Fifty-three middle-aged adults (50–65 years of age) were included in this phase IV single-center and open-label study. Participants were recruited from an urban area in the middle of the Netherlands in March 2015. Potential participants were excluded based on the following criteria: antibiotic use or fever (>38°C) within the last 14 days, diseases demanding immune suppressive treatment within the last 3 months, a known or suspected immune deficiency, a blood coagulation disorder, a neurologic disorder, administration of blood products in the past 6 months, serious surgery within the last 3 months, the use of hormone supplementation, a suspected allergy toward the vaccine components, a history of serious adverse events after previous vaccinations, a previous varicella zoster episode, a previous Zostavax vaccination, and any vaccination in the month before enrollment. Written informed consent was obtained from all participants before the study. All procedures were in accordance with the Declaration of Helsinki. The study was approved by the Medical Research Ethics Committees United (Mec-U) in Nieuwegein, the Netherlands (NTR4636).

All participants received a single dose of the live-attenuated varicella zoster vaccination (Zostavax; Sanofi Pasteur MSD), containing not less than 19,400 PFU. The vaccine was subcutaneously administered. A blood sample was taken from all participants before the vaccination. Post-vaccination, blood samples were taken after 14 days, 28 days, and 1 year (Figure 1). Peripheral blood mononuclear cells (PBMCs) were isolated at all time points using vacutainer cell preparation tubes (CPT) containing sodium citrate (BD Biosciences), according to the manufacturer’s instructions (23). Plasma samples were collected after initial spinning of the CPT tubes and were filtered using a Corning^®Costar^® Spin^® X filter (Sigma-Aldrich) and were frozen at −20°C before further use. After the initial collection, the PBMCs were washed with RPMI-1640 medium (Gibco) supplemented with 1% heat inactivated fetal calf serum (FCS, Gibco) and 1% penicillin and streptomycin (Lonza). PBMCs were counted and frozen in a 90% FCS, 10% DMSO solution at −135°C until further use. Moreover, pre-vaccination, and at 14 and 28 days post-vaccination, an additional blood sample was collected in tubes containing lithium heparin (BD) for detailed cellular immune phenotyping. Both the vaccinations and blood samplings were performed in the evening hours.

The whole blood cellular immune phenotyping was performed within 18 h after blood collection. Pre-vaccination, the absolute numbers of lymphocytes, monocytes, granulocytes, CD3+ T-cells, NK cells, B-cells, and the different B-cell subsets (naïve mature, transitional, natural effector, memory B-cells, and plasma cells) were determined with a lyse no-wash protocol using TruCOUNT tubes (BD). The following antibodies were used: CD3(UCHT1)-BV711, CD16(B73.1)-PE, and CD38(HB7)-APC-H7 (all from BD Biosciences), CD45(GA90)-OC515 and CD56(C5.9)-PE (both from Cytognos, Salamanca, Spain), CD27(M-T271)-BV421 and IgD(IA6-2)-FITC (both from BioLegend, San Diego, CA, USA), and CD19(J3-119)-PE-Cy7 (Beckman Coulter, Fullerton, CA, USA). Moreover, a detailed immune phenotyping was performed separately on the whole blood pre-vaccination samples using: CD4(RPA-T4)-BV510, CD45RA(HI100)-BV605, and CD28(CD28.2)-PerCP-Cy5.5 (all from BioLegend), CCR7(150503)-PE-CF594, CD8(SK1)-APC-H7, CD25(2A3)-FITC, TCRgd(11F2)-PE-Cy7, and CD127(hIL-7R-M21)-PE (all from BD Biosciences), and CXCR5(51505)-APC (R&D Systems, Minneapolis, MN, USA). Absolute numbers of T-cell subsets were calculated using the CD3+ T-cell numbers from the TruCOUNT analysis. In addition, 14 and 28 days post-vaccination whole blood TruCOUNT analysis was performed using the following antibodies: CD27(M-T271)-BV421, CD45RA(HI100)-BV605, IgD(IA6-2)-FITC, CD8(SK1)-FITC, CD4(OKT4)-PerCP-Cy5.5 (all BioLegend), CD3(UCHT1)-BV711, CD25(2A3)-PE, CCR7(150503)-PE-CF594, TCRgd(11F2)-PE-Cy7, and CD38(HB7)-APC-H7 (all BD Biosciences), CD45(GA90)-OC515 (Cytognos), CD19(J3-119)-PE-Cy7 (Beckman Coulter), and CXCR5(51505)-APC (R&D Systems). Gating strategies for the different cell subsets were applied as previously described (24). An overview of the different markers used to specify the different cell populations is given in Table S2 in Supplementary Material. Flow cytometric analyses were performed on a 4-laser LSR Fortessa (BD Biosciences) using standardized measurement settings (25), and data analysis using FacsDiva V8 (BD Biosciences) and FlowJo V10 (FlowJo Company, Ashland, OR, USA).

