| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Department of Pathology, University of Cambridge, Cambridge, UK
1Correspondence: Tennis Ct. Rd., Cambridge CB2 1QP, UK. E-mail: pr284{at}cam.ac.uk or ht20{at}hermes.cam.ac.uk
 |
ABSTRACT |
|---|
 TOP ABSTRACT BACKGROUNDRECENT RESULTSCONTROVERSY/UNANSWERED QUESTIONSPROSPECTS AND PREDICTIONSREFERENCES |
|---|
Key Words: membrane connective structure/nanotube
|
BACKGROUND |
|---|
|
TOPABSTRACTBACKGROUNDRECENT RESULTSCONTROVERSY/UNANSWERED QUESTIONSPROSPECTS AND PREDICTIONSREFERENCES |
|---|
. The IS was first described during the interaction between T cells and antigen-presenting cells (APC), but formation of an immune synapse also occurs during B cell activation and function (2
, 3)
. The IS has been suggested to act as a platform for the extended interactions required to trigger the differentiation of naive cells to the effector state, and occurs for both CD4+ helper and CD8+ cytotoxic T cells (CTLs). The mature T cell synapse comprises a central (c) supramolecular activation cluster (SMAC) that contains the T cell receptor (TCR) and intracellular signaling molecules, as well as a peripheral (p) SMAC composed of integrins like LFA-1, and cytoskeletal components such as talin (1)
. After the initial activation CTLs act to clear virally infected and tumor cells; this function is mediated by the formation of a secretory, cytotoxic synapse between the effector and the target cell. This CTL secretory synapse is characterized by a cSMAC that contains a secretory domain along with the signaling region (4)
. Although the majority of these data have been obtained from in vitro studies, mature IS have been observed to form in vivo between CTLs and virus-infected astrocytes (5)
.
Natural killer (NK) cell function is regulated by a balance of intracellular signals triggered by activating receptors specific for ligands expressed on the surface of surrounding cells and inhibitory receptors specific for the MHC class I molecules (6)
. In humans, a wide repertoire of activating receptor complexes has been described including the natural cytotoxicity receptors (NCR), NKp46/CD3
, NKp30/CD3, NKp44/DAP12, NKG2D/DAP10, the Fc receptor CD16/CD3/Fc
R and the coactivating receptor 2B4/LAT (7)
. Cellular ligands for the NCRs are unknown, although they can bind certain viral hemagglutinins (8)
. Binding of NKG2D/DAP10 to the MHC class I-related chains (MIC) A and B or the UL16 binding proteins (ULBP1–4) (9
, 10)
leads to the recruitment and activation of PI3K and Grb2. The 2B4 receptor binds CD48, and upon receptor phosphorylation the adaptor protein signaling lymphocyte activation molecule-associated protein (SAP) and the kinase Fyn are recruited to the cytoplasmic tail of the receptor (11
12
13)
. The major inhibitory receptors are the lectin-like heterodimer CD94/NKG2A and the killer cell immunoglobulin-like receptors (KIRs) (14
15
16)
. These receptors, on MHC class I binding, switch activating signals off by recruiting the phosphatase SHP-1. Thus immunosurveillance by NK cells involves multiple rounds of both cytotoxic and noncytotoxic interactions. Moreover, activation and inhibition are spatially regulated, as a single NK cell can interact with, and spare, a normal cell while simultaneously killing a malignant target (17)
.
Studies realized with both cultured primary NK cells or tumor NK cell lines have shown that the NK cell cytolytic process involves the formation of a cytotoxic NK cell immune synapse (cNK-IS). The cNK-IS contains both a cSMAC and a pSMAC, and is similar to the mature T cell synapse (i.e., signaling molecules at the center surrounded by LFA-1 and cytoskeletal components at the periphery) (Fig. 1
) (18
19
20)
. We have shown that cSMAC localization of the activating receptor NKG2D correlates with receptor-mediated activation of NK cytotoxicity, suggesting that the segregation of activating receptors to the cSMAC is an important step for the cytotoxic activity of IL-2 cultured NK cells (19)
. Another critical step for cytotoxicity is translocation of the MTOC and cytotoxic granules toward the activating synapse. An inability to achieve this step is the basis of the reduced cytotoxicity of decidual NK cells (21)
.
