Enhancement of cytokine‐driven NK cell IFN‐γ production after vaccination of HCMV infected Africans

Human cytomegalovirus (HCMV) infection drives the phenotypic and functional differentiation of NK cells, thereby influencing the responses of these cells after vaccination. NK cell functional differentiation is particularly advanced in African populations with universal exposure to HCMV. To investigate the impact of advanced differentiation on vaccine‐induced responses, we studied NK‐cell function before and after vaccination with Trivalent Influenza Vaccine (TIV) or diphtheria, tetanus, pertussis, inactivated poliovirus vaccine (DTPiP) in Africans with universal, lifelong HCMV exposure. In contrast to populations with lower prevalence of HCMV infection, no significant enhancement of NK‐cell responses (IFN‐γ, CD107a, CD25) occurred after in vitro re‐stimulation of post‐vaccination NK cells with TIV or DTPiP antigens compared to pre‐vaccination baseline cells. However, both vaccinations resulted in higher frequencies of NK cells producing IFN‐γ in response to exogenous IL‐12 with IL‐18, which persisted for up to 6 months. Enhanced cytokine responsiveness was restricted to less differentiated NK cells, with increased frequencies of IFN‐γ+ cells observed within CD56brightCD57−, CD56dimCD57−NKG2C− and CD56dimCD57−NKG2C+ NK‐cell subsets. These data suggest a common mechanism whereby different vaccines enhance NK cell IFN‐γ function in HCMV infected donors and raise the potential for further exploitation of NK cell “pre‐activation” to improve vaccine effectiveness.

