Links between fecal microbiota and the response to vaccination against influenza A virus in pigs

Swine influenza A virus (IAV) is widespread globally and causes significant morbidity, economic losses, and animal welfare concerns in pig production. Vaccination is a key sanitary measure to reduce susceptibility and transmission, yet substantial inter-individual variability in vaccine responses is observed. Beyond host genetics, growing evidence in humans and mice indicates that the gut microbiota modulates vaccine efficacy and disease severity through microbial metabolites and components influencing innate and adaptive immunity. This study investigates in pigs: (i) whether pre-vaccination fecal microbiota composition is associated with the intensity of the humoral response to an inactivated IAV vaccine, and (ii) whether vaccination affects fecal microbiota dynamics compared with non-vaccinated controls. The work aims to identify microbial features predictive of strong or weak vaccine responses and to understand microbiota–immunity interactions during a critical early-life window around weaning.

Prior studies showed microbiota can enhance or dampen vaccine responses. Bacterial components can stabilize enteric viruses (e.g., histo-blood group antigen-expressing bacteria stabilizing norovirus; LPS and N-acetylglucosamine for poliovirus) and shape respiratory immunity via gut–lung axis. Dysbiosis increases influenza severity in mice and alters influenza vaccine response in humans without pre-existing immunity. Mechanisms include short-chain fatty acids, desaminotyrosine, and flagellin acting as adjuvants via TLR pathways; bacterial products are being explored as vaccine adjuvants. In pigs, previous work linked higher abundance of Prevotella OTUs in pre-vaccination feces with stronger responses to Mycoplasma hyopneumoniae vaccine. Enterotypes in pigs around 60–70 days relate to mucosal immunity. These data support examining microbiota–vaccine response links in swine IAV vaccination.

Design and animals: 131 piglets (Large White) from 26 litters were reared under standard indoor conditions; 98 were vaccinated and 33 served as non-vaccinated controls. Piglets were weaned at ~28 days (range D24–D30). Vaccine: Commercial inactivated trivalent IAV vaccine (Respiporc Flu3; H1N1, H3N2, H1N2) given intramuscularly at D28 with a booster at D49. Sampling: Blood at D28, D49 (21 dpv), D56 (28 dpv), D63 (35 dpv), D146 (118 dpv) for humoral responses; feces at D28, D49, D63, D146 for microbiota. Fecal score (diarrhea) and body weight recorded at weaning. Immunoassays: IAV-specific IgG quantified by semi-quantitative ELISA against the three homologous strains; HAI assays performed with standard protocols using turkey RBCs; HAI titer defined as reciprocal of highest inhibiting dilution. Microbiota sequencing: DNA extracted from 200 mg feces; V3–V4 16S rRNA amplicons sequenced (Illumina MiSeq, 2×300). Processing with FROGS pipeline: read assembly, primer trimming, length filtering (300–490 bp), Swarm clustering (d=1 then 3), chimera removal (VSEARCH), rare OTU filtering (<0.005% total), taxonomic assignment via BLAST+ against SILVA v132, phylogeny by MAFFT/FastTree. Counts rarefied at 10,000 per sample for alpha/beta diversity. Final datasets after QC: D28 n=89 vaccinated; D49 n=97; D63 n=98; D146 n=83; 74 pigs had all four time points. Diversity and composition analyses: Alpha diversity (richness, Shannon, evenness, inverse Simpson) and beta diversity (Bray–Curtis; NMDS). Effects of sex, batch, age and weight at weaning assessed; Adonis used for beta diversity associations. Enterotypes at D63 identified using Jensen–Shannon divergence at genus level. Association with vaccine response: Linear mixed-effects models tested associations between alpha diversity (at D28, D49, D63) and contemporaneous/subsequent log10 IgG and log2 HAI (fixed covariates: weaning age, sex, batch; random: litter). Extreme responder groups (high responders, HR; low responders, LR) defined per time point: for IgG, HR > mean+1 SD, LR < mean−1 SD; for HAI, thresholds set per dpv (e.g., D49 HR ≥160, LR ≤20; D56/D63 HR ≥640, LR ≤160; D146 HR ≥80, LR ≤10). Differential abundance: metagenomeSeq with cumulative-sum scaling and zero-inflated Gaussian models tested OTU differences between HR and LR, adjusting for sex, weaning age, batch; FDR 0.05. Co-abundance network among significant OTUs via FastSpar bootstrap correlations (p<0.05; |rho|≥0.4/0.5). Predictive modeling: sPLS-DA (mixOmics) identified minimal OTU sets predicting HR vs LR at each time point for IgG and HAI; tuning via repeated 5-fold CV to select number of OTUs/components; performance by balanced error rate (BER), AUC, p-values. All predictive OTUs merged (81 unique) for PLS-DA to improve classification. PLS regression using the 81 OTUs on the whole vaccinated cohort (n≈86–89) predicted continuous IgG at D63; evaluated by Q2, Pearson correlation, and residual normality. Raw sequences: NCBI SRA PRJNA647267.

