|Home | About | Journals | Submit | Contact Us | Français|
Natural killer (NK) cells play a vital role in the defense against viral infections and tumor development. NK cell function is primarily regulated by the sum of signals from a broad array of activation and inhibitory receptors. Key to generating the input level of either activating or inhibitory signals is the maintenance of receptor expression levels on the cell surface. Although the mechanisms of endocytosis and trafficking for some cell surface receptors, such as transferrin receptor, and certain immune receptors, are very well known, that is not the situation for receptors expressed by NK cells. Recent studies have uncovered that endocytosis and trafficking routes characteristic for specific activation and inhibitory receptors can regulate the functional responses of NK cells. In this review, we summarize the current knowledge of receptor endocytosis and trafficking, and integrate this with our current understanding of NK cell receptor trafficking.
There is a large amount of literature regarding the endocytosis and trafficking of membrane receptors. Some of these studies have focused on immune receptors. With the exception of major histocompatibily complex (MHC) molecules, most immune receptors that have been analyzed are “activation” receptors and their downregulation after ligation often serves to regulate the immune response. For example, it has been shown that the T cell receptor (TCR) is constitutively recycled in resting T cells and that the ligation of TCR enhances endocytosis and degradation, which serves to control excessive inflammation (1–3). B lymphocytes function as efficient antigen presenting cells mainly through presentation of antigens internalized via the B cell receptor (BCR) (4). The ligation of cell surface receptors usually initiates a signaling process. The signaling process can continue after internalization as the receptors traffic through the endocytic compartments (5–7). The trafficking and endocytosis of some immune receptors, such as Toll–like receptors (TLRs) (8) Fc Receptors (9–11) and MHC molecules (12–14), have also received significant attention; however, only a few studies have focused on receptor trafficking in natural killer (NK) cells.
Here, we briefly summarize the various mechanisms known to be involved in receptor mediated endocytosis. Endocytosis is the process by which cells uptake molecules from the extracellular environment. This process is historically divided into two broad types: phagocytosis, which involves ingestion of large particles, such as bacteria and dead cells, and pinocytosis, a process associated with fluid phase uptake. The pinocytic uptake of solutes can be receptor-mediated (e.g. clathrin mediated) or non-specific in nature such as in macro or micropinocytosis (15).
Many cell surface proteins that are internalized into endosomal vesicles by non-phagocytic means utilize the well-characterized clathrin-mediated endocytic (CME) pathway (15–17). Others utilize less defined routes that are mostly categorized as caveolin-dependent, lipid raft-dependent, or macropinocytosis (18–20). One should realize, however, that it is not always easy to draw clear distinctions among these pathways (21) and that some receptors, depending on the circumstances, are not necessarily restricted to a single endocytic pathway (22–24). CME requires the formation of clathrin triskelion and concerted action of adaptor proteins (such as AP2, EPS-15 etc) to form clathrin coated buds at the cell surface (25–27). The clathrin coated vesicles pinch off the cell membrane through the action of dynamin (17). The internalization mechanisms of non-clathrin-dependent pathways are less well characterized (18, 19, 28). A major clathrin-independent pathway is thought to be mediated by caveolae, but the role of caveolae in immune cells remains controversial. Even though the characteristic caveolae-like structures are absent in lymphocytes (29, 30), the expression of caveolin-1, the essential component of caveolae has been reported in immune cells (31–34). Some receptors appear to rely at least partially on a little understood mechanism that is lipid raft-mediated. This includes such important immune system receptors as TCR (1–3), BCR (35), the high affinity receptor for IgE (FcεRI) (36) and the interleukin (IL)-2 receptor (37, 38). Receptor endocytosis is currently considered as lipid raft- dependent if internalized vesicles co-localize with proteins known to be lipid raft associated (for example, flotillins and GM1 gangliosides (39)), are endocytosed by a mechanism independent of clathrin-associated machinery (40) and the internalized proteins reside in the detergent resistant membrane (DRM) fraction (36).
