Differences in Influenza-Specific CD4 T-Cell Mediated Immunity Following Acute Infection Versus Inactivated Vaccination in Children

Abstract

Background

Early childhood influenza infections imprint influenza-specific immune memory, with most studies evaluating antibody specificity. In this study, we examined how infection versus inactivated influenza vaccination (IIV) establish pediatric CD4 T-cell mediated immunity to influenza and whether this poises the immune system to respond differently to IIV the following year.

Methods

We tracked influenza-specific CD4 T-cell responses in 16 H3N2 infected and 28 IIV immunized children following both initial exposure and after cohorts were revaccinated with IIV the following fall. PBMCs were stimulated with peptide pools encompassing the translated regions of the H3 HA and NP proteins and were then stained to assess CD4 T-cell specificity and function.

Results

Compared to IIV, infection primed a greater magnitude CD4 T-cell response specific for the infecting HA and NP proteins, with more robust NP-specific immunity persisting through year 2. Post infection, CD4 T cells preferentially produced combinations of cytokines that included interferon-γ. Interestingly, age-specific patterns in CD4 T-cell reactivity demonstrated the impact of multiple influenza exposures over time.

Conclusions

These data indicate that infection and vaccination differentially prime influenza-specific CD4 T-cell responses in early childhood, with these differences contributing to the lasting immunologic imprinting established following early influenza infection.

Clinical Trials Registration

NCT02559505.

In this study, influenza-specific CD4 T-cell responses were compared in children either acutely infected with influenza or vaccinated with IIV. H3- and NP-specific CD4 T-cell reactivity was found to vary depending on both subject exposure history and age.

Influenza causes yearly epidemics and intermittent pandemics with a substantial burden of disease in the pediatric population. Children are first exposed to this virus in early childhood, with historically most exposures occurring through infection with a seasonal influenza strain. Both past and more recent studies have demonstrated the importance of these early life influenza exposures in shaping long-term influenza-specific immunologic memory. Historically, the preferential boosting of antibodies specific for earlier influenza strains upon repeated viral encounter was termed “original antigenic sin” [1–3]. More recently, the importance of early influenza infections in establishing lifelong biases in immune memory able to shape seasonal influenza risk patterns within a given influenza subtype [4] and provide some degree of heterosubtypic protection against viruses within the same hemagglutinin (HA) group [5] has been established. The ability of early influenza exposures to shape lifelong immunity has been termed imprinting, with our data as well as others confirming the importance of early immunologic memory in shaping subsequent responses to both infection and vaccination [6–10].

While historic and recent data demonstrate the unique importance of early life influenza infections in establishing lasting anti-influenza immunologic memory, many countries, including the United States, have instituted recommendations for universal inactivated influenza vaccination (IIV) starting at the age of 6 months [11]. While early vaccination provides critical protection from severe influenza disease in infants and young children, it also results in increasing numbers of children receiving IIV prior to their first natural infection. It is not yet known how immunity established by IIV differs from that established through an active infection. Further, the impact of early acute infection on immunity subsequently elicited by IIV, and how initial IIV influences immunity to a later influenza infection, remains unknown.

CD4 T cells provide multifaceted protection against influenza through the provision of help for B-cell and CD8 T-cell responses, early recruitment of innate immune effectors to the lung, and independent cytotoxicity [12–16]. Importantly, our previous work has demonstrated that in the setting of preexisting CD4 T-cell memory, responses to conserved CD4 T-cell epitopes were boosted at the expense of responses to unique epitopes, influencing the specificity of the resulting antibody response [10, 17]. This provides evidence that preexisting immunity can bias the CD4 T-cell compartment and supports a potential role for CD4 T cells in imprinting. Given the multiplicity of CD4 T-cell functions and their potential to cross-react with diverse influenza strains [10, 18–20] and provide independent protection from severe disease [21], an improved understanding of how early life CD4 T-cell–mediated immunity is shaped following infection versus vaccination is critical to the development of next-generation influenza vaccines.