Recent evidence increasingly suggests that NK cells can display certain characteristics of immunological memory. However, this phenomenon has been described in different models. First, murine cytomegalovirus-and hapten-induced memory NK cells were described in human and mice [12,13]. Second, cytokine-induced memory-like NK cells generated through the pre-activation of NK cells with interleukin (IL)-12/18/15 cytokines [14,15]. Finally, "adaptive" NK cells mainly identified in HCMV infected individuals with increased expression of CD57, leukocyte immunoglobulin-like receptor, and CD2 with or without NKG2C as well as decreased CD161 and CD7 and lack of the signaling adaptor FcɛRγI-, and transcription factors such as Syk, EAT-2 and promyelocytic leukemia zinc finger (PLZF) [11,[16][17][18]. All of these scenarios demonstrate characteristics of enhanced NK cell functionality post-infection and/or stimulation.
We, and others, have shown that NK cells make a significant contribution to vaccine-induced immune responses to intracellular pathogens, including viruses. In vitro stimulation of peripheral blood mononuclear cells (PBMC) from vaccinated individuals with vaccine antigens including rabies [5], malaria [19], tetanus toxoid [20], whole cell Bordetella pertussis [6], and inactivated influenza virus [2,6], results in NK cell degranulation (surface expression of CD107a), IFN-γ production and upregulation of IL-2Rα/ CD25. NK cell IFN-γ responses to vaccine antigens are enhanced after vaccination with trivalent influenza vaccines, rabies or therapeutic HIV, in a T cell IL-2 dependent process [2,5,7]. Moreover, antibodies contribute to NK cell CD107a and CD25 expression through formation of immune complexes and cross-linking of FcγRIII (CD16) and these responses may also be enhanced by vaccination [2,[21][22][23].
In particular, human cytomegalovirus (HCMV) is a known modulator of NK cell phenotype and function, predominantly associated with ligation of the activating C-type lectin-like receptor, CD94/NKG2C, by an HCMV peptide (UL40) bound to HLA-E, but also with expansions of NK cells expressing activating Killer cell Immunoglobulin-like Receptors (KIR) [10,[29][30][31].
HCMV-seropositive individuals accumulate "adaptive" NK cells, which are mostly CD56 dim , CD57 + and CD16 bright , and frequently (but not always) express NKG2C [3,11,29]. In keeping with this phenotype, adaptively expanded NK cells are less responsive to exogenous IL-12 and IL-18 but are highly efficient mediators of ADCC (including killing of HCMV infected target cells) and produce IFN-γ in response to cross-linking of CD16 [9,11].
We have observed that NK cells from HCMV-seropositive subjects make low frequency IFN-responses to killed or inactivated pathogens or vaccine antigens and that responses are not strongly enhanced by vaccination [2,6,20]. However, these impacts of HCMV-infection are not only due to skewing of NK cell populations towards an "adaptive" phenotype as IFNγ responses are lower in all NK subsets of HCMV seropositive individuals when compared to HCMV seronegative individuals [6].
The seroprevalence of HCMV increases with age and varies enormously between populations [32,33]. HCMV serostatus is therefore likely to be a significant confounder of comparisons of immune responsiveness over the life course and between populations. We have previously studied the phenotype and function of NK cells in a Gambian population (age range 1 and 49 years) where HCMV infection is essentially universal by the age of 12 months (>97% HCMV seropositive) [28]. In this population, differentiation of NK cells from CD56 bright through CD56 dim CD57 − to CD56 dim CD57 + occurs very early in life and is maximal by the age of 10 years; acquisition of CD57 is often associated with expression of NKG2C and NK cell differentiation is delayed in individuals lacking NKG2C due to a homozygous deletion of KLRC2, the locus encoding NKG2C [28].
We therefore investigated whether NK cells from Gambian children would produce IFN-γ in response to pathogenassociated vaccines and whether vaccination would significantly boost these responses. Moreover, we have recently observed that vaccination can enhance NK cell IFN-γ responses to exogenous cytokines (in a manner analogous to in vitro "pre-activation" of NK cells by IL-12 and IL-18 [14,15,34] in HCMV-seropositive but not HCMV-seronegative donors [2]. We therefore hypothesized that vaccination would enhance NK-cell responsiveness to cytokines in this HCMV-infected population. We investigated the impact of two different vaccinations on NK cell function and phenotype in The Gambia. First, we studied the NK cell response to primary vaccination with inactivated, trivalent, seasonal influenza vaccine (TIV) in a previously unvaccinated, age-stratified cohort. Second, the NK cell response to booster vaccination against diphtheria, tetanus, pertussis, and polio (DTPiP) was examined in adults who had been vaccinated in infancy. We observed only limited enhancement of TIV-driven T-cell-dependent NK-cell responses after vaccination in all age groups. However, significant enhancement of vaccine antigen independent responses to exogenous pro-inflammatory cytokines was observed, persisting for at least 6 months after vaccination.

High HCMV seroprevalence in Gambian children and adults
For the TIV study, complete sample sets (baseline, 1, 3, and 6 months post-vaccination) were obtained from 68 individuals; 22 children (2-6 years of age), 21 young adults (20-30 years or age) and 25 older adults (60-75 years of age) ( Table 1). Similar numbers of male and female subjects were recruited within each age group ( Table 1). As expected, HCMV was highly prevalent with only one child and one elderly adult being seronegative ( Table  1). Median anti-HCMV IgG concentrations were slightly, but not significantly, higher in children than 20-30 year young adults and were significantly higher among 60-75 year olds compared to young adults, indicating potential HCMV reactivation in the oldest individuals. The vast majority of subjects were also seropositive for EBV infection with a tendency toward increasing seroprevalence with increasing age but without any significant age-dependent effect on EBNA-1 IgG titres (Table 1). Additionally, the overall frequencies of individuals homozygous for the NKG2C wild type gene was 44.1%; 42.6% were heterozygous and 13.2% were homozygous for the NKG2C gene deletion (an allele frequency of 34.6%) consistent with known frequency in The Gambia [35] (Table 1).