As stated previously, some receptors can utilize several alternative pathways for internalization and trafficking depending on the circumstances (22–24). For example, BCR is internalized via both CME and clathrin-independent lipid raft associated pathways (35, 41). FcεRI has been reported to undergo endocytosis by both CME and clathrin-independent pathways (36, 42). CME is the major pathway of epidermal growth factor receptor (EGFR) internalization; however, it can also be internalized via a clathrin-independent mechanism with different outcomes. EGFR internalized via CME is not targeted for degradation, but instead recycled back to the cell surface. On the other hand, EGFR internalized by clathrin-independent route is preferentially targeted for degradation (23). Transforming growth factor (TGF)-β receptor can also be internalized by CME or a clathrin-independent/lipid raft endocytic pathway (22, 43). In the case of CD317, clathrin- and lipid raft-mediated endocytic processes act in concert to mediate efficient endocytosis (44). Endocytic routes are often determined by structural motifs in the cytoplasmic portion of cell surface receptors (25, 45). For example, it has been shown that ligation promotes phosphorylation of the Tyr in the ITAM motifs of a subset of BCR that are selectively retained at the cell surface to serve as scaffolds for the assembly of signaling molecules. In contrast, the non-phosphorylated receptors are rapidly endocytosed (46).
Several internalization pathways, including clathrin- dependent and -independent mechanisms, share the requirement for dynamin (15, 36, 37, 47) and Arf6. Dynamins are large GTPases (MW ~100 kDa) responsible for fission of newly formed endocytic vesicles from the cell membrane (15, 48). The GTP hydrolysis provides the force required for membrane budding events at the neck of the nascent endocytic vesicle (49). Development of mutant dynamin constructs (50, 51) and dynamin specific inhibitors such as dynasore (52) have helped determine the requirement for dynamin involvement in endocytic pathways.
Arf6 is a small GTPase, implicated in various cellular events such as cell adhesion, structural organization at the cell membrane, endocytosis, macropinocytosis and phagocytosis (53). Arf6 has been shown to regulate both clathrin-dependent and -independent endocytic pathways (54). MHC class I molecules are endocytosed via a clathrin independent, Arf6 mediated pathway (14, 55). Other plasma membrane proteins such as α1 integrins, M2-muscaranic acetylcholine receptors and the peripheral myelin-membrane protein are also internalized by Arf6 mediated pathways (54). In addition, Arf6 is known to be required for endosomal recycling of IL-2 receptor alpha subunit, GPI-anchored proteins and MHC class I molecules (56, 57) and for induction of membrane protrusions and uptake of plasma membrane proteins into macropinosomes (58).
Macropinocytosis is a specialized form of endocytosis, whereby extracellular fluid is engulfed forming large endocytic vesicles called macropinosomes (59–62). Although there are no specific cargoes or markers that define this endocytic mechanism, sensitivity to inhibitors of Na+/H+ exchange, such as amiloride, and enhanced fluid-phase uptake in response to growth factor stimulation are used to discriminate this mechanism (62–64). Co-endocytosis of fluid phase markers such as lucifer yellow and dextran into large vesicles (0.5–5 µm) allows the tracing of macropinosomes in cells. Macropinocytosis is associated with plasma membrane ruffling mediated by Ras superfamily GTPases (65, 66). Phorbol esters stimulate macropinocytosis by inducing membrane ruffles via protein kinase C and Rac (67, 68). The formation of macropinocytic vesicles has been shown to involve actin polymerization (64, 69), dynamin (70, 71), phosphoinositide 3-kinase (PI3K) (72) and small GTPases, such as Arf6 (54, 73), Cdc42 and Rac1 (74). The GTPase function of Cdc42 is thought to be essential for macropinocytosis in dendritic cells (DC) (66). However, none of these biochemical characteristics in itself can be considered a hallmark for macropinocytosis. For example, amiloride does not specifically inhibit macropinocytosis in DC (75) and, ICAM-1 and PECAM-1 internalization, which is known to involve macropinocytosis does not require PI3K (70). The macropinosomes mature intra-cellularly by fusing with endosomes and later with lysosomes (59, 62, 76). Aquaporins (AQP), the membrane water channel proteins, have been shown to play an important role in controlling the cellular volume during macropinocytosis in immature DC (77). Along with amiloride sensitive Na+/H+ channels, AQP3 and AQP7 facilitate concentration of macromolecules and efficient antigen presentation.