In this article, we compared CD4 T-cell reactivity between cohorts of children either acutely infected with H3N2 influenza or vaccinated with IIV. Compared to IIV, acute infection primed a greater magnitude CD4 T-cell response specific for the infecting HA and NP proteins, with the postinfection boost in NP- but not H3-specific reactivity persisting following vaccination in study year 2. CD4 T-cell functionality also varied depending on the route of influenza exposure in study year 1, with CD4 T cells primed by infection as opposed to vaccination preferentially producing combinations of cytokines that included interferon-γ (IFN-γ). Marked differences in immunity were noted with age, suggesting that the CD4 T-cell response to influenza infection is shaped by repeated exposures over time and supporting the critical importance of early immunity in establishing long-term anti-influenza CD4 T-cell memory.

METHODS

Human Subjects

Following approval from the University of Rochester Research Subjects Review Board (protocol RSRB00058437), venous blood was obtained from 28 vaccinated and 16 acutely infected pediatric subjects between 2016 and 2018. These subjects were part of a larger cohort and consisted of all subjects acutely infected with H3N2 influenza with age-matched subjects vaccinated with seasonal Quadrivalent Fluzone (a split virus inactivated vaccine) tested concurrently. All subjects had influenza vaccination histories documented by review of the New York State Immunization Information System. Subjects had blood drawn 8 to 14 days and again 20 to 28 days following enrollment. All subjects were then followed and vaccinated with Quadrivalent Fluzone the following fall, with blood again obtained pre- and at days 8 to 14 and 20 to 28 post vaccination. In completion of this study, experimentation guidelines of the United States Department of Health and Human Services and the University of Rochester were followed and study procedures were performed in accordance with the ethical standards of the Helsinki Declaration. All parents provided written informed consent prior to study participation. The study was registered in the clinicaltrials.gov database (NCT02559505).

Isolation of PBMCs From Human Blood

Within 6 hours of isolation, blood was gently spun and plasma was removed. Peripheral blood mononuclear cells (PBMCs) were isolated using density gradient centrifugation (Ficoll-Paque Plus; GE Healthcare), counted, and frozen at a controlled rate in fetal bovine serum (Gibco) containing 10% dimethyl sulfoxide at approximately 5–10 million cells per mL. Frozen PBMCs were thawed into RPMI-1640 (Gibco) containing 10% fetal calf serum.

Synthetic Peptides and Libraries

Peptides, 17-mer overlapping by 11 amino acids, encompassing the entire translated sequences of the examined viral proteins were obtained from the Biodefense and Emerging Infections Research Repository, National Institute of Allergy and Infectious Diseases, National Institutes of Health. The peptide arrays included A/New York/384/2005(H3N2) hemagglutinin (H3; NR-2603) and A/New York/348/2003(H1N1) nucleoprotein (NP; NR-2611). Although the NP peptide array was derived from an H1N1 virus, it had 91.8% genomic identity and 96% amino acid sequence homology with the NP protein from the more recently circulating A/Michigan/29/2016 (H3N2) influenza virus. A peptide pool encompassing 111 peptides from Sin Nombre Virus (NM H10) glycoprotein precursor protein (NR-4764) was used as a negative control. All peptides from a given protein were pooled, with each peptide present at a final concentration of 1 μM in assays.

Intracellular Cytokine Staining

PBMCs were thawed and rested unstimulated for approximately 16–18 hours at 37°C and 5% CO₂. Cells were then counted and between 800 000 and 1.5 million PBMCs were stimulated with negative, H3, or NP peptide pools for 16 hours, with brefeldin A and monensin (BD Biosciences) added for the last 8 hours. Following stimulation, cells were washed and stained for 30 minutes with Fixable Live/Dead Aqua (ThermoFisher Scientific). Surface and intracellular staining were performed per published protocol [22]. Antibodies used included CD8 (RPA-T8), CD4 (RPA-T4), CD3 (SK7), CD45RA (HI100), IL-2 (MQ1-17H1), IFN-γ (B27), Ki67 (B56), and CD69 (FN50) (BD Biosciences) as well as tumor necrosis factor-α (TNF-α; MAb11) and CD19 (HIB19) (Biolegend), with cell fixation and permeabilization performed using the eBioscience FoxP3 Transcription Factor Fixation/Permeabilization Kit (ThermoFisher Scientific). Fluorochrome-stained UltraComp e-Beads (Invitrogen) and ArC Amine Reactive Compensation Beads (ThermoFisher Scientific) were used for compensation. Data were acquired using a BD LSR-II instrument, configured with 488, 633, 407, and 532-nm lasers using FACS DIVA software (BD Bioscience). Sequential gating was performed using FlowJo v10 (TreeStar). Activated cells of interest were defined as CD69⁺CD4⁺ T cells (CD3⁺CD8⁻) positive for cytokine staining (IFN-γ, IL-2, or TNF-α). Reactivity to the negative peptide pool was generally negative to extremely low when quantified by expression of IFN-γ and slightly higher for IL-2 and TNF-α.

