Systems immunology reveals the molecular mechanisms of heterogeneous influenza vaccine response in the elderly

Demographic characteristics and their association to vaccine response in the elderly

We collected whole blood and plasma samples from 234 trivalent inactivated influenza vaccine (TIV)-vaccinees aged 65 and above in Hannover, Germany across two influenza seasons and generated a multimodal dataset covering whole blood transcriptome, plasma proteome, plasma/serum metabolome, and serology from up to five time points (pre-vaccination, days one-three, six-seven, 21-35 and 60-70 post-vaccination). By doing this, different stages of the vaccination response were covered (Figure 1A). Examining the serological data, we observe heterogeneity across the elderly donors, with the majority of donors showing sufficient antibody responses expected to confer protection. Donors were evaluated based on their antibody fold change (day 35 vs pre-vaccination) against the three strains included in the TIV vaccine⁷, and categorised as Triple Responders (TR, >4-fold rise in antibody titres against all three strains), Non Responders (NR, <4-fold rise in titres against all three strains), or Other (>4-fold rise in titres against one or two of the vaccine strains; Figure 1B). As we have access to data from two independent seasons, we used the larger cohort (season 2015/2016, n=200) as a discovery cohort, and the smaller cohort to replicate our findings (season 2014/2015, n=34).

• Download figure
• Open in new tab

Figure 1. Influenza vaccination cohort overview and transcriptome of highest and lowest responders

(A). Overview of the different omics datasets generated in this study. We generated multi-modal datasets covering the transcriptome, proteome and metabolome response to influenza vaccination from up to 234 individuals for different timepoints spanning both pre and post-vaccination, covering two independent seasons of influenza vaccination. The serological data is adapted from earlier publications on the same cohort^{7, 17} (B). Circos plot showing the serological response to trivalent inactivated influenza vaccine (TIV). Each heatmap depicts pre-vaccination antibody titres against the particular strains included in the vaccine (H3N2, H1N1, B), whereas the corresponding histograms show the log2 fold-change in antibody titres against the strains upon vaccination. Donors were classified as TR, NR or Other based on their serological response (outer tiles). (C) Heatmap showing mean gene expression levels across 10 TRs and 10 NRs for each time point. Each row is a gene that is significantly differentially expressed between TRs and NRs, with purple indicating higher gene expression levels and green indicating lower gene expression levels (adj. p-value < 0.01). (D) Top 40 BTMs listed based on gene set enrichment analysis (GSEA) of genes that are differentially expressed between TRs and NRs. Positive (red) normalised enrichment scores (NES) correspond to BTMs upregulated in TRs, whereas negative NES corresponds to BTMs upregulated in NRs. (E) Line plots depicting immune cell population proportions for all timepoints as estimated by CIBERSORT, stratified for TRs and NRs. Each dot is a sample coloured for TRs (red) or NRs (blue).

We observed another layer of heterogeneity among donors based on pre-vaccination antibody titres. Previous studies have discussed the impact of high pre-vaccination hemagglutination inhibition (HAI) titres that lead to reduced antibody fold change upon vaccination^{8, 20–22}. Within our cohort, approximately one-third of the donors (H1N1:36%, B:34%, H3N2:33%) followed this pattern. Nevertheless, the majority of responders in this cohort (TRs and Others) still showed a high antibody response to vaccination that was independent of pre-vaccination titres, suggesting that pre-vaccination titres are not the only variable defining the vaccine response (Figure 1B). Consistent positive correlations among antibody titres against the three influenza strains and positive correlations between HAI titres and microneutralization (MN) titres per strain were observed (Supplementary Figure 1A-1B). We did not observe statistically significant differences in serological responses between elderly males and females (Supplemental Figure 1C). Across two seasons, we observed a significant negative correlation of antibody titres with age for the H1N1 strain, but not for the H3N2 and B strains (Supplementary Figure 1D). These observed patterns were replicated in the smaller cohort.

