The final study cohort was comprised of 147 adult subjects divided into two subgroups on the basis of age: 18–40 years of age (n = 75) and ≥65 years of age (n = 72). All subjects in the study were recruited in 2014 from Olmsted County, MN, and the surrounding area. Each subject completed a blood draw prior to receipt of the 2014–2015 standard dose TIV, at 24 h post-vaccination, and at 28 days post-vaccination. Peripheral blood mononuclear cells (PBMCs) and serum samples collected at each timepoint were processed for cryogenic storage using previously standardized protocols (Umlauf et al., 2012). All study procedures were reviewed and approved by the Mayo Clinic Institutional Review Board, and written informed consent was obtained from each study participant.

HAI assays were performed using a standardized procedure set forth by the Human Immunology Project Consortium (Wang et al., 2008). Briefly, serum samples were treated overnight with receptor-destroying enzyme (RDE) from Vibrio cholerae (MilliporeSigma; Burlington, MA) and subsequently heat-inactivated for 45 min at 56°C. Serial two-fold dilutions of RDE-treated serum (25 μl; beginning at 1:10) were prepared in 1x PBS and incubated with influenza virus A/H1N1/California/7/2009 (25 μl; hemagglutination titer = 8 HA units) in a 96-well V-bottom plate for 30 min at room temperature. A 0.5% solution of turkey red blood cells (Lampire Biological Laboratories; Pipersville, PA) was subsequently added to each sample and incubated for 45 min at 4°C. Serum samples were analyzed in duplicate with a third measurement being conducted when the initial two runs differed. The HAI titer was reported as the reciprocal of the highest dilution where hemagglutination was inhibited.

CD14⁺ monocytes were isolated from PBMCs and cultured for 2 weeks under conditions supporting differentiation into macrophages. Briefly, CD14⁺ monocytes were isolated from whole PBMCs through negative selection using the human Monocyte Isolation Kit II (Miltenyi Biotec, Inc.; Auburn, CA). Purified CD14⁺ monocytes were incubated in serum-free Macrophage-SFM culture media (Thermo Fisher Scientific; Waltham, MA) supplemented with 20 mM HEPES, 10 ng/mL GM-CSF, 1% penicillin/streptomycin, and 1% amphotericin B, for 2 weeks at 37°C. Fresh culture media was added every 48 h. Cells were subsequently washed with 1x PBS, and non-adherent cells were discarded. Differentiated macrophages were detached using 0.25% trypsin-EDTA, harvested, and plated for all subsequent assays. Differentiation status and purity of the macrophage population was determined visually and by flow cytometry, by decreased expression of CD14 and increased expression of CD16 (Fluorochrome-labeled antibodies obtained from BD Biosciences, CA) (Akagawa et al., 2006; Jin and Kruth, 2016).

CD14 ⁺ macrophages were isolated from differentiation cultures and 5 × 10⁴ cells added to each well of a 96-well tissue culture plate (round bottom). Cells were incubated at 37°C with media alone (unstimulated), influenza virus A/H1N1/California/7/2009 (MOI = 0.5), resiquimod (R848, 1 μg/ml; Mabtech, Inc.; Cincinnati, OH), or both influenza virus and R848. After 24 h, cell culture supernatants were collected and assayed for IL-1β, pro-IL-1β, caspase-1, and IL-18 using Quantikine® ELISA kits (R&D Systems, Inc.; Minneapolis, MN) and Human IL-18 Instant ELISA kits (Thermo Fisher Scientific; Waltham, MA) according to the respective manufacturer’s protocols. The coefficient of variation (CV) for all cytokine ELISA assays are presented in Supplementary Table S1.

CD14⁺ macrophages from the cytokine assays detailed above were stored at −80°C in RNAprotect Cell Reagent (Qiagen, Inc.; Valencia, CA) until further use. Cell pellets were subsequently thawed and total RNA was extracted using RNeasy Mini kits (Qiagen, Inc.; Valencia, CA) according to the manufacturer’s protocol. Isolated RNA was quantitated, normalized for concentration across all samples, and sent to the Mayo Clinic Medical Genome Facility for quality control and analysis. All RNA samples were analyzed using an inflammasome pathway-focused RT² Profiler™ PCR Array (PAHS-097Z, Qiagen, Inc.; Valencia, CA).

Difference between age groups in demographic variables were assessed using Pearson’s chi-square test. Correlations between variables are calculated using Spearman’s rank correlation, Wilcoxon’s rank-sum test was used to test for differences between age groups and sexes for all continuous variables. Wilcoxon’s signed-rank test was used to test for differences when the data is paired (e.g. stimulated vs unstimulated values). Gene expression values (ΔCt) were first normalized by subtracting the cycle threshold (Ct) from the corresponding B2M housekeeping gene from the Ct value for each gene. Log₂ fold-changes for stimulated vs unstimulated samples were calculated by taking the difference between the ΔCt values for the unstimulated minus the ΔCt from the stimulated samples.

