Hypersensitivity of MAIT cells to Leukocidin ED indicates a Staphylococcus aureus immune evasion mechanism targeting the innate effector cell response

Mucosa-associated invariant T (MAIT) cells recognize bacterial riboflavin metabolite antigens presented by MR1 and play an important role in antimicrobial immune defense. Staphylococcus aureus is a pathobiont expressing a range of virulence factors including the secreted toxin Leukocidin ED (LukED), which binds to certain chemokine receptors and causes cell death by osmolysis. Here, we investigated the effect of LukED on subsets of human T cells and NK cells that are involved in the early innate response to infection. MAIT cells were strikingly hypersensitive to LukED-mediated lysis and rapidly lost from the peripheral blood T cell pool upon exposure to the toxin, leaving a T cell population devoid of MAIT cells. The cytolytic effect of LukED on MAIT cells was rapid, occurred at lower LukED concentration compared to effects on the overall T cell pool, and coincided with extraordinarily high and uniform expression of CCR5. Furthermore, loss of MAIT cells was efficiently inhibited by the CCR5 inhibitor Maraviroc. Interestingly, pre-activation of MAIT cells with IL-12 and IL-18 also partially rescued these cells from LukED toxicity. Among NK cells, LukED targeted the more mature and cytotoxic CD57+ NK cell subset in a CXCR1-dependent manner. Overall, these results indicate that LukED efficiently eliminates cells of the human immune system that have the capacity to respond rapidly to S. aureus in an innate fashion, and that MAIT cells are exceptionally vulnerable to this toxin. Thus, the findings support a model where LukED functions as a S. aureus immune evasion mechanism to avoid recognition by the rapid cell-mediated responses mediated by MAIT cells and NK cells.


Introduction
Mucosa-Associated Invariant T (MAIT) cells belong to the broad and diverse group of unconventional non-MHC restricted T cells [1,2]. MAIT cells recognize non-peptide microbial antigens presented in complex with the MHC-class Ib-related protein (MR1) [3,4].
Antigens recognized by MAIT cells include intermediates of the vitamin B2 (riboflavin) synthesis pathway expressed by many microbes, and this allows immune surveillance of a broad range of microorganisms in an MR1-restricted fashion [5,6]. The presence of MAIT cells in high numbers across blood and mucosal tissues poise them to rapid effector responses in response to microbial infection [7]. Upon recognition of antigen presented by MR1 they secrete cytokines including IFNγ, TNF, IL-17A, and IL-22 [7,8], and can participate in tissue repair and wound healing [9][10][11][12]. Furthermore, MAIT cells can kill bacterially infected cells via the release of cytotoxic effector molecules such as Granzyme A, Granzyme B, and Granulysin [13][14][15][16], and have direct antimicrobial properties against both cell-associated and free-living bacteria [17]. Their role in defense against bacterial infection was demonstrated in several mouse models of infection [18][19][20]. In humans, MAIT cells expand in response to Salmonella enterica subsp. enterica serovar Paratyphi A challenge [21], and migrate to lung tissue during tuberculosis [22,23]. NK cells are the prototypical innate effector cells and mediate rapid immune responses against microbes and tumor cells [24,25]. Differentiation of CD56 bright NK cells to mature cytolytic NK cells is characterized by lowered CD56 expression and the gain of CD16, CD57 and KIR expression [26,27]. During bacterial infection, NK cells can be indirectly activated by cytokines or through interaction with other cell types, and can also directly recognize bacteria through Toll-like receptor (TLR) sensing [28]. Recently, the activating receptor KIR2DS4 was found to recognize a bacterial HLA-C*05:01-presented epitope derived from a protein conserved in many bacterial species including S. aureus [29].
S. aureus is a bacterial pathobiont that colonizes 30% of the human population through nasal and skin carriage [30]. The shift from commensal microbe to pathogen requires the expression of virulence factors as well as barrier breach [31,32]. S. aureus affects local tissue and can spread systemically to cause life-threatening diseases such as pneumonia, endocarditis, and sepsis. Virulence factors that are critical for pathogenesis include superantigens, cytolytic peptides and pore-forming toxins [33]. One of the pore-forming toxins is the leucocidin ED (LukED), which is expressed by a majority of S. aureus isolates [34]. This bicomponent toxin is composed of two water-soluble monomers and acts in two steps: LukE first binds to target proteins on the cell membrane and recruits the LukD subunit.
The complex then oligomerizes, inserts in the cell membrane as a β-barrel pore, leading to disruption of the cellular osmotic balance and cell death [34]. LukED binds to the chemokine receptors CCR5, expressed on macrophages, dendritic cells (DC) and T cells [35]. It also binds to CXCR1 and CXCR2 expressed on NK cells and neutrophils [36], as well as to DARC expressed on erythrocytes and endothelial cells [37,38]. LukED contributes to S. aureus pathogenesis in vivo [35][36][37]39], and the binding to DARC expressed on epithelial and endothelial cells leads to vascular leakage and organ failure [38]. S. aureus ΔLukED mutants are less invasive with reduced bacterial burden and mortality [36,37,39].
The role of MAIT cells in S. aureus immunopathogenesis remains unclear. MAIT cells are activated by S. aureus stimulation in vitro and produce IFNγ [22,23], and their frequencies are increased in tonsils and blood of individuals with S. aureus tonsillitis [40].
MAIT cell are also significant contributors to the massive cytokine release, "cytokine-storm", in response to Staphylococcal enterotoxin B (SEB), but this response renders them anergic and unable to respond to further MR1-dependent stimulation [41,42]. In the present study, we investigate the effect of LukED on MAIT cells in comparison with other subsets in the peripheral blood T cell pool and dissect the effect of LukED on MAIT cell recognition of S. aureus. Furthermore, we investigate how LukED affects the NK cell compartment.
Altogether, the findings indicate that LukED secretion by S. aureus constitutes an immune evasion mechanism to interfere with responses mediated by human innate effector cells.