|
An inhibitory NK cell immune synapse (iNK-IS) is observed when NK cells interact with targets that express ligands for the NK inhibitory receptors (22
, 23)
. After just 1 min of interaction, iNK-IS is characterized by the central distribution of a single cluster of the phosphatase SHP1 surrounded by talin. In contrast, in cytolytic interactions this molecule is redistributed to the periphery at this time. The activating signaling molecule Lck is seen dispersed in multiple small clusters at the iNK-IS (23)
, consistent with the requirement for early tyrosine phosphorylation for KIR inhibition (24)
. KIRs and their MHC class I ligands colocalize at the contact area between NK cells and resistant target cells (25)
. Although inhibitory KIRs are clustered across the immune synapse, phosphorylation of these receptors at iNK-IS has recently been observed to be restricted to discrete microclusters (26)
. This finding could explain why Lck and SHP-1 have been observed in dispersed small clusters at the iNK-IS monitored at 10 min (23)
. Finally, ligation of either the KIR or CD94/NKG2A inhibitory receptors abolishes the redistribution of the multimolecular signaling complex associated with lipid rafts to the contact site (27
, 28)
.
While the majority of studies of the NK-IS have been carried out using NK cells that have been cultured in vitro with IL-2, resting NK cells also form cytotoxic synapses with MHC class I negative targets (29
, 30)
. Dynamic studies have shown that accumulation of 2B4 at these synapses occurs rapidly, suggesting an important role for this receptor in the initial adhesion of resting NK cells to targets before initiation of the killing process. Moreover, 2B4/SAP complexes form stable clusters at the synapse, consistent with the idea that a sustained signaling triggered by this receptor complex is required for activation of lysis by resting NK cells (29)
. It is worth noting here that although dendritic cells are now recognized as being critical in the activation of resting NK cells (31
32
33)
, NK-IS formation between naive peripheral NK cells and APCs has not yet been studied. Given that immature dendritic cells can both activate resting NK cells and be targets for cytotoxicity by activated NK cells, it would be interesting to compare the immune synapses formed in these distinct interactions; perhaps the organization of the synapse for initial activation of the NK cell is different to the cytotoxic NK-IS in the same way as the formation of a cSMAC is not required for activation of naive CD8+ T cells (34)
.
Intercellular transfer of molecules is a process that seems to occur commonly as a consequence of immune synapse formation during communication between immune cells. Acquisition of APC molecules by T cells, called "trogocytosis" (35)
, has been reviewed by Hudrisier and Bongrand (36)
. This process of peptide-MHC acquisition involves the transfer of membrane fragments and depends on signaling via the TCR (37)
. The biological consequences of antigen capture by T cells are poorly understood. A number of different hypotheses, including T cell-T cell antigen presentation, fratricidal killing, and regulation of memory cell homeostasis, have all been proposed as possible consequences of peptide/MHC transfer (38
, 39)
. T cell activation correlates with an increased proportion of antigen among the materials acquired from target cells (40)
. Acquisition of antigen from target cell has also been reported after formation of the B cell immune synapse. Both soluble antigen and membrane-attached antigen acquired by the B cell receptor (BCR) are internalized, processed, and eventually presented to T cells (2)
.
Recently, bidirectional transfer of receptor and ligands during cytolytic or noncytolytic NK cell/target cell interactions has been documented (19
, 41
42
43
44)
. The functions of ligand acquisition by NK cells and receptor transfer to target cells are still unclear. The possible cellular mechanisms and the physiological or pathological significance of these processes are the main topics of this review.
|
RECENT RESULTS |
|---|
|
TOPABSTRACTBACKGROUNDRECENT RESULTSCONTROVERSY/UNANSWERED QUESTIONSPROSPECTS AND PREDICTIONSREFERENCES |
|---|
. This study described the stable segregation of KIR receptors and their specific MHC class I ligands in a peripheral region surrounding a central cluster of LFA-1/ICAM-1. More recently, different distributions of human leukocyte antigen (HLA) molecules at synapses formed in different cell conjugates or in the same cell conjugate over time have been observed. After accumulation of HLA-C at the iNK-IS, some HLA molecules transfer to the membrane and cytoplasm of NK cells (41)
. This was the first observation of protein acquisition by NK cells, an active phenomenon shown to require NK cell ATP but not active polymerization of the NK cell actin cytoskeleton.
