Zoster Vaccination Increases the Breadth of CD4⁺ T Cells Responsive to Varicella Zoster Virus

Abstract

Background. The live, attenuated varicella vaccine strain (vOka) is the only licensed therapeutic vaccine. Boost of varicella zoster virus (VZV)–specific cellular immunity is a likely mechanism of action. We examined memory CD4⁺ T-cell responses to each VZV protein at baseline and after zoster vaccination.

Methods. Serial blood samples were collected from 12 subjects vaccinated with Zostavax and immunogenicity confirmed by ex vivo VZV-specific T-cell and antibody assays. CD4⁺ T-cell lines enriched for VZV specificity were generated and probed for proliferative responses to every VZV protein and selected peptide sets.

Results. Zoster vaccination increased the median magnitude (2.3-fold) and breadth (4.2-fold) of VZV-specific CD4⁺ T cells one month post-vaccination. Both measures declined by 6 months. The most prevalent responses at baseline included VZV open reading frames (ORFs) 68, 4, 37, and 63. After vaccination, responses to ORFs 40, 67, 9, 59, 12, 62, and 18 were also prevalent. The immunogenicity of ORF9 and ORF18 were confirmed using peptides, defining a large number of discrete CD4 T-cell epitopes.

Conclusions. The breadth and magnitude of the VZV-specific CD4⁺ T-cell response increase after zoster vaccination. In addition to glycoprotein E (ORF68), we identified antigenic ORFs that may be useful components of subunit vaccines.

Herpes zoster (shingles) is caused by reactivation of varicella zoster virus (VZV). This alphaherpesvirus establishes latent infection of neurons innervating skin and other structures during primary infection. Zoster typically presents in a single dermatome as a unilateral, painful, vesicular rash. Chronic, long-lasting pain known as postherpetic neuralgia (PHN) is common, may require prolonged analgesic use, and can impair quality of life. Other complications include encephalitis, ocular and cranial nerve dysfunction, and sometimes death. Higher risk for vascular diseases, such as stroke, transient ischemic attack, and myocardial infarction, is also observed postzoster [1, 2]. Disseminated infection and complications are more likely in immunocompromised and older persons [3, 4].

Recent estimates of zoster rates in high-income countries are 3.4 to 5.0 per 1000 person-years and 8 to 11 per 1000 person-years in those ≥65 years old [3]. Zoster incidence appears to be rising [5]. Current strategies to prevent zoster rely on therapeutic vaccination of older adults with a live attenuated (vOka) virus vaccine. As the only licensed immunotherapeutic vaccine, this represents an important paradigm for boosting acquired immunity without stimulating immunopathology [6]. The zoster vaccine is licensed for use in persons ≥50 years and recommended for use in those ≥60 years by the Advisory Committee on Immunization Practices. Zoster vaccination reduces risk for shingles (70% in persons aged 50–59 years and 51%–55% in persons ≥60 years) and other complications, including PHN (59%–67%) [6–9], thus has clear benefits yet is only partially protective in aged populations. In addition, as a live attenuated virus, it is not recommended for immunocompromised persons [4, 9]. There is, therefore, a need for alternative and improved vaccines.

Alternative vaccines in development include heat-treated vOka [10] and an adjuvanted VZV glycoprotein E (gE) protein subunit [11, 12]. While safe and immunogenic, their protective efficacy is largely unknown. gE is a target of neutralizing antibodies, making a gE-subunit vaccine a rational choice for preventing primary infection. It is also a known CD4⁺ T-cell antigen, but how gE compares to other VZV proteins with regard to cellular immunity is unknown. The current zoster vaccine boosts both arms of the adaptive immune system [13]. Fold increase in anti-VZV immunoglobin G (IgG), when accompanied with a high absolute titer, correlates with protection against zoster [14, 15]. Vaccine-induced T-cell levels may also correlate with protection [16].

This study was conducted to more fully understand the nature of the immune boost from the licensed zoster vaccine. The mechanism(s) of action should be understood to improve efficacy, create a vaccine that is efficacious and safe for immune-compromised persons, and extend therapeutic vaccination to other chronic infections. We hypothesized that zoster vaccination would increase the breadth of the T-cell response to VZV in addition to magnitude. By screening blood-derived T cells for reactivity to each VZV protein, an overall increase in CD4⁺ T-cell breadth was observed. Population-prevalent responses to novel VZV-derived T-cell antigens were discovered and confirmed to the peptide level. This information may benefit the development of future vaccines.

