Tumor Necrosis Factor- a Activation of the c-Jun N-terminal Kinase Pathway in Human Neutrophils INTEGRIN INVOLVEMENT IN A PATHWAY LEADING FROM CYTOPLASMIC TYROSINE KINASES TO APOPTOSIS*

The intensity and duration of an inflammatory response depends on the balance of factors that favor perpetuation versus resolution. At sites of inflammation, neutrophils adherent to other cells or matrix components are exposed to tumor necrosis factor- a (TNF a ). Although TNF a has been implicated in induction of pro-inflammatory responses, it may also inhibit the intensity of neutrophilic inflammation by promoting apoptosis. Since TNF a is not only an impor-tantactivatorofthestress-inducedpathwaysleadingtop38MAPkandc-JunN-terminalkinase(JNK)butalsoapotenteffectorofapoptosis,weinvestigatedtheeffectsofTNF a on the JNK pathway in adherent human neutrophils and the potential involvement of this pathway in neutrophil apoptosis. Stimulation with TNF a was found to result in b 2 integrin-mediated activation of the cytoplasmic tyrosine kinases Pyk2 and Syk, and activation of a three-part MAPk module composed of MEKK1, MKK7, and/or MKK4 and JNK1. JNK activation was attenuated by blocking antibodies to b 2 integrins, the m tophosphorylation PhosphorImager

The rapid influx of polymorphonuclear leukocytes (neutrophils) into sites of injury is an important component of the acute inflammatory response in humans (1). The resulting adherence and chemotaxis of leukocytes in the appropriate cytokine milieu are not only necessary for efficient clearance of microorganisms (2) but may also participate in appropriate resolution of the inflammatory response through the subsequent induction of cell death by apoptosis.
Tumor necrosis factor-␣ (TNF␣), 1 a pluripotent cytokine, is produced by a variety of leukocytes in response to stimulation by lipopolysaccharide (3). In turn, exposure to TNF␣ induces wide ranging biological effects including cell differentiation, proliferation, apoptosis, and multiple pro-inflammatory effects. TNF␣ exerts potent effects on neutrophils that have been suggested to contribute to the inflammatory response in sepsis (4,5) and the adult respiratory distress syndrome (6). However, TNF␣ also exerts anti-inflammatory effects as demonstrated by an enhanced inflammatory response in TNF␣ Ϫ/Ϫ mice after infection (7), an effect that has been ascribed to the ability of TNF␣ to induce apoptosis in neutrophils (8).
Whereas TNF␣ can prime neutrophils in suspension (9, 10), many TNF␣-induced cellular responses, including oxidant release, have been reported in association with ␤ 2 -mediated, integrin-mediated adherence (11)(12)(13). Neutrophils utilize a variety of adhesion molecules (14), but studies employing either genetically manipulated mice (15), antibody blocking studies (16), and studies with adhesion molecule-deficient patients (17) have shown that integrins, in particular ␤ 2 and ␤ 3 , are fundamentally important in adhesive events (16,18). Integrin ligation in neutrophils results in reorganization of the cytoskeleton, cell spreading across an adhesive surface, and release of O 2 Ϫ and granule constituents. Activation of tyrosine kinases plays a central role in human neutrophil integrin signaling (14). Studies have linked adherence-associated neutrophil responses in TNF␣Ϫstimulated neutrophils to ␤ 2 integrin-dependent phosphorylation of tyrosine kinase substrates and augmentation of p58 c-Fgr and p53/ 56 Lyn kinase activity (19 -22). In addition, neutrophils induced to spread over fibrinogen in the presence of TNF␣ show integrindependent activation of Syk (a non-Src-like cytoplasmic) protein tyrosine kinase (PTK) which associates with p58 c-Fgr and p53/56 Lyn in co-localized complexes. The TNF␣-induced phosphorylation of another tyrosine kinase, Pyk2, has also observed (23).
Recent studies have identified the integrin-associated cyto-plasmic tyrosine kinase Pyk2 (proline-rich tyrosine kinase 2 (24)), also known as CAK␤-(cell adhesion kinase ␤ (25)), RAFTK-(related adhesion focal tyrosine kinase (26)), CADTKcalcium-dependent tyrosine kinase CADTK (27)), as a member of the FAK family. Pyk2 has been identified in brain cells, cell lines and hematopoietic cells (24,25,28,29). Studies in adherent neutrophils exposed to TNF␣ demonstrate that tyrosine phosphorylation of Pyk2 is ␤ 2 integrin-dependent and that Pyk2 appears to lie downstream of Lyn, Syk, protein kinase C, and reorganization of the cytoskeleton (30). Furthermore, overexpression of Pyk2 in PC12 cells, which have been stimulated with TNF␣, induces JNK activation in a calcium-independent manner (31). Thus the altered response of the adherent neutrophil to TNF␣ may reflect in part differences in tyrosine kinase signaling through ␤ 2 integrin-mediated mechanisms. Three distinct MAPk families have been described to date that may act downstream of tyrosine kinases. The prototypical p42/p44 ERK MAPk are activated in response to growth factors and chemoattractants, whereas the p38 MAPk and c-Jun Nterminal kinase (JNK) families modulate stress-activated responses (32,33). Each MAPk cascade requires sequential activation of a MAPk kinase kinase (or equivalent) which can then activate a MAPk kinase, resulting in turn in the phosphorylation and activation of a MAPk (34,35). In the neutrophil, only the p38 MAPk cascade has been clearly shown to be utilized in response to TNF␣ stimulation (36). Involvement of the p42/p44 ERK MAPk pathway in response to TNF␣ is controversial. Activation of a JNK pathway in the neutrophil has not as yet been shown in any context. Furthermore, regulation of apoptosis may represent a functional consequence of JNK, and this has yet to be delineated in the context of the neutrophil inflammatory response.
