MAIT cell activation augments adenovirus vector vaccine immunogenicity

Abstract

Mucosal-associated invariant T (MAIT) cells are innate sensors of viruses and can augment early immune responses and contribute to protection. We hypothesized MAIT cells may have inherent adjuvant activity in vaccine platforms that employ replication-incompetent adenovirus vectors. In mice and humans, ChAdOx1 (chimpanzee adenovirus Ox1) immunization robustly activated MAIT cells. Activation required plasmacytoid dendritic cell (pDC)–derived IFN-α and monocyte-derived IL-18. IFN-α-induced, monocyte-derived TNF was also identified as a key secondary signal. All three cytokines were required in vitro and in vivo. Activation of MAIT cells positively correlated with vaccine-induced T cell responses in human volunteers and MAIT cell-deficient mice displayed impaired CD8⁺ T cell responses to multiple vaccine-encoded antigens. Thus, MAIT cells contribute to the immunogenicity of adenovirus vectors, with implications for vaccine design.

Mucosal-associated invariant T (MAIT) cells are unconventional T cells that recognize microbe-derived metabolites of vitamin B2 biosynthesis like 5-(2-oxopropylideneamino)-6-ribitylaminouracil (5-OP-RU) (1). However, MAIT cells can also be activated by cytokines, and thereby respond to viruses, which do not synthesize vitamin B2. In vivo, MAIT cells respond to influenza virus to amplify early local immune responses and protect against lethal infection (2–4). We hypothesized that the ability of MAIT cells to augment early immune responses may play a key role in viral vector vaccine immunogenicity. Replication-incompetent adenovirus (Ad) vectors are highly potent vaccine platforms for many human diseases (5). They have been recently licensed for use against Ebola virus (6) and show promise for SARS-CoV-2 infection (7, 8). We sought to determine if such vectors activate MAIT cells and if this impacts vaccine immunogenicity.

To determine if MAIT cells respond to Ad vectors, we stimulated human peripheral blood mononuclear cells (PBMCs) with Ad5 and chimpanzee adenovirus Ox1 (ChAdOx1), which are leading SARS-CoV-2 candidate vaccines (7, 8). ChAdOx1 induced dose-dependent upregulation of CD69, granzyme B, and interferon (IFN)-γ by MAIT cells (Fig. 1, A to C and fig. S1, A to D), whereas Ad5 only weakly activated MAIT cells. This activation was confirmed using the MR1/5-OP-RU tetramer to identify MAIT cells (fig. S1E).

Figure 1. Activation of human and murine MAIT cells by adenovirus vectors.

(A to C) Human PBMCs (n=9; four experiments) were stimulated with Ad5-GFP or ChAdOx1-GFP (multiplicity of infection (MOI)=0 to 10⁴ vp (viral particles)). MAIT cell CD69 (A), granzyme B (GzmB) (B), and IFN-γ (C) expression was measured after 24 hours. (D and E) Human PBMCs (n=5; two experiments) were stimulated with the indicated vectors (species in parentheses). MAIT cell GzmB (D) or IFN-γ (E) expression were measured after 24 hours. (F and G) C57BL/6J mice (n=6 per group; representative of two experiments) were immunized intramuscularly (i.m.) with 10⁸ IU (infectious units) of Ad5-GFP or ChAdOx1-GFP. Inguinal LN MAIT cell CD69 (F) and GzmB (G) expression was measured after 24 hours. (H and I) Healthy human volunteers (n=14) were immunized with a 5×10¹⁰ vp dose of ChAdOx1 MenB.1. (H) MAIT cell CD69 expression 1 day pre- and 1 day post-immunization. (I) Pearson correlation of change in plasma IFN-γ levels following vaccination with the change in MAIT cell CD69 expression. *, P<0.05; **, P<0.01; ***, P<0.001. Unpaired t test (A to C), two-way ANOVA (D and E), one-way ANOVA with Sidak correction for multiple comparisons (F and G), or Wilcoxon rank-sum test (H). Symbols indicate average response (A to C) or individual mice/volunteers (D to I). Mean ± SEM are shown.