MATERIALS AND METHODS

Subjects and Specimens

Twelve VZV-seropositive adults age 50 or older (seronegative for human immunodeficiency virus and hepatitis B and C) with no history of zoster or recent VZV exposure were vaccinated. Zostavax was administered (day 0) as a single 0.65-mL dose subcutaneously in the deltoid region of the upper arm. Blood was obtained before (day 0) and after (days 14, 28, 182) vaccination. Additional blood was collected to obtain antigen-presenting cells (APCs). Peripheral blood mononuclear cells (PBMCs) were isolated from acid citrate dextrose–anticoagulated blood by density centrifugation and cryopreserved within 8 hours [17]. Serum samples were collected for anti-VZV IgG titers by end-point glycoprotein enzyme-linked immunosorbent assay [18]. Class II human leukocyte antigen (HLA) genotyping was performed using next-generation sequencing (Geraghty Lab, Fred Hutchinson Cancer Research Center, Seattle, Washington) [19]. The protocol was approved by the University of Washington institutional review board, and all participants signed a written informed consent.

Viral Antigens

Expression constructs containing 71 sequence-verified VZV genes or fragments were made as detailed elsewhere [20, 21] covering each known VZV open reading frame (ORF) [22]. Most genes were from parental Oka and some from a clinical VZV isolate (Seattle, Washington). Briefly, VZV genes were polymerase chain reaction–amplified using primers containing adaptors for cloning into pDONR207 or pDONR221 (Invitrogen, Grand Island, New York) (Supplementary Table 1). Resultant pENTR constructs facilitated subcloning into pDEST203 [21]. VZV proteins were synthesized as N-terminal 6-His fusions by cell-free Escherichia coli in vitro transcription translation (IVTT) (Expressway, Invitrogen) and expression confirmed by dot-blot Western via anti–6-His monoclonal antibody (mAb) (Roche, Indianapolis, Indiana) [21]. A negative control antigen set contained IVTT proteins from Francisella tularensis genes, empty pDEST203, or no DNA.

VZV was propagated on human embryonic tonsil (HET) cells. Vaccine Oka strain (vOka, Zostavax, Merck, West Point, Pennsylvania) was reconstituted per the manufacturer to approximately 3 × 10⁴ plaque-forming units (PFUs)/mL assuming 19 400 PFU per dose as per the package insert. HET in Eagle's modified essential medium (EMEM; Corning, Tewksbury, Massachusetts) with 10% fetal calf serum (FCS), 2 mM l-glutamine (Hyclone, Logan, Utah), and 1% penicillin/streptomycin (Hyclone) at 80% confluence were infected at a multiplicity of infection of 0.01. At 50%–60% cytopathic effect (CPE), trypsin-harvested cells were cocultured with uninfected HET 1:3 (infected:uninfected) to generate cell-associated virus, which was cryopreserved. For virus revival, thawed infected cells were overlaid onto uninfected subconfluent HET monolayers at 1:3.

For ultraviolet (UV)–inactivated VZV antigen (UV-VZV), 80% confluent HET were overlaid with vOKA-infected HET at 1:6. At 50%–60% CPE, cells removed by scraping were resuspended in complete EMEM, sonicated, and centrifuged (400g, 10 minutes). Supernatant (100 µL droplets) was exposed to UV light for 30 minutes, pooled, aliquoted, and frozen [23]. Mock UV-HET antigen was made in parallel.

