The integrin LFA-1 is essential for efficient activation and for cytotoxicity of NK cells because it initiates the assembly of the immunological synapse and mediates firm adhesion to the target. LFA-1 is also needed to polarize the cytotoxic machinery of the NK cell toward the target cell. The binding affinity and avidity of integrins can be regulated via inside-out signals from other receptors. In this article, we investigate the signals necessary to activate LFA-1 in human NK cells. Our data show that LFA-1 has a low ligand-binding activity in resting human NK cells, but it can be stimulated by triggering activating receptors, such as 2B4 or CD16, or by coactivation of different receptor combinations. Short-term stimulation of freshly isolated NK cells with cytokines, such as IL-15, IL-12, or IL-18, does not activate LFA-1 but increases the responsiveness of the cells to subsequent receptor stimulation. Different NK cell subsets vary in their ability to induce LFA-1 binding activity after activating receptor stimulation. Interestingly, the NK cell subsets that are more mature and possess higher cytotoxic potential also show the highest activation of LFA-1, which correlated with the expression of the small calcium-binding protein S100A4. Our data suggest that regulation of LFA-1 is one reason for the different activity of NK cells during differentiation.
This article is featured in In This Issue, p.1763
Natural killer cells are innate lymphocytes that participate in the first line of defense against transformed or infected cells (1). Mature NK cells can be subdivided into different groups with distinct functional properties. The CD56bright subset is considered the most responsive to cytokines, has a high expression of the inhibitory NKG2A/CD94 receptor complex, and shows the highest cytokine production. CD56dim NK cells are mostly positive for CD16, express more killer cell Ig-like receptor (KIRs) and higher amounts of perforin and granzymes than does the CD56bright subset, and have a higher cytotoxic potential. Among the CD56dim subset, the cells expressing the marker CD57 have the highest maturation state (2). It was reported that these cells react strongly to activating receptor stimulation and are highly cytotoxic (3).
A fine-tuned panel of activating and inhibitory receptors regulates the activation of NK cells. For resting NK cells, triggering of at least two activating receptors is essential for efficient cytotoxicity and cytokine release (4). The Fc receptor CD16, which is responsible for triggering Ab-dependent cellular cytotoxicity (ADCC), is the only exception. CD16 is able to induce effector functions without additional stimuli. Specific receptor combinations that result in a synergistic increase in cytotoxicity or cytokine release are called coactivating receptors (5). Coactivation can also be observed with IL-2–expanded NK cells (6). Inhibitory receptors, such as KIR and NKG2A/CD94, recruit the tyrosine phosphatase SHP-1 to their cytoplasmic ITIM sequence to interrupt activating signaling events. Inhibitory receptors mainly interfere with signaling pathways, leading to the polarization of the NK cell (7).
Conjugate formation between NK and target cells is an important step that leads to NK cell cytotoxicity (8). The integrin LFA-1 is essential for stable contact with the target (9). LFA-1 is a heterodimer consisting of αL (CD11a) and β2 (CD18) chains. A unique property of integrins is that their adhesiveness can be regulated (10). Depending on the conformation of LFA-1 and its arrangement in the membrane (clustering), it has different binding abilities to its ligands, the ICAMs. LFA-1 can change among the bent, the closed/extended, and the open/extended conformation that have low, intermediate, and high affinity, respectively, for ICAM. Divalent cations are indispensable for LFA-1 function. Binding of Mg2+ or Mn2+ to the metal ion–dependent adhesion site, a region near the N terminus, induces a conformational change into the high-affinity state, whereas Ca2+ stabilizes LFA-1 in the low-affinity conformation (11, 12). Changes in the lateral mobility of LFA-1 in the membrane can cause substantial relocalization of the receptor. Clusters of the receptor have a higher avidity; therefore, LFA-1 can bind ICAM more efficiently, even if it is not in the high-affinity conformation.