Micropinocytosis is the process of non-specific, constitutive uptake of extracellular molecules by the formation of small vesicles from the plasma membrane in the absence of activation or membrane ruffles (61). Micropinocytosis is morphologically distinguishable from macropinocytosis by the size of the pinosome, i.e., lesser than 0.2 µm. In addition, micropinocytosis is not dependent on actin cytoskeleton and PI3K (78, 79). Recently, dynamin 2 has been shown to mediate micropinocytosis in epithelial cells (80).
Ubiquitination is a post-translational and reversible modification in which the small (76-residue) protein ubiquitin (Ub) is attached to a substrate protein. This modification has been shown to be involved in the endocytosis and intracellular trafficking of cell surface receptors (81, 82). This modification occurs through a multi-step process that requires the action of a hierarchical set of enzymes (E1, E2 and E3) leading to the linkage of Ub to the ε-amino group of lysine residues in the acceptor protein (83). Receptors can be mono-, multi-mono-, or polyubiquitinated. The addition of a single molecule of Ub to a single Lys residue in the target protein (monoubiquitination) or the addition of single Ub to different residues (multiubiquitination), among other things, can dictate endocytic fates (84). The Ub itself has seven different Lys residues that can be used to form chains of polyubiquitin. The formation of Lys-48 polyubiquitin chains is known to target proteins for degradation in the 26S–proteasome (85). Lys-63 linked polyubiquitinated chains is known to involved in receptor endocytosis and signaling, as in the case of EGFR (86) and the lysosomal degradation of major histocompatibility complex (MHC) class I molecules (87). Additional studies on ubiquitination of EGFR (86, 88), TGFβR (22) and FcεRI (89) underline the role of ubiquitination in determining receptor endocytosis and trafficking.
Once internalized, the receptor-ligand complexes enter early endosomes (90, 91). The classical early endosomal compartment is defined by the presence of early endosomal antigen (EEA)1. Recent studies indicate that early endosomes are a more heterogeneous than previously believed and that they possess distinct membrane domains (92–95). Two distinct populations of early endosomes have been discovered, one that is highly dynamic and rapidly maturing and another that is static and mature much more slowly (95). Subsequent maturation of endosomes is regulated by recruitment of Rab GTPases. Membrane domains of endosomes contain a mosaic of different Rab proteins (96, 97). Many molecular switches are likely involved in the transition between endosomal compartments (98, 99). From the early endosomes, the cargo that is recycled to plasma membrane are sequentially transported from Rab5 to Rab4 and Rab11 domains, the cargo destined for degradation are transported from Rab5 to Rab7 domains (93, 96, 100). Consequently, GFP-tagged Rab proteins can be used to identify endosomal compartments. Rab5 and Rab4 identify distinct domains of early sorting endosomes; Rab4 is associated with regions in the early endosomes destined for fast recycling to the plasma membrane. Recycling endosomes are characterized by the presence of Rab11, whereas Rab7 is the key determinants for the late endosomes. Cargos from late endosomes are subsequently targeted to lysosomes for degradation (93, 96).
NK cells are large granular lymphocytes that play an important role in the defense against viral infections and tumorigenesis through the production of cytokines and cytotoxicity. NK cell functions are regulated by the expression of a broad number of activating and inhibitory receptors. During NK cell-target cell interactions, whether NK cells are activated or not depend on the balance of the signals emanating from these receptors (101). The NK cell activation receptors represent a family of heterogeneous molecules, including NKG2D, CD16, 2B4 and members of natural cytotoxicity receptors (NCR), NKp30, NKp44 and NKp46 among others (102). CD94/NKG2A, members of killer cell immunoglobulin-like receptor (KIR) family and immunoglobulin-like transcripts (ILT) are important inhibitory receptors expressed by NK cells (101, 103, 104). These inhibitory receptors bind to MHC class I molecules on normal cells and prevent excessive NK cell activation resulting in untoward cytotoxicity and inflammation (104). Other inhibitory receptors expressed on NK cells, although on other immune cells as well, include LAIR-1 (105–107), CD300a (108) and KLRG1 (109). For a more complete list of NK cell receptors see the recent review of Biassoni (102).