Distinct transcriptional signatures between elderly triple and non-responders

We first evaluated transcriptome data generated from the 10 highest (TRs) and 10 lowest (NRs) responders to identify transcriptional differences between the groups and the transcriptomic dynamics of vaccine response resulting in 1466 significantly differentially expressed genes (DEGs) (adj. p-value < 0.01, Figure 1C). Functional significance of the differential transcriptome profiles between TRs and NRs (independent of time) was identified using gene set enrichment analysis (GSEA) with blood transcriptome modules (BTMs)²³ as background (Supplementary Data 1). We found that the top 40 most significantly enriched modules showed a consistent upregulation of T and B cell related, and T cell activation modules in TRs, whereas NRs showed upregulation of inflammatory and inflammatory signalling modules (Figure 1D). These pathways were consistently upregulated in TRs and NRs at all time points, including pre-vaccination. These differences were also reflected by different proportions of immune cell populations in these donors using transcriptome deconvolution (CIBERSORT²⁴, Figure 1E and Supplementary Figure 2A). In line with our enrichment results, we found that TRs had higher proportions of T and B cells, whereas NRs showed elevated proportions of neutrophils. These results agree with reported flow cytometry results⁷ and with the observed trend of higher neutrophils in NRs at the pre-vaccination time point (Supplementary Figure 2B). Despite enrichment of monocyte-related BTMs in NRs, we did not observe differences in monocyte proportions between TRs and NRs in deconvolution or cell blood counts. This suggests higher activity of monocytes in NRs irrespective of cell proportions (Figure 1E and Supplementary Figure 2B).

We next evaluated the pre-vaccination BTMs alone that separate TRs from NRs. We found TRs showed enrichment for T cell activation and signalling, Th2 differentiation and plasma/B cell modules (Figure 2A, Supplementary Data 2). A recent study described a high inflammatory pre-vaccination endotype that is predictive of high vaccine responders in adults, but not in the elderly¹³. In line with these results, we did not observe enrichment of inflammatory BTMs at the pre-vaccination time point in TRs. Instead, we observed an opposite pattern, in which TRs had lower enrichment of inflammatory modules compared to NRs (Supplementary Figure 3A). When we evaluated the expression patterns of the genes enriched in these pathways, we found only CD83 and IL23A to be significantly upregulated in TRs at all time points, suggesting that only select genes are important in elderly TRs and support the immune response (Supplementary Figure 3B). While we observed a gene expression pattern associated with cells engaged in humoral responses in TRs (which is in line with their high responsiveness to vaccination), the NRs presented a consistently upregulated inflammatory signature.

• Download figure
• Open in new tab

Figure 2. Low inflammatory status at pre-vaccination stage indicator of high responders

(A) BTMs significantly upregulated in TRs at the pre-vaccination timepoint. Boxplots show median with interquartile range and Tukey whiskers. (B) Line plot summarising changing gene expression in TRs (red) and NRs (blue) upon vaccination indicated by the percentage of significantly differentially expressed genes over time compared to pre-vaccination. (C) Line plots for two clusters of genes, where the cluster "Transient day 3 TRs" contains genes upregulated in TRs at day 3 post-vaccination. Genes in the cluster "consistent up NRs" are upregulated in NRs at all timepoints. Each dot represents the mean of genes per donor for each cluster, line plot stratified for TR and NRs. (D) Mean expression of genes for significant BTMs and HALLMARK pathways clustered shows split based on transient upregulation in TRs (top) and downregulation in TRs at all time points (bottom). (E) Association of CIBERSORT cell proportions with cell cycle BTMs (left) and cell type signature BTMs (right) at seven days post-vaccination. Size of the dots represent adjusted p-values from linear mixed models, whereas colour represents the estimate from the model. Associations with adjusted p-values < 0.05 plotted, for TRs, while all associations for NRs with cell cycle BTMs only irrespective of adjusted p-value threshold (grey). Associations with adj. p-value < 0.05 for cell type signature BTMs for NRs shown.