Modifications in the affinity and avidity of LFA-1 can be caused by inside-out signals from other receptors, and various activating NK cell receptors were shown to activate LFA-1 (13, 14). However, which signaling events initiated by activating receptors lead to changes in integrin conformation and mobility is only partially understood. In addition to mediating the tight adhesion to target cells, LFA-1–mediated signals are important for the polarization of lytic granules toward the target cell (15, 16) and, thereby, contribute to efficient NK cell cytotoxicity.
We used a ligand complex–based adhesion assay (LC-AA) (17) to analyze the signals that are involved in the activation of LFA-1 on NK cells. Using ICAM-1–Fc complexes, the LC-AA evaluates the overall binding capacity of LFA-1, the affinity, and the avidity. Our data show that activation of LFA-1 is a highly regulated process that is influenced by inside-out signaling of different (co)activating receptors, cytokine signaling, and NK cell differentiation.
Materials and Methods
PBMCs were purified by density centrifugation over LSM solution (PAN Biotec). NK cells were isolated using the NK cell negative isolation kit (Dynal; Invitrogen). PBMCs and resting NK cells were kept in NKpop medium at 1.5–3 × 106 cells per milliliter (IMDM with GlutaMAX, 10% FCS, 1% Penicillin/Streptomycin; all from Life Technologies). All HEK 293T cells were cultured in DMEM with 10% FCS and 1% Penicillin/Streptomycin, with additional selection for the transfectants (CD48: 50 μg/ml Geneticin; ULBP2: 1 μg/ml puromycin). K562 cells were cultured in IMDM with 10% FCS and 1% Penicillin/Streptomycin. For some experiments, resting NK cells were preincubated in NKpop medium with the following cytokines for the indicated times: 5 ng/ml IL-12, 5 ng/ml IL-15 (both from PAN Biotec), and 25 ng/ml IL-18 (MBL).
The LC-AA (17) was performed in a modified version. Basic buffer for all incubations is PBS containing 0.5% BSA, with or without cations (1 mM CaCl2 and 2 mM MgCl2). The ICAM-1–Fc complexes were prepared by mixing 50 μg/ml recombinant human ICAM-1–Fc chimera (R&D Systems) and F(ab)2 fragments of goat anti-human Fcγ fragment specific (40 μg/ml PE labeled or 160 μg/ml FITC labeled; Jackson ImmunoResearch) in buffer without cations for ≥30 min at room temperature (RT). As a negative control, CD99–Fc (18) was used instead of ICAM-1–Fc. Purified NK cells or PBMCs were incubated for 10 min with 1 μg/ml of the appropriate primary Abs: IgG (MOPC21; Sigma), NKp30 (p30-15; BioLegend), NKp46 (BAB281; Beckman Coulter), NKG2D (149810; R&D Systems), DNAM-1 (DX11; BD), 2B4 (C1.7; Beckman Coulter), NTB-A (NT-7) (19), and CD16 (3G8; BioLegend). After washing, cells were resuspended in buffer and ICAM-1–Fc complexes were added (dilution 1:20), the cells were transferred to a 37°C water bath, and goat anti-mouse was added to a final concentration of 5 μg/ml for Ab cross-linking. Stimulation controls were performed by adding PMA (10 nM) or Mg2+/EGTA (10 mM/1 mM) instead of Abs. Unless stated otherwise, cells were fixed after 10 min by adding paraformaldehyde to a final concentration of 2%. When HEK 293T cells were used for stimulation, they were coincubated at an E:T ratio of 1:1 for 30 min at 37°C. Where indicated, the cells were preincubated for 30 min at 37°C with inhibitors (10 μM PP1, 100 nM wortmannin, 2.5 μM U73122, 25 μM SB202190, 25 μM PD98059 [all from Biomol] or 20 μM trifluoperazine [TFP; Sigma]), and these inhibitors were kept in the medium during stimulation. In some experiments, the cells were also stained with anti-KIR2DL2/3–PE (GL183; Beckman Coulter), NKG2A-PE (Z199; Beckman Coulter), CD56–Brilliant Violet 421 (NCAM16.2; BD), CD3-PerCP (UCHT1; BioLegend), or CD57–Alexa Fluor 647 (HCD57; BioLegend). Data were acquired on a BD FACSCalibur or LSRFortessa and analyzed with FlowJo software (TreeStar). All overlaid graphs are normalized to equal peak height.