Species C-derived Ad vectors have been shown to poorly stimulate innate immune responses as compared to non-species C vectors (9–11). We tested the relative ability of three species C vectors (Ad5, Ad6, and ChAdN13) and five non-species C vectors (Ad24, Ad35, ChAd63, ChAd68, and ChAdOx1) (fig. S1F) to activate MAIT cells. Following stimulation, we observed greater average activation by non-species C compared to species C vectors (Fig. 1, D and E).

We next tested the ability of Ad vectors to activate MAIT cells in vivo. Intramuscular (i.m.) ChAdOx1 immunization of C57BL/6J mice strongly induced upregulation of CD69 and granzyme B by MAIT cells, whereas Ad5 induced significantly weaker activation (Fig. 1, F and G and fig. S2, A to C). We also observed significant upregulation of CD69 on MAIT cells 1 day following immunization of human volunteers with a candidate ChAdOx1 vaccine (Fig. 1H and fig. S3, A to C). Plasma IFN-γ levels markedly increased post-vaccination (fig. S3D), as seen in non-human primates (10). This increase correlated with levels of MAIT cell activation (Fig. 1I).

To investigate the pathways involved, RNA sequencing (RNA-Seq) of MAIT cells was performed. Eighty-four genes were significantly upregulated in human MAIT cells following vaccination (Fig. 2A and data S1). GSEA (12) identified the strong induction of type I IFN, interleukin (IL)-1 family, IL-12 family, and IL-2 family signaling pathways (Fig. 2B). Changes in post-vaccination plasma IFN-α or CCL2, an IFN-regulated chemokine (13), strongly correlated with MAIT cell activation (Fig. 2C and fig S3, D and E). Comparison of genes upregulated in MAIT cells following human vaccination, vaccination of mice, or in vitro stimulation showed a high degree of overlap. Ninety-eight percent of vaccine-upregulated genes in humans were upregulated in at least one of the other two conditions and 63% were upregulated in both (Fig. 2D and fig. S4, A and B and data S2 to S4). GSEA on murine MAIT cells and in vitro-stimulated human MAIT cells identified similar enrichment of these cytokine signaling pathways (fig. S4, C and D).

Figure 2. Activation of MAIT cells by adenovirus vectors requires pDC-derived IFN-α and monocyte-derived IL-18.

(A and B) Gene expression analysis of MAIT cells isolated from the PBMCs of human volunteers 1 day pre- and 1 day post-vaccination with ChAdOx1 MenB.1 (n=14). (A) Volcano plot of differentially expressed genes (log₂ FC>1, adjusted P<0.05). The top 10 upregulated genes are annotated. (B) Selected cytokine signaling pathways from the Reactome database enriched by Gene Set Enrichment Analysis. (C) Pearson correlation of change in plasma IFN-α level following vaccination with the change in MAIT cell CD69 expression. (D) Overlap of genes upregulated in MAIT cells from ChAdOx1-vaccinated volunteers, from human PBMCs stimulated with ChAdOx1, and from the draining inguinal LNs of ChAdOx1-vaccinated mice. (E and F) Human PBMCs were stimulated with ChAdOx1-GFP and the following inhibitors were used: vaccinia virus-derived type I interferon antagonist B18R (1 or 10 μg/ml; n=7; three experiments) or anti-IFNAR2 antibody (10 or 25 μg/ml; n=5 or 3; two or one experiments, respectively) (E) or anti-IL-12, anti-IL-15, or anti-IL-18 antibodies (10 μg/ml; n=5; two experiments) (F). MAIT cell IFN-γ expression was measured after 24 hours. (G and H) PBMCs were depleted of CD123⁺ pDCs or left untreated and stimulated with ChAOx1-GFP. MAIT cell IFN-γ expression (n=8; three experiments) (G) or levels of IFN-α in the cell culture supernatant (n=4; one experiment) (H) were measured after 24 hours. (I and J) PBMCs were depleted of CD14⁺ monocytes or left untreated and stimulated with ChAdOx1-GFP. MAIT cell IFN-γ expression (n=4; two experiments) (I) or IL-18 levels in the supernatant (n=4; three experiments) (J) were measured after 24 hours. *, P<0.05; **, P<0.01; ***, P<0.001. Repeated-measures one-way ANOVA with Dunnett Correction (E and F), or unpaired t test (G to J). Symbols indicate individual donors. Mean ± SEM are shown.

