Special Viewpoints

Natural Killer Cell Signaling Pathways

See allHide authors and affiliations

Science  26 Nov 2004:
Vol. 306, Issue 5701, pp. 1517-1519
DOI: 10.1126/science.1103478


Natural killer (NK) cells are lymphocytes of the innate immune system that are involved in the early defenses against foreign cells, as well as autologous cells undergoing various forms of stress, such as microbial infection or tumor transformation. NK cell activation is controlled by a dynamic balance between complementary and antagonistic pathways that are initiated upon interaction with potential target cells. NK cells express an array of activating cell surface receptors that can trigger cytolytic programs, as well as cytokine or chemokine secretion. Some of these activating cell surface receptors initiate protein tyrosine kinase (PTK)–dependent pathways through noncovalent associations with transmembrane signaling adaptors that harbor intracytoplasmic ITAMs (immunoreceptor tyrosine-based activation motifs). Additional cell surface receptors that are not directly coupled to ITAMs also participate in NK cell activation. These include NKG2D, which is noncovalently associated to the DAP10 transmembrane signaling adaptor, as well as integrins and cytokine receptors. NK cells also express cell surface inhibitory receptors that antagonize activating pathways through protein tyrosine phosphatases (PTPs). These inhibitory cell surface receptors are characterized by intracytoplasmic ITIMs (immunoreceptor tyrosine-based inhibition motifs). The tyrosine-phosphorylation status of several signaling components that are substrates for both PTKs and PTPs is thus key to the propagation of the NK cell effector pathways. Understanding the integration of these multiple signals is central to the understanding and manipulation of NK cell effector signaling pathways.

Natural killer (NK) cells are lymphocytes of the innate immune system that are involved in early defenses against both allogeneic (nonself) cells and autologous cells undergoing various forms of stress, such as infection with viruses, bacteria, or parasites or malignant transformation. NK cells act by means of direct cytotoxic attack on their targets or by producing a large array of cytokines and chemokines. The latter approach contributes to initiation of the antigen-specific immune response, making NK cells an important link between innate and adaptive immunity. Although NK cells do not express classical antigen receptors of the immunoglobulingene family, such as the antibodies produced by B cells or the T cell receptor expressed by T cells, they are equipped with various receptors whose engagement allows them to discriminate between target and nontarget cells. Only a minority of these receptors, such as the natural cytotoxicity receptors (NCRs), are NK cell specific. Most are also found on other hematopoietic cells, particularly T cells. NK cells have been instrumental in revealing a general theme in cell activation, which is that effector cell function results from a dynamic equilibrium between multiple and sometimes opposing pathways that can be simultaneously engaged (1, 2). Integration of these numerous inputs culminates in a graded NK cell response—in other words, cytotoxicity and/or cytokine production. Furthermore, analysis of various mouse mutant models as well as blocking experiments using panels of monoclonal antibodies against human NK cells have revealed that NK cell natural cytotoxicity is a very robust effector function, which can be triggered through apparently similar or redundant signaling pathways. We focus here on NK signaling pathways that are dependent on protein tyrosine kinase activation, as well as on the equilibrium between excitatory and inhibitory signals.

NK cells express several cell surface activating complexes that are formed by noncovalent association between distinct transmembrane ligand-binding and signaling adaptor polypeptides. Most signaling adaptor polypeptides, such as killer cell–activating receptor-associated protein [KARAP, also called DNAX-activating protein of 12 kD (DAP12)], FcRγ, and CD3ζ, contain one or three ITAMs (immunoreceptor tyrosine-based activation motifs; YxxL/Ix6-8YxxL/I, where x is any amino acid), whereas another one, DAP10, contains a YxxM motif, where x is any amino acid (Fig. 1). All of these adaptor molecules contain a transmembrane aspartic acid residue, whose negative charge is required for oligomer formation. We will concentrate on these oligomeric activating complexes, because their modular design appears to represent an evolutionarily conserved architecture common to several signaling pathways (for example, B cell receptor, T cell receptor, Toll-like receptors, and cytokine- as well as growth factor receptor–mediated signaling pathways). This oligomeric architecture allows plasticity in the composition of the receptor, so that it can be readily adapted to mediate disparate responses (3).

Fig. 1.