Statistical Analysis

Throughout the manuscript, data are presented as percentage of live CD4 T cells with background subtracted. Pie graphs were created using SPICE software (NIAID), while bar and scatter graphs were created using Prism 8 (GraphPad Prism). A repeated ANCOVA was applied to study cytokine, polyfunctionality, and Ki67 data, controlling for age as a continuous variable. Data were then reexamined using a repeated ANOVA analysis, controlling for age as a group. Other demographics were either homogeneous between study groups (sex, race, and ethnicity) or nearly contained in group and age distribution, and thus were not included in repeated-measurement models. Plots demonstrate individual data, the unadjusted population mean (horizontal line), and estimated percentages from the statistical analysis with age controlled as an overlaid bar graph. For all analyses, a P value < .05 was considered significant. Statistical analyses were performed using Software SAS 9.4 (SAS Institute).

RESULTS

The goal of this study was to evaluate how early life CD4 T-cell reactivity is differentially primed by acute infection versus vaccination and determine the impact this has on the immune response to vaccination the subsequent year. We evaluated a cohort of 16 children between 3 months and 7 years of age enrolled upon infection with H3N2 influenza and compared these children to 28 children between 7 months and 7 years of age initially vaccinated with seasonal Quadrivalent Fluzone (Table 1). All subjects were longitudinally followed and reevaluated post vaccination with seasonal Quadrivalent Fluzone the following fall (Figure 1). As the vaccinated cohort consisted of children age-matched to all infected subjects, including the smaller numbers of children enrolled with influenza B and H1N1 infections, this cohort contained a greater subject number. Subjects acutely infected with H3N2 influenza were enrolled from the Golisano Children’s Hospital at Strong Pediatric Emergency Department in 2016–2017 (n = 12) and 2017–2018 (n = 4). The vaccinated cohort was largely enrolled and first vaccinated between September and December of 2016. More subjects in the vaccination cohort had completed the primary influenza vaccination series (71.4% compared to 31.25% acutely infected subjects). Cryopreserved PBMCs were evaluated by intracellular cytokine staining following stimulation with complete overlapping peptides pools representing the entire translated sequences of the H3 or NP proteins. Cells were gated on live, CD3⁺CD4⁺ cells to evaluate CD4 T-cell specificity and function, with cytokine production quantified as the percentage of activated (CD69+) cytokine-positive cells after subtracting background (Supplementary Figure 1).

Table 1.

Subject Demographic Data

Characteristic	Vaccinated	Acute (H3N2) Infection
Subjects
Total	28	16
Enrolled 2016–2017	26	12
Enrolled 2017–2018	2	4
Sex, male	13 (46)	8 (50)
Race
White	14 (50)	7 (44)
Black	8 (29)	4 (25)
Native American	1 (4)	0 (0)
Mixed	5 (18)	2 (13)
Not reported	0 (0)	3 (19)
Ethnicity
Non-Hispanic	25 (89)	11 (69)
Hispanic	3 (11)	5 (31)
Age
Mean, y	2.3	2.7
1A/B, 3–12 mo	8 (28.6)	2 (12.5)
2A/B, 13–35 mo	13 (46.4)	8 (50)
3A/B, 36 mo–5 y	4 (14.3)	4 (25)
4A/B, 6–8 y	3 (10.7)	2 (12.5)
Previous influenza vaccination
Complete primary series	20 (71.4)	5 (31.25)
Incomplete primary series	NA	3 (18.75)
Never	8 (28.6)	8 (50)

Data are No. (%) except where indicated.

