The functional capacity of NK cells is dynamically tuned by integrated signals from inhibitory and activating cell surface receptors in a process termed NK cell education. However, the understanding of the cellular and molecular mechanisms behind this functional tuning is limited. In this study, we show that the expression of the adhesion molecule and activation receptor DNAX accessory molecule 1 (DNAM-1) correlates with the quantity and quality of the inhibitory input by HLA class I–specific killer cell Ig-like receptors and CD94/NKG2A as well as with the magnitude of functional responses. Upon target cell recognition, the conformational state of LFA-1 changed in educated NK cells, associated with rapid colocalization of both active LFA-1 and DNAM-1 at the immune synapse. Thus, the coordinated expression of LFA-1 and DNAM-1 is a central component of NK cell education and provides a potential mechanism for controlling cytotoxicity by functionally mature NK cells.
Natural killer cells are lymphocytes that belong to the innate immune system and have been implicated in the early control of viral infection and tumor immunity (1, 2). Challenging the notion that NK cells are short-lived lymphocytes with a static phenotype, a more complex picture is currently emerging suggesting that NK cells undergo discrete stages of differentiation and tune their responsiveness in a dynamic fashion (3). NK cells are functionally regulated by an array of germline-encoded receptors, including the stochastically expressed killer cell Ig-like receptors (KIRs) (4). Integration of receptor–ligand interactions determines the outcome of target cell recognition during the effector phase. Thus, inhibitory receptors abrogate early steps in the signaling pathways for activation (4). However, the same interactions are also involved in calibrating the intrinsic functional responsiveness of NK cells in a process referred to as NK cell education (or NK cell licensing) (5, 6). The strength of the inhibitory interaction determines the “educating impact” (the level of responsiveness to target cells). Together, the tight control of NK cell activation by multiple germline-encoded inhibitory receptors during target cell interactions and the dynamic tuning of the intrinsic functional capacity serve to maintain tolerance to self and allow recognition of cells with aberrant expression of HLA class I.
NK cell education determines the efficiency of the in vivo rejection of transplantable mouse tumors (7). In humans, immunogenetic studies suggest that KIR–HLA interactions, influencing the education of donor-derived NK cells, are important for the outcome in allogeneic stem cell transplantation (8). Despite its profound impact on NK cell functionality, we still have limited understanding of the cellular and molecular basis for NK cell education and its temporospatial regulation. Furthermore, there is no existing marker to distinguish educated and noneducated NK cells. Remarkably, phenotypic assessment and transcriptional profiling of educated and uneducated NK cells have revealed only few differences between the subsets (9, 10). Rather, the plasticity of NK cell function has been suggested to depend on upstream organization of receptors in the cell membrane (9). Thus, a unique feature of educated NK cells may be their confinement of activating receptors, including the natural cytotoxicity receptor NKp46, in nanodomain-like structures (9). Such membrane organization of activating receptors may influence signaling and, as a consequence, the ability of educated cells to form conjugates and engage in downstream functional responses (11).
An early and critical step for target cell conjugation and the formation of the immune synapse in cytotoxic lymphocytes is the inside-out signaling leading to a conformational change in LFA-1 to an open, active configuration (12–14). Active LFA-1 binds ICAM-1 and is essential for granule polarization and efficient killing of the target cell (15). In the absence of LFA-1/ICAM-1 interactions, NK cell activation leads to nonpolarized degranulation and limited killing. In mouse T cells, LFA-1 forms a functional pair with the coactivation/adhesion molecule DNAX accessory molecule 1 (DNAM-1), also known as CD226, PTA-1, and TLiSA-1, as they colocalize at the immune synapse (16). Intriguingly, DNAM-1 was one of few molecules with a distinct expression pattern in educated and uneducated NK cells in the original description of education in the human (10). Thus, DNAM-1 could serve as an important link between lymphocyte activation and the downstream engagement in functional immune synapses with target cells. In favor of this notion, DNAM-1 facilitates LFA-1–mediated T cell costimulation (17) and promotes the formation of stable target cell conjugates (18, 19). Similarly, triggering of DNAM-1 on human NK cells induces inside-out signals for LFA-1 (20), thus boosting conjugate formation. Furthermore, the impaired immunological control of chemically induced tumors in DNAM-1–deficient mice (21) was recently attributed to defects in conjugate formation (18).
