Defining mucosal immunity using mass cytometry following experimental human pneumococcal challenge

Streptococcus pneumoniae (Spn) is a common cause of respiratory infection, but also frequently colonises the nasopharynx in the absence of disease. We used mass cytometry to study immune cells from nasal biopsy samples, collected following experimental human pneumococcal challenge, in order to identify immunological changes that follow and control spn colonisation. Using 37 markers, we characterized 293 nasal immune cell clusters, of which 7 were associated with Spn colonisation. B cell and CD8+CD161+ T cell clusters were significantly higher in non-colonised than in colonised subjects. Spn colonization led to recirculation of not only Spn-specific but also aspecific nasal B cells. This associated with increased numbers of circulating plasmablasts and increased antibody levels against the unrelated bacterium Haemophilus influenzae. In addition, we demonstrated that baseline functionality of blood mucosal associated invariant T (MAIT) cells associated with protection against Spn. These results identify new host-pathogen interactions at the mucosa upon Spn colonisation.


Introduction
twelve subjects remained carriage - (Fig. 1A). Biopsies yielded a median of 2.3x10 5 cells (IQR: Viable immune cells were manually gated from all acquired events and subsequently clustered by hierarchical-stochastic neighbour embedding (h-sne) using Cytosplore software (Fig. 1C,2) 48 (Abdelmoula et al., 2018;Hollt et al., 2016;van Unen et al., 2017). H-sne is a recently developed 49 method in which t-distributed stochastic neighbor embedding (t-sne) is performed sequentially to 50 cluster first global cell populations, each of which is then in turn clustered into subpopulations.

51
Based on the expression of 37 markers, a total of 199,426 immune cells from all subjects were 52 divided into nine lineages (CD8 + T cells, CD4 + T cells, myeloid cells, innate lymphoid cells, B cells, 53 double-negative T cells, granulocytes, CD117 + cells and plasma cells, in order of decreasing 54 abundance). These cell lineages were further divided into twenty-two subpopulations and 293 55 clusters ( Fig. 1C and Table 2). Cell numbers were normalized to the number of stromal cells for 56 each subject to correct for varying biopsy yields. Normalized abundancies were then compared 57 between carriageand carriage + subjects for each of the lineages, subpopulations and clusters.

58
There were no significant differences in frequencies between total lineages or subpopulations 59 between carriageand carriage + subjects. However, at a finer level seven clusters were 60 significantly higher in carriagethan in carriage + subjects (Fig. 1C, blue bars). Of note, three B cell 61 clusters were higher in carriagesubjects (Fig. 1C). Moreover, three CD8 + T cell clusters, all 62 expressing CD161, and one CD8 dim T cell cluster were higher in carriagesubjects than in 63 carriage + subjects (Fig. 1C). The seven significant clusters strongly correlated (r>0.70) with eighty-64 eight clusters in other lineages/subpopulations, sixty-eight of which were in B or T cell lineages, highlighting an interconnectivity between B and T cell subpopulations in the human nasopharynx ( Fig. 1C).
was not due to other shifting B cell populations (Supplementary Fig. 2A). We then investigated 95 CCR10 expression on these plasmablasts, which has been reported to mark IgA secreting cells 96 (Morteau et al., 2008) and is potentially important for homing of B cells to mucosal tissues 97 including the airways (Kato et al., 2013;van Splunter et al., 2018). The total population of 98 plasmablasts post carriage displayed reduced numbers of CCR10 + cells, in contrast to 6B-specific 99 plasmablasts, indicating differential expansion between specific and non-specific B cell populations 100 (Fig. 4B). This is supported by the observation that increased circulating levels of 6B

116
Nasal CD8 Tissue-resident memory T cells are higher in carriagesubjects

117
The three clusters of CD8 + T cells and the cluster of CD8 dim T cells that were higher in carriage -118 subjects all expressed CD69, a marker of tissue-resident memory (Trm) cells (Fig. 5A). To verify 119 that these CD69 + CD8 + T cells represented Trm cells, we measured the expression of CD103 and CD49a on CD69 + and CD69-cells by flow cytometry from a representative biopsy (Supplementary 121 Fig. 3A). Indeed, 89.1% of nasal CD69 + CD8 + T cells expressed CD103 and CD49a, confirming 122 that these were Trm cells (Fig. 5B) (Kumar et al., 2017). The markers CD5, CD38, HLA-DR, 123 CCR6, CD127, CCR7 and CD11c were expressed in cluster-specific patterns and at varying 124 intensities among the significant clusters. This suggests that clusters of cells with varying degrees 125 of activation and memory types were enriched in carriagesubjects. One cluster expressed only 126 low levels of CD8 (cluster 10 of CD8 dim T cells, 2.0-fold higher, p = 0.016), which could reflect 127 cytotoxic effector memory cells (Trautmann et al., 2003). We then stimulated nasal biopsy cells 128 and PBMC overnight with PMA and ionomycin to assess the functional capacity of nasal CD8 + T 129 cells (Fig. 5C). Among nasal CD8 + T cells, 94.8% produced tumor necrosis factor alpha (TNF) 130 and/or interferon gamma (IFN-) following stimulation, compared to 36% of blood CD8 + T cells,

