clone H4A3) FITC, and CD107b (clone H4B4) FITC (all from BioLegend). The following anti-mouse Abs were used to characterize mouse CD8+ TCM, CD8+ TEM, and CD8+ TRM subpopulations: anti-mouse CD8 (clone 53–6.7) PE-Cy7, anti-mouse CD44 (clone IM7) allophycocyanin-Cy7, anti-mouse CD11a (clone M17/4) FITC, anti-mouse CD103 (clone M290) allophycocyanin, anti-mouse CD62L (clone MEL-14) A700, anti-mouse CCR7 (clone 4B12) A647, anti-mouse CD69 (clone H1.2F3) PE-Cy7, CD107a (clone ID4B) FITC, CD107b (clone M3/84) FITC (all from BD Pharmingen), anti-mouse IFN-γ (clone XMG1.2) PE-Cy7 (BioLegend), TIGIT (clone GIGD7) PE (eBioscience), and VISTA (clone 13f3, a gift from R.J.N.). For surface staining, mAbs against various cell markers were added to a total of 1 × 106 cells in 1× PBS containing 1% FBS and 0.1% sodium azide (FACS buffer) for 45 min at 4°C. After washing with FACS buffer, cells were permeabilized for 20 min on ice using the Cytofix/Cytoperm Kit (BD Biosciences) and then washed twice with Perm/Wash Buffer (BD Biosciences). Intracellular cytokine mAbs were then added to the cells and incubated for 45 min on ice in the dark. Cells were washed again with Perm/Wash Buffer and FACS buffer and fixed in PBS containing 2% paraformaldehyde (Sigma-Aldrich, St. Louis, MO). For each sample, 200,000 total events were acquired on BD LSR II. Ab capture beads (BD Biosciences) were used as individual compensation tubes for each fluorophore in the experiment. To define positive and negative populations, we used fluorescence minus one controls for each fluorophore used in this study when initially developing staining protocols. In addition, we further optimized gating by examining known negative cell populations for background level expression. The gating strategy was similar to that used in our previous work (6). Briefly, we gated single cells, dump− cells, viable cells (Aqua Blue), lymphocytes, CD3+ cells, and CD8+ cells before finally gating human epitope-specific CD8+ T cells using HSV-specific tetramers. Data analysis was performed using FlowJo version 9.9.4 (TreeStar). Statistical analyses were done using GraphPad Prism version 5 (GraphPad, La Jolla, CA).
Production of IFN-γ and expression of cytotoxic markers
To detect cytolytic CD8+ T cells recognizing peptides in freshly isolated PBMCs, we performed intracellular IFN-γ and CD107a/b cytotoxicity assay. Betts and colleagues (37) have described the CD107a/b assay as an alternative cytotoxicity assay that is able to address some of the shortcomings of the [51Cr]-release assay. The intracellular assay to detect IFN-γ and CD107a/b in response to in vitro peptide stimulations was performed as described (38, 39), with a few modifications. On the day of the assay, 1 × 106 PBMCs were stimulated in vitro with peptide pools (10 μg/ml per peptide) at 37°C for an additional 6 h in a 96-well plate with BD GolgiStop (BD Biosciences), 10 μl each of CD107a FITC and CD107b FITC. PHA (5 μg/ml; Sigma-Aldrich) or and no peptide were used as positive and negative controls, respectively. At the end of the incubation period, the cells were transferred to a 96-well round-bottom plate, washed twice with FACS buffer, and stained with PE-conjugated anti-human CD8 for 45 min at 4°C. The intracellular staining for the detection of IFN-γ and CD107a/b was performed as outlined above. The cells were washed again and fixed, 500,000 total events were acquired on a BD LSR II, and data analysis was performed using FlowJo version 9.9.4 (TreeStar). Peptide pools yielding positive responses were deconvoluted by testing individual peptides in vitro at 10 μg/ml.
T cell–proliferation assay
CD8+ T cell proliferation was measured using a CFSE assay, as we recently described (5, 18). Briefly, PBMCs were labeled with CFSE (2.5 μM; Molecular Probes) in 1× PBS at room temperature. Cold FCS was added, and cells were washed extensively with RPMI 1640 plus 10% FCS. CFSE-labeled cells were incubated or not with peptide pools and individual peptides (10 μg/ml) and incubated for 6 d. As a positive control, 2 μg/ml of PHA was used to stimulate T cells for 3 d. Then cells were washed and stained with PE-conjugated mAbs specific to human CD8+ molecules. The numbers of dividing CD8+ T cells per 300,000 total cells were analyzed by FACS. Their absolute number was calculated using the following formula: (number of events in CD8+/CFSE+ cells) × (number of events in gated lymphocytes)/(number of total events acquired).
Design and construction of adeno-associated virus type 8 vector expressing mouse chemokine under neurotropic promoters (CamKIIα and CMV)
The ORFs for mouse chemokines (CCL5, CXCL9, or CXCL10) were cloned into SalI/EcoRI sites of a pAAV-CamKIIα-MCS-CamKIIα-EGFP vector or a pAAV-CMV-MCS-CMV-EGFP vector to determine the best vector for targeted delivery of chemokines in neuronal cells of mouse TG. These pAAV vectors were cotransfected with helper plasmid in HEK293 cells to produce the adeno-associated virus (AAV). Two days after transfections, cell pellets were harvested, and viruses were released through three cycles of freeze/thaw. AAVs were purified through CsCl-gradient ultracentrifugation, followed by desalting. Viral titers (genome copies per milliliter) were determined through real-time PCR.
Prime/pull therapeutic vaccine in humanized HLA-A*0201–Tg mice
HLA-A*0201–Tg mice were kindly provided by Dr. Lemonier (Pasteur Institute) and were described recently (1–3). To screen for neurotropic vector (CMV and CamKIIα), 6–8-wk-old HLA-A*0201–Tg mice were ocularly infected (without scarification of cornea) using 2 × 105 wild-type McKrae (HSV-1) strain. Thirty-five days PI, when latency was fully established, reactivation of latent HSV-1 infection was induced following UV-B irradiation. Each eye was irradiated with 250 mJ of UV-B light per square centimeter (60-s exposure on the transilluminator). Anesthetized mice were placed on the transilluminator, and each mouse was positioned on a piece of cardboard containing a hole the same size as the mouse’s eye. This allowed just the eye to be irradiated by the UV-B source. Mice were divided into three groups for various days post–UV-B irradiation (i.e., day 1, day 2, day 5, day 7 after UV-B exposure). Group 1 mice were administered AAV8-CamKIIα-mCXCL10-CamKIIα-eGFP vector, group 2 mice were administered AAV8-CMV-mCXCL10-CMV-eGFP vector, and group 3 mice received an empty vector. Mice were sacrificed at various times after chemokine administration,. The frequencies of GFP+NeuN+ neuronal cells were evaluated in TG of mice from all three groups.
In another experiment evaluating the prime/pull vaccine model, 6–8-wk-old HLA-A*0201–Tg mice were ocularly infected (without scarification of cornea) using 2 × 105 wild-type McKrae (HSV-1) strain. Twenty-one days postinfection (PI), when latency was fully established, mice were divided into three groups. Group 1 and group 2 were vaccinated with ASYMP epitopes selected from HSV-1 genome-wide epitopes, along with PADRE and CpG1826 adjuvant. Group 3 mice were mock vaccinated with CpG1826 adjuvant without ASYMP epitopes (mock vaccinated). After 2 wk of vaccination (day 35 PI), eyes of all three groups of mice were exposed to UV-B radiation to induce virus reactivation. On day 37 PI, group 2 mice received additional treatment with 107 PFU rAAV8-CamKIIα-GFP-CamKIIα-CXCL10 vector (vaccinated + CXCL10). The virus shedding induced by UV-B exposure was quantified in eye swabs collected every day post–UV-B light exposure (up to 10 d). Eyes were swabbed using moist type 1 calcium alginate swabs that were then placed in 1 ml of titration media (Media 199, 2% penicillin/streptomycin, 2% newborn calf serum) and frozen at −80°C until titrated on RS cell monolayers, as described previously (6). Animals were examined for signs of recurrent corneal herpetic disease by slit lamp for 30 d post–UV-B radiation; this was performed by investigators who were blinded to the treatment regimen of the mice and scored according to a standard 0–4 scale (0, no disease; 1, 25% staining; 2, 50% staining; 3, 75% staining; and 4, 100% staining), as previously described (1–3). Following CXCL10 administration, mice were sacrificed, and the frequency and function of HSV-specific memory CD8+ T cells were evaluated in TG of mice from all three groups. All animal experiments were conducted following National Institutes of Health guidelines for the housing and care of laboratory animals in facilities approved by the University of California Irvine Association for Assessment and Accreditation of Laboratory Animal Care and according to Institutional Animal Care and Use Committee–approved animal protocols (IACUC #2002-2372).