NKG2D belongs to the C-type lectin protein family and was once thought to act as an activation receptor for NK cells and costimulatory receptor for T cells, but more recent evidence indicates that it may also be co-stimulatory for NK cells (110, 111). NKG2D plays a major role in immunity against tumors and pathogens (112). Human NKG2D has several ligands, including MHC class I-chain related proteins (MIC)-A and MIC-B and UL16-binding protein (ULBP). These ligands are usually upregulated when normal cells are transformed, subjected to stress or infected (112). Ligand binding to human NKG2D induces activation signals through the associated DAP10, resulting in cytotoxicity towards the susceptible target cells and/or secretion of inflammatory cytokines (112).
Soluble and membrane bound ligands can lead to NKG2D downmodulation, which promotes tumor evasion (112–116). Colorectal cancer patients have abnormally higher levels of soluble MIC-A and –B in circulation. Interestingly, incubation of NK cells from these patients with autologous serum that contains soluble MIC-A and –B, not only resulted in downregulation of NKG2D, but also of the activation receptor NKp44 (113). Moreover, internalization of the chemokine receptors CXCR1 and CCR7 was also observed. The secondary downregulation of NKp44, CXCR1 and CCR7 seems to be dependent on NKG2D downmodulation, although how this occurs is not obvious. Antibody induced internalization of NKG2D in NK cells involves the actin cytoskeleton, as disruption of actin polymerization impairs the rate of endocytosis (21). As on NK cells, tumor-derived NKG2D ligands impair expression of NKG2D and its activation on CD8+ T cells (115). Groh et al. showed that the total amount of human NKG2D on CD8+ T cells was decreased after contact with ligand expressing target cells, as a consequence of lysosomal degradation (as shown by treating cells with the two inhibitors of lysosomal degradation bafilomycin A and chloroquine) (115). NKG2D expressed by murine NK cells is also downregulated when exposed to ligand (RAE-1ε) (114). This downregulation was sensitive to hypertonic medium suggesting that it is clathrin-mediated. Although there was no direct evidence of the adaptor protein AP2 binding to DAP10 YXXM motif, the motif is required for ligand-induced endocytosis, as is PI3K activation (114). [The cytoplasmic YXXΦ motif (Φ, a bulky hydrophobic amino acid) of several membrane receptors serves as a docking site for AP2 (25, 45)].
Roda-Navarro and Reyburn (117) have found that in resting human NK cells NKG2D resides primarily on the cell surface, while in cytokine activated NK cells, an intracellular pool of receptor can also be found that cycles to the plasma membrane (117). They also observed that exposure of NK cells to MIC-B expressing target cells resulted in degradation of NKG2D and DAP10 in the lysosomes (as indicated both by treatment with chloroquine and confocal microscopy observations). In contrast, for murine NKG2D, Ogasawara et al. reported that for Ba/F3 cells expressing mouse NKG2D and DAP10, the cell surface expression of NKG2D was reduced after interaction with RAE-1γ expressing RMA cells, but this downmodulation did not affect the total cellular NKG2D, suggesting that the internalized NKG2D was not degraded (114). The reasons for the discrepancies between studies are not known but differences between mouse and human NKG2D receptors and the cells expressing NKG2D, i.e., primary cells vs transfectants, need to be considered.
NKp46 is expressed on the surface of both activated and resting NK cells (118). Viral encoded ligands for NKp46 have been identified (119, 120), but the tumor ligands still remain unknown. Sivori et al, showed that incubation of human NK cell clones with immobilized anti-NKp46 antibody resulted in the downregulation of NKp46 from the cell surface (121). NKp46 downmodulation induced by incubation with the tumor cell line 721.221 was not blocked by the ligation of inhibitory receptors (122). Unexpectedly, NKp46 expression is decreased when NK cells are treated with the proteasome inhibitor bortezomib (123). To explain this apparent contradiction, these investigators suggested that the decreased levels of NKp46 may be, at least in part, is attributable to the known inhibitory effects of bortezomib on NF-κB function. In fact, treatment of NK cells with a selective pharmacological inhibitor of NF-κB substantially decreased NKp46 cell surface levels (123).