In vitro, inhibition of type I IFN signaling reduced MAIT cell IFN-γ production by >50% (Fig. 2E). Blockade of IL-18 (an IL-1 family member) or IL-12 also reduced MAIT cell activation. By contrast, blockade of IL-15 (an IL-2 family member) had no effect (Fig. 2F and fig. S5A). MAIT cell activation by Ad vectors was independent of TCR signaling (fig. S5B) (2, 3).

To understand the cellular origins of these critical cytokines, we examined the cell populations transduced by Ad5 and ChAdOx1. Monocytes or conventional dendritic cells (cDCs) were the major transduced population by both vectors (>80% of GFP⁺ cells) (fig. S5, C to F). ChAdOx1 also efficiently transduced CD123⁺ plasmacytoid dendritic cells (pDCs), whereas Ad5 did not (fig. S5F) (11). Importantly, depletion of CD123⁺ pDCs resulted in a significant (67%) reduction in IFN-γ production by MAIT cells (Fig. 2G) and reduced IFN-α levels by >99% following ChAdOx1 stimulation (Fig. 2H).

Depletion of CD14⁺ monocytes significantly reduced MAIT cell activation after ChAdOx1 stimulation (Fig. 2I and fig. S5G) and abrogated the secretion of IL-18 (Fig. 2J). The Cathepsin B–NLRP3 inflammasome pathway (14) was the source of IL-18 in response to ChAdOx1 (fig. S6). Thus, pDC-derived IFN-α and monocyte-derived IL-18 play critical roles in activating MAIT cells in response to Ad vectors. Ad5 induced negligible amounts of IFN-α (fig. S7, A and B) (10, 11). Despite transducing monocytes, Ad5 did not induce IL-18 or IL-12p70 (fig. S7, C and D). By sharp contrast, ChAdOx1 induced robust production of IFN-α and IL-18.

Notably, although IFN-α/β + IL-18 induced production of IFN-γ by MAIT cells in PBMC culture, this was not seen using isolated CD8⁺ T cells (~75% of human MAIT cells express CD8 (15)) (Fig. 3A), despite the induction of CD69 (fig. S8A). Depletion of monocytes reduced MAIT cell IFN-γ production following IFN-α + IL-18 stimulation (fig. S8B). The addition of monocytes rescued the response (Fig. 3B). Since conditioned supernatant, or provision of PBMCs across a transwell significantly rescued MAIT cell IFN-γ production (Fig. 3C and fig. S8C), this suggested the presence of a soluble, monocyte-derived, IFN-α-dependent signal. Unexpectedly, IFN-α-treated monocytes secreted TNF (Fig. 3D and fig. S8D). Additionally, TNFR2 signaling pathways were strongly induced in stimulated MAIT cells (fig. S8, E to G). Thus, we investigated if TNF was the IFN-α-dependent intermediary signal. Addition of TNF or an anti-TNFR2 agonist to isolated CD8⁺ T cells stimulated with IFN-α + IL-18 enhanced MAIT cell IFN-γ production by >300% (Fig. 3E and fig. S8H). Addition of an anti-TNF antibody (adalimumab) inhibited MAIT cell IFN-γ production in response to IFN-α + IL-18 stimulation, or conditioned supernatant (fig. S8, I and J). TNF blockade using either adalimumab or recombinant TNFR2–Fc fusion protein (etanercept), but not a control antibody (vedolizumab), inhibited IFN-γ production by MAIT cells in response to ChAdOx1 (Fig. 3F). Depletion of monocytes reduced ChAdOx1-induced TNF production by 94% (Fig. 3G). Ad5 induced minimal TNF (fig. S8K), consistent with the poor ability to stimulate IFN-α (fig. S7A).

Figure 3. IFN-α acts directly and indirectly through the induction of TNF to activate MAIT cells.