Ex Vivo Assays

Ex vivo PBMC analyses used duplicate interferon (IFN)–γ–enzyme-linked immunospot (ELISPOT) and intracellular cytokine cytometry (ICC) for interleukin (IL)–2 and IFN-γ [24, 25]. For ELISPOT, overnight-rested PBMCs (5 × 10⁵/well) in R10 (Roswell Park Memorial Institute 1640, 10% FCS, 2 mM l-glutamine, 1% penicillin/streptomycin) were added to anti-IFN-γ (1-D1K-γ; Mabtech, Cincinnati, Ohio)-coated multiscreen-Immobilon-P ELISPOT plates (Millipore, Billerica, Massachusetts) with UV-VZV (1:100), UV-mock (1:100), medium, or phytohemagglutinin (PHA-P; 1.6 µg/mL; Remel, Kansas City, Kansas). Secreted IFN-γ was detected after 18 hours with 7-B6–1-biotin antibody, streptavidin-alkaline phosphatase (AP), and AP–conjugate (Bio-Rad, Hercules, California). Spot-forming units were counted (Immunospot, CTL, Shaker Heights, Ohio). For ICC, rested PBMCs (10⁶ PBMCs/well) were incubated (18 hours) with T-cell medium (TCM), costimulatory antibodies (αCD28 and αCD49d; BD, San Jose, California) and antigens: UV-VZV or UV-mock (1:100), medium, or staphylococcal enterotoxin B (SEB, 2.5 µg/mL, Sigma) [26]. Brefeldin-A (Sigma) was added after 2 hours. After Live/Dead violet (Invitrogen) stain, cells were treated with fluorescence-activated cell sorting (FACS) lyse/perm II (BD) and stained with: αCD3-PE (S4.1; Invitrogen), αCD4–fluorescein isothiocyanate (S3.5; Invitrogen), αCD8-PerCP/Cy5.5 (SK1; BD), αIFN-γ-PE/Cy7 (4S.B3; BD), αIL-2-APC (MQ1-17H12; BD). Data collected on a BD FACSCanto II cytometer were compensated/analyzed with FlowJo (8.6.6; Treestar, Ashland, Oregon). Single live CD3⁺CD4⁺CD8^– lymphocytes were assessed for IFN-γ and IL-2.

VZV-reactive T-cell Lines and Assays

UV-VZV (1:200) was added to PBMCs (4 × 10⁶/2 mL TCM). After 4 days (37^oC/5% CO₂), natural human interleukin-2 (nIL-2; Hemagen, Columbia, Maryland) was added (32 U/mL) and replenished twice a week for 2 weeks. Enrichment for VZV reactivity was tested by ICC in 2.5 × 10⁵ T cells combined with equal numbers of carboxyfluorescein succinimidyl ester (CFSE, Invitrogen)–stained autologous APCs [17]. Staining and analysis was as described above, substituting αCD4-APC/H7 (Clone RPA-T4; BD) and dump-gating CFSE-stained cells. VZV-reactivity-enriched T-cell lines were amplified for 2–3 weeks with anti-CD3 mAbs, recombinant IL-2 (Chiron, Emeryville, California), and feeders [27], and cryopreserved. To detect responses to each VZV protein or fragment, a ³H-thymidine incorporation assay was used [28]. Briefly, 10⁵ VZV-enriched T cells were combined with 10⁵ irradiated autologous PBMCs. Antigens included IVTT products (1:2000, duplicates) and whole UV-VZV or UV-mock (1:100, 6 replicates). ³H-thymidine (0.5 µCi, Perkin-Elmer, Boston, Massachusetts) was added at 72 hours, cells harvested 24 hours later, and counts per minute determined (TopCount NXT, Perkin-Elmer).

Epitope Mapping

Peptides (13-mers, overlapping by 9 amino acids) were synthesized for VZV ORF9 and ORF18 (Genscript, Piscataway, New Jersey), reconstituted in dimethyl sulfoxide (DMSO), and arrayed in matrices to form pools of ≤12 peptides with each peptide in 2 pools. VZV-enriched T cells (10⁵) and autologous irradiated PBMCs (10⁵) were tested with peptide pools (1 µg/mL each peptide), DMSO, medium, or PHA-P (1.6 µg/mL). Positive pools were deconvoluted to identify antigenic peptide(s) [28]. Confirmation included dose-response (serial 10-fold dilutions of peptide beginning at 1 µg/mL) analysis of single peptides in triplicate using a representative response-confirmed cell line per subject.

Statistics

Positive response to VZV proteins in lymphoproliferation assays was determined on a per-cell-line basis surpassing this threshold: at least 2.33 × median absolute deviation of all negative-control antigen samples above the median response of those same samples [29]. This calculation has a theoretical false-positive rate of 1.0%. VZV antigen wells were scored positive only if both replicates exceeded the threshold. ORFs synthesized as fragments (eg, ORF22) were scored positive once if any fragment was deemed positive.