Stimulation and staining with mAb 24
Purified NK cells were resuspended in HEPES/NaCl buffer (20) containing 10 mM HEPES, 0.14 mM NaCl, 0.5% BSA, 1 mM CaCl2, and 2 mM MgCl2. mAb 24 or an IgG control Ab (MOPC21; Sigma) was added to a final concentration of 2 μg/ml, and cells were transferred to a 37°C water bath. Stimulation with PMA, Mg2+/EGTA, or by coincubation with HEK 293T cells was performed as described for the LC-AA. After fixation, cells were stained with PE-labeled goat anti-mouse (BioLegend).
To identify NK cells and NK cells in conjugates, PBMCs were stained with anti-CD3–PerCP and anti-CD56–Brilliant Violet 421 and for some experiments preincubated for 30 min at 37°C with 20 μM TFP or with DMSO as solvent control. The inhibitor concentration was kept constant throughout the assay. Then cells were washed and mixed at an E:T ratio of 1.5:1 with CellTracker Orange CMRA (Molecular Probes)–stained K562 or K562-CD48. The samples were centrifuged for 1 min at 300 × g, incubated at 37°C for the indicated times, mixed thoroughly by vortexing, and fixed with a final concentration of 2% paraformaldehyde.
Staining of S100A4
For analysis by flow cytometry, PBMCs were stained with anti-CD3–PerCP, anti-CD56–Brilliant Violet 421, and CD57–Alexa Fluor 647, fixed with a final concentration of 2% paraformaldehyde, permeabilized using BD FACS Permeabilizing Solution 2, and stained with rabbit anti-S100A4 [clone EPR2761(2); GeneTex] 1:400 or rabbit control Ab for 20 min at RT, followed by donkey anti-rabbit–PE (BioLegend). For microscopic analysis, resting NK cells were incubated for 30 min at 37°C with 20 μM TFP or with DMSO as solvent control. Then, the NK cells were mixed with 0.5 μM CFSE (Molecular Probes)–stained K562 at a 1:1 ratio, centrifuged for 1 min at 300 × g, incubated at 37°C for 20 min, mixed thoroughly by vortexing, fixed and permeabilized as before, and stained with Alexa Fluor 647–labeled rabbit anti S100A4 [clone EPR2761(2); Abcam] 1:100 at RT. The cells were mounted on coverslips using ProLong Diamond Antifade (Molecular Probes) and analyzed using an Evos FL Auto Cell Imaging System. Microscopic pictures were edited using ImageJ, displaying the S100A4 staining with the “fire” pseudocolor scale.
Cytotoxicity was assessed by [51Cr]-release assay, as described previously (21). Fresh NK cells were incubated at an E:T ratio of 4:1 with 51Cr-labeled K562 for 4 h in the presence of inhibitor or DMSO as solvent control.
Cell lysis, SDS-PAGE, and Western blotting were performed as described previously (22). The membrane was cut at the 25-kDa marker band; the lower part was incubated with rabbit anti-S100A4 [clone EPR2761(2); GeneTex] 1:2,000, and the upper part was incubated with rabbit anti-actin (Sigma) 1:800, followed by anti-rabbit HRP (CST) 1:20,000.