(A) Human PBMCs or purified CD8⁺ T cells (n=3; one experiment) were stimulated with the indicated cytokines (50 ng/ml). MAIT cell IFN-γ expression was measured after 24 hours. (B) Purified CD8⁺ T cells ± CD14⁺ monocytes (n=4; one experiment) were stimulated with IFN-α + IL-18 (50 ng/ml). MAIT cell IFN-γ expression was measured after 24 hours. (C) Purified monocytes (n=3; one experiment) were stimulated with IFN-α (50 ng/ml), or left untreated. After 24 hours, supernatants were transferred ± IL-18 (50 ng/ml) to autologous purified CD8⁺ T cells. MAIT cell IFN-γ expression was measured after 24 hours. (D) TNF production by IFN-α-treated CD14-purified monocytes was measured after 24 hours (n=3; one experiment). (E) Purified CD8⁺ T cells (n=10; four experiments) were stimulated with IFN-α and IL-18 ± TNF (50 ng/ml) or anti-TNFR2 agonist antibody (2.5 μg/ml). MAIT cell IFN-γ expression was measured after 24 hours. (F) PBMCs were stimulated with ChAdOx1 and the following inhibitors were added: vedolizumab (anti-α4β7 integrin antibody, n=8; two experiments), adalimumab (anti-TNF antibody, n=11; three experiments), or etanercept (TNFR2-Fc fusion protein, n=8; two experiments) (10 μg/ml). MAIT cell IFN-γ expression was measured after 24 hours. (G) PBMCs ± CD14-depletion were stimulated with ChAdOx1. Concentration of TNF in the supernatant was measured after 24 hours (n=4; one experiment). (H to J) C57BL/6J (n=4), Il18rap^−/− (n=3), Tnfrsf1a^−/−Tnfrsf1b^−/− (n=4), or Ifnar^−/− (n=4) mice were immunized intramuscularly with 10⁸ IU of ChAdOx1-GFP. Naive C57BL/6J mice (n=4) were used as a control. After 24 hours, MAIT cells were isolated from the inguinal LNs and sorted for RNA sequencing (one experiment). (H) Principal component analysis. (I) Heatmap of the upregulated differentially expressed genes (log₂ FC>1, adjusted P<0.05) between MAIT cells from ChAdOx1-immunized and naive C57BL/6J mice, with all other groups shown for comparison. (J) Overlap of the genes upregulated (log₂ FC>1, adjusted P<0.05) in MAIT cells from ChAdOx1-immunized and naïve C57BL/6J mice, and the genes upregulated in MAIT cells from ChAdOx1-immunized C57BL/6J mice as compared to each of the ChAdOx1-immunized knockout strains. *, P<0.05; **, P<0.01. Unpaired t test (B, C and G), repeated-measures one-way ANOVA with Dunnett Correction (E and F). Symbols indicate individual donors. Mean ± SEM are shown.

These data suggest a model in which pDC-derived IFN-α acts directly and indirectly via induction of TNF by monocytes (with IL-18) to activate MAIT cells in response to ChAdOx1 (fig. S9). To test this model in vivo, wild-type (WT) C57BL/6J, Il18rap^−/−, Ifnar^−/−, and Tnfrsf1a^−/−Tnfrsf1b^−/− mice were immunized with ChAdOx1. MAIT cells from these animals were then analyzed by RNA-seq (fig. S10, A and B and data S5 to 8). Principal component analysis identified a strong gradient of activation, where MAIT cells from Ifnar^−/− mice were most similar to those from naïve animals and MAIT cells from Tnfrsf1a^−/−Tnfrsf1b^−/− and Il18rap^−/− mice had intermediate transcriptional profiles (Fig. 3, H and I). The effector genes Cd69 (and at the protein level), Cxcl10, Cxcl11, Ccl5, and Gzmb were all regulated along this gradient (fig. S10, C and D). Other genes (like Cxcl9) were only regulated by TNF signaling (fig. S10D). In total, 51% of the genes induced by vaccination were regulated by ≥1 of these cytokine pathways, and 11% were co-regulated by ≥2 of these pathways (Fig. 3J and data S9). Thus, TNF, IL-18, and especially type I IFN play a critical role in vivo in Ad vector-induced MAIT cell activation.

