Immune cellular networks underlying recovery from influenza virus infection in acute hospitalized patients

Seasonal influenza causes substantial global hospitalization and mortality. Determinants of disease severity remain incompletely defined, spanning viral, host genetic, and immune factors. Prior studies have implicated overactivation of innate immunity (hypercytokinemia) and/or impaired humoral and cellular immunity in severe disease. In the absence of neutralizing antibodies, pre-existing influenza-specific CD8+ and CD4+ T cells can reduce disease severity. Host genetic factors (e.g., HLA type and IFITM3 polymorphisms) have been associated with outcomes, with rs12252-C/C IFITM3 and early elevations of IL-6, IL-8, IL-10, and MIP-1β linked to fatal H7N9 disease. Elevated cytokines (IL-8, MCP-1, IP-10) have correlated with fatal H5N1, and IL-6/IL-8 have been linked to severe seasonal/pandemic influenza in adults and children. However, most patient studies assessed limited immunological parameters and rarely integrated innate and adaptive cellular responses with humoral immunity over time. There is a need to delineate how innate and adaptive responses collectively drive recovery from acute influenza pneumonia. This study leverages longitudinal sampling in hospitalized adults to map a comprehensive panel of immune parameters, integrating clinical and genetic factors, to define networks underlying recovery and severity.

The authors synthesize evidence that memory T cells contribute to reduced influenza severity when neutralizing antibodies are absent. Prior associations include IFITM3 variants (rs12252, rs34481144) and specific HLA alleles with severe outcomes. Early hypercytokinemia (IL-6, IL-8, IL-10, MCP-1, MIP-1α/β, IP-10) has been associated with fatal outcomes in H5N1, H7N9, and severe seasonal/pandemic influenza. Few studies have concurrently measured innate (cytokines/monocytes/NK), adaptive T cell, and antibody responses; SHIVERS is one notable study with acute and 2-week convalescent sampling, indicating prolonged activation in hospitalized patients. The literature points to gaps in understanding the interplay of cytokine milieu, B cell help (Tfh), antibody breadth, and antigen-specific T cells driving recovery, which this study addresses by integrated longitudinal profiling.

Study design and participants: The DISI cohort enrolled consenting adults with influenza-like illness admitted to The Alfred Hospital (Melbourne, Australia) during 2014–2017. A total of 44 influenza PCR-positive (influenza+) and 20 influenza-negative (influenza−) patients were sampled longitudinally. Blood was collected within 24–72 h of admission (Visit 1, V1), then every 2–5 days until discharge (V2, V3, etc.), and follow-up (Fup) blood ~30–41 days after disease onset/discharge. Nasal swabs were collected at V1 and V2. Clinical data included vaccination status, SOFA scores (disease severity), comorbidities, days in hospital. One influenza+ and one influenza− patient died; one influenza+ patient required ICU. Comparators: Data from a previously described H7N9 cohort (n=18; 12 recovered, 6 died) and healthy vaccinated adults (2015 TIV n=16, 2016 QIV n=26; samples at day 0, 7, 14, 28) were referenced for context. Healthy blood donors were also used for cellular baseline comparisons. Genetics: HLA class I and II molecular typing (Luminex, LABType SSO). IFITM3 SNPs rs12252 (exon 1) and rs34481144 (promoter) were genotyped via PCR and sequencing using specified primers. Analyses assessed distributions of risk vs non-risk alleles. Virology: Viral HA sequencing from nasal swabs by multiplex RT-PCR and NGS (Ion Torrent PGM); phylogenetics with Geneious. Clinical viruses used in assays included A/H1N1/California/7/2009, A/H3N2/Switzerland/9715293/2013 or Hong Kong/4801/2014, B/Phuket/3073/2013 (YAM) and B/Brisbane/60/2008 (VIC), grown in eggs and titrated by plaque assay. Cytokines/chemokines: Serum/plasma assayed by BD Cytometric Bead Array for IL-2, IL-4, IL-6, IL-8, IL-10, IL-12p70, IL-17A, IL-1β, IFN-α, IFN-γ, TNF, MCP-1, MIP-1α, MIP-1β, RANTES, CD178/FasL, and granzyme B. Partial correlations adjusted for age and days since onset. Antibody responses: Hemagglutination inhibition (HAI) assays (WHO protocols) measured titers against contemporaneous and historical strains. Titers were log-transformed as log2(HAI/10). Antibody landscapes constructed across birth-year exposures. Geometric mean titers (GMTs) compared across timepoints and cohorts. B cell and Tfh assays: Absolute lymphocyte and subset counts via BD TruCount. Circulating T follicular helper (cTfh) cells phenotyped as CD4+CXCR5+ with subsets cTfh1 (CXCR3+CCR6−), cTfh2 (CXCR3−CCR6−), cTfh17 (CXCR3−CCR6+); activation markers PD-1 and ICOS assessed. Antibody-secreting cells (ASCs) identified as CD19+CD20−/lowCD27++CD38++. Recombinant HA probes (H1, H3, B) detected class-switched HA-specific IgD− B cells; phenotype (CD21/CD27) and isotype (IgA, IgG, IgM) distributions assessed. Influenza-specific functional responses: PBMCs cultured with cognate live influenza virus (MOI 4) for 22 h; intracellular NP staining confirmed infection; IFN-γ production measured in NK cells, γδ T cells, MAIT (CD161+TRAV1-2+ or MR1-5-OP-RU tetramer+), CD4+ and CD8+ T cells. Cytolytic molecules: In flow-through fractions after tetramer enrichment or PBMCs from healthy donors, expression of granzymes A, B, K, M and perforin quantified in MAIT, NK, CD8+, and CD4+ T cells; polyfunctionality analyzed (Pestle/Spice, permutation tests). Tetramer-associated magnetic enrichment (TAME): Extensive panel of peptide-MHC class I and II tetramers (n=21 specificities) spanning common HLAs (A*01:01, A*02:01, A*24:02, A*68:01, B*07:02, B*08:01, B*18:01, B*35:01, B*57:01; DRB1*01:01, *04:01, *11:01) used to detect ex vivo influenza-specific CD8+ and CD4+ T cells. Activation markers (PD-1, CD38, HLA-DR, CD71/Ki-67) and differentiation (CD27, CD45RA, CCR7, CD95) profiled in tetramer+ vs parent populations. Statistics: Nonparametric tests (Mann–Whitney, Wilcoxon), Kruskal–Wallis, Friedman, Tukey’s multiple comparisons, LOESS for kinetics, OLS for trends in IFN-γ+ frequencies, Spearman correlations, partial correlations with FDR adjustment for cytokine networks, permutation tests for polyfunctionality. Heatmaps (empirical percentile transformation) integrated immune, clinical, and genetic parameters. Significance threshold p<0.05.

