Dependence of Both Spontaneous and Antibody-Dependent, Granule Exocytosis-Mediated NK Cell Cytotoxicity on Extracellular Signal-Regulated Kinases¹

Natural killer cells control viral infections and the generation of cell-mediated immune responses through MHC-nonrestricted cytotoxicity and cytokine production (reviewed in Refs. 1 and 2). These functions are elicited upon binding of IgG Ab-coated target cells by the low affinity receptor for the Fc fragment of IgG (FcγRIIIA)³ (3), and of a variety of target cells (including normal, virus-infected, and transformed cells) (4) via receptors still poorly characterized. In both cases, the “lethal hit” step leading to target cell lysis follows the binding and recognition of the target cells, and the consequent migration of intracytoplasmic granules to the sites of effector/target cell interaction, regulated by the rearrangement of cytoskeleton proteins (5). The actual cytotoxic event depends, for certain target cells, on Ca²⁺-dependent exocytosis of cytotoxic proteins contained in the intracytoplasmic granules, among which perforin and granzymes are most relevant (reviewed in Ref. 6), as indicated from the analysis of perforin^−/− mice (7). An alternative, Ca²⁺-independent, mechanism of cytotoxicity involves Fas-mediated apoptosis upon interaction of Fas⁺ target cells with Fas ligand, expressed or induced on the effector cells (8, 9).

The earliest signal transduction events elicited upon engagement of surface molecules that trigger cytotoxicity, be it spontaneous or Ab-dependent, include activation of protein tyrosine kinases (PTK) (reviewed in Ref. 10) (3), among which Syk has been reported necessary for both types of cytotoxicity (11), and phospholipase C-γ1 and -γ2-induced increase in intracellular Ca²⁺ concentration ([Ca²⁺]_i) (12, 13, 14). However, the downstream targets used in the two conditions are, at least in part, distinct. For example, ZAP-70 PTK is activated upon binding Ig-coated, but not nonsensitized, target cells (11, 15); protein kinase C (PKC), but not phosphatidylinositol 3-kinase, is involved in the regulation of spontaneous cytotoxicity, whereas the reverse is true for the FcγRIIIA-dependent granule release and killing (16). Both types of cytotoxicity are abolished upon inhibition of PTK (17, 18), or activation of protein tyrosine phosphatases (19, 20) elicited upon engagement of killer inhibitory receptors binding MHC class I Ags on the target cells, concomitant with binding of IgG or other ligands on the target cells by receptors activating cytotoxicity (reviewed in Ref. 21). At least one of these killer-activating receptors (p50 KAR) on NK cells is associated with one or more ∼12 kDa phosphoproteins (22), and a molecule of similar m.w. has been reported phosphorylated on tyrosine residues upon KAR cross-linking (23). These observations, together with the data indicating direct association, in vitro, between Syk family kinases and phosphopeptides from DAP12 (a molecule containing immune-based tyrosine-associated motifs) (24) support the notion that, like FcγRIIIA (15, 25), one or more of these, and possibly other yet to be identified NK cell surface receptors involved in triggering spontaneous cytotoxicity are associated, directly or indirectly, with PTK.

Several receptors associated with PTK or endowed with intrinsic tyrosine kinase activity utilize extracellular signal-regulated kinases (ERK), also known as mitogen-activated protein kinases (MAPK), for signal transduction leading to gene expression (26). These are serine-threonine kinases that become activated, in a Ca²⁺-dependent (27) or -independent fashion depending on the stimulus, upon phosphorylation of threonine (T 183) and tyrosine (Y 185) in their TEY motif (28). We have shown that ERK2 activation is prerequisite to FcγRIIIA-induced expression of early activation genes (e.g., c-fos) and TNF-α in NK cells (29), and others reported that the same kinase controls FcγRIIIA-induced granule exocytosis in NK cells (30), likely independently from effects on gene transcription.

Here, we tested the hypothesis that ERK, if activated upon recognition of Ab nonsensitized target cells, are involved in regulating spontaneous cytotoxicity. Our data indicate that ERK2 is activated in NK cells upon binding K562, the prototype target cell for spontaneous Ca²⁺-dependent cytotoxicity. This kinase regulates not only expression of cytokine genes, but also both Ca²⁺-dependent spontaneous and Ab-dependent cytotoxicity (ADCC), acting, at least in part, by inhibiting early events necessary for granule exocytosis. These data demonstrate that a common, not PTK, molecule serves to regulate NK cell effector functions triggered upon target cell recognition mediated both by FcγRIIIA and by other yet to be identified receptors.