A, vaccination cohort; B, infection cohort; NA, not applicable.

Figure 1.

Design of the study. Pediatric subjects between 3 months and 7 years of age were enrolled and vaccinated with IIV (n = 28) or enrolled upon presentation with an acute influenza infection (n = 16). Blood was obtained for PBMCs at baseline, day 8 to 14, and day 20 to 28 post exposure. Subjects were then followed for the remainder of the influenza season and were revaccinated with IIV the following fall, with blood again obtained at baseline, day 8 to 14, and day 20 to 28 post vaccination. Abbreviations: IIV, inactivated influenza vaccination; PBMC, peripheral blood mononuclear cell; V, visit.

Initially, influenza reactivity to the H3 and NP proteins was compared between infected and vaccinated subjects by quantifying production of IFN-γ by activated CD4 T cells (Figure 2). In this analysis, several interesting patterns emerged. While influenza-specific H3 reactivity was initially more robust at approximately day 10 post exposure in acutely infected compared to vaccinated subjects, this advantage did not persist to either later time points post infection or to visits following vaccination in study year 2. This contrasted with the pattern of NP-specific CD4 T-cell reactivity, as NP-specific CD4 T-cell responses were consistently more robust in infected as compared to IIV immunized subjects, including post-IIV administration the following fall. The full dataset, quantified as a percentage of activated (CD69+) cytokine positive cells after subtracting background, is presented in Supplementary Table 1, with additional data available upon request.

Figure 2.

Compared to the vaccination cohort, the percentages of IFN-γ-producing H3- and NP-specific CD4 T cells were greater by visit 2 post-H3N2 infection, with NP-specific cells in acute cohort subjects preferentially boosted upon vaccination the following fall. PBMCs from either infected or vaccinated subjects were stimulated with pools of overlapping peptides encompassing the coding regions of the H3 and NP proteins and were stained by intracellular cytokine staining. Results were gated on CD4⁺ T cells expressing CD69 and IFN-γ. A repeated ANCOVA was applied to examine differences in IFN-γ production between acutely infected and vaccinated subjects, controlling for age as a continuous variable. Filled circles represent CD4 T-cell responses of individual subjects following year 1 acute infection, while open squares represent CD4 T-cell responses post year 1 vaccination. Horizontal lines are means of the individual responses, while bars represent estimated percentages controlled for age as an overlaid graph. Abbreviations: ANCOVA, analysis of covariance; IFN-γ, interferon- γ; PBMC, peripheral blood mononuclear cell.

As children may have less T helper 1 (Th1)-biased CD4 T-cell reactivity in response to influenza [23], we next used combination gating to simultaneously quantify production of IFN-γ, IL-2, and TNF-α. These data are depicted as pie graphs in Supplementary Figure 2 and as a scatter plot of individual subject reactivity overlaid with bar graphs showing age-adjusted mean concentration estimates in Figure 3. Overall, these data revealed that at about day 24 following acute infection, a greater percentage of H3- and NP-specific CD4 T cells produced IFN-γ, with a larger fraction of cells producing the combination of IFN-γ and TNF-α following stimulation with peptide pools representing both proteins. However, only differences in NP-specific reactivity persisted following administration of IIV to the acute cohort in year 2, with a greater percentage of CD4 T cells in the acutely infected cohort producing IFN-γ, IL-2, and TNF-α in combination, IFN-γ and TNF-α together, and IFN-γ alone. These data support the importance of acute infection in driving the typical Th1-biased CD4 T-cell response to the NP protein.

Figure 3.

There are differences in CD4 T-cell functionality between subjects that were acutely infected with H3N2 influenza or vaccinated with IIV in study year 1. PBMCs were stimulated and stained as previously described with gating as shown in Supplementary Figure 1. Combination gates were created using FlowJo v10 software (TreeStar) to highlight single, double, and triple cytokine producing cells. A repeated ANCOVA analysis was applied to examine differences in polyfunctionality between acutely infected and vaccinated subjects, controlling for age as a continuous variable. Filled circles represent CD4 T-cell responses of individual subjects acutely infected in year 1, while open squares represent CD4 T-cell responses in subjects vaccinated in year 1. Horizontal lines are means of the individual responses, while bars represent estimated percentages controlled for age as an overlaid graph. Data obtained following stimulation of PBMCs with the H3 HA protein are shown in (A) and (B), and with NP-specific responses depicted in (C) and (D). Abbreviations: ANCOVA, analysis of covariance; IIV, inactivated influenza vaccination; PBMC, peripheral blood mononuclear cell.