Age-related changes in NK-cell differentiation phenotype
Both HCMV infection and age influence the differentiation and function of NK cells and may therefore affect vaccine responses [25,26,28]. PBMC collected at baseline (prior to vaccination) from participants in the influenza study were therefore analysed ex vivo for NK cell ( Fig. 1; flow cytometry gating strategies are shown for NK cells in Supporting Information Fig. 1, blood myeloid and lymphoid lineages in Supporting Information Fig. 2 and memory T cells in Supporting Information Fig. 3). The overall frequency of NK cells (CD3 − CD56 + ) among the peripheral lymphocyte population increased significantly with increasing age (Fig. 1A) and, within the NK cell population, the frequency of CD56 bright NK cells was significantly higher among children than among adults (Fig. 1B). While there was a gradated increase in the frequencies of cells expressing the late differentiation marker CD57 (Fig. 1C), similar frequencies of NKG2C + NK cells were observed at all ages (Fig. 1D). As expected, the frequency of cells expressing NKG2A decreased, and the frequency of cells expressing NKG2C increased, as NK cells differentiated from CD56 bright via CD56 dim CD57 − to CD56 dim CD57 + (Fig. 1E and F). No significant difference was observed in the frequency of highly differentiated CD57 + NKG2C + NK cells between children and adults in this cohort ( Fig. 1E and F).
B-cell frequencies were significantly higher in 2-6 year-old children than in adults and there was a tendency for the frequencies of blood myeloid cell populations to increase with age Boxes indicate median values with interquartile ranges and whiskers indicate 95th percentiles. Statistical analysis was performed on samples using (A-D) Kruskal-Wallis test, *p < 0.05, **p < 0.01, ***p < 0.001 and (E,F) using linear trend ANOVA with correction for multiple comparisons ****p < 0.0001.
(Supporting Information Fig. 2). While the overall proportion of CD3 + T cells did not differ between age groups, both CD4 + and CD8 + T cells differentiated toward effector memory cell populations with increasing age (Supporting Information Fig. 3). There was a particularly marked accumulation of highly differentiated CD28 − CD57 + CD4 + T cells in the oldest age group (Supporting Information Fig. 3, E), consistent with previous observations in the elderly [36,37]. While the proportions of CD28 − CD57 + CD8 + T cells that were highest in the oldest age group, high frequencies were also present in children, as observed previously by ourselves and others (Supporting Information Fig. 3, J) [37,38].

Effect of vaccination on NK-cell responses to influenza vaccine antigens
We have previously observed, in UK subjects, that natural exposure to influenza, or vaccination with TIV, promotes T-cell-dependent NK-cell IFN-γ responses and antibody dependent NK cell degranulation [2,6]. Importantly, upregulation of CD25 and production of IFN-γ by NK cells after in vitro restimulation with vaccine antigens was consistently greater among HCMV seronegative than HCMV seropositive subjects, whereas degranulation responses were relatively unaffected by HCMV infection [2,6]. Thus, given the very high prevalence of HCMV infection in The Gambia (Table 1), we hypothesized that vaccination of Gambian subjects with TIV might potentiate antigen/antibody-induced degranulation responses but not IFN-γ production. A potential exception to such a response pattern could be in 2-6 year-old children, among whom the effects of persistent HCMV infection may not yet be fully apparent.
In vitro re-stimulation of PBMC with TIV antigen revealed only limited induction of IFN-γ and CD25 compared to the (background) response to adjuvanting low concentrations of IL-12 + IL-18 alone ( Fig. 2A and C). Significant induction of NK cell CD107a expression in response to TIV was, however, observed at both baseline and all post-vaccination time points, although there was no significant impact of vaccination (Fig. 2B). A similar pattern of responses was observed to antigen alone, in the absence of adjuvanting cytokines (data not shown). By comparison, NK cells from both HCMV seropositive and seronegative UK subjects made significant IFN-γ and CD25 responses to TIV, although these were enhanced by vaccination only in HCMV seronegative donors ( Fig. 2D and F). Moreover, both HCMV seropositive and seronegative UK donors made more potent CD107a responses to TIV than did the Gambian donors, although these were not enhanced after vaccination (Fig. 2E). None of the responses to TIV differed significantly between the different age groups of Gambian subjects, potentially reflecting the very early functional differentiation of NK cells in this population ( Fig. 1) [28].
Similar patterns of NK-cell responses occurred in Gambian adults vaccinated with DTPiP, with minimal NK-cell responsiveness to DTPiP antigens either before or after vaccination (data not shown). Overall, these data are consistent with loss of antigendriven NK cell IFN-γ responses in HCMV seropositive individuals while maintaining degranulation responses [6]. There was limited induction of IL-2 from CD4 + T cells, which could, in part explain the paucity of CD4 + T-cell-dependent NK-cell responses on restimulation with influenza antigen in vitro (Supporting Information Fig. 4).