The human monocytic THP-1, erythroleukemic Fas⁻ K562, B lymphoblastoid RPMI 8866, T lymphoid Fas⁺ Jurkat (clone J32), and the murine mastocytoma P815 cell lines were maintained in culture in RPMI 1640 medium (BioWhittaker, Walkersville, MD) supplemented with 10% heat-inactivated FBS (Sigma Chemical Co., St. Louis, MO) and 100 μg/ml glutamine (Life, Technologies, Gaithersburg, MD).

Homogeneous NK cell preparations were obtained from 10-day cocultures of PBL from healthy individuals with 30-Gy-irradiated RPMI 8866 cells following negative selection using a mixture of anti-CD14, -CD3, and -CD5 mAb and indirect anti-Ig rosetting, as previously described (31). The cell preparations were >98% CD16⁺/CD56⁺/CD3⁻, and <3% CD3⁺ cells, as determined in indirect immunofluorescence (flow cytometry) using a panel of mAb.

mAb 3G8 (anti-CD16), B159.5 (anti-CD56), B36.1 (anti-CD5), OKT3 (anti-CD3), and B52.1 (anti-CD14) have been previously described (31). When indicated, IgG were purified from ascites on protein G-Sepharose columns (Pierce, Rockford, IL). The goat anti-mouse Ig (GaMIg) used to cross-link the mAb used for NK cell stimulation was produced in our laboratory, adsorbed on human Ig, and affinity purified on mouse IgG-Sepharose 4B columns (Pharmacia Biotechnology, Uppsala, Sweden) before use. The polyclonal rabbit sera ERK2 (anti-MAPK ERK2, detecting, primarily, ERK2 but cross-reacting with ERK1) and ERK1 (cross-reacting with MAPK ERK1 and ERK2) were from Santa Cruz Biotechnology (Santa Cruz, CA); the anti-active ERK Ab, specifically detecting enzymatically active ERK dually phosphorylated on T 183 and Y 185 (27), was from Promega (Madison, WI); the anti-P815 rabbit serum has been produced in our laboratory.

When indicated, NK cells were metabolically labeled with H₃³²PO₄ (see below) and incubated (5 × 10⁶/ml serum-free medium, 15 min, 37°C) with or without trypsin (0.25 mg/ml) in the presence of DNase, 50 μg/ml (both from Sigma). After washing, the cells were incubated for the indicated times (5 × 10⁶/ml, 37°C) with the different stimuli. These were: K562 cells (5:1, NK to target cell ratio), mAb 3G8 (anti-FcγRIIIA), or B159.5 (anti-CD56) (both 2 μg/ml) with added or not: 10 μg/ml of GaMIg, 50 ng/ml of PMA (Sigma), 1 μM ionomycin (Sigma), immune complexes [rabbit IgG-sensitized bovine erythrocytes (EA) prepared as described (32), or E (negative control) 0.5% suspension]. When indicated, K562 were fixed with paraformaldehyde (3 × 10⁶ cells/ml, 1% paraformaldehyde in PBS, 30 min on ice) and washed extensively before use. Preliminary experiments (not shown) indicated that these target cells stimulate a pattern of tyrosine phosphorylation in NK cells similar to that detected upon stimulation with nontreated cells and at similar level. When K562 or E were used to stimulate NK cells, the effector/target cell mixtures were centrifuged (600 rpm, 2 min) before incubation. Treatment with the MAPK kinase (MEK) inhibitor PD098059 (33) (kindly provided by Dr. A. Saltiel, Parke-Davis Pharmaceutical Research/Warner-Lambert Co., Ann Arbor, MI) was for 40 min at 37°C.

NK cells, washed three times with 0.15 M NaCl, were resuspended in phosphate-free RPMI 1640 medium (ICN Pharmaceuticals, Costa Mesa, CA) supplemented with 1% glutamine, 0.1% BSA, and 10 mM HEPES, pH 7.4. After 1-h incubation at 37°C, the cells were resuspended (15 × 10⁶ cells/ml) in phosphate-free medium containing 0.5 mCi H₃ ³²PO₄/ml (spec. act. ∼400–800 mCi/ml, ICN Pharmaceuticals). After 3-h additional incubation at 37°C, the cells were washed with the same medium without H₃ ³²PO₄ and incubated (5 × 10⁶/ml phosphate-free medium) with the different stimuli for the indicated times. The cells were then lysed (10⁸ cells/ml lysis buffer: 1% Nonidet P-40, 10 mM HEPES, pH 7.5, 0.15 M NaCl, 10% glycerol, 10 μg/ml each aprotinin and leupeptin, 1 mM PMSF, 1 mM Na₃VO₄, 50 mM NaF, and 1 mM EDTA), as described (34).

