Humoral Correlates of Protection Against Influenza A/H3N2 Virus Infection

Abstract

Background

Influenza virus remains a threat to human health, but gaps remain in our knowledge of the humoral correlates of protection against influenza virus A/H3N2, limiting our ability to generate effective, broadly protective vaccines. The role of antibodies against the hemagglutinin (HA) stalk, a highly conserved but immunologically subdominant region, has not been established for influenza virus A/H3N2.

Methods

Household transmission studies were conducted in Managua, Nicaragua, across 3 influenza seasons. Household contacts were tested for influenza virus infection using reverse-transcription polymerase chain reaction. We compared preexisting antibody levels against full-length HA, HA stalk, and neuraminidase (NA) measured by enzyme-linked immunosorbent assay, along with hemagglutination inhibition assay titers, between infected and uninfected participants.

Results

A total of 899 individuals participated in household activation, with 329 infections occurring. A 4-fold increase in initial HA stalk titers was independently associated with an 18% decrease in the risk of infection (adjusted odds ratio [aOR], 0.82 [95% confidence interval {CI}, .68–.98]; P = .04). In adults, anti-HA stalk antibodies were independently associated with protection (aOR, 0.72 [95% CI, .54–.95]; P = .02). However, in 0- to 14-year-olds, anti-NA antibodies (aOR, 0.67 [95% CI, .53–.85]; P < .01) were associated with protection against infection, but anti-HA stalk antibodies were not.

Conclusions

The HA stalk is an independent correlate of protection against A/H3N2 infection, though this association is age dependent. Our results support the continued exploration of the HA stalk as a target for broadly protective influenza vaccines but suggest that the relative benefits may depend on age and influenza virus exposure history.

The humoral correlates of protection against influenza A/H3N2 virus are poorly understood. We found that the hemagglutinin stalk is independently associated with protection against A/H3N2 virus infection, and that responses measured using hemagglutinin inhibition assays varied across season and age.

Influenza virus remains a persistent threat to human health, with up to 1 billion cases, 5 million severe cases, and 650 000 deaths occurring globally each year [1]. Influenza vaccination is one of the most important tools available to reduce the morbidity and mortality of influenza in the community, but the effectiveness of influenza vaccination remains suboptimal, with vaccine effectiveness estimates ranging from 10%–60% by season, primarily due to mismatch between predicted and circulating strains [2, 3]. The development of a more broadly protective or “universal” influenza vaccine has long been a goal of influenza researchers, but progress has proven to be difficult [4–6]. Specifically, our understanding of the correlates of protection against influenza virus infection, and how these correlates may change across the lifespan, is limited. Antibody responses against the hemagglutinin (HA) head, often measured by hemagglutination inhibition assay (HAI), have long been known to correlate with protection against influenza virus infection, as have responses against neuraminidase (NA) [7–10]. However, these targeted epitopes are susceptible to antigenic drift, meaning that repeatedly targeting responses against these epitopes is of limited utility across seasons and influenza virus strain [5]. The HA stalk, a highly conserved domain of influenza virus HA, is of great interest for the development of a universally protective influenza virus vaccine because antibody responses against the stalk domain neutralize influenza virus in vitro and to protect broadly across influenza A virus subtypes and influenza B virus strains in animal models [11–13]. Antibody responses against the HA stalk independently correlate with protection against A/H1N1pdm virus infection in a household transmission study of influenza; however, the role of the HA stalk as a correlate of protection against influenza A/H3N2 virus infection, and whether the effect of anti-HA stalk varies with age or by influenza virus exposure history, is currently unknown [14]. Most studies have focused on the group 1 stalk, potentially because group 1 stalk antibodies were discovered first and because a group 1 subtype caused a pandemic more recently in 2009. There are differences between group 1 and group 2 stalks, including monoclonal antibody (mAb) epitopes identified so far. The presence of a glycan at position 38 of HA1 in most group 2 subtypes also blocks binding of “typical” VH1-69 germline central stalk mAbs that target group 1 stalks [15]. It has also been harder to generate reagents to measure the group 2 stalk response like recombinant chimeric HAs (cHAs) or headless HA constructs due to issues with their expression levels. For these reasons, it is important to study group 2 stalk antibodies and their role in protection. This study aims to explore the humoral correlates of protection against influenza A/H3N2 virus infection with a particular emphasis on the role of the HA stalk as an independent correlate of protection against A/H3N2 virus infection.