Major differences in post-vaccine transcriptional dynamics between elderly triple and non-responders

We then assessed transcriptional changes resulting from vaccination over time, in each group separately. Based on significant DEGs at each time point compared to pre-vaccination, we observed distinct post-vaccination dynamics between TRs and NRs. TRs showed an early response at three days post-vaccination, evident from a small proportion of DEGs, followed by the highest change in gene expression at day seven before going back to pre-vaccination gene expression levels. However, NRs did not show significant DEGs at any time point post-vaccination compared to pre-vaccination (Figure 2B and Supplementary Figure 3C). Interestingly, ∼18% of genes significantly upregulated at day three in TRs were also significantly upregulated in NRs at all time points (Figure 1C and 2C). Functionally, the day three transcriptomic profile was enriched in innate immune modules related to antiviral, interferon and complement activation (Supplementary Figure 3D, Supplementary Data 3). Consequently, the upregulated inflammatory genes in NRs were categorised in two clusters depending on their expression patterns: "Transiently up day 3 TRs" or "Consistently up NRs" (Figure 1C and 2C). While “Transiently up day 3 TRs” genes were involved in antiviral response, the "Consistently up NRs" genes were enriched for modules related to AP-1 transcriptional network, neutrophils and other inflammatory signalling modules, which may suggest a chronic inflammatory status in NRs (Figure 1C, 2D). These results highlight that TRs mounted an immune response evident from transcriptional changes at different time points (percentage of DEGs at day three and day seven) post-vaccination. However, NRs maintained an inflammatory profile at all time points, which in turn may impede mounting the vaccine response.

Elderly non-responders show higher activation of NK cells seven days post vaccination

We next examined day seven transcriptional changes in TRs and NRs separately in greater detail, as day seven showed the highest change in gene expression in TRs compared to pre-vaccination (Figure 2B). GSEA results in TRs showed strong positive enrichment of multiple modules particularly related to cell cycle, plasma cells, immunoglobulins and T cell activation, among others. Even though NRs did not show statistically significant DEGs seven days post vaccination compared to pre-vaccination, GSEA at this time point showed very few enrichments related to cell cycle, T cell activation and plasma cells. However, the inflammatory modules were downregulated in both groups compared to pre-vaccination (Supplementary Figure 3E, Supplementary Data 3).

We associated the enriched modules with cell proportions identified through transcriptome deconvolution. Cell cycle modules were strongly associated with plasma cells in TRs (FDR < 0.05) suggesting clonal expansion of plasma cells, whereas cell cycle modules in NRs did not show significant associations with any cell subsets (Figure 2E, left). Transcriptional changes, particularly in plasma cells and B cells seven days post vaccination have also been shown through single-cell studies²⁵. Plasma cell proportions transiently increased at this time point for TRs, whereas this was not seen in NRs (Figure 1E). This observation is supported by previous studies which showed that an increase in antibody secreting cells at day seven correlates positively with high antibody fold change²⁶. Instead, activation modules were strongly positively associated with NK and CD8 T cells in NRs. We also observed similar associations in TRs, but they were weaker than in NRs (Figure 2E). The deconvolution results also show higher proportions of activated NK cells in NRs (Figure 1E). Flow cytometry on donors from the replication cohort also confirmed these findings (Supplementary Figure 4). A higher proportion of NK cells in NRs compared to TRs has also been reported in the single-cell dataset⁷. Increased NK cell proportions and activity in the elderly compared to younger adults has been previously reported⁸. In addition, a significant negative association of neutrophils with T cell activation BTMs in NRs suggests modulation of T cell activation by neutrophils. Finally, in NRs, Tregs were negatively associated with immune activation BTM, which may suggest attempts at suppression of inflammation in response to vaccination. Altogether, the observed associations link cell subsets to the functional changes characterised through BTMs and highlight multiple differences in TRs and NRs post-vaccination. Particularly, we observe higher NK cell proportions and higher activation of NK cells as a characteristic of the elderly NRs.