CD16, also known as the low affinity receptor for IgG (FcγRIIIa), is the receptor responsible for the induction of the NK antibody-dependent cellular cytotoxicity (ADCC) (124). CD16, expressed by most NK cells, is able to transduce an activating signal through its association with CD3ζ and FcεRIγ chains (125, 126). Upon CD16 cross-linking using specific mAb, the receptor is rapidly internalized by an ATP independent mechanism that requires an intact actin cytoskeleton (127). Others have shown that specific mAb mediated downregulation of CD16 is blocked by the Zn2+-dependent metalloproteases inhibitor, 1,10-phenanthroline, suggesting that the downregulation of CD16 from the cell surface is more a consequence of metalloprotease induced shedding of receptor rather than internalization (128). Shedding of CD16 has also been shown after treatment of NK cells with the PKC activator PMA (128, 129). Additional studies are required to determine the contribution of internalization and shedding to the mAb mediated downregulation of CD16. It has been shown that the phosphatidylcholine-specific phospholipase C (PC-PLC) is associated with CD16 in lipid rafts and is involved in the regulation of CD16 membrane expression, as a specific inhibitor of PC-PLC strongly down-modulates CD16 expression (127). The CD16 CD3ζ subunit is rapidly ubiquitinated by lck after aggregation of CD16, probably directing this subunit to degradation (130).
2B4 is one of the three members of the signaling lymphocytic activating molecule (SLAM) family receptors expressed by NK cells that has been shown to have both activation and inhibitory potential (131–137). 2B4 crosslinking with mAb or incubation with target cells expressing its ligand CD48 induces downregulation from the cell surface by internalization. Inhibitory receptors do not block ligand induced 2B4 downmodulation. 2B4 internalization requires Src kinases, although it has been suggested that the internalization is independent of PI3K activity or actin polymerization (122). Other examples of receptors belonging to the SLAM family that are downregulated upon ligand binding are: SLAM (CD150) (138) and CD229 (LY9) (139). However, another member of this family, NTB-A was not down-regulated after culturing NK cells with NTB-A expressing cell lines [NTB-A exhibits homophilic binding (122)]. Within the SLAM family, CD229 (Ly9) is the only member that binds to the μ2 chain of the AP-2 complex (139).
CD94/NKG2A is an inhibitory receptor of the C-type lectin family found in humans and mice, which binds to the non-classical class I molecule HLA-E (140–142). Ligation of the receptor with HLA-E promotes phosphorylation of the Tyr in the cytoplasmic ITIM domain of CD94/NKG2A leading to the recruitment and activation of src homology 2 domain-bearing tyrosine phosphatase (SHP)-1 and/or SHP-2 (143–145). SHP-1 activation results in dephosphorylation of Vav and ezrin/radixin/moesin (ERM) proteins leading to the actin depolymerization and disruption of the cytoskeleton. As a consequence, lipid rafts and activating receptors are excluded at the site of contact between NK cell and target cell, negating the propagation of positive signals (146–148).
The investigation of CD94/NKG2A endocytosis and trafficking revealed some surprising results (21, 149). CD94/NKG2A continuously recycles between the cell surface and intracellular compartments in a process that is independent of inhibitory signaling (149). Endocytosis of CD94/NKG2A resembles classical macropinocytosis in that it is independent of clathrin, lipid rafts and PI3K activity, requires Rac1, and is sensitive to amiloride. It differs from classical macropinocytosis by not requiring actin, dynamin or Arf6. Actin polymerization is required for the recycling of CD94/NKG2A back to the cell surface (21). Once endocytosed, CD94/NKG2A localizes in the early endosomal compartment, as demonstrated by co-localization with EEA1 and Rab5, and recycles back to the cell surface.
CD94/NKG2C is a DAP12 associated activation receptor closely related to CD94/NKG2A (103). The extracellular region of NKG2C is 92% identical to NKG2A and the receptor also recognizes HLA-E (albeit with lower affinity) (103, 150). Using a mouse cell line Ba/F3 transfected with CD94/NKG2C/DAP12, we have shown that CD94/NKG2C is a long-lived receptor that differs in its trafficking route from CD94/NKG2A (149). Like transferrin receptor, the cell surface levels of CD94/NKG2C/DAP12 are decreased after brefeldin A treatment, suggesting that CD94/NKG2C/DAP12 traffics through brefeldin A sensitive intra-cellular trafficking compartments. Brefeldin A is known to cause tubulation and fusion of early endosomes with the trans-Golgi network thereby blocking recycling, as shown for the transferrin receptor, and to obstruct anterograde transport from the endoplasmic reticulum to the Golgi complex (151, 152).