NK cell effector signaling pathways. Oligomeric activating receptors, which include ITAM-bearing molecules (red) or DAP10 (green), as well as inhibitory MHC class I–specific receptors (blue), are represented. h, human; m, mouse. Asterisks indicate the presence of a transmembrane charged amino acid. KIR-S are activating killer cell Ig-like receptors with a short intracytoplasmic domain and no intrinsic signaling properties, whereas KIR-L are inhibitory receptors with an intracytoplasmic ITIM. In the mouse, but not in humans, two alternative spliced forms of NKG2D coexist: Whereas NKG2D-L associates with DAP10, NKG2D-S can associate with both DAP10 and KARAP/DAP12 (3). The link between DAP10 and SLP-76 is not fully characterized but could occur by way of Grb2 (5). The signaling pathways leading to cytokine secretion appear to be strictly dependent on ITAM-bearing receptors, but their precise delineation remains to be completed. The substrates for SHP-1 and SHP-2 tyrosine phosphatases downstream of ITIM-bearing molecules include Vav1. For additional details, refer to STKE Connections Maps (http://stke.sciencemag.org/cgi/cm/stkecm;CMP_13625 and http://stke.sciencemag.org/cgi/cm/stkecm;CMP_14358).

In human NK cells, these receptors include the low-affinity Fcγ receptor CD16 (FcγRIIIA), which enables recognition of antibody-coated target cells and thus antibody-dependent cell cytotoxicity (ADCC) and cytokine production. CD16 couples to the ITAM-bearing polypeptides CD3ζ and FcRγ, as do the NKp46 and NKp30 NCRs (4). In addition, NK cells express the ITAM-bearing adaptor KARAP/DAP12, which associates with activating killer-cell Ig-like receptors (KIR-S), the NCR NKp44, and the activating receptors for human leukocyte antigen E (HLA-E), namely the CD94-NKG2C and the CD94-NKG2E heterodimers (4). There are many differences between human and mouse NK cell surface receptors. In the mouse, NK cells express CD16 in association with FcRγ, as well as activating Ly49, the mouse NKG2D-S isoform, and CD94 heterodimers that associate with KARAP/DAP12. In a manner reminiscent of T cell receptor– and B cell receptor–mediated signaling pathways, engagement of NK cell ITAM-bearing receptors activates “first line” protein tyrosine kinases (PTKs) of the Src family (such as p56lck and p59fyn), which phosphorylate ITAMs on tandem tyrosine residues, thereby leading to the recruitment and activation of “second line” tandem-SH2 PTKs of the Syk family (such as Syk and ZAP70). Various transmembrane adaptors (such as linker of activated T cells, LAT) and cytosolic adaptors [such as SH2 domain–containing leukocyte phosphoprotein of 76 kD (SLP-76) and SH3-binding protein 2 (3BP2)] then enter into the signaling pathway. These adaptors play an important role in integrating and propagating the signals that lead to cell activation. Indeed, a major function of LAT resides in its capacity to provide multiple tyrosine-phosphorylation docking sites for the SH2 domains of intracellular signaling molecules (such as cytosolic adaptor proteins), thereby targeting them to the inner leaflet of the plasma membrane. In T cells, numerous signaling molecules bind to LAT either directly through SH2 domains [for example, Grb2, Gads, Grap, PLC-γ1 (phospholipase C–γ1), PI3K (phosphatidylinositol 3-kinase), Cbl-b, 3BP2, and Shb] or indirectly, through other adaptor molecules (for example, SLP-76, Sos, Itk, Vav, Nck, SLAP-130, and SKAP55) (5). LAT thus acts as a scaffold for the organization of macromolecular T cell signalosome complexes. In NK cells, LAT is tyrosine phosphorylated following direct contact with target cells and stimulation of CD16. Among the various downstream LAT signaling partners described in T cells, only PLC-γ1 has been reported to directly interact with the phosphorylated form of LAT in activated human NK cells (6), but other possible associations have not been studied in these cells. Despite the apparent wiring conservation between ITAM-dependent signaling pathways in T cells and NK cells, no drastic alterations of NK cell development have been described in mice deficient in the ITAM-bearing polypeptides, p56lck, Syk-family members, LAT, or SLP-76, in contrast to the major impairment in T cell development observed in these mutant mice. These results thus reveal that the pathways involved in NK and T cell development are markedly different. As far as NK cell effector function is concerned, only Syk–/–ZAP70–/– NK cells present notable signaling defects in ADCC as well as in cytokine secretion (7). These results initially suggest that ITAM-bearing polypeptides, p56lck, LAT, or SLP-76 might play redundant roles in the initiation and propagation of NK cell effector programs. However, NK cells use distinct combinations of receptors (and thus distinct proximal signaling pathways) to induce cytotoxicity or produce cytokines when encountering distinct target cells. Therefore, a definitive statement about the involvement of NK signaling components must await a more extensive analysis of NK cell effector function when triggered through a large panel of cell surface receptors.