Antibody titers, T-cell frequencies, and T-cell breadth were compared between the different time points using the nonparametric Friedman's test for repeated measures. If significant, subsequent analysis of measures using Dunn's test identified which time points were different from prevaccine samples. Correlation analyses were performed using the Spearman's test. Nonparametric tests were used because the data were highly left skewed, thus did not satisfy assumptions of normality. Two-sided P values <.05 were considered statistically significant. Statistical tests and graphs were generated using Prism (v6.04, GraphPad Software, La Jolla, California).

RESULTS

Immunogenicity Ex Vivo

We enrolled 12 subjects who met the Food and Drug Administration indication for zoster vaccination. Among these, 8 (67%) were men, 10 (83%) were white, and median age was 56 (range, 51–67). We first confirmed that each subject was VZV IgG seropositive at baseline (median titer = 3500; interquartile range [IQR], 2000–5000). Boosting of adaptive immune responses with vaccination was readily observed. Postvaccine median anti-VZV IgG titers increased 1.4-fold by day 14 (5000; IQR, 4000–17 500; P = .043) and 2.1-fold by day 28 (7500; IQR, 4250–10 000; P = .017), returning to 1.1-fold above baseline by 6 months (4000; IQR, 2500–7000; P = .805) (Figure 1A). Ex vivo ICC, verified T-cell reactivity to UV-VZV was boosted and was mediated via CD3⁺CD4⁺ cells, but not CD3⁺CD8⁺ cells (Figure 1B). Median postvaccine IL-2 or IFN-γ responses were >2-fold higher than baseline levels (0.13%; IQR, 0.0–0.25) at day 14 (0.29%; IQR, 0.07–0.36; P = .008) and day 28 (0.30%; IQR, 0.14–0.38; P = .0002), but not at day 182 (0.18%; IQR, 0.08–0.41; P = .173) (Figure 1C). Results for ELISPOT generally mirrored and correlated with the CD3⁺CD4⁺ ICC data (net response, r = 0.782; P < .0001) (Figure 1D). Neither ICC (r = 0.206; P = .16) nor ELISPOT (r = 0.120; P = .42) CD4⁺ T-cell endpoints were correlated with IgG titers (see also Supplementary Table 2).

Figure 1.

Adaptive immune responses to zoster vaccine. A, Anti-VZV IgG titers prior to and after vaccination, with each color representing a given subject. B, Representative dot-plots showing the frequency of cytokine positive events in CD4⁺ and CD8⁺ T cells by ICC in a postvaccine (subject 6, day 28) sample. C, Net frequency (after subtraction of background values) of VZV-specific CD4⁺ T cells before and after vaccination determined by ICC for IFN-γ and IL-2; colors/subject numbers match (A). Statistical significance determined using the Friedman's test followed by Dunn's post-hoc analysis (significant P values are shown). D, Correlation between ELISPOT and ICC using data from each time point, analyzed with the Spearman's test (P value and r value shown). Abbreviations: ELISPOT, enzyme-linked immunospot; ICC, intracellular cytokine cytometry; IFN-γ, interferon γ; IgG, immunoglobin G; IL-2, interleukin 2; SFU, spot-forming unit; VZV, varicella zoster virus.

Breadth of the VZV-specific CD4⁺ T-cell Response

We next determined the number of VZV proteins that scored positive or negative (response breadth) in prevaccine (day 0) and postvaccine (day 28 and 182) polyclonal VZV-reactive cell lines. A lymphoproliferative pattern to discrete VZV proteins was typical with responses to most other proteins near background levels (Figure 2A). Prior to vaccination, CD4⁺ T cells responded to a median 2.5 VZV proteins (IQR, 0.0–9.8) (Figure 2B). At 28 days postvaccination, the median CD4⁺ T-cell response breadth increased 4.2-fold to 10.5 VZV ORFs (IQR, 5.3–20.0; P = .0495). Breadth returned to near baseline levels by 6 months after vaccination (median = 5.5, IQR, 2.0–7.0; day 0 vs day 182, P > .999).

Figure 2.