Multiscreen 96-well ELISpot plates (Merck Millipore) were shortly activated with 70% ethanol after which they were thoroughly washed with PBS (Tritium Microbiology) and either coated with antihuman IFNγ antibody 1-D1K or antihuman GrzB GB10 (1 mg/mL; Mabtech). The coated plates were kept overnight at 4°C. The next day, the plates were blocked with AIM-V medium (ThermoFisher) containing 5% human AB serum (Sigma-Aldrich), hereafter called T-cell medium, for at least 1 h at 37°C to avoid non-specific binding. The same batch of human AB serum was used for all samples to avoid differences in antigen concentration during T-cell stimulation. After a quick thawing of the PBMCs in T-cell medium, PBMCs were thoroughly washed and distributed in concentrations of 3 × 10⁵, 1.5 × 10⁵, and 7.5 × 10⁴ cells/well over the pre-coated IFNγ or GrzB plates. The PBMCs were stimulated with the mock (negative control), 6 µg/mL VZV-specific purified antigen (VZ10 strain; Genway), or 1 µg/mL Staphylococcus aureus enterotoxin B (SEB, positive control) (Sigma-Aldrich), again in a threefold dilution. All stimulations were performed in duplicate. All time points of one participant were tested on the same ELISpot plate, to reduce assay variation. The plates were incubated for 48 h at 37°C with 5% CO₂. Next, supernatants of the stimulated cells were collected and frozen at −80°C. Subsequently, the plates were washed thoroughly and incubated for 2 h at 37°C with either mouse antihuman IFNγ antibody 7-B6-1 or mouse antihuman GrzB antibody GB11 (both Mabtech). After a washing step, a mixture containing Extravidin–alkaline (Sigma-Aldrich) was added for 1 h at room temperature. Finally, plates were washed and spots were developed by addition of a SIGMAFAST™ BCIP/NBT (Sigma-Aldrich) solution. After drying, plates were scanned with the Epson ELISpot Scanner, and the spots were counted with a standardized protocol using the AELVIS software. Numbers of VZV-specific IFNγ and GrzB producing cells are presented per 10⁶ PBMCs after subtraction of the spots in the mock control. Mock controls on average contained 17 spots/10⁶ PBMCs for IFNγ and 20 spots/10⁶ PBMCs for GrzB. An NK-cell depletion experiment (using CD56 magnetic bead separation) was performed to estimate the role of NK cells in the IFNγ production as measured in the ELISpot assays.

Varicella zoster virus-specific IgG concentrations (IU/mL) at the different time points were measured using a bead-based immunoassay as described previously (26). VZV IgA concentrations (NTU) were measured using an enzyme-linked immunoassay (Genway Biotech, Inc., San Diego, CA, USA) according to the manufacturer’s instructions.

CMV IgG concentrations were determined in the plasma samples using an enzyme-linked immunoassay (ETI-CYTOK-G Plus, P002033, DiaSorin, Saluggia, Italy) according to the manufacturer’s indications and as described earlier (24).

Peripheral blood mononuclear cells were thawed as before. Thereafter, 10⁶ cells/well were stimulated with mock (negative control) or 6 µg/mL VZV-specific purified antigen (VZ10 strain; Genway) in a 48-well plate. Moreover, 5 × 10⁵ cells/well were stimulated with 1 µg/mL SEB (positive control) (Sigma-Aldrich). The cells were incubated for 72 h at 37°C with 5% CO₂. During the last 5 h GolgiPlug protein transport inhibitor containing Brefeldin A (1,000× dilution, BD) was added to each well. After a thorough washing, cells were incubated for 30 min with a mixture of Life-Death Zombie Aqua fluorescent dye (BioLegend) and surface antibodies in FACS buffer, containing PBS with 0.5% BSA and 2 mM EDTA. The following antibodies were used for surface staining: CD3(SK7)-PerCP, CD4(RPA-T4)-ACP, CD45RA(L48)-PE-Cy7, CCR7(3D12)-BV605, CD56(NCAM16.2)-BV711 (all BD), CD38(HIT2)-BV786 (BioLegend), CD8(CLBT8/4, H8)- FITC (Sanquin), and HLA-DR(LN3)-APCefluor780 (eBiosciences). Subsequently, cells were washed with PBS, and permeabilized for 20 min with cytofix/cytoperm (BD). A perm/wash solution (BD) was used to wash the cells. In addition, the cells were stained for 30 min with IFNγ (25723.11)-PE. After an additional washing step, the cells were resuspended in FACS buffer and immediately measured on a 4-laser LSR Fortessa (BD), and the data were analyzed using FlowJo V10. Frequencies of activated (CD38+ HLA-DR+) and CD4+ IFNγ+ (data not shown) VZV-specific cells were calculated after subtraction of the mock control.

