Study design
We selected samples from a longitudinal population-based cohort of Norwegian healthy seniors before and after two doses of vaccine against SARS-CoV-2 (Table 1) 16. Response to
Table 1
Category
|
N (male/female)
|
Median age (range), yrs
|
Low responders (LR)
|
7 (5/2)
|
72 (66–78)
|
High responders (HR)
|
7 (4/3)
|
70 (69–73)
|
vaccination was quantified by measuring serum levels of anti-RBD IgG, as well as by the frequencies of responding spike-specific CD40L+TNF-α+ CD4 T cells in peripheral blood upon in vitro activation by SARS-CoV-2 spike peptides (Fig. 1a). Based on the change in frequency of spike-specific CD4+ T cells after the second vaccine dose compared to baseline, we selected 14 individuals and categorized them as low and high responders according to specific T cell responses (frequency of CD40L+TNF-α+ CD4+ T cells lower than or greater than 0.05% (pre-vaccination – post dose 2, respectively, n = 7 each) (Fig. 1b). Despite displaying heterogeneity, the low responder (LR) group was found to have significantly lower amounts of circulating anti-RBD IgG than the high responders (HR) (Fig. 1c).
Vaccine low and high responders have comparable frequencies of major circulating cell types
To test if differences in circulating immune cell frequencies and function could explain the differential induction of humoral and cellular immunity in response to vaccine between low and high responders, we deployed mass cytometry on unstimulated, SARS-CoV-2 spike peptide-, and Cytostim-treated cells utilizing a custom panel of 40 antibodies against both surface and intracellular antigens (Supplementary table 1). Projection of this multidimensional data onto uniform manifold approximation and projection (UMAP) embedded space revealed distinct clusters of cells corresponding to the major cell types within human peripheral blood mononuclear cells (PBMCs), identifiable by the expression of canonical markers (Supplementary table 2): B cells, CD4+ T cells, CD8+ T cells, γδ T cells, NK cells, NK T cells, and myeloid cells (Fig. 1d). Comparison of frequencies of these major cell types between low and high responders did not reveal any significant differences, irrespective of vaccination or stimulation condition (Supplementary Fig. 1).
Pre-vaccination differences in frequencies of B cell subsets are associated with vaccine responses
Initially, we tested whether preexisting differences in cell subset frequencies could discriminate between low and high responders to vaccination. We reclustered each of the major cell types based on expression of all phenotypic and functional markers as visualized by UMAP (Supplementary Fig. 2). After reclustering of the B cells, we identified nine distinct clusters and plotted them onto UMAP-embedded space to reveal phenotypically distinct subclusters within the B cell compartment (Fig. 2a). Overlaying canonical B cell markers on the UMAP plot revealed the natural separation of B cells into CD27+IgD- class-switched memory B cells (SwM), CD27+IgD+ non-class-switched memory B cells (NSwM), CD27-CD24+CD38+ transitional B cells (Trans B), and CD27-CD24-CD38+/- mature B cells, which could be further split into IgD+ naïve B cells and IgD- double negative (DN) B cells (Figs. 2a and b, Supplementary table 2). Moreover, the naïve B cells could be further separated into CD11c- resting naïve cells (resN) and CD11c+ activated naïve cells (Figs. 2a and b, Supplementary table 2). While the frequencies of these subsets did not differ between low and high responders, we found smaller clusters whose frequencies significantly differed between low and high responders before vaccination (Fig. 2c). Specifically, we observed that clusters bearing signatures of Trans B, resN (three phenotypically distinct subsets; CD38-, CCR6- and CCR6+ subsets), and IgM+ NSwM were more abundant within the high responders in pre-vaccinated samples (Fig. 2c). The annotations were performed by inspecting canonical marker expressions for these significantly different clusters (Fig. 2d).
Taken together, our results indicate that higher pre-vaccination frequencies of subsets of mature and memory B cell populations are associated with high responders to SARS-CoV-2 vaccination.