Post-vaccination plasma proteome and metabolome differences between triple and non-responders highlight key molecules involved in vaccination response

We next evaluated the proteome (Olink-Explore 384 panel) profile between all TRs (n=71) and NRs (n=10) in our cohort, measuring up to 311 proteins. Differential abundance testing between TRs and NRs irrespective of time showed higher CLEC7A in TRs in the discovery cohort (adj. p = 0.03, moderated t-test,), but not in the replication cohort possibly because of the small effect size (Supplementary Figure 5A, 5B). We next examined protein dynamics induced by vaccination separately in TRs and NRs by comparing each time point to pre-vaccination. Unlike the distinct dynamics observed at the transcriptome level in TRs and NRs, changes in protein abundance in both TRs and NRs seven days post-vaccination were observed (Figure 3A, Supplementary Data 4). In TRs, 14 proteins were upregulated at seven days post-vaccination (Figure 3B), among which TNFRSF13B (also known as TACI) has been widely described for its role in B cell proliferation and plasma cell differentiation^{20, 27, 28}. Other proteins upregulated in TRs include IFNLR1^{29, 30}, a receptor for type three interferons; CCL3³¹ or MIP1a, CD48 and SLAMF7. TNFRSF13B, IFNLR1 and CCL3 also showed a trend of increased abundance in the replication cohort at day seven post-vaccination (Supplementary Figure 5B). On the other hand, five distinct proteins were upregulated in NRs at day seven post-vaccination, including PARP1, EGLN1 and SERPINB8. PARP1 plays a role in oxidative stress-induced inflammation³² and also in recruiting NK cells in a viral response³³. EGLN1/PHD2 and SERPINB8 are involved in anti-inflammatory roles^{34, 35}. At day 35, we found relatively fewer changes (Supplementary Figure 5C). Altogether, differences in plasma proteome dynamics in TRs and NRs highlights the differential response to vaccination induced in the two groups.

• Download figure
• Open in new tab

Figure 3 Plasma proteins and metabolite changes upon vaccination are distinct in TRs and NRs

(A) Line plot showing significant changes (padj < 0.05) in protein abundance in TRs (N = 71) and NRs (N=10) post-vaccination. TRs show a higher percentage of significantly regulated proteins, while NRs show a comparably dampened response. (B) Volcano plot of proteins upregulated in TRs and NRs 7 days post vaccination with significant proteins coloured in red and labelled; Proteins were considered significant at adj. p-value < 0.05 (C) Line plot showing significant changes in metabolite abundance in TRs (N=71) and NRs (N=10) post vaccination. (D) Significantly differentially regulated metabolites categorised based on their taxonomic class as annotated by the Human Metabolome Database (HMDB). The upper and lower parts of the bars represent up- and down regulated metabolites, respectively. The "Measured" column represents the total universe of metabolites that we considered (192 endogenous metabolites, see Methods). The TR and NR columns indicate significantly regulated metabolites at each timepoint compared to day 0. (E) Pathways enriched for metabolites with increased or decreased abundance 7 days post vaccination in TRs. Enrichment of pathways was calculated using IMPALA. Pathways with adj. p-value < 0.05 are shown. (F) Integration of proteins and metabolites with increased abundance at 7 days post vaccination with transcriptome pathways in TRs. Linear mixed models were used to estimate the association between protein / metabolite abundance and mean transcriptomic pathway activity (Methods). Drawn links are significant with adj. p-value < 0.05. Colour and width for the links are based on p-values, where wide red links indicate increased significance.

We also assessed untargeted metabolomic data measured by liquid chromatography-gas spectrometry (LC-MS) and focused specifically on 192 endogenous metabolites (see Methods). No significant differentially abundant metabolites between TRs and NRs were found, irrespective of the time points. However, metabolites in TRs showed significant changes at day seven and 35 compared to pre-vaccination, whereas this was less evident in NRs (Figure 3C top, Supplementary Data 5). Summarising the metabolites into taxonomic classes, TRs showed significant increased abundance of organic acids, particularly amino acids, crucial for adequate immune function both at seven and 35 days post-vaccination¹⁴. On the other hand, metabolites with reduced abundance both at seven and 35 days post-vaccination were lipids and lipid-like molecules, more specifically fatty acyls (Figure 3C bottom, Supplementary Figure 5D, 5F). These findings were validated in the replication cohort, suggesting that overlapping metabolic signatures were induced by vaccination in both seasons (Supplementary Figure 5E). Interestingly, four out of five upregulated lipids and lipid-like molecules in TRs are bile acids. Bile acid perturbation has been described in antibiotics-treated donors with low antibody response to vaccination¹². In contrast, NRs did not show any significant differences in metabolite abundance seven days post-vaccination, but instead showed a decrease in lipid abundance 35 days post-vaccination (Figure 3C top and bottom). Pathway over-representation analysis in TRs at day seven showed significant upregulation of metabolic pathways involving amino acid, glycine/serine metabolism and glucose homeostasis (Figure 3E, Supplementary Data 6). On the other hand, downregulated metabolic pathways were related to fatty acids, linoleic acid metabolism and signal transduction, among others. In summary, the increased and sustained abundance of amino acids with reduced abundance of fatty acyls over time suggests the importance of these metabolite classes in vaccine response.

