Extensive CD4 and CD8 T-cell cross-reactivity between alphaherpesviruses

Abstract

The alphaherpesvirinae subfamily includes HSV types 1 and 2 and the sequence-divergent pathogen varicella zoster virus (VZV). T cells, controlled by TCR and HLA molecules that tolerate limited epitope amino acid variation, might cross-react between these microbes. We show that memory PBMC expansion with either HSV or VZV enriches for CD4 T cell lines that recognize the other agent at the whole virus, protein, and peptide levels, consistent with bi-directional cross-reactivity. HSV-specific CD4 T cells recovered from HSV seronegative persons can be partially explained by such VZV cross-reactivity. HSV-1-reactive CD8 T cells also cross-react with VZV-infected cells, full-length VZV proteins, and VZV peptides, and kill VZV-infected dermal fibroblasts. Mono- and cross-reactive CD8 T cells use distinct TCRB CDR3 sequences. Cross-reactivity to VZV is reconstituted by cloning and expressing TCRA/TCRB receptors from T-cells that are initially isolated using HSV reagents. Overall, we define 13 novel CD4 and CD8 HSV-VZV cross-reactive epitopes and strongly imply additional cross-reactive peptide sets. Viral proteins can harbor both CD4 and CD8 HSV/VZV cross-reactive epitopes. Quantitative estimates of HSV/VZV cross-reactivity for both CD4 and CD8 T cells vary from 10-50%. Based on these findings, we hypothesize host herpesvirus immune history may influence the pathogenesis and clinical outcome of subsequent infections or vaccinations for related pathogens, and that cross-reactive epitopes and TCRs may be useful for multi-alphaherpesvirus vaccine design and adoptive cellular therapy.

Introduction

The epidemiology of infections with members of the Alphaherpesvirinae subfamily is geographically and temporally complex, showing variation between regions and over time. Close to 100% of the US population are seropositive for VZV due to infection or vaccination. Since commencement of universal vaccination with attenuated VZV in 1995 (1) the relative proportion of persons with natural and vaccine-induced VZV immunity is shifting, with uncertain consequences for VZV transmission and recurrence. The age-specific incidence of recurrent varicella infection (zoster) is increasing in the US (2). Pediatric varicella vaccination is not practiced in most countries, where primary varicella remains ubiquitous (1). Seronegative adults remain susceptible to primary varicella and curiously, VZV seropositivity amongst adults is considerably under 100% in some areas near the equator (3). Conversely, herpes simplex virus seroprevalence is higher in some equatorial regions (4) than in the US. Amongst US adults aged 14-45, 50% are infected with HSV-1 and 16% with HSV-2. As with VZV, HSV infection and resulting seroconversion are thought to be permanent due to latent infection of neural ganglia. Modest decreases have occurred in the age-specific prevalence of HSV-1 over recent decades (5). Reflecting this, more individuals are commencing sexual activity while seronegative for HSV-1. Indeed, HSV-1 accounts for the majority of clinical first episode genital herpes both in the US (6).

The immune boost hypothesis of Hope-Simpson suggests that periodic re-exposure to wild-type VZV stimulates favorable immune responses that inhibit zoster. These antigenic encounters may be decreasing as an unintended consequence of pediatric vaccination (7, 8). However, the causal link between varicella vaccination and zoster is controversial (9). The relative order of acquisition of immunity to HSV-1 and VZV is likely heterogeneous within populations. Varicella vaccine, where used, is recommended at 12 to 15 months of age. HSV-1 seroprevalence also rapidly increases during the first few years of life. Overall, infection and vaccination patterns with HSV-1, HSV-2, and VZV vary with location and age group and are changing dynamically within communities, creating a complex pattern within which diverse immune interactions may operate to modulate the clinical manifestations of these infections.

Given that HSV and VZV have 65 homologous genes (10), it is rational that immunity related to VZV infection or vaccination could exert heterologous effects on HSV-1 or HSV-2 infection, and vice versa. Boosting of antibody levels to HSV by VZV infection, and the reciprocal, occur in primary and recurrent infection (11-13), but far less is known about T-cell responses. Our group has observed T-cell reactivity to HSV in HSV-1/HSV-2 seronegative persons. This could be due to VZV cross-reactivity, albeit a limited number of HSV-2-reactive CD4 clones reactive did not exhibit this property (14, 15). This report focuses on T-cell cross-reactivity to structurally-related, sequence-homologous peptides. More broadly, T-cell cross reactivity includes recognition of unrelated peptides, in the context of either the index or unrelated MHC molecules, and is now thought to underlie minor histocompatibility antigen graft rejection, HLA-linked drug hypersensitivity, and possibly heterologous immunity effects between unrelated organisms. The T-cell repertoire seems to be less diverse than the non-self peptide set, requiring ubiquitous cross-reactivity to minimize gaps in non-self recognition.

Zoster is the target of the only licensed therapeutic vaccine. This attenuated varicella strain modestly boosts serum antibody and VZV-specific CD4 T-cells (16). It is thought to work via T-cells, as shingles risk correlates with HLA-region single nucleotide polymorphisms (17), and with age-related declines in VZV-specific CD4 T cells (18). The apparent correlation of efficacy with antibody responses could reflect CD4 T-cell help (19). A recent study found only 1 in 8 (12.5%) persons had a vaccine-related CD8 T-cell boost (20). Indeed, very little is known overall about the CD8 response to VZV. CD8 T-cell lysis of VZV-infected cells has never been demonstrated. Concerns about the live vaccine’s safety in immune compromised persons and its limited, 50-70% efficacy (21, 22), and the identification of CD4 T-cell responses to VZV glycoprotein E (open reading frame 68 =ORF68 product) have motivated development of an ORF68 subunit vaccine in a novel adjuvant platform. This vaccine stimulates high CD4/antibody responses and appears very efficacious (23). Cross-reactivity between gE or other ORFs shared between VZV and HSV might inform extension of this subunit vaccine format to HSV immunotherapy.

We have developed proteome-wide tools to study CD8 and CD4 T-cell responses to every protein in each human Alphaherpesvirinae. Responder cells are prepared by enriching and expanding memory virus-specific polyclonal T-cell lines that contain multiple epitope-specific populations. Herein, we extend previous virus-specific reports in which we discovered many viral T-cell antigens and epitopes (16, 24-26) to delineate and measure T-cell cross-reactivity across the Alphaherpesvinae subfamily.

Materials and Methods

Subjects and specimens

Participants were healthy adults. Some were studied previously (Supplemental Table I). HSV-1 immunity was studied in participants 21 and 23 (25, 27), zoster vaccine immunity in participants 1, 2, 3, 4, and 26 (16), T-cell responses to HSV-2 in subjects 30-34 (15), and drug reactions in subject 35 (28). Subjects 6-19 and 26-29 were genetically confirmed HSV-1-infected monozygotic twins (29, 30), with consecutively numbers denoting twin pairs. PBMC were cryopreserved. VZV (16) and HSV-1 and HSV-2 IgG were detected as reported (16, 31).

Viruses and antigens

Cell lines were Mycoplasma-negative. VZV isolated from zoster was expanded in human embryonic tonsil (HET). Trypsin-recovered were cryopreserved for a cell-associated stock of wild-type VZV. VZV strain OKA from Zostavax (Merck, Kenilworth, NJ) was cryopreserved as infected cells and used to make UV-cell-associated whole VZV antigen (16). For PCR template for truncation analyses, VZV DNA was prepared as described (25). HSV-1 17+ and HSV-2 186 antigens and mock Vero (for HSV) and HET antigens were similarly prepared (32). HSV-1, HSV-2, and VZV ORFeome-covering proteins are described (16, 24, 26). For CD4 epitope mapping, regions of VZV ORF24 and ORF68 were cloned into pDEST203 (25) (primers available on request). Proteins were expressed by in vitro transcription translation (IVTT) (16). For CD8 assays, VZV (16), HSV-1 (25), and HSV-2 (24) ORFs in pDONR207 or pDONR221 (Invitrogen, Carlsbad, CA) were moved to pDEST103 (25). In-frame fusion to green fluorescent protein was sequence-confirmed. New or previously obtained (15, 16, 33) peptides (Sigma, St. Louis, MO; Genscript, Piscataway, NJ) were dissolved in DMSO. Peptide/primers were from Genbank HSV-1: JN555585.1; HSV-2: JN561323.1; VZV: NC_001348.1. HLA A*0201-restricted peptide recognized by CMV-specific TCR was NLVPMVATV (34).