Acquisition of MHC class I molecules by mouse NK cells during noncytolitic interactions was also described at this time. In the mouse, NK cell inhibition is mediated by the specific binding of different MHC class I (H-2) alleles by different members of the C-type lectin-like receptor family Ly49 (45)
. Two papers described the uptake of MHC by NK cells expressing Ly49 molecules (43
, 44)
. This process was observed to occur both in vivo and in vitro and depended on specific binding of an Ly49 receptor to a specific H-2 allele. Moreover, a strong down-modulation of Ly49 paralleled the uptake of H-2 molecules. This phenomenon has been studied intensively in the interaction of Ly49A with its specific high-affinity ligands H-2Dd and H-2Dk, but has also been observed for the Ly49C/H-2Kb interaction. While Sjostrom et al. (43)
found that Ly49A+, H-2Dd+ NK cells could still acquire H-2Dd, Zimmer et al. (44)
observed that H-2Dd or H-2Dk expression by Ly49A+ NK cells precluded the uptake of H-2Dk or H-2Dd by these NK cells, consistent with masking of the receptor by a specific association with ligand in cis (i.e., on the effector cell) (46)
. The basis of this discrepancy between these different studies is not clear.
More recently, the transfer of human and murine inhibitory receptors from NK cells to target cells has been described (42)
. Inhibitory KIR and Ly-49 receptors transfer to cells expressing a cognate MHC class I ligand and the amount of receptor transferred correlates with the levels of target cell MHC class I expression. Other surface molecules, but not a cytoplasmic dye, can also transfer to target cells. Thus, a bidirectional transfer of receptors and MHC class I ligands occurs at the inhibitory NK cell immune synapse.
Intercellular protein transfer at the cNK-IS
Trans-synaptic capture of membrane fragments by NK cells was first documented during cytotoxic interactions (47)
. This process was controlled by Src kinases, ATP, Ca2+, and PKC and involved rearrangement of the actin cytoskeleton. Specific ligand recognition by NK cell-activating receptors was postulated as the mechanism mediating this process. Inhibitory signals provided by MHC class I molecules abrogated this acquisition of membranes. Thus, both activating and inhibitory receptors at the immune NK synapse appear to regulate membrane capture as well as killing of target cells. Recently the transfer of activating molecules from effector cells to target cells during effector cell activation has also been observed. Time lapse videomicroscopy has shown that the NK cell-coactivating receptor 2B4 is stably clustered at the cNK-IS, concentrated at membrane connective structures (MCS) formed as the NK cell detaches, and left behind on the target cells after the NK cell detaches completely (29)
(Fig. 2
). MCS were first observed during cytotoxic interactions in mouse NK cells, where they were proposed to be involved in inducing physical damage to the target cells; they were not found during noncytolytic interactions (17)
. Recently MCS have also been found communicating other immune cells and have been called nanotubes (48)
. Our recent results indicate that transfer of NK cell-activating receptors could be a more general process (19)
. We have observed specific transfer of NKG2D from activated NK cells to target cells expressing MICB. Moreover, the transfer depends on the receptor/ligand binding and is accompanied by a down-modulation of NKG2D from the NK cell surface. As in the case of the 2B4 molecule, NKG2D is found in MCS-communicating NK cells and target cells, and transferred NKG2D colocalizes with transferred 2B4. Synaptic acquisition of MICB by activated NK cells is also observed in these experiments. The ligand molecules transferred are displayed in patches and colocalize with NKG2D. Thus, a bidirectional transfer of receptor/ligand pairs occurs at the cNK-IS, as observed for iNK-IS. This phenomenon correlates with a reduction in the NK cell cytotoxic function as discussed below (19)
(Fig. 2)
.
|
During immunosurveillance NK cells are frequently close to pathogens that infect surrounding cells; indeed, synaptic transfer of molecules can contribute to the spread of pathogens. For example, CD21 is the primary receptor used by Epstein-Barr virus (EBV) for entry into the cell, and although NK cells do not express CD21, occasionally they are infected by the EBV. An important finding is that this infection may depend on synaptic uptake of CD21 by NK cells interacting with EBV-infected cells that allows EBV viral particles to enter NK cells (49)
. A related phenomenon is the NK cell acquisition of poliovirus receptor (PVR) on binding of CD96 expressed in NK cells (50)
.
|
CONTROVERSY/UNANSWERED QUESTIONS |
|---|
|
TOPABSTRACTBACKGROUNDRECENT RESULTSCONTROVERSY/UNANSWERED QUESTIONSPROSPECTS AND PREDICTIONSREFERENCES |
|---|
Mechanisms of protein transfer
Cell membrane proteins could be released in two ways: shedding of proteins as soluble fragments or release of the whole membrane proteins (36)
. These two possible mechanisms will be discussed separately.
