Blood in EDTA tubes was analysed for immune cell parameters using the Ac.T 5 diff CP haematology analyser (Beckman Coulter) following the manufacturer’s instructions. IgG antibody levels to the EBV viral capsid protein VCA were measured in plasma using a multiplex bead assay on a Luminex BioRad Bio-plex200 system as previously reported (27).

IFN-γ T cell responses to a cocktail of latent and lytic EBV peptides ( Supplementary Table 1 ) were measured using the ELISpot assay. The MABTECH Human IFN-γ ELISpot kit (Code: 3420-2AST-2) was used for the assay, with a few alterations to the manufacturer’s protocol. Briefly, the ELISpot plates with the capture antibody from the kit were washed five times with 200μl of 1xPBS per well. Afterwards, thawed cells were added to the plates in a volume of 100μl AIM-V medium containing 150,000 cells per well. The plates were covered with the lid, wrapped in aluminium foil and transferred to a 5% CO₂ 37°C incubator for a 24-hour resting period. To stimulate them, 100μl per well of the EBV peptide pool, anti-CD3 and media (AIM-V media, Gibco 12055091) at working concentrations of 5μg/ml/peptide were added to the wells. The plates were then incubated at 5% CO₂ 37°C for a further 46-48 hours. Following stimulation, cells were washed 5 times with 200μl of PBS per well and 100μl of anti-human IFN-γ IgG conjugated to alkaline phosphatase (Code: 7-B6-ALP) was added at a dilution of 1/200 in PBS + 0.5% FBS. The plates were incubated at room temperature (25°C) for 2 hours. After the incubation, the plates were washed 5 times with 200μl of 1xPBS per well and 100μl of filtered 5-bromo-4-chromo-3-indolyl-phosphate (BCIP)/nitroblue tetrazolium (NBT)-plus substrate were added per well. The plates were then incubated at room temperature for 6.5 minutes and the reaction was stopped by washing the plate with running tap water. The plates were dried in the dark overnight and the spots were subsequently counted using an ELISpot reader (CTL ImmunoSpot Analyzer). This protocol has been reported elsewhere (28, 29).

Fluorochrome antibody conjugate titration and reference control type selection were carried out prior to study participants’ PBMCs staining ( Table 1 ). For conventional flow cytometry, beads (BD CompBeads (BD Biosciences, 552843) were used for compensation of all fluorochrome antibody conjugates apart from the live/dead stain where PBMCs were used ( Table 1 ). The most appropriate reference control type beads (Ultra Comp eBeads Invitrogen, 01-2222-42) or PBMCs were used for full spectrum flow cytometry ( Table 1 ). Live/dead staining using the fixable viability dye eFluor 780 (eBioscience) for conventional flow cytometry, or fixable blue dead stain kit for full spectrum flow cytometry (Thermo Fisher) was carried out in 1mL of PBS containing 1 million PBMCs. IgG F_C receptor (F_CR) blocking was performed prior to fluorochrome antibody conjugate staining using a human F_CR binding inhibitor (eBioscience). Fluorochrome antibody conjugate cocktails were made in FACS buffer (1X PBS, 0.5% BSA, 0.5M EDTA and 0.05% sodium-azide) using the predetermined optimal concentration ( Table 1 ). Brilliant stain buffer (BD Biosciences 566349) was added to the antibody cocktail following the manufacturer’s recommendations. Study participants’ PBMCs were stained with a cocktail of fluorochrome antibody conjugates in 100ul or 50ul of FACS buffer for full spectrum or conventional flow cytometry, respectively, for 30 minutes at 4^°C. Stained PBMCs were fixed using the FluoroFix buffer (BioLegend, 422101) before acquisition on BD LSR-II flow cytometer (conventional flow cytometer) or a 5-laser Cytek Aurora (full spectrum flow cytometer). A total of 200,000 events from each study participant sample were recorded.

Three different panels were used to identify T and B cells using conventional flow cytometry. Panel one contained CD3, CD4, CD8, CCR7, CD45RA and live/dead, panel two contained CD3, CD4, CD8, CD57, CD28, HLADR, PD-1 and live/dead. Panel three contained CD19, CD10, CD5, CD27, IgD, CD21, CD38, IgM, IgG and the live/dead stain. Single cells were gated using forward scatter area and forward scatter height. Lymphocytes were gated using side scatter and forward scatter followed by the exclusion of dead cells using the live/dead stain ( Supplementary Figure 1 and Supplementary Table 1 ).