Virus-reactive CD4 T-cells

Stimulation of PBMC with whole VZV and expansion with IL-2 to create polyclonal VZV-reactive cell lines has been described (16). For some subjects, we stimulated PBMC 18 hours with whole UV-treated VZV (1:100). Live CD3+, CD4+ lymphocytes were sorted into CD137^low and CD137^high populations, expanded once with phytohemagglutinin (PHA) and then once with anti-CD3 mAb (25). HSV-1-reactive polyclonal CD4 T-cell lines were created using CD137 (25). CD4 T-cell clones from persons who are seronegative for HSV-1 and HSV-2 have been described (14, 15).

Virus-reactive CD8 T-cells

HSV-1-reactive polyclonal CD8 T-cell lines were created as described (25). Briefly, HSV-1-infected HeLa were added to myeloid dendritic cells (DC), which were co-incubated with autologous CD8+ PBMC. After 20 hours, CD3+CD8+CD137^high cells were sorted and polyclonally expanded. For HSV-2, a similar workflow used a HSV-2 UL41 deletion mutant. For polyclonal peptide-specific CD8 lines or clones, PBMC or the polyclonal CD8 lines were stained with phycoerythrin (PE)-labeled tetramers of HLA A*0201 and either HSV-1 UL25 372-380 or HSV-1 UL40 184-192 (35). CD8+tetramer^high cells were either expanded polyclonally or plated in limited dilution and expanded with PHA and then anti-CD3 (25).

TCR sequencing and expression

V-CDR3 regions of TCRA and TCRB mRNA from T-cell clones were sequenced using 5′ RACE (Clontech, Mountain View, CA) (36). TCR expression plasmid was created by placing TCRA/TCRB V-CDR3 (Genbank KT955849/KT955850) into pMP71flex (34). An extracellular murine TCRB-C epitope allows detection of transduced cells with anti-TCRB (H57-597, eBioscience, San Diego, CA). TCR cassettes were inserted into pRRL.PPT.MP.GFPpre (37). Lentiviral stocks were created by co-transection of HEK293 and used to transduce anti-CD3/anti-CD28 (Dynal, Grand Island, NY)-stimulated CD8 PBMC, which were expanded using anti-CD3 (38). For bulk CDR3 sequencing, 10⁶ polyclonal HSV-2-reactive CD8 T-cells or 3 × 10³ tetramer-sorted cells were submitted to Adaptive Biotechnologies (Seattle, WA) for TCRB clonoSEQ^™ (39).

CD4 T-cell assays

PBMC were stimulated with UV-VZV antigen, media or mock control and analyzed for CD4 T-cell expression of IFN-γ and IL-2 (40). To test expanded CD4 T-cell lines, equal numbers of responders and CFSE-labeled autologous PBMC used as antigen-presenting cells (APC) were added to test stimuli for 18 hours (16). Live, CFSE-negative, CD3+,CD4+ cells were analyzed for IFN-γ and IL-2 (25). Analyses used FACSCanto II (Becton Dickinson, Franklin Lakes, NJ) and Flowjo (Treestar, Ashland, OR). Proteome-spanning proliferation assays admixed 5-10 × 10⁴ polyclonal expanded CD4 T-cells with antigen and an equal number of γ-irradiated autologous PBMC in 96-well U bottom plates (duplicate for recombinant antigens, triplicate for whole viral antigens) (35), measuring incorporation of ³H thymidine at 72 hours as counts per minute (CPM) (32). Results are expressed as net cpm = mean cpm for experimental antigen minus mean cpm for negative control antigen, or as stimulation index (SI) = mean cpm for experimental antigen divided by mean com for control antigen. Antigens were UV-treated VZV, HSV-1, or HSV-2 strain 186 or mock preparations (1:100), PHA (1.6 μg/ml, Remel, Kansas City, KA), or IVTT proteins (1:1,000-1:2,000). Negative controls were 27 IVTT-expressed P. falciparum proteins to set cutoffs (16). Other assays used 1 μg/ml peptide in triplicate with 0.1% DMSO control. For studies of HSV-2-reactive CD4 T-cell clones from HSV-1/HSV-2 seronegative persons (14, 15), ORF-level specificity was assigned using responder clones seeded at 2 × 10⁴ cells/well in duplicate with 1 × 10⁵ autologous, gamma-irradiated PBMC/well as APC, and each HSV-2 protein (24) at 1:2000 (24) an ³H thymidine readout at 72 hours. Alternatively, HSV-2 peptide specificity was determined using a peptide set covering a portion of the HSV-2 proteome (15). Subsequent assays to study VZV cross-reactivity used responder cells and APC as above with whole HSV-2 and VZV antigens, selected IVTT-synthesized HSV-2 and VZV proteins, or 1 μg/ml peptide, with relevant negative controls.

CD8 T-cell assays

For whole virus, killing assays used dermal fibroblasts (DFB) infected by co-culture (3:1) with VZV-infected HET. After several co-culture passages (3:1) of uninfected DFB (UN-DFB) with infected DFB (IN-DFB), IN-DFB were used at >50% cytopathic effect (CPE) typically 3 days after passage. eGFP-VZV showed fluorescent viral infection well outside of visible plaques. Killing was measured by VITAL-FR (41) using UN-DFB and IN-DFB labeled with 5 μM FarRed and 10 μM CFSE (Invitrogen), respectively. DFB (infected or uninfected into separate wells) were re-plated in 24-well plates in triplicate (5 × 10^5/well) and HSV-1-reactive CD8 T-cell effectors (1 × 10⁶/well) added 4 hours later. After 24 hours, the trypsin-recovered contents of one uninfected and one infected DFB well were paired, mixed, fixed, and dye-labelled cells counted by flow cytometry. For each tube the ratio R between the percent IN-DFB (green) and the percent UN-DFB (red) was calculated. Killing was calculated as (R (no T-cells added) – R (T cells added))/ R (no T-cells added). For each effector:target combination, we report the mean and standard deviation of three calculated values derived from 6 pairs/12 wells. To test cytokine synthesis, HSV-1-reactive CD8 T-cells effectors were co-incubated with autologous or HLA-mismatched EBV-LCL or DFB. LCL were infected with HSV-1 strain E-115 at MOI 10 for 24 hours and DFB were infected with vOKA as above. APC and responder cells (2.5 × 10⁵/well each) in 48 well plates DFB or 96 well U bottom wells for EBV-LCL. Cells were stained with Violet live/dead and antibodies to CD8, CD3, IL-2 and IFN-γ, fixed and analyzed by flow cytometry (25).

To test processing/presentation of full-length VZV ORFs, Cos-7 cells were co-transfected with HLA cDNA and HSV-1, HSV-2, or VZV ORFs cloned into pDEST103 to create artificial APC, co-incubated with CD8 T-cell lines, and assayed for IFN-γ secretion (25). In a modification used for antigen discovery, polyclonal HSV-2-reactive CD8 T-cells were screened with Cos-7 co-transfected with HLA B*1502 cDNA and each HSV-2 ORF as described (25).

To test recognition of peptides, polyclonally-expanded CD8 T-cell lines (above) were incubated with autologous LCL and 1 μg/ml peptide and processed for intracellular ICS IFN-γ accumulation (25). To screen candidate peptide-reactive CD8 T-cell, responders (1/4 of a clonal microculture, or 5 × 10⁴ polyclonal cells) were co-incubated 48 hours with an equal number of HLA-appropriate EBV-LCL and 1 μg/ml peptide followed by supernatant IFN-γ ELISA. Tests of cloned TCRs used 1 × 10⁵ each lentivirally-transduced CD8 T-cells and EBV-LCL/well and 1 μg/ml peptide.