Immunoprecipitation and Western blotting were performed according to our published protocols (29). MAPK were immunoprecipitated from the cell lysates using the polyclonal ERK2 Ab (2 h, 4°C, 15 × 10⁶ cell equivalent and 1 μg Ab/sample) and protein A-Sepharose beads (Pharmacia Biotechnology). Immunoprecipitated proteins were separated in SDS-PAGE under reducing conditions, and ERK2 were visualized after exposure of the filters to X-AR films (Eastman Kodak, Rochester, NY). Molecular weight markers were run in every gel. Western blot analysis was performed as previously described (29) using anti-ERK1, anti-ERK2, or anti-active ERK Ab, as indicated, and Ab-reactive proteins were detected with horseradish peroxidase-labeled sheep anti-rabbit Ig sera and enhanced chemiluminescence (ECL) (Amersham Corp., Arlington Heights, IL). MAPK assays were performed using myelin basic protein (MBP) as a substrate, as previously described in detail (29).

cDNA encoding the wild-type (wt) and dominant-negative ERK1 (T192A, T → A mutation at position 192), both in frame with the influenza hemagglutinin (HA) nonapeptide YDVPDYASL epitope in the pcDNA-neo vector (35, 36), were kindly provided by Dr. J. Pouyssegur (Université de Nice, Nice, France). To generate recombinant Vac encoding wt and T192A ERK1, cDNA fragments encoding the fusion proteins were obtained after XbaI digestion and HindIII partial digestion, the blunt-ended cDNA were inserted into the NheI cloning site of the psC11 vector, and introduced into Vac, WR strain, by homologous recombination, as previously described (37). For infection, NK cells were incubated with the indicated Vac recombinant at 10–20 multiplicity of infection (37°C, 1.5 h, 10 × 10⁶/ml, and additional 4 h, 2 × 10⁶/ml RPMI 1640 supplemented with 10% FBS). After a washing, the cells were used immediately. Expression of the wt or dominant-negative ERK1-HA fusion recombinant proteins was confirmed in Western blot with anti-ERK1 Ab.

Total cellular RNA was extracted from control or stimulated cells using Trizol Reagent (Life Technologies) following the manufacturer’s specifications, size fractionated in 1% agarose-formaldehyde gels, transferred by capillarity onto Hybond-nylon membranes (Amersham), and hybridized to cDNA probes specific for human IFN-γ. TCR β-chain (detecting a nonfunctional, truncated, 1.0-kb mRNA species in NK cells) and β₂-microglobulin cDNA were used for normalization, as described (38). RNA was also visualized in ethidium bromide-stained gels. cDNA probes were labeled with [α-³²P]dCTP (spec. act. 3000 Ci/mmol; ICN Pharmaceuticals) by random priming (Boehringer Mannheim, Indianapolis, IN). Filters were exposed to X-AR films for autoradiography. Levels of expression of each mRNA species were quantitated by densitometric analysis using a laser scanner (Personal Densitometer SI, Molecular Dynamics, Sunnyvale, CA) with proprietary software (ImageQuant). Computer-assisted imaging was performed on the scanned autoradiograms; the background in the figures shown are typical of those in the original film.

This was performed as described by Visonneau et al. (39). Briefly, 3G8 and B159.5 IgG (10 μg/ml carbonate buffer, pH 9.6) were immobilized (18 h, 4°C) on 96-well microtiter plates. After the plates were washed with RPMI 1640 medium supplemented with 5% FBS and 10 mM HEPES buffer to saturate protein-free sites, NK cells were added to the wells (2 × 10⁵ cells/triplicate wells, 100 μl RPMI 1640-10% FBS). PMA, 50 ng/ml, and ionomycin, 1 μM (both from Sigma), were used as positive control. Cell-free supernatants were collected after 4-h incubation at 37°C. Cells were lysed following suspension in either 100 μl of medium and freeze/thawing (3 cycles), or 100 μl of 0.2% Nonidet P-40. Samples from supernatants and cell lysates (20 μl) were mixed with 180 μl of 0.1 M Tris-HCl (pH 8.0) containing 2 × 10⁻⁴ M BLT (Calbiochem, San Diego, CA), 1.1 × 10⁻⁴ M 5,5′-dithiobis-2-nitrobenzoic acid (Pierce). After 30-min incubation at room temperature, the ODs of the samples were measured in an ELISA plate reader (Bio-Rad Laboratories, Hercules, CA) using a 412-nm filter. The percentage of released BLT esterase activity was calculated for each sample according to the formula (S/S + C) × 100, where S is OD in the supernatants, and C is that in the corresponding cell lysates.