In this study, we show that expression of DNAM-1 and conformational changes in LFA-1 are carefully coordinated in NK cells and correlate with the education and differentiation status of the cell. Thus, educated NK cells with a high intrinsic functional capacity display higher expression of DNAM-1 but also a more restricted and target cell–dependent regulation of the conformational state of LFA-1. Our results suggest that the coordinated regulation of DNAM-1 and LFA-1 and their colocalization at the immune synapse are essential components contributing to the heightened effector function of educated NK cells.
Materials and Methods
Human subjects and cell lines
This study was approved by the Regional Ethics Committee of Stockholm, Sweden.
PBMCs were isolated by density gravity centrifugation (Ficoll-Hypaque; GE Healthcare) from buffy coats of healthy donors. Staining for surface markers was performed on freshly isolated PBMCs, and functional assays of either PBMCs or purified NK cells (Miltenyi Biotec) were performed after overnight resting in complete medium (RPMI 1640 supplemented with 10% FBS, 5 mM l-glutamine, and 50 mM streptomycin/penicillin). The K562 and 721.221–wild-type (wt) cell lines were cultured in complete medium, whereas the 721.221 transfectants 0.221-PVR, 0.221–Nectin-2 (provided by Dr. O. Mandelboim, Hebrew University of Jerusalem), and 0.221-MICA (provided by Dr. K. Söderström, University of Oxford) were grown in complete medium supplemented with selection antibiotics (1.6 ml/ml Geneticin, Invitrogen). Transfection and maintenance of Drosophila Schneider line 2 (S2) cells were under control of the metallothionein promoter in plasmid pRmHa3 as described in detail elsewhere (15, 22). The S2 cells were induced to express the NK receptor ligands by adding 1 mM CuSO4 48 h prior to experiment.
Genomic DNA was isolated from 5 million PBMCs using the DNeasy blood and tissue kit (Qiagen). KIR ligands were determined using the KIR HLA ligand kit (Olerup SSP) for detecting the HLA-Bw4, -C1, and -C2 motifs.
Abs and flow cytometry
The following conjugated Abs were used: anti-CD57 (NK-1), anti-CD14 (M5E2), anti-CD19 (HIB19), anti-CD107a (H4A3), anti–IFN-γ (B27), anti-2B4 (2-69), anti-NKG2D (1D11), anti-ICAM1 (HA58), and anti-DNAM1 (Dx11) were obtained from BD Biosciences; anti-KIR2DL1/S1 (EB6), anti-KIRDL2/S2/L3 (GL183), anti-NKG2A (z199), anti-CD3 (UHCT1), and anti-CD56 (N901) were from Beckman Coulter; anti-DNAM1 (Tx25), anti-KIR3DL1 (Dx9), anti-CD57 (HCD57), anti-PVR (SKII.4), and anti–Nectin-2 (Tx31) were from BioLegend; anti-KIR2DL1 (143211), anti-KIR2DL3 (180701), anti-ULBP1 (170818), and anti-MICA (159227) were from R&D Systems; anti-CD48 (MEM-102) was from Exbio; anti-CD11a (25.3) was from Immunotech; and anti-NKp30 (AF29-4D12) was from Miltenyi Biotec. The anti-KIR3DL2 hybridoma Dx31 was provided by Dr. J. Philips (DNAX Research Institute, Palo Alto, CA). Purification and biotinylation of anti-KIR3DL2 was performed by Mabtech. The LFA-1 conformation-specific mAb anti-CD11/CD18 (mab24) was obtained from Hycult Biotech and was biotinylated with NHS-LC-biotin from Thermo Fisher according to standard procedures. Streptavidin Qdot605 (Invitrogen) was used to detect biotinylated mAbs. A Live/Dead fixable aqua dead cell stain kit (Invitrogen) was used to exclude dead cells from the analyses. For blockade of HLA class I, clone Dx17 (BD Biosciences) and clone A6131 (provided by Dr. D. Pende, Istituto Nazionale per la Ricerca sul Cancro, Genoa, Italy) were used in combination. DNAM-1 was blocked using the clone dx11 (BD Biosciences) and mouse IgG1 clone mopc-21 as control (BioLegend).