Using a single panel of 23 antibody-fluorochrome conjugates and one live/dead stain (fixable blue dead cell stain kit (Thermo Fisher, L34961) T cell, B cell, NK cell and monocyte subsets were identified ( Figure 1 ; Supplementary Table 1 ). Single cells were gated using forward scatter area and forward scatter height. Lymphocytes and monocytes were gated using side scatter and forward scatter followed by exclusion of dead cells using the live/dead stain ( Figure 1 ). Using CD3 and CD19 three main subsets were classified including CD3+ (T cells) CD19+ (B cells) and CD19-CD3- (NK cells and monocytes).

EBV DNA was quantified in PBMCs and saliva using primers (Balf5 EBV forward: 5’ – CGG AAG CCC TCT GGA CTT C – 3’, - Balf5 EBV reverse: 5’ – CCC TGT TTA TCC GAT GGA ATG – 3’) and probe (Balf5 EBV Probe: 5’ -/56-FAM/TGT ACA CGC ACG AGA AAT GCG CCT/3BHQ_1/- 3’) previously reported to be specific to the Balf5 gene (6, 30). Additionally, B-Actin was amplified in the same sample as an internal positive control using primers (B-Actin forward: 5’ – TCA CCC ACA CTG TGC CCA TCT ACG A – 3’, B-Actin reverse: 5’ – CAG CGG AAC CGC TCA TTG CCA ATG G – 3’) and probe (B-Actin Probe: 5’ -/5HEX/ATG CCC TCC CCC ATG CCA TCC TGC GT/3BHQ_1/- 3’) as previously reported (31).

We compared CD4+ and CD8+ T cell subsets in the 4-15, 16-30, 31-55 and 56-89 age groups using the Kruskal Wallis test. Overall, the percentage of naïve CD4+ and CD8+ T cells decreased with increasing age groups ( Figures 2A, C , 3A, C ; Supplementary Figures 2 - 5 , Supplementary Tables 2 , 3 ). The median of naïve CD4+ T cells was 46% interquartile range-IQR (41-54) of total CD4+ T cells in the 4-15 age group, 28% IQR (25-37) in the 16-30 age group, 30% IQR (18-34) in the 31-55 age group and 26% IQR (15- 32) in the 56-89 age group ( Figure 3A ; Supplementary Table 3 ). The median percentage of naïve CD8+ T cells of total CD8+ T cells was 39% IQR (31-45) in the 4-15 age group, 34% IQR (23-38) in the 16-30 age group, 28% IQR (22- 32) in the 31-55 age group and 30% IQR (24-31) in the 56-89 age group ( Figure 3C ). Overall, percentages of central memory CD4+ and CD8+ T cells increased with increasing age groups ( Figures 2A, C , 3A, C ). Individuals aged 4-15 years had a lower percentage of effector memory CD4+ T cells compared to their older counterparts ( Figure 3A ). The percentage of effector memory CD4+ T cells didn’t change between 16 to 89 years ( Figures 2A , 3A ). When age was analysed continuously, the percentages of effector memory CD4+ T cells slightly increased with increasing age ( Supplementary Figures 2 , 3 ). The percentage of effector memory CD8+ T cells increased with increasing age groups ( Figure 2C ). The percentage of total CD4+ T cells, total CD8+ T cells and T_EMRA didn’t change by age group ( Figures 2A, C , 3A, C ). However CD4+ T cells, CD8+ T cells and T_EMRA had a weak positive correlation with increasing age ( Supplementary Figures 2 - 5 ). However, the ratio of CD4+:CD8+ T cells was highest in the 16-30 age groups, and lowest in the 31-89 groups ( Figure 4C ). Direct comparison between conventional flow cytometry and full spectrum flow cytometry was not possible with the current data, however, CD4+ naïve CD4+, effector memory CD4+, T_EMRA CD4+, naïve CD8+, central memory CD8+, T_EMRA CD8+ T cells followed a similar pattern in both results from the two flow cytometry methods. On the other hand, the pattern of central memory CD4+, CD8+, effector memory CD8+ by full spectrum flow cytometry was different from the pattern of the same cell subsets by conventional flow cytometry ( Figures 2 , 3 ; Supplementary Tables 2 , 3 ). We were not equipped to investigate the differences in cell sub sets by the two flow cytometry methods due to sample limitations but these differences could be partially attributed to the inclusion of the 4-15 years age group in conventional flow cytometry analysis but not the full spectrum flow cytometry.