Statistics

For ³H assays with ORFeome panels, a 1% false positivity (FP) threshold was set as CPM > median plus 2.3*median absolute deviation (MAD). The MAD = 1.4826*median(abs(qi-mq)), where qi is the log₁₀ for each control well, and mq is the median of control quantities. The threshold for positivity is mq + Z*MAD, where Z is the standard quantile such the probability of exceeding abs(Z) is the false positive rate: p(qi > mq + Z*MAD)=p(FP) (42, 43). The control quantities were derived from a panel of 27 P. falciparum proteins run in duplicate parallel with the viral ORFeomes within each assay. T cell clonality indices were estimated as described (44, 45). CD4 T-cell cross-reactivity to HSV-1 and VZV ORFs were analyzed using InStat 3.10 (GraphPad Software, La Jolla, CA).

Human subjects

Investigations conformed to Declaration of Helsinki principles and were approved by the relevant institutional review boards. Informed written consent was received from participants prior to inclusion in the study.

Results

Cross-reactivity to whole virus amongst polyclonal alphaherpesvirus-reactive CD4 T-cells

The abundance of VZV-reactive CD4 T-cells in blood is < 0.2%, even after boosting with attenuated VZV, while HSV-specific CD4 T-cells are more abundant, in the 0.2%-3.0% range (16, 40). Possibly, this reflects more frequent HSV reactivation. We created VZV-reactive CD4 T-cell lines before and after VZV vaccination (16). These cells showed brisk IFN-γ/IL-2 responses to HSV-1 in addition to VZV whether obtained before or after VZV vaccination, and regardless of the presence or absence of HSV infection (Fig. 1). The proportion of VZV-reactive CD4 T-cells that showed cross-reactivity towards HSV-1 ranged from 18 to 100% as judged by the ratio between reactivity to the two whole virus preparations. To explore reciprocal cross-reactivity, we created HSV-1-reactive CD4 T-cell lines from 7 pairs of HSV-1-seropositive, HSV-2-seronegative monozygotic twins (Supplemental Table I lists participant characteristics) using HSV-1 re-stimulation and CD137 selection (25). These cell lines had positive HSV-1 stimulation indices (SI > 5.0 using ³H thymidine proliferation assays) for 13 of 14 donors (93%). Amongst these, 12 of 13 (92%) were also positive for VZV, with a good correlation for mean CPM values between viruses (R² = 0.48). We also compared SI and delta CPM values for HSV-1 and VZV whole viral antigens as very rough estimates of CD4 T-cell cross-reactivity. The mean value of the stimulation index for VZV was 42% of that for HSV-1, while the mean delta CPM value for VZV was 47% of that for HSV-1.

Fig 1.

CD4 T-cell cross-recognition of whole HSV-1. VZV-driven polyclonal CD4 T-cell responder lines were created from PBMC before (day 0) of 28 days after vOKA vaccination from persons of defined HSV serostatus (far left). The indicated antigens and PBMC as APC were used to detect cytokine responses by ICS. Gating is on live, CD3+ CD4+ cells. Numbers are percent gated live CD3+CD4+ responders with cytokine expression. At right, net responder cells expressing either IFN-γ or IL-2 or both for VZV compared to mock HET, or HSV-1 compared to mock Vero, are shown the proportion of HSV-1 responder cells compared to VZV responder cells. Day 0 in blue, day 28 in red.

CD4 T-cell cross-reactivity to alphaherpesvirus proteins

After showing cross-reactivity to whole viral antigens, we then tested polyclonal virus-reactive T-cells for responses to proteome-covering recombinant protein sets. Polyclonal CD4 HSV-1- or VZV-reactive cell lines were obtained with or without using activation marker CD137 expression (25, 27). From a HSV-1/HSV-2/VZV triple-seropositive participant (subject 5), VZV-reactive CD4 T-cells were rare (0.1-0.2%) ex vivo by cytokine and CD137 analyses (Fig. 2A). However, bulk-expanded CD3+, CD4+, CD137^high cells were about 100-fold enriched for VZV reactivity (Fig. 2B).

Fig. 2.

A: Direct ex vivo PBMC cytokine responses (left) to VZV and mock viral antigens gated on live CD3+ CD4+ cells (not shown), and expression of CD137 18 hours after stimulation gated on live CD3+ cells. B: responses of sorted, expanded cells sorted for CD137 expression. Numbers in quadrants are percent of gated live CD3+CD4+ cells. C: Proliferative responses of bulk CD137^high-origin UV-VZV-reactive CD4 T-cell line from subject 5 to whole virus and control antigens, and to each VZV protein. Top 8 VZV proteins indicated; O= ORF. Reactivity of same responders to each HSV-1/HSV-2 protein are also shown. The homologs of the top 8 VZV reactive proteins are color-coded. D: Proliferative responses of UV-HSV-1-reactive polylclonal CD4 T-cell lines from 18 participants to each HSV-1 and VZV protein. Proteins with homologs at top, followed by those found only in HSV-1 in the middle and only in VZV at bottom.

The alphaherpesvirus protein sets (16, 24, 25) include 77 HSV-1 and HSV-2 and 70 VZV ORFs. They drive CD4 T-cell proliferative and cytokine responses at 1:1,000-1:36,000 dilutions with autologous PBMC as APC (16, 24-27). Polyclonal VZV-driven CD137^high CD4 T-cells show proliferation to many VZV proteins (Fig. 2C). Amongst the eight VZV proteins stimulating the strongest proliferative responses in this subject, responses to five (HSV UL40, homolog of VZV ORF18, US8, homolog of VZV ORF68, UL39, homolog of VZV ORF 19, UL29, homolog of VZV ORF 29, and UL23, homolog of VZV ORF 36) were also detected (Fig. 2C). Additional cross-reactive and VZV-only responses were noted (not labeled in Fig. 2C to preserve readability). In contrast, while responses to VZV ORFs 8, 50 and 67 were positive, neither the HSV-1 nor the HSV-2 homologs stimulated proliferation. No instances of reactivity to HSV-1 or HSV-2 proteins were noted for HSV proteins that were homologs of non-reactive VZV proteins. Overall, amongst the 30/70 (40%) VZV proteins that stimulated CD4 responses for this donor, HSV-1 cross-reactivity was noted for 8 proteins (27%).

Reciprocal cross-reactivity, by HSV-1-reactive CD4 T-cells towards their VZV homolog, was measured in HSV-1-reactive polyclonal cell lines enriched from monozygotic twins (n=18) (subjects 6-19 and 26-29, Supplemental Table I) (Fig. 2D). In aggregate, amongst the 18 donors and 65 proteins with homologs in HSV-1 and VZV, HSV-driven cell lines recognized HSV-1 proteins on 322 occasions. Amongst these, cross-reactivity to both the HSV-1 and VZV protein was noted 33 times (Fig. 2D). Thus, 10.2% of the antigen-level CD4 responses to HSV-1 in this population were cross-reactive towards VZV. Amongst the 65 proteins with homologs in HSV-1 and VZV (upper portion of Fig. 2D), 13 (20%) showed shared recognition for both the HSV-1 and VZV homologs for at least one participant. There was an apparent relationship between the breadth of potentially cross-reactive HSV-1 antigens recognized, and the presence of VZV cross-reactivity. Amongst the 14 persons with cross-reactivity to at least one VZV protein, the mean and standard deviation number of HSV-1 ORFs recognized was 14 ± 8.3, while amongst the 4 persons without a positive response to a discrete VZV protein, the values were 9.2 ± 3.1, with a two-tailed p value of 0.025 using the two-tailed Mann-Whitney test. However, amongst the 14 persons with CD4 cross-reactivity to at least one VZV protein, there appeared not to be a correlation between the number of potentially cross-reactive HSV-1 proteins recognized and the number of VZV proteins recognized (Spearman rank correlation, r=0.06, p = 0.83).

These cross-reactive proteins, using HSV-1 nomenclature, included enzymes and factors involved in DNA replication (UL5, UL9, UL12, UL29, UL40), structural capsid (UL19) and tegument (UL21, UL46, UL47, UL49) proteins, and two proteins involved in nuclear egress (UL31 and UL34). The protein homolog set with the most prevalent cross-reactive recognition was HSV UL40/VZV ORF18, the small subunit of ribonucleotide reductase, for which we detected cross-reactivity amongst 8 of 18 subjects (44%). Cross-reactivity in HSV-1 ICP4, an essential repressor/transactivator encoded by gene RS1 in HSV and ORF62 in VZV, was detected in 7 of 18 subjects.