- Vaccine responses: In vaccinated pigs (n=98), IAV-specific IgG and HAI rose after vaccination and booster, peaking around D56–D63, with persistence at D146 (Table 1). - Pre-vaccination microbiota composition (D28): 4,182,259 reads yielded 2,402,310 sequences; mean 26,992 per sample (11,693–48,357). Identified 1,172 OTUs across 10 phyla, 41 families, 152 genera. Dominant phyla: Firmicutes 54.8%, Bacteroidetes 30.5%; also Fusobacteria 6.0%, Spirochaetes 4.6%, Proteobacteria 3.2%. Eight core OTUs found in all samples (cumulative abundance 11.3%). A subset of 12/89 piglets showed high Fusobacteria (>12%) and had more diarrhea at D28 (Kruskal–Wallis p=6.8×10⁻¹⁰). - Post-weaning dynamics: From D49–D146, Bacteroidetes dominant but decreasing (64% to 48%), Firmicutes increasing (28% to 46%). Prevotella 9 markedly increased (≈44% at D63) and remained dominant to D146. Two enterotypes at D63 (Prevotella/Mitsuokella vs Treponema) were identified but showed no association with vaccine response. - Vaccination effect on microbiota: No significant alpha (richness, Shannon) or beta (Bray–Curtis) diversity differences between vaccinated (n=89) and non-vaccinated (n=29) at D49, D63, D146 (all nominal p>0.2–0.3; FDR>0.5–0.9). Some rare OTUs showed differential abundance at D28 (113), D49 (74), D63 (11), D146 (21), suggesting minor effects amid high inter-individual variability. - Alpha diversity vs vaccine response: Higher D28 richness associated with stronger responses at peak: IgG at D56 (p=0.023; FDR=0.092) and D63 (p=0.036; FDR=0.096); HAI at D56 (p=0.049; FDR=0.098) and D63 (p=0.016; FDR=0.092). Suggestive positive relationship between microbial richness and vaccine response magnitude. - Beta diversity vs vaccine response: Significant D28 community composition difference between HR and LR defined by IgG at D63 (Adonis p=0.014; FDR=0.112), with clustering in NMDS; other time points non-significant or only trends. - Differential OTUs at D28 predicting peak response: 23 OTUs differed between future IgG HR vs LR at D63. Enriched in HR (8 OTUs): three Prevotella 2 (Clusters 520, 11, 75), Paludibacteraceae (Cluster_199; 2.0% HR vs 0.01% LR; FDR=0.0008; Log2FC≈−5.07 as defined), Bradymonadales (Cluster_55; 1.07% vs 0.03%; FDR=0.004), Ruminococcaceae UCG-005, Ruminococcus 1, CAG-873 (all <1%). Enriched in LR (15 OTUs): Helicobacter (Cluster_230 FDR=0.02; Cluster_492 FDR=0.03), Bacteroides (Cluster_44 FDR=0.03; Cluster_330 FDR=0.02), Christensenellaceae R-7 group (Clusters 169 FDR=0.004; 165 FDR=0.03), Succinivibrio, Escherichia-Shigella (Cluster_60; mean 3.3% in LR; FDR=0.03), among others. Co-abundance network showed correlated Prevotella 2 OTUs (rho up to 0.68) and associations within Paludibacteraceae/Bradymonadales and within LR-enriched taxa. - Predictive OTU sets (sPLS-DA): D28 OTUs predicted HR vs LR with high accuracy. • IgG: D49 (7 OTUs, AUC 0.952, p=0.007), D56 (9 OTUs, AUC 0.959, p=8×10⁻⁴), D63 (3 OTUs, AUC 0.982, p=2×10⁻⁹), D146 (23 OTUs, AUC 1.0, p=6×10⁻⁵). Prevotellaceae (notably Prevotella 2) frequently contributed to early predictions; Ruminococcaceae dominated late predictors (D146). • HAI: D49 (5 OTUs, AUC 1.0, p=1×10⁻¹⁰; strong contributors Mailhella, Lactobacillus), D56 (16 OTUs, AUC 0.923, p=4×10⁻⁷), D63 (22 OTUs, AUC 0.953, p=5×10⁻⁹), D146 (3 OTUs, AUC 0.821, p=7×10⁻⁶). Key families: Lachnospiraceae, Ruminococcaceae, Muribaculaceae, Prevotellaceae, Spirochaetaceae (Treponema 2). • Overlaps: Three OTUs predicted both IgG and HAI at specific time points (e.g., Erysipelotrichaceae Cluster_1094 for HAI D56 and IgG D63). - Combined predictors: Aggregating 81 unique D28 OTUs improved PLS-DA classification; best AUCs reached 1.0 for IgG at D63 and D146; strong performance also for IgG/HAI at D49. PLS regression using 81 OTUs predicted continuous IgG at D63 with Pearson r=0.70 (p=6.5×10⁻¹⁴), Q2=0.095 (just below 0.0975 threshold), and normal residuals. - Later time points: No stable OTU signature maintained over time across sampling; most differentially abundant OTUs at D49/D63/D146 were rare (<0.1%), and patterns varied by time point. Enterotypes at D63 were not associated with HR/LR groups. - Overall, stronger vaccine responses were positively associated with Prevotella spp. (notably Prevotella 2) and Muribaculaceae (e.g., CAG-873/Cluster_53), while weaker responses associated with Helicobacter, Bacteroides, and Escherichia-Shigella.