MAIT cells are hypersensitive to LukED cytotoxicity
To investigate the effect of the LukED toxin on human peripheral blood T cells, we incubated peripheral blood mononuclear cells (PBMC) in the presence of the recombinant toxins LukE and LukD and assessed population changes using flow cytometry. The total lymphocyte population was first analyzed using Uniform Manifold Approximation and Projection (UMAP) analysis [43] of healthy donor PBMC exposed to LukED. Cell populations were identified by the projection of defining markers on the UMAP topography ( Figure 1A).
Projection of LukED-treated versus untreated conditions revealed differences between the two settings and the loss of some cell populations ( Figure 1B). Strikingly, the UMAP area defined by the 5-OP-RU-hMR1 tetramer was almost completely absent after exposure to LukED, suggesting that exposure to the toxin depletes MAIT cells. The profound reduction of MAIT cells (defined as CD3+, CD161 high , 5-OP-RU-hMR1+) was confirmed both as percentage and absolute count ( Figure 1C). In contrast, non-MAIT T cell populations were only slightly affected by the toxin with a decline of 25% ( Figure 1D), versus 97% for MAIT cells ( Figure   1E). The single components of the toxin alone, LukE and LukD, did not affect the T cell compartment in a detectable manner (Suppl. Figure 1A and 1B). Since LukED was previously shown to lyse T cells in a CCR5-dependent fashion (30), we zoomed in on the CCR5+ non-MAIT T cell subset. We noticed a decrease of this population ( Figure 1F), which was 80% depleted upon LukED exposure ( Figure 1G) compared to 97% for MAIT cells. This is in line with a strong CCR5-dependency, as MAIT cells have homogenous expression of CCR5 (Suppl. Figure 1C) and display higher level of expression compared to CCR5+ non-MAIT T cells ( Figure 1H and Suppl. Figure 1D).
Within the MAIT cell compartment, CD8+ and CD8-CD4-double negative (DN) MAIT cells appeared to be slightly more sensitive to LukED than the minor CD4+ MAIT cell subpopulation ( Figure 1I), probably due to their relatively higher expression of CCR5 (Suppl. Figure 1E). No significant difference was noted in CD8+, CD4+ and DN non-MAIT T cells upon LukED exposure (Suppl. Figure 1F). Next, we explored the composition of the non-MAIT T cell population based on the expression of differentiation markers CCR7 and CD45RA. Terminally differentiated effector memory RA+ (TEMRA) and effector memory T (TEM) cell populations decreased in frequency upon LukED exposure ( Figure 1J and 1K) contrary to central memory (TCM) and naïve T cells (Suppl. Figure 1G). This coincided with higher levels of CXCR1 and CCR5 expression by TEMRA cells and TEM cells, respectively (Suppl. Figure 1H). Projection of the LukED-treated condition on the UMAP also revealed a dramatic loss of cells double expressing CD8 and CD56 ( Figure 1B). The decrease of CD8+CD56+ non-MAIT T cells was confirmed by manual gating ( Figure 1L), and coincided with co-expression of CCR5 and CXCR1 on this subset (Suppl. Figure 1K). Overall, the loss of TEMRA, TEM and CD8+CD56+ non-MAIT T cells was not as severe as the depletion of MAIT cells (Suppl. Figure 1I and 1L).
To investigate if the patterns obtained with the recombinant toxin occur with live bacteria, we incubated PBMC with culture supernatants of three S. aureus clinical isolates previously characterized regarding their toxin gene profile [44]. The three strains studied were selected based on varying presence of the lukED genes and low expression of the bacterial αtoxin, another pore-forming toxin highly effective in killing cells by binding the surface protein ADAM-10. Interestingly, PBMC cultures exposed to the supernatants of lukED positive strains lost a larger fraction of their MAIT cells, as compared to cultures exposed to lukED negative strains ( Figure 1M). CCR5+ non-MAIT T cells followed a similar pattern (Suppl. Figure 1M). Taken together, these findings indicate that LukED depletes T cell subsets with effector and effector memory characteristics, with MAIT cells being the major targeted population.