Five VZV ORFs (ORF1, ORF2, ORF13, ORF32, ORF57) have no known HSV homolog (10, 46), while 9 HSV ORFs (γ34.5, UL45, US2, US4, US5, US6, US8.5, US11, US12) have no known VZV homolog. Amongst these 9 (mid-lower portion of Fig. 2D), plentiful reactivity to some HSV-1 proteins was detected, particularly for the HSV vaccine candidate glycoprotein D (15 of 18 subjects, gene US6, lower portion of Fig. 2D) (47). There were no instances of cross-reactivity to any of the five VZV proteins that do not have HSV-1 homologs (bottom of Fig. 2D) in any donor.

CD4 T-cell cross-reactive alphaherpesvirus epitopes

We validated selected hits from the ORFeome scans to the peptide level. Our goals were to the contributions of sequence-identical regions, as in the one known HSV/VZV cross-reactive CD4 epitope (48), with divergent but sequence-related homologous regions, and to compare the functional avidities of the T-cells for VZV and HSV peptide homologs. We used polyclonal CD4 CD137^high-origin responder cells (Fig. 2B) or UV VZV/IL-2 driven polyclonal cell lines from subjects 2 and 25 (Supplemental Table I) from our previous vaccine study (16) as responders. Cross-reactive epitopes were defined by two strategies. The first used existing (25, 33) HSV-1 and HSV-2 overlapping peptide sets (OLP) covering the HSV homolog of a reactive VZV ORF to obtain an initial hit, followed by titration assays with the VZV and HSV-1/HSV-2 homolog peptides. The second started with truncated VZV proteins (primer details on request) and progressed to synthetic peptides within reactive VZV fragments. Autologous PBMC were required as APC for IVTT-expressed polypeptide antigens, and EBV-LCL were used as APC for peptides.

An example for each strategy is detailed for polyclonal CD137^high origin VZV-reactive CD4 T-cells and VZV ORFs 29 and 68. Each was cross-reactive when studied as full-length polypeptides with their HSV-1 and HSV-2 homologs (Fig. 2C). ORF68 truncation showed several reactive fragments, with AA 264-420 the strongest (Fig. 3A). ORF68 is a promising vaccine candidate, and data consistent with multiple ORF68 CD4 epitopes per person is not unexpected based on previous studies (23, 49). We limited the search for cross-reactive peptides to the AA 348-420 region because VZV and HSV proteins are less divergent therein than in the other active VZV ORF68 fragments. CD4+ T-cell reactivity was present for 15-mers 388-402 and 396-410 (Fig. 3B). The HSV-1 and HSV-2 homologs of VZV 388-402, QPMRLYSTCLYHPNA, with sequences AEMRIYESCLYHPQL and ADMRIYEACLYHPQL, respectively, were active (Fig. 3C). The HSV homolog of VZV 396-410, which is identical between HSV-1 and HSV-2, was inactive at 1 μg/ml (data not shown). Dose-response (Fig. 3C) showed equal potency for both HSV homologs of VZV 388-402. As an example of the second strategy, we tested HSV-2 UL29 OLP and detected strong responses to HSV-2 UL29 529-543. Comparison of the homologous alphaherpesvirus peptides (Fig. 3D) showed divergence for potency, with the HSV-2 peptide showing brisk responses down to 1 pg/ml. These HSV peptides differ from their VZV homolog only at amino acids 1 and 12.

Fig. 3.

Cross-reactive CD4 T-cell antigens and epitopes. A: Proliferative responses by VZV-reactive polyclonal CD137^high-origin CD4 T-cells to full-length or truncated VZV ORF68, HSV homologs, and negative control. B: Responses to 15-mer peptides at 1 μg/ml within reactive fragment 262-240. C, D: Dose-response for the same effectors to the indicated viral peptides for VZV-reactive polyclonal CD4 T-cells from a recent zoster vaccinee. All data from triplicate assays. E: Responses of CD4 T-cell clone from an HSV-seronegative person to whole viral and recombinant alphaherpes proteins. × axis labels are participant numbers (30-34) followed by clone number. F: Responses of CD4 T-cell clones to whole viral and peptide antigens (HSV-2 peptide details in Supplemental Table II).

Additional cross-reactive CD4 peptide epitopes (Table I, Supplemental Fig. 1) were obtained by the truncation strategy for VZV ORF24 and the OLP strategy for VZV ORFs 29, 31, 19, 18, and 9. Only peptides active at 1 μg/ml or lower concentration in two separate assays in duplicate or triplicate are reported. For the epitope in VZV ORF24/HSV UL34, detailed truncation analyses of the active 15-mer were performed defining 11-mer VZV ORF24 84-94 as the shortest active peptide (Supplemental Fig. 2). Interestingly, every cross-reactive epitope we have identified to date is non-identical between HSV and VZV, at least at the 13-mer to 15-mer level (Table I). Overall, 11 peptide sets driving cross-reactive HSV/VZV CD4 T-cell responses were discovered. Six epitopes show identity between the HSV-1 and HSV-2 homologs of the reactive VZV peptide, while five display divergence between the biologically active HSV-1 and HSV-2 peptides. In the case of VZV ORFs 29 and 31, the same polyclonal responder cell population responded to two contiguous 15-mer peptides overlapping by 11 AA from each virus. As CD4 T-cells can recognize peptides as short as 8 AA, we conservatively estimate discovery of at least 9 cross-reactive CD4 T-cell peptides if these overlapping pairs are counted as a single epitope each. VZV ORF18 contained a minimum of 3 spatially separated and thus distinct epitopes that were cross-reactive with their homologs in HSV UL40. With the exception of the VZV ORF29 epitope (Fig. 3D), the visually estimated 50% activity concentrations (EC₅₀) tended to be equivalent between peptides from different viruses and mostly in the 0.1 μg/ml range.

Table I.

Novel cross-reactive Alphaherpesvirinae CD4 T-cell epitopes.

subjecta	virus	gene	AA	sequence	EC50 μg/mlb
2	VZV	ORF9	177-189	NKRVFCEAVRRVA	0.1-1.0
	HSV-1	UL49	197-209	------A--G-L-	0.1-1.0
	HSV-2	UL49	197-209	------A--G-L-	Invariantc
5	VZV	ORF18	57-69	FIFTFLSAADDLV	0.1-1.0
	HSV-1	UL40	85-97	-L-A---------	0.1-1.0
	HSV-2	UL40	82-94	-L-A---------	invariant
2	VZV	ORF18	85-97	IHHYYIEQECIEV	0.1-1.0
	HSV-1	UL40	113-125	-L---V-------	0.1-1.0
	HSV-2	UL40	110-122	-L---V-------	invariant
25	VZV	ORF18	165-177	SSFAAIAYLRNNG	0.1-1.0
	HSV-1	UL40	193-205	----------T-N	0.1-1.0
	HSV-2	UL40	190-202	----------T-N	invariant
5	VZV	ORF19	598-612	NSQFLALMPTVSSAQ	0.01
	HSV-1	UL39	960-974	---V------AA---	0.01
	HSV-2	UL39	965-979	---I------AA---	0.01
5	VZV	ORF24	83-94	PYIKIQNTGVSV	0.1
	HSV-1	UL34	83-94	--LR--------	0.1-1.0
	HSV-2	UL34	83-94	--LRV-------	0.1-1.0
5	VZV	ORF29	527-541	TRQPIGVFGTMNSQY	0.01
	HSV-1	UL29	529-543	A------------M-	0.01
	HSV-2	UL29	529-543	A------------A-	0.000001
	VZV	ORF29	531-545	-----------SDCD	0.01
	HSV-1	UL29	529-543	---------M-----	0.01
	HSV-2	UL29	529-543	---------A-----	0.01
5	VZV	ORF31	233-247	STGDIIYMSPFFGLR	0.1
	HSV-1	UL27	290-304	A---FV-----Y-Y-	0.1
	HSV-2	UL27	285-299	A---FV-----Y-Y-	invariant
	VZV	ORF31	237-251	-----------DGAY	0.1-1.0
	HSV-1	UL27	294-308	FV-----Y-Y-E-SH	0.1
	HSV-2	UL27	289-303	FV-----Y-Y-E-SH	invariant
5	VZV	ORF68	388-402	QPMRLYSTCLYHPNA	0.1
	HSV-1	US8	272-286	AE--I--ES----QL	0.1
	HSV-2	US8	267-281	AD--I--EA----QL	0.1

^aSource PBMC with subjects from Supplemental Table I.