METHODS

Study Population and Design

This study analyzes data from 2 household observational studies based at the Health Center Sócrates Flores Vivas in Managua, Nicaragua: the Household Influenza Transmission Study (HITS) and the Household Influenza Cohort Study (HICS). HITS is a case-ascertained study that began in 2012 and concluded in 2017, and HICS is a prospective household-based cohort study that began in 2017 and is currently ongoing. In both studies, A/H3N2 virus–positive individuals are initially detected at the health center, where household members are enrolled (HITS) or activated (HICS) into intensive monitoring for a period of approximately 2 weeks. During this period, household members are tested repeatedly for influenza virus regardless of symptoms, and daily symptom occurrence is documented. Blood samples are collected both at the beginning of the monitoring period and 30–45 days after [16]. These studies were approved by the institutional review boards at the Nicaraguan Ministry of Health and the University of Michigan and are in accordance with the Helsinki Declaration of the World Medical Association. Written consent to participate or parental permission was obtained for all participants; in children aged >6 years, verbal assent was obtained where possible.

Laboratory Methods

Influenza swabs collected from household members were tested for influenza virus with real-time reverse-transcription polymerase chain reaction (RT-PCR) using validated Centers for Disease Control and Prevention protocols. If positive for influenza virus, subtype or lineage determination was performed using additional RT-PCR assays [17–19]. Preexposure antibody levels were assessed by testing each sample using (1) HAI along with enzyme-linked immunosorbent assays (ELISAs) directed at antibodies against (2) full-length HA, the (3) HA stalk, and (4) NA. HAIs were performed to test the ability of participant antibodies to neutralize influenza virus HA's agglutination of turkey red blood cells, a proxy for antibody responses against the HA head. ELISAs were performed to test participant antibody responses against full-length HA, the HA stalk, and NA. The HAI of samples against A/Hong Kong/4801/2014 and A/Singapore/INFIMH-16-0019/2016 was tested. Immunoglobulin G (IgG) antibodies against trimeric H3 A/Hong Kong/4801/2014 and tetrameric N2 A/Hong Kong/4801/2014 were measured by ELISA using an anti-human IgG (Fab specific) horseradish peroxidase detector [20]. A 4-fold rise in HAI titer across the monitoring period demonstrated reasonable agreement with PCR-confirmed A/H3N2 virus infection (Supplementary Figure 1). Antibody responses against the HA stalk were measured using a cHA (cH7/3) protein that expresses a head domain to which participants should be naive (A/Anhui/1/2013) along with the stalk of H3 (A/Hong Kong/4801/2014), ensuring that detected immune responses against the cHA are directed at only the HA stalk. The primary correlates of protection analyses utilized assay data from initial, baseline samples, which were collected either before or shortly after household activation, when anti-influenza IgG levels should approximate preexposure levels. In total, 95.8% of initial samples were collected no later than 3 days after symptom onset (symptomatic participants) or household activation (asymptomatic or negative participants), and 99.3% of initial samples were collected no later than 7 days after symptom onset or household activation.

Statistical Analysis

Antibody titer levels and area under the curve measurements (AUCs) were log-transformed for all analyses. Differences in the distribution of initial HAI titers and ELISA antibody levels between infected and uninfected participants were tested using Wilcoxon rank-sum tests. The antibody level necessary to achieve 50% and 80% protection against infection in the population was estimated using 3-parameter logistic regression using the R package nlpr version 0.1–7. This method improves upon conventional logistic approaches by allowing for incomplete protection at high titer levels and some degree of protection at low titer levels, which is more biologically plausible than a model with strict asymptotes at 0 and 1, and has been used successfully for similar analyses in the past [14]. Finally, the independence of the association between each antibody measure and protection against infection was evaluated using multi-assay conventional logistic regression models, which adjust for age and the antibody levels of other assays to account for correlation between antibody responses.

RESULTS

A total of 899 individuals across 169 influenza A/H3N2 virus–activated households from the 2014, 2016, and 2017 influenza seasons were included, with 329 individuals testing positive for A/H3N2 and 570 individuals testing negative for A/H3N2 by RT-PCR. The age of the study population ranged from <6 months to 89 years, with a median age of 14.7 years for the A/H3N2-positive individuals and 26.7 years for the A/H3N2-negative individuals. This study population is largely unvaccinated for influenza; <0.5% of the population had been vaccinated for influenza in the last 6 months, and 8.7% had ever been vaccinated according to self-report and vaccine records (Table 1).

Table 1.