The study cohort of 147 subjects ranged from 18.1 to 92.3 years of age and was designed to equally represent both younger (18–40 years of age, n = 75) and older (≥65 years of age, n = 72) age groups. The study cohort was predominantly White (89%), with the remaining individuals identifying as either Asian (3%), Other (1%), or electing not to disclose their ethnicity (7%). Most of the study participants (89%) identified as non-Hispanic or Latino. The younger and older age subgroups were 58.7 and 59.7% female, respectively. See Table 1 for a complete summary of subject demographics.

As expected, HAI titers against influenza A/H1N1 were significantly higher 28 days after vaccination with TIV (Day 28 median HAI titer = 1:160; p = 8.6 × 10⁻¹⁰) compared to pre-vaccination titers (Baseline median HAI titer = 1:80) in all subjects as well as within both younger (p = 1.9 × 10⁻⁵) and older (p = 8.3 × 10⁻⁶) age subgroups (Figure 1A). Younger subjects exhibited significantly higher median HAI titers compared to older subjects both at Baseline (1:160 vs 1:40; p = 2.3 × 10⁻⁸) and Day 28 (1:160 vs 1:80; p = 2.5 × 10⁻⁸) (Figure 1B). Following vaccination, 7 of 114 (6%) of the subjects were found to seroconvert (as defined by a 4-fold increase in HAI titer), and 94% of the subjects were seropositive, defined as an HAI titer ≥1:40, at Day 28 which is also considered a correlate of protection (Cox, 2013). We compared seroconversion rates in subjects with low HAI titers (<40) before vaccination to subjects with higher HAI titers (≥40) and found: 4/22 (18.2%) versus 3/92 (3.3%%) seroconverted, respectively. When comparing the younger (18–40) and older (≥65) subgroups, we found that 10% of the younger subjects seroconverted and 100% were seropositive at Day 28; while 2% of the older subjects seroconverted and 87% were seropositive at Day 28. Recognizing that the rates of seroconversion were low, we relaxed the definition to a 2-fold rise in HAI titer and repeated the analysis, revealing a similar seroconversion trend with respect to baseline HAI titer (40.9% of low HAI subjects seroconverted compared to 10.0% of high HAI subjects), and a similar rate of seroconversion for younger and older subjects (16.7% for both).

No significant difference in HAI titer was detected between men and women in the overall cohort; however, sex-based differences were apparent within each age subgroup (Figure 1C; Supplementary Figure S1). Younger men exhibited slightly higher HAI titers than young women at Baseline (p = 0.048), and this disparity became more apparent following vaccination (p = 0.019). Conversely, women in the older age group exhibited markedly higher HAI titers than men following vaccination (p = 0.0021), despite having similar HAI titers at Baseline (p = 0.38).

In order to assess inflammasome activity, we measured IL-1β, pro-IL-1β, IL-18, and caspase-1 production by PBMCs in response to in vitro stimulation with the following: A/H1N1 influenza virus, resiquimod (R848, a TLR7 agonist), or a combination of A/H1N1 and R848. Stimulation of whole PBMCs resulted in high but variable levels of pro-IL-1β and IL-1β, while production of IL-18 and caspase-1 was much lower but also inconsistent (data not shown). Because monocytes and macrophages have been suggested to play a key role in immune responses to influenza with age (Roberts, 2020), we refocused our analyses on CD14⁺ peripheral blood monocytes. Monocytes were isolated using negative selection to minimize manipulation of the cells and cultured in vitro under conditions to promote macrophage differentiation prior to stimulation (Jin and Kruth, 2016). As expected, differentiated macrophages had an oval, “fried egg morphology” and expressed high levels of CD16 with little to no CD14 (Akagawa et al., 2006).

We first evaluated the extent of inflammasome activity in macrophages at baseline. Levels of inflammasome-related proteins and cytokines are summarized in Figure 2 and Supplementary Tables S2–S4. Influenza stimulation resulted in a significant percent increase in caspase-1 (27.2%, p = 0.007), pro-IL-1β (19.1%, p = 2.22 × 10⁻¹⁶), and IL-1β (22.8%, p = 2.22 × 10⁻¹⁶) secretion relative to unstimulated macrophages. Treatment with R848 induced similar increases in pro-IL-1β (24.5%, p = 2.22 × 10⁻¹⁶) and IL-1β (33.6%, p = 2.22 × 10⁻¹⁶) production, but a slight decrease in caspase-1 levels (-3.1%, p = 0.02). Combined stimulation with influenza and R848 resulted in markedly increased production of pro-IL-1β (153.9%, p = 2.22 × 10⁻¹⁶) and IL-1β (339.8%, p = 2.22 × 10⁻¹⁶), with more moderate but significant increases in caspase-1 (32.0%, p = 0.009) and IL-18 (8.5%, p = 2 × 10⁻⁵).