K562, THP-1, Jurkat, or P815, as indicated, were used as target cells in 4- or 6-h ⁵¹Cr-release assays (32). For redirected ADCC, mAb 3G8 (supernatant, at a 1:4 predetermined optimal concentration) was present throughout the assay with THP-1 cells; for ADCC, P815 cells were sensitized with anti-P815 Ab rabbit serum at predetermined optimal (10⁻⁴) concentration. When indicated, 1 mM EGTA and 2 mM MgCl₂ were added throughout the assay. A constant number of target cells (5 × 10³–10⁴/well, as indicated) and serial dilutions of effector cells were used, in triplicates. Spontaneous release from any of the target cells used was always <10% in the assay time. Lytic units (LU) per 10⁶ cells were calculated at 20% specific ⁵¹Cr release (LU_20%) (40).

NK/target cell mixtures (5:1) were centrifuged (500 rpm, 2 min) and incubated at 37°C. After 1 h, the cells were gently resuspended, conjugates were counted, and cytospin preparations were made. The 1-h incubation time was chosen based on results of preliminary experiments (not shown), performed after 15-, 30-, or 60-min incubation, indicating optimal (maximal) percentages of effector cells with detectable actin or tubulin rearrangement at this time point. The percentage of NK:target cell conjugates was similar before and after centrifugation. After 10-min fixation in PBS-3.7% formaldehyde (Sigma), the cells were permeabilized in PBS-0.1% Triton X-100 (Sigma) for 5 min. To detect actin, the slides were incubated (30 min, 37°C) with rhodamine-phalloidin (Molecular Probes, Eugene, OR) (10⁻² in PBS). To detect tubulin, fixed cells were permeabilized as above and incubated (15 min, 20°C) with 50 mM NH₄Cl in PBS. After 30-min incubation in PBS-10% rabbit serum, anti-α tubulin mAb (Amersham) and FITC-GaMIgG (Vector Laboratories, Burlingame, CA) (PBS-3% BSA) were sequentially added to the slides (1.5 h, 20°C, each incubation). Coverslips were mounted using Slow Fade Antifade Kit (Molecular Probes), conjugates were counted scoring at least 500 cells/duplicate slides, and the percentage of conjugates in which actin and tubulin rearrangements were present was calculated out of ∼100 conjugates/sample. Blind-coded slides were analyzed by two individuals using a fluorescence microscope (Leitz Diaplan, Leica, Malvern, PA).

To determine whether binding of sensitive target cells induces MEK activation, we analyzed ERK2 phosphorylation in NK cells metabolically labeled with [³²P]orthophosphate incubated with K562 cells for 2 min. Effector cells used as control were trypsin treated to abolish target cell-induced triggering of spontaneous cytotoxicity (32), and incubated with K562 cells with or without added anti-FcγRIIIA mAb 3G8, or PMA (Fig. 1,A). After SDS-PAGE analysis of MEK immunoprecipitates, phosphorylated ERK2 was detected in the samples from NK cells incubated with K562 and in those from trypsin-treated cells incubated with the same target cells with added mAb 3G8 (i.e., FcγRIIIA stimulation), or with PMA (Fig. 1,A, top), but not in those from trypsin-treated cells incubated with K562 only. Western blot analysis with anti-ERK2 Ab confirmed the presence of similar amounts of the kinase in each immunoprecipitate (Fig. 1,A, bottom). To confirm that the phosphorylated ERK detected in NK cells upon interaction with K562 cells corresponded to enzymatically activated ERK, Western blot analysis with anti-active (T 183, Y 185 phosphorylated) ERK Ab was performed on the lysates from NK cells stimulated with fixed K562 target cells. This protocol prevents possible activation of the target cells ERK and allows to focus specifically on the analysis of NK cell activation, avoiding confounding variables. As shown in Fig. 1 B, the basal level of active ERK2 detectable in NK cells was significantly increased within 2-min stimulation of the cells with K562 target cells and declined to control levels within 10-min stimulation. Western blot analysis with anti-ERK1 Ab (detecting both ERK1 and ERK2) confirmed the presence of similar amounts of ERK in all samples. These data confirm functional transient activation of the detected phosphorylated ERK2.