Furthermore, we integrated plasma molecules upregulated at day seven with transcriptome BTMs to derive their functional significance. In TRs, protein concentrations of TNFRSF13B and IFNLR1 were strongly positively associated with B cells and plasma cell immunoglobulin modules from transcriptome (False Discovery Rate (FDR) <0.05). The CCL3, GZMA and CD48 protein concentrations were also significantly positively associated with T cell transcriptional modules, suggesting their role in the activation of T cells. Among metabolites, uridine and tyrosine also showed positive associations with T cell activation and plasma cell numbers, which may imply their involvement in mounting the immune response. Uridine has previously been shown to be important for T cell proliferation³⁶. Beta-alanine, lactic acid and serine all positively correlated with NK cells (Figure 3F). In contrast, no significant associations between the proteome and transcriptional modules were found in NRs. However, EGLN/PHD2, SERPINB8 and NBN were negatively associated with inflammatory pathways (nominal p-value < 0.05, Supplementary Figure 6). Combining these results, we show distinct changes in the proteome and metabolome following vaccination and describe their potential significance in shaping the vaccination response in the elderly vaccine responders.

Multi-omics integration reveals a common axis of variation across layers separating high and low responders

Next, we integrated the three omics layers of the highest (TRs, n=10) and lowest responders (NRs, n=10) to examine the interconnected processes underlying the heterogeneous influenza vaccine response in the elderly. Using multi-omics factor analysis (MOFA)³⁷, an unsupervised method, 15 latent factors were identified which explained different levels of data variation along with varying contributions of omics layers within each latent factor (Figure 4A). Among these, factor three showed a common axis of variation among three omics layers that separated TRs from NRs irrespective of time (Figure 4B). Examining significance of the other factors showed that factors four and six were consistent with the time dynamics differences observed in TRs and NRs (Supplementary Figure 7A).

• Download figure
• Open in new tab

Figure 4 Multiomics integration of high and low responders

(A) Bar plot showing the proportion of explained variance per factor resulting from unsupervised factor analysis method MOFA. Colours within each bar indicate the contribution of each modality. (B) MOFA Factor 3 separates TRs from NRs irrespective of time and across modalities. Y-axis represents the Factor 3 variance value attributed to each donor. P-values are generated using a Wilcoxon ranked-sum test. (C) Top most positive and negative molecules within the metabolome and proteome modalities as calculated by MOFA. Molecules are ranked on MOFA scaled weights for factor 3. Positive weights are for TRs, whereas negative weights correspond to NRs. (D) Abundance of CCL25 and L-arginine in all 81 TR and NRs. P-values are generated using a Wilcoxon ranked-sum test. (E) Association of arginine with cytokine response post influenza stimulation in 500 FG cohorts. P-values are generated using t-test. Partial correlation estimate was calculated, corrected for age and gender (F) Integrative network of transcriptome, proteome and metabolome data. Nodes in the network correspond to molecules (proteome, metabolome) or BTMs (transcriptome). Only molecules and pathways identified by MOFA weights (Supplementary Figure 6F) are plotted. Edges in the network (all adj. p-value < 0.05) are statistical associations from linear mixed models (see Methods), where red edges represent positive associations and blue edges represent negative associations.