Maraviroc rescues MAIT cells from LukED toxicity
To investigate if LukED toxicity against MAIT cells is CCR5-dependent, we evaluated the ability of Maraviroc (MVC), a CCR5 antagonist used in HIV therapy, to protect MAIT cells from the recombinant toxin. Addition of MVC to the assay largely rescued the MAIT cell population from LukED toxicity ( Figure 2A). While this protective effect was incomplete, it was observed for all the main CD8, CD4 and DN MAIT cell subsets (Suppl. Figure 2A).
MVC did not have a detectable effect on the overall non-MAIT T cells bulk population ( Figure 2B), but seemed to partially rescue the CCR5+ subset of non-MAIT T cells ( Figure   2C). Similarly, there was a trend towards partial MVC rescue of TEMRA, TEM and CD8+CD56+ non-MAIT T cells but this effect did not reach statistical significance (Suppl. Figure 2B and 2C). As some of those subsets express CXCR1, LukED binding to this receptor would not be expected to be inhibited by MVC. Altogether, these results indicate that MVC inhibits the toxicity of LukED against MAIT cells in vitro.  Figure 2E). Altogether these findings showed that the lethal effect of LukED on MAIT cells is rapid, dose-dependent, and occurs at lower doses compared to conventional T cells.

MAIT cell activation with IL-12 and IL-18 partly prevents LukED-mediated loss
MAIT cells can be activated in response to innate cytokines produced in the setting of myeloid cell activation, such as IL-12 and IL-18. Surprisingly, MAIT cells activated by IL-12 and IL-18 for 20 h, and then exposed to LukED for the last two hours of incubation, were partly preserved as compared to the non-activated control MAIT cells ( Figure 3A and 3B).
This effect coincided with a decrease in CCR5 expression upon IL-12 and IL-18 stimulation ( Figure 3C). It is thus possible that activation by innate cytokines in the inflammatory milieu may render MAIT cells partly resistant to LukED.

Sub-lethal doses of LukED have no major impact on MAIT cell responsiveness
Since LukED is a pore-forming toxin it is possible that it may affect signaling in MAIT cells exposed to concentrations insufficient for direct killing of the cells. We therefore next sought to determine if exposure to sub-lethal doses of LukED affects TCR-dependent activation of MAIT cells. To test this, MAIT cells were pre-treated with sub-lethal doses of LukED before stimulation with the mildly fixed LukED-positive S. aureus strain 134, or the LukEDnegative strain 289. No detectable difference was observed in MAIT cell IFNγ, TNF, IL-17A and GzB production between the sub-lethal dose LukED-treated and -untreated conditions ( Figure 3D and 3E). These results indicate that sub-lethal levels of LukED have little observable impact on MAIT cell function and ability to respond to bacteria.