KIR comprise a group of polymorphic molecules that consist of both activation and inhibitory receptors that are expressed on NK cells and a subset of T lymphocytes (153). Individual KIR bind to MHC HLA-A, -B or -C class I molecules. KIR have two (KIR2D) or three (KIR3D) immunoglobulin domains in the extracellular portion of the molecule. The inhibitory KIR have a long cytoplasmic tail (KIR2DL and KIR3DL) that contain ITIM sequences, while the activating KIR contain a short cytoplasmic tail (KIR2DS and KIR3DS) that lack ITIMs but contain lysine in the transmembrane portion. The positively charged amino acid is required for pairing with the ITAM containing adaptor molecule DAP12, which is required for signaling (154). An exception to this general classification is KIR2DL4. KIR2DL4 has a transmembrane arginine and one ITIM in the intracellular domain. Ligation of this receptor with mAbs or with its ligand HLA-G has shown that it has mostly activating function. This is manifested by the potent production of IFN-γ, as well as of other cytokines, and weak cytotoxic activity (155–157). The activating functions of KIR2DL4 depend on the transmembrane association with the adaptor molecule FcεRIγ instead of DAP12 (158); however, it has also been shown that some KIR2DL4 mediated activation signals are independent of the association with FcεRIγ (159).
On CD8+ T cells, in the absence of TCR stimulation, engagement of KIR2DL with the ligand HLA-Cw7 induces downregulation of KIR2DL. The internalized KIR2DL is routed to lysosomes and degraded. When TCR is stimulated with appropriate antigen, the KIR2DL downregulation is inhibited and stable cell surface levels of KIR2DL are maintained, likely serving to elevate the activation threshold of cytotoxic T cells to avoid excessive stimulation leading hyper inflammation (160). For NK cells conflicting results have been reported. Huard et al. have shown that KIR2DL2 is downregulated after incubation of NK cell clones with target cells bearing the appropriate ligand (161). As on T cells, this downregulation showed a slow kinetics and, more importantly, was associated with the downregulation of other activating receptors, such as CD16, rendering the NK cells unresponsive to activating signals (161). However, others have shown that crosslinking of a cell line transfected with KIR2DL1 with specific mAb antibodies did not induce downmodulation of the receptor (122). It is possible that the different experimental settings may explain these conflicting results. In any case, more detailed studies in more physiological settings are required to study the fate of KIRs after interaction with their ligands.
A role for PKCδ in controlling expression of KIR has been proposed. Chwae et al. have used a panel of Jurkat T cells transfected with different constructs expressing wild type and mutant forms of KIR, as well as chimeric fusion receptors composed of the extracellular and transmembrane part of CD8α and the intracellular tail of KIR3DL1 (162, 163). They have shown that PKCδ activation upregulates membrane expression of KIR. This upregulation is associated with increased processing in the ER-Golgi and also increased recycling of KIR via EEA1+ endosomes. They also showed, using pharmacological inhibitors, that PKCδ plays a role in the exocytosis of KIR from secretory lysosomes. Others have shown that phosphorylation of a specific Ser residue (Ser394) in the cytoplasmic tail of KIR3DL1 inhibits the inhibitory function of this receptor, as well as its internalization and turnover (164).
KIR2DL4 binds soluble HLA-G and is endocytosed into Rab5+ endosomes in a dynamin-dependent process. The result of this internalization is the secretion of inflammatory and angiogenic factors, such as TNF-α, IL-1β, IL-8 and IFN-γ (156). As the expression of HLA-G is restricted to trophoblast cells that invade maternal deciduas during early pregnancy, these findings suggest that interaction of KIR2DL4 with HLA-G promotes vascularization of the maternal decidua (156). The KIR2DL4 activation signaling has been proposed to occur in the endosomes, as only soluble but not membrane-bound or plate-bound antibodies to KIR2DL4 induced IFN-γ secretion. However, it has also been shown that uterine NK cells can be stimulated not only by soluble HLA-G, but also by membrane bound HLA-G. Co-incubation of uterine NK cells with cell lines transfected with membrane bound HLA-G leads to proliferation and IFN-γ production (165).