To evaluate the effect of age on CD4 T cell reactivity, we next divided subjects into 2 discrete age groups and compared subjects < 3 years of age to subjects between 3 and 8 years of age who were likely to have had multiple previous influenza infections and vaccinations. When H3-specific CD4 T-cell reactivity was subgrouped by age, an age-dependent accrual of H3-specific CD4 T cells in the vaccinated cohort was seen at both visit 3 and visit 6 (Figure 4A). At visit 3, greater reactivity to the H3 HA protein in H3N2-infected compared to IIV immunized subjects < 3 years of age was also seen. This trend reversed in older children, possibly due to multiple previous immunizations in the 3 to 8-year-old vaccinated cohort (Figure 4A; left). When examining NP-specific responses, greater CD4 T-cell reactivity was demonstrated following acute infection in both age groups at visit 3 (Figure 4B; left), with this difference accentuated in the 3 to 8-year-old age group at visit 6 (Figure 4B; right). These data are summarized in stacked bar graphs demonstrating reactivity by age at visit 3 (Figure 4C) and visit 6 (Figure 4D) and highlight the age dependence of H3-specific CD4 T-cell responses, especially among the vaccinated cohort. In addition, the more robust NP-specific CD4 T-cell response in older, acutely infected children is clearly demonstrated, suggesting NP-specific CD4 T-cell mediated immunity may be dependent upon repeated viral exposures.

Figure 4.

The specificity of the influenza-specific CD4 T-cell response varies depending on subject age. Subjects were banded into 2 groups consisting of those < 3 years of age and those between 3 and 8 years of age, with H3- and NP-specific CD4 T-cell reactivity examined. A, H3-specific CD4 T-cell response at visit 3 (left) and visit 6 (right) and (B) CD4 T-cell reactivity to NP at the same 2 time points. Data are presented as scatter plots demonstrating individual reactivity, overlaid by bar graphs showing the age-adjusted mean concentration estimate. Statistical analysis was completed using a repeated ANOVA analysis, controlling for age as a group: <3 years vs 3–8 years. C and D, Stacked bar graphs highlighting differences in reactivity by age: (C) CD4 T-cell reactivity at visit 3 (day 20–28 post infection or vaccination in year 1) and (D) reactivity at visit 6 (day 20–28 post vaccination in study year 2). Abbreviations: ANOVA, analysis of variance; IFN-γ, interferon- γ; yo, years old.

We next examined variability in CD4 T-cell response amongst individual subjects ordered by age (Figure 5). To be included in this analysis, subjects had to have an adequate number of PBMCs to evaluate both H3- and NP-specific CD4 T-cell reactivity at either visit 3 (Figure 5A) or visit 6 (Figure 5B), with subjects excluded if there were insufficient cells to examine both conditions. These data demonstrate an increased fractional response to NP compared to H3 in acutely infected subjects, with this pattern persisting post vaccination 1 year later. A trend demonstrating increasing H3-specific CD4 T-cell reactivity with age in the vaccinated cohort is also visualized.

Figure 5.

Individual subjects acutely infected with influenza in study year 1 demonstrated an increased fractional response to NP compared to H3. Processed data from intracellular cytokine staining is represented as a stacked bar for each subject, with reactivity to H3 HA in black and NP in gray. Individual subject identities are indicated on the bottom of the graph, arranged by age group. Subjects shown had an adequate quantity of PBMCs to stimulate with both the H3 HA and NP proteins. A, Results from visit 3 (day 20–28 post initial exposure) and (B) visit 6 (day 20–28 following revaccination). Abbreviations: IFN-γ, interferon- γ; PBMC, peripheral blood mononuclear cell.