To test the binding activity of LFA-1 in primary human NK cells, we performed an LC-AA (using soluble ICAM-1–Fc complexes) or we used the conformation-specific Ab mAb 24, which only binds to LFA-1 in its high-affinity state (23). Both reagents were tested for their staining and specificity on resting, freshly isolated human NK cells. In NK cells from many donors, we detected only very weak staining with these reagents (Fig. 1A, Supplemental Fig. 1), demonstrating that LFA-1 is in its low-affinity, low-avidity state in these cells. However, in some donors, we detected staining with mAb 24 on freshly isolated NK cells, suggesting some level of LFA-1 activity that might be due to NK cell activation in vivo (Supplemental Fig. 1B). Treating cells with PMA increases the avidity of LFA-1 without affecting its affinity (24). Incubating resting human NK cells with PMA resulted in an intermediate staining in the LC-AA, whereas minimal change compared with unstimulated cells was detectable with mAb 24 (Supplemental Fig. 1A). This confirms that the LC-AA can detect avidity changes of LFA-1. In contrast, divalent cations have a strong influence on the conformation of LFA-1, and high affinity can be induced by addition of Mg2+, whereas Ca2+ is chelated with EGTA (11). As expected, NK cells showed a homogeneous strong staining in the LC-AA and with mAb 24 after incubation with Mg2+/EGTA (Supplemental Fig. 1A). This confirms that, although mAb 24 can detect affinity changes in LFA-1, the LC-AA can be used to detect affinity and avidity changes in LFA-1.
Coactivation of distinct pairs of activating receptors is essential for the induction of cytotoxicity and cytokine production by resting human NK cells (5). Therefore, we wanted to assess how the stimulation of different activating receptors influences the binding activity of LFA-1. Stimulation of resting human NK cells via cross-linking of NKp30, NKp46, NKG2D, or DNAM-1 resulted in only very weak activation of LFA-1, as detected by LC-AA (Fig. 1A). Interestingly, engagement of 2B4 was sufficient to induce substantial LFA-1 activation. 2B4 belongs to the SLAM family of receptors (25, 26). Stimulation of NTB-A, another member of this receptor family, induced weaker, but still considerable, LFA-1 activation, demonstrating that this receptor family is particularly good at inducing LFA-1 binding activity in resting human NK cells. To test for the effect of coactivation on LFA-1 binding activity, we performed pairwise stimulation of different activating receptor combinations. Stimulating 2B4 in combination with NKp30, NKp46, or DNAM-1 merely had an additive effect on LFA-1 compared with the single stimulations (Fig. 1A). Coengagement of 2B4 and NKG2D, which was shown to be very effective at coactivating resting NK cells (5), resulted in a synergistic increase in LFA-1 activation. A weaker, but similar, effect was seen when we combined NTB-A with NKG2D. This demonstrates that coactivation of resting human NK cells affects the binding activity of LFA-1.
Activation of LFA-1 after cross-linking of activating receptors occurred very rapidly and peaked after ∼10 min. The kinetics were similar for different activating receptors and their combinations (Fig. 1B, 1C), and the observed synergistic effect of 2B4 and NKG2D was not due to changes in the reaction speed of the cells. Also, the number of cells reacting to stimulation did not change dramatically by coactivation (Fig. 1B), but the cells showed a much higher signal intensity (mean fluorescence intensity [MFI]) in the LC-AA (Fig. 1C). This indicates that coactivation quantitatively activates more LFA-1 molecules per cell, or it induces a higher ligand-binding activity by affinity and/or avidity changes in LFA-1.
To test for changes in LFA-1–binding activity after stimulation of activating receptors by target cell–expressed ligands, we used HEK 293T cells that were transfected to overexpress the ligands for 2B4 (CD48), NKG2D (ULBP2), or a combination of both. In this assay, we detected LFA-1 affinity changes by staining NK cells with mAb 24, and we measured LFA-1 affinity and avidity changes using the LC-AA. Incubation of resting human NK cells with nontransfected HEK 293T cells did not cause any significant activation of LFA-1 (Fig. 2A). Similar to receptor stimulation with Abs, 2B4 stimulation by CD48-expressing cells induced a stronger activation of LFA-1 compared with NKG2D activation by ULBP2-expressing cells. The combination of both ligands resulted in an additional increase in LFA-1–binding activity. Interestingly, results of the LC-AA and the staining with the mAb 24 followed the same pattern (Fig. 2B). This suggests that the engagement of 2B4 and the engagement of 2B4 in combination with NKG2D induce high-affinity LFA-1.