131
demonstrating that nasal CD8 + T cells are highly functional.

142
MAIT cells from carriage + subjects did not produce increased levels of any cytokine upon 143 stimulation. This was specific to MAIT cells as conventional CD8 + T cells responded by producing 144 small amounts of IFN-and no TNF ( Supplementary Fig. 3D). The baseline responses of MAIT 145 cells in blood upon restimulation showed a positive correlation with numbers of nasal cells at ten days post pneumococcal challenge in CyTOF CD161 + CD8 + T cell cluster 9, which was the human nasopharynx. Of the twenty-five clusters defined in the myeloid lineage, fifteen 152 expressed CD14 ( Supplementary Fig. 4). Of these, only two also expressed CD16. Four CD14 + 153 clusters expressed the macrophage markers CD163 and CD206 and an additional three clusters

159
Characterization of nasal CD4 + memory T cells

160
CD4 + T memory cells, in particular Th17 cells, were previously found to be critical for Spn immunity 161 in mice models of nasal colonisation (Lu et al., 2008;Zhang et al., 2009). Of all cells in the CD4 + T 162 cell lineage, 89.6% expressed the memory marker CD45RO. Of these, 60.3% expressed CD161, a 163 marker that has been proposed to identify Th17 cells (Cosmi et al., 2008;Kleinschek et al., 2009).

164
Another 4.6% of memory cells was defined by expression of high levels of CD25, a marker for 165 regulatory T cells. We defined twenty-three clusters of CD161 -CD4 + T memory cells, twenty-one 166 clusters of CD161 + CD4 + T memory cells and nine clusters of CD25 hi CD4 + T memory cells

186
This study comprehensively characterised immune cells in biopsies collected from the human 187 nasal mucosa. As nasal samples were collected ten days following experimental human 188 pneumococcal challenge, we were able to associate the frequency of specific immune populations

231
In murine models, depletion of CD8 + T cells was protective against Spn lung infection but did not

233
We found here that CD8 + MAIT cell functionality before pneumococcal challenge associated with a 234 resistance to carriage acquisition. MAIT cells are a recently identified T cell subset that is common 235 in humans, consisting of up to 10% of all T cells in the circulation, but that is very rare in mice 236 (Wakao et al., 2017). It is possible that this difference has led to an underappreciation of the CD8+ One limitation of this study is that the number of granulocytes measured was very low due to the 247 overnight resting step following enzymatic digestion. While this resting step allowed for the return 248 of markers that were cleaved by the enzymatic digestion, neutrophils quickly become apoptotic 249 after being removed from the body (Autengruber et al., 2012;Goodyear et al., 2014;Pongracz et 250 al., 1999). Consequently, the characterization of granulocytes reported here is incomplete and we 251 were not able to assess whether specific neutrophil subsets are associated with protection against 252 pneumococcal colonisation. In addition, due to the invasiveness of sample acquisition, sample size 253 was limited and we were not able to characterize nasal biopsies at various time points. Thus, no 254 baseline was available and transient responses early after bacterial inoculation could not be 255 assessed. As subjects received antibiotics prior to biopsy collection, we were unable to associate 256 levels of any of the monocyte or CD4 + T cell clusters with Spn clearance.

257
In conclusion, this study provides both a broad and an in-depth view of the adult human nasal             by CyTOF for that subject to account for number of cells isolated from a given biopsy. This 414 normalization strategy has the advantage that the normalized frequencies of cells in a cluster is not 415 dependent on other clusters, which is a major disadvantage of normalizing against total immune 416 cells. Normalized cluster abundances were then compared between carriageand carriage + 417 subjects for each of the clusters using the Mann-Whitney test, without correcting for multiple 418 testing. Data was analysed and graphs were created using 'pheatmap' and 'ggplot2' packages in R software and circular graph (Fig. 1C)