MHC class I molecules co‐stimulate NK1.1 signaling and enhance Ca2+ flux in murine NK cells

Murine NK cells express MHC class I specific inhibitory receptors of the Ly49 family [1]. MHC class I molecules act as gatekeepers for NK cell tolerance and only in the absence of MHC class I on surrounding cells, tolerance is broken and NK cell react against “missing self” [2]. MHC class I molecules are not only ligands for inhibitory NK cell receptors, but may also inhibit NK cell function themselves [3–5]. MHC class I molecules lack signaling motifs in their intracellular domains [6], and their direct inhibitory roles have been suggested to depend on MHC class I associated proteins at the NK cell surface [5]. We used a recently developed Ca2+ flux assay in mouse NK cells to ask if crosslinking of MHC class I would influence signaling downstream of activating receptors [7]. When the activating receptor NK1.1 and MHC class I molecules were co-crosslinked, we found an additive effect on Ca2+ flux, characterized by an enhanced peak and a sustained slope of the calcium release curve (Fig. 1A–C; Supporting

Murine NK cells express MHC class I specific inhibitory receptors of the Ly49 family [1]. MHC class I molecules act as gatekeepers for NK cell tolerance and only in the absence of MHC class I on surrounding cells, tolerance is broken and NK cell react against "missing self" [2]. MHC class I molecules are not only ligands for inhibitory NK cell receptors, but may also inhibit NK cell function themselves [3][4][5]. MHC class I molecules lack signaling motifs in their intracellular domains [6], and their direct inhibitory roles have been suggested to depend on MHC class I associated proteins at the NK cell surface [5].
We used a recently developed Ca 2+ flux assay in mouse NK cells to ask if crosslinking of MHC class I would influence signaling downstream of activating receptors [7]. When the activating receptor NK1.1 and MHC class I molecules were co-crosslinked, we found an additive effect on Ca 2+ flux, characterized by an enhanced peak and a sustained slope of the calcium release curve ( Fig. 1A Fig. S1). Cross-linking MHC class I molecules alone also triggered a Ca 2+ flux response, but with a different kinetics characterized by late onset and a slower rise (Fig. 1A-C; Supporting information Fig. S1). In the absence of extracellular Ca 2+ , the magnitude of Ca 2+ flux was decreased but co-stimulation persisted, showing that the effect of MHC class I cocross-linking did not depend exclusively on extracellular Ca 2+ (Fig. 1D). The MHC class I molecule H2K b bind the inhibitory Ly49C receptor in cis [8]. To allow analysis of Ly49C+ NK cells independently, we included an antibody in the panel that stained Ly49C+ NK cells but did not trigger NK cell inhibition. The effects of crosslinking H2K b was similar on Ly49C+ and Ly49C-NK cells (Fig. 1E), showing that the simultaneous expression of a Ly49 receptor with the potential to bind H2K b in cis was not a prerequisite for the costimulatory effect of H2K b .
Because of the unusual calcium flux profile following H2K b cross-linking, we asked if this response was targeted by Ly49 receptor-mediated inhibition. To test this, we included a fluorochrome-conjugated antibody against the inhibitory receptor Ly49G2, which in this case transmits inhibitory signals following cross-linking [7]. When Ly49G2 was co-cross-linked with H2K b , the Ca 2+ signal was aborted compared to Ly49G2-NK cells in the same sample (Fig. 1F). Similarly, the additive effect between NK1.1 and H2K b was reduced in the presence of Ly49G2 crosslinking (Fig. 1F). This experiment revealed that the Ca 2+ flux response downstream of H2K b triggering was indeed a real signaling response, since it could be actively dampened by cross-linking of an inhibitory Ly49 receptor. To exclude that that the additive effect of the anti-H2K b antibody was dependent on FcγRIIIa (CD16), we added two blocking antibodies against FcγRIIIa (clone 2.4G2 and clone 93) to our assay. This did not prevent the additive effect of anti-H2K b on NK1.1-mediated Ca 2+ flux (Fig. 1G), suggesting that the enhancing effect of anti-H2K b was independent from Fc receptor binding.