To investigate the signals necessary for LFA-1 activation by engagement of activating receptors, we used inhibitors against important signaling molecules involved in NK cell activation (Fig. 3A). Inhibiting Src family kinases by PP1 or phospholipase C-γ by U73122 completely blocked 2B4- and/or NKG2D-induced LFA-1 activation. Inhibiting PI3K by wortmannin had a similar effect, although 2B4-mediated LFA-1 activation was not completely inhibited. Inside-out signals are known to be dependent on small GTPases and their GEFs. Therefore, blocking receptor proximal signaling events upstream of small GTPases was expected to affect LFA-1 activation. Inhibiting more downstream signaling events, such as p38-MAPK by SB202190 or MEK by PD98059, had only a very small or no effect on 2B4- and/or NKG2D-mediated inside-out signals, leading to LFA-1 activation (Fig. 3B). Interestingly, NK cell stimulation via a single receptor and coactivation via 2B4 and NKG2D were similarly affected by the inhibitors. This suggests that the same signaling pathways are used by the different receptors. Therefore, coactivation of NK cells may induce a stronger signal that results in synergistic LFA-1 activation. To test this, we titrated the Src family kinase inhibitor PP1. At a high dose (10 μM) all inside-out signaling was completely blocked (Fig. 3C). A suboptimal dose of 5 μM PP1 was still sufficient to block 2B4- or NKG2D-mediated LFA-1 activation. However, coactivation via 2B4 and NKG2D was still able to induce inside-out signals at this dose. This confirms that, at least at the level of Src family kinases, coactivation by 2B4 and NKG2D induces a stronger signal for LFA-1 activation and making it more resistant to PP1-mediated inhibition, similar to coactivation making expanded NK cells less sensitive to inhibitory receptor signals (6).
Stimulation with cytokines can effectively enhance the effector function of NK cells (27, 28). Therefore, we wanted to investigate how cytokines can influence LFA-1 activation. Incubation of resting human NK cells with cytokines alone for 1 or 2 d had only a small effect on the activation state of LFA-1 (Fig. 4A). Incubation with IL-15, IL-12, or IL-18 or the combination noticeably increased ICAM-1–Fc binding, but this effect was small compared with activating receptor stimulation. However, pretreatment with cytokines was very effective in enhancing inside-out signals of activating receptors, leading to enhanced LFA-1 activation (Fig. 4B). The combination of IL-12, IL-15, and IL-18 was most effective in enhancing LFA-1 activation in response to 2B4 and/or NKG2D engagement. Interestingly, cytokine pretreatment could not replace the need for coactivation, because we still observed a synergistic effect on LFA-1 activation when we coengaged 2B4 and NKG2D. However, in the presence of IL-12, IL-15, and IL-18, the triggering of NKG2D alone was now sufficient to induce some LFA-1 activation. These data demonstrate that cytokine stimulation of resting human NK cells does not result in activation of LFA-1 by itself, but it can potently enhance the inside-out signaling induced by activating receptors.
Inhibitory NK cell receptors can interfere with the signaling of activating receptors. Therefore, we wanted to investigate their effect on inside-out signaling and LFA-1 activation. In line with the finding that educated NK cells show stronger adhesion (29), the NK subset expressing the inhibitory receptor KIR2DL2/3 showed a higher background activity of LFA-1 and reacted more strongly to activating receptor stimulation compared with KIR2DL2/3− NK cells (Supplemental Fig. 2). Coengagement of KIR2DL2/3 on cells that were stimulated via 2B4 or NKG2D significantly reduced, but did not completely block, activation of LFA-1 on the KIR+ subset (Fig. 4C, Supplemental Fig. 2). This demonstrates that inhibitory KIRs can effectively interfere with inside-out signaling of activating receptors. Compared with the KIR2DL2/3+ subset, the NKG2A+ subset showed a much weaker LFA-1 activation after activating receptor stimulation (Fig. 4C). Surprisingly, coengagement of NKG2A with 2B4 or NKG2D did not interfere with LFA-1 activation.