Description of Study Population

Characteristic	A/H3N2 Positive (n = 329)	A/H3N2 Negative (n = 570)	Total (n = 899)
Sex
Female	183 (55.6)	376 (66.0)	559 (62.2)
Male	146 (44.4)	194 (34.0)	340 (37.8)
Age, y, mean (SD)	14.7 (14.9)	26.7 (18.9)	22.3 (18.5)
0–1	30 (9.1)	10 (1.8)	40 (4.5)
2–4	68 (20.7)	34 (6.0)	102 (11.4)
5–14	121 (36.8)	166 (29.1)	287 (31.9)
≥15	110 (33.4)	360 (63.2)	470 (52.3)
Vaccination
Ever vaccinated	40 (12.2)	38 (6.7)	78 (8.7)
Recently vaccinateda	2 (0.6)	0 (0.0)	2 (0.2)
Season
2014–2015	50 (15.2)	133 (23.3)	183 (20.4)
2016–2017	123 (37.4)	168 (29.5)	291 (32.4)
2017–2018	156 (47.4)	269 (47.2)	425 (47.3)
Symptomatic
Yes	279 (84.8)	NA	NA
No	50 (15.2)	NA	NA
Influenza-like illnessb
Yes	143 (43.5)	NA	NA
No	186 (46.5)	NA	NA

Abbreviations: NA, not applicable; SD, standard deviation.

^aVaccinated within 6 months of the start of the intensive monitoring period.

^bObjective fever ≥38°C with cough and symptom onset in the last 10 days.

As expected, the distribution of initial antibody responses was lower among individuals who became infected with influenza A/H3N2 virus than individuals who remained negative throughout the monitoring period for responses directed at the HA head, full-length HA, the HA stalk, and NA (Figure 1). The median HAI titer among those who became infected was 1:40, compared to a median titer of 1:80 in the uninfected (P < .0001). The median anti-full-length HA antibody levels among those who became infected was 164, compared to 363 among the uninfected (P < .0001). The median anti-HA stalk antibody levels among those who became infected was 25, compared to 54 among the uninfected (P < .0001). Finally, the median anti-NA antibody levels among those who became infected was 33, compared to 97 among the uninfected (P < .0001).

Figure 1.

Distribution of initial antibody levels by influenza A/H3N2 infection status. The distribution of initial antibody levels for hemagglutination inhibition assay and enzyme-linked immunosorbent assay against full-length hemagglutinin (HA), HA stalk, and neuraminidase between uninfected (left) and infected (right) participants. The differences between distributions were tested using Wilcoxon rank-sum tests. Abbreviations: AUC, area under the curve; HA, hemagglutinin; HAI, hemagglutination inhibition assay; NA, neuraminidase; PCR, polymerase chain reaction.

Estimated 50% protective levels, or the antibody level necessary to reduce the risk of infection by 50% relative to participants with no detectable antibodies, were calculated for all 4 assays. The 50% protective HAI titer across all participants was found to be 1:612, indicating that an HAI titer of at least 1:612 is necessary to reduce the risk of infection by 50% in this population. Due to this higher than expected 50% protective titer for HAI and the potential impact of antigenic mismatch in the HAI viral strain, the HAIs were repeated using other antigens for the 2016 and 2017 seasons. HAIs were repeated using the A/Singapore/INFIMH-16-0019/2016, which was not a better match. When SI16 was used, increasing titers necessary for protection were still observed in 2016 and 2017, though the overall independent association between anti-HA stalk antibodies and protection against infection still held (Supplementary Tables 1 and 2, Supplementary Figure 2). The 50% protective AUC for the ELISAs for full-length HA, HA stalk, and NA was estimated to be 106, 61, and 67, respectively (Figure 2). Compared to adults, children demonstrated lower estimated 50% protective levels for antibodies against the HA head (HAI), full-length HA, and NA, and higher 50% protective levels for antibodies against the HA stalk. The 50% protective HAI titer was 1:142 for children aged 0–14 years and 1:355 for adults aged ≥15 years, and the 50% protective antibody level for the ELISAs for full-length HA, HA stalk, and NA was estimated to be 276, 172, and 123, respectively, for children and 1600, 134, and 786, respectively, for adults (Figure 2). Generally, lower antibody levels were necessary to protect against RT-PCR–positive influenza-like-illness (ILI) than against PCR positivity alone (Supplementary Figure 3).

Figure 2.