^bConcentrations with 50% maximal activity were estimated graphically.

^cHSV-1 and HSV-2 peptides are identical and the peptide listed was tested.

CD4 T-cell reactivity to VZV proteins and peptides in HSV-seronegative persons

We previously characterized healthy immune seronegative (IS) adults who repeatedly test negative for antibodies to HSV-1 and HSV-2 but have direct ex vivo CD4 T-cell proliferative responses to whole killed HSV-2 antigen (14, 15). Limiting dilution cloning of bulk T-cell cultures after one cycle of whole HSV-2 stimulation of PBMC from these individuals readily yield HSV-2-specific CD4+ T-cell clones. Most of these individuals have had negative, serial studies of genital tract swabs using a sensitive PCR assay for HSV-2 DNA negative, in contrast to the almost universal finding of at least some low level HSV-2 shedding in seropositive persons (50), consistent with a true lack of infection. Many have been in sexual relationships with persons with HSV-2 infection, such that mucosal exposure leading to antigen exposure and T-cell sensitization has been hypothesized to lead to sensitization. In an earlier report, we evaluated four CD4 T-cell clones from an IS individual reactive with HSV-1 UL21, UL29, UL46, and UL47 (14). None reacted with a commercial VZV antigen preparation, despite the presence of a genomic homolog of each HSV-2 ORF in VZV (10).

To examine the hypothesis that some T-cell reactivity to HSV in HSV seronegative persons could be due to Alphaherpesvirinae cross-reactivity, two strategies were used. In the first approach, 12 CD4 T-cell clones reactive with whole UV-HSV-2 were screened against both HSV-2 (24) and VZV genome-covering ORF polypeptide sets. Two clones from subject 34 (Supplemental Table I) cross-reactive against HSV-2 UL2 and the VZV homolog ORF59 were observed (representative clone 6 from subject 34, Fig. 3E). The 10 additional HSV-2-reactive CD4 T-cell clones from IS subjects 30-33 reacted with eight other HSV-2 proteins but react with their VZV homolog (data not shown). By this approach, one in 9 (11%) HSV-2 proteins recognized by clones from HSV-seronegative persons could be ascribed to possible VZV cross-reactivity.

The second method used a panel of 31 CD4 T-cell clones from 8 IS participants and determined their reactivity to HSV-2 15-mer peptides by screening a partial proteome-covering peptide set at 1 μg/ml peptide, similar to (15) (data not shown). A total of 21 peptides were recognized amongst these clones, which were then tested for reactivity to whole VZV. Using a conservative criterion of a net 5,000 cpm of ³H thymidine incorporation, 8 of 31 clones (26%) recognizing 6 of 21 distinct HSV-2 peptides (29%) had proliferative responses to VZV (Fig. 3F). The HSV-2 specificities (Supplemental Table II) included HSV-2 UL27, encoding envelope glycoprotein gB2, a candidate viral entry protein and vaccine antigen. Several clones reacted with the 15-mer peptide HSV-2 protein ICP4 1024-1038, containing a known (48) cross-reactive VZV ORF62 CD4 T-cell epitope in a region with 14 contiguous amino acids identical to the HSV-2 15-mer. A separate CD4 T-cell clone was specific for a second HSV-2 ICP4/VZV ORF62 epitope. Two clones with VZV cross-reactivity recognized distinct epitopes in HSV-2 major capsid protein UL19. Not all clones reactive with HSV-2 UL19 1233-1247 cross-reacted with whole VZV (data not shown), implying that cross-reactivity may vary with T-cell clonotype. The reactive HSV-2 peptides and their putative VZV cross-reactive homologs are each relatively sequence-similar to each other (Supplemental Table II).

The cross-reactive peptide in HSV-2 UL29 was particularly interesting. We noted convergence of epitopes discovered using VZV re-stimulation of PBMC to obtain polyclonal responder cells, and using HSV-2 to obtain CD4 T-cell clones from HSV-seronegative persons. The VZV ORF29 527-541/HSV-2 UL29 529-543 peptide pair was immunogenic for VZV-stimulated polyclonal CD4 cells from a HSV-1/HSV-2/VZV triple-seropositive subject (Table I) and as noted above, the HSV-2 peptide was extraordinarily potent in this context (Fig. 3D). CD4 reactivity to the same HSV-2 peptide was detected independently (Fig. 3F, Supplemental Table II) in subject 33, a 56 year-old who is HSV-1 and HSV-2 seronegative (Supplemental Table I).

Cross-reactivity to whole VZV amongst polyclonal HSV-1-reactive CD8 T-cell lines

We previously showed 100-fold enrichment of polyclonal HSV-1-specific CD8 T-cells from PBMC by DC-based cross-presentation and selection of CD137^high activated cells, and confirmed specificity with polypeptide, peptide, and tetramer assays (25). We began CD8 T-cell cross-reactivity studies using these polyclonal T-cell lines and whole VZV-infected APC in cytokine and cell killing assays. Unlike the CD4 T-cell data above, we have not yet reliably enriched polyclonal VZV-specific memory CD8 T-cells from PBMC by direct- or cross-presentation, so we have no reciprocal cross-reactivity data of VZV-reactive polyclonal CD8 T-cells against HSV.

For a representative HSV-1/VZV-seropositive subject (subject 21, Supplemental Table I), 36.9% of polyclonal HSV-1-driven CD8 T-cells expressed IFN-γ and/or IL-2 in response to autologous HSV-1-infected LCL, with a background of 1.7% for uninfected APC (Fig. 4A). A lower level (16.8%) of cytokine response was detected for HSV-1-infected HLA-mismatched allogeneic EBV-LCL used as APC. This was not simple alloreactivity, as the same cell line was only 0.5% stimulatory in the absence of HSV-1 infection (please see Discussion). The CD8 T-cell line exhibited 10.9% cytokine responses to VZV-infected autologous DFB, and no response to uninfected negative control stimulators and no response to VZV-infected DFB from the same HLA-mismatched subject, 36 (Supplemental Table I) used for EBV-LCL (Fig. 4B, 4A). Thus, about 30% of the HSV-1-reactivity in polyclonal CD8 cultures from this subject showed cross-reactivity to VZV. We conclude that CD8 T-cells derived using HSV-1 have significant cross-reactivity against VZV, and that VZV-infected DFB are competent for antigen presentation to these effectors.

Fig. 4.

Recognition of VZV-infected cells by polyclonal HSV-reactive CD8 T-cells derived from HSV-1 seropositive, HSV-2 seronegative healthy adults. A: CD137-selected effectors show brisk virus-specific-recognition of HSV-1-infected EBV-LCL with lesser activation when co-cultured with HLA-mismatched, HSV-1-infected EBV-LCL. B: The same effectors display virus- and self-restricted activation when co-cultured with VZV-infected autologous DFB. Cells in A and B are gated in live, CD8+ small lymphocytes and data are representative of triplicate assays. C: Polyclonal CD8 effectors purified with HLA A*0201 tetramers containing epitopes in the indicated HSV-1 proteins and specific for cross-reactive epitopes in VZV ORF18 or ORF34 show specific killing of DFB infected with VZV wild-type or vaccine strains in triplicate assays.