KIR and NKG2A are differentially expressed on different NK cell subsets. Therefore, we tested whether NK cell subsets differ in LFA-1 activation via inside-out signaling. We discriminated three subsets of resting human peripheral blood NK cells (Supplemental Fig. 3): CD56bright, CD56dim/CD57−, and CD56dim/CD57+. After triggering of 2B4, CD56dim/CD57+ NK cells showed the strongest binding of ICAM-1–Fc complexes in the LC-AA (Fig. 5A). CD56dim/CD57− cells showed intermediate staining, and CD56bright NK cells had very little active LFA-1 on their surface after 2B4 engagement. This tendency was the same for triggering CD16, NKG2D, or 2B4 and NKG2D in combination (Fig. 5A). This demonstrates that the different NK cell subsets differ greatly in their LFA-1 activation upon the engagement of activating receptors. Importantly, these receptors, with the exception of CD16, are expressed at comparable levels on the different subsets (Supplemental Fig. 3), excluding that differences in expression levels are responsible for this effect. Because stimulation with Mg2+/EGTA results in complete staining with ICAM-1–Fc complexes on all subsets, the expression and functionality of LFA-1 on the subsets are also comparable. This suggests that differences in the inside-out signaling processes cause the different behavior of \the NK subsets. The higher activation state of LFA-1 on the CD56dim subsets also had functional consequences. In a conjugate assay using K562 target cells, CD56dim cells adhered much better to the target compared with CD56bright cells (Fig. 5B). Additionally, engaging 2B4 by using K562 expressing the 2B4 ligand CD48 increased the overall number of cells in contact, but the CD56bright subset still formed much fewer conjugates.
Because the difference in LFA-1 activation between the NK cell subsets pointed to variations in the signal transduction, we took advantage of a proteomics study (30) that investigated expression differences between these subsets. The small calcium-binding protein S100A4 shows a strong and previously unknown increase in expression through the maturation stages, from lowest in the CD56bright subset to highest in the CD56dim/CD57+ subset. Proteins of the S100 family are involved in signal transduction and controlling the cytoskeleton (31). S100A4 was primarily known for its metastasis-promoting effect in tumor cells (32), but we speculated that the differential expression among NK cell subsets could be relevant for NK cell functions. We confirmed the differential expression using multicolor flow cytometry, with a very consistent pattern among all tested donors (Fig. 5C). Incubation with proinflammatory cytokines increases the expression level of S100A4 compared with the actin control (Fig. 5D). S100A4 could provide a functional link between the early activation signals induced by the activating NK cell receptors and cytoskeletal rearrangements needed for efficient LFA-1 activation (33, 34). Unfortunately, we were not able to generate a sufficient small interfering RNA–mediated knockdown of S100A4 to check whether there really is a functional connection. As an alternative to manipulation by small interfering RNA, we chose to use drugs of the phenothiazine group, which were described to inhibit S100A4 (35). We used TFP because it showed the most robust effects, inhibiting cytotoxicity completely (Fig. 6A) at concentrations not affecting cell viability. Treating the cells with 20 μM TFP effectively inhibited LFA-1 activation mediated by triggering different activating receptors (Fig. 6B). Formation of conjugates with K562 cells was also reduced in the presence of the inhibitor (Fig. 6C). To test that TFP treatment indeed affected S100A4, we performed microscopic analysis of conjugates (Fig. 6D). In the majority of control conjugates, S100A4 was polarized toward the contact with the K562 target cell, in accordance with a previous study (30). Fewer conjugates showed polarization of S100A4 in the TFP-treated samples, and the strong polarization observed in some control conjugates was almost completely absent. Therefore, TFP affects the functionality of S100A4, blocking its normal polarization and resulting in reduced conjugate formation and LFA-1 activation.