After 48 h of PBMC stimulation, IL2, TNFα, IL5, IL13, and IL10 concentrations in the supernatants were determined as previously described (27). Samples with concentrations below the lower limit of quantification were assigned half the concentration of the lowest measurement. Cytokine concentrations in the mock controls were subtracted from those in the antigen stimulated samples.

Only data of participants with measurements at all different time points were used for analysis. The numbers of VZV-specific IFNγ-producing cells, cytokine concentrations, activated cells, and the whole blood TruCOUNT responses at all the different time points were compared with the Friedman test. Only if this test yielded significant outcomes, the Wilcoxon signed rank test was applied to determine significant differences between two separate time points analyzed. The VZV-specific IgG and IgA responses were log-transformed after which the ANOVA was used to compare the different time points. Corrections for multiple testing were used as indicated under the figures/tables. Participants with low and high pre-vaccination immunity were compared at all time points by using the Mann–Whitney U test. Correlations were determined by the Spearman’s rho correlation test. Geometric means with the 95% confidence intervals were indicated in the graphs. The boxplots used in the figures are plotted from the min to max values with indication of the median. The whole blood absolute cell numbers were compared at the different time points between the participants with low and high pre-vaccination CMI with the Mann–Whitney U test. The geometric means of these groups were normalized to z-scores using the geometric means and standard deviation of the total group to be presented in the heat maps. GraphPad V7 and SPSS V22.0 were used.

A total of 53 middle-aged adults participated in this study (mean age: 57.6; range 50–65; 66% male) (Figure 1). After a pre-vaccination blood drawing, all participants received the Zostavax vaccine. Follow-up blood samples were drawn at 14 days, 28 days, and 1 year post-vaccination. Finally, 49 (92.5%) of the participants completed the study. Additional participant health characteristics are presented in Table S1 in Supplementary Material. Due to sample availability, distinct numbers of participant samples were used for the various assays (Figure 1).

Pre-vaccination, highly variable numbers of VZV-specific IFNγ-producing cells per 10⁶ PBMCs were observed in the circulation of middle-aged adults (geometric mean 53.4 [35.8–79.6]) (Figure 2A). These cells significantly increased shortly after vaccination (14 days: geometric mean 101.6 [79.0–130.8]; 28 days: geometric mean 115.9 [94.5–142.3]) and returned to baseline values 1-year post-vaccination (geometric mean 71.5 [55.1–92.8]) (Figure 2A). Importantly, we observed a negative correlation between pre-vaccination VZV-specific IFNγ-producing cell numbers and the fold change of VZV-specific IFNγ-producing cells post-vaccination (14 days: rho: −0.669, 28 days: rho: −0.714, and 1 year: rho: −0.610; all p < 0.0001). We did not observe correlations between the cellular and humoral responses, indicating that these responses develop independently (data not shown). A preliminary NK cell depletion experiment indicated that the majority of this IFNγ was produced by the T-cells (data not shown).

All participants possessed pre-vaccination IgG antibody concentrations above the VZV seropositivity level of 0.26 IU/mL, as established by van Lier et al. (16) (geometric mean 2.0 [1.7–2.5]), indicative of previous infection with the VZV virus in all participants (Figure 2B). These VZV-specific IgG antibodies were found significantly increased at days 14 (geometric mean 12.4 [9.2–17.6]) and 28 (geometric mean 8.2 [6.1–11.1]) post-vaccination. Similar to the IFNγ T-cell responses, no significantly elevated IgG responses were observed 1-year post-vaccination (geometric mean 2.5 [1.9–3.1]) (Figure 2B).