Apoptosis is an important mechanism for regulating the extent of an inflammatory response, making it self-limiting and thus preventing massive tissue damage. TNF␣ is a potent inducer of apoptosis in neutrophils, but its mechanisms of action are unknown. Recent reports have suggested that TNF␣ in conjunction with ␤ 2 integrins induces apoptosis in a tyrosine kinase-dependent manner (37), whereas other studies have shown activation of the cysteine proteases, caspase-3 and -8 (8). The role of MAPK members, in particular JNK pathway activation, in the TNF␣ induction of apoptosis remains controversial (38,39).
The present study was undertaken to determine the conditions under which TNF␣ might activate the JNK pathway in human neutrophils, to define components of a putative MAPKmodule, and to delineate key upstream components, as well as downstream functional consequences (in particular apoptosis). We report that in adherent neutrophils, activation of JNK1 in response to TNF␣ results from the mobilization of an integrinmediated cascade involving activation of tyrosine kinases, in particular Syk and Pyk2, and activation of a MAPk module in which MEKK1 activates either or both of the MAPk kinases, MKK7 and MKK4. Among possible functional consequences of this pathway, we show that TNF␣-elicited JNK activation is accompanied by a dramatic acceleration of neutrophil apoptosis which is modulated both by integrins and tyrosine kinases. Together, these results provide the first description to date of the activation of the JNK MAPk cascade in the human neutrophil, while providing a potential mechanism for TNF␣-induced apoptosis in the human neutrophil.
Preincubation of Neutrophils to Promote Adherence-Neutrophils were resuspended at 20 ϫ 10 6 /ml in complete KRPD containing 0.25% human serum albumin, 1 mM PMSF, 10 g/ml leupeptin, and 10 g/ml aprotinin. Following this, 20 ϫ 10 6 cells were either preincubated in a tube under static conditions for 55 min at 37°C (cell-cell/homotypic adherence; hereafter referred to as adherent neutrophils) or preincubated in a tube under static conditions for 30 min at 37°C and then aliquoted into 1 well of a 12-well plate (cell-substratum adherence) for an additional 25 min prior to stimulation with 10 ng/ml TNF␣ for the intervals indicated hereafter. Cells were either harvested from wells into a microcentrifuge tube and pelleted at 15,000 rpm for 20 s or those stimulated in microcentrifuge tubes were directly pelleted.
JNK activity was expressed as a ratio percentage of unstimulated control (% BL). Baseline-BL.
MKK4/MKK7 in Vitro Kinase Assays-Adherent neutrophils were exposed to TNF␣, lysed in JNK lysis buffer, and immunoprecipitated with either anti-MKK4 or anti-MKK7. MKK4-bound beads or MKK7bound beads were then washed once in JNK lysis buffer and twice in JNK reaction mix. For direct in vitro kinase assays, samples were incubated with 40 l of JNK reaction mix containing 10 Ci of [␥-32 P]ATP and 500 ng of JNK1 km for 30 min at 30°C. For coupled in vitro kinase assays, samples, immunoprecipitated and washed as above, were incubated with 40 l of JNK reaction mix containing 10 Ci of [␥-32 P]ATP, 250 M ATP, 50 ng JNK1 wt , and 500 ng of c-Jun-(1-79) for 30 min at 30°C. All reactions were terminated with the addition of 5ϫ Laemmli sample buffer, subjected to SDS-PAGE, and proteins transferred to nitrocellulose. Phosphorylation of substrates (JNK1 km , c-Jun- , and JNK1 wt ) was quantified by PhosphorImager (Molecular Dynamics) analysis and visualized by autoradiography.
Inhibitor Studies-Neutrophils were preincubated with the PI3-kinase inhibitor wortmannin and tyrosine kinase inhibitor genistein as described (40,41). 10 and 30 M piceatannol were used for Syk and Lyn in vitro kinase autophosphorylation studies. Neutrophils were preincubated with tyrphostin A9 (5 M) or tyrphostin A63 (50 M) a negative control, for 1 h at 37°C prior to stimulation with TNF␣.
Apoptosis Assessment-For annexin V binding, externalized phosphatidylserine was determined by annexin V binding. Adherent neutrophils were exposed to TNF␣ for 1, 1.5, and 2 h. Aliquots containing 2.5 ϫ 10 5 cells were resuspended in 100 l of NBD buffer, pH 7.5 (137 mM NaCl, 2.7 mM KCl, 2 mM MgCl 2 , 5 mM glucose, 10 mM Hepes, and 2.5 mM CaCl 2 ), incubated with 400 ng of annexin V fluorescein (Caltag Laboratories, Burligame, CA) and 500 ng of propidium iodide for 15 min at room temperature and placed on ice with the addition of 400 l of cold NBD buffer. FACs analysis was performed by flow cytometry on a FACScan (Becton Dickinson), and annexin V binding was determined from the non-necrotic cell population, which stains negatively for propidium iodide. 5000 cells were analyzed in each sample. For caspase-3 cleavage, 20 ϫ 10 6 adherent neutrophils were exposed to TNF␣, 10 ng/ml for 0, 1, 1.5, and 2 h. Cells were lysed in RIPA, and whole cell lysates were prepared from the Triton-soluble fraction. Samples were subjected to SDS-PAGE using a 14% gel, and proteins were transferred to nitrocellulose membrane that was then probed with antibodies to caspase-3 by Western blot analysis.