Data were acquired in FACSDiva software on a BD LSRFortessa equipped with a 488-nm laser, a 633-nm laser, a 405-nm laser, and a 562-nm laser. Anti-mouse IgGk beads (BD Biosciences) were stained with each of the fluorochrome-conjugated Abs separately and used as compensation controls. Fluorescence minus one stainings were used as controls to all receptor median fluorescence intensity measurements. Acquired data were analyzed in FlowJo 9.3 (Tree Star).
Functional flow cytometry assays
PBMCs or purified NK cells were mixed with target cells at a 10:1 or 1:1 ratio in U-bottom 96-well plates and incubated for 2 or 6 h at 37°C and 5% CO2. In blocking experiments, PBMCs were preincubated with 10 μg/ml indicated Ab for 30 min at room temperature. GolgiStop (1:1500, BD Biosciences) and GolgiPlug (1:1000, BD Biosciences) were added after 1 h of coincubation. At the end of the assay, the cells were stained for CD107a and the open conformation of LFA-1 as well as additional surface markers to identify subsets of NK cells. This was followed by fixation and permeabilization (Fix/Perm kit, eBioscience) and subsequent intracellular staining for IFN-γ.
PBMCs were plated in U-bottom 96-well plates, stimulated with IFN-α (10 ng/ml), IL-2 (500 U/ml), IL-12 (10 ng/ml), IL-15 (10 ng/ml), IL-18 (100 ng/ml), or IL-21 (20 ng/ml). and incubated at 37°C and 5% CO2 overnight.
Conjugation assay for confocal microscopy and ImageStream analysis
NK cells were labeled with CellTrace Violet and the membrane dye DiO or calcein-AM (Invitrogen), mixed with target cells at a 1:1 ratio, incubated for 30 min at 37°C, 5% CO2, followed by Ab staining for ImageStream analysis. Conjugates for confocal analysis were first fixed in 4% paraformaldehyde and then stained with LFA-1open (mAb24, Hycult Biotech) or DNAM-1 (Tx25, BioLegend) and secondary goat anti-mouse Alexa Fluor 546 (Molecular Probes). Conjugates were also stained with phalloidin–Abberior Star 635 (Aberrior) in the final staining step to visualize F-actin. Ab stainings were performed in 4°C for 1 h and each step was followed by washing/blocking with PBS supplemented with 10% goat serum.
NK cell–target cell conjugates were imaged using a ×63/1.4 numerical aperture oil objective in an LSM 510 confocal microscope (Carl Zeiss). Conjugates between NK and target cells were selected for imaging based on having a flattened intercellular contact. The fluorescence intensity images from LFA-1 and DNAM-1 were corrected for background fluorescence by reducing the intensity value in each pixel by a mean value found in a region over the target cell involved in the conjugate. Pixels with intensity values below the background intensity were set to 0 to avoid negative intensities. Quantification of the fluorescence intensity inside and outside the synapse was done by manually drawn mask in the ImageJ software guided by the fluorescence signal from either F-actin or DiO. Conjugates with background fluorescence ≥25% of that measured in the synapse were excluded from the analysis.
Samples were acquired at ×40 magnification on a four-laser, 12-channel, ASSIST-calibrated ImageStreamX (Amnis, Seattle, WA) imaging flow cytometer. Laser excitations of 405 (125 mW), 488 (100 mW), 561 (200 mW), and 658 nm (120 mW) were set to avoid saturation in any of the spectral channels. Single-stained controls were collected for generation of a compensation matrix to correct for spectral crosstalk. The NK cell mask was defined based on the CellTrace Violet fluorescent signal and used to extrapolate the target cell mask. Selecting images with a low NK cell mask aspect ratio was done to identify images containing one NK cell (23) By selecting images with a low target cell mask aspect ratio as well as a low NK cell mask aspect ratio, images of conjugates with one NK cell and one target cell were identified. Using the “interface” masking function in IDEAS (23) the NK cell synapse was defined in all of the images. Polarization of DNAM-1 and LFA-1 toward the synapse was quantified by calculating the fluorescent signal in the synapse as a ratio to the fluorescent signal of the whole NK cell, where values above >1 indicate polarization toward the synapse. To adjust for cell-to-cell variation, the signals were normalized for area of the respective masks. A threshold polarization of >1.25 was set based on the polarization observed in conjugates with unmodified target cells (S2-WT), defined as background.