Mouse TG were cut into 8-μm-thick sections using a cryostat. Sections were washed with 1× PBS, permeabilized using 0.05% Triton X-100 in 1× PBS for 15 min, and blocked using 10% FBS in 1× PBS for 1 h. Sections were costained using anti-mouse mAbs specific to A5 or KH10 neurons (1:200 dilution) and CD8+ anti-rabbit Ab (1:200) overnight at 4°C. Sections were washed with 1× PBS and costained with anti-mouse PE (1:500, indicating A5+ or KH10+ neurons) and Cy5 (1:200, indicating CD8+ T cell proliferation). After secondary fluorescent staining, sections were washed with 1× PBS and mounted after DAPI staining (1:10,000 dilution). Immunofluorescence localization of A5+ or KH10+ and CD8+ with AAV8-GFP was examined using a Keyence BZ-X700 fluorescent microscope at 40× magnification and imaged using z-stack.
We examined the distribution of each immunological parameter. In the case of two-group comparisons, we used the parametric two-sample t test or nonparametric Wilcoxon rank sum test, as appropriate. For parameters satisfying a normal distribution, we used the F-test for equality of variances to determine whether it would be appropriate to apply the Satterthwaite test assuming unequal variances between groups. For paired comparisons involving multiple peptides, we used the conservative Bonferroni procedure to adjust for the 22 groups of peptides (Fig. 1), 22 or 23 individual peptides (Fig. 2), and four group comparisons (Fig. 4E, Supplemental Fig. 2) that were examined. We report the nominal p values in the corresponding figures after correction for multiple comparisons. In the specific case of three groups (i.e., comparing two subgroups with a baseline subgroup), we used the general linear model procedure and compared the least squares means using the Dunnett procedure for multiple comparisons (Fig. 4B). Flow cytometry data were analyzed with FlowJo software (TreeStar). For analysis, we used SAS v.9.4 (Statistical Analysis System, Cary, NC). Graphs were prepared with GraphPad Prism software. Data are expressed as mean + SD. Error bars show SEM.
Frequent HSV-specific IFN-γ–producing CD8+ T cells, expressing cytotoxic markers, detected in seropositive individuals are directed against a limited number of epitopes from the HSV-1 genome
PBMCs from 20 HLA-A*0201+ and 8 HLA-A*0201− HSV-seropositive individuals were stimulated in vitro for 72 h with each of the 22 groups of peptide epitopes (i.e., grp.1 to grp.22) spanning the entire HSV-1 genome. We then used intracellular FACS staining to measure IFN-γ and CD107a/b produced by stimulated CD8+ T cells, as described in 2Materials and Methods.
Fig. 1A shows representative contour plots of the percentages of IFN-γ+CD8+ T cells detected in one HLA-A*0201+ and one HLA-A*0201− HSV-1–seropositive individual following stimulation with equimolar amounts of peptides from grp.6, grp.11, and grp.16. Fig. 1B shows the average percentages of IFN-γ+CD8+ T cells detected from 20 HLA-A*0201+ and 8 HLA-A*0201− HSV-1–seropositive individuals. The highest percentages of IFN-γ+CD8+ T cells detected from 20 HLA-A*0201+ HSV-1–seropositive individuals were directed against three groups of peptides: grp.6, grp.11, and grp.16. Grp.6 HSV-1 peptides induced up to 4.4% IFN-γ+CD8+ T cells from 13 of the 20 HLA-A*0201+ HSV-1–seropositive individuals. Similarly, significant percentages of IFN-γ+CD8+ T cells (up to 6.6%) were detected in 17 of 20 individuals in response to grp.11 HSV-1 peptides. Grp.16 induced 1.6–3.9% IFN-γ+CD8+ T cells in 15 of 20 HSV-1–seropositive individuals. Grp.12 and grp.20 induced a significant percentage of IFN-γ+CD8+ T cells in only 2 of 20 HLA-A*0201+ HSV-1–seropositive individuals. In contrast, none of the remaining 16 groups of HSV-1 peptides stimulated a significant percentage of IFN-γ+CD8+ T cells in any of the 20 HLA-A*0201+ HSV-1–seropositive individuals tested. In parallel experiments, no significant percentages of IFN-γ+CD8+ T cells were detected from PBMCs of the eight HLA-A*0201− HSV-1–seropositive individuals when stimulated with equimolar amounts of the 22 groups of peptides, indicating the expected HLA specificity of the T cell responses. These results indicate that the HSV-1 peptides in grp.6, grp.11, and grp.16 contain at least one immunodominant IFN-γ–producing CD8+ T cell epitope.
We next examined the ability of each group of HSV-1 peptides (i.e., grp.1 to grp.22) to induce cytotoxic CD8+ T cells from 20 HLA-A*0201+ and 8 HLA-A*0201− HSV-1–seropositive individuals. For this evaluation, we assessed the expression of CD107a and CD107b on stimulated CD8+ T cells by FACS and determined the percentage of CD107a/b+CD8+ T cells in HLA-A*0201+ and HLA-A*0201− HSV-1–seropositive individuals. Lysosomal-associated membrane glycoproteins CD107a and CD107b are two markers of cytotoxic function that surround the core of the lytic granules in CTLs. Upon TCR engagement and epitope stimulation, CD107a/b expression is upregulated on the cell membrane of CD8+ cytotoxic T cells.
As shown in the representative data in Fig. 1C, grp.6, grp.11, and grp.16 HSV-1 peptides induced high percentages of CD107a/b+CD8+ T cells in PBMCs from an HLA-A*0201+ individual but not an HLA-A*0201− individual. Fig. 1D shows the average percentages of CD107a/b+CD8+ T cells detected from 20 HLA-A*0201+ and 8 HLA-A*0201− HSV-1–seropositive individuals. The highest percentages of CD107a/b+CD8+ T cells detected from the 20 HLA-A*0201+ HSV-1–seropositive individuals were also directed against grp.6, grp.11, and grp.16.
Up to 5.7% of CD107a/b+CD8+ T cells were recorded in PBMCs from 16 of 20 HLA-A*0201+ individuals following stimulation with HSV-1 peptides from grp.6. Seventeen of 20 HLA-A*0201+ individuals responded strongly (up to 5.2%) to grp.11 peptides, whereas 18 of 20 HLA-A*0201+ individuals showed up to 4.4% CD107a/b+CD8+ T cells in response to grp.16 peptides. Grp.12 induced 2–3.6% CD107a/b+CD8+ T cells in 8 of 20 HLA-A*0201+ individuals. Modest, yet significant, percentages of CD107a/b+CD8+ T cells were also detected against three additional groups of HSV-1 peptides: grp.8, grp.12, and grp.20. In contrast, the remaining 13 groups of HSV-1 peptides did not induce significant percentages of CD107a/b+CD8+ T cells following stimulation of PBMCs from 20 HLA-A*0201+ HSV-1–seropositive individuals. As expected, none of the 22 groups of HSV-1 peptides induced significant percentages of CD107a/b+CD8+ T cells in the eight HLA-A*0201− HSV-seropositive individuals tested, indicating an HLA specificity in the cytotoxic CD8+ T cell responses (Fig. 1D).
Because a lack of IFN-γ production and CD107a/b expression may not always reflect the lack of a T cell response (20, 40), we studied the proliferative response of CD8+ T cells to the 22 groups of HSV-1 peptides from 20 HSV-1–seropositive individuals. PBMCs were labeled with CFSE and then restimulated in vitro for 6 d with each of the 22 groups of HSV-1 peptides. As shown in Fig. 1E (upper panel), significant percentages of CFSElowCD8+ T cells were detected from one HLA-A*0201+ individual when stimulated with grp.6, grp.11, grp.12, and grp.16 HSV-1 peptides. Grp.6, grp.11, and grp.16 induced the highest proliferation of CD8+ T cells (30.2, 32.5, and 35.7% of dividing CFSElowCD8+ T cells, respectively). Grp.12 induced a modest, yet significant, 11.8% of dividing CFSElowCD8+ T cells. In contrast, none of the remaining 18 groups of HSV-1 peptides induced significant proliferation of CD8+ T cells in HLA-A*0201+ individuals. No significant percentages of dividing CFSElowCD8+ T cells were induced by any of the 22 groups of HSV-1 peptides in HLA-A*0201− individuals, confirming the HLA specificity of the T cell responses (Fig. 1E, lower panel).