To examine the significance of this common axis (latent factor three), we examined the top proteins and metabolites identified through this factor (Figure 4C) and evaluated these in all TRs (n=71) and NRs (n=10). Among the proteins, CCL25 and CCL3 were consistently more abundant in TRs than in NRs. Similarly, for metabolites, the amino acids arginine and methionine were more abundant in TRs (Figure 4D and Supplementary Figure 7B). To further assess the significance of these amino acids in immune function, we examined the association of amino acids with cytokine production capacity, particularly induced upon influenza stimulation in an independent cohort of 500 healthy individuals^{38, 39}. Arginine and methionine showed significant positive associations with IL1β, IL6 and TNFL (Figure 4E, Supplementary Figure 7C). The increased abundance of these amino acids in TRs and their positive association to pro-inflammatory cytokine production may indicate their involvement in shaping the adaptive immune response to influenza vaccination via modulation of cytokine production.

Finally, we then created a network of significant intra- and inter-omic connections by associating plasma molecules with transcriptome BTMs (Methods), overlaying the weights assigned for TRs (red) and NRs (blue) from MOFA latent factor three (Figure 4E). Across omic layers, the transcriptome and proteome were highly connected, whereas the metabolome was less connected to the other two layers. The metabolite arginine was positively associated with TR-related BTMs involving plasma cells and immunoglobulins. Similarly, xanthosine was positively associated with immune activation BTMs, which was estimated as a predictor for NRs. Both xanthosine and orotidine are involved in nucleotide metabolism which plays an important role in inflammation^{40, 41}. We also observed strong positive associations of glyceric acid 1,3-bisphosphate/1,3-BPG with up to seven proteins that are predictors for NRs. In summary, we describe a common axis of variation in transcriptome, proteome and metabolome which highlights the interconnected differences in influenza vaccine response between elderly TRs and NRs.

Pre-vaccination high plasma IL-15 concentration negatively correlates with antibody response

One of the prerequisites for complete protection is the ability to produce sufficient antibodies against all three strains of the influenza vaccine. To prioritise the important pre-vaccination plasma proteins for antibody response to three strains of influenza vaccine concurrently, we used partial least square regression (PLSR) in 200 donors (discovery cohort). This allowed us to perform a systematic investigation of heterogeneity at the pre-vaccination time point. The resulting proteins from the top three PLSR components explain ∼40% variance of the antibody fold change within the cohort (Figure 5A). We evaluated the top 30 proteins from each of the three components to examine the directionality of the association with antibody response per strain (Figure 5B). Overall, the majority of the prioritised proteins showed contrasting associations with the three strains. Interestingly, TNFSF13/APRIL and IL-15 were negatively associated with antibody fold changes for all three strains (Figure 5B,5C and Supplementary Figure 8A left). Both these proteins have been described as pro-inflammatory cytokines and in the pathogenesis of influenza infection^42–44. In the replication cohort, we also found IL-15 negatively associated with high antibody changes for two strains, whereas this was not true for TNFSF13, perhaps due to the small effect size that is hard to replicate in a smaller cohort (Supplementary Figure 8A right, 8B). Single-cell RNA sequencing on human PBMCs showed that IL-15 is mainly expressed by monocytes⁴⁵ (Supplementary Figure 8C). Based on these results, the contrasting relationship of the majority of prioritised proteins against the three strains suggests a complex relationship of these proteins involved in providing complete protection against influenza infection. However, the consistent negative association of IL-15 with antibody response in both the discovery and replication cohort suggests a detrimental role in protection against influenza.