TNF␣ Activates JNK in Adherent Neutrophils-Adherent
human neutrophils were stimulated with TNF␣, and JNK1 was immunoprecipitated from Triton-soluble cell lysates, and JNK activation was measured by the phosphorylation of an N-terminal fragment of c-Jun (c-Jun-(1-79)) in the presence of [ 32 P]ATP. Phosphorylation of c-Jun-(1-79) by JNK1 immunoprecipitates was augmented following TNF␣ stimulation of neutrophils (Fig. 1A). Activation was detected at 5 min and was maximal between 10 and 15 min. Immunoblots, probed with anti-JNK1 antibody, confirmed that equivalent amounts of JNK were immunoprecipitated and that in neutrophils the predominant immunoreactive JNK1 species is a 55-kDa protein (Fig. 1C). A change in the electrophoretic mobility of JNK1, consistent with phosphorylation, was detected as early as 1 min after TNF␣ exposure. JNK1 activation was further supported by immunodetection with a phosphospecific anti-JNK1 antibody, indicating increased phosphorylation of JNK in the Triton-soluble fraction of JNK1 immunoprecipitates 5 min after TNF␣ stimulation of neutrophils (Fig. 1B). The time course of JNK1 phosphorylation mirrored the phosphorylation of c-Jun-(1-79).
The Role of Adherence in JNK Activation-JNK activation was studied in neutrophils preincubated under different conditions that promote adherence and then exposed to TNF␣. Cells were maintained either stationary in a tube under conditions that favor cell-cell interactions or stimulated on a surface, favoring cell-substratum interactions ( Fig. 2A). Neutrophil exposure to TNF␣ induced comparable amounts of JNK1 activation under both sets of conditions (Fig. 2B). Since integrins, in particular ␤ 2 and ␤ 3 , are important components of neutrophil-adhesive interactions, we questioned whether ␤ 2 integrins were up-regulated by TNF␣ on the neutrophil surface. Neutrophils, exposed to TNF␣ for various time intervals, were stained with anti-CD11b, an antibody directed against ␤ 2 integrins, and their fluorescence was analyzed by flow cytometry. We observed a time-dependent up-regulation of ␤ 2 integrins on the neutrophil surface in those cells that had been exposed to TNF␣ (Fig. 2C).
␤ 2 Integrin Dependence of JNK Activation-Previous studies have indicated a role for integrins in mediating a variety of neutrophil responses, including the activity of the Na/H ϩ antiport and oxygen radical release (12,42). To determine whether the TNF␣-induced JNK activation we had observed was ␤ 2 integrin-mediated, neutrophils were pretreated with a blocking antibody directed against CD11b, prior to TNF␣ stimulation, and JNK activity was determined on JNK immunoprecipitates. The resultant JNK activation was significantly diminished compared with controls (Fig. 3, A and B).
To validate the specificity of blocking by anti-CD11b, two types of control experiments were performed. The first, using mouse IgG as an isotype-matched antibody, ruled out the possibility that the inhibition observed was due to anti-CD11b competition for protein A-Sepharose (preventing anti-JNK1 binding), whereas it also tested the blocking capabilities of anti-CD11b per se (Fig. 3, A and B). The second, carried out to determine whether anti-CD11b inhibited the TNFR and thus its downstream signaling, examined a TNF␣-induced function, actin assembly. We have previously shown in neutrophils and HL-60 promyelocytic cells, stimulated with either chemoattractants or lipopolysaccharide, that actin assembly is differentiable from integrin-mediated adherence (43)(44)(45). Neutrophils, pretreated in the absence and presence of anti-CD11b and stimulated with TNF␣, demonstrated identical actin assembly in both cases (Fig. 3C), indicating that anti-CD11b does not modulate the TNFR or its immediate downstream signaling.
MKK7 and MKK4 Activate JNK1-Whereas JNK activation has been reported as a consequence of MKK4/SEK1 activation in many cell systems (46), the recent description of MKK7 (homologous to hemipterous, the Drosophila JNK kinase) (47) as a downstream effector of TNF␣ suggests its potential role (48). Activation of MKK4 and MKK7 following TNF␣ exposure was determined in MKK4 and MKK7 immunoprecipitates by direct phosphorylation of a kinase-inactive recombinant human JNK1 (JNK1 km ). A time-dependent increase in JNK1 km phosphorylation by both MKK4 and MKK7 immunoprecipitates from TNF␣Ϫstimulated neutrophils was detected (Fig.  4A). To determine whether the JNK1 km phosphorylation induced by MKK4 and MKK7 resulted in enhanced activation of JNK1, we determined whether these MKK homologues could initiate a coupled reaction leading to c-Jun-(1-79) via kinaseactive JNK1. Both MKK4 and MKK7 immunoprecipitates from TNF␣Ϫexposed neutrophils induced phosphorylation of the JNK1 wt and c-Jun-(1-79) substrates with a time course identical to that shown for JNK1 km (Fig. 4B). However, MKK7 induced much stronger phosphorylation than did MKK4. Immunodetection of immunoprecipitated MKK4 and MKK7 confirmed that within each experiment, equivalent amounts of  -79) band under the conditions described above. By using the Student t test, we determined a lack of statistically significant differences in JNK activation. C, graph depicts TNF␣-induced up-regulation of CD11b. Neutrophils, exposed to TNF␣ (10 ng/ml) for 15, 30, and 60 min or buffer and stained with anti-CD11b, were analyzed by fluorescence analysis on the FACScan (Becton Dickinson) for surface expression of CD11b. CD11b up-regulation was expressed as a relative fluorescence index (RFI), (Ⅺ, buffer; f, TNF␣).