Statistical analyses were performed with GraphPad Prism software 5.0 and 6.0. For comparisons of matched groups, a Friedman test followed by a Dunn multiple comparison test or Wilcoxon matched test were used. A Mann–Whitney t test was used for comparisons of unpaired groups, and correlations were tested using a Spearman correlation test. A p value <0.05 was considered statistically significant.
DNAM-1 expression correlates with expression of educating NK cell receptors
We examined the surface expression of an array of adhesion molecules and activation receptors on human NK cells relative to expression of inhibitory KIRs and CD94/NKG2A. Strikingly, expression of DNAM-1, but not other key activating receptors and adhesion molecules, correlated with the presence of educating self-specific inhibitory receptors (Fig. 1A, 1B). Thus, the level of DNAM-1 expression was higher in KIR2DL3 single-positive (2DL3-SP) NK cells compared with 2DL1-SP NK cells in donors harboring the HLA-C1 (C1+) ligand but not the HLA-C2 (C2+) ligand. Conversely, the expression of DNAM-1 was higher in 2DL1-SP and 3DL1-SP NK cells in C2+, Bw4+ donors. Human NK cells are also functionally tuned by NKG2A/HLA-E interactions (24, 25). Hence, corroborating the link to education, NKG2A+ NK cells also expressed higher levels of DNAM-1 compared with uneducated NK cells. Furthermore, the level of DNAM-1 expression correlated quantitatively with the number of educating receptor–ligand interactions (Fig. 1C). In line with the detuned NK cell functionality mediated by KIR2DS1 (26), we detected a low DNAM-1 expression on 2DS1-SP cells in C2+ donors (Fig. 1D). In agreement with the original description of NK cell education in the human (10), none of the other markers showed a systematic correlation to education, albeit NKG2D was slightly reduced, and the closed configuration of LFA-1 was slightly increased on educated NK cells. NKp30 was highly expressed on NKG2A+ NK cells, which is in line with the relatively less differentiated phenotype of NKG2A+ NK cells and the previously shown downregulation of NKp30 during NK cell differentiation (27).
In addition to education, late stage differentiation of CD56dim NK cells is associated with a functional reprogramming of the cells primarily manifested as an increased ability to perform Ab-dependent cellular cytotoxicity and a gradually declining responsiveness to cytokines (27, 28). Therefore, we also assessed the potential impact of NK cell differentiation on the expression of DNAM-1. Intriguingly, NK cell differentiation, defined in this study by CD57 expression, was associated with increased DNAM-1 expression in uneducated NK cell subsets and a further increase in educated subsets (Fig. 1E). When controlling for the effect of differentiation by limiting the analysis to CD57− cells, we still observed a higher DNAM-1 expression level in NK cells educated by KIRs or NKG2A (Fig. 1F, 1G). Therefore, these data reveal that DNAM-1 expression is a marker of education, with expression reflecting an additive input of education by different receptors.
DNAM-1 is an intrinsic marker of NK cell education
Because many experimental platforms used to quantitatively assess NK cell education, including stimulation by K562 cells, involve triggering via DNAM-1 receptors, the high expression of DNAM-1 could potentially explain why educated NK cells respond better to target cells in such assays. To examine this possibility, we plotted degranulation as well as IFN-γ production in resting NK cells as a function of DNAM-1 expression following stimulation with target cells with and without expression of the DNAM-1 ligands PVR and Nectin-2 (29). Corroborating the phenotypic link to education, these experiments revealed a striking correlation between DNAM-1 expression and NK cell function (Fig. 2). Moreover, the correlation between DNAM-1 expression and functional responses also applied for stimulations with target cells that lacked known DNAM-1 ligands such as 721.221-wt and 721.221 cells transfected with MICA (Fig. 2, Supplemental Fig. 1), albeit the amplitude of the response were generally lower.
In T cells, DNAM-1 can form a functional pair with LFA-1 at the immune synapse where the two receptors seem to partly depend on each other for optimal signaling (17, 30–32). Thus, the correlation between NK cell responses and DNAM-1 expression against targets that lacked DNAM-1 ligands could theoretically be explained by DNAM-1–mediated facilitation of LFA-1/ICAM-1 interactions. To examine this possibility, overnight IL-2–activated NK cells were coincubated with Drosophila S2 cells transfected with a combination of ligands for human NK cell receptors (15, 22). Notably, S2 cells transfected with CD48 and ULBP-1, ligands for 2B4 and NKG2D, with and without ICAM-1 triggered strong degranulation responses and IFN-γ production, which correlated with the expression of DNAM-1 regardless of whether ligands for DNAM-1 or LFA-1 were expressed (Fig. 2D, 2E).