We next evaluated if age or biological sex were associated with baseline inflammasome activity (Supplementary Tables S3 and S4). Interestingly, only IL-1β secretion in response to influenza virus at day 1 (33.8 vs 22.4 pg/ml; p = 0.025) and IL-18 secretion in response to R848 at baseline (1.35 vs 0.37 pg/ml; p = 0.021) were significantly associated with age and both were higher in older adults. Regarding sex-based differences, females exhibited significantly higher caspase-1 (21.16 vs 3.16 pg/ml; p = 0.029) and pro-IL-1β levels (50.97 vs 32.9 pg/ml; p = 0.016) in response to influenza stimulation at baseline, whereas males secreted higher levels of pro-IL-1β (49.39 vs 24.9 pg/ml; p = 0.034) and IL-18 (3.06 vs 0.83 pg/ml; p = 0.0093) at Day 1 in response to stimulation with R848 or influenza, respectively.

Influenza vaccination did not appear to have an immediate effect on macrophage inflammasome activity, although IL-1β levels were slightly elevated in unstimulated macrophages 24 h post-vaccination compared to baseline (161.27 vs 152.01 pg/ml, p = 0.01). Stimulation of Day 1 macrophages with influenza virus elicited a significant increase in caspase-1 production (35.7%, p = 9.8 × 10⁻⁵), but more moderate increases in IL-18 (4.2%, p = 0.0013), IL-1β (15.7%, p = 2.22 × 10⁻¹⁶), and pro-IL-1β (13.4%, p = 3.43 × 10⁻¹⁴) compared to unstimulated. Treatment with R848 induced moderate increases in IL-1β (36.7%, p = 2.22 × 10⁻¹⁶) and pro-IL-1β (15.9%, p = 4.49 × 10⁻¹⁵) production, while combined influenza and R848 stimulation resulted in significant increases for all four proteins (Figure 2, Supplementary Table S2). Interestingly, the magnitude of caspase-1 (p = 0.012) and IL-18 (p = 0.024) production in response to influenza + R848 was greater at Day 1 compared to baseline, while the opposite was true for IL-1β (p = 0.031) and pro-IL-1β (p = 0.024) secretion in response to influenza alone. We did not observe any associations between inflammasome-mediated cytokine production and change in HAI titer after vaccination (R ranged from -0.057 to 0.031 and p > 0.56 for all 4 outcomes). Correlations with inflammasome activity were also weak to non-existent for baseline and Day 28 HAI titer.While we believe that signaling via TLR7 plays an important role (as evidenced by our data from cells stimulated with R848 alone), but we do not think a TLR7-related mechanism/pathway solely explains our observations. Stimulation of cells with A/H1N1 alone often resulted in similar levels of cytokine production as R848 alone (see Supplementary Table S2). Only when both signals were provided to the cells did we note a synergistic effect on cytokine/protein production–which is presumably due to complete inflammasome activation.

To further characterize the contributions of inflammasome activation to vaccine-induced influenza immunity, we isolated RNA from cultured macrophages at Baseline and evaluated the expression levels of 84 inflammasome-related genes using a custom RT-PCR array. Baseline unstimulated gene expression was associated with age in a single gene (P2RX7) which had a significantly lower expression in older subjects (p = 0.006). Comparing baseline unstimulated gene expression between sexes, eight genes (i.e., CFLAR, NFKBIB, NFKBIA, CCL7, CHUK, HSP90AB1, RIPK2, PYCARD) were identified with significantly higher expression levels (p < 0.01) among women. There were fewer differences observed between old and young individuals, suggesting that macrophages from both age groups had a similar ability to activate inflammasome pathways upon stimulation. Older subjects had greater expression of CASP5 (p = 0.013) and lower expression of CCL2 (0.043) and CFLAR (0.006). Following stimulation of baseline macrophages with A/H1N1 and resiquimod, we identified 51 genes with significantly altered expression levels (28 upregulated, 23 downregulated; p < 0.01) (Figure 3, Table 2). When evaluating the effects of both age and sex on gene expression in response to influenza stimulation, only CFLAR expression was statistically significant for age (p = 0.006), but not with sex (p = 0.057).

We detected several correlations between changes in gene expression at baseline and the secretion of inflammasome-related proteins. Caspase-1 secretion at baseline was moderately associated with TXNIP expression (r = 0.278, p = 0.01), while IL-18 secretion at Day 1 was positively associated with CASP1 expression (r = 0.344, p = 0.002) and negatively associated with TNF expression (r = −0.372, p = 0.007). Secreted levels of pro-IL-1β were positively associated with the expression of several inflammasome-related genes at baseline (PTGS2: r = 0.421, p < 0.005; RELA: r = 0.318, p = 0.002; RIPK2: r = 0.265, p = 0.01) and 24 h post-vaccination (CCL2: r = 0.385, p = 0.002; HSP90AB1: r = 0.311, p = 0.006). IL-1β secretion at baseline was positively associated with CXCL1(r = 0.437, p < 0.005) and PTGS2 (r = 0.423, p < 0.005) expression but negatively associated with NAIP (r = -0.423, p = 0.004), PEA15 (r = -0.297, p = 0.004), and MAPK3 expression (r = −0.272, p = 0.009).

Interestingly, we did not detect any robust correlations between the expression of inflammasome-related genes in response to influenza stimulation at baseline and HAI titers at either timepoint.