Impact of post-vaccination antibody on vaccine antigen-induced responses
Previous studies have shown a strong dependence of vaccine or influenza virus-driven NK cell degranulation (and CD25 induction) on serum IgG antibodies [2,6,21,22]. We therefore tested sera collected after vaccination could promote or enhance vaccineantigen-driven NK-cell responses.
Anti-TIV IgG concentrations were significantly boosted by vaccination in children (baseline, median 574 AEU/mL and interquartile range (IQR) 460-653; 4 weeks, 1033 and 908-1073) and in young adults (baseline, median 722 and 402-879; 4 weeks, 975 and 759-1104), whereas no significant increase was observed in older adults (baseline, median 723 and IQR 622-862; 4 weeks, 813 and 583-1031) (Fig. 3A-C). High median levels of IgG were present at baseline in all age groups, consistent with widespread natural exposure to influenza within the study population. Similar anti-TIV specific antibody levels were observed in a TIV vaccinated UK cohort (baseline, median titre = 430, 299-767; 4 weeks, 962, 398-961). As shown for cells cultured in pooled control sera (Fig. 2), strong degranulation responses were observed in pre and post vaccination NK cells cultured with autologous plasma collected either at baseline or 4 weeks after vaccination (Fig. 3D). Similar observations were made in TIV vaccinated UK donors (Fig. 3E). However, no additional induction of CD107a expression was observed after vaccination, confirming that in both study cohorts, sufficient antibody was already present at baseline to drive these responses ( Fig. 3D and E). TIV-induced degranulation in Gambians was associated with the down-regulation of CD16, which was further enhanced in post-vaccination plasma, as previous described in UK donors [22]

Vaccination enhances NK-cell IFN-γ responses to accessory cytokines
In UK subjects, influenza vaccination induces intrinsic changes in NK cells characteristic of "pre-activated" NK cells; IFN-γ and degranulation responses to exogenous cytokines were enhanced after vaccination [2].
To assess whether enhancement of these responses also occurred after TIV or DTPiP vaccination of Gambians with lifelong HCMV infection, pre and post-vaccination PBMC were cultured with high concentrations of IL-12 (5 ng/mL) plus IL-18 (50 ng/mL) (HCC). Significant enhancement of NK-cell IFN-γ responses to HCC were observed by 4 weeks after TIV vaccination, which was stable and detectable for at least 24 weeks (Fig. 4A). By comparison, the responses of UK donors were only enhanced after vaccination for HCMV+ individuals, who had significantly lower baseline HCC-stimulated IFN-γ production compared to HCMV-individuals (Fig. 4B). In Gambians, this effect was highly significant among 60-75 year old individuals (Fig. 4C).  Paired statistical analysis was performed using Wilcoxon signedrank test and comparisons between groups were made using Mann-Whitney U test.*p < 0.05, **p < 0.01, ***p < 0.001,****p < 0.0001. production had high baseline production (median 10.9%, range 2.04-22.8%) which tended to decrease after vaccination (median 6.0%, range 1.3-13.3%). Comparing the distribution of NK cell subsets at baseline between these vaccine 'responder' and 'nonresponder' groups, individuals where boosting of the HCC driven IFN-γ response was observed after vaccination tended to have lower frequencies of CD57-cells at baseline, a trend also observed for children alone (Fig. 4D and E).
Enhancement of responses to exogenous cytokines also occurred 4 weeks after DTPiP booster vaccination in adults (Fig. 4F). In contrast to previous observations in UK donors, IL-12/IL-18-induced NK cell CD107a or CD25 expression was not

