Turning on Natural Killer Cells

NK cells preferentially recognize and kill cells that lack expression of MHC class I 1. While inhibitory receptors expressing immunoreceptor tyrosine-based inhibition motifs prevent NK cells from harming tissues expressing normal levels of classical or nonclassical MHC class I (for a review, see reference 2), what turns on NK cells? Recent reports 3 4, including Smith et al. in this issue 4, are beginning to unfold the receptors and signaling pathways involved in positive activation of NK cells.

Unlike T cells or B cells, which recognize antigen using clonally restricted receptors generated by gene rearrangement, NK cells appear to use a variety of different, nonrearranging receptors to initiate cytolytic activity and cytokine production (for a review, see reference 5). Many of these activating receptors are not restricted to NK cells; most are also present on T cells, and these serve as costimulatory or adhesion receptors in both cell types. Like the ability of antibodies against CD3 or the T cell antigen receptor to trigger killing or cytokine production by cytotoxic T cells, a productive strategy to search for activating NK cell receptors has involved screening for mAbs that initiate NK cell killing of Fc receptor–bearing tumor cell lines. Using this approach, several membrane receptors have been implicated in NK cell activation, including CD2 6, CD16 7, Ly6 8, CD44 9 10, CD69 11, NKR-P1 8 12, 2B4 13 14 15, DNAM-1 16, NKG2D 17, CD94/NKG2C 18, NKp44 19, NKp46 20, and NKp30 3. The ligands for some of these receptors are known (e.g., CD2 for CD58 21, 2B4 for CD48 22, CD16 for IgG [23, 24], NKG2D for MICA 17, and CD94/NKG2C for HLA-E 25). Expression of the ligands on target cells can initiate NK cell lysis or cytokine production, provided the NK cells are not turned off by inhibitory NK receptors recognizing class I molecules on the targets.

Although several different activating NK cell receptors have been identified, many share a common adaptor molecule or signaling pathway. Recently, Pende et al. 3 reported that NKp30 lacks known signaling motifs in its cytoplasmic domain, but noncovalently associates with CD3ζ to provide for cellular activation. CD16 also associates with CD3ζ 26, or the structurally similar FcεRIγ chain 27 28, and this receptor complex mediates antibody-dependent cellular cytotoxicity. NKp46 29 and mouse NKR-P1 30 are other NK cell receptors linked to CD3ζ or FcεRIγ, respectively. Although not physically associated, CD3ζ has been implicated in CD2-mediated activation of NK cells 31. NK cells express both CD3ζ and FcεRIγ, and these adaptor molecules form disulfide-bonded homodimers, as well as heterodimers with each other 28. CD3ζ and FcεRIγ have immunoreceptor tyrosine-based activation motifs (ITAMs) in their cytoplasmic domains that upon phosphorylation caused by receptor ligation recruit the cytoplasmic tyrosine kinases Syk and ZAP70 (both of which are expressed in NK cells; for a review, see reference 32). Ligands for NKp30, NKp46, and NKR-P1 have not been identified; however, there are suggestions that these receptors are involved in NK cell recognition of certain tumors 3 19 33.

Other activating NK cell receptors signal by association with another ITAM-containing adaptor protein, DAP12. Like CD3ζ, DAP12 has a very short extracellular domain (of ∼14 amino acids) and is expressed as a disulfide-bonded homodimer on the surface of NK cells, myeloid cells, and a subset of T cells 34 35. Whereas CD3ζ has three ITAMs, DAP12 and FcεRIγ each have a single ITAM. DAP12 associates with the activating isoforms of the human killer cell Ig-like receptors (KIRs) 34, mouse Ly49D and Ly49H 36 37, the CD94/NKG2C receptor 38, and NKp44 39. In this issue, Smith et al. 4 describe an mAb against Ly49H and demonstrate that cross-linking this receptor results in the activation of a subset of mouse NK cells. Known ligands of these DAP12-associated receptors include HLA-C for KIR, HLA-E for CD94/NKG2C, and H-2D^d for Ly49D (to date, the ligand for Ly49H has not been identified). As with CD3ζ or FcεRIγ, ligation of these DAP12-associated receptors results in tyrosine phosphorylation of the ITAM of DAP12 and recruitment of Syk or ZAP70 34 40, thereby initiating cytolytic activity and cytokine production. DAP12, CD3ζ, and FcεRIγ all possess an acidic amino acid (aspartic acid) in their transmembrane domains that provides for their association with charged amino acids in the transmembrane domains of the ligand-binding receptor subunits. While CD3ζ and FcεRIγ may be interchangeable in their association with certain partners, DAP12 interacts with completely distinct glycoproteins, indicating exquisite specificity in these multisubunit receptor complexes.

NK cell development and cytolytic function are intact in mice or humans with disrupted ZAP70 41 42 or Syk 43 genes, suggesting the existence of receptors using other signaling pathways. Activating NK cell receptors not linked to ITAM-containing adaptor proteins have been identified. The activating NK cell receptor 2B4 13 14 44 uses a signaling pathway involving the cytoplasmic tyrosine phosphatase SHP-2 and the Src homology 2 (SH2)-containing intracellular adaptor protein SAP 15. SAP, identified by its association with the T cell costimulatory molecule SLAM 45, likely serves a common role for several receptors in NK and T cells. NKG2D, an activating receptor on T and NK cells for the nonclassical MICA and MICB antigens 17, uses the DAP10 adaptor protein for signal transduction 46. DAP10 is a membrane-anchored, disulfide-linked homodimer with an acidic amino acid (aspartic acid) in the transmembrane that permits association with a basic amino acid (arginine) in the transmembrane of NKG2D 46. Phosphorylation of the tyrosine residue in the YxxM motif in the DAP10 cytoplasmic domain recruits the p85 subunit of phosphatidylinositol 3-kinase (PI3-kinase) 46. Inhibitors of PI3-kinase prevent NKG2D-mediated NK cell cytotoxicity and cytokine production, implicating this signaling pathway in DAP10-induced activation. Expression of DAP10 in myeloid cells, which lack NKG2D, predicts the existence of other receptors that will use DAP10 for signal transduction 46.

Although many receptors are now being implicated in NK cell activation, it is uncertain whether any individual receptor alone will be sufficient to initiate effector function. Rather, additive or synergistic interactions between multiple receptors may be necessary to stimulate cytotoxicity and cytokine production. Experimental findings support the latter. In the recent studies of Pende et al. 3, the mAb against NKp30 alone only marginally blocked NK cell killing of several tumors. However, more dramatic effects were observed when blocking antibodies against several NK cell receptors and adhesion molecules were used in combination. Thus, the emerging concept is that NK cell function is determined by the integration of signals from several activating receptors that are regulated by the inhibitory receptors for MHC class I, based on the density and array of ligands and class I molecules expressed by the antigen-presenting cell. Infection and inflammation are often accompanied by the release of local cytokines (e.g., IL-15, TNF-α, type I IFNs, chemokines). These cytokines may serve not only to activate NK cells, but also to upregulate ligands (e.g., the intercellular adhesion molecules [ICAMs], CD48, CD58) for adhesion or costimulatory receptors that are present on NK cells. In this environment, rich in stimulatory cytokines and ligands for activating receptors, NK cells may provide a critical bridge between the innate and adaptive immune systems.