The interaction of LFA-1 on NK cells with its ligands on potential target cells is important for the stable adhesion, formation of the immunological synapse, polarization of lytic granules, and subsequent killing of the target cell. In this study, we investigated the requirements necessary for efficient activation of LFA-1.
The LC-AA provides an easy and exact method to investigate LFA-1 activation. The advantage of this assay is that avidity and affinity changes of LFA-1 can be monitored by binding of the natural ligand and analyzed on a single-cell level. Staining with LFA-1 conformation–specific Abs can give more detailed information about the affinity-maturation process of LFA-1 (36), but it gives no information about the overall adhesiveness of the cells. Classical adhesion assays with plate-bound ICAM-1 also reveal avidity and affinity changes but cannot be used for single-cell or subpopulation analysis. Additionally, the contact with plate-bound ICAM-1 might distort the effect of inside-out signals, because mechanical triggers can influence the LFA-1–ICAM interaction (37, 38). These mechanical triggers (i.e., the drag caused by binding to a tethered ligand and shear forces tearing on the receptor–ligand pair) have an effect on LFA-1 that is independent of inside-out signals (39). The LC-AA and staining with mAb 24 should not trigger these mechanisms, because the assays are performed with soluble agents and, therefore, only measure inside-out activation of LFA-1.
We found that LFA-1 had a low activity on freshly isolated, resting human NK cells from most donors. This may be connected to the low cytotoxic activity of resting NK cells. Interestingly, triggering of 2B4 or NTB-A was sufficient to induce affinity maturation of LFA-1, whereas triggering of NKG2D, NKp30, or NKp46 was not sufficient to induce substantial LFA-1 activation. Therefore, members of the SLAM receptor family seem to be particularly good at inducing LFA-1 activation and target cell adhesion. Indeed, we observed better adhesion to K562 cells when the 2B4 ligand CD48 was expressed by the target cells; however, 2B4 engagement on resting NK cells is not sufficient to trigger cytotoxicity. For this, the coengagement of at least two coactivating receptors is necessary (13). For receptor combinations that have a coactivating effect on cytotoxicity, we also noted a synergistic increase in the induction of LFA-1–binding activity. This suggests that coactivating signals converge at the level of LFA-1 activation. It is interesting to speculate that these complementary signals induce LFA-1 affinity and avidity increases to provide optimal target cell adhesion. Engagement of 2B4, but not of NKG2D, increased LFA-1 affinity. Therefore, NKG2D may contribute more to LFA-1 avidity changes, thus complementing the 2B4 signal and resulting in efficient LFA-1–binding activity upon coactivation.
Stimulation by cytokines can greatly enhance the activity of NK cells. Interestingly, incubation with IL-12, IL-15, and IL-18 for 2 d induced only a very weak increase in LFA-1 activity. Therefore, cytokine signaling itself is insufficient to induce inside-out signaling in NK cells. However, preincubation with cytokines made the NK cells more responsive to subsequent activating receptor stimulation. Interestingly, NKG2D triggering on cytokine-primed NK cells was now sufficient to induce LFA-1 activation, demonstrating that cytokine stimulation can overcome the need for coactivation at the level of LFA-1 activation. Stimulation with IL-12, IL-15, and IL-18 was shown to produce cytokine-induced memory-like NK cells (28, 40) with enhanced effector functions. Our data show that these cytokines greatly enhance the inside-out signaling by different activating receptors to stimulate LFA-1 activation, suggesting that enhanced LFA-1 activity is at least one reason for the improved effector functions of memory-like NK cells.