FIG. 3. ␤ 2 integrin involvement in JNK activation.
A, autoradiograph showing JNK activation following preincubation with anti-CD11b. Neutrophils were pretreated with either buffer (Ϫ), isotype control antibody, mouse IgG 20 g/ml (IgG), or the ␤ 2 integrin blocking antibody, anti-CD11b10 g/ml, 20 g/m,(CD11b) for 55 min at 37°C, stimulated with buffer (B) or TNF␣ (T, 1 ng/ml for 15 min). JNK activity was determined on JNK immunoprecipitates by in vitro kinase assay, in which c-Jun-(1-79) was used as substrate. Samples were separated by SDS-PAGE, transferred to nitrocellulose, and the phosphorylated c-Jun-(1-79) visualized by autoradiography. B, graph depicting JNK activity under the same conditions demonstrated in A. JNK activation was determined from c-Jun-(1-79) phosphorylation by the protocol described for samples from ( Fig. 2A) and was quantified by PhosphorImager analysis. The graph depicts mean JNK activity Ϯ S.E. of three experiments. Statistical analysis employing a one-way ANOVA determined that in the presence of 20 g/ml anti-CD11b, JNK activation was significantly reduced compared with control, * (p Ͻ 0.01). C, graph depicts TNF␣-induced increase in actin assembly in the absence and presence of anti-CD11b. Neutrophils were pretreated in the absence (Ⅺ) and presence (f) of anti-CD11b (10 g/ml), stimulated with TNF␣ (10 ng/ml) for 0, 2, 5, and 15 min, and then stained with NBD-phallacidin. The fluorescence was analyzed by flow cytometry on an Epics (Coulter) for determination of actin assembly as described previously (45). each kinase were immunoprecipitated (data not shown). Since MKK4 appeared to be only a weak activator of JNK1, we further verified that MKK4 itself was being activated by probing with an antibody specific for the phosphorylated (active) form of MKK4. MKK4 was activated in a time-dependent manner in TNF␣Ϫstimulated neutrophils (Fig. 4C). Immunoblotting the same samples with an antibody to MKK4 confirmed that equivalent amounts of MKK4 had been immunoprecipitated.
TNF␣ Induces MEKK1 Activation-MAP kinase kinase kinases reported to activate MKK4 include MLK, MEKK1, and MEKK4 (49 -51). Since it has been shown that MEKK1 activates JNK1 in TNF␣-stimulated macrophages (52), we employed a coupled assay to determine whether MEKK1 activity was induced in neutrophils by TNF␣ exposure. MEKK1 immunoprecipitates were incubated together with the kinase-active (wild type) recombinant human kinase substrates MKK4 and JNK1 and a c-Jun fragment, c-Jun- . The consequence of this coupled co-incubation is visualized in a representative autoradiograph (Fig. 5A) wherein increased phosphorylation of both the c-Jun-(1-79) and JNK1 wt species was detected, peaking 5 min after TNF␣ stimulation of adherent neutrophils. The time course of MEKK1 activity, determined from several experiments, was quantified from c-Jun-(1-79) phosphorylation (Fig. 5B). MEKK1 activation peaked 5 min after TNF␣ stimulation of neutrophils and declined to a plateau between 10 and 15 min. Phosphorylation of wild type recombinant JNK1 was also detected, confirming that JNK1 is itself a substrate for recombinant MKK4 following MEKK1 activation in neutrophils.
By having determined the TNF␣-stimulated JNK1 activation in neutrophils to be ␤ 2 integrin-dependent, we next questioned whether integrin input into the signaling pathway leading to JNK1 activation might occur upstream from MEKK1 or between MEKK1 and JNK1. MEKK activity was determined in response to TNF␣ in the presence and absence of anti-CD11b by a coupled in vitro kinase assay. MEKK1 immunoprecipitates were prepared from neutrophils, pretreated in the absence and presence of anti-CD11b, and then exposed to TNF␣.
The phosphorylation of c-Jun-(1-79) and JNK1 wt induced by these immunoprecipitates is shown (Fig. 5C). We observed greatly decreased phosphorylation of both the c-Jun-(1-79) and JNK1 wt species in TNF␣Ϫstimulated neutrophils that had been pretreated with anti-CD11b, suggesting that the integrindependent activation of JNK occurs at, or upstream of, MEKK1.
Involvement of Tyrosine Kinases in the ␤ 2 Integrin-mediated JNK Activation-Since studies in adherent neutrophils exposed to TNF␣ have shown that phosphorylation and activation of tyrosine kinases are associated with integrin activation, we sought to determine whether tyrosine kinases were involved in the TNF␣Ϫinduced JNK activation in adherent neutrophils. Activation of JNK1 from TNF␣Ϫstimulated neutrophils was assessed following pretreatment with genistein, an inhibitor of tyrosine kinases, and wortmannin, a PI3-kinase for 3 consecutive experiments. C, autoradiograph displaying ␤ 2 integrin-associated MEKK1 activity. Neutrophils, pretreated in the absence (Ϫ) or presence (ϩ) of anti-CD11b (10 g/ml), were exposed to TNF␣ (10 ng/ml) for 5 min, and lysates were immunoprecipitated with anti-MEKK1. Immunoprecipitates were subjected to coupled in vitro kinase assay as described under "Experimental Procedures." Samples were subjected to SDS-PAGE and transferred to nitrocellulose. Phosphorylated c-Jun-(1-79) and JNKwt were visualized by autoradiography (B, buffer and T, TNF␣).
inhibitor. TNF␣Ϫinduced JNK activation was significantly inhibited in cells that had been pretreated with genistein but unaffected by wortmannin (Fig. 6). Since the Src family kinase member, Lyn, and the non-Src cytoplasmic tyrosine kinase, Syk, have been implicated in the stress-activated JNK pathway in B-cells (53)(54)(55)(56) and in integrin signaling in platelets (57) and monocytes (58), we questioned whether either kinase was activated in adherent neutrophils exposed to TNF␣ and, if so, whether this activation contributed to JNK pathway activation in the neutrophil.