Shedding of protein fragments
Shedding of proteins cleaved/released by activation of specific proteases/phospholipases has been observed in multiple systems (51)
and could represent a widespread molecular mechanism explaining protein transfer. This process is tightly regulated by cytokines secreted by neighbor cells, although a dependence on intracellular signaling has also been reported. The functional significance of this process during immune responses has been demonstrated (52
53
54)
. Release of soluble ICAM-1 by melanoma cells can abrogate the non-MHC-restricted cytotoxicity mediated by NK and lymphokine-activated killer cells, which may be one of the mechanisms by which neoplastic cells escape immunosurveillance (52)
. Similarly, soluble HLA molecules found in the serum of healthy donors, and patients with inflammatory diseases, allografts, and autoimmune diseases, can also down-modulate the T cell response (55)
. Soluble forms of NKG2D ligands (sMICs and sULBPs) have been found in the serum of patients with different tumor malignances, and their presence is related to an impairment of CTL and NK cell cytotoxic function (56
57
58)
. Although intracellular signals triggered in targets on NK-IS formation might induce the release of soluble MIC fragments, there are no data at this time to support this hypothesis. Most probably cell-cell contact is not required for the release of soluble NKG2D ligands. Shedding of tumoral MIC and ULBP-2 fragments by metalloproteases has been described (58
59
60)
; however, the release of NKG2D ligands in tumor exosomes has also been noted (61)
. Since several ULBPs are GPI-linked proteins, the extracellular release of some of these could also be mediated by phospholipases. All these possible mechanisms leading to the release of soluble fragments represent open questions that require careful investigation. Since soluble fragments of membrane molecules such as the adhesion molecule ICAM-1, HLA alleles, and NKG2D ligands have a serious impact on the immune response by NK cells and T cells, precise identification of the molecular mechanisms involved in the shedding of these proteins constitutes an important goal in biomedicine.
Transfer of entire membrane proteins
Transfer of whole membrane proteins between cells could be mediated by several mechanisms: vesicles released from cells that eventually will be acquired by other cells, movement of proteins in cytoplasmic bridges or membrane nanotubes formed between cells after fusion of apposed membranes, and the direct extraction of proteins, or membrane fragments that contain proteins, from neighboring cells. It is plausible to imagine that these mechanisms would not be exclusive and could operate simultaneously. Since much of the force (membrane load) resisting the membrane movement necessary to form vesicles or tethers derives from the strength of membrane-cytoskeleton adhesion, release of vesicles or formation of tethers should strongly depend on local disruption of the cytoskeleton (62)
. Moreover, proteins that are not strongly bound to the cytoskeleton can be uprooted from the plasma membrane by moderate active traction. For example, L-selectin has been described to be easily extractable from the plasma membrane when the cytoplasmic tail is deleted and, as a consequence, its capacity to bind cytoskeletal elements is prevented (63)
.
Several examples of transfer of whole membrane proteins among immune cells have been described: Acquisition by T cell of MHC class II molecules from APCs is mediated by extracellular vesicles (64)
. Acquisition of APC-derived exosome-like vesicles by CD8+ T cells has also been demonstrated recently (65)
. In the absence of a viable APC, these vesicles efficiently prime T cells for tumor rejection in vivo (66)
. On the other hand, T cells are also able to acquire, by tearing, APC membrane fragments during T cell activation. Since the efficiency of membrane capture and the strength of T cell stimulation are correlated, it is reasonable to speculate that specific T cell stimulation triggers the cellular machinery, providing the driving force for the specific capture of membrane proteins (36
, 40)
.