- Cohort and severity: 44 influenza+ and 20 influenza− hospitalized adults sampled longitudinally; median hospital stay 4 days; one influenza+ ICU case and one death in each group. Influenza A patients were older (median 58 years) than B or influenza− patients. Compared to H7N9 cohorts, seasonal influenza patients had significantly shorter hospital stays. - Cytokine networks: Early (V1) inflammatory cytokines IL-6, IL-8, MIP-1α, MIP-1β, and IFN-γ were significantly, strongly positively inter-correlated after adjusting for age and days since onset; the IL-6/IL-8/MIP-1α/MIP-1β cluster persisted at convalescence with an additional IFN-α/IL-17/IL-10 cluster emerging. - Antibodies: H3N2-infected patients showed significant increases in HAI titers at follow-up despite antigenic drift/mismatch, with many achieving ≥40 (protective) titers. GMT acute→follow-up ratios in influenza+ were approximately: H1N1 8→122 (6.9-fold), H3N2 18→552 (9.2-fold), B/YAM 41→60 (1.6-fold), B/VIC 26→38 (1.1-fold). Antibody landscapes showed back-boosting to older strains of the infecting subtype at follow-up; influenza− patients showed low, unchanged titers. - cTfh and ASC dynamics: Activated PD-1+ICOS+ cTfh1 cells and ASCs peaked around days 7–10 after onset, preceding recovery. ASCs were significantly higher during acute illness than follow-up (~2–8-fold), and higher than vaccine-induced peaks in healthy controls. Activated cTfh1 were higher in influenza+ than influenza− patients and higher than cTfh2/17 subsets. Acute correlations in influenza+: cTfh1 with ASCs (r=0.7060, p<0.0001), total cTfh with ASCs (r=0.5397, p=0.0003); ASCs associated with higher acute HAI titers (≥40). cTfh did not correlate with antibody titers during acute infection. - HA-specific memory B cells: Numbers of rHA+ IgD− B cells were similar between acute and follow-up, but phenotype and isotype shifted: at acute, more activated memory (CD21loCD27hi) with IgA and IgG (H3/B probes), transitioning to resting memory (CD21hiCD27hi) and predominantly IgG+ at follow-up. Vaccination induced activated memory without major isotype shifts (predominantly IgG+), highlighting infection-vaccination differences. - Functional cellular responses: IFN-γ-producing CD8+ and CD4+ T cell responses increased during the first ~15 days, stabilizing at convalescence, whereas innate cell IFN-γ responses (NK, MAIT, γδ) did not increase over time. Early decreases in IFN-γ+ γδ T and MAIT cells at V1 were associated with higher SOFA (2–6) severity (significant p-values reported), while at convalescence, lower IFN-γ+ CD4+ and CD8+ T cell responses were associated with higher severity. - Cytolytic molecules: MAIT cells showed highest total cytotoxic molecule expression, followed by NK and CD8+ T cells; CD4+ T cells were minimal. Compared to healthy donors, influenza+ patients had higher NK granzyme A/B/M and perforin but lower granzyme K; MAIT and CD8+ T cells had lower granzyme K and perforin. Co-expression of ≥4 cytolytic molecules was significantly reduced in patients (acute and follow-up) across subsets (e.g., NK p=0.0002 acute vs healthy), consistent with degranulation during infection. - Tetramer-defined T cells: Broad ex vivo detection of influenza-specific CD8+ and CD4+ T cells across 21 tetramers covering common HLAs. Tetramer+ CD8+ T cell precursor frequencies were substantially higher than CD4+ but similar between acute and follow-up; frequencies rarely exceeded 1% of CD8+ in blood. Activation markers (PD-1, CD38, CD71, HLA-DR) were elevated on tetramer+ cells during acute phase (especially days 6–10), declining by follow-up; parent populations were less activated. Phenotypes were enriched for central memory (CD27+CD45RA−) in both tetramer+ CD8+ and CD4+ T cells. - Integrated heatmaps: Acute samples showed regions where lower cytokines (IL-6, IL-8, MIP-1α/β, MCP-1) coincided with higher IFN-γ-producing cells, ASCs, and activated cTfh1; reciprocal regions showed higher cytokines with lower functional responses. These patterns were not strictly defined by SOFA or genetic parameters. - Genetics: No significant associations of disease severity with HLA categories or IFITM3 SNPs. rs12252 C/C was absent; T/T predominated. rs34481144 A/G was most common; severity did not differ by genotype.

