Recent biochemical studies have suggested that DAP10 may associate with NKRs other than NKG2D, including Ly49D and Ly49H (39
). We now demonstrate that not only do these Ly49H–DAP10 complexes form but that they are able to induce NK cell functions, including proliferation, cytotoxicity, and cytokine production, and, most importantly, that Ly49H–DAP10 complexes are required for optimal control of MCMV infection. By comparing WT, DAP10-deficient, DAP12-deficient, and DAP10–DAP12-deficient mice, we found that DAP10 alone is sufficient to mediate some Ly49H-dependent NK cell functions and that DAP10 and DAP12 together are necessary for optimal Ly49H-mediated NK activity and control of MCMV infection. In the absence of DAP12, DAP10 was sufficient to support the surface expression of Ly49H, Ly49H-dependent NK cell–mediated cytotoxicity, NK cell proliferation, and control of MCMV. However, DAP10 alone was not sufficient for Ly49H-induced IFN-γ production. Even in the presence of DAP12, DAP10 was required for optimal Ly49H expression, IFN-γ production, proliferation, and control of MCMV infection. Thus, we hypothesize that Ly49H signaling via DAP10 augments the DAP12-mediated proliferation and cytokine production of Ly49H+
NK cells, resulting in optimal NK cell control of MCMV infection.
Surface expression of Ly49H on NK cells requires either DAP10 or DAP12, as demonstrated by the total absence of Ly49H on NK cells in DAP10–DAP12 doubly deficient mice. In singly deficient NK cells, the reduction in Ly49H surface level is more pronounced in DAP12-deficient cells than in DAP10-deficient cells. At least two possibilities might explain this difference. First, Ly49H–DAP10 receptor complexes might be less stable than Ly49H–DAP12 receptor complexes. Second, the amount of DAP10 protein in NK cells might be limiting or being preferentially used by other receptors, such as NKG2D. Given the sequence similarity of the transmembrane domains of DAP10 and DAP12, which determine receptor–adaptor pairing, it is anticipated that other DAP12-associated receptors will be found to associate with DAP10. Indeed, we have also observed an association of Ly49D and mouse CD94/NKG2C with DAP10 (unpublished data), which is in agreement with recent findings (39
). It is important to note that our results indicate that DAP10 deficiency not only abrogates the functions of the NKG2D-L receptor but also affects the functions of other activating NKRs that were previously only known to associate with DAP12 in vivo.
During MCMV infection, the Ly49H+
subset of NK cells proliferates extensively. Using DAP12ki mice in which the ITAM of DAP12 was mutated to be nonfunctional, French et al. (36
) showed that the selective BrdU incorporation by Ly49H+
NK cells during MCMV infection required DAP12 signaling. Conversely, we found that DAP12-deficient Ly49H+
NK cells preferentially incorporated BrdU during MCMV infection. It is possible that the DAP12ki homodimer might be acting as a dominant negative, preventing DAP10 from associating with Ly49H and thus accounting for the differences between the DAP12ki and the DAP12-deficient NK cells. By adoptively transferring CFSE-labeled cells, we found that either DAP10 or DAP12 was sufficient to mediate proliferation of Ly49H+
NK cells during MCMV infection. However, maximal proliferation required both signaling subunits because WT NK cells outcompeted DAP10-deficient NK cells when cells of both genotypes were transferred into the same host.
Engaging receptors associated with ITAM adaptors (DAP12, CD3-ζ, or FcϵRIγ), such as the KIR2DS1–DAP12 receptor complex, results in both cytotoxicity and cytokine secretion. In contrast, triggering the DAP10-associated NKG2D receptor leads to cytotoxicity but not efficient cytokine secretion (19
). We previously reported that when we simultaneously cross-linked both KIR2DS2–DAP12 and NKG2D–DAP10 on human NK cells, DAP10 activation augmented DAP12-induced IFN-γ production (22
). We now show that by triggering a single receptor, i.e., Ly49H, both the DAP10 and DAP12 signaling pathways are engaged and lead to optimal IFN-γ production in WT NK cells, whereas either signaling subunit is sufficient to trigger NK cell–mediated cytotoxicity. Signaling downstream of DAP10 and DAP12 engages some of the same pathways, including the ERK1/2 pathway (21
). Compared with WT NK cells, cross-linking Ly49H on DAP10-deficient NK cells lead to only a modest induction of ERK1/2, suggesting that the DAP10 and DAP12 signaling cascades intersect at this molecule in an additive or synergistic manner. ERK1/2 activation is known to be upstream of IFN-γ transcription; thus, the partial defect in IFN-γ production by DAP10-deficient NK cells downstream of Ly49H might be caused by inefficient activation of ERK1/2 (48
). Similarly, CD45-deficient NK cells exhibit minimal phosphorylation of ERK1/2 and IFN-γ production upon triggering of multiple ITAM-associated receptors; however, cytolytic activity induced by these same receptors was maintained in the Cd45−/−
NK cells, suggesting that weak activation of ERK1/2 may be sufficient for NK cell–mediated killing but not IFN-γ production (50
Ly49H is the dominant receptor in NK cell–mediated control of MCMV in B6 mice (34
). In addition to the previously reported DAP12-dependent mechanism of MCMV control (38
), we find a DAP12-independent mechanism of control of MCMV that is dependent on both Ly49H and DAP10. Compared with DAP12-deficient mice, mice lacking both DAP10 and DAP12 demonstrated elevated MCMV titers in both the spleen and the liver similar to that seen in DAP12-deficient mice treated with blocking Ly49H antibody or depleted of NK cells. The most striking defect in MCMV control in DAP10-deficient mice is evident in the salivary glands 1 wk after infection. As Ly49H+
NK cells in DAP10-deficient mice are partially impaired in their proliferative response to MCMV, it is possible that efficient control of MCMV in the salivary glands is dependent on expansion of Ly49H+
NK cells. Furthermore, a DAP10-dependent contribution to IFN-γ production might also be involved in control of viral replication in the salivary gland. The DAP10-dependent control of MCMV in the absence of DAP12 is unlikely to be caused by NKG2D because MCMV encodes at least four proteins that prevent expression of NKG2D ligands on the surface of infected cells (45
), and NKG2D-deficient mice control MCMV infection as efficiently as WT mice (Dan Serna and David Raulet, personal communication). Thus, although DAP12-deficient mice are impaired in their Ly49H-dependent response against MCMV infection (54
), we now show that DAP10 enables some Ly49H function in the absence of DAP12 and is necessary for optimal activity in the presence of DAP12.
Although DAP10 and DAP12 make unequal contributions to Ly49H-mediated function, it is clear that both are required for optimal NK cell responses to MCMV. The signaling events known to be induced by DAP10 receptor triggering are also induced by DAP12 (48
). Thus, DAP12 activation could theoretically supersede DAP10 activation. However, it is possible that engaging both pathways induces a quantitatively more robust signal, as demonstrated by the stronger ERK1/2 activation seen in WT NK cells compared with DAP10-deficient NK cells triggered via Ly49H. Alternatively, DAP10 triggering might engage an unknown signaling pathway that is not shared with DAP12. Regardless, we demonstrate that DAP12-induced DAP10-augmented NK cell responses initiated by Ly49H-mediated recognition of m157 provide for the optimal NK cell response during MCMV infection. Thus, our findings provide a new paradigm for adaptor usage, signaling, and function of activating NKRs, specifically the Ly49H receptor which is necessary for NK cell–mediated resistance to MCMV and potentially other DAP12-associated receptors.