ERK2 phosphorylation was analyzed in Western blots of lysates of NK cells pretreated with different concentrations of PD098059, a selective MEK inhibitor (33), and incubated with anti-FcγRIIIA mAb or PMA (Fig. 2, top and bottom, respectively). The FcγRIIIA- and PMA-induced ERK2 phosphorylation, indicated by decreased electrophoretic mobility of the protein in SDS-PAGE, and activation, detected in in vitro kinase assays using MBP as a substrate (PMA stimulation, bottom panel; not shown for FcγRIIIA stimulation), were inhibited in a dose-dependent fashion in the cells treated with PD098059. ERK2 phosphorylation induced upon anti-FcγRIIIA or PMA treatment were inhibited by >90% using, respectively, 30 and 100 μM concentration of the inhibitor. At these concentrations, PD098059 was not toxic for NK or tumor target cells, and did not inhibit early signaling events (e.g., tyrosine phosphorylation) induced in the effector cells (29) (data not shown).

To control for the inhibitory effect of PD098059, and to determine whether the ERK activation observed in NK cells upon binding Ab nonsensitized target cells, like that induced via FcγRIIIA stimulation, is involved in the regulation of target cell-induced cytokine synthesis, target cell-induced IFN-γ mRNA, accumulation was analyzed. As previously reported (4), (41), significant levels of IFN-γ mRNA, undetectable in nonstimulated cells, were detected in NK cells incubated for 1.5 h with K562 cells (Fig. 3,A, top) or EA (Fig. 3 B, top). Instead, IFN-γ mRNA was detected neither in the control, trypsin-treated, or E-stimulated cells, nor in K562 cells. In PD098059-pretreated NK cells, K562- and EA-induced IFN-γ mRNA expression was inhibited by 90.0 ± 17.3 and 73.7 ± 15.1% (mean ± SD, n = 3 and 5, respectively), as determined after densitometric analysis.

The effect of PD098059 on BLT esterase release in response to different stimuli was analyzed to determine the role of ERK in granule exocytosis (Fig. 4). Using 50 μM PD0989059, which inhibited FcγR-induced BLT esterase release by 85% (Fig. 4, inset), both FcγRIIIA- and PMA + ionomycin-induced BLT esterase secretion were inhibited significantly (>65%) (Fig. 4). Inhibition of BLT esterase release from NK cells was also observed upon K562 binding, but the very low levels of BLT esterase secretion, detected reproducibly upon target cell binding, prevented quantitation of the results.

To determine whether ERK is involved in regulating Ca²⁺-dependent granule exocytosis-mediated spontaneous cytotoxicity and ADCC, these were measured in PD098059-treated NK cells, using as target K562, or THP-1 cells in FcγRIIIA-redirected lysis (Fig. 5). In a dose-dependent fashion, PD098059 inhibited both spontaneous (left) and FcγRIIIA-induced cytotoxicity (right) (in two separate experiments using K562 target cells; the number of LU_20%/10⁶ cells was reduced upon PD098059 treatment (50 μM) from 66 to 15, and from 105 to 20. In the same experiments, cytotoxicity against THP-1/3G8 target cells was reduced from 19 to 4, and from 26 to 7 LU_20%/10⁶ cells)). Control THP-1 cells in the presence of the anti-CD56 mAb were not lysed under any condition, whereas ADCC against IgG Ab-coated P815 cells was inhibited in the same conditions (data not shown).

In all cases, the percentages of NK cells forming conjugates with K562 target cells (not shown), and those of conjugates in which actin was rearranged in the effector cells (Fig. 6, A and B) were similar in control (42.0 ± 3.7%) and PD098059-treated NK cells (39.0 ± 6.7%, mean ± SD, n = 5). Similarly, target cell-induced tubulin rearrangement was not affected in NK cells pretreated with the inhibitor (Fig. 6, C, and D).