This study integrates innate and adaptive immune metrics with clinical and genetic data to delineate immune networks associated with recovery from acute influenza in hospitalized adults. The findings indicate that robust, coordinated adaptive responses—particularly activation of cTfh1 cells with concurrent ASC expansion and HA-specific memory B cell maturation, alongside increasing influenza-specific IFN-γ-producing CD4+ and CD8+ T cells—emerge within the first 1–2 weeks and precede clinical recovery. Conversely, elevated inflammatory cytokine clusters (IL-6, IL-8, MIP-1α/β) are associated with diminished functional cellular responses. Early deficits in innate-like γδ T and MAIT IFN-γ responses relate to higher clinical severity, while stronger adaptive T cell function at convalescence associates with lower severity, underscoring a temporal hand-off from innate to adaptive immunity in effective disease resolution. Tetramer analyses confirm that influenza-specific memory T cells are present and highly activated during acute illness, transitioning to central memory phenotypes post-recovery, with most effector activity likely concentrated in respiratory tissues. The distinct phenotypic/isotypic profiles of HA-specific B cells after natural infection versus vaccination have implications for vaccine strategies aiming to induce mucosal IgA and systemic IgG memory. Overall, the data support that coordinated humoral help (cTfh→B cell axis) and functional T cell responses, in a milieu of controlled inflammation, are key to recovery from influenza.

The study provides a comprehensive longitudinal map of immune cellular networks during hospitalized seasonal influenza, demonstrating that: (1) activated cTfh1 and ASC responses peak around days 7–10 and correlate with HA-specific memory B cells; (2) antibody titers increase in magnitude and breadth at convalescence with back-boosting to prior strains; (3) influenza-specific CD4+ and CD8+ T cells are broadly detectable, highly activated during acute illness, and retain memory phenotypes after recovery; (4) early inflammatory cytokine clusters correlate inversely with functional cellular responses; and (5) early lower γδ T and MAIT responses and later weaker adaptive T cell responses are associated with higher clinical severity. These insights highlight immune correlates of recovery and potential biomarkers of severity. Future research should include tissue-resident immune sampling (e.g., airway/BAL), larger and more diverse cohorts (including mild community cases), mechanistic studies linking cytokine milieu to cellular function, and vaccine designs that elicit both IgG and mucosal IgA memory and robust CD8+ T cell responses.

- Cohort skew: Hospitalized adults with high prevalence of comorbidities (86% in influenza+), limiting generalizability to mild community cases, young healthy individuals, or pediatric populations. - Sample size: Modest cohort (approximately 64–65 enrolled; 44 influenza+ and 20 influenza− analyzed), limiting power for some subgroup analyses (e.g., genetic associations). - Disease severity: Seasonal influenza cases were relatively less severe (few ICU admissions), potentially underrepresenting immune features of critical illness. - Causality: Observational design precludes firm causal inferences regarding cytokines versus cellular responses driving outcomes. - Tissue compartmentalization: Lack of respiratory tract samples (e.g., BAL) limits insight into tissue-resident immune dynamics; blood frequencies of antigen-specific T cells may underestimate tissue responses. - HLA/epitope coverage: Although extensive, tetramer panel may miss responses to rarer epitopes/HLAs.