We measured the ability of cross-reactive CD8 T-cells to kill VZV-infected cells using mono-specific effectors purified using HLA A*0201-containing tetramers formulated with known HSV CD8 T-cell epitopes. One tetramer included HSV-1 UL40 181-189 (25), previously described (20) as HSV-VZV cross-reactive, and the other tetramer included HSV-1 UL25 372-380 (33). HSV infection down-regulates HLA class I in DFB (31) and eliminates CD8 T-cell recognition, so we typically use EBV-LCL as CD8 T-cell targets and demonstrate virus-specific, HLA-restricted lysis (57). VZV also down-regulates DFB HLA class I, but because EBV-LCL are not permissive for VZV replication, we had to try DFB as APC for VZV killing experiments involving viral infection (51). We observed strong cytotoxicity towards VZV-infected DFB for both VZV vOKA and a wild-type VZV (Fig. 4C). Killing was HLA-restricted.

Processing of VZV full-length polypeptides for presentation to cross-reactive CD8 T-cells

From the preceding data, we predicted that some VZV ORFs should be processed to antigenic peptides and presented to HSV-1-driven polyclonal CD8 T-cell lines. To test this, we transiently co-transfected Cos-7 cells with subject-specific HLA class I cDNA and with individual viral ORFs. The viral genes were expressed as fusions with eGFP under the control of a strong constitutive promoter, while cell surface HLA expression after co-export with C. aethiops β2-microglobulin was documented by flow cytometry (25, 26, 38). Polyclonal CD8 T-cell responders from subject 21 showed HLA A*0201-restricted cross-reactive IFN-γ secretion responses to the VZV ORF homologs of HSV UL25 and UL40 (Fig. 5, top). Application of the HSV-1 CD8 polyclonal re-stimulation and ORFeome screen platform (26, 52) to HSV-2 revealed an additional cross-reactive ORF. In studies focused on subject 35 and HLA B*1502-restricted T-cells (Phillips et al. manuscript in preparation), HSV-2-specific CD8 T-cells recognized full-length HSV-2 UL48 and also the HSV-1 and VZV ORF10 homologs after co-transfection with HLA B*1502 cDNA (Fig. 5, bottom). We conclude that VZV ORF 10, 18, and 34 polypeptides are intracellularly processed to peptide epitopes recognized by HSV-1- or HSV-2-reactive CD8 T-cells.

Fig. 5.

Presentation of full-length alphaherpes proteins to polyclonal CD8 T-cell lines selected from PBMC using cross-presentation of HSV-1 (top) or HSV-2 (bottom). Cos-7 cells were co-transfected in triplicate with the indicated HLA cDNA and viral genes. IFN-γ secretion was detected by ELISA.

CD8 T-cell cross-reactive epitopes

We previously defined 39 HSV-1 peptides recognized by polyclonal HSV-1-driven CD8 T-cell lines (25). We tested each VZV homolog of these HSV-1 CD8 epitopes (n=30) alongside the HSV-1 peptide, using the presence of a VZV homolog protein and 6 or more identical amino acids as selection criteria. Using IFN-γ ICS, we detected cross-reactive responses in 4 of 39 (10%) of epitopes. Our data (not shown) for a HSV UL40/VZV ORF18 cross-reactive CD8 T-cell epitope confirm the earlier report by Chiu et al. (20) that these ORFs contain a cross-reactive HLA A*0201-restricted epitope. Novel cross-reactive CD8 epitopes were detected in HSV-1/VZV ORFs UL25/ORF34 (subjects 21 and 23, two distinct epitopes) and UL29/ORF29 (subject 23) (Table II). While background IFN-γ was above zero for some responder lines, levels observed after peptide addition were reproducibly higher. Each novel cross-reactive epitope had strong predicted (53) binding to the relevant HLA class I molecule for both the HSV and VZV peptide homologs (Table II).

Table II.

Novel cross-reactive Alphaherpesvirinae CD8 T-cell epitopes.

virus	gene	AA	sequence	HLA restriction	predicted HLA binding, nMa
VZV	ORF10	164-172	ELRAREEAY	B*1502	163
HSV-1	UL48	160-168	-------S-		221
HSV-2	UL48	158-166	-------S-		Invariantb
VZV	ORF29	893-901	YMANLILKY	A*2902	2
HSV-1	UL29	895-903	----Q--R-		2
HSV-2	UL29	895-903	----Q--R-		invariant
VZV	ORF34	232-240	AVKCLYLMY	A*2902	22
HSV-1	UL25	235-243	-------L-		8
HSV-2	UL25	240-248	-------L-		Invariant
VZV	ORF34	361-369	FLMEDQTLL	A*0201	3
HSV-1	UL25	367-375	--W------		4
HSV-2	UL25	372-380	--W------		Invariant

^aFrom Immune Epitope Database (http://www.iedb.org/) using ANN. nM = nanomolar.

^bHSV-1 and HSV-2 peptides have identical sequence.

Transfection assays (Fig. 5) indicated the presence of at least one cross-reactive HLA B*1502-restricted epitope in HSV-1 UL48/VZV ORF10. Inspection of the sequences and HLA binding prediction (54) suggested that HSV-2 UL48 160-168 and the homologous VZV ORF10 164-172, with sequences ELRAREESY and ELRAREEAY, respectively, were candidate cross-reactive peptides. Initial mapping with HSV-1 overlapping peptides defined a HSV UL48 CD8 epitope in this region (data not shown). Peptide truncation (Fig. 6A) mapped the nonamer peptide epitope to UL48 160-168, a region identical between HSV-1 and HSV-2. Addition of C terminal amino acids reduced activity while an N-terminal alanine did not change potency. The antigenic region is sequence-identical in HSV-1 and HSV-2. For the polyclonal responders, the VZV nonamer ELRAREEAY had a lower maximal response compared to HSV peptide ELRAREESY, but both were definitely and repeatedly positive (Fig. 6B and not shown).

Fig. 6.

Cross-reactive CD8 T-cell epitopes. A: Definition of minimal epitope in HSV-1 UL48 recognized by polyclonal HLA B*1502-restricted CD8 T-cells. EBV-LCL used as APC are matched to responder cells only at HLA B*1502. B: Comparison of shortest fully active HSV-1 UL48 and homolog VZV peptide for same responders and APC. C: Dual staining of same responder cell line with tetrameric complexes of HLA B*1502 and either the HSV-1 or VZV peptide. Numbers are percent of cells in square regions with overall gating on live CD3+CD8+ cells. D: Reactivity of polyclonal HSV-1-specific CD8 T-cell line to DMSO control or the indicated peptides at 1 μg/ml with autologous EBV-LCL as APC. Numbers are percent of gated live CD3+ cells. E: Reactivity of T-cell clones from subjects 21 and 24 to the same nonamer peptides. Clone TCC17 used for TCR sequencing and cloning indicated. F: Reactivity of lentivirally transduced CD8 PBMC expressing a CMV-specific control TCR or the TCR from TCC17, to HLA A*0201 (+) EBV-LCL as APC and 1 μg/ml viral peptide.

Heterogeneity for CD8 cross-reactivity with peptide-specific T-cell populations and reconstruction of cross-reactivity with cloned TCR

The CD8 T-cell response to individual viral peptides is usually polyclonal (55). The higher IFN-γ responses for HSV than for VZV peptides amongst polyclonal responders suggested that these populations contained cross-reactive and HSV mono-reactive clonotypes. To study cross-reactivity at the clonal level, peptide-HLA tetramers (33) constructed for HSV-1 UL25 372-380/HLA A*0201 were used to stain peptide-specific cells. CD8 (+) tetramer (+) T-cells were selected from PBMC or from polyclonal HSV-1-reactive CD8 T-cell lines, cloned by limiting dilution, expanded, and then tested in split-well format. For subject 21, all 36 clones derived from pre-expanded polyclonal line were cross-reactive while for subject 24, all reactive clones were HSV mono-reactive. Clones derived directly from PBMC showed discrete HSV-1-only reactivity or HSV-1/VZV cross-reactivity (Fig. 6E).