Findings address the central hypothesis that early-life gut microbiota influences vaccine responsiveness. Higher pre-vaccination microbial richness was linked to greater peak IgG and HAI responses, supporting the concept that a diverse, resilient microbiome promotes effective humoral immunity. Community composition differences at D28, notably enrichment of Prevotella 2 and Muribaculaceae in future high responders and of Helicobacter, Bacteroides, and Escherichia-Shigella in low responders, suggest specific microbial taxa may modulate vaccine responsiveness, potentially via metabolites (e.g., SCFAs like butyrate), LPS immunogenicity, or TLR ligands (e.g., flagellin). Predictive modeling demonstrated that pre-vaccination microbiota profiles can accurately classify and even predict the magnitude and persistence of antibody responses months later, highlighting a critical window around weaning when microbiota may shape immune competence. Vaccination itself had minimal long-term impact on overall microbiota diversity and structure, indicating observed associations predominantly reflect baseline inter-individual microbiota variation rather than vaccine-induced shifts. These results parallel prior pig and human studies implicating Prevotella and S24-7/Muribaculaceae with favorable immune outcomes and enterobacteria or Helicobacter with poorer responses. Functional and species/strain-level resolution will be needed to clarify mechanisms and identify probiotic or adjuvant candidates.

This study demonstrates that fecal microbiota composition before vaccination is associated with variability in humoral responses to an inactivated IAV vaccine in pigs. Greater microbial richness and specific taxa (Prevotella 2, Muribaculaceae) correlate with stronger responses, while Helicobacter, Bacteroides, and Escherichia-Shigella correlate with weaker responses. A combined set of 81 pre-vaccination OTUs robustly predicts early, peak, and persistent antibody responses, indicating predictive utility of early-life microbiota. Vaccination had little effect on overall microbiota diversity/composition. Practically, farm and breeding practices that enhance gut microbial richness may improve vaccine efficacy. Future work should include species/strain-level and functional profiling via whole-metagenome sequencing, mechanistic validation (e.g., fecal microbiota transplantation, gnotobiotic models), and exploration of microbial metabolites or components as adjuvants or probiotics to optimize vaccine responses.

- Observational associations do not establish causality; intervention studies are needed. - 16S rRNA amplicon sequencing limits taxonomic resolution to genus/OTU and infers function indirectly; species/strain-level effects remain unresolved. - Some statistical signals (e.g., beta diversity Adonis FDR) did not remain significant after multiple testing correction; many differentially abundant OTUs were rare. - HR/LR group membership varied by time point; no single OTU signature persisted across ages. - Potential unmeasured environmental factors around weaning could contribute to microbiota variability despite adjustments (sex, batch, weaning age) and random litter effects. - Predictive PLS Q2 was near but slightly below recommended threshold, indicating modest generalizability requiring external validation.