Single antibody models to estimate 50% and 80% protective levels. Protection curves estimated using 3-parameter logistic regression for all ages (top row), children aged 0–14 y (middle row), and adults aged ≥15 y (bottom row). The y-axes represent the proportion of individuals that fall within a titer grouping (vertical bars) and the estimated infection risk within each titer grouping (fitted model). Abbreviations: AUC, area under the curve; ELISA, enzyme-linked immunosorbent assay; HA, hemagglutinin; HAI, hemagglutination inhibition assay; NA, neuraminidase.

The association between HAI titer and 50% protective level varied greatly by season. In 2014, the estimated 50% protective level was 1:69, which is reasonably in line with prior literature [7, 8, 21, 22]. The 50% protective level is higher in subsequent seasons, falling to 1:309 in 2016 and to 1:1316 in 2017. In contrast, the estimates for the other assays remained relatively stable by season (Figure 3). This pattern was also observed when using Singapore 16 for the HAIs (Supplementary Table 1, Supplementary Figure 2)

Figure 3.

Single antibody models to estimate protective levels, by season. Protection curves estimated using 3-parameter logistic regression for samples from the 2014 season (top row), 2016 season (middle row), and 2017 season (bottom row). The y-axes represent the proportion of individuals that fall within a titer grouping (vertical bars) and the estimated infection risk within each titer grouping (fitted model). Abbreviations: AUC, area under the curve; ELISA, enzyme-linked immunosorbent assay; HA, hemagglutinin; HAI, hemagglutination inhibition assay; NA, neuraminidase.

Single-assay models, though informative for understanding general associations between each assay and protection against infection, do not account for the possibility of non-independent correlations between each antibody level and protection against A/H3N2 virus infection due to correlated immune responses. Therefore, we constructed several multi-assay models, each with a different combination of antibody measures, to investigate which, if any, measures are correlated with protection against infection independent of the association of the other antibody measures. Assay set 1 includes NA, HA stalk, and HAI to capture antibodies against the HA head, stalk, and NA. Assay set 2 includes NA and full-length HA to capture non-domain-specific antibodies against HA along with NA. Assay set 3 includes HAI and NA to capture antibodies against the HA head and NA, without accounting for antibodies against the stalk. All models are age-adjusted.

As expected, protective associations measured from multi-assay models were less strongly protective than those measured from single-assay models, as the multi-assay models account for the correlation between the antibody measures (Figure 4A). However, all 4 antibody measures still demonstrated independent correlation with protection against A/H3N2 virus infection. Specifically, a 4-fold increase in anti-HA stalk antibodies was associated with protection against infection in both the single-assay model (adjusted odds ratio [aOR], 0.64 [95% confidence interval {CI}, .54–.76]; P < .001) and the multi-assay model (aOR, 0.82 [.68–.99]; P = .04) across all participants, indicating that antibody responses against the HA stalk are an independent correlate of protection against influenza A/H3N2 virus infection even when accounting for responses against the HA head and NA.

Figure 4.

Results of age-adjusted logistic models estimating the odds of A/H3N2 virus infection with a 4-fold increase in initial antibody level across the population. A, Single-assay models (left) and multi-assay models (right) for each of the 3 assay sets. B, Multi-assay models, by age, demonstrating the different associations between each antibody level and protection against A/H3N2 virus infection by age group. Abbreviations: ELISA, enzyme-linked immunosorbent assay; HA, hemagglutinin; HAI, hemagglutination inhibition assay; NA, neuraminidase.

The association between each antibody measure and protection against A/H3N2 virus infection was not consistent across age (Figure 4B). Among 0- to 14-year-olds, responses against NA and the HA head (HAI) were consistently associated with protection against infection across assay sets; however, a 4-fold increase in anti-HA stalk antibodies was not associated with protection against A/H3N2 virus infection in this age group (aOR, 0.95 [95% CI, .73–1.24]; P = .71). In individuals aged ≥15 years, anti-NA and anti-HA head responses were not consistently associated with protection against A/H3N2 virus infection, but responses against the HA stalk were associated with protection (aOR, 0.72 [95% CI, .54–.95]; P = .02).