Finally, blocking experiments corroborated the role of DNAM-1 in target cell recognition by educated NK cells, albeit background degranulation by uneducated NK cells was also partly dependent on DNAM-1 (Fig. 2F).
These results conclusively show that DNAM-1 serves as an intrinsic, readout-independent marker for NK cell education. Furthermore, because many tumor types overexpress PVR (17–20), differential expression of DNAM-1 in NK cells may also be critically involved in regulating the function of educated NK cells during the effector phase.
Tight control of the conformational state of LFA-1 in educated NK cells
Given the biochemical interaction between LFA-1 and DNAM-1, we next focused our attention on the conformational state of LFA-1 in educated NK cells in the absence or presence of cytokines and following target cell stimulation. Incubation of NK cells overnight in the absence of stimulation resulted in a conformational change of LFA-1 in a fraction of the cells into the active open form (LFA-1open). This effect was boosted by cytokines (Fig. 3A, 3B) and was significantly more pronounced in uneducated cells compared with educated NK cells (Fig. 3C). Thus, educated NK cells have a more stringent control of LFA-1open in response to exogenous cytokines. We speculated that the absence of conformational change in LFA-1 on educated NK cells was due to inhibition via self-specific KIRs or NKG2A interacting with HLA class I on neighboring NK cells. Indeed, blockade of HLA class I led to a similar fraction of cells expressing an active LFA-1 conformation in educated and uneducated NK cells (Fig. 3D).
Notably, upon stimulation with target cells, the situation was reversed and the shift to the open conformation was more pronounced in educated NK cells (Fig. 3A, 3C). The differential regulation of LFA-1open in educated and uneducated NK cells may provide a mechanism to ensure NK cell tolerance under inflammatory conditions and yet strong responsiveness to cells with aberrant expression of HLA class I.
Recruitment of DNAM-1 and LFA-1open at the immune synapse
LFA-1/ICAM-1 interactions are essential for polarization of cytolytic granules during NK cell–target interactions and necessary for a successful immunological response (14). In mouse T cells, stimulation via the TCR leads to inside-out signaling and a conformational shift of LFA-1 followed by association of LFA-1 and DNAM-1 in lipid rafts at the immune synapse (32). To investigate the dynamics of such events in NK cells, we monitored the cellular localization of LFA-1open and DNAM-1 in resting human NK cells following coincubation with S2 cells expressing distinct combinations of NK receptor ligands. The polarization of DNAM-1 and LFA-1open during synapse formation was quantified by measuring their mean fluorescence intensity inside and outside the synapse. Coincubation of NK cells with S2 cells transfected with ICAM-1, ULBP-1, and CD48 led to a pronounced accumulation of LFA-1open at the immune synapse (Fig. 4A, 4C). Notably, DNAM-1 was partially recruited to the target cell interface even in the absence of known DNAM-1 ligands (Fig. 4A, 4C). In contrast, coincubation with S2 cells transfected with ULBP-1 and CD48, which triggered robust degranulation responses, did not lead to polarization of either LFA-1open or DNAM-1 (data not shown). These results suggest that LFA-1 engagement and subsequent inside-out signaling can assist in recruiting DNAM-1 to the immune synapse. However, polarization of DNAM-1 and LFA-1open was more pronounced following NK cell incubation with K562 cells expressing high levels of DNAM-1 ligands (Fig. 4B, 4C), suggesting that these play an active role in recruiting DNAM-1 to the immune synapse.
To further corroborate these findings, we quantified the polarization of DNAM-1 and LFA-1open at the NK cell immune synapse using the high-throughput imaging flow cytometer, ImageStreamX (Fig. 5A). Using carefully defined masks to identify NK–target cell conjugates, receptor polarization toward the immune synapse was assessed in an automated and unbiased fashion (Supplemental Fig. 2). The experiments on the ImageStream platform further supported the notion that outside-in signaling via activating receptors and LFA-1 led to recruitment of DNAM-1 to the immune synapse and that this was further enhanced in the presence of DNAM-1 ligands (K562 cells) (Fig. 5B). The coordinated regulation of DNAM-1 and the conformational state of LFA-1 may serve as a tunable quantitative checkpoint for the effector function of educated and differentiated NK cells with high cytolytic potential.