NK cells also express NKG2D, a cell surface receptor that is noncovalently associated with the DAP10 (DNAX-activating protein of 10 kD) transmembrane adaptor polypeptide. The proximal DAP10 signaling pathway includes Grb-2, PLC-γ2, SLP-76, and PI3K (8), and several features distinguish it from ITAM-dependent signaling pathways. First, DAP10 signaling is independent of Syk-family protein tyrosine kinases (8). Second, it has been shown in the mouse that DAP10 is coupled to the Rac-Cdc42 family exchange factor Vav1, whereas ITAM-bearing polypeptides couple to its relatives Vav2 and Vav3 (9). Third, DAP10 signals appear sufficient to trigger mouse NK cell cytoxicity against certain tumor cell lines but insufficient for inducing cytokine secretion (e.g., interferon-γ). In contrast, ITAM-dependent signals can trigger both cytotoxicity and cytokine secretion (3, 7).

Both ITAM-dependent signaling and DAP-10–dependent signaling seem to converge on a common NK cell cytoxicity pathway. This common pathway involves the activation of Rac, which sequentially leads to activation of the mitogen-activated protein kinase (MAPK) extracellular signal-regulated kinase (ERK) by way of p21-activated kinase (PAK) and mitogen-activated or extracellular signal-regulated protein kinase kinase (MEK) (10), culminating in the redistribution of intracytoplasmic granules and their exocytosis at the interface between the NK cell and its target (Fig. 1). The intracytoplasmic signaling molecule 3BP2 might be a common integrator for both DAP10- and ITAM-dependent cytotoxic pathways. Indeed, 3BP2 act as a positive regulator of NK cell–mediated cytotoxicity and couples to both Vav1 and Vav2 (11). Nonetheless, it remains clear that cytotoxicity and cytokine secretion depend on distinct signaling pathways (8).

A major finding resulting from the study of NK cell signal transduction has been the elucidation of inhibitory pathways that control NK cell effector function (12, 13). NK cells express cell surface receptors, such as the human killer cell inhibitory receptor–L (KIR-L) and mouse Ly49, that transduce inhibitory signals upon engagement with their major histocompatibility complex (MHC) class I ligands. In contrast to oligomeric ITAM-dependent receptors, inhibitory receptors are monomeric and express one or more intracytoplasmic immunoreceptor tyrosine-based inhibition motifs (ITIMs; I/S/T/LxYxxL/V). Upon tyrosine phosphorylation, ITIMs recruit and activate the tandem SH2 protein tyrosine phosphatases SHP-1 and SHP-2 (12, 13). Inhibition by ITIM-bearing receptors of signaling mediated by ITAM-coupled receptors is saturable and proportional to the magnitude of engagement of both types of receptors by their ligands. Inhibitory and activating receptors thus act in concert, and coaggregation between activating and inhibitory receptors is required for inhibition to occur (14). A dynamic equilibrium is set between the “strength” of activating and inhibitory signals and is relayed by the level of tyrosine phosphorylation of critical signaling components (such as Vav1) that can serve as targets for both Syk-family protein tyrosine kinase and SHP-1 or SHP-2 protein phosphatases.

Many other receptors, including adhesion molecules, cytokine receptors, and Toll-like receptors, contribute to the dynamic balance between activating and inhibitory signals. In particular, β2 integrin (LFA-1) signaling is coupled to perforin degranulation, as well as to the formation of NK cell–target cell conjugates through the Wiskott-Aldrich syndrome protein (WASp), a regulator of actin cytoskeleton (15, 16). Interestingly, WASp is also critically involved in both natural cytotoxicity and ADCC, exemplifying the cross talk between the signaling pathways that are initiated by distinct NK cell surface receptors (15). Another example of cross talk between apparently independent receptors has been recently described for the human NKp30, NKp44, and NKp46 NCRs, in which the engagement of a single NCR (e.g., NKp46) results in the activation of the signaling cascades associated with the other NCRs (e.g., NKp30 and NKp44). NCRs thus appear to form a functionally coordinated activating molecular complex, possibly resulting in the amplification of activating signals (17). It remains necessary to find a way to quantify the “strength” of the multiple signals that are initiated in NK cells upon interaction with encountering cells and to understand how these inputs are integrated. Finally, the differential association of KARAP/DAP12 or DAP10 with mouse NKG2D at various stages of mature NK cell activation (3) suggests that during NK cell differentiation, the wiring of NK cell surface molecules can be variable, paving the way for future dissection of signaling pathways during NK cell development.


View Abstract

Stay Connected to Science

Navigate This Article