In initial studies, the tyrosine kinase activities of Syk and Lyn were determined in vitro by autophosphorylation. Anti-Syk immunoprecipitates from TNF␣-stimulated adherent neutrophils exhibited tyrosine kinase activity with greatly enhanced autophosphorylation of a 40-kDa species, consistent with the almost universal proteolytic cleavage product of Syk (59) (Fig. 7A). Incubation with piceatannol, a Syk-specific inhibitor (60), inhibited this autophosphorylation in a dose-dependent fashion. In contrast to the augmented Syk activation, we observed minimal activation of the Src family kinase, Lyn, in TNF␣-stimulated adherent neutrophils (Fig. 7B). Additionally, the autophosphorylation activity of immunoprecipitated Lyn was diminished only slightly at the highest concentration of piceatannol.
Since integrin signaling may proceed through Syk activation (19), we next questioned whether Syk activation in TNF␣stimulated neutrophils was also integrin-dependent. The Syk autophosphorylation activity, determined on Syk immunoprecipitates, was diminished by preincubation of neutrophils with anti-CD11b antibody prior to TNF␣ exposure (Fig. 7C). This observation provided a link between ␤ 2 integrins and Syk activation in our system, raising the possibility that Syk may be a component of the integrin-signaling pathway leading to JNK1 activation.
To test this hypothesis, we studied the effects of a Sykspecific inhibitor, piceatannol, on the TNF␣Ϫinduced activation of MEKK1 and JNK1 in adherent neutrophils. Piceatannol pretreatment of TNF␣-stimulated neutrophils resulted in reduced MEKK1 activation (Fig. 8A) as determined by the diminished phosphorylation of both the c-Jun-(1-79) and JNK1 wt species. Similarly, piceatannol pretreatment induced a concentrationdependent inhibition of JNK1 activation in TNF␣-stimulated neutrophils (Fig. 8B). Pretreatment of neutrophils with 10 M piceatannol, a concentration that significantly inhibited Syk autophosphorylation but had no effect on Lyn autophosphorylation, decreased JNK1 activation by 66%.
Recent studies have linked activation of another tyrosine kinase, Pyk2, a FAK family member, to TNF␣-induced JNK signaling in PC12 cells (31), whereas ␤ 2 integrin-mediated Pyk2 phosphorylation in adherent neutrophils, exposed to TNF␣ (23) has been also shown. Furthermore, in the latter system, Pyk2 has been shown to lie downstream of Syk and Lyn (30). Hence, we questioned whether Pyk2 was phosphorylated under our experimental conditions and, if so, was this linked to the JNK activation we had observed earlier. Enhanced tyrosine Samples were subjected to SDS-PAGE, transferred to nitrocellulose, and the autophosphorylated p40 Syk visualized by PhosphorImager analysis. B, autoradiograph depicting Lyn activity in TNF␣-stimulated (10 ng/ml, 5 min) adherent neutrophils. P53/56 Lyn autophosphorylation was determined on anti-Lyn immunoprecipitates in presence of 0.1% Me 2 SO, piceatannol (10 or 30 M). C, ␤ 2 integrin-associated inhibition of Syk autophosphorylation. Neutrophils were pretreated with anti-CD11b (10 g/ml for 55 min) exposed to TNF␣ (10 ng/ml) for 5 min, and an in vitro kinase assay was performed on the anti-Syk immunoprecipitates. Samples were subjected to SDS-PAGE, transferred to nitrocellulose, and the Syk autophosphorylation visualized (B, buffer and T, TNF␣). phosphorylation of Pyk2 was detected in Pyk2 immunoprecipitates from TNF␣-stimulated neutrophils compared with the unstimulated controls (Fig. 9A). Tyrphostin A9, a tyrosine kinase inhibitor, has been shown to inhibit specifically Pyk2 phosphorylation in TNF␣-stimulated neutrophils (23). Adherent neutrophils, pretreated with tyrphostin A9 and stimulated with TNF␣, exhibited reduction in JNK activation compared with nonpretreated controls (Fig. 9B).
TNF␣Ϫinduced Apoptosis in Adherent Neutrophils-Exposure to TNF␣ induces apoptosis in many cells by mechanisms that, while controversial, may involve the JNK pathway. Annexin V binding and caspase-3 cleavage were used to examine the onset of apoptosis in human neutrophils. Annexin V binding to neutrophils, studied at 1, 1.5, and 2 h post-TNF␣ stimulation was significantly increased at 1.5 and 2 h post-TNF␣ (Fig. 10A). Pretreatment of neutrophils with anti-CD11b F(abЈ) 2 fragments prior to TNF␣ stimulation attenuated this increase that was significant at 2 h post-TNF␣. The onset of nuclear condensation, as determined from cytospin preparations, and the susceptibility of this process to anti-CD11b fragments were similar to annexin V binding (data not shown).