To exclude that the inhibition of cytotoxicity observed in the above experiments depends on either nonspecific effects or on effects on the target cells, and to prove that ERK activation is a prerequisite to NK cell-mediated cytotoxicity, this was analyzed in NK cells expressing a hemagglutinin (HA)-tagged kinase-inactive ERK1 mutant (T192A, demonstrated to function as dominant negative inhibiting activation of both ERK1 and ERK2 endogenous kinases) (36), or wt ERK1 upon infection with recombinant Vac. Expression of the exogenous proteins after infection was confirmed in Western blot with an Ab recognizing both ERK1 and ERK2 MAPK (Fig. 7, top panel). Endogenous ERK1 and ERK2 were detected in all lysates analyzed. As expected, an extra band of about 46 kDa, corresponding to HA-tagged T192A or wt ERK1, was detected in the lysates of cells infected with the corresponding recombinant Vac. Quantitative analysis of the cytotoxicity data, based on calculation of LU_20%, indicated that spontaneous and FcγRIIIA-dependent NK cell-mediated cytotoxicity (K562 and THP-1+3G8 target cells, respectively) were inhibited by 62.4 ± 11.8 and 66.9 ± 4.5%, respectively, in T192A vs wt ERK1-expressing NK cells (mean ± SD, n = 4). In the same experiments, the levels of cytotoxicity in NK cells infected with either wt ERK1 (68.7 ± 39.9 and 24.5 ± 21.9 LU_20%/10⁶ cells against K562 and THP-1+3G8 target cells, respectively) or with the nonrecombinant Vac (79.2 ± 19.4, and 32.7 ± 14.7) were not significantly different from those mediated by control noninfected cells (80.7 ± 40.8 and 32.7 ± 8.8).

To determine whether Ca²⁺-independent (Fas/FasL-dependent) cytotoxicity involves ERK activation, cytotoxicity experiments were performed using Jurkat target cells, with or without adding the Ca²⁺ chelator EGTA (Fig. 8). As expected (42), K562, but not Jurkat cell killing was completely Ca²⁺ dependent. In both cases, PD098059 inhibited cytotoxicity significantly, although never completely. The levels of cytotoxicity mediated against Jurkat cells by NK cells treated with the MEK inhibitor added to EGTA were similar to those mediated by NK cells in the presence of the Ca²⁺ chelator only, indicating that the Ca²⁺-independent portion of the cytotoxicity against Jurkat cells (Fas mediated, not shown) was not inhibited upon inactivation of the MEK pathway.

The serine/threonine MAPK are activated upon ligand binding to several receptors directly or indirectly endowed with tyrosine kinase activity (26), and regulate a variety of functions (e.g., gene transcription (43), proliferation (36), differentiation (44), and secretion (30) in both hemopoietic (29, 30) and nonhemopoietic cells (36, 44)). In this article, we demonstrate that 1) the activation of ERK2 is elicited in human primary NK cells, not only upon FcγRIIIA stimulation (29, 30), but also upon binding of Ab-nonsensitized target cells, and 2) in both cases, this kinase regulates not only cytokine gene expression but also granule release-dependent cytotoxicity induced upon target cell recognition. These data provide novel information indicating that both immediate (degranulation) and later functional events (cytokine production) following interaction of primary human NK cells with target cells are regulated not only by PTK- but also by a serine/threonine kinase (ERK2)-mediated mechanism. This is shared in the signal transduction pathways activated in these cells upon binding of their targets both via FcγRIIIA and via receptors (still to be defined precisely) that trigger cytotoxicity upon recognition of nonsensitized target cells.

Phosphorylation of ERK2 upon interaction with target cells has been demonstrated in NK cells metabolically labeled with [³²P]orthophosphate in order to exclude confounding results due to the presence of ERK of mixed (i.e., target and effector cell) origin in the lysates analyzed. ERK are activated upon phosphorylation on threonine and tyrosine residues (28). In agreement with this, we show that stimulation of NK cells with their targets induces enzymatic activation of ERK2, based on double phosphorylation of the kinase on both T 183 and Y 185, as detected with the anti-active ERK Ab (27). This extends our previous findings that ERK phosphorylation reflects functional activation of its enzymatic activity in FcγRIIIA-stimulated cells, as determined in in vitro kinase assays on ERK immunoprecipitates (29). Additional direct proof that the functionally activated ERK derives from the NK cells in the effector:target cell mixed population is provided by the results of Western blot analysis of lysates from NK cells stimulated with metabolically inactive (fixed) target cells using the anti-active ERK Ab. Further evidence is provided by the observation that no phosphorylation is associated with ERK immunoprecipitated from identical samples containing trypsin-treated NK cells. The use of these effector cells as control allows direct comparison between effector cells interacting with identical target cells in conditions that prevent triggering of the cytotoxic machinery upon spontaneous recognition (32). In addition, K562, expressing CD32, are capable of eliciting redirected cytotoxicity in the presence of Ab to activatory receptors on NK cells, thus allowing to control, under identical conditions, for the functionality of the cells, elicited via the trypsin-insensitive FcγRIIIA (32) and to exclude nonspecific effects of trypsin on MEK activation. Identification of phosphorylated active ERK in trypsin-treated NK cells incubated with K562 in the presence of anti-FcγRIIIA mAb, together with lack of ERK activation both in trypsin-treated NK cells incubated with K562 and in NK cells incubated with the NK-insensitive THP-1 target cells (not shown), indicate that activating signals transduced by trypsin-cleavable surface receptors are needed to induce activation of this kinase upon target cell binding, exclude functional deficiency of the trypsin-treated cells, and confirm that recognition of IgG Ab-coated target cells, like artificial stimulation with anti-FcγRIIIA mAb (29 , 30), results in ERK activation.