Helper T cell subsets respond differently to activation and antigen recall post-vaccination in low and high responders
Our next objective was to investigate differences in helper T cell responses between low and high responders. Reclustering of the CD4+ T cell compartment identified 17 distinct subsets including regulatory CD25+Foxp3+ T cells (Tregs), as well as smaller subsets of TH1 (CD4+Tbet+), and TH17 (CD4+CCR6+) cell types (Fig. 3a, Supplementary Fig. 3a). Annotation of cell types was based on the expression of key surface antigens and transcription factors that are traditionally known to be associated with CD4+ T cell subtypes (Fig. 3b and Supplementary table S2).
We observed a high degree of heterogeneity for spike peptide- and Cytostim-induced cytokine production between these cell types (Supplementary Fig. 3a).
In contrast to the B cell subsets, we found no significant differences in CD4+ T cell subset frequencies between low and high responders before vaccination (Supplementary Fig. 3b). However, after vaccination, unsupervised clustering revealed significantly different cluster abundances between low and high responders in both spike- and Cytostim-stimulated samples (Fig. 3c). We observed a significantly higher expansion of a subset of CD27−CD40L+ TH cells in high responders with spike recall and a concurrent decrease in frequency of two subsets of CD27+ TH17 cells in high responders (Fig. 3c). The more abundant of these two subsets was also positive for the NK cell receptor CD161 and showed a trend of decreased frequencies in the high responder group irrespective of stimulant (Fig. 3c). The high responders also had lower frequencies of a CD38+CD138+IL-7Rα+ TH cell subset that was apparent for all stimulations but significantly different only for Cytostim. We observed an expansion of IFN-γ+ TH cells with spike stimulation, with high responders responding significantly more (Fig. 3c). Finally, a population of MIP-1β−IFN-γ−GranzymeB+ TH1 cells displayed a trend of higher frequencies in the high responders, but it reached statistical significance only with spike stimulated cells (Fig. 3c).
We then tested if vaccine-induced changes in the 17 CD4+T cluster cell frequencies were different between low and high responders. As shown in Fig. 3d, we observed significantly higher vaccine-induced expansions of IFN-γ+, IL-2+, and MIP-1β+ TH cells in high responders after stimulation with spike peptides. Of the three, the IFN-γ+ and IL-2+ populations also expressed CD40L, while the IL-2+ population expressed low levels of Foxp3 and CD25 (Supplementary Fig. 3a). On the other hand, a subset of CD27+CD161+ TH17 cell type was preferentially expanded in the low responders upon vaccination. We also observed a vaccine-induced increase of a Foxp3+CD25low regulatory T cell subset in the low responders.
A similar analysis to investigate differences in cytotoxic CD8+ T cell compartment did not reveal any cell subset that separated low and high responders, regardless of vaccination status or experimental activation (data not shown).
Distinct subsets of γδ T cells may differentiate between low and high responders to SARS-CoV-2 vaccination
The role of γδ T cells in vaccine response is underexplored. Therefore, our next objective was to test if low and high responders to SARS-CoV-2 vaccination had different abundances of functionally different γδ T cell subsets. Reclustering of the total γδ T cell population revealed nine different subsets (Figs. 4a, b and c). In the pre-vaccination samples, we did not find any differences between low and high responders for these clusters (Supplementary Fig. 4). However, after vaccination we observed marked differences between the two response groups (Fig. 4d) for a CD56+ subset that was significantly more abundant in the high responders, regardless of the stimulation condition (cluster 4, Fig. 4c). We also found two small clusters within the γδ T cell compartment that expanded with Cytostim activation to be significantly higher in the high responders (clusters 6 and 9, Fig. 4d). These last two clusters were relatively lower for CD3 and TCRγδ expression (Fig. 4c), suggestive of their status as early γδ T cells. Moreover, cluster 9 also had high expression of the inhibitory receptor NKG2A and production of the cytokines IFN-γ, TNF-α, MIP-1β and CD40L (CD154) while cluster 6 was NKG2A− but also exhibited MIP-1β expression (Fig. 4c). Among the other subsets, we identified a CD16+ cluster of γδ T cells which was lower in abundance in high responders across all stimulation conditions, although the difference did not reach statistical significance (cluster 8, Fig. 4d).
Taken together, our findings identified functionally distinct subsets of γδ T cells in the peripheral blood that differed between low and high responders to SARS-CoV-2 vaccination.