• Download figure
• Open in new tab

Figure 5 Pre-vaccination plasma proteome correlates to the antibody response to vaccination

(A) Proportion of co-variation in proteome and antibody fold change explained by each component for either each strain or the predictors examined through PLSR analysis with 10-fold cross validations (N = 200). The first three components using the predictors were able to explain ∼ 40% co-variation in the antibody response and pre-vaccination protein abundance (B) Heatmap of t-test statistics for top predictors with antibody fold change for top three components as calculated using rank product analysis for these components. Most proteins show heterogenous associations with the antibody fold change against the three influenza strains, with notable exceptions for IL-15 and TNFSF13 in component 1. (C) Correlation plots of IL-15 against antibody fold change for all 200 donors. (D) IL15RA^+/+ and IL15RA^−/− mice were immunised with OVA-Alum or OVA-IFA in the footpad, followed by analysis of antibody titres, and lymphocyte populations from inguinal lymph nodes 11 days post-immunization. (E) Proportion of B (CD19+), T (CD3+), NK (NK1.1+) cells, as well as CD4+ and CD8+ T cell subsets, among splenocytes of unimmunized IL15RA^+/+ and IL15RA^−/− mice. (F) Ratio of Tfh/Tfr cells, calculated from their frequency among total CD4+ cells, within LNs of immunized IL15RA^+/+ and IL15RA^−/− mice. (G) Serum concentration of OVA-specific IgG1 in mice immunized with OVA-alum, or (H) immunized with OVA-IFA. (I) Schematic summarising the mechanism of IL-15-mediated activation of NK cells leading to suppression of germinal centre responses and low production of antibodies.

High IL-15 concentration leads to reduced antibody titres through NK cells mediated suppression of germinal centre responses

IL-15 is important for NK cell proliferation and maturation^46–48. From both bulk and previously reported single-cell transcriptomics⁷, we observed higher frequencies of activated NK cells and total NK cell numbers in NRs (Figure 1E). NK cells have been described to suppress germinal centre (GC) responses^{49, 50}. We therefore examined the impact of a perturbation of the IL-15-NK cell axis on antibody production with IL15RA-deficient mice. We immunised IL15RA^+/+ and IL15RA^−/− mice with ovalbumin (OVA-IFA or OVA-Alum) to examine the resulting humoral response 11 days post-immunization, at the time of maximum GC-response in this animal model⁵¹ (Figure 5D). IL15RA^−/− mice showed a marked reduction of NK cells in the spleen (as anticipated, given the importance of IL-15 for NK cell numbers), without significant differences on B cell populations, but with slight differences on the frequency of CD4+ and CD8+ T cells (a population that can also be affected by IL-15 reduction) (Figure 5E, Supplementary Figure 9A). In the draining lymph node, we observed a consistent increase of Tfh cells and a decrease of Treg and Tfr cells in IL15RA^−/− mice, although without reaching statistical significance (p-values > 0.05) (Supplementary Figure 9B). However, the Tfh/Tfr ratio, critical in the regulation of GC responses⁵², was significantly higher in IL15RA^−/− mice (p-value = 0.02, Wilcoxon ranked sum-test, Figure 5F). In the same mice that had an increased Tfh/Tfr ratio, we observed higher production of antibodies, across two different immunisation protocols (p-values = 0.006 (OVA-Alum) and 0.048 (OVA-IFA), Wilcoxon ranked-sum test, Figure 5G-H). In conclusion, these results suggest that the reduction of IL-15 and NK cells led to more effective antibody production, in line with the data from elderly NRs who displayed higher IL-15 and NK cell activation together with poor antibody response (Figure 5I).

Pre-vaccination plasma malic and citric acid concentrations negatively correlate with antibody response

To evaluate the pre-vaccination metabolite levels with antibody responses against three strains, PLSR was used in a similar way as described above. Herein, the metabolites in the top eight PLSR components explain ∼40% variation of the antibody fold change among donors suggesting high inter-individual variation in the donors as a consequence of metabolic differences (Figure 6A). Metabolites prioritised in the first component showed malic acid and citric acid as top molecules, with negative associations to antibody response (Figure 6B, 6C and Supplementary Figure 9A). To derive the role of citric acid and malic acid in immune function, we associated pre-vaccination abundance of these metabolites with deeply phenotyped immune cell counts and cytokine production induced upon influenza stimulation in two independent cohorts of 300 and 500 healthy individuals, respectively^{38, 39, 53, 54}. Citric acid showed significant negative associations with CD4+CCR6+CCR5+CCR7+ T cells along with three different subsets of immature neutrophils (FDR<0.05) (Supplementary Figure 9B). Malic acid showed strong negative associations with TNFL, IL6 and IL1β cytokine levels upon influenza stimulation (Figure 6D). Furthermore, we found betaine and cysteine as the top predictors from components one and two, respectively, to be consistently positively associated with antibody response across all strains (Figure 6B and Supplementary Figure 9C). Betaine has been described for its inhibition of IL1β production and release^{55, 56} and for its suppression of pro-inflammatory signalling during ageing⁵⁷. These results reflect the impact of pre-vaccination metabolite abundance linked to different physiological immune states across responders culminating in differential vaccination response in the elderly.