CD4⁺ T-cell responses to VZV increase in breadth after vaccination. A, Representative plot showing reactivity of a bulk VZV-reactive CD4⁺ T-cell line to VZV and control proteins (duplicate wells), and to whole UV-treated VZV (UV-VZV) or HET (UV-Mock) antigens. Bars represent mean CPM and error bars represent standard deviation. The threshold for positivity—calculated as (median response of all negative control antigens) + (2.33 times median absolute deviation of all negative control antigens)—is represented by a dashed horizontal line. Both duplicates needed to exceed the threshold for the ORF to be determined positive. Positive ORFs are named. B, The number of VZV ORFs recognized by CD4⁺ T-cell lines enriched for reactivity to VZV from samples of 12 study subjects obtained before (D0), 1 (D28), and 6 months (D128) after vaccination. Horizontal bars represent the median of each time point. Each subject is shown with 1 color. Statistical significance determined using the Friedman's test followed by Dunn's post-hoc analysis (significant P values are shown). Abbreviations: CPM, counts per minute; HET, human embryonic tonsil; ORF, open reading frame; UV, ultraviolet; VZV, varicella zoster virus.

Population-prevalent CD4⁺ T-cell Antigens in VZV

The immunoreactivity pattern revealed several VZV polypeptides repeatedly recognized by CD4⁺ T cells (Figure 3A). Vaccination prompted rises in response rates across multiple ORFs. CD4⁺ T-cell responses were observed to 42 (60%) of the 70 unique VZV proteins (Supplementary Table 1) evaluated in at least 1 subject for at least 1 time point. In general, a similar set of proteins (structural and nonstructural) were immunoprevalent (ie, gave the most frequent responses) at baseline and after vaccination.

Figure 3.

CD4⁺ T-cell responses of adult zoster vaccine recipients before (Day 0) and after (Day 28 and Day 182) vaccination. A, VZV ORFs eliciting responses within polyclonal VZV-specific T-cell lines are indicated by colored boxes. Each column represents a study subject with colors consistent at each time point; colors/subject numbers match Figure 1A and 1C. Each row represents a VZV ORF. Ranking is from highest to lowest overall response frequency. An additional 29 ORFs without responses detected are not shown. B, Frequency of protein-specific responses among zoster vaccine recipients at before (Day 0) and after (Day 28 and Day 182) vaccination, or at any time point (bottom). Abbreviations: ORFs, open reading frames; VZV, varicella zoster virus.

Prevaccine responses represent long-term CD4⁺ T-cell memory (Figure 3A, left). Overall, 28 (40%) of the VZV ORFs were recognized by at least 1 person at baseline. The most prevalent baseline response was to ORF68 (membrane glycoprotein E; 5/12 subjects; 42%), followed by ORF4 (regulatory/tegument), ORF37 (membrane glycoprotein H), and ORF63 (regulatory/tegument; 4/12 subjects each; 33%) (Figure 3B).

One month postvaccination, the integrated response breadth of our cohort and the level of highest immunoprevalence both increased. In total, 36 (51%) ORFs were targeted with highest prevalence (9/12 subjects, 75%) for ORF40 (major capsid protein). At least 50% of the study subjects also had responses to ORFs 68 and 63, mentioned above, in addition to ORF67 (membrane glycoprotein), ORFs 9, 12, 62, and 22 (all tegument), and ORFs 18, 36, and 59 (enzymes), 1 month after vaccination. Six months postvaccination, responses to several proteins returned to near prevaccine prevalence levels, although ORFs 40, 59, 63, and 67 retained a high frequency of responses, and 31 (44%) ORFs remained immunogenic.

Integration of pre- and postvaccine reactivity revealed the broad view of T-cell specificity to VZV (Figure 3B, bottom). Largely reflecting the immunoreactivity patterns observed in day 28 samples, responses to ORF40 and ORF67 (83%; 10/12 subjects each) were the most prevalent, followed by ORFs 68, 9, and 59 (75%; 9/12 each), ORFs 37, 12, 62, 63, and 18 (67%; 8/12 each), and ORF36 (58%; 7/12). High response prevalence to relatively short ORFs (eg, ORF9) highlighted that CD4⁺ T-cell immunogenicity is not purely related to polypeptide length.