Human volunteers showed a significant increase in IFN-γ-producing T cells following ChAdOx1 boosting immunization (Fig. 4A). The degree of expansion positively correlated with MAIT cell activation (Fig. 4B). To determine if this was a causal relationship, WT and Mr1^−/− mice, which lack MAIT cells (16), were used (fig. S11, A to C). Following vaccination with ChAdOx1 expressing an optimized HCV antigen (17), Mr1^−/− mice had significantly reduced frequencies of HCV-specific CD8⁺ T cells compared with WT mice (Fig. 4C; fig. S11D). No significant defect in HCV-specific CD4⁺ T cells was observed (fig. S11E). We also observed defects in the CD8⁺ T cell responses of Mr1^−/− mice vaccinated with the candidate SARS-CoV-2 vaccine, ChAdOx1-nCoV-19 (Fig. 4D and fig. S11F) (7). WT and Mr1^−/− mice were then given a homologous ChAd63-OVA prime-boost immunization (fig. S11G) (18). Mr1^−/− mice displayed reduced OVA-specific CD8⁺ T cell responses after both priming and boosting (Fig. 4, E and F). Differences in the microbiome (19) or general immunodeficiency of Mr1^−/− mice did not explain these differences in immunogenicity (fig. S12).

Figure 4. MAIT cell deficiency impacts on T cell responses following ChAdOx1 or ChAd63 immunization.

(A) Frequency of IFN-γ-producing PBMCs measured by peptide ELISPOT in ChAdOx1 MenB.1 vaccinated volunteers pre-boost (n=14) or day 14 post-boost (n=13). (B) Spearman rank correlation analysis of the change in MAIT cell CD69 expression from pre-boost to day 1 post-boost versus the increase in IFN-γ-producing PBMCs from pre-boost to day 14 post-boost. (C) C57BL/6J (n=12) or Mr1^−/− (n=9) mice were immunized intramuscularly (i.m.) with 10⁸ IU of ChAdOx1-HCV-GT1-6_D_TM-Ii+L (two experiments). On day 16, HCV-specific CD107a⁺, IFN-γ ⁺, TNF⁺, or IFN-γ ⁺TNF⁺ CD8⁺ T cell responses were measured. (D) C57BL/6J (n=12) or Mr1^−/− (n=11) mice were immunized i.m. with 10⁸ IU of ChAdOx1-nCoV-19 (two experiments). On day 13, SARS-CoV-2 spike-specific CD107a⁺, IFN-γ ⁺, TNF⁺, or IFN-γ ⁺TNF⁺ CD8⁺ T cell responses were measured. (E and F) C57BL/6J (n=12) or Mr1^−/− (n=12, n=11 post-boost) were primed i.m. with 10⁷ IU of ChAd63-OVA and boosted intravenously on day 28 with 10⁸ IU (squares) or 10⁹ IU (circles) of ChAd63-OVA. SIINFEKL-specific CD107a⁺, IFN-γ ⁺, TNF⁺, or IFN-γ ⁺TNF⁺ CD8⁺ T cell responses were measured either 3 weeks post-prime (E) or 3 weeks post-boost (F). *, P<0.05; **, P<0.01; ***, P<0.001. Wilcoxon rank-sum test (A), or two-way ANOVA (C to F). Symbols indicate individual volunteers/mice. Mean ± SEM are shown.

In summary, MAIT cells can sense the diversity of the Ad vector-induced innate immune activation landscape (e.g. IFN-α, TNF, IL-18), integrating these signals to augment vaccine-induced CD8⁺ T cell immunity. The blend of signals required to maximally trigger MAIT cells described here includes a critical pathway via type I IFN-dependent TNF release and relies on cross-talk between two distinct populations of transduced cells, and varies between adenovirus serotypes. Our data, coupled with studies in the lung (4, 20, 21), support a model that places MAIT cells in a critical bridging position between innate and adaptive immunity. The mechanism by which MAIT cell activation promotes antigen-specific CD8⁺ T cell responses remains to be defined. However, local production of chemokine CXC10 represents a promising candidate as it can promote CD8⁺ T cell priming (22).

It is striking that the activation of MAIT cells is tightly linked to the immunogenicity of adenovirus vectors. This technology has emerged as a potent platform for T cell immunogenicity in clinical trials for HIV (23), and as vaccines for emerging viruses such as Ebola (6) and SARS-CoV-2 (7, 8). This knowledge can be harnessed to improve the design of these vaccines against major pathogens and cancers.

One-sentence summary.

Maximal immunogenicity of candidate adenovirus vaccine vectors requires the activation of MAIT cells.

Data Availability

All gene expression data are deposited under GSE158835. All data are available in the manuscript or the supplementary materials.