• Download figure
• Open in new tab

Figure 6 Pre-vaccination plasma metabolites as modulators for vaccination response

(A) Proportion of co-variation in endogenous metabolites and antibody fold change explained by each component for either each strain or the predictors examined through PLSR analysis with 10-fold cross validations (N = 200). The first 8 components were able to explain ∼ 40% co-variation in the antibody response and pre-vaccination metabolite abundance (B) Plot of metabolites from component 1 with their ranks and estimated percentage of false predictions (PFP) (left), Heatmap of t-test statistic for the top predictors with antibody fold change for the top component as calculated using rank product analysis for these components (right). Most metabolites show heterogeneous associations to the different influenza strains in the vaccine, except for malic acid and citric acid showing negative association while betaine shows positive association. (C) Correlation plots of malic acid against antibody fold change for all three influenza strains for all 200 donors. (D) Association of malic acid to cytokine production upon influenza stimulation in an independent cohort of 500 younger healthy individuals, showing a negative trend. P-values are generated using t-test. Partial correlation estimate was calculated, corrected for age and gender (E) Top metabolites for component 2 with their ranks and estimated percentage of false predictions shows long chain fatty acids as top candidates. (F) Negative association of these long chain fatty acids identified in PLSR component 2 with IL-15. P-values are generated using t-test. Partial correlation estimate was calculated, corrected for age and gender (G) Schematic describing the role of IL-15 in delayed neutrophil apoptosis, maturation and activation of NK cell populations, leading to suppression of GC responses and reduced antibody production and a chronic inflammatory status. Evidence for each observation is derived through different omics layers and experiments. Connections among different observations were also supported by published findings.

Unsaturated long chain fatty acids are negatively associated with pre-vaccination IL-15 abundance

Given the putative role of high IL-15 in suppressing antibody production in the elderly, we evaluated the relationship of pre-vaccination IL-15 abundance and metabolites. Pre-vaccination IL-15 showed strong negative associations with odd-chain (pentadecanoic acid) and certain unsaturated long-chain fatty acids (LCFAs, e.g. palmitoleic acid) (p value < 0.02) (Figure 6F and Supplementary Figure 9D). These LCFAs were also among the top 30 predictors for PLSR component two, explaining variation in antibody fold change among the donors (Figure 6E and 6F). LCFAs and especially poly-unsaturated fatty acids have been described as important immunomodulatory metabolites to protect against infections⁵⁸. To test the significance particularly of LCFAs in suppressing inflammatory proteins, the relationship of all pre-vaccination lipid and lipid-like molecules (taxonomy class) with these proteins was examined. LCFAs were negatively associated with inflammatory proteins, whereas other lipid molecules showed both positive and negative associations (Supplementary Figure 9E). We also examined the role of LCFA in a younger cohort (500FG^{38, 39}, with 98% donors less than 65 years of age), however, the LCFA-specific negative associations with inflammatory proteins were absent (Supplementary Figure 9E). Among the LCFAs, the most significantly negatively associated metabolite was pentadecanoic acid, an odd-chain fatty acid with anti-inflammatory properties and resistance to oxidation^{59, 60}. Taken together, the data suggests an age-dependent role of LCFAs in the control of inflammation. We described different metabolites associated with IL-15 and other inflammatory proteins, as well as the antibody fold change, suggesting their role in modulating the vaccine response in the elderly, particularly LCFAs as potential pre-vaccination suppressors of high inflammation.