Since we observed a negative correlation between the numbers of pre-vaccination IFNγ-producing cells and the vaccine-induced increase in these cells, we determined whether differential vaccine responses were observed in participants with low and high numbers of pre-vaccination VZV-specific IFNγ-producing cells (Figure 3), hereafter called participants with low and high pre-CMI. The median number of pre-vaccination spots was determined, and participants below the median were considered to have low pre-CMI whereas participants above the median were considered to have high pre-CMI. Remarkably, only participants with low pre-CMI showed enhancement of the number of IFNγ-producing cells post-vaccination, which was maintained up until 1 year (fold change 14 days: 2.7, 28 days: 4.3, and 1 year: 2.2) (Figure 3A). On the contrary, numbers of IFNγ-producing cells were not significantly enhanced post-vaccination in participants with high pre-CMI (fold change 14 days: 1.3, 28 days: 1.1, and 1 year: 0.8) (Figure 3A). One-year post-vaccination, some of these participants even showed trends toward reduced numbers of IFNγ-producing cells compared with pre-vaccination. Interestingly, participants with low pre-CMI were significantly older than the participants with high pre-CMI, even within this small age-range (Figure 3B).

The two groups showed similar VZV-specific IgG concentrations (Figure S1A in Supplementary Material), confirming the independence of the VZV-specific IgG and CMI responses. Noteworthy, IgA responses significantly differed between the two groups, with stronger responses in the participants with low pre-CMI (Figure S1B in Supplementary Material).

To enlarge our understanding of the difference in T-cell responses between the participant with low and high pre-CMI, we compared the production of several cytokines in the cell culture supernatants after 48 h of culture. Significant differences were found between the two groups in VZV-specific Th1 and Th2 cytokines. Post-vaccination, significant enhancement of the VZV-specific IL2, TNFα, IL13, and IL5 secretion was observed in the low pre-CMI group, whereas no significant responses were found in the high pre-CMI group. Noteworthy, no significant IL10 and GrzB responses were observed in both groups (Figure S2 in Supplementary Material).

Overall, our results indicate strong VZV-specific cytokine and IgA responses in the circulation of middle-aged adults with low pre- VZV-specific CMI, whereas this response was absent in the circulation of adults with high pre-CMI. Importantly, a low pre-CMI was associated with higher age in this group.

Subsequently, we investigated the frequencies of VZV-specific activated T-cells, both within the CD4 and CD8 T-cell compartment (gating strategy presented in Figure S3 in Supplementary Material). Activated cells were based on the double expression of HLA-DR and CD38 (gating strategy in Figure S4 in Supplementary Material). Activated VZV-specific CD8 CM cells could not be detected, due to the low frequencies of these cells in the circulation (28). We observed elevated frequencies of activated VZV-specific CD4 T-cells at 14 and 28 days post-vaccination, mainly consisting of CM, TemRO, and TemRA cells (Figure 4A). After 1 year, elevated frequencies of activated CD4 CM and TemRO cells were still found. The numbers of VZV-specific activated CD8 T-cells were significantly enhanced at all time points post-vaccination, and mainly consisted of TemRO and TemRA cells (Figure 4B). Remarkably, activated VZV-specific cell responses were significantly different between the participants with low and high pre-CMI (Figures 4C–F). Participants with low pre-CMI showed strong enhancement of the activated VZV-specific CD4 cells (CM, TemRO, and TemRA) and only a small enhancement of the CD8 TemRA response (Figures 4C,E). By contrast, participants with high pre-CMI showed a CD8 T-cell oriented response, with increased frequencies of VZV-specific CD8 TemRA cells up until 1 year (Figure 4F). However, 1-year post-vaccination, this group also showed some enhancement of the VZV-specific CD4 CM T-cells (Figure 4D). Overall, these responses suggest differential induction of T-cell responses in participants with low and high pre-CMI, with a more CD4 T-cell oriented response in participants with low pre-CMI as compared with a more CD8 TemRA oriented response in participants with high pre-CMI.

Finally, we determined whether the blood leukocyte responses were different between participants with low and high pre-CMI (Figure S5 in Supplementary Material; Figure 5A). Pre-vaccination, a significant difference was found in the CD4/CD8 T-cell ratio between the two groups: participants with low pre-CMI in general possessed a CD4/CD8 ratio above 3, whereas participants with high pre-CMI showed a ratio below 3. This possibly suggests a compositional difference in the immune phenotype between participants with low and high pre-CMI. Participants with low pre- CMI showed slightly higher numbers of CD4 T-cells (mainly naïve cells), whereas participants with high pre-CMI showed slightly higher numbers of CD8 T-cells. This difference in CD4/CD8 T-cell ratio continued to exist at the different time points after vaccination (Figures 5A,F). Correspondingly, at 14 days post-vaccination, significantly higher numbers of circulating CD4 naïve cells, CD4 TemRA cells, B-cells, and naïve mature B-cells were found in the participants with low pre-CMI (Figures 5A–E). This result indicates differential lymphocyte subset kinetics in participants with low and high pre-CMI 14 days post-vaccination and therefore suggests that these responses can be used as a short-term biomarker to predict the long-term vaccine responses in middle-aged adults.