We next obtained the percentages of CD4+ and CD8+ senescent (CD57+ CD28-), activated (HLADR+CD38+) and exhausted (PD1+ or LAG3+) subsets from the total CD4+ and CD8+ T cells respectively. Whilst Follicular Helper T cells do express PD1, they don’t express LAG3. Whilst we cannot rule out TFH cells, we can be reasonably sure that LAG3 expressing cells are not this subset and are likely exhausted. Although percentages of senescent CD8+ T cells were higher than the percentages of senescent CD4+ T cells [median 27% IQR (18.6 – 39.7) vs. 1% IQR (0.37 – 2.85)], both senescent CD8+ and CD4+ T cells increased with increasing age ( Figures 3B, D ). The percentages of activated (HLADR+) CD4+ and CD8+ T cells were low and increased with increasing age groups ( Figures 3B, D ). The percentages of exhausted T cells were very low and did not change with increasing age ( Figures 2B, D , 3B, D ).

Concurrently, we obtained the percentages of B cell subsets out of the total CD19+ B cells and compared them the 4-15, 16-30, 31-55 and 56-89 age groups using the Kruskal Wallis test. The percentages of total CD19+ B cells decreased with increasing age groups ( Figure 3E ). Naïve B cells also reduced with increasing age groups ( Figures 2E , 3E ). Resting naïve B cells were relatively high in individuals aged 4-55 and lower in 56-89 year-olds ( Figure 3E ). Activated naïve B cells increased with age ( Figure 3E ). Atypical B cells are mature B cells double negative for both CD27 and IgD. These double negative (DN) mature B cells were classified into DN1 and DN2 using CD38 and CD21 (32). Atypical B cells, DN2 and IgG+ double negative B cells increased with increasing age groups ( Figure 3F ). DN1 B cells decreased with increasing age groups ( Figure 2F ). DN IgM+ B cells increased with increasing age groups and started reducing in the 31-55 age group ( Figure 3F ). Memory B cells increased with increasing age groups and dropped in the 56-89 age group ( Figure 3G ). Activated memory B cells increased with increasing age groups while IgG+ memory B cells increased with increasing age groups and reduced in the 56-89 age group ( Figure 3G ). Other B cell subsets did not change over time. The pattern of naïve B cells, CD38+ CD21+ naïve B cells with increasing age, measured using full spectrum flow cytometry were similar to the pattern of the same B cell subsets measured using conventional flow cytometry. However, the pattern of CD19+ CD38- CD21- naïve B cells, DN, DN1, DN2 measured using full spectrum flow cytometry were different to the pattern of the same B cell subsets measured using conventional flow cytometry ( Figures 2 , 3 ; Supplementary Figures 6 - 8 ).

Using the Kruskal Wallis test, NK cells and monocyte subsets did not change with changing age ( Figures 2G, H ; Supplementary Figure 9 ). Neutrophils and lymphocytes obtained from full blood counts were the most abundant and basophils were the least abundant ( Figure 4 ). Using the one-way ANOVA, neutrophils increased while lymphocytes decreased with age, both plateauing in the 16-30 age group ( Figure 4A ). Both eosinophils and basophils from full blood counts decreased with increasing age groups, plateauing at the 6-30 year age group, while monocytes did not change in the different age groups. Although the different T, B, NK cells, neutrophils and monocyte subsets were not different between males and females ( Supplementary Figures 10 , 11 ) using a student T test, eosinophil and basophil percentages were higher in males compared to females ( Figure 4B ).

All individuals tested had antibodies to the EBV VCA antigen, implying that all were infected with EBV. Using the Chi² test, the percentage of individuals with a positive T cell response to EBV was highest in individuals aged 31-55 years and lowest in the youngest age group (3-12 years) ( Figure 5A ). After adjusting for sex and testing batch using logistic regression modelling, individuals in the older age groups were more likely to have a positive EBV T cell response compared to the youngest age group (3-12 years) ( Figure 5B ). Nonetheless, using the Wilcoxon Ranksum test individuals with a positive EBV T cell response had lower percentages of exhausted CD8+ T cells (LAG3+) compared to those without a detectable EBV T cell response ( Figure 5C ). No difference in other cell types were observed between individuals with and without an EBV T cell response ( Figure 5C ; Supplementary Figure 12C ). Furthermore, the proportion of naïve B cells but not memory or atypical B cells ( Supplementary Figures 12A, B ) negatively correlated (Sperman’s rank correlation) with the amount of IgG to EBV-VCA antigen ( Figure 5D ). There was no association between EBV viral load in PBMCs and saliva with the presence of a positive T cell response to EBV ( Supplementary Figure 12D ). Furthermore, EBV viral load in PBMCs was not associated with the frequency of T cell subsets ( Supplementary Figure 12E ).