Mechanisms of protein transfer at iNK-IS
As described above, bidirectional transfer of receptor and ligands has been described at inhibitory or activating NK-IS. In the case of the human iNK-IS, biochemical data indicate that the full-length MHC class I molecule seems to be transferred (41)
, and this transfer is specific to cells expressing the cognate KIR. The majority of human MHC class I captured is at the plasma membrane, but transport of vesicles moving away from iNK-IS into the NK cell cytoplasm has also been observed. Secretion of exosomes containing MHC class I is probably too slow to be involved in the process of rapid MHC class I capture described in this work. Pharmacological blockade of target cell actin polymerization increased the amount of HLA acquired by NK cells (42)
. This treatment may favor detachment of the actin cytoskeleton from the plasma membrane, lowering the membrane tension and facilitating target cell release of vesicles or the formation of membrane blebs (67)
.
Although the whole MHC class I molecules are transferred to mouse NK cells, specific acquisition of H-2 alleles does not seem to involve uptake of target membrane domains, as bystander molecules and the membrane dye DiI are not transferred (43)
. Whether these molecules are pulled up from the target cell membrane as a consequence of forces dependent on the Ly49/H-2 interaction as the NK cell moves away from the target is still an unanswered question. However, the association between interaction strength and acquisition of MHC class I would support this idea.
In humans, KIRs are transferred at iNK-IS to target cells expressing cognate MHC class I ligands. Although inhibitory receptor transfer correlates with MHC class I expression levels on target cells, many other molecules are also transferred to the target cells. As with MHC class I molecules, the full-length receptor is transferred (42)
. Whether fusion of membranes or endocytosis of vesicles are involved in this process is unknown. Approximately half of the KIR transferred is not stably embedded at the target cell plasma membrane, an observation that favors the hypothesis that at least some KIR molecules may be uprooted from the NK cell plasma membrane. The remaining protein seems to be stably incorporated into the target cell membrane consistent with hypotheses that postulate the involvement of membrane fusion or vesicle endocytosis as mechanisms important for protein transfer. Since staining of the KIR extracellular domain was found to be greater in permeabilized cells than for nonpermeabilized, it is possible that some receptor might be internalized. In summary, it is plausible to think that different molecular mechanisms might operate during a noncytolytic interaction to produce KIR transfer to target cells, although the different KIR forms found in targets may also correspond to different temporal steps of the same process.
Finally, the observation that MHC class I transfer at the murine iNK-IS seems to be highly specific to cells expressing high-affinity receptors, where membrane fragments seem not to be involved whereas human MHC class I transfer seems to happen in the context of vesicles or membrane domains may indicate that different molecular mechanisms are operating in the different species.
Mechanism of protein transfer at cNK-IS
Trans-synaptic acquisition of target cell membrane fragments during the cytotoxic process has been described for NK cells (47)
and CTLs (4)
. In the latter case, electron microscopy (EM) reveals the formation of membrane bridges between cells, which may help the extraction of target cell membranes and explain the synaptic transfer of membrane proteins without changing their orientation (Fig. 3
A). These observations thus suggest that NK cells could also acquire integral membrane proteins during cytotoxic interactions either inserted in acquired membrane fragments or by diffusion through membrane bridges. These processes depend on the actin cytoskeleton, which is necessary for cNK-IS formation. F-actin anchors adhesion molecules at the synapse and so may constitute a strong platform to generate the force necessary to extract membrane fragments and proteins from the target cell. In this context, target cell membranes should be easily extractable from apoptotic cells where disruption of cytoskeletal-membrane adhesion has happened and cells show characteristic membrane blebs (68)
. However, perforin-deficient CTLs (37)
or noncytolytic cells such as CD4+ T cells (69
, 70)
have a similar ability to capture target cell molecules, suggesting that this process does not depend exclusively on the initiation of target cell death. Another molecule acquired by NK cells is PVR (50)
. In this case, acquisition of protein is observed after 2 h of incubation at 4°C; however, this is markedly increased at 37°C, indicating that this protein capture may occur via an active process of transport as well as by detaching from the targets cells when NK cells are pulled apart. We observed a similar behavior of MICB when it is transferred to NK cells (P. Roda-Navarro and H. T. Reyburn, unpublished observations). Finally, how membrane fusion occurs or how membrane bridges are disrupted as the effector and target cells separate is still unclear.