Altogether, these results indicate that grp.6, grp.11, and grp.16 HSV-1 peptides contain one or several immunodominant epitopes that induce HLA-A*020–restricted IFN-γ–producing CD8+ T cells expressing cytotoxic markers in HLA-A*0201+ HSV-seropositive individuals. Therefore, the remainder of this study focused on identifying the individual HSV-1 peptides from grp.6, grp.11, and grp.16 that produced the CD8+ T cell responses.
Immunodominant HSV-1 genome-wide–derived epitopes targeted by IFN-γ–producing CD8+ T cells, expressing cytotoxic markers, from HLA-A*0201+ HSV-seropositive individuals
We next compared the ability of individual HSV-1 peptides within grp.6, grp.11, and grp.16 to induce HSV-specific CD8+ T cell responses in 10 HLA-A*0201+ individuals and 8 HLA-A*0201− individuals.
As shown in the representative contour plots in Fig. 2A, 4.3% of IFN-γ+CD8+ T cells specific to the UL9196–204 epitope were detected in an HLA-A*0201+ HSV-1–seropositive individual, whereas only 0.5% of IFN-γ+CD8+ T cells were detected in an HLA-A*0201− HSV-1–seropositive individual following stimulation with equimolar amounts of the same peptide. Fig. 2B shows the average percentage of IFN-γ+CD8+ T cells detected in 10 HLA-A*0201+ individuals and 8 HLA-A*0201− individuals for each of the 23 individual peptides in grp.6. The highest percentage of IFN-γ+CD8+ T cells detected from 10 HLA-A*0201+ HSV-1–seropositive individuals were directed against two immunodominant epitopes: the UL9196–204 epitope (up to 4.1% of IFN-γ+CD8+ T cells) and the UL10162–170 epitope (up to 4.3% of IFN-γ+CD8+ T cells). The UL9196–204 epitope belongs to the HSV-1 replication-initiator UL9 protein, and the UL10162–170 epitope belongs to the membrane UL10 protein. Fig. 2C shows representative contour plots of the percentage of IFN-γ+CD8+ T cells induced by the immunodominant UL25572–580 epitope from grp.11. Fig. 2D shows the average percentage of IFN-γ+CD8+ T cells detected in 10 HLA-A*0201+ individuals and 8 HLA-A*0201− individuals for each of the 23 peptides in grp.11. The highest percentage of IFN-γ+CD8+ T cells was directed against three epitopes that belong to the HSV-1 DNA packaging tegument protein (UL25) and the capsid maturation protease (UL26): UL25572–580 epitope (up to 3.7%), UL26480–488 epitope (up to 4.0%), and UL26221–229 epitope (up to 3.6%). Fig. 2E shows representative contour plots of the percentage of IFN-γ+CD8+ T cells induced by the immunodominant UL43272–280 epitope from grp.16. Fig. 2F shows that the highest average percentage of IFN-γ+CD8+ T cells detected in 10 HLA-A*0201+ individuals was directed primarily against six epitopes from grp.16 that belong to the highly conserved sequence in alphaherpesvirus and gammaherpesvirus envelope proteins UL43 and UL44: UL43148–156 epitope (1.3–3.9%), UL43272–280 epitope (2.3–6.2%), UL43386–394 epitope (3.6–7.9%), UL43302–310 epitope (3.1–7.0%), UL44400–408 epitope (3.0–6.0%), and UL44443–451 epitope (2.9–6.3%).
We then examined the ability of individual HSV-1 peptides from grp.6, grp.11, and grp.16 to induce CD8+ T cells expressing cytotoxic markers from 10 HLA-A*0201+ and 8 HLA-A*0201− HSV-1–seropositive individuals. Fig. 2G shows the percentage of CD107a/b+CD8+ T cells specific to the UL9196–204 epitope from grp.6 detected in one HLA-A*0201+ and one HLA-A*0201− HSV-1–seropositive individual. Fig. 2H shows that the highest percentages of CD107a/b+CD8+ T cells detected in 10 HLA-A*0201+ individuals were primarily directed against two UL9 and UL10 epitopes: UL9196–204 (up to 5.1%) and peptide UL10162–170 (up to 6.8%). Fig. 2I shows representative contour plots of the percentages of CD107a/b+CD8+ T cells detected in one HLA-A*0201+) and one HLA-A*0201−) HSV-1–seropositive individual following in vitro stimulation with the immunodominant peptide epitope UL25572–580 from grp.11. Fig. 2J shows that the average highest % CD107a/b+CD8+ T cells was detected against three epitopes from UL25 and UL26: UL25367–375 (up to 2.5%), UL25572–580 (up to 4.5%), and UL26221–229 (up to 4.7%). Fig. 2K shows the percentages of CD107a/b+CD8+ T cells detected in one HLA-A*0201+ and one HLA-A*0201− HSV-1–seropositive individual following in vitro stimulation with the immunodominant peptide epitope UL43272–280 from grp.16. Fig. 2L shows that the average highest percentages of CD107a/b+CD8+ T cells were detected against five peptide epitopes from UL43 and UL44: UL43272–280 (up to 4.9%), UL43386–394 (up to 5.2%), UL43302–310 (up to 5.6%), UL44400–408 (up to 5.7%), and UL44443–451 (up to 5.1%). In parallel experiments, no significant percentages of IFN-γ–producing CD8+ T cells expressing cytotoxic markers were detected in eight HLA-A*0201− HSV-1–seropositive individuals when stimulated with equimolar amounts of peptide epitopes from grp.6, grp.11, and grp.16 (Fig. 2G–L), confirming the HLA specificity of the CD8+ T cell responses.
Altogether, these results demonstrate that IFN-γ–producing CD8+ T cells expressing cytotoxic markers in the majority of HLA-A*0201+ HSV-1–seropositive individuals are directed against 12 epitopes derived from the whole HSV-1 genome: UL9196–204, UL10162–170, UL25367–375, UL25572–580, UL26480–488, UL26221–229, UL43272–280, UL43386–394, UL43148–156, UL43302–310, UL44400–408, and UL44443–451. Furthermore, we demonstrated that the 10 immunodominant epitopes above also induced significant CD8+ T cell proliferative responses in the majority of HLA-A*0201+ HSV-1–seropositive individuals (up to 32.4% of CFSElow CD8+ T cells) (Supplemental Fig. 1). In contrast, a very low frequency of proliferative CD8+ T cells (2.2–4.3% CFSElow CD8+ T cells) was detected from HLA-A*0201− controls in response to equimolar amounts of these individual peptides. These results confirm the immunodominance of the 12 epitopes from the HSV-1 genome.
Most immunodominant HSV-1 genome–derived CD8+ T cell epitopes bind with high affinity to, and stabilize the expression of, HLA-A*0201 molecules on the surface of target cells
Ten potential epitopes with high predicted affinity to the HLA-A*0201 molecule were identified from the amino acid sequence of the HSV-1 genome (strain 17) using BIMAS, SYFPEITHI, and MAPPP computational algorithms (Fig. 3A). We next assessed whether the 10 HSV-1 genome-wide–derived CD8+ epitopes predicted above would bind and stabilize HLA-A*0201 molecules on the surface of target cells. Binding affinities were determined in classical competition assays that measure the capacity of each peptide to competitively inhibit the binding of a high-affinity radiolabeled reference flu peptide to purified HLA-A*0201 molecules, as described in 2Materials and Methods and by Sidney et al. (35). The concentration of HSV-1 peptide yielding 50% inhibition of the binding of the radiolabeled reference peptide (IC50) was calculated. Under the conditions used, the measured IC50 values are reasonable approximations of the true Kd values (41). As shown in Fig. 3B, 9 of 10 HSV-1 genome-wide–derived peptide epitopes bound with high affinity to soluble HLA-A*0201 molecules, with IC50 < 100 nM. Notably, the peptides UL25572–580, UL26480–488, UL26221–229, and UL44400–408 bound with high affinity to HLA-A*0201 molecules, with IC50 < 25 nM. The UL9196–204 peptide bound with high to intermediate binding affinities to HLA-A*0201, with IC50 ∼ 46 nM. Peptides UL10162–170, UL43272–280, UL43386–394, and UL43302–310 bound with intermediate affinities to HLA-A*0201, with IC50 between 50 and 100 nM. The remaining UL44443–451 peptide bound to HLA-A*0201 with low binding affinity (IC50 of 1099 nM).