Understanding the cellular and molecular mechanisms governing NK cell education represents one of the major challenges in immunology. In line with this notion, in this study we provide evidence that NK cell education is strongly linked to the dynamic and coordinated expression of the adhesion/activation molecule DNAM-1 and conformational changes in LFA-1. We show that DNAM-1 serves as a unique intrinsic sensor of the cytolytic potential of human NK cells. Upon target cell interaction, DNAM-1 rapidly colocalizes with LFA-1open at the immune synapse. Given the essential role for LFA-1/ICAM-1 interactions in the early activation of human NK cells, our findings provide important mechanistic insights into how human NK cells are functionally regulated during NK cell education.
NK cell education is a tunable process and the responsiveness of NK cells reflects the strength of the net inhibitory signals (7, 33, 34). Thus, discrete subsets of NK cells expressing distinct combinations of inhibitory receptors mount functional responses ranging from weak/absent to high upon stimulation with target cells lacking the cognate MHC class I molecules or upon cross-linking of activating NK cell receptors. Differences in the functional potential at the single cell level translate into variability in the overall responsiveness of NK cells in each individual and are determined by the composition of the NK cell repertoire in combination with the HLA background (34, 35). NK cell repertoire diversity and global NK cell responsiveness, as predicted by KIR/HLA genetics, have been implicated in a number of clinical settings, including outcomes of allogeneic stem cell transplantation (8, 36, 37), infectious diseases (38), autoimmunity, and disorders of pregnancy (1). Thus, elucidating the mechanisms underlying NK cell education is important and may provide means to manipulate the responsiveness of NK cells to alleviate autoimmunity and boost antitumor immunity.
In this study, we report that the expression of DNAM-1, which is a coactivating receptor and adhesion molecule, correlates with the expression of self-specific inhibitory KIR and NKG2A as well as the functional tuning by multiple inhibitory receptor–ligand interactions. In line with numerous previous reports (10), none of the other assessed NK cell receptors displayed a similar positive correlation with NK cell education. DNAM-1 is crucial for the NK cell recognition of several types of primary human tumors, including ovarian carcinoma, melanoma, acute leukemia, myelodysplastic syndromes, and neuroblastoma (39–43). The increased expression of DNAM-1 by educated NK cells points to a specific role of this subset in surveillance against such tumor types. More recently, DNAM-1 was found to be essential for the differentiation of memory NK cells during mouse CMV infection (44). Thus, our data indicate a possible link between the phenotype of the educated NK cell and the downstream adaptive behavior of the subset. Notably, DNAM-1 expression was highest on the subset with a terminally differentiated memory-like phenotype.
Our understanding of the biological basis behind NK cell education is tightly linked to the assays used to monitor the functional responsiveness of the cell, including the type and length of stimulation as well as the downstream readout (6). In this respect, most work in the human has been based on stimulation with K562 cells. Because K562 cells express the DNAM-1 ligands PVR and Nectin-2, the increased DNAM-1 expression on educated NK cell subsets could potentially explain their heightened responsiveness in vitro. Although we considered this possibility to be rather unlikely, because experiments using plate-bound Abs have demonstrated clear differences between educated and uneducated NK cells in the mouse (7), we went on to examine the intrinsic responsiveness of NK cells as a function of DNAM-1 expression following stimulation by target cells lacking DNAM-1 ligands. Indeed, these experiments revealed a correlation between DNAM-1 expression and the functional response of the NK cell subset regardless of the target cell used. By stimulating NK cells with Drosophila S2 cells transfected with CD48 and ULBP-1, ligands for 2B4 and NKG2D, respectively, we could demonstrate a correlation between DNAM-1 expression levels and function in NK cells stimulated without LFA-1 and DNAM-1 ligands. This was important, because ICAM-1 is expressed on all other cell lines used to read-out NK cell education, and ligation of LFA-1 involves signaling via DNAM-1 (data not shown) (32). Taken together, these experiments corroborate the view that education is linked to an intrinsic potential of the cell and reveal that DNAM-1 can serve as a marker for this functional potential.