Confirmation of Responses to VZV ORF9 and ORF18 by Peptide Epitope Mapping

VZV ORF9, orthologous to tegument protein herpes simplex virus (HSV) UL49, and ORF18, the homolog of HSV UL40 (both known T-cell antigens) [30, 31], were selected for peptide confirmation of T-cell reactivity. The selection was based on novelty as VZV T-cell antigens, high CD4⁺ T-cell immunoprevalence, and short lengths (302 and 306 amino acids, respectively). Using a matrix peptide pooling approach, a discrete pattern of positive and negative responses was observed in the polyclonal VZV-reactive responder cell lines (Figure 4A). Candidate single-antigenic peptides identified at the intersections of the positive row and column pools were retested. Reactive and nonreactive peptides were clearly differentiated at 1 µg/mL (Figure 4B). Dose-response studies in serial 10-fold dilutions showed that most peptides remained stimulatory at 0.1 µg/mL, and lost activity below 0.01 µg/mL (data not shown). Peptide-specific responses were detected in all but 1 person among persons with reactivity to ORF9 (n = 9) and ORF18 (n = 8) proteins. Most persons had multiple active peptides within ORF9 and ORF18.

Figure 4.

Peptides in VZV ORFs 9 and 18 recognized by polyclonal VZV-reactive CD4⁺ T-cell lines. A, Representative data showing reactivity to ORF9 peptide pools for subject 3 at day 28. Each 13-amino-acid peptide was present in 1 row pool and 1 column pool. B, Hits in row and column pools were deconvoluted by screening each candidate reactive peptide present in both a reactive row and reactive column pool at 1 µg/mL. C, VZV ORFs 9 and 18 CD4⁺ T-cell epitopes confirmed by single-peptide reactivity. Numbers represent subjects 1–12 as in Figure 3A. Shaded columns represent subjects that had no response to the full-length ORF. Black boxes indicate subject reactivity to listed peptides. Abbreviations: CPM, counts per minute; ORFs, open reading frames; VZV, varicella zoster virus.

We identified 13 antigenic peptides in ORF9 and 14 antigenic peptides in ORF18 (Figure 4C). Reactivity was regularly observed against 2 or more overlapping peptides, such as ORF18_53–65 and ORF18_57–69 for subject 2. Even using a conservative approach in which 2 overlapping 13-amino-acid peptides are counted as 1 epitope, we noted multiple epitopes within several individuals tested for ORF9 (median 2, max 5) and ORF18 (median 1.5, max 7). Several 13-mers were immunogenic in multiple persons (eg, ORF18_169–181 in subjects 6, 8, and 10) with shared HLA types—these persons share the HLA-DPB1*04:01 allele but not HLA-DR or -DQ alleles (Supplementary Table 3). Reactivity of subjects 8 and 10, but not subject 6, to the overlapping peptide ORF18_165–177 suggests a potential heterogeneous response to this region. Some peptides appear to be relatively promiscuous T-cell activators; for example, ORF9_121−133 activated T cells in subjects 3, 7, and 8, and ORF9_185−197 activated T cells in subjects 2, 6, 7, 8, and 9, among whom there was no consensus HLA allelic variant. No sequence variations were noted in the epitope-containing regions (Figure 4C) upon examination of VZV sequences in GenBank (2015).

DISCUSSION

In the present report, we evaluated the CD4 T-cell response to zoster vaccination at the level of individual VZV proteins and peptide epitopes. Antigen-specific responses were readily detectable prior to vaccination, suggesting that memory to childhood varicella is durable, albeit boosting by periodic subclinical reactivation or exposure, or cross-reactive antigens [32] is also possible. Applying a standardized protocol to blood obtained before and after vaccination, we confirmed previous estimates of the overall antibody and CD4 T-cell boost with vaccination [15, 33] and showed that the ORF level breadth of the CD4 T-cell responses approximately quadrupled at 1 month, which declined by 6 months postvaccination theoretically via T-cell recruitment from blood to the skin or lymphoid tissue, loss during contraction, or negative regulation (eg, T-regs [34]). In addition to the subunit vaccine candidate antigen ORF68 (gE) currently in clinical development [35], we uncovered several other population-prevalent CD4 T-cell antigens from VZV that are, therefore, rational for investigation as vaccine immunogens. Little is known about the fine specificity or phenotype of VZV-specific T-cell responses [30, 31]. The few CD4 T-cell epitopes, documented in literature or a central database [36] prior to our study, at the well-defined level of short peptides and HLA class II–restricting alleles were limited to ORFs 4, 31, 62, 63, 67, and 68 [30, 37–40]. We greatly expanded the number of known VZV-specific CD4 T-cell epitopes with our studies, targeting VZV ORFs 9 and 18, and provide a clear pathway to define epitopes in detail from any ORF reactive in our system.