|
NK cell receptors are also transported from NK cells to target cells during NK cell activation, which demonstrates that the transfer of proteins at cNK-IS is a bidirectional process (29)
. Receptors tagged with GFP at the cytoplasmic tail are readily observed on target cells during cytotoxic interactions, which indicates that the full activating receptor molecules are transferred. Extracellular NKG2D ligand binding domain epitopes are labeled under nonpermeabilized conditions, indicating that the transferred molecule is oriented correctly on the surface of target cells (19)
. All these data indicate that traffic of vesicles or membrane domains are probably involved in this transfer process. NKG2D has been found in intracellular vesicles colocalizing with perforin, which suggests traffic of the receptor through NK cell lytic granules (19)
. One possibility would be that the internal vesicles of lytic granules that contain receptors fuse to the target cell plasma membrane to liberate the lytic machinery (perforin and granzymes), and in that way receptors reach the target (Fig. 3A
). This hypothesis is supported by EM experiments, where a DAP10-GFP chimeric molecule is observed in internal vesicles of lytic granules during cytotoxic interactions of the NKL cell line with targets expressing MICB (P. Roda-Navarro and H. T. Reyburn, unpublished results). Consistent with this idea, the presence of molecules relevant for T cell-target cell interaction (TCR, CD3, and CD8) in lytic granules from human T cells has been described (71
, 72)
. On the other hand, the redistribution of activating receptors to lamelipodia and MCS/nanotubes formed during cytotoxic interactions suggests that these structures may also be involved in protein transfer (19
, 29)
(Fig. 3B
). The movement of bulges along the MCS suggests the dynamic transport of vesicles inside the tube, and at least some of these bulges contain NKG2D (19)
. The presence of F-actin within MCS supports a transport mediated by actin, as has been described for intercellular organelle transport via nanotubular highways during communication between PC12 cells (19
, 73)
. Although these data indicate intracellular transport of molecules through the structure, the transport of receptors at the surface of MCS cannot be ruled out (48)
. In addition to these mechanisms of protein transport along the MCS, activating receptors are found clustered at cSMAC before MCS formation (29)
. As mentioned above, receptor clusters remain at the end of MCS as NK cells detach and are transferred after disruption of the filament (Fig. 2)
. The cSMAC is a region where no actin cytoskeleton is organized under the membrane but an active secretion of vesicles occurs. Vesicle secretion may lower the membrane tension (62)
, which would favor the formation of initial tethers as soon as the NK cell starts to move away from the target. These tethers could cover long distances at rapid rates, as occurs in neuronal growth cones (74)
. It has been observed that the separation of the membrane from the end of a filopodia with laser tweezers lowers the load of the membrane and stimulates the rate of assembly of actin in the filopodia (75)
. A similar process could operate during MCS formation that explains the presence of F-actin inside the structure just until the place where the receptors are clustered at the contact site with the target (19)
. When the membranes at the end of MCS are disrupted, transfer of the receptor cluster is completed. NK cell membrane resealing could be favored in a place where secretion of vesicles lowers the membrane tension (76
, 77)
.
Thus, during cytolytic interactions it is worth noting that whereas effector cells acquire membrane fragments from target cells, NK cells can also transfer components to target cells. Probably these two processes will be temporally separated: first, before the killing process, activating receptors would be transferred to target cells; and second, after the lethal hit and favored by the induction of apoptosis in the target cell and formation of membrane blebs; consequently, membrane fragments will be more easily extracted from targets by effector cells.
Fate and physiopathological significance of protein transfer during NK cell immunosurveillance
At the human iNK-IS, MHC class I molecules seem to be internalized along with the KIR molecule inside the NK cell, but the physiological consequences of this process have not been studied. Similarly, the physiological significance of the transfer of KIR or Ly49 molecules to healthy cells remains enigmatic.