Poly-I:C, a TLR 3 agonist, induced NK cell activation in vivo as shown by upregulation of the activation marker CD69 but also of NK1.1, NKp46, and H2K b (Fig. 2B). Poly-I:C enhanced the response to all stimuli in our system, including both anti-H2K b alone and of anti-H2K b + anti-NK1.1 (Fig. 2B), suggesting that cross-linking of H2K b augments NK1.1 signaling by a general enhancement of signaling capacity additive to TLR priming. These data also demonstrate that Poly I:C stimulation operates, at least partly, at the level of early Ca 2+ signaling. Finally, to test if the augmented Ca 2+ flux response after H2K b and NK1.1 co-cross-linking affected IFN-γ secretion and degranulation, we switched to a stimulation protocol based on coating plates with antibodies, adding NK cells and allowing them to be stimulated by the antibodies over 4 h. NK1.1 stimulation resulted in both IFN-γ and degranulation, but adding an antibody to H2K b to the plate did not affect these responses ( Fig. 2C and D).
In contrast to the dampening effect of MHC class I cross-linking seen in other systems [3][4][5], we found that stimulation of MHC class I molecules in primary mouse NK cells triggered a Ca 2+ flux response in itself as well as provided an additive effect on immunoreceptor tyrosinebased activating (ITAM) motif mediated Ca 2+ flux. The slow onset response using is subject to inhibition from inhibitory receptors. Dot plot shows gating strategy to identify Ly49G2-and Ly49G2+ NK cells, middle two graphs show one representative Ca 2+ flux experiment on Ly49G2-and Ly49G2+ NK cells, and the bar graph shows the AUC from a summary of three experiments (n = 2 mice/experiment). One-way ANOVA with multiple comparisons with Tukey's correction was used. * * p < 0.01, * * * p < 0.001, * * * * p < 0.0001. (G) MHC class I induced Ca 2+ flux is not affected by FcR blockade. Dot plot shows gating strategy, the two following graphs show binding of 2.4G2 and Clone 93 on the four subsets defined in the dot plot, the fourth graph shows Ca 2+ flux from one experiment, and the bar graph is a summary of three experiments (n = 1 mouse/experiment). Statistics used one-way ANOVA with multiple comparisons with Tukey's correction was used for statistical calculation. ns = not significant.
MHC class I triggering alone was distinct from the Ca 2+ flux response reported after MHC I cross-linking on T cells and B cells, in which a rapid, sharp onset of Ca 2+ flux was followed by a slow decline [9]. The response in our system bears more similarities to triggering of some NK cell receptors in human, for example, 2B4 [10], which is interesting in light of a previous study in humans showing that MHC class I molecules associate with 2B4 at the NK cell surface and that MHC class I cross-linking dampens 2B4mediated killing responses [5]. It remains to be tested if MHC class I molecules bind 2B4 also in mouse NK cells, and if so, to which extent our results would be dependent on 2B4 signaling. Intriguingly, the enhanced Ca 2+ flux did not translate into increased IFN-γ secretion or degranulation, suggesting that the enhanced Ca 2+ flux induced by NK1.1+H2K b may be qualitatively different compared to the response triggered by ITAM signaling alone. Ca 2+ flux is crucial for effector lymphocyte function but sig-naling via activating receptors is complex and may not necessarily imply effector cell function. Strong immune cell activation might induce an exhaustion phenotype rather than a powerful effector cell response. Perhaps the costimulatory role of H2K b in our model is geared towards enhanced proliferation or secretion of factors that we did not measure. Another more trivial, yet possible explanation, could be that Ca 2+ flux was measured in solution using two antibodies, while IFN-γ and CD107a were measured after 4 h of stimulation using plate-bound primary antibodies alone. Two parameters differ in these settings: the type and time of stimulation, both of which might affect the experimental result and determine downstream effects resulting from the additive Ca 2+ flux.
Further work needs to be performed to identify the signaling pathways involved in MHC class I signaling and which cisbinding NK cell surface proteins that might be responsible for transmitting this signaling response. The physiological relevance of MHC class I cross-linking also needs to be identified, but it is not unlikely that any protein with signaling capacity that bind to MHC class I, in trans or in cis, could be involved. Identifying the functional consequences of MHC class I induced Ca 2+ flux is an important next step, requiring novel experimental models in which MHC class I triggering can also be induced by more physiological ligands in cellular systems. SG performed experiments, acquired and analyzed data, and wrote the manuscript. PH analyzed data, wrote the manuscript, and provided funding for the study.

Conflict of interest:
The authors declare no financial or commercial conflict of interest.

Data availability statement:
The data that support the findings of this study are available from the corresponding author upon reasonable request.