Although we defined T-cell responses in blood, relevant memory immune responses to VZV could reside in susceptible tissues: zoster can attack any skin site, the liver, retina, middle ear, and large arteries, and can cause devastating noncutaneous syndromes [41]. This clinical variety is explained by VZV latent infection in and reactivation from extensive sensory or autonomic ganglia [42], although VZV reactivates rarely [43]. The related alphaherpesviruses, HSV-1 and HSV-2, reactivate more frequently, provoke antigen-specific T-cell influx to sites of recurrent infection, and lead to persistent deposition of tissue resident–memory (T_RM) T cells [44–46]. We have yet to confirm that zoster vaccine–invoked T cells have chemokine and adhesion receptors suitable for migration to relevant tissues or that VZV-specific T_RM persist in ganglia or skin. However, circulating VZV-specific CD4 T cells with a CD45RO⁺ T_CM (central memory) pattern—expressing CD27 and CD28 for costimulation, CCR7 and variable CD62L for lymph node homing, but low cutaneous lymphocyte-associated antigen for skin homing—have been described, some of which show signs of recent antigen exposure (CD38⁺) and possible T-cell exhaustion (PD1⁺) [47, 48]. Analogous to our proteome-wide HSV studies [27, 45], the VZV tools described herein provide opportunities to study acute zoster lesion-infiltrating lymphocytes and T_RM. Current efforts to define the HLA restriction of our epitopes will facilitate tetramer-based study of VZV-specific T-cell phenotype.

Limitations of this study include the relatively small number of subjects and time points studied and the lack of vaccine clinical efficacy data in these subjects. However, repositories from phase III trials could provide suitable large cohort specimens for clinical end-point correlative studies, for which our methods could be suitably simplified and scaled up. We used cell-associated VZV antigen to purposefully restimulate T cells reactive with both virion and nonvirion proteins, and the prevalence of responses to enzymes such as ORF18 showed we were successful. However, it is possible that restimulation and expansion of VZV-specific memory CD4 T cells was not unbiased, such that some antigenic specificity was under- or overrepresented, or extremely rare responses were missed. Nonetheless, we did reliably demonstrate increased response breadth to vaccination.

As a relatively recent vaccination program, insufficient data exist regarding longevity of protection from the current zoster vaccine, although protection is still evident at 7 years with an apparent decline over time [7]. Consistent with our presented observations, short-term immunogenicity of zoster vaccination (in clinical trials) approximate a 2-fold boost in both T-cell and B-cell response magnitude, which declines to levels moderately above prevaccine levels within a year [49]; the declines we observe in magnitude and response breadth may explain reduced protection over time. Although not confirmed mechanistic correlates of protection, vaccinees with both a high fold-rise and high endpoint titer of anti-VZV IgG after vaccination have less risk for zoster [15]. A correlation between T-cell boosting and protection has yet to be reported. However, T cells should not be overlooked in the context of vaccination because they are essential in VZV control in the absence of vaccination. Lack of correlation in VZV-specific T-cell magnitude and antibody titer observed during our investigation (r = 0.206; P = .16) and earlier studies suggest both arms of adaptive immunity have independent relationships with zoster, making antibody rises an unlikely biomarker for T-cell boosts (or vice versa). We did note a relationship between ex vivo magnitude of VZV-reactive T cells and response breadth (r = 0.58; P = .0002), suggesting that dominant responses to single ORFs were not the driving force for T-cell elevation following vaccination. Consequently, CD4 T-cell magnitude may serve as a good surrogate for predicting breadth, albeit it is not known if the number, breadth, specificity, or phenotype of CD4 T cells is the most important for protection from, or resolution of, zoster or its complications.

In conclusion, the CD4 T-cell response to VZV is polyspecific, both before and after therapeutic vaccination of seropositive adults, with myriad VZV proteins documented for the first time to elicit specific immunity. Our data support the hypothesis that zoster vaccination works at least in part by increasing the magnitude and breadth of the T-cell response to VZV.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.