In an additional set of assays, we also determined the capacity of each peptide to stabilize the expression of HLA-A*0201 molecules on the surface of target T2 cells. As shown in Fig. 3C, 8 of 10 peptides, UL9196–204, UL10162–170, UL25572–580, UL26480–488, UL26221–229, UL43386–394, UL43302–310, and UL44400–408, significantly increased the level of HLA-A*0201 molecules on the surface of T2 cells, as detected by FACS, which suggests their high affinity for HLA-A*0201 molecules. The increase in HLA-A*0201 molecule expression was dose dependent. The remaining peptides (UL43272–280 and UL44443–451) demonstrated a moderate stabilization of HLA-A*0201 expression on the surface of T2 cells, indicating their medium affinity for HLA-A*0201 molecules (Fig. 3C). Altogether, these results indicate that UL9196–204, UL10162–170, UL25572–580, UL26480–488, UL26221–229, UL43272–280, UL43386–394, UL43302–310, UL44400–408, and UL44443–451 bind with high to moderate affinity to HLA-A*0201 molecules and stabilize their expression on the surface of target cells.
ASYMP individuals have higher frequencies of multifunctional CD8+ TEM cells specific to HSV-1 genome-wide–derived immunodominant epitopes, whereas SYMP individuals have higher frequencies of monofunctional CD8+ TCM cells
Next, we investigated the frequency, phenotype, and function of HSV-1 genome-wide–derived epitope-specific CD8+ T cells in HLA-A*0201+ HSV-1–seropositive SYMP and ASYMP individuals (Table I). SYMP individuals have frequent clinical ocular herpetic diseases, such as herpetic lid lesions, herpetic conjunctivitis, dendritic or geographic keratitis, stromal keratitis, and iritis, consistent with recurrent HSK. This is in stark contrast to ASYMP individuals who, despite their HSV-1 seropositivity, have never had any recurrent herpes disease, ocular, genital or elsewhere, based on self-reporting and physician examination. The characteristics of the SYMP and ASYMP study populations, with respect to gender, age, HLA-A*0201 frequency distribution, HSV-1/HSV-2 seropositivity, and status of ocular herpetic disease, are presented in Table II and are detailed in 2Materials and Methods. Because HSV-1 is the main cause of ocular herpes, only individuals who were HSV-1 seropositive and HSV-2 seronegative were enrolled in the current study. The low frequencies of PBMC-derived HSV-specific CD8+ T cells complicate a direct ex vivo tetramer detection of CD8+ T cells using a typical number of PBMCs (i.e., 1 × 106 cells), and a prior expansion of CD8+ T cells by HSV-1 or peptide stimulation in an in vitro culture would hamper the reliable determination of the frequencies of HSV-1 epitope–specific CD8+ T cells. Therefore, we opted to determine the frequencies of HSV-1 epitope–specific CD8+ T cells ex vivo using a large number of PBMCs (∼10 × 106) per tetramer/CD8 mAb panel.
|All Subjects (n = 803)|
|Gender (n [%])|
|Race (n [%])|
|Age (y; median [range])||39 (21–67)|
|HSV status (n [%])|
|HSV-1 and HSV-2+||36 (4)|
|HLA status (n [%])|
|Herpes disease status (n [%])|
|All Subjects (n = 803)|
|Gender (n [%])|
|Race (n [%])|
|Age (y; median [range])||39 (21–67)|
|HSV status (n [%])|
|HSV-1 and HSV-2+||36 (4)|
|HLA status (n [%])|
|Herpes disease status (n [%])|
|Enrolled Subjects (n = 38)|
|Gender (n [%])|
|Race (n [%])|
|Age (y; median [range])||39 (21–67)|
|HSV status (n [%])|
|HSV-1 and HSV-2+||0 (0)|
|HLA status (n [%])|
|Herpes disease status (n [%])|
|Enrolled Subjects (n = 38)|
|Gender (n [%])|
|Race (n [%])|
|Age (y; median [range])||39 (21–67)|
|HSV status (n [%])|
|HSV-1 and HSV-2+||0 (0)|
|HLA status (n [%])|
|Herpes disease status (n [%])|
As shown in Fig. 4A (representative data) and Fig. 4B (average of frequencies), the highest and most significant frequencies of CD8+ T cells detected in eight SYMP and eight ASYMP individuals were directed against 4 of 10 peptide epitopes: UL9196–204, UL25572–580, UL43302–310, and UL44400–408 (1.7–5.9%). Interestingly, the average frequencies of CD8+ T cells specific to the UL9196–204 epitope were consistently and significantly higher in HLA-A*0201+ ASYMP individuals (Fig. 4B, black circles) compared with HLA-A*0201+ SYMP individuals (Fig. 4B, white circles) (p = 0.005). In contrast, the average frequencies of CD8+ T cells specific to the UL43302–310 epitope were significantly higher in HLA-A*0201+ SYMP individuals compared with HLA-A*0201+ ASYMP individuals. The average frequencies of CD8+ T cells specific to the remaining eight epitopes, UL10162–170, UL25572–580, UL26480–488, UL26221–229, UL43272–280, UL43386–394, UL44400–408, and UL44443–451, were at similarly low levels in HLA-A*0201+ SYMP and ASYMP individuals. The average frequencies of CD8+ T cells specific to the 10 immunodominant epitopes in HLA-A*0201− controls did not yield any significant percentages, confirming the HLA specificity (Fig. 4B, gray circles).
CD8+ T cells specific to the UL9196–204 ASYMP epitope and to the UL43302–310 SYMP epitope were visualized from SYMP and ASYMP individuals using an ImageStream assay as an alternative and complementary assay to FACS. Fig. 4C shows the strong response of CD8+ T cells from one ASYMP individual when stained with a tetramer specific to the UL9196–204 ASYMP epitope (upper left panel) in contrast with weak staining with a tetramer specific to the UL43302–310 SYMP epitope (upper right panel). Conversely, a strong response of CD8+ cells from one SYMP individual was detected when stained with a tetramer specific to the UL43302–310 SYMP epitope (lower right panel) in contrast to a weak staining with a tetramer specific to the UL9196–204 ASYMP epitope (lower left panel).
We next determined whether the memory CD8+ T cells specific to the immunodominant UL9196–204, UL25572–580, UL43302–310, and UL44400–408 epitopes are composed of distinct subpopulations of central memory cells (i.e., CD44highCD62highCD8+ TCM cells) and effector memory cells (i.e., CD44highCD62lowCD8+ TEM cells). As shown in Fig. 4D (representative data) and Fig. 4E (average of frequencies), ASYMP individuals had significantly higher percentages of CD44highCC62LlowCD8+ TEM cells specific to UL9196–204, UL25572–580, and UL44400–408 epitopes compared with SYMP individuals (p ≤ 0.003, unpaired t test). In contrast, significantly higher percentages of UL9196–204 and UL25572–580 epitope-specific CD8+CD44highCC62Lhigh TCM cells were consistently detected in SYMP individuals compared with ASYMP individuals (p ≤ 0.002, unpaired t test).
Fig. 4F shows that, overall, 92% of ASYMP individuals had HSV-specific CD8+ TEM cells with three to six functions, indicating their ability to display concurrent polyfunctional activities: production of high levels of IFN-γ, expression of high levels of cytotoxic markers (CD107a/b/perforin/granzyme B) (Supplemental Fig. 2), proliferation, frequency of tetramer-specific CD8+ T cells, stabilization of soluble HLA molecule, and T2 binding assay. In contrast, only 42% of SYMP patients had HSV-specific CD8+ T cells with three functions, whereas the majority of them had just one or two functions.