H-2 molecules are down-modulated in murine NK cells upon interaction with high-affinity ligands. Although acquired H-2 is rapidly turned over when surrounding H-2 positive cells are removed, in vivo NK cells will be continuously surrounded by H-2 positive cells and so a more stable presence of H-2 on NK cells could occur. Ly49A+ cells that have acquired H-2Dd showed reduced killing against MHC class I negative targets (43)
. One explanation of this observation is that the specific activating Ly49 receptor for H-2Dd could be also down-modulated on the uptake of H-2Dd. However, these data have to be reconciled with cold target inhibition experiments, where MHC class I expressed on target cells did not affect the NK cell cytotoxic activity (78)
. Whether the acquisition of specific H-2 has any impact on the capability of NK cells to be inhibited is not clear, since results obtained in different studies do not reach consistent conclusions. Sjostrom et al. (43)
found that Ly49A+ cells that have acquired H-2Dd were still equally inhibited by H2-Dd positive target cells, leading these authors to claim that expression of H2-Dd (endogenously encoded or acquired by surrounding cells) by the NK cell itself is still compatible with an Ly49/H2-Dd interaction in trans. However, Zimmer et al. (44)
reported a correlation between reduced H-2Dd ligand uptake by H2-Dd positive Ly49A+ NK cells and an inefficient inhibition of this H2-Dd positive effector by H-2Dd positive target cells. More recent experiments by these authors show that cis interaction between Ly49A and H-2-Dd restricts the number of Ly49A available for binding of H-2Dd on target cells (46)
. As a consequence, NK cell inhibition through Ly49A is reduced, and this allows NK cells to more easily detect induced self (detection of up-regulated NKG2D ligands). These data would support the idea that acquired H-2 alleles could bind in cis the Ly49 receptor and affect the capability of NK cells to be inhibited. In addition, since the cis association negatively influences the binding of all reagents to detect Ly49A, this phenomenon, instead of receptor down-modulation, can in fact explain the "receptor calibration" model (79)
. Whether similar processes operate in humans is unknown.
As described above, one consequence of cytotoxic interactions is the acquisition of target cell molecules. Although this process could help the final lethal hit, it can also have a negative outcome. Bidirectional transfer of NKG2D and MICB is correlated with a reduction in the NK cell cytotoxic function (19)
. Our data indicate that the entire MICB molecule is acquired by NK cells; moreover, it clearly colocalizes with NKG2D on the surface of the NK cells. Different hypotheses on how MICB transferred to the NK cell surface could affect NKG2D function are summarized in Fig. 4
. One possibility is that acquisition of MICB could induce the internalization and degradation of NKG2D. NKG2D-MICB clusters have not been observed in intracellular compartments in this study; however, a down-modulation of NKG2D parallels this process, giving some support to the receptor internalization and degradation hypothesis. Another possibility is that transferred MICB could stay on the NK cell surface and decrease the available interaction sites of NKG2D for MICB expressed in other cells. Synaptic acquisition of membrane-attached MICB then emerges as a tumor evasion strategy like the phenomenon described for sMICs or sULBPs (57
, 80
, 81)
. Although sMICB does not down-modulate NKG2D expression on NK cells, membrane attached MICB does (81)
. Thus, NKG2D ligands could affect cytotoxic effector cells by a decrease in ligand expression on target cells (as sMICs and sULBPs) and by down-modulation of functional NKG2D receptor expression (as sMICA and membrane-attached MICB). Acquisition of other molecules expressed in tumor cells, like the PVR, also induces down-modulation of the ligand (CD96) in NK cells (50)
and this could indicate that tumor evasion via synaptic acquisition of molecules by NK cells is a more general process.
|
In a nonpathological context, acquisition of MICB could serve to dampen the NK cell response after clearance of the infection. In this regard, "expression" of MICB by NK cells could induce fratricidal killing (Fig. 4)
, and could favor the recognition of NK cells by CTLs that also express NKG2D. Another hypothesis would be that receptor-ligand pairs such as NKG2D/MICB or CD96/PVR could be internalized and maintain a signaling necessary for NK cell activation inside the cell. Such is the case for the EGFR/epidermal growth factor (EGF) system (82)
. However, currently there are no data available supporting this hypothesis for NK cell receptors.
Why do NK cells leave receptors on target cells during cytotoxic interactions? It is tempting to speculate that the receptors acquired by target cells signal that these targets are already committed to die and should not be recognized again. This would be an efficient and dynamic way to control the NK cell response (Fig. 2)
, but it is not clear how the NK cell would recognize the transferred receptor molecules unless the process depended on simply masking the ligands for activating receptors by transferred receptors. On the other hand, receptors might be transferred as a consequence of the lethal hit after the release of lytic granule vesicles at synapse (see above and Fig. 3A
). This process could help the unidirectional delivery of lytic substances and explain why other bystander cells (not expressing activating NK cell receptor ligands) are not killed during the NK cell immune response.
|
PROSPECTS AND PREDICTIONS |
|---|
|
TOPABSTRACTBACKGROUNDRECENT RESULTSCONTROVERSY/UNANSWERED QUESTIONSPROSPECTS AND PREDICTIONSREFERENCES |
|---|
) at the target side could favor the detachment of membrane fragments, as these membrane domains contain a higher fusion temperature that makes them more resistant to mechanical stress (84)
. Fragment detachment along with fusion of membranes to form cytoplasmic bridges between cells would help the intercellular transfer of molecules embedded in the membrane in the right orientation. We predict that during the organization of the iNK-IS and cNK-IS, different molecular mechanisms such as secretion of vesicles or soluble protein fragments and direct extraction of proteins from the plasma membrane will operate during NK cell immunosurveillance.