Vaccination-induced enhancement of IFN-γ production among CD56 bright and CD56 dim CD57 − subsets
As cytokine responsiveness varies by NK-cell differentiation status [20], we analyzed the distribution of vaccine-enhanced responses within NK-cell subsets. As expected, IFN-γ producing cells were more frequent in the CD56 bright NK cell subset than in the CD56 dim CD57 + NK cell subset, with intermediate frequencies in CD56 dim CD57 − cells (Fig. 5A). Moreover, the proportion of cells producing IFN-γ in response to IL-12/IL-18 was significantly enhanced 4 weeks post-vaccination in both the CD56 bright and the CD56 dim CD57 − subsets but not in the CD56 dim CD57 + sub-set (Fig. 5A). Strikingly, when grouping NK cells according to expression of CD57 and/or NKG2C, only the CD57 − NKG2C − and CD57 − NKG2C + CD56 dim demonstrated increased frequencies of IFN-γ producing cells after vaccination (4 weeks, p = 0.004) which persisted up to 24 weeks (Fig. 5C). This enhancement was not due to enrichment of CD57/NKG2C subsets, since no change was observed in the proportions of CD56, CD57 and NKG2C-defined subsets after vaccination (Supporting Information Fig. 5F). Significant enhancement of cytokine-driven IFN-γ responses was also observed within CD57-(NKG2C-or NKG2C+) NK cells for the 2-6 and 60-75 year old age groups (Supporting Information Fig.  6). Consistent with the effects of TIV, DTPiP booster vaccination of adults resulted in significant enhancement of the IFN-γ response to rIL-12/IL-18 among CD56 dim CD57 − NKG2C + NK cells (p = 0.009, 4 weeks) and there was a trend for enhanced responsiveness among CD56 dim CD57 − NKG2C − cells (Fig. 5D).