Separate populations of CD8 T-cells with or without cross-reactivity was also observed for the B*1502-restricted HSV UL48 epitope. Tetramers of B*1502 with the HSV or VZV peptide were differentially fluorochrome labelled. Each stained a well-separated subset of polyclonal, HSV-2-driven CD8 T-cells by single staining (not shown). When used simultaneously, we observed HSV/VZV double tetramer-positive and HSV tetramer-only-positive groups (Fig. 6C). These were sorted for TCRB CDR3 sequencing. The cross-reactive cells had a single dominant (68%) TCRB CDR3 using TCRBV12 and TCRBJ01-01*01 encoding CASNRQGARNTEAFF (Table III). The HSV mono-reactive cells appeared more polyclonal, with two TCRB CDR3 sequences comprising 45 and 31% of reads, using distinct TCRBV genes but the same TCRBJ 01-06*01 gene. Two subdominant clonotypes had related CDR3 sequences (CASSQYRVDEQFF and CASSQLRRETQYF) and used similar gene segments. The major clonotypes in each group were rare in the other. The tetramer staining and CDR3 sequence data indicate that alphaherpesvirus-specific mono- and cross-reactive CD8 T-cells are distinct.

Table III.

TCRB CDR3 sequences and gene use by polyclonal HSV-2-reactive CD8 T-cells subpopulations sorted with tetramers of HLA B*1502 and HSV UL48 160-168 or VZV ORF10 164-172.

TCRBV	TCRBJ	CDR3	bulk HSV-2- reactive	HSV tetramer (+) VZV tetramer (+)	HSV tetramer (+) VZV tetramer (−)
			read % (rank) a
27-01*01	01-05*01	CASSLSEDSNQPQHF	9.41 (1st)	1.06 (5th)	0.99 (8th)
06-05*01	02-07*01	CASSRDLLEQYF	4.21 (2nd)	0.26 (12th)	0.15 (18th)
06	01-06*01	CASSHDRGGHNSPLHF	4.15 (3rd)	3.58 (3rd)	30.74 (2nd)
02-01*01	02-01*01	CASSMSSYNEQFF	3.45 (4th)	0.35 (11th)	0.25 (12th)
06-01*01	01-01*01	CASGRQDENTEAFF	2.31 (5th)	0.07 (112th)	not found
27-01*01	01-06*01	CASKIGTGASPLHF	2.31 (7th)	0.23 (13th)	45.48 (1st)
12	01-01*01	CASNRQGARNTEAFF	0.79 (21st)	68.33 (1st)	0.71 (8th)
14-01*01	02-01*01	CASSQYRVDEQFF	0.33 (70th)	0.24 (13th)	5.31 (3rd)
14-01*01	02-05*01	CASSQLRRETQYF	0.12 (137th)	1.02 (6th)	2.79 (4th)
06-05*01	01-01*01	CASSYGRGRRNTEAFF	0.15 (119th)	5.83 (2nd)	1.66 (5th)
01-01*01	02-07*01	CTSSHCDTGTTSVYEQYF	0.08 (184th)	1.67 (4th)	0.22 (14th)
reads			4,614,956	27,407	97,518
productive reads (%)			85.01	89.5	94.5
unique productive reads			3,039	98	173
clonalityb			0.42	0.72	0.69

^aPercent of CDR3 reads and rank amongst total reads. Top five CDR3 sequences bolded.

^bPer Adaptive algorithm. A monoclonal population is 1.0. No enrichment for any CDR3 sequence is 0.

To confirm TCR cross-reactivity, TCRA and TCRB CDR3 regions were cloned from cross-reactive CD8 T-cell clone TCC17 (Fig. 6E) into a lentivirus (34, 37). The TCRA and TCRB used TRAV13-1*02/TRAJ18*01 and TRBV28*01/TRBJ2-7*02, encoding CDR3 sequences CAATRGSTLGRLYF and CASSSAMTSGSSYEQYF. Transduced T-cells specifically recognized both the HSV-1 and VZV peptides, but not a CMV peptide, with a reciprocal pattern for the control CMV-specific TCR (Fig. 6F).

Discussion

In this report, we demonstrate recognition of VZV at the virus, ORF, and peptide levels by polyclonal HSV-1-driven CD4 responder T-cells. Reciprocally, we detect responses to HSV-1 virus, antigen, and peptide by CD4 T-cells reactivated using VZV. This is largely recapitulated for CD8 T-cells, for which polyclonal HSV-1-reactive cells recognize VZV-infected cells, proteins, and peptides. Various methods for measuring the degree of CD4 and CD8 T-cell cross-reactivity each yield numbers in the 10-50% range. We propose that this degree of cross-reactivity is likely to be clinically significant in some situations, and is suitable for exploitation during vaccine design.

Homo sapiens harbor two Simplexviruses, HSV-1 and HSV-2, that are antigenically related including T-cell cross-reactivity (56, 57). The Betaherpesvirinae and Gammaherpesvirinae that infect humans are more distantly related (58). Earlier analyses have detected single instances of CD4 and CD8 T-cell HSV/VZV cross-reactivity (20, 48) with CD8 cross-reactivity extending to EBV. Thus far we have not detected cross-reactivity to additional herpesvirus peptides outside of Alphaherpesvirinae (not shown), but our data support HSV/VZV cross-reactivity as a generalizable phenomenon. Some of the cross-reactive epitopes appear to be immunodominant, for example HLA *0201-restricted responses specific for HSV UL25 (Fig. 4, 5) and cross-reactive with VZV ORF34 can have quite high magnitude in direct ex vivo PBMC tetramer stains (33). For most of the epitope studied herein, further research is required to determine their place within immunodominance hierarchies.

Clinically or epidemiologically important reciprocal interactions during the natural history of primary and recurrent infections by HSV and VZV have not been described, but arguably have not yet been carefully sought. Perhaps the best opportunity to determine if HSV-specific immunity might modify VZV infection is zoster. Almost all adults harbor latent VZV, but only about 30% of adults experience clinical zoster. The reasons for this heterogeneity are incompletely understood. Analysis of specimens from the placebo arms of the zoster prevention vaccines (21-23) by sero-testing for HSV-1 and HSV-2 could determine if HSV-infected persons have variation in zoster incidence or post-herpetic neuralgia. Another puzzling clinical phenotype is the variable efficacy of the current vaccine, which protects 50-70% of persons from shingles. It is possible that the currently licensed live attenuated VZV vaccine could be more active in HSV-infected persons, as abundant HSV-specific memory CD4 and CD8 T-cells are present and the subset that cross-react with VZV might be boosted by the vaccine.

Negative immune interactions are possible. Memory responses have a lower threshold for activation than does priming, so prior infection could favor expansion of pre-existing cross-reactive memory clonotypes, referred to as original antigenic sin (59). Longitudinal cross-reactivity assays in persons in the process of primary seroconversion to one virus and with prior immunity to the other, and viral accurate proteome-wide comparisons of the relative contributions of mono-reactive vs. cross-reactive T-cell memory from mono- and dually-infected persons may detect this phenomenon. Cross-modulation due to T-cell exhaustion from chronic infection is also possible. Essentially all HSV-2-infected persons have intermittent reactivations (50). VZV is thought to reactivate far less frequently. Despite reactivations, HSV-specific CD4 T-cells show brisk proliferative and polyfunctional cytokine responses (40) and CD8 T-cells proliferate well to antigen (25, 60). There is some evidence that proliferation, cytokine secretion and cytotoxic potential may vary with HSV severity and fine epitope specificity (61, 62). VZV-specific CD4 T-cells are polyfunctional and proliferate well during VZV quiescence (16), show characteristics of exhaustion when obtained during zoster, and revert to a non-exhausted phenotype after recovery (63, 64). HSV-VZV cross-reactive CD8 T-cells studied with a previously identified tetramer showed a spectrum of exhaustion markers between persons and irrespective of HSV immune status (20).

Proteome-wide reagents allowed us to detect many examples of Alphaherpesvirinae T-cell cross-reactivity. Our genome-covering screening assays are by nature high throughput and it is possible that positive responses may vary near the cutoffs that are determined using irrelevant antigens. Sampling of microbe-specific cells isolated from blood using CD137-based sorting could also lead to variable results for rare specificities. Synthetic peptides allow us to validate and extend the observations made using our virtual libraries. We focus on antigens in candidate vaccines and those showing both CD4 and CD8 T-cell cross reactivity. VZV ORF68 is a virulence factor (65), neutralizing antibody target (66), and vaccine candidate showing over 95% efficacy as an adjuvanted subunit (23). Bergen et al. detected T-cell responses to VZV ORF68 388-400, and based on sequence, speculated it might be HSV-cross-reactive (67). We now provide data that this is true (Fig. 3C). Similar to Bergen et al. (67), we show that individuals can recognize multiple ORF68 epitopes (Fig. 3A).