Altogether, the frequency and phenotypic and functional properties of human CD8+ T cell–specific HSV-1 genome-wide–derived epitopes suggest that some ASYMP epitopes (e.g., UL9196–204) were associated with the natural protection seen in ASYMP individuals, whereas other SYMP epitopes (e.g., UL43302–310) were associated with the lack of protection seen in SYMP individuals; a clear dichotomy in HSV-specific memory CD8+ T cell subpopulations exists in HSV-1–infected individuals, with ASYMP individuals featuring higher frequencies of CD8+ TEM cells and SYMP individuals featuring higher frequencies of CD8+ TCM cells; and the majority of ASYMP individuals had multifunctional HSV-specific CD8+ TEM cells, with three to six functions, whereas the majority of SYMP individuals had monofunctional HSV-specific CD8+ TCM cells.
The above findings imply that, following HSV-1 reactivation from latency, ASYMP individuals, but not SYMP individuals, would be better armed immunologically to control herpes infection and disease by mounting faster and stronger protective antiviral CD8+ TEM cell responses. To test this hypothesis, throughout the remainder of the study, we used our UV-B–induced HSV-reactivation mouse model (13, 14) to determine whether therapeutic immunization of latently infected HLA-A*0201–Tg mice with ASYMP epitopes will preferentially induce HSV-specific TEM cells associated with protection against recurrent herpes.
Successful delivery of CXCL10 chemokine to HSV-1 latently infected TG using the neurotropic recombinant nonreplicating AAV8 vector
We next investigated whether a chemokine-encoding recombinant replication-deficient AAV8 vector that preferentially infects the neurons would deliver the T cell–attracting chemokines CCL5, CXCL9, and CXCL10 into HSV-1 latently infected TG and whether that would be accompanied by an increase in the number of CD8+ T cells infiltrating TG.
To optimize such a delivery, we first searched for a promoter that would lead to a specific and optimal expression of GFP in sensory neurons of HSV-1 latently infected TG. Two types of promoters were tested: the universal CMV promoter and the neurotropic CamKIIα promoter. The CMV or CamKIIα promoters were used to control expression of GFP and CXCL10 chemokine, resulting in two types of AAV8 vectors: the rAAV8-CMV-GFP-CMV-CXCL10 vector and the rAAV8-CamKIIα-GFP-CamKIIα-CXCL10 vector (Fig. 5A).
A group of HLA-A*0201–Tg mice (n = 30) latently infected with HSV-1 was exposed to UV-B light on day 35 PI to induce HSV-1 reactivation. Then mice were divided into three groups (n = 10 per group) and received topical ocular application of 107 PFU the rAAV8-CMV-GFP-CMV-CXCL10 vector, the rAAV8-CamKIIα-GFP-CamKIIα-CXCL10 vector, or an empty AAV8 vector as control on day 37 PI. The percentages of NeuN+ neuronal cells expressing GFP (i.e., NeuN+GFP+) were determined from total cells isolated from TG by FACS at various days postinoculation.
As shown in Fig. 5B (left panel, representative data, right panel, average of frequencies), 6 d following topical ocular application of rAAV8-CamKIIα-GFP-CamKIIα-CXCL10, 27.1–39.4% of GFP+NeuN+ neuronal cells in TG of HLA-A*0201–Tg mice expressed GFP. In contrast, only about half (∼14.5%) of GFP+NeuN+ neuronal cells in TG expressed GFP following topical ocular application of the rAAV8-CMV-GFP-CMV-CXCL10 vector (p < 0.05). The expression of GFP in TG persists for ≥3 wk (data not shown). As expected, no significant percentage of GFP+NeuN+ neuronal cells was detected in control TG that received the empty AAV8 vector. These results indicate that the neurotropic CamKIIα is an optimal promoter for expression of T cell–attracting chemokines in HSV-1 latently infected TG.
Using the optimal CamKIIα promoter, we then constructed two additional recombinant AAV8 vectors expressing the remaining T cell–attracting chemokines CCL5 and CXCL9 (in addition to CXCL10) (Fig. 5C): rAAV8-CamKIIα-GFP-CamKIIα-CCL5 and rAAV8-CamKIIα-GFP-CamKIIα-CXCL9. We next determined which one of the three chemokine-encoding recombinant nonreplicating AAV8 vectors shown in Fig. 5C would recruit more CD8+ T cells in HSV-1 latently infected TG. Three groups of HSV-1 latently infected HLA-A*0201–Tg mice (n = 30 per group) were treated with UV-B light on day 35 PI to induce HSV-1 reactivation and were inoculated 2 d later with the rAAV8-CamKIIα-GFP-CamKIIα-CCL5 vector, the rAAV8-CamKIIα-GFP-CamKIIα-CXCL9 vector, or the rAAV8-CamKIIα-GFP-CamKIIα-CXCL10 vector (Fig. 5D). An empty AAV8 vector was used as control. The percentages of GFP+NeuN+ neuronal cells and CD103+CD8+ T cells were determined in TG cell suspension using FACS assay. As shown in Fig. 5E, inoculation of the rAAV8-CamKIIα-GFP-CamKIIα-CXCL10 vector led to significantly more CD8+ T cells in TG of latently infected HLA-A*0201–Tg mice compared with the rAAV8-CamKIIα-GFP-CamKIIα-CCL5 or rAAV8-CamKIIα-GFP-CamKIIα-CXCL9 vector (p < 0.05). Similar percentages of GFP+NeuN+ neuronal cells were detected in TG of latently infected HLA-A*0201–Tg mice for all three vectors, indicating that similar levels of chemokines were expressed (data not shown). No corneal pathology or recurrent HSK was associated with any of the three chemokine-encoding recombinant nonreplicating AAV8 vectors. Interestingly, GFP appeared to be expressed almost exclusively in KH10+ neurons and not in A5+ neurons (Fig. 5F, upper panels). As a result, more CD8+ T cells were located around KH10+ neurons than A5+ neurons (Fig. 5F, lower panels).
Altogether, these results indicate that the neurotropic CamKIIα promoter, but not the CMV promoter, is optimal for T cell–attracting chemokine expression in HSV-1 latently infected TG; inoculation of the rAAV8-CamKIIα-GFP-CamKIIα-CXCL10 vector led to significantly more CD8+ T cells in TG of latently infected HLA-A*0201–Tg mice compared with the rAAV8-CamKIIα-GFP-CamKIIα-CCL5 and rAAV8-CamKIIα-GFP-CamKIIα-CXCL9 vectors; and GFP and CXCL10 chemokine under the neurotropic CamKIIα promoter appeared to be highly expressed in KH10+ neurons but not in A5+ neurons.
Bolstering the number and function of HSV-specific CD8+ T cells in latently infected TG through a prime/pull therapeutic vaccine
We next determined whether priming with multiple HSV-1 ASYMP human CD8+ T cell epitopes, followed by topical ocular treatment with the rAAV8-CamKIIα-GFP-CamKIIα-CXCL10 vector, would pull more functional antiviral CD8+ T cells within HSV-1 latently infected TG and provide protective immunity at the site of virus reactivation (i.e., HSV-1 latently infected TG) associated with reduced virus shedding and recurrent ocular herpes following UV-B–induced reactivation.
As shown in Fig. 6A, HLA-A*0201–Tg mice (n = 30) were ocularly infected with HSV-1; 21 d PI, when latency was fully established, mice were divided into three groups (n = 10 per group). On day 21, groups 1 and 2 received a s.c. immunization with the three HSV-1 ASYMP CD8+ T cell epitopes (UL9196–204, UL25572–580, and UL44400–408) associated with natural protection seen in ASYMP individuals, mixed with the promiscuous CD4+ Th epitope PADRE and CpG1826 adjuvant. Group 3 served as a negative control and received adjuvant alone (mock vaccinated). Two weeks postvaccination (i.e., on day 35 PI), all animals were exposed to UV-B light to induce HSV-1 reactivation, as we recently described (13, 14). On day 37 PI, group 2 received additional treatment with 107 PFU the rAAV8-CamKIIα-GFP-CamKIIα-CXCL10 vector (vaccinated + CXCL10), as above.
Eye swabs were collected every day post–UV-B light exposure (up to 10 d), and recurrent corneal herpetic disease was recorded for 30 d and scored on a scale of 0–4, as described in 2Materials and Methods. As shown in Fig. 6B and 6C, four mice from the vaccinated + CXCL10 group have no disease (score 0), five mice have minor disease (score of 1), and only one mouse has a score of 3. In contrast, two mice from the vaccinated group, which did not receive CXCL10 treatment, had minor disease (score of 1), six mice have a disease score of 2, and two mice have a score of 4. As expected, most mice from the mock-vaccinated group had notable recurrent herpetic disease with a score of 3–4. Additionally, we recorded significantly low virus titers from mice that were vaccinated and treated with CXCL10 compared with mice that were vaccinated only and with mock-vaccinated mice (p = 0.01, Fig. 6D). These results suggest that the prime/pull vaccine based on ASYMP epitopes and CXCL10 did improve protection against virus shedding and recurrent disease.