Receptor downregulation upon ligand binding is a well documented means for modulating receptor signaling (1, 90, 160, 161). As shown for the TCR (1), this downregulation serves to dampen the intensity and duration of a cell’s response as means of regulating the overall response. Most if not all of the NK cell activation receptors studied have been shown to be downregulated upon ligand binding (112, 121, 122). After ligand-induced downmodulation, the default intracellular trafficking pathway seems to lead to lysosomes for degradation (90). By limiting the availability of activation receptors by downmodulation from the cell surface, the cells not only become refractory to more stimulation, but may also prevent cell death associated with sustained activation (activation induced cell death) (166, 167). However, it is not yet clear if signaling from NK cell activation receptors continues in endocytic compartments.
As for inhibitory receptors, there is a requirement that they be continuously expressed on the NK cell surface to prevent undesirable activation. The primary NK cell inhibitory receptors, CD94/NKG2A and inhibitory KIRs, responsible for recognition of self MHC class I molecules, are constantly exposed to ligand expressed by surrounding cells. It could be potentially disastrous for these receptors to be downregulated, as the balance of signals could then tip toward activation leading to auto-reactive NK cells, as many ligands for NK cell activation receptors are expressed by normal cells. In order to restrain this unwanted activation of NK cells, it would be desirable for these inhibitory receptors to remain in constant supply at the cell surface. This could be achieved by new synthesis, rapid recycling or a combination of both. It would seem advantageous for inhibitory receptor trafficking to be designed in such a way to insure this continuous supply at the cell membrane. From an energetic point of view, it would be beneficial for the cell to achieve this by quickly directing the internalized receptors from early endosomal compartments back to the cell membrane rather than rely on de novo protein synthesis. In this light, our studies on human CD94/NKG2A have shown that cell surface expression levels are maintained in the face of ligand exposure and that the receptors are long lived and, upon endocytosis, rapidly recycle to the cell surface using an abbreviated trafficking pattern (21, 149). Current data indicates that all NK cell inhibitory receptors do not share this trafficking scheme as KIR inhibitory receptors may be downregulated by ligand exposure under certain experimental conditions (see above). At this moment, it is not clear if these differing results for CD94/NKG2A and KIR are due to the different experimental settings or something else. Clearly, more experiments are warranted to address endocytosis and trafficking of inhibitory receptors in situations that reflect physiological conditions.
A curious aspect of CD94/NKG2A trafficking is that its endocytosis is apparently actin-independent. The actin cytoskeleton is absolutely essential for NK cytotoxicity (168) and is shown to be a target for inhibitory signaling (147, 169, 170). The actin cytoskeleton also provides the scaffold for the formation of the signalosome, a multi-molecular assembly that is formed by aggregation of pre-assembled lipid raft platforms and it plays a major role in regulating the endocytosis of cell surface proteins from these signalosomes. The actin cytoskeleton is known to be involved in most internalization processes including macropinocytosis (64, 171, 172). Interestingly, the engagement of CD94/NKG2A, as well as inhibitory KIR, on NK cells disrupts actin reorganization (147, 170, 173). So, perhaps it is not surprising that the endocytosis of CD94/NKG2A apparently does not depend on actin polymerization. CD94/NKG2A does, however, require actin for de novo and recycling protein trafficking to the cell surface (21). It is possible that by-passing the requirement of actin allows CD94/NKG2A to quickly internalize and recycle to a different location within a short period of time, thereby, maintaining inhibitory receptor presence on the cell surface.
In the complex scenario of receptor endocytosis and intracellular trafficking, much has been learned, but much more needs to be explored. Particularly, much less in known about NK cell receptor endocytosis and trafficking. We have to be aware of the multiplicity of scenarios, such as cell type, environment, ligand conditions, that can alter a receptors trip through the cell, perhaps, in some cases, affecting cell function. In particular, for immune system receptors, such as those on NK cells, a deeper understanding of the factors regulating trafficking could potentiate the development of drugs or other means for controlling receptor surface expression that would have therapeutic value. A good example of this might be having the ability to maintain NKG2D expression in the face of ligand exposure.
This work is supported by funds from the Division of Intramural Research, NIAID/NIH to JEC, Intramural program, FDA to FB and by The Jaffe Food Allergy Institute, Mount Sinai School of Medicine to MM. We would like to thank Dr. Konrad Krzewski for his critical reading of the review.
The authors have no financial conflicts of interest.