Caspase-3 cleavage has been widely described in cells undergoing apoptosis, including neutrophils. We observed a time-dependent cleavage of pro-caspase-3 (32 kDa) to a caspase-3 cleavage product (17 kDa) in lysates from adherent neutrophils that had been exposed to TNF␣ (Fig. 10B). When neutrophils were preincubated with tyrphostin A9 (the Pyk2-associated tyrosine kinase inhibitor shown in Fig. 9B to inhibit JNK activation) and then stimulated with TNF␣, the cleavage of caspase-3 was greatly reduced (Fig. 10C). DISCUSSION Regulation of cellular responses to specific stimuli occurs in part through selective activation of MAPk-signaling cascades. In human neutrophils, chemoattractant-induced activation of c-Raf and MEKK1 leads to the activation of p42/p44 ERK MAPk pathways (40,41,61,62), whereas chemoattractants and lipopolysaccharide activate p38 MAPk as well (63,64). In contrast, studies to date in human neutrophils have not as yet provided evidence for activation of JNK and its associated signaling partners. This study, performed in primary human neutrophils, describes key elements leading to activation of the JNK pathway following stimulation with TNF␣ under adherent conditions. Our data support the conclusion that integration of two signals, one from integrin-mediated adherence and the other through the TNF receptor(s), regulates a pathway in which activation of tyrosine kinases, Syk and Pyk2, is linked to activation of a MAPK module consisting of MEKK1, MKK4/7, JNK1, and associated with cleavage of caspases-3 and externalization of phosphatidylserine (Fig. 11). We propose that induction of apoptosis, a critical mechanism for maintaining a self-limiting inflammatory response, may be associated with activation of these elements. This pathway not only constitutes the first report of activation and utilization of a JNK pathway in human neutrophils but joins a very limited group of studies on apoptosis induced through stimulation of the JNK pathway.
JNK activation in response to TNF␣ exposure has been reported for a variety of cell types (65). Ten different human JNK isoforms, found as either 46-or 55-kDa proteins, have been described, arising from the alternative splicing of the three genes jnk1, jnk2, and jnk3. We detected by Western blot anal-FIG. 8. Syk involvement in JNK pathway activation. A, MEKK activation following piceatannol pretreatment. Neutrophils, preincubated in the absence (Ϫ) and presence (ϩ) of piceatannol (30 M), were stimulated with TNF␣ (10 ng/ml for 5 min), and anti-MEKK1 immunoprecipitates were exposed to coupled in vitro kinase assay (as described under "Experimental Procedures"). Samples were subjected to SDS-PAGE and transferred to nitrocellulose. Phosphorylated substrates were visualized by autoradiography (B, buffer and T, TNF␣). B, JNK activity following pretreatment with piceatannol. Neutrophils, pretreated with 0.1% Me 2 SO (control) or piceatannol (3, 10, 30, and 100 M) for 55 min, were stimulated in the absence (⌬) or presence of TNF␣ (OE), 10 ng/ml, for 15 min. JNK1 immunoprecipitates were subjected to in vitro kinase assay using c-Jun-(1-79) as substrate, and samples separated by SDS-PAGE were transferred to nitrocellulose. Phosphorylated c-Jun-(1-79) was quantified by PhosphorImager analysis. Graph represents mean c-Jun phosphorylation Ϯ S.E. of three consecutive experiments. ysis both the 46-and 55-kDa species, although the predominant species was a 55-kDa protein. This p55 splice variant we detected is probably JNK1␣ 2 or JNK1␤ 2 (65,66). Evidence of TNF␣-induced neutrophil JNK activation comes both from observations that immunoprecipitated, endogenous JNK is itself specifically phosphorylated following TNF␣ stimulation and that it phosphorylates a c-Jun fragment substrate.
The MAP kinases kinases reported to be directly upstream of JNK include MKK4 and the recently cloned MKK7 (67). Overexpression of MKK4 in a variety of cell types leads to activation of JNK(s) (46,68), and its targeted deletion blocks the activation of JNK in embryonic fibroblasts (69). Recent studies, however, suggest that MKK7 may be the preferred MKK leading to JNK activation following TNF␣ exposure (48). In our studies, both MKK4 and MKK7 were activated in a stimulus-dependent manner when neutrophils were exposed to TNF␣, indicating the likelihood that both MKK4 and MKK7 are involved in JNK1 activation in the neutrophil. Our data, while suggesting that MKK7 may be a more potent activator of JNK1 than MKK4 in this system, do not preclude the possibility that immunoprecipitation may not allow detection of the full extent of activation. In keeping with the upstream location of MKK4 and MKK7, their peak activation (5 min) precedes that of JNK1 (10 -15 min). Similar time courses of activation for JNK and MKK4 have been reported in macrophages following TNF␣ exposure (52).
MAP kinase kinase kinases shown to act upstream of JNK include MEKK family members (49,50,70,71), MLK3 (72), ASK-1 (51), and NIK (73). We have focused on MEKK1, which phosphorylates and activates MKK4 in vitro (49). Overexpression of MEKK1 in a variety of cell types results in activation of both MKK4 and JNK(s) (49,74) and also leads to apoptosis in fibroblasts (75). In this study, we demonstrate, with the use of recombinant constituents of the JNK pathway, MEKK1 activation in adherent neutrophils following exposure to TNF␣. In light of our previous findings that formyl-methionyl-leucylphenylalanine induces MEKK1-associated p42/44 ERK MAPk activation (conditions in which MKK4 is not activated (40)), we speculate that the neutrophil may respond to specific stimuli by using the same upstream kinase but allowing different downstream effectors to be activated. The yeast accomplishes such specificity through the use of scaffolding proteins (76). Additionally, or alternatively, other MEKK family members may also activate MKK4 and MKK7 in response to TNF␣.