DISCUSSION

By utilizing intensive monitoring and testing strategies in households with known influenza A/H3N2 virus infection, this study was able to compare the impact of preexisting antibody responses to the HA head, full-length HA, the HA stalk, and NA between infected and uninfected household members to examine the effect of these responses on protection against infection. Unsurprisingly, individuals who remained uninfected had, on average, higher initial antibody levels to the HA head, full-length HA, the HA stalk, and NA when compared to individuals who became infected with influenza A/H3N2. This is consistent with the well-established role of general anti-HA and anti-NA responses on protection against influenza virus infection in humans, and additionally suggests that antibody responses against the HA stalk are important for protection against A/H3N2 virus infection [9, 10, 23, 24]. Importantly, though there is significant separation in the distribution of responses between infected and uninfected individuals, a large amount of overlap remains for each antibody type, a reminder that other immunologic factors such as innate immune responses and T-cell responses are important for protection against infection and that no single assay, even HAI, can cleanly separate those who are susceptible from those who are not.

We estimated the association between a 4-fold increase in initial antibody levels across the population and protection against infection in multi-assay models that accounted for the known correlation between assays. We found that across the population, all measures were independently correlated with protection even when accounting for correlations between measures. Specifically, we observed that a 4-fold increase in initial antibody levels against the HA stalk is associated with an 18% reduction in the odds of A/H3N2 virus infection, controlling for age and the value of the other assays. This observation was not consistent across age, however, with children aged <14 years demonstrating no protection offered from anti-HA stalk antibody levels, and adults aged ≥15 years demonstrating stronger protection from anti-HA stalk responses. Previous work has demonstrated that children and adults require different titer levels to achieve protection against influenza infection; this work expands that by suggesting that different epitope targets, in addition to different levels of antibodies against the same target, may be important for different age groups [25]. Young children, whose anti-HA head responses are closer in antigenic proximity to the circulating strain, may experience more benefit from the direct neutralization potential of anti-HA head responses as opposed to the breadth of protection offered by anti-HA stalk responses, whereas in adults, who are more likely to have more cross-reactive anti-HA head antibodies, the broad protection of anti-HA stalk responses is more beneficial [11]. Though speculative, this would suggest that anti-HA stalk responses are able to effectively protect against A/H3N2 infection if strong, on-target anti-HA head responses are ineffective, and supports the continued efforts to utilize strong humoral responses against the HA stalk for the development of a more broadly protective influenza vaccine. Notably, these analyses are of individuals who are naturally infected with influenza virus and have generated a corresponding immune response, and thus the results are dependent on the distribution of these prior responses. In this study, children have lower overall levels of anti-HA stalk antibodies than adults, and this distributional difference in anti-HA stalk antibody levels could partially explain the lack of correlation between anti-HA stalk antibody levels and protection against infection in children. This is in line with prior work, which showed that children generate much narrower immune responses to influenza A/H1N1pdm when compared to adults, and children tend to generate antibodies against the HA head, compared to more broad epitopes in adults [26, 27]. Children may still benefit from anti-HA stalk antibody boosting from vaccination even though we do not observe an association between anti-HA stalk antibody levels and protection against infection among children in this study. Children and adults also differ with respect to antibody subclass distribution and cell-mediated immune responses, which could also affect the association between anti-HA stalk antibodies and protection against A/H3N2 virus infection.

Additionally, we calculated the 50% protective level for each assay, and found that, though all antibody measures were associated with protection against infection, the 50% protective level varied widely by age group and season. An HAI titer of 1:40 is commonly presumed to correlate with a 50% reduction in influenza virus infection risk, and we found that, though a similar 50% protective level was observed in some age groups and seasons, this was not universal [8, 28]. Though the estimated protective level from HAI in 2014 was close to 1:40, later seasons demonstrated far higher HAI titers necessary to achieve a 50% reduction in influenza infection risk. Though this may be partially explained by antigenic drift creating a mismatch between the HAI antigen and circulating strain in more recent seasons, we observed similar results when HAIs were conducted using A/Singapore/INFIMH-16-0019/2016, a strain that should be a reasonable match to A/H3N2 virus circulation in 2016 and 2017, and Hong Kong 14 HAIs showed reasonable agreement with PCR for detecting infection. Additionally, children and adults had different associations between HAI and protection against infection, while antigenic mismatch would be expected to be consistent across age groups. Together, these results suggest that antigenic mismatch is not fully to blame for the low correlation between HAI titers and protection against A/H3N2 virus infection, and that this low correlation may be due to technical issues with HAI as an assay and its validity as a construct for anti-HA head neutralizing antibodies with recent H3N2 viruses. Though our primary outcome was RT-PCR–confirmed infection, protection against RT-PCR–confirmed ILI was achieved at lower titers, which is consistent with other studies [7, 8, 29–31]. Though HAI titers are often viewed as the consensus correlate of protection against influenza virus for the development of vaccines, the variance in protective HAI titer by season, age, and influenza exposure history suggests a limitation in the utility of a simple 1:40 HAI titer cutoff for assessing protection against influenza A/H3N2 virus [8, 24]. Previously using the same study design and methods, we found that a 1:40 HAI titer corresponded with a 50% reduction in the probability of A/H1N1pdm virus infection, so this limitation may be relatively specific to influenza virus A/H3N2 [14]. Estimates of protection from the ELISAs were more stable across season and age group compared to HAI, suggesting a potential role for these assays in the assessment of antibody-induced protection against A/H3N2 virus. The deterioration in protection offered by anti-NA responses in 2016–2017 compared to 2014 is consistent with the circulation of an A/H3N2 strain with an N-linked glycosylation site at residue 245, which has been shown to restrict the binding of some monoclonal anti-NA antibodies [32, 33]. Unlike HAI, ELISAs are not a functional assay, and thus the antibody binding that is detected by ELISA may overestimate the level of antibodies that are truly protective against infection. Therefore, this study demonstrates the importance of utilizing robust assessment of influenza virus antibody responses rather than relying on one assay as a complete proxy for protection against infection, especially infection caused by influenza A/H3N2.