Representative FACS plots (Fig. 6E), average frequencies (Fig. 6F), and absolute numbers (Fig. 6G) of ASYMP UL9196–204 epitope–specific functional CD8+ T cells were determined in TG. Significantly higher frequencies of HSV-specific IFN-γ+CD8+ T cells and HSV-specific CD107a/b+CD8+ T cells and significantly lower frequencies of HSV-specific PD-1+CD8+ T cells and HSV-specific VISTA+CD8+ T cells were detected in TG of mice that were vaccinated and treated with CXCL10 compared with mice that received the vaccine alone (p = 0.02), indicating that CXCL10 chemokine treatment contributed to the increased function of TG-resident HSV-specific T cells.
Furthermore, there was a positive correlation between the percentage (Fig. 7A) and the numbers (Fig. 7B) of CD8+ T cells in TG of HLA-A*0201–Tg mice and protection against recurrent disease, indicating a CD8+ T cell–dependent protective mechanism. This suggests that local encounters with HSV-1 Ag are critical for activation of TG-resident CD8+ T cells. The effect of prime/pull therapeutic vaccine persisted for ≥4 wk and occurred only in mice that were exposed to UV-B reactivation (data not shown).
To identify the CD8+ T cell subpopulations (i.e., TRM, TCM, TEM) that correlated with protection in a final group of experiments, 30 mice were infected, divided into three groups, vaccinated, and exposed to UV-B light, as shown in Fig. 6A. A single-cell suspension was obtained from TG; stained for markers of CD8+ TRM, TCM, and TEM cells, including CD8, CD11a, CD69, CD103, CD44, CD62L, and CCR7; and analyzed by FACS. Significantly higher frequencies (Fig. 7C, 7D) and absolute numbers (Fig. 7E) of ASYMP UL9196–204 epitope–specific CD8+ TRM cells (top panels), CD8+ TCM cells (middle panels), and CD8+ TEM cells (bottom panels) were detected in TG of mice that were vaccinated compared with mice that were vaccinated and treated with CXCL10 (p < 0.05). The frequency of UL9196–204 epitope–specific CD8+ TEM cells doubled in TG of protected HLA-A*0201–Tg mice that were vaccinated and treated with CXCL10 (i.e., 6.1 versus 14.9%, respectively). In contrast, no significant difference in the frequency of CD8+ TCM cells was detected in TG of mice that were vaccinated compared with mice that were vaccinated and treated with CXCL10 (p > 0.05, middle panels). Expansion of TRM and TEM cells occurs only in mice that were exposed to UV-B reactivation (data not shown). This confirms that local exposure to HSV-1 Ags is critical for expansion of these TG-resident CD8+ T cell subpopulations.
These results demonstrate that bolstering the number of functional HSV-specific CD8+ TEM cells and TRM cells in TG of latently infected HLA-A*0201–Tg mice reduced viral shedding in tears and recurrent ocular herpetic disease following UV-B light induced reactivation, and they suggest that the protective antiviral CD8+ T cell population in latently infected TG is not compartmentalized only into the TRM cell subpopulation but can also be replenished by the CD8+ TEM cell subpopulation that can migrate from circulation.
HSV-specific CD8+ T cells that reside in latently infected TG appeared to play a critical role in aborting attempts of HSV-1 reactivation from sensory neurons (18, 29). In designing a human herpes therapeutic vaccine, it is critical to identify the protective herpes epitopes derived from the 84+ ORFs of the HSV-1 genome that are exclusively recognized by the human CD8+ T cells from “naturally” protected ASYMP individuals who, despite being infected, never develop recurrent herpetic disease. In the current study, we have implemented several in silico, ex vivo, in vitro, and in vivo phenotypic and functional immunological methods to efficiently generate a genome-wide map of the responsiveness of HSV-specific CD8+ T cells in ASYMP individuals. This leads to the identification of several previously unknown protective human ASYMP CD8+ T cell epitopes. These ASYMP epitopes are associated with frequent, robust, and polyfunctional CD8+ TEM cell responses. Moreover, therapeutic immunization of a novel susceptible humanized HLA-A*0201–Tg mouse model with ASYMP epitopes induced a strong CD8+ T cell–dependent protective immunity against recurrent ocular herpes. Furthermore, we report a novel prime/pull therapeutic vaccine strategy that is based on priming CD8+ T cells with multiple ASYMP epitopes, followed by treatment with the CXCL10 T cell–attracting chemokine. This approach significantly boosts the number of functional antiviral CD8+ TEM and TRM cells in TG of HSV-1 latently infected humanized HLA-Tg mice and improves protection against recurrent herpetic disease following UV-B–induced HSV-1 reactivation. These findings have profound implications in the development of T cell–based immunotherapeutic vaccine strategies to treat recurrent herpes infection and disease in man.
Development of CD8+ T cell epitope-based vaccines in the era of omics is taking advantage of new technologies to tackle infectious diseases for which vaccine development has been unsuccessful (12). Cutting-edge technologies and screening strategies have been developed recently to mine genomic sequence information for state-of-the-art rational vaccine design (38). However, the large size of the HSV-1 genome and the low frequency of HSV-specific CD8+ T cells have hampered genome-wide identification of protective HSV-1 CD8+ T cell epitopes and characterization of the full repertoire of CD8+ T cells that are associated with the “natural protective immunity” seen in ASYMP individuals. The HSV-1 genome is a linear dsDNA sequence of ∼152 kb and contains two unique regions called the long unique region (UL) and the short unique region, which together encode 84+ ORFs (42). The recent availability of comprehensive genomic datasets of HSV-1 has shifted the paradigm of herpes vaccine development from virological to sequence-based approaches (42). Our genome-wide screening of the HSV-1 sequence using PBMC-derived T cells from cohorts of SYMP and ASYMP individuals identified several previously unknown CD8+ T cell epitopes that span a wide range of envelope, structural, and tegument HSV-1 proteins. We have characterized the phenotype and function of CD8+ T cells specific to these epitopes in ASYMP humans and in HLA-A*0201–Tg mice. Importantly, the prime/pull therapeutic vaccine strategy that increases the size and function of HSV-specific CD8+ T cells in latently infected TG has been demonstrated for the first time, to our knowledge, in a humanized animal model of recurrent ocular herpes. Thus, our novel vaccination strategy has produced scores of new protective ASYMP CD8+ T cell epitopes, as well as an innovative Ag delivery system that is prime for clinical trial testing.
One of the advantages of ASYMP epitope-based vaccines over protein-based vaccines is the avoidance of SYMP epitopes that might inadvertently drive unwanted immunopathological responses (39). Such immunopathological responses might contribute to exacerbation of disease, as recently found for an HLA-A*0201–restricted CD8+ T cell epitope from the HSV-1 gK protein (43). Thus, identification of human SYMP epitopes from the HSV-1 genome and their elimination from a therapeutic herpes vaccine would be beneficial, because these SYMP epitopes may exacerbate recurrent herpetic disease.