LukED targets the mature CD57+ NK cell subset
In the UMAP topography ( Figure 1A and 1B), apart from MAIT cells another area severely affected by LukED was the one dominated by CD56+CD3-cells corresponding to the NK cell population. Indeed, LukED severely reduced both CD56 bright and CD56 dim NK cells among PBMC both as percentage and as absolute count ( Figure 4A), and the CD56 dim subset was particularly severely impaired ( Figure 4B). LukE and LukD alone did not affect NK cells (Suppl. Figure 3A). In line with the differential activity of LukED on the two NK cell subsets, CD56 dim NK cells had higher expression of CXCR1 and CXCR2 than CD56 bright NK cells ( Figure 4C and 4D). In general, NK cells expressed low levels of CCR5 ( Figure 4D and Suppl. Figure 3B). Dissection of the NK cell compartment revealed that LukED targeted the most mature NK cells known to have cytolytic effector properties. In particular, the toxin diminished CD56 dim NK cells expressing CD16, CD57, KIRD2L1, perforin or co-expressing CD57 and KIRD2L1 ( Figure 4E and 4F). LukED exposure did not change the representation of NK cell populations expressing granulysin or CD27 (Suppl. Figure 3C). Taken together, these observations indicate that LukED preferentially targets mature cytolytic NK cells.

Discussion
The soluble virulence factors secreted by S. aureus play a major role in the pathogenesis of this infection. However, it is still not fully understood how these virulence factors affect the human immune system. MAIT cells are an attractive target for development of vaccines and therapeutics due to their specificity towards conserved bacterial antigen, their secretion of IFNγ, IL-17, and IL-22, as well as their presence in skin and blood, the main sites of S. aureus infection and pathogenesis [45]. In this study, we explored the interplay between the LukED toxin and the peripheral blood immune system, and found that MAIT cells are hypersensitive to LukED, due to their very high surface expression of CCR5. LukED depleted MAIT cells more efficiently, with lower IC50, than other T cells subsets, such as conventional TEM and TEMRA cells. The mature effector CD56 dim NK cells were also depleted to a large extent, while CD56 bright NK cells were less affected.

MAIT cells and NK cells are important effector cells involved in the host response
against bacterial infection. MAIT cells have cytotoxic properties and can directly kill bacteria [14,15,17]. NK cells are activated by S. aureus [46] and contribute to immune defense against this infection in vivo [47]. In the absence of Th1 and Th17 immunity the susceptibility to S. aureus infection is increased [45]. Thus, the findings of the present study indicate that LukED targets cells that have the capacity to mount a rapid effector response against S. The CCR5 antagonist MVC inhibits the effect of LukED on T cells. As MVC is approved for use in humans, it is possible that this inhibitor could be used to block MAIT cell depletion and restore immune control in patients with invasive S. aureus infection. It is interesting to note that several new treatments are in development against S. aureus infection including monoclonal antibodies aimed to neutralize S. aureus toxins [49]. In experimental animal models of infection in mice [50] and rabbits [51], the combination of monoclonal antibodies, including one targeting LukED, shows enhanced efficacy over single antibody administration with decreased bacterial burden [50] or increased survival [51]. In addition, new neutralizing agents called centyrins are able to block the binding of the five bicomponent leukocidins to their receptors and protect in vivo against S. aureus infection [52].   Statistics. Statistical analyses were performed using Prism software v.7 (GraphPad). Data sets were first assessed for normality of the data distribution. Statistically significant differences between samples were determined as appropriate using the unpaired t-test or Mann-Whitney's test for unpaired samples, and the paired t-test or Wilcoxon's signed-rank test for paired samples. Correlations were assessed using the Spearman's rank correlation. Two-sided pvalues < 0.05 were considered significant.