Signaling events upstream of MEKK remain to be elucidated, although the pleiotropic actions of TNF␣ (7) are known to be mediated by two TNF receptors, RI and RII, that belong to a superfamily of cell surface receptors including Fas and FIG. 10. TNF␣؊induced apoptosis. A, effect of anti-CD11b on TNF␣-induced annexin V binding. Unstimulated neutrophils (Ⅺ) or neutrophils exposed to TNF␣ (10 ng/ml, for 1, 1.5, and 2 h) in the absence (f) and presence (OE) of anti-CD11b F(abЈ) 2 fragments (2 g/ml) were incubated with a mixture of annexin V fluorescein and propidium iodide, and their fluorescence was analyzed by flow cytometry. (10,000 cells were analyzed, and the percent of cells binding annexin V was determined from the propidium iodide-negative staining population.) Graph represents mean annexin V binding Ϯ S.E. of three experiments. Statistical analysis was carried out by two-way ANOVA. * designates comparison of (i) TNF␣/buffer with p Ͻ 0.0136 at 1.5 h and p Ͻ 0.0001 for 2 h and (ii) TNF␣-CD11b/buffer with p Ͻ 0.002. ** represents (p Ͻ 0.01) for samples stimulated with TNF␣ for 2 h in the presence and absence of anti-CD11b F(abЈ) 2 fragments. B, TNF␣ induces caspase-3 cleavage in neutrophils. Anti-caspase-3 immunoblot shows the appearance of the 17-kDa cleavage product with time. Triton-soluble lysates were prepared from neutrophils stimulated with either buffer (B) or TNF␣ (T) 10 ng/ml for 0, 60, 90, and 120 min. Samples were separated by SDS-PAGE, transferred to nitrocellulose, and probed with anticaspase-3. The 32-kDa proenzyme is cleaved into a heterodimer of 17-21 and 12 kDa. In neutrophils we have only observed the 17-kDa species. C, immunoblot showing the effect of tyrphostin A9 on caspase-3 cleavage. Neutrophils were preincubated in the absence (Ϫ) or presence of 5 M tyrphostin A9 (ϩ) prior to stimulation with buffer (B) or TNF␣ (T), 10 ng/ml, for 1, 1.5, and 2 h. Triton-soluble lysates were prepared, and samples separated by SDS-PAGE were transferred to nitrocellulose and probed with anti-caspase-3 by Western blotting.
FIG. 11. Proposed scheme for activation of JNK pathway in TNF␣-stimulated adherent neutrophils. TNF␣-stimulated activation of JNK1 proceeds through TNFR and ␤ 2 integrin-mediated activation of the Syk and Pyk2 tyrosine kinases, MEKK1 and MKK4/MKK7. Functional outcomes of JNK1 activation may be associated with induction of apoptosis.
CD40 (77). Although most of the biological effects of TNF␣ are ascribed to TNFRI (78,79), neutrophils express both types. 2 Binding of TNF␣ induces receptor aggregation and recruitment of TNFR-associated proteins (80 -82), including TRAF2 and RIP, thought to be points of divergence that lead to many downstream events, including activation of the JNK pathway and of the transcription factor NF-B (83).
In spite of the many activities linked to activation of TNFRI and -II, we demonstrate that an additional component, neutrophil adherence, is necessary for JNK1 activation in TNF␣stimulated neutrophils. Although neutrophils utilize a variety of adhesion molecules, ␤ 2 integrins are fundamentally important in both homotypic and cell-substratum adhesive events, and neutrophil interactions with the extracellular matrix have been shown to require integrins (16). In our studies, both the JNK and MEKK1 activation in TNF␣-stimulated adherent neutrophils could be blocked by pretreatment of cells with an antibody to CD11b, the ␣ subunit of the ␤ 2 integrin, Mac-1 (CD11b/CD18, aM␤ 2 ). Although JNK activation following integrin activation alone has been reported (84), we found that in neutrophils adherence in the absence of TNF␣ stimulation was insufficient to initiate activation of JNK1. By using an isotype control antibody, we demonstrated that the anti-Cd11b blocking was indeed specific. Additionally, we present evidence that anti-CD11b does not block the TNFR itself by demonstrating that TNF␣-induced actin assembly in neutrophils is an integrin-independent function (Fig. 2C). Initial actin assembly in the neutrophil in response to variety of agonists occurs in suspension and is independent of ␤ 2 integrin blockade. Hence, this provides a system in which integrin-independent signaling events downstream of the TNFR and its expression can be assessed.
Neutrophil interactions with the extracellular matrix have been shown to induce phosphorylation of specific cytoplasmic tyrosine kinases, including the Src family member p58 Fgr (19 -22). Fuortes and colleagues (20) have demonstrated that TNF␣ activation of adherent neutrophils resulted in tyrosine phosphorylation of several protein species that were not seen in suspended cells. Our studies show that the tyrosine kinase inhibitor, genistein, attenuated TNF␣-induced JNK activation, providing initial evidence for upstream participation of tyrosine kinases. A recent study in T-cells reports that activation through the T-cell receptor, if associated with ligation of a co-receptor, results in activation of Syk and downstream activation of the JNK pathway (85), whereas Syk has also been reported to be associated with integrins (57,58). We have not only demonstrated Syk activation, which could be inhibited by anti-CD11b antibodies, but also inhibition of both Syk and JNK activation by piceatannol, the Syk-specific inhibitor, thus linking Syk and JNK activation in neutrophil. Although we were unable to ascertain from our present studies clear involvement of the Src family member, Lyn, we feel that at present we cannot discount its contribution.