HSV-1 expresses its genes sequentially in four kinetic classes of proteins: the immediate early α (most of which are transcriptional factors), the early β (mostly enzymes), the early late γ1 (most of which are structural proteins), and the true late γ2 (mostly structural proteins and viral particles). The repertoire of HSV-specific CD8+ T cells typically targets epitopes from only a fraction of the viral proteins. Those viral epitopes usually fall into a dominance hierarchy consisting of one or a few dominant epitopes and several other subdominant epitopes. Most studies of HSV-specific CD8+ T cells in animal models, including ours, use C57BL/6 mice (44, 45). A genome-wide approach was recently used to predict 376 potential HSV-1 CD8+ T cell epitopes in HSV-1–infected C57BL/6 mice (46, 47). Functional CD8+ T cells in HSV-1–infected C57BL/6 mice actually target only 19 of the 376 potential epitopes (46, 47). These 19 epitopes identified virtually all or the vast majority of the HSV-specific mouse CD8+ TCR repertoire. The 19 mouse HSV-1 epitopes are derived from only 11 of 84+ HSV-1 proteins (46, 47). Of the four kinetic classes of HSV proteins, no immediate early α gene products are targeted by mouse CD8+ T cells, 11 of 19 epitopes are derived from four early β gene products, 4 of 19 epitopes are derived from one early late γ1 gene product, and the remaining 4 epitopes are from four true late γ2 gene products that require viral DNA synthesis (46, 47). Strikingly, an incredible 50–70% of all HSV-specific CD8+ T cells in infected C57BL/6 mice target a single glycoprotein B–immunodominant epitope: gB498–505. Through the use of a similar genome-wide comprehensive approach in this study, we identified several new human CD8+ T cell epitopes from HSV-1 proteins not previously considered as vaccine candidates. In contrast to mouse CD8+ T cells, human HSV-specific CD8+ T cells targeted each of the four kinetic classes of HSV proteins (i.e., α, β, γ1, and γ2) (48, 49). These targets include epitopes from four immediate early α gene products, four early β gene products, and four early late γ1 gene products, whereas the remaining seven are from γ2 gene products (48, 49). Thus, unlike in C57BL/6 mice, human CD8+ T cell epitopes are derived from viral proteins that are produced before and after viral DNA synthesis. It seems clear that the HSV-specific CD8+ T cell repertoire detected in C57BL/6 mice during acute infection does not reliably represent the human CD8+ T cell repertoire (23).
In the present study, the complexity of the CD8+ T cell responses ranged from 3 to 20 reactive ORF/HLA combinations per ASYMP individual. We found that most immunodominant human epitopes derived from the whole HSV-1 genome, UL9196–204, UL10162–170, UL25367–375, UL25572–580, UL26480–488, UL26221–229, UL43272–280, UL43386–394, UL43148–156, UL43302–310, UL44400–408, and UL44443–451, are from the UL of the HSV genome. Of 12 immunodominant epitopes, 3 epitopes were from the early stage of HSV-1 replication (UL9196–204, UL26480–488, and UL26221–229), whereas 9 epitopes (UL10162–170, UL25367–375, UL25572–580, UL43148–156, UL43272–280, UL43386–394, UL43302–310, UL44400–408, and UL44443–451) were from the late stage of assembly and replication of HSV-1. It has been proposed that the most important T cell epitopes in HSV may be in tegument or other noncapsid proteins (50, 51). Tegument proteins, such as those encoded by UL41, UL46/VP11/12, UL47/VP13/14, UL48/VP16, and UL49, as well as immediate early proteins, including RL2/ICP0 and RS1/ICP4, are major targets for effector T cells (50, 51).
Overall, based on our own results (48), as well as those of other investigators (49), unlike the mouse HSV-specific CD8+ T cell repertoire in the aforementioned C57BL/6 mouse studies, the human HSV-specific CD8+ T cell repertoire appears to target a much wider range (20 of 84+) of HSV proteins. Interestingly, HSV-specific CD8+ T cells from mice and humans are commonly directed against 5 of the 84+ HSV proteins. However, human CD8+ T cell dominance hierarchy is also different from that of mice, because no human CD8+ T cell immunodominant epitope equivalent to HSV gB498–505 has been reported in humans. In contrast, the epitopes recognized by human CD8+ T cells appeared to be evenly distributed among the 20 HSV-1 proteins (48, 49). These differences in the array, dominance, and hierarchy of HSV proteins targeted by human and mouse HSV-specific CD8+ T cells make the relevance of the C57BL/6-HSV model to the human HSV host–pathogen natural combination questionable (23). The distribution and the profile of HSV-specific CD8+ T cell responses detected in the C57BL/6-HSV mouse model appear to be “artificially” uneven (50). This unusual distribution of CD8+ T cell responses does not reflect the CD8+ T cell profile of responses that we (48) and other investigators (49) have detected in the human HSV host–pathogen natural combination.
After primary infection of epithelial surfaces, the alphaherpesviruses, such as HSV-1, often establish latent infection in the sensory neurons of their natural hosts (33, 53). During this process, antiviral memory CD8+ T cells develop and coexist for years along with latent HSV-1 (26, 33, 53). Sufficient numbers of functional memory CD8+ T cells in TG of ASYMP individuals are critical for suppressing (or aborting) virus reactivation from latency (33). However, the apparent feeble numbers of existing memory CD8+ T cells in TG of SYMP individuals might not be sufficient to eliminate virus reactivation and would result in the observed repetitive SYMP shedding and recurrent herpetic disease (18, 29). Therapeutic vaccines can be designed to boost the number and function of existing memory CD8+ T cell subpopulations in TG for increased protection against virus reactivation and, hence, eliminate or reduce reinfection of the cornea and recurrent herpetic disease. Boosting the apparent feeble numbers of functional memory CD8+ TEM cells that we detected in SYMP individuals may also contribute to reducing the HSV-1 reservoir in latently infected TG. Consequently, this would actively suppress or permanently silence virus reactivation that occurs in latently infected sensory neurons of TG, thus stopping or reducing recurrent disease. However, the relative contribution of tissue-resident, peripheral, and recirculating memory CD8+ T cells (i.e., TCM, TRM, TRM cells) in protection against recurrent herpes infection and disease remains to be fully elucidated. In the present preclinical study, we demonstrate that a novel prime/pull therapeutic vaccine strategy, consisting of priming antiviral CD8+ T cells with ASYMP human epitopes, followed by neurotropic CXCL10 treatment, can bolster the number and broaden the repertoire of functional antiviral CD8+ TEM and TRM cells, but not CD8+ TCM cells, in TG of latently infected HLA-A*0201–Tg mice. This boost was associated with a significant reduction in viral shedding and recurrent ocular herpetic disease following UV-B–induced reactivation. This finding suggests that TRM cells and recirculating TEM cells contribute greatly to protection against recurrent herpes infection and disease and that CD8+ TEM cells in latently infected TG can be replenished from an apparent recruitment from the blood, provided that appropriate T cell–attracting chemokines, such as CXCL10, are delivered locally into TG. Additionally, TEM and TRM cells likely expand locally following encounter with reactivated virus and viral Ag in sensory neurons. These findings further confirm our hypothesis that the HSV-1 latency/reactivation cycle occurring in sensory neurons might be eliminated or permanently silenced if sufficient numbers of functional HSV-specific CD8+ TEM and TRM cells can be mobilized into latently infected TG by a prime/pull therapeutic vaccine.
In the current study, frequent IFN-γ+CD107a/b+CD44highCD62LlowCD8+ TEM cells specific to HSV-1 genome–derived epitopes were detected in ASYMP individuals. In contrast, CD8+ T cells in SYMP individuals consisted predominantly of CD44highCD62LhighCD8+ TCM cells. This is consistent with our previous findings of significantly higher proportions of CD8+ TEM cells specific to epitopes from gB (31), from VP11/12 (15), and from VP13/14 (1) in ASYMP individuals. Although we are aware that information gained from PBMC-derived T cells may not be completely reflective of TG-resident CD8+ T cells, our investigations were limited to human PBMC-derived CD8+ T cells because of the ethical and practical challenges in obtaining human TG-resident CD8+ T cells. Nevertheless, our human and humanized HLA Tg mouse results converge into suggesting that ASYMP individuals mount faster and stronger polyfunctional HSV-specific CD8+ TEM and CD8+ TRM cell responses that allow for a fast and better clearance of herpes reactivation and recurrent disease.