A subunit vaccine containing truncated HSV-2 ICP4 has partial efficacy against recurrent HSV-2 (68, 69). We observed that HSV-1-specific CD4 T-cell lines frequently recognized HSV-1 ICP4 and VZV ORF62 (Fig. 2D; last gene in upper section amongst genes with homologs in both viruses). Ouwendjik et al. detected HSV-VZV ORF62 918-927, sequence LLLSTRDLAF, as a minimal HSV-VZV cross-reactive epitope (48). We recovered CD4 T-cell clones from HSV-uninfected persons that recognized HSV-2 ICP4 1024-1038, QGVLLLSTRDLAFAG, homologous to VZV ORF62 915-929, RGVLLLSTRDLAFAG. These clones recognize whole VZV, thus likely detecting the epitope described above. A CD4 T-cell clone reactive with HSV-2 ICP4 505-519 also cross-reacts with whole VZV (Supplemental Table II). ICP4/ORF62 are also CD8 T-cell targets (68). These data support use of ICP4 in vaccines.

In this report, we extend HSV-VZV CD4 T-cell cross-reactivity to non-identical peptides. The shortest active peptides in the cross-reactive HSV UL34/VZV ORF24 epitope are different. Most of the other proven (Supplemental Fig. 1, Fig. 3, Table I) and probable (Supplemental Table II) cross-reactive peptide sets also contain several variant amino acids. Whole-virus and ORF-level cross-reactivity (Fig. 1, 2) thus likely integrates identical and non-identical epitopes. Heterogeneity for mono- and cross-reactivity T-cell clonotype level, as shown for CD8 T-cell (Fig. 6C, 6E), is also likely for CD4 responses.

Vaccine antigens recognized by both T-cell subsets are rational. We report that VZV ORF29/HSV UL29 and VZV ORF18/HSV UL40 contain cross-reactive CD4 and CD8 T-cell epitopes. VZV ORF18/HSV UL40 is an enzyme that is a prevalent CD4 T-cell antigen for both viruses and contains additional documented epitopes (16, 25). Chiu et al. determined that CD8 T-cells purified with tetramers of HLA A*0201 and a VZV ORF18 peptide could cross-react against HSV-1 and HSV-2 homologs. We used a similar tetramer, containing the HSV-1 variant epitope we initially described (25) to extend these findings to recognition of full-length, naturally processed viral proteins (Fig. 5), and whole virus (Fig. 4A, 4B). To our knowledge, this is the first demonstration of CD8 T-cell recognition of VZV-infected cells by CD8 T-cell effectors. The recognition of HSV-1 by these polyclonal effectors had a component that appeared not to be HLA-restricted (Fig. 4A, right-most panel). We hypothesize that this is due to transmission of HSV-1 from infected fibroblasts to CD8 T-cells during this prolonged assay, followed by T-cell recognition of infected T-cells. We extended CD8 effector functions to cytotoxicity, showing that HSV UL40/VZV ORF18-reactive and HSV UL25/VZV ORF34-reactive CD8 T cells kill VZV-infected fibroblasts in a virus- and HLA-restricted manner (Fig. 4C).

VZV ORF10/HSV UL48 is a third dually CD4/CD8 cross-reactive target. The CD8 data in this report (Fig. 6A, 6B, 6C, Table II) is complemented by our prior reports that the VZV, HSV-1 and HSV-2 UL48 homologs contain multiple CD4 and CD8 T-cell epitopes recognized by cells from blood, genital skin biopsies, and human trigeminal ganglia (24, 26, 32, 70-73) (16). Several of the minimal CD4 epitopes defined in our previous work are sequence-conserved between these viruses.

There is increasing interest in adoptive T-cell therapy for severe viral infections in immune compromised hosts (39). Both CD4 and CD8 T-cell therapy have been explored with natural and engineered T-cells, and clinical-grade CD8 T-cell products have been developed for VZV (74). We have purified polyclonal HSV/VZV cross-reactive CD8 T-cells restricted by HLA A*0201, the commonest HLA class I allele present in 40-50% of persons in diverse ethnic groups (Fig. 5); these cells can be expanded rapidly as in our polyomavirus work (75). We show here that TCR-engineered cells using TCRs recovered from CD8 clones show HSV/VZV cross-reactivity. This approach holds promise as an off-the-shelf method to provide cross-reactive CD8 T-cells to HLA-defined patients. Interestingly, tetramer sorting-based recovery of cross-reactive versus mono-reactive CD8 clones for the purpose of TCR harvesting was somewhat dependent on the cell source used. While both direct ex vivo PBMC and cell lines yielded cross-reactive clones, the additional recovery of HSV mono-reactive clones directly from PBMC (Fig. 5E) suggest this cell source might best capture TCR diversity.

TCR sequencing has also implies structural information concerning peptide-HLA contacts. The cross-reactive HSV UL48/VZV ORF10 peptides are identical at amino acids 1-7 and 9. From TCRB CDR3 sequencing (Fig. 6C supplying cells sequenced in Table III), we conclude that TCRB CDR3 amino acids contribute to differential recognition of the HSV and VZV peptide variants. This is consistent with crystal structures of TCR-HLA-peptide (76, 77). We are testing the hypothesis that TCRA CDR3 sequences may be similar for mono- and cross-reactive cells. Interest in HLAB*1502-restricted TCRs and epitopes is enhanced by the high risk for severe T-cell mediated skin reactions in HLA B*1502 (+) persons receiving carbamazepine (28). Abacavir, a cause of severe, HLA-linked toxicity (78) binds the peptide binding groove in an allele-specific manner (79) and alters the HLA-bound peptide repertoire (80). We hope to use full TCR sequences from HLA B*1502-restricited T-cells to elucidate HLA B*1502-related hypersensitivity.

T-cell cross-reactivity is likely quite ubiquitous given the large size of the set of potential T-cell epitopes and the lesser observed complexity of TCRs (81). T-cell responses amongst apparently uninfected persons and mice can have memory phenotypes and be linked to specific responses to sequence-unrelated peptides present in environmental flora (82, 83). In contrast, the cross-reactivity studied in this report is likely due to sequence-related, homologous peptides present in phylogenetically related pathogens. However, while we observed cross-reactivity in both dually HSV-VZV infected persons and VZV mono-infected persons, it does not explain all of the reactivity to HSV we observe in HSV-uninfected persons. Thus, priming by more distantly related antigens may account for some of our observations. We have not yet done reciprocal studies to measure if VZV-seronegative persons, who are quite uncommon in most populations, can harbor VZV-reactive T-cells, possibly as a result of HSV cross-reactivity.

In summary, we have demonstrated bi-directional, CD4 and CD8 T-cell cross-reactivity between HSV and VZV, detected cross-reactivity for many protein homolog sets, defined several novel cross-reactive fine T-cell epitopes, and provided quantitative estimates for the extent of cross-reactivity. The clinical significance of these observations should be testable. Several viral ORFs that elicit both cross-reactive CD4 and CD8 T-cells have been discovered that can be up-prioritized as subunit vaccine candidates, and synthetic immunogens that link cross-reactive moieties can be envisioned. Licensed primary varicella and shingles vaccines are live attenuated VZV strains, but we have not yet had the chance to study whether this elicits or boosts cross-reactive T-cells. Cross-reactivity due to VZV infection can account for a proportion of T-cell reactivity to HSV that has been noted in HSV seronegative persons. Additionally, we provide the first evidence that specific CD8 T-cells can recognize VZV-infected skin cells, and reconstitute CD8 T-cell cross-reactivity using expression of cloned TCR to provide a prototype for parsimonious adoptive T-cell therapy of multiple viral infections. Using current proteome-wide antigen and TCR tools to dissect T-cell responses and recognizing that most microbes occur in large, related families, we now appreciate that cross-reactivity can make a considerable contribution to overall pathogen-specific responses.