Given the ability of Ki67 to quantify recently proliferating CD4 T cells in circulation at early time points post infection or post vaccination, CD69⁺Ki67⁺ CD4 T cells were examined at visit 2 (about day 10 following initial exposure) and visit 5 (about day 10 post vaccination study year 2). Greater recruitment of CD4 T cells into the immune response was seen following acute infection compared to vaccination at visit 2, with this difference reaching statistical significance for H3 (Figure 6A) and trending towards significance for NP (Figure 6B). This increased CD4 T-cell proliferation was not seen post vaccination in study year 2. This suggests that there is greater recruitment of CD4 T cells into the influenza-specific response post infection compared to vaccination; however, this advantage is not sustained post vaccination the following year.

Figure 6.

Acutely infected subjects showed significantly increased proliferation following infection compared to noninfected vaccinated children. PBMCs were stimulated with complete peptide pools comprising the coding sequences of the H3 HA (A) or NP (B) proteins and then were stained using intracellular cytokine staining. An increase in antigen-specific Ki67⁺ CD4 T cells was present at approximately day 10 post infection (visit 2) when compared to the same time point post vaccination. This increase was not maintained following vaccination of both cohorts in study year 2. Statistical analysis was performed using a repeated ANCOVA analysis, controlling for age as a continuous variable. Filled circles represent Ki67⁺CD69⁺ CD4 T-cell responses of individual subjects following year 1 acute infection; open squares represent data following vaccination. Horizontal lines are means of the individual responses, while overlaid bars are the estimated percentages controlled for age. Abbreviations: ANCOVA, analysis of covariance; PBMC, peripheral blood mononuclear cell.

DISCUSSION

In this article, we detail differences in CD4 T-cell specificity and function between children initially infected with H3N2 influenza versus vaccinated with IIV, and then evaluate how this initial difference in exposure shapes postvaccination CD4 T cell-mediated immunity the following year. We found that there was a greater magnitude H3-specific CD4 T-cell response following acute infection; however, this advantage was not sustained post vaccination in study year 2. More robust NP-specific CD4 T-cell mediated immunity was also established by infection compared to vaccination and persisted following vaccination in the second year of the study. Important differences in CD4 T cell functionality were also identified, with generally a greater percentage of CD4 T cells primed by infection as opposed to vaccination producing IFN-γ, often in combination with TNF-α. In addition, age-based differences in reactivity supported an accrual of H3-specific CD4 T cells over time with repeated exposures via both vaccination and infection, as opposed to CD4 T cells directed against epitopes within NP that appeared to be preferentially primed following acute infection.

Consideration of how early childhood influenza exposures shape lifelong influenza-specific immune memory is of critical importance. There is a growing understanding that the influenza strains encountered in childhood imprint some degree of lasting protection against seasonal [4] and potentially pandemic influenza strains [5] and shape lifelong responses to influenza vaccination [6–9]. However, most of this research has focused on the B-cell and antibody responses [24–27]. This research demonstrates that the degree of back-boosting upon encounter with a drifted influenza virus is likely related to multiple factors, including the individual’s immune history, the antigenic relatedness between the viruses encountered, and the type of antigenic encounter (infection versus vaccination) [26, 28].

Despite the documented importance of early childhood immune history in the anti-influenza immune response, there is little yet known about how early childhood influenza encounters shape CD4 T-cell reactivity. Increases in influenza-specific CD4 T cells producing IFN-γ have been demonstrated post vaccination in both younger [29] and older children [9]; however, these studies did not examine either response specificity or how reactivity was established by different exposure routes. Our previous work has demonstrated an overall lower frequency of influenza-specific CD4 T cells in young children, most markedly to the internal virion protein NP, with a less Th1-biased CD4 T-cell response in children compared to the young adults [23]. However, these responses were only examined at steady state. Our current research adds to these findings, supporting the observation that acute infection is able to prime a greater magnitude NP-specific CD4 T-cell response. Interestingly, H3-specific CD4 T-cell reactivity appeared to accumulate over time in the vaccination cohort, while NP-reactive CD4 T cells were primed mainly by acute infection, with these cells boosted almost exclusively in older subjects who had a documented infection in study year 1. This suggests that, while NP is contained within IIV [30], it may not be present in large enough quantities to establish a primary CD4 T-cell response.