The mechanisms by which ERK activation occurs remain to be determined. Our preliminary data indicate that pretreatment of NK cells with the PI-3K inhibitors wortmannin and Ly294002 abrogates FcγRIIIA-mediated ERK activation (P. Kanakaraj, R. Trotta, and B. Perussia, unpublished observations), suggesting that PI-3K is an upstream molecule in the FcγRIIIA-induced signal transduction pathway leading to ERK activation. It is reasonable to hypothesize that, in the case of spontaneous cytotoxicity, other intermediary molecules are involved. We (45) and others (16, 46) have reported that PKC regulates spontaneous cytotoxicity, and PKC-dependent mechanisms have been demonstrated to mediate granule release by NK cells upon recognition of Ab-nonsensitized target cells (47). As indicated here, the phorbol diester PMA, which activates the serine-threonine kinase PKC directly (48), induces ERK phosphorylation and activation in NK cells. The observation, better discussed later, that PKC-dependent (PMA-induced) BLT esterase release is inhibited by the MEK inhibitor PD098059 makes it reasonable to speculate that PKC-dependent mechanisms activated upon recognition of nonsensitized target cells serve to regulate ERK activation in this case. Also, the observation that both cytotoxicity and cytokine production triggered upon target cell recognition (both Ab-dependent and not) are inhibited by PTK inhibitors (17, 18), and the fact that PTK are likely the first kinases also activated upon triggering spontaneous cytotoxicity against K562 cells, strongly suggest that molecules with PTK activity may contribute to induce ERK2 phosphorylation. Whether the same kinases are involved in this effect both via FcγRIIIA stimulation and recognition of Ab-nonsensitized target cells, and their identity, remain to be determined.

We have reported that ERK activation is a common event following engagement of FcγRI, IIA, and IIIA on human leukocytes mediating innate immunity, and that this regulates the induced expression of both early activation genes (e.g., c-fos) and genes encoding cytokines (TNF-α) (29). The involvement of ERK in the control of gene expression via phosphorylation of several transcription factors (e.g., Elk 1, c-Myb, TAL1) is well established (43, 49, 50). The observation that a specific MEK inhibitor abrogates accumulation of IFN-γ mRNA induced upon target cell recognition confirms that ERK regulate target cell-induced cytokine gene expression, likely via regulation of the activity of specific transcription factors. Whether this is the only mechanism used to regulate cytokine gene expression in NK cells upon tumor target cell binding or upon FcγRIIIA stimulation remains to be investigated.