Overall, this study is strengthened by the large sample size, the utilization of data from longitudinal studies that allows for the comparison of A/H3N2 virus correlates of protection across several seasons, and the ability to compare different assays in the same study population. This study is limited by the lack of vaccinated individuals enrolled, potentially limiting the generalizability of these results to vaccine-induced protection against infection. Additionally, the level of exposure to influenza may not have been consistent across participants in ways that were unmeasured and thus cannot be adjusted for. Finally, this work utilizes serum antibody measurements, primarily of IgG. The epitope targets of antibodies measured in the serum mimic the targets of mucosal antibodies, though the relative antibody levels, and the antibody class distribution, likely differ [10]. Additional work that characterizes the correlates of protection of mucosal antibodies, especially immunoglobulin A, is needed.

CONCLUSIONS

Utilizing data from 2 household transmission studies, we found that the HA stalk is an independent correlate of protection against A/H3N2 virus infection, though this protection is age dependent; adults with higher preinfection anti-HA stalk antibody levels are protected against infection, but children are not, indicating that the role of anti-HA stalk responses in natural infection is dependent on age and influenza exposure history. These findings support the continued exploration of the HA stalk as a target for broadly protective influenza vaccines but suggest that the importance of anti-HA stalk responses depends on age and individual influenza virus exposure history. Additionally, we found that the correlation between HAI titer and protection was differential by age and season, which emphasizes the potential importance of other assays such as ELISA in the assessment of protection against A/H3N2 virus infection.

Supplementary Data

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

Notes

Acknowledgments. We thank the study participants and the many dedicated study personnel in Nicaragua at the Centro Nacional de Diagnóstico y Referencia and the Sócrates Flores Vivas Health Center.

Author contributions. G. H.: data curation, formal analysis, methodology, software, validation, visualization, writing—original draft, writing—review and editing. D. S., J. M. C. Q., T. Y., and A. F.: data curation, writing—review and editing. A. B.: investigation, methodology, supervision, project administration, writing—review and editing. G. K., S. O., and M. P.: investigation, project administration, writing—review and editing. R. L. and N. S.: investigation, writing—review and editing. F. K.: conceptualization, funding acquisition, investigation, project administration, supervision, writing—review and editing. A. G.: conceptualization, funding acquisition, investigation, methodology, project administration, resources, supervision, writing—review and editing.

Data availability. The data that support the findings of this study are available from the corresponding author upon reasonable request and following institutional review board approval.

Disclaimer. The funding agencies had no role in the design and conduct of the study; collection, management, analysis, or interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.

Financial support. This work was supported by the National Institute of Allergy and Infectious Diseases through the Collaborative Influenza Vaccine Innovation Centers (grant number 75N93019C00051 to F. K. and A. G.); the National Institute of Allergy and Infectious Diseases through the St Jude Center of Influenza Research and Surveillance (grant number HHSN272201400006C to A. G.); National Institute of Allergy and Infectious Diseases through the St Jude Center of Excellence for Influenza Research and Response (grant number 75N93021C00016 to A. G.); and National Institute of Allergy and Infectious Diseases (grant number R01 AI120997 to A. G.). A. G. is supported by the Biosciences Initiative at the University of Michigan through a Mid-career Biosciences Faculty Achievement Award.