The conformational change of LFA-1 into the open active state is essential for lymphocyte adhesion to target cells (15). This conformational change represents an early step in NK and T cell activation, occurring within minutes of receptor ligation and at levels below those required to trigger degranulation and release of cytokines (45). Heterogeneity in LFA-1open expression patterns within resting and activated NK cell populations have been observed (45) and linked to expression of inhibitory self-receptors (11). However, it remained unclear whether particular NK cell subsets are more prone to conformational changes following different modes of NK cell activation. Upon IL-2 stimulation of PBMCs in the absence of target cells, uneducated but not educated cells changed to the open conformation of LFA-1 at the cell surface. This difference was abolished by blocking HLA class I, suggesting that educated NK cells are kept in check and prohibited from shifting to the LFA-1open conformation by their expression of self-specific inhibitory KIR or NKG2A. In reductionist systems, engagement of NKG2A has been shown to control inside-out signals from multiple coactivating NK cell receptors for LFA-1 conformational changes (20). Such tight regulation of LFA-1 in educated NK cells may be important to avoid unspecific conjugate formation and targeting of normal tissues by educated NK cells in the context of inflammation. In contrast, corroborating the work of Thomas et al. (11), conformational changes in LFA-1 were more pronounced in educated NK cells following stimulation with target cells.
By using Drosophila cells transfected with individual ligands for NK cell receptors and combinations thereof, we could probe the dynamics of DNAM-1 and LFA-1open mobilization to the immune synapse under distinct conditions in resting human NK cells. In line with the previously described uncoupling of signals for degranulation and granule polarization (15), polarization of LFA-1open at the immune synapse was observed only when NK cells were stimulated by S2 cells transfected with ICAM-1. Notably, stimulation of NK cells with S2 cells transfected with CD48 and ULBP-1, ligands known to trigger robust degranulation responses (20), in combination with the adhesion molecule ICAM-1, which triggers granule polarization, was associated with recruitment of DNAM-1 to the immune synapse. The recruitment of DNAM-1 to the immune synapse in the absence of cognate DNAM-1 ligands suggested that LFA-1–mediated outside-in signaling facilitates polarization of DNAM-1 to the target cell interface. This interpretation is supported by observations in mouse T cells that have conserved tyrosine residues in the cytoplasmic tail of DNAM-1 (32). Hence, the physical association of DNAM-1 and LFA-1 was shown to be dependent on phosphorylation of DNAM-1 at residue 329 (46), which could be achieved independently of LFA-1 by triggering T cells with anti-CD3 and anti-CD28 (32). In CD4 T cells from LFA-1–deficient mice, TcR-mediated stimulation failed to recruit DNAM-1 to the immune synapse, suggesting that this process requires a physical interaction between DNAM-1 and LFA-1 (32).
Ligation of several different activating NK cell receptors, including DNAM-1, as well as LFA-1 itself, has been shown to induce a conformational change of LFA-1 in human NK cells (47–50). Given the high expression of DNAM-1 in educated and differentiated NK cells, such DNAM-1–mediated inside-out signaling via LFA-1 may constitute a major checkpoint in the regulation of target cell conjugation and thereby affect NK cell cytotoxicity.
In this study, we reveal that DNAM-1 serves as an intrinsic marker of the cytolytic potential of educated NK cells. Although our data point to the coordinated expression of DNAM-1 and LFA-1 as a central component of NK cell education, several challenges remain. It will be important to carefully assess the temporal and spatial coordination of DNAM-1/LFA-1 in the immune synapse, their association with signaling molecules, and the consequences for the downstream signaling in educated and uneducated NK cells. Furthermore, it represents a major challenge to define the molecular pathways that link the strength of the inhibitory input via KIR and NKG2A to the expression of DNAM-1. Decoding such pathways might provide unique insights into the molecular basis behind the tunable intrinsic cytolytic potential of educated NK cells.
This work was supported by grants from the Swedish Research Council, the Swedish Children’s Cancer Society, the Swedish Cancer Society, the Karolinska Institutet, the Wenner-Gren Foundation, the Anders Jahres Foundation, the Norwegian Cancer Society, the Norwegian Research Council, the South-Eastern Norway Regional Health Authority, the European Research Council, and the K.G. Jebsen Center for Cancer Immunotherapy.
The online version of this article contains supplemental material.
Abbreviations used in this article:
DNAX accessory molecule 1
killer cell Ig-like receptor
Schneider line 2
The authors have no financial conflicts of interest.