All subjects in this study were vaccinated with seasonal Quadrivalent Fluzone, a split virus vaccine known to contain greater amounts of contaminating internal virion proteins. Koroleva et al recently demonstrated that split vaccines such as seasonal Fluzone contained abundant NP compared to subunit vaccines [30], in agreement with studies from other groups [31, 32]. In the study by Koroleva et al, the 2014–2015 seasonal Fluzone vaccine was shown to contain approximately 215 ng NP per μg HA, with a similar amount of NP quantified in the 2015 and 2016 vaccine formulations by relative densitometry [30]. It would be expected that split vaccinations such as Fluzone would have the best opportunity to establish CD4 T-cell responses to internal virion proteins when compared to subunit vaccines such as Fluvirin, including in vaccinated pediatric subjects

In this study, we compared CD4 T-cell responses specific for the H3 HA protein to the NP protein, as both proteins were present in the infecting H3N2 influenza strain and within IIV, although in disparate amounts. Ideally, we would have also examined CD4 T-cell reactivity in H1-infected subjects to evaluate how early infection with different influenza strains differentially imprinted cellular immunity and affected the development of the influenza-specific memory response. Unfortunately, this was not possible as H3N2 was the predominant influenza strain circulating in both the years we were actively enrolling in this cohort. Although cell numbers were often limited, whenever possible responses to the H1 HA protein were simultaneously examined, enabling this comparison to be completed in future studies. In addition, studies of the B-cell, antibody, and CD8 T-cell response are currently ongoing.

One potential confounder in this study was the disparity in previous IIV immunizations between the acute and vaccinated cohorts, with 71% of the vaccine cohort having completed the primary vaccination series before study enrollment compared to only 31% of the acute cohort. This was in part driven by the natural history of pediatric influenza, as unvaccinated or undervaccinated children are more likely to be diagnosed with an acute influenza infection. This inequity in prior influenza vaccine exposures could have confounded some of our findings, including the relative lack of an H3-specific immune response in the older children acutely infected with H3N2 influenza, many of whom had not been previously vaccinated. Small sample size in the analysis segregated by age could also have biased these results.

Overall, our findings support the importance of early childhood acute influenza infections in determining the specificity and functionality of the anti-influenza CD4 T-cell response. It is interesting to consider these results in the context of both B-cell help and other CD4 T-cell effector functions. While our study time points were not ideally scheduled to examine the emergence of circulating T follicular helper (Tfh) cells, the ability of both vaccination and infection to establish an H3-specific CD4 T-cell response suggests that some CD4 T-cell help will likely be available to B cells specific for the H3 HA protein from a young age. This early availability of CD4 T-cell help for B cells is supported by previous research documenting effectiveness of a prime-boost IIV series in infants and young children [33], with generation of antibody secreting cells and protective antibody levels [34–36]. However, the relative lack of NP-specific CD4 T-cell responses and fewer CD4 T cells able to produce IFN-γ following early life vaccination could indicate that IIV in children is less able to establish a conserved CD4 T-cell response, with the response established possibly skewed away from traditional Th1 function. Further studies designed to meticulously evaluate diverse CD4 T-cell functions in the pediatric population, including focused studies on Tfh, will provide much needed data on how age affects the CD4 T-cell response and clarify the role of different effector functions in protective anti-influenza immunity in children.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgment. The authors thank Professor Andrea J. Sant for thoughtful discussions and editorial suggestions, the NYICE clinical core, and the study participants for their willingness to contribute to scientific research.

Financial support. This work was supported by the Doris Duke Charitable Foundation (grant number 2015098); and the National Institute of Allergy and Infectious Diseases, National Institutes of Health, Department of Health and Human Services (Centers of Excellence for Influenza Research and Surveillance grant number HHSN272201400005C).

Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Presented in part: Annual Centers of Excellence for Influenza Research and Surveillance (CEIRS) Network Meeting 22–25 July 2018, New York, NY; and Annual CEIRS Network Meeting 24–26 June 2019, Baltimore, MD.