The licensing of NK cells via their inhibitory receptors has a big impact on the response to activating receptor stimulation (41). Recently, it was shown that NK cells that lack inhibitory receptors and, therefore, are uneducated show reduced adhesion (29). This was caused by differences in the inside-out signals from activating receptors (14), which we also observed in our experiments: 2B4 and NKG2D were more potent in inducing LFA-1 activation of KIR2DL2/3+ cells compared with KIR2DL2/3− cells.
Inhibitory receptors are also important to control the signals of activating NK cell receptors, and they can effectively inhibit the polarization of lytic granules toward a target cell (7). Our data show that KIR2DL2/3 could control activating receptor-mediated LFA-1 activation. Because LFA-1 signaling is necessary for granule polarization (15, 16), the KIR-mediated control of LFA-1 activity may be the mechanism through which inhibitory receptors control granule polarization.
Surprisingly, the triggering of KIR2DL2/3, but not NKG2A, inhibited the activation of LFA-1 mediated by activating receptor signals in our system. Although we cannot exclude that the inhibitory signals differ between KIR and CD94/NKG2A, another explanation for this surprising effect might be the expression pattern of the receptors. KIRs are primarily expressed on the mature CD56dim NK cell subset, whereas NKG2A is highly expressed on more immature CD56bright NK cells. Because the NKG2A+ subset, as well as CD56bright NK cells, showed very little LFA-1 activation mediated by activating receptor engagement, we may not be able to detect a significant inhibitory effect of the CD94/NKG2A receptor in our system. The CD56bright subset is considered the least mature among peripheral blood NK cells. It reacts better to cytokine stimulation, but it only shows weak degranulation and cytokine production upon stimulation with target cells. In contrast, the CD56dim subset shows strong cytotoxicity against target cells. Among the CD56dim cells, the additional expression of CD57 marks the most mature cells that were shown to be even more reactive (3). In our assay, we found that LFA-1 activation corresponds to these differences in activity. Activating receptor stimulation induced the greatest LFA-1 activation in CD57+ NK cells, and it was almost ineffective for CD56bright cells. Our data confirm that the expression of the small Ca2+-binding protein S100A4 correlates with these different activities of NK cell subsets (30), with high expression in CD57+ CD56dim NK cells, intermediate expression in CD57− CD56dim NK cells, and lower expression in CD56bright NK cells. S100A4 is well known for its metastasis-promoting effect in tumors (42), but its effects on the motility of healthy cells and its possible involvement in receptor proximal activation signals are also documented (33, 43). TFP and other drugs of the phenothiazine group can oligomerize S100A4 and inhibit its functions (35). We cannot exclude that TFP has unspecific side effects, but we were able to show that it reduced the polarization of S100A4 at the immunological synapse, reduced the activation of LFA-1 through inside-out signals, and inhibited conjugate formation and cytotoxicity. Incubation with proinflammatory cytokines increased the S100A4 expression level, as well as the activation of LFA-1, upon activating receptor stimulation. Therefore, we speculate that S100A4 is needed to properly convert the inside-out signal from activating NK receptors to cytoskeleton remodeling and LFA-1 activation. Therefore, the different expression levels of S100A4 could be an important factor in determining LFA-1 function and, ultimately, the cytotoxic activity of the different NK cell subsets.
Although further studies are needed to clarify the role of S100A4 and other members of the S100 family in the signaling processes of NK cells, our data clearly show that the regulation of LFA-1–binding activity is involved in important processes, such as NK cell maturation, cytokine-induced memory, and the differential activities of NK cell subsets.
We thank Lothar Jänsch and Maxi Heyner (née Scheiter) for helpful discussions and Linda Drenkelforth for expert technical assistance.
This work was supported by the Leibniz Association (SAW-2013-IfADo-2) and the Deutsche Forschungsgemeinschaft (WA-1552/5-1).
The online version of this article contains supplemental material.
Abbreviations used in this article:
Ab-dependent cellular cytotoxicity
killer cell Ig-like receptor
ligand complex–based adhesion assay
mean fluorescence intensity
The authors have no financial conflicts of interest.