The fate of the transferred receptor or ligand molecules and the physiopathologycal significance of this process needs to be carefully examined. In the case of MHC class I acquisition by NK cells at iNK-IS, the functional consequences are not clear, but in a site of infection, acquired MHC class I can protect NK cells from neighboring NK cells. Perhaps NK cells may also act as APCs to spread the processed antigen acquired from professional APCs (85)
. On the other hand, shed soluble receptors can induce receptor signaling in surrounding cells (86)
. In this regard, when receptors such as NKG2D localize with the ligand at the same membrane domains on target cells, they might trigger signaling of the ligand. This idea is supported by colocalization of transferred NKG2D with MICB (P. Roda-Navarro and H. T. Reyburn, unpublished observations). Cross-linking of GPI-anchored proteins has been described to trigger apoptotic signals (87)
. It will be interesting to test whether NK cell receptors ligands such as CD48 or ULBPs are able to trigger apoptotic signals upon binding of the transferred receptors.
The idea that acquisition of receptor ligands can impair the NK cell cytotoxic function has more experimental support. Experiments done in mouse models both in vivo and in vitro have shown how sustained exposure to NKG2D ligands impairs the cytotoxic function mediated by this receptor (88
, 89)
. The synaptic acquisition of membrane-attached MICB observed recently correlates with a reduction in the NK cell cytotoxic function, and this may represent a different mechanism of tumor immune evasion in addition to the effect of soluble ligands (80)
. Experiments currently under way in our laboratory show how interactions of just 1 h between IL-2 cultured human NK cells and MICB-expressing cells induce a marked decrease in the total amount of NKG2D (P. Roda-Navarro and H. T. Reyburn, unpublished observations). These data indicate that NKG2D ligands can rapidly affect the intracellular traffic of NKG2D and complement previously published data regarding the effect that sustained exposure (>24 h) to NKG2D ligands has during the activation of mouse NK cells (88)
. How the NKG2D/DAP10 receptor complex traffics in human NK cells during the NK cell immune response is an important question to address in order to understand how NKG2D ligands affect this process. Alternatively, in T and B cells, a temporal functional disconnection between receptors and signaling adaptor molecules after specific receptor stimulation has been described (90
, 91)
, and a similar process could be hypothesized to be involved in the loss of NK function observed in our experiments. Finally, NK cell fratricidal killing on NKG2D ligand acquisition should be more carefully tested.
Receptor transfer to target cells may also be responsible, at least in part, for the reduced NKG2D expression observed in NK cells after MICB expressing cells encounter (19)
. It is tempting to speculate that NK cells lose some functional receptor to help the specific delivery of lytic substances to target cells, but that process leads to a temporary and partial loss of NK function. It should be also investigated whether target cells that acquire activating receptors are indeed committed to die. Videomicroscopy approaches will help to follow the fate of cells that have acquired NK cell membrane fragments containing activating receptors.
Overall, immune synapses formed between lymphocytes and target cells offer a platform for the intercellular transport of proteins between cells. While these exogenous proteins can have an important impact on the immune response in physiological situations, this platform can also have a negative effect. The acquisition of molecules that decrease the immune response of the effector cell or the transfer of molecules involved in pathogen entry to the cells, since this fact favors the spread of the pathogen. Pathogens have even been observed to subvert the immune synapse for an efficient propagation between cells (92)
. Thus, synaptic interchange of molecules between cells is a phenomenon where multiple distinct physiological and pathological aspects of the immune response converge. A better understanding of the molecular mechanisms leading this process and the functional significance of this acquired material at the new cell will hopefully offer new strategies to understand immune responses.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication November 22, 2006. Accepted for publication January 11, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTBACKGROUNDRECENT RESULTSCONTROVERSY/UNANSWERED QUESTIONSPROSPECTS AND PREDICTIONSREFERENCES |
|---|