The inadequacy of animal models of primary herpes infection and immunity have made it challenging to explore the immune mechanisms that lead to protection against recurrent herpetic disease (17). A critical question is which animal model would be the most appropriate to mimic the immunopathological aspects of recurrent herpetic disease as occurs in humans? Mice have been the animal models of choice for most herpes immunologists, and the results from mouse models have yielded tremendous insights into the protective mechanisms during primary acute infection (reviewed in Ref. 19). Characterization of the phenotype and function of protective HSV-specific memory CD8+ T cells in mice have been largely limited to T cells specific to the immunodominant HSV-1 gB495–505 epitope studied during acute infection (34, 46). Unlike humans, spontaneous HSV-1 reactivation in latently infected mice and virus shedding in tear film either does not occur at all or occurs at very low levels in mice (54). Thus, despite evidence that the majority of clinical recurrent HSK is due to reactivation (42, 55), most preclinical animal studies investigating the mechanisms that orchestrate recurrent HSK have used primary acute infection of mice (56, 57). Only a handful of studies have employed the mouse model of UV-B light induced recurrent herpetic corneal disease, mostly using C57BL/6 and BALB/c mice (58–60). Moreover, considering the wealth of data addressing the protective mechanisms of CD8+ T cells specific to mouse HSV-1 epitopes (mostly in acutely infected C57BL/6), it is surprising how few reports exist exploring the protective mechanisms of CD8+ T cells specific to human HSV-1 epitopes (58, 60). The present study validated the UV-B light–induced recurrent herpetic corneal disease in the humanized HLA-Tg mouse model of ocular herpes (4, 29). The study showed that shedding of reactivated virus in tears and recurrent corneal HSV can be induced by UV-B exposure of latently infected HLA-Tg mice. Moreover, our humanized HLA-Tg mice express the human HLA-A*0201 molecule, instead of mouse H2b or H2d MHC molecules. Thus, they develop “human-like” CD8+ T cell responses to HLA-restricted epitopes (29). In addition, after UV-B light–induced reactivation, high numbers of functional CD8+ TRM and TEM cells were detected in TG and were associated with protection against induced recurrent herpetic disease. In contrast, high numbers of exhausted CD8+ T cells in TG are associated with severe recurrent disease in these mice. Until recently, it was not possible to directly evaluate therapeutic vaccine efficacy of human HLA–restricted CD8+ T cell epitopes in a small animal model of recurrent ocular herpes. In our opinion, the HLA-Tg mice combined with UV-B light–induced recurrent disease is arguably the best available small animal model to study the role of HLA-restricted CD8+ T cells specific to human HSV-1 epitopes in protection against virus shedding and recurrent herpetic disease. Ongoing studies in our laboratory are using the HLA-Tg mouse model of UV-B light–induced recurrent disease to investigate the dynamics of TG- and cornea-resident antiviral CD8+ T cells that may lead to SYMP versus ASYMP infections.
Following acute infection, three major HSV-specific memory CD8+ T cell subpopulations (TCM, TEM, and TRM) develop, infiltrate, and sequester in infected TG in response to a high level of T cell–attracting CXCL9, CXCL10, and CCL5 chemokines and other factors early during acute infection (1, 15, 31). Memory CD8+ T cell infiltrates were reported in latently infected TG of human cadavers (61). Functional CD8+ T cells in these infiltrates likely help to decrease viral shedding by preventing virus reactivation from latency (18, 29). However, the apparent low level of T cell–attracting CXCL9, CXCL10, and CCL5 chemokines in latently infected TG may not be sufficient to control sequestration of CD8+ TRM cells or to guide CD8+ TEM and TCM cells from the circulation into latently infected TG. Thus, during latency, TG appear to be more immunologically restrictive and are not open to homing CD8+ TEM and TCM cells that would migrate from the circulation into infected TG (1). In this study, we found that local delivery of the T cell–attracting CXCL10 chemokine, but not of CXCL9 or CCL5 chemokines, dramatically increased the number and function of HSV-specific CD8+ TEM and TRM cell subpopulations in latently infected TG, improving protection against recurrent ocular herpes infection and disease following UV-B light–induced reactivation from latently infected TG of HLA-A*0201–Tg mice. The effect of CXCL10 treatment on activation and expansion of CD8+ T cells in TG occurs only in mice that were exposed to UV-B light reactivation, indicating that local exposure to HSV-1 Ags is crucial. These results suggest that the number and/or the function of CD8+ T cells specific for ASYMP epitopes is suppressed in latently infected human TG, that local delivery of the T cell–attracting CXCL10 chemokine will expand the repertoire and function of HSV-specific CD8+ T cells and reduce the likelihood of viral reactivation from latency, and that a prime/pull vaccine strategy that increases the number and/or function of CD8+ T cells in latently infected TG over a certain threshold would likely lead to protection in humans. The success of this innovative prime/pull therapeutic vaccine is likely due to the expansion and survival characteristics of CD8+ T cell precursors, as well as the increase in the number and function of other immune cells in TG, including CD4+ T cells and APCs, following CXCL10 treatment. This contributes to homing of CD8+ TEM cell subpopulations in latently infected TG, as well as to increased expansion and survival of the local CD8+ TRM cells that already exist within the TG.
In this study, we detected high frequencies of HLA-A*0201–restricted CD8+ T cells specific to HSV-1 epitopes. These high frequencies are surprising because the tetramer analyses were performed ex vivo and because the frequencies of T cells binding each tetramer were expected to account for a few percentages of the total cells. If all of the tetramer-specific CD8+ T cells were added together, then large percentages of the entire CD8+ T cell repertoire would be accounted for by HSV-1 epitopes that are binding to HLA-A2 molecules. Given that there are other immunodominant viruses (e.g., EBV), as well as most likely other HLA restrictions, the percentages measured in this study should be interpreted with caution. It is possible that cross-reactivity with other epitopes within or outside the herpes family might also account for the high frequencies of at least some epitopes. Broadly reactive human CD8+ T cells that recognize an epitope conserved among varicella zoster virus, HSV, and EBV have been recently reported by Chiu et al. (53). An extensive CD4+ and CD8+ T cell cross-reactivity between alphaherpesviruses has also been recently reported by Jing et al. (26). Cross-reactivity between seemingly unrelated epitopes has also been reported by Welsh, Selin, and colleagues (62, 63). Moreover, compared with the frequencies of CD8+ T cells specific to HSV-2 epitopes detected from genital herpes patients (49, 64), we have consistently detected higher frequencies of HLA-A*0201–restricted CD8+ T cells specific to HSV-1 epitopes from ocular herpes patients (1, 6, 15, 29, 31). At least some of these differences could also be due to differences in cohorts of ocular and genital herpes patients. A side-by-side comparison of the frequencies of CD8+ T cells specific to HSV-1 and HSV-2 epitopes from ocular and genital herpes patients is warranted and will be the subject of a future study.
The AAV8 vector was not directly involved in reduced recurrent infection and disease seen in the AAV8-CXCL10 prime/pull vaccine because no protection was observed in mice that received empty AAV8 vector alone. The AAV8 vector used in our prime/pull vaccine has rapidly moved to the forefront of human therapies in the past few years (65). Several naturally occurring, tissue-specific AAV serotypes have been isolated (65). We chose AAV8 because it is a neurotropic virus with the potential of persistent transgene expression in neurons (66, 67), it can superinfect HSV latently infected neurons (66), it can accommodate up to 4.7 kb of DNA, and it is nonpathogenic. Clinical trials using AAV vectors showed only transient inflammation while demonstrating clinical benefits (65). No side effects have been observed in HSV-1–infected mice during the 30 d of monitoring following treatment with the neuroinvasive AAV8 vector. However, long-term monitoring of the corneas and TG for pathology associated with AAV8 will be necessary to ensure the safety of the vectors for eventual use in patients with herpes keratitis.
In conclusion, the current study identifies previously unreported protective ASYMP epitopes that are potentially useful if included in a therapeutic herpes simplex vaccine, characterizes the phenotype and the function of the protective CD8+ T cell subpopulations associated with immunologic control of recurrent herpetic disease, and demonstrates that bolstering the number of functional HSV-specific CD8+ TEM and TRM cells in latently infected TG through HSV-1 human epitopes/CXCL10-based prime/pull therapeutic vaccine protected against virus shedding and reduced recurrent herpetic disease. The results from this preclinical study should pave the way toward developing a novel clinical T cell–based vaccine immunotherapy against recurrent ocular herpes.
We thank Dale Long from the National Institutes of Health Tetramer Facility (Emory University, Atlanta, GA) for providing the tetramers used in this study, Barbara Bodenhoefer (University of California Irvine’s Institute for Clinical and Translational Science) for helping with blood collections from HSV-1–seropositive SYMP and ASYMP individuals, and Wen-Pin Chen for helping with the statistics. This work is dedicated to the memory of Professor Steven L. Wechsler (1948–2016), whose numerous pioneering works on herpes latency laid the foundation for this line of research.
This work was supported by Public Health Service Research Grants R01 EY026103, R01 EY019896, and R01 EY024618 from the National Eye Institute, Grant R01 AI110902 from the National Institute of Allergy and Infectious Diseases, The Discovery Center for Eye Research, and a Research to Prevent Blindness grant.
The online version of this article contains supplemental material.
Abbreviations used in this article:
herpes stromal keratitis
HSV type 1
HSV type 2
mean fluorescence intensity
open reading frame
central memory T
effector memory T
tissue-resident memory T
long unique region.
The authors have no financial conflicts of interest.