Recent studies, however, have shown that another tyrosine kinase, the FAK family Pyk2, localizes to podosomes and focal adhesion-type structures in TNF␣-stimulated adherent neutrophils (23). TNF␣-induced Pyk2 phosphorylation in neutrophils is decreased in the presence of ␤ 2 integrin blocking antibodies (30) and Pyk2 integrates signals from integrins and TNFR (23). In B-cells, aggregation of ␤ 1 integrins (86) and in T-cells, ␤ 3 integrins (87) results in the phosphorylation of Pyk2. Our studies in adherent neutrophils show both phosphorylation of immunoprecipitated Pyk2 following TNF␣ stimulation and inhi-bition of JNK activation by the tyrphostin A9, shown to inhibit specifically Pyk2 (23). These observations provide preliminary evidence that Pyk2 may be associated with JNK activation in the neutrophil and are in agreement with studies in which Pyk2 overexpression in 293 cells leads to activation of coexpressed JNK, whereas dominant-negative Pyk2 inhibits JNK in PC12 (31). Pyk2 has been linked to JNK via p130 Cas and p130 Crk (87,88). Whereas in THP1 cells the PI3-kinase pathway is associated with Pyk2 (89), it is not necessary for Pyk2 activation in neutrophils (30). This observation agrees with our findings that wortmannin does not affect TNF␣-induced JNK activation in neutrophils. Our studies on Syk and Pyk2, together with our CD11b blocking studies in TNF␣induced JNK activation in the neutrophil, further strengthen our hypothesis that the JNK pathway may require two interacting signals, one of which involves ␤ 2 integrins whose actions are mediated in part by activation of tyrosine kinases. However, the TNFR-and integrin-mediated signals may not necessarily be independent, since they may reflect the action by TNF␣ to modify integrin function (19) as demonstrated by both the capability of TNF␣ to induce up-regulation of CD18/CD11b (Fig. 3C) and stimulate ␤ 2 integrin-mediated adherence in neutrophils (90). Since both antibodies to ␤ 2 integrins and piceatannol inhibited MEKK1 activation ( Fig. 5 and Fig. 8), these data indicate that protein tyrosine kinases and ␤ 2 integrins effects impinge on the JNK pathway upstream of MEKK1.
In the neutrophil, as in many other cells, TNF␣ enhances apoptosis. We have shown herein that in adherent neutrophils TNF␣ induces apoptosis with a rapid time course comparable to that induced by UV exposure (91). Whereas recent studies invoke a role for JNK as a co-participant of apoptosis in response to ionizing radiation, UV, TNF␣ (38), and trophic factor deprivation (92), the role of JNK in apoptosis may be stimulusand cell-specific. Certain studies suggest the possibility that JNK lies upstream of caspase-3 (93). Our studies on caspase-3 cleavage and annexin V binding demonstrated the induction of apoptosis in TNF␣-stimulated adherent neutrophils.
Walzog and colleagues (37) have previously suggested that integrin involvement is important in TNF␣-induced apoptosis and, furthermore, that tyrosine kinase inhibitors modulate the induction of apoptosis. Similarly, it has been demonstrated that wound macrophages induce neutrophil apoptosis through a mechanism requiring both TNF␣ and integrin-ligand interactions (94). Of considerable interest, Coxon and colleagues (95) have reported that CD11b Ϫ/Ϫ mice exhibit an increase in elicited neutrophils, ascribed, in part, to a failure of apoptosis. Similarly, we found that preincubation of neutrophils with F(ab)Ј 2 fragments of anti-CD11b partially inhibited annexin V binding. Furthermore, a recent report that overexpression of Pyk2 leads to apoptosis in rat and mouse fibroblasts (27) supports our studies in which caspase-3 cleavage is both attenuated and delayed following preincubation of neutrophils with tyrphostin A9. These effects on caspase-3 cleavage were induced by the same concentrations of tyrphostin A9 that inhibited JNK activation.
While TNF␣ alone is a potent inducer of apoptosis in transformed cells, an additional insult is required to induce apoptosis in normal cells (e.g. cycloheximide). We propose that in human neutrophils, integrin involvement is required to induce apoptosis in the presence of TNF␣. Hence, unlike many cells, in which integrin-mediated detachment from substratum leads to apoptosis (anoikis), we have found, in contrast, that the neutrophil requires integrin-mediated attachment for apoptosis to occur. As we have previously reported, UV-induced apoptosis of neutrophils involves the MAP kinase family member p38 (91). Thus, neutrophil apoptosis may involve different MAP kinase family members for apoptotic responses under diverse conditions.
The acute inflammatory response is characterized both by large numbers of adherent neutrophils and the presence of TNF␣. It is precisely these conditions, as we have demonstrated here, that allow the neutrophil to utilize the JNK pathway. The consequences of such activation including apoptosis may act to down-regulate the intensity of the inflammatory response. Although further studies, including investigations using genetically manipulated mice, are required to determine the relationship between the JNK pathway and apoptosis, this complex activation sequence provides a mechanism for enabling the neutrophil to both acquire specific functional capabilities and to ensure they are only used transiently.