Discussion
In contrast to the enhancement of CD4 T-cell-dependent NK-cell responses to vaccine antigens in previous studies of UK vaccinees [2,5], we consistently failed to detect TIV or DTPiP vaccine antigen-driven NK cell IFN-γ or CD25 responses in Gambian donors. Although antigen-driven CD107a responses were detected, these were not boosted by vaccination. The very high prevalence of HCMV among the Gambian population may partially explain these observations since augmentation of NK cell responses by vaccination was significantly more pronounced among HCMV-than among HCMV + UK subjects [2]. Nonetheless, the IFN-γ and CD25 responses to vaccine antigens among NK cells from Gambian donors are typically much lower than those of HCMV + UK donors, suggesting that HCMV infection status alone does not fully explain the lack of response after vaccination. In addition, we have previously observed NK cell phenotypical and functional discrepancies between HCMV + Gambian and UK donors (23). Further studies are required to determine whether differences in the timing of HCMV infection, differences in strains of infecting virus, or other factors including co-infections, might explain this lack of vaccine antigen-driven NK function in Gambians of all ages.
Despite the lack of "boosting" of antigen-induced NK cell responses after vaccination in Gambian subjects, this study provides further evidence for enhancement of cytokine-driven NK cell IFN-γ responses. TIV, DTPiP and likely other vaccines may enhance these responses through a common mechanism, with similar mechanisms potentially underlying protracted enhancement of NK cell responses after BCG booster vaccination in other African populations [39] and in European populations after yellow fever vaccination [40]. In this study, such vaccine-stimulated NK cells persist for at least 6 months whereas in our UK-based studies, these effects were lost by 3 months. Possible explanations for this difference could be that responses in Gambian individuals are boosted by more frequent exposures to environmental pathogens or that the relative stronger responses of both HCMV − and HCMV + UK donors are less amenable to sustained enhancement.
Restriction of this effect to the CD56 bright , CD56 dim CD57 − NKG2C − and CD56 dim CD57 − NKG2C + subset of NK cells is reminiscent of the less differentiated NK cells generated in vitro by cytokine pre-activation [14,41]. Heightened cytokine-driven IFN-γ responses within CD56 dim CD57 − NKG2C −/+ NK cells also suggests that these cells are equivalent to the less differentiated CD57 − cells observed in UK donors with enhanced cytokine sensitivity and higher intrinsic proliferative capacity (Ki67) after influenza vaccination [2]. These data are also consistent with the acquisition of NKG2C, and expansion of NKG2C + cells, prior to functional differentiation and CD57 acquisition [28]. Interestingly, TIV vaccinated individuals showing enhanced responses to cytokines after vaccination tended to have fewer cytokine-driven IFN-γ + NK cells and of CD56 dim CD57 − cells at baseline compared to those with no boosting. Differences in cytokine driven responses observed at baseline were, however, not attributable to ongoing environmental background activation as we saw no differences in ex-vivo expression of CD25, CD107a or IFN-γ comparing 'responder' and 'non-responder' groups. This pattern of responses parallels that observed in UK subjects where enhancement of cytokine driven responses after TIV vaccination occurs principally in HCMV + individuals with suboptimal baseline responses and not in HCMV-individuals, who already have maximal frequencies of IFN-γ + cells [2].
Differentiation subsets of NK cells could have varying thresholds of cytokine (Interleukins-2, 12, 15, and 18) responsiveness, thereby influencing the effective potency of the different cytokines induced by vaccination. The acquisition of CD57 correlates with decreased expression of IL-12Rβ2 and IL-18Rα [20]. However, when comparing HCMV + to HCMV − individuals we have observed lower responsiveness within each CD57 defined subset, which cannot be explained by cytokine receptor expression alone [2,24].
Vaccine-induced enhancement of IFN-γ secretion in less differentiated NK cells could reflect preferential proliferative expansion, epigenetic modifications or a combination of both of these processes. Epigenetic modifications have been demonstrated in human and murine NK cells after cytokine pre-activation which, in turn, promote increased IFN-γ transcription [42,43]. One of the limitations of this study is that, we had insufficient available paired pre and post-vaccination cells investigate whether CD56 dim CD57/NKG2C + cells are particularly sensitive to vaccine induced epigenetic modifications. Examination of other NK cell cytokines such as tumor necrosis factor-α and granulocytemacrophage colony-stimulating factor would also reveal whether vaccine-induced enhancement is specific for IFN-γ production.
In summary, two very different vaccines are able to re-set the threshold for human NK cell IFN-γ responses to cytokines (in an antigen-independent manner) in individuals where HCMV has had a major effect on NK-cell functional differentiation. Enhancement of cytokine-driven NK-cell IFN-γ responses is particularly evident in elderly individuals and within a distinct population of CD56 dim CD57 − NKG2C + NK cells that accumulate in HCMV-infected individuals. This phenomenon may partially compensate for the terminal differentiation of NK cells, away from cytokine responsiveness, observed in HCMV seropositive donors [3,11,25,26,28] and may also contribute to the expansion of NKG2C + NK cell populations that is observed among HCMV-infected subjects after systemic viral infections [44][45][46][47]. Vaccine-driven enhancement of effector functions in CD56 dim CD57 − NKG2C − and CD56 dim CD57 − NKG2C + NK merits further investigation as a route to improve vaccine efficacy in HCMV infected elderly individuals.