In the attempt to determine whether ERK-dependent mechanisms also control ADCC and spontaneous cytotoxicity, as suggested by more recent data indicating that FcγRIIIA-induced granule exocytosis in NK cells in MAPK-dependent (30), we utilized a combination of biochemical and genetic approaches. Both ADCC and spontaneous Ca²⁺-dependent and granule exocytosis-mediated cytotoxicity (K562 target cells) are inhibited in NK cells treated with a specific MEK inhibitor or expressing dominant-negative, but not wt, ERK. Under the same conditions, no modulation of the receptors involved in target cell recognition was observed. This was determined in immunofluorescence (not shown), and based on analysis of conjugate formation. Moreover, biochemical assays indicated that stimulation of the Vac infected NK cells via FcγRIIIA stimulation induced a pattern of protein tyrosine phosphorylation similar to that observed in control cells (not shown). We demonstrated that ERK2 is activated upon target cell recognition, and we used an ERK1 dominant-negative mutant to analyze the functional role of this kinase. The ERK1 dominant-negative mutant used here has been previously reported to inhibit the functional activity of ERK1 and ERK2. Although the reason for this effect remains to be determined, it may depend on competition for MEK, the known activatory molecule common to MAPK (51), or on utilization of common substrates by ERK1 and ERK2. Data indicating a role for ERK2 in NK cell-mediated cytotoxicity have been reported very recently using an NK-like cell line overexpressing the transfected active or inactive form of ERK (52). Our data add significantly to those in that they prove a role for ERK in the regulation of cytotoxicity mediated by primary human NK cells, and exclude possible artifacts associated with the sole use of chemical inhibitors, or of transfected cell lines. Additionally, our data demonstrate that ERK plays a role in the Ca²⁺-dependent granule exocytosis-mediated, and not in the Fas-dependent, cytotoxicity (measured using Jurkat as target cells in the presence of the Ca²⁺ chelator EGTA). As shown in Fig. 8, and in agreement with previous observations by others (42), K562 killing is abrogated in the presence of the Ca²⁺ chelator EGTA. The reduced Jurkat cell killing in the presence of the MEK inhibitor likely reflects the involvement of ERK in the granule exocytosis-dependent component of the cytotoxicity against these target cells, as also indicated by the observation that the inhibitor did not further reduce the EGTA-independent (Fas-dependent) lysis of the same targets. This serves to confirm lack of nonspecific effects of the inhibitor itself, indicates that target cell-induced expression of functional Fas ligand (antigenically undetectable on nonstimulated NK cells) (9) (data not shown) occurs independently from ERK, and is in agreement with previous reports indicating ERK-independent regulation of Fas-mediated apoptosis (53).

Confirming a previous report (30), our data indicate that the inhibition of NK cell-mediated cytotoxicity by the MEK inhibitor depends on the regulatory effect of ERK on NK cell degranulation. Although this is demonstrated unambiguously only in the case of FcγRIIIA stimulation, the trend of the results of BLT esterase assays with K562 target cells as the stimulus, and the above considerations on the inhibition of PMA-induced degranulation, make it reasonable to suggest that the same is true in the case of stimulation with nonsensitized target cells. Both spontaneous cytotoxicity and Ab-mediated effector-target cell conjugate formation were unaffected by the MEK inhibitor (data not shown), excluding a role of ERK in target cell recognition by the receptors involved. Previous studies have demonstrated the importance of some cytoskeletal components in the killing mechanisms (54), and rearrangement of microtubules and microfilaments has been determined to follow NK:target cell recognition (55, 56) and to be essential for NK cell-mediated cytotoxicity. In particular, in experiments using specific inhibitors (54), reorganization of actin microfilaments has been suggested to be related to the binding step, whereas that of microtubules has been proposed to play a role only at the lethal hit phase. We observed that both the actin and tubulin rearrangement induced in NK cells upon interaction with K562 cells are not inhibited by the MEK inhibitor PD098059, suggesting that either ERK do not regulate degranulation affecting the rearrangement of cytoskeleton components, or their activity can be substituted by other effector molecules.

Microtubules provide a polarized scaffold along which granules can bind and move (5). In CTL, kinesin supports the motility of lytic granules toward the microtubules, as assayed in vitro (57), and several kinesin-associated proteins have been identified, the state of phosphorylation of which affects the degree of kinesin motor activity and granule release (58). The molecular mechanisms used by ERK to regulate NK cell-granule exocytosis and cell-mediated cytotoxicity remain to be investigated. Recent data have indicated that ERK2 controls mobilization of perforin and granzyme B in an NK-like cell line upon target cell contact (52). Whichever mechanism is at the basis of this, and independently from whether kinesin and/or kinesin-associated proteins will be demonstrated to be among the direct or indirect targets of ERK, our data provide novel information to indicate that the biochemical pathways elicited in primary NK cells upon engagement of receptors responsible for either spontaneous or Ab-dependent recognition of target cells, although distinct, converge distally on ERK as one of their downstream molecules regulating NK cell effector functions.

We thank Dr. D. Perrotti for helpful discussions, Dr. A. Saltiel for the gift of PD098059, Dr. J. Pouyssegur for kindly providing the vectors encoding wild type and T192A ERK1, and Mr. D. Dicker for assistance with flow cytofluorimetry.