Study subjects
The In order to minimize the impact of natural exposure to influenza during the follow-up period, sample collection times fell outside the main Gambian influenza season [48]. Following written informed consent, 68 healthy participants were recruited from three villages (Keneba, Manduar, and Kantong Kunda) in West Kiang, The Gambia. Participants fell into one of three age groups: 2-6, 20-30 and 60-75 years of age. Individuals with chronic disease, infections, or influenza-like signs and symptoms during the previous 3 months or with an axillary temperature of ࣙ 38°C were excluded. Pregnant women, individuals potentially allergic to vaccine products, and those with a history of previous influenza vaccination were also excluded. Venous blood was collected at baseline (pre-vaccination) and partici- For vaccination with tetravalent DTPiP, 18 adult males between the ages of 21 and 29 years (median age = 24.2 years) were recruited. All subjects previously received DTP and oral polio vaccine as children and had not since been boosted. This study was conducted in Sukuta, West Coast Region, The Gambia. A single dose of DTPiP (Repevax TM Sanofi Pasteur MSD) was given intramuscularly after collecting a baseline blood sample. A follow-up blood sample was collected 4 weeks later.

Serology
Baseline samples from the influenza study were screened for IgG antibodies to HCMV (Biokit, Barcelona, Spain) and IgG antibody against Epstein Barr virus (EBV) nuclear antigen-1 (EBNA-1) (Euroimmun, Lubeck, Germany). Seropositivity was determined according to HCMV-and EBV-negative standards provided by the manufacturer. Influenza vaccine antigen-specific IgG was determined by an in-house ELISA as previously described [2,6].

Cell preparation and culture
Peripheral Blood mononuclear cells (PBMC) were prepared by centrifugation of whole blood over Lymphoprep TM (Sigma, UK). PBMC were reserved for ex vivo staining or cryo-preserved in freezing medium (5% DMSO in Fetal Calf Serum, FCS) at -80°C (1 × 10 7 PBMC per vial) overnight in Mr. Frosty TM Freezing Containers, prior to transfer into liquid nitrogen. Cells were recovered by rapid thawing into culture medium and repeated washing, and rested for 3 h prior to stimulation. Viability of recovered cells was >95% and yields ranged from 30 to 65%.
In vitro restimulation assays were performed simultaneously on cells from all pre and post vaccination time points for a given individual. PBMC (3 × 10 5 / well) were cultured for 18 h in 96well U-bottom plates (Costar R cell culture plates, USA) at 37°C, 5% CO 2. For T-cell assays, cells were stimulated for a total of 5 h. TIV and DTPiP vaccine antigens were used at final concentrations of 2.5 and 0.5 μg/mL, respectively. Low Concentration of Cytokines (LCC) comprised 12.5 pg/mL recombinant rIL-12 (Peprotech, London, UK) + 10 ng/mL rIL-18 (R&D systems, Abingdon, UK). High Concentration of Cytokine (HCC) comprised 5 ng/mL rIL-12 + 50 ng/mL rIL-18. GolgiStop, containing Monensin and GolgiPlug, containing Brefeldin A (both from BD Biosciences, Oxford, UK) were added after 2 h for T cell and after 15 h for NKcell assays. Culture medium was RPMI 1640 supplemented with Penicillin-Streptomycin-Glutamine (Gibco, UK) and 10% pooled human male AB serum (Sigma-Aldrich R , Saint Louis, USA) unless otherwise stated.

Statistical analysis
Nonparametric Wilcoxon matched paired tests were used for intra-group comparisons between pre and post vaccination data. Kruskal-Wallis tests were used for unpaired comparisons between age groups, for nonpaired analysis Mann-Whitney U test was used. Linear trend analyses were performed using repeated measures ANOVA. GraphPad Prism (GraphPad Software 6) was used for statistical analysis. Significant differences are reported as *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.