Tumor Necrosis Factor Stimulates Multiple Serine/Threonine Protein Kinases in Swiss 3T3 and L929 Cells IMPLICATION OF CASEIN KINASE-2 AND EXTRACELLULAR SIGNAL-REGULATED KINASES IN THE TUMOR NECROSIS FACTOR SIGNAL TRANSDUCTION PATHWAY*

Incubation of Swiss 3T3 or L929 cells with tumor necrosis factor (TNF) leads to the rapid stimulation of several cytosolic Ser/Thr kinases active toward myelin basic protein, the Ss peptide (RRLSSLR), the G peptide (SPQPSRRGSESSEE), and Kemptide (LRRASLG). This confirms the hypothesis that kinases other than protein kinases A and C may be involved in the TNF signal transduction. Chromatograp~y on Mono Q re- solved multiple kinase peaks with each substrate tested and moreover revealed a TNF-mediated casein kinase-2 activation in both cell lines, measurable with the specific RRREEESEEE peptide or with the G peptide. The TNF-stimulated myelin basic protein kinases-1 and -2 were identified as extracellular signal-regu-lated kinases-2 and -1, respectively, based on their elution pattern on Mono Q chromatography, their inactivation by protein phosphatase action, their reac- tion with phosphothreonine and phosphotyrosine antibodies, and by their migration on sodium dodecyl sulfate-polyacrylamide gel electrophoresis as 42- and 44-kDa proteins recognized by anti-extracellular sig- nal-regulated kinase antibodies.


Incubation of Swiss 3T3 or L929 cells with tumor necrosis factor (TNF) leads to the rapid stimulation of several cytosolic Ser/Thr kinases active toward myelin
basic protein, the Ss peptide (RRLSSLR), the G peptide (SPQPSRRGSESSEE), and Kemptide (LRRASLG). This confirms the hypothesis that kinases other than protein kinases A and C may be involved in the TNF signal transduction. Chromatograp~y on Mono Q resolved multiple kinase peaks with each substrate tested and moreover revealed a TNF-mediated casein kinase-2 activation in both cell lines, measurable with the specific RRREEESEEE peptide or with the G peptide. The TNF-stimulated myelin basic protein kinases-1 and -2 were identified as extracellular signal-regulated kinases-2 and -1, respectively, based on their elution pattern on Mono Q chromatography, their inactivation by protein phosphatase action, their reaction with phosphothreonine and phosphotyrosine antibodies, and by their migration on sodium dodecyl sulfate-polyacrylamide gel electrophoresis as 42-and 44-kDa proteins recognized by anti-extracellular signal-regulated kinase antibodies.
Tumor necrosis factor (TNF)' is a trimer, consisting of 17-kDa polypeptide subunits, produced mainly by monocytes and macrophages. It has an unusual wide range of biological activities, and depending on cell type and growth state, TNF can be either mitogenic, cytostatic, or cytotoxic. Differentiation processes are also dramatically influenced by T N F myeloid leukemia cell differentiation is promoted, but adipogenic cell differentiation is inhibited. The pleiotropic effects of this * This work was supported by grants from the Nationaal Fonds voor Geneeskundig Weten~happelijk Onderzoek and the Geconcerteerde Onderzoeksacties van het Ministerie van de Vlaamse Gemeenschap. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. cytokine make it an important mediator in processes as diverse as cachexia, septic shock, inflammation, tissue remodeling, infection, and immunity (1)(2)(3)(4).
Little is known about the molecular mechanisms responsible for the multiple biological activities of TNF. Receptors for TNF are present on almost all cells, and two types of receptor have been characterized and cloned. A 55-kDa receptor is rather ubiquitous, whereas the presence of a 75-kDa receptor seems to be more restricted to cells of hematopoietic origin. The extracellular domains of both receptor types show significant homology to the extracellular part of the nerve growth factor receptor as well as to the CD40 antigen, the OX40 antigen and the T2 antigen of the Shope fibroma virus; their cytoplasmic domains show no homology to any known protein (5)(6)(7)(8).
The pleiotropic nature of the TNF response in different cell types and conditions can only be explained by multiple signaling pathways. At the membrane level, some TNF-mediated molecular events have been reported GTPase activity and GTP binding are stimulated by TNF in HL-60 cells (9), and the involvement of phospholipase A2 in this signal transduction pathway is suggested by the release of arachidonic acid in the medium of TNF-stimulated cells (10, 11). There is evidence for an increased CAMP production and resulting protein kinase A activation (12) which may contribute to the activation of some genes by TNF, while not being essential for many of the TNF actions (13). TNF-induced activation of protein kinase C has also been demonstrated in several cell lines, but down-regulation of the protein kinase C activity does not seem to block the major TNF effects (12, 14, 151, suggesting that protein kinase C is not involved in the regulation of these actions or that alternative pathways can fully compensate for the loss of protein kinase C. Neve~heless, the earliest event occurring in many cells after TNF stimulation is a rapid rise in protein phosphorylation. TNF stimulates the phosphorylation of a 26-kDa protein in U-937 cells (a myeloid cell line) and in CRL1500 cells (a breast cancer cell line) (16), a 27-kDa heat shock protein in human fibroblasts (X), and several 28-kDa proteins in endothelial cells (17) and in Me-180 human cervical carcinoma cells (18). In HL-60 cells, a 75-kDa protein is phosphorylated both on tyrosine and on serine, whereas 70-and 42-kDa proteins are exclusively phosphorylated on tyrosine (19). In Swiss 3T3 fibroblasts, 41-and 43-kDa proteins are phosphorylated on tyrosine (20).
Since several TNF responses and/or phosphorylations were noticed to be protein kinase C-and protein kinase A-independent (12-14), we set out to investigate whether other

TNF-stimulated SerlThr Protein
Kinases 25917 kinases besides the "classical" second messenger-dependent protein kinase A and protein kinase C could be involved in the TNF signal transduction pathway.

Materials
Swiss mouse 3T3 cells were obtained from the American Type Culture Collection (Rockville, MD); the murine fibrosarcoma cell line L929, which is sensitive to the cytotoxic action of TNF, was from the Rega Institute (Leuven, Belgium). Aprotinin, pepstatin, and leupeptin were purchased from Boehringer Mannheim (Germany). Recombinant murine TNF from Escherichia coli was purified to 99% homogeneity and stored in phosphate-buffered saline at -80 "C; it contained < 13 ng/mg of endotoxin and had a specific activity of 1.9 X 10' IU/mg (10). The synthetic peptides RRLSSLR (s6 peptide), SPQPSRRGSESSEE (G peptide), RRREEESEEE (casein kinase-2 peptide), and LRRASLG (Kemptide) were synthesized with a Milligen 9050, using the N-(9-fluorenyl)methoxycarbonyl (Fmoc) mode and purified using reverse phase high performance liquid chromatography on a Delta-Pack C18 column from Waters. The protein kinase A inhibitor peptide TTYADFIASGRTGRRNAIHD (protein kinase inhibitor peptide) was obtained from Multiple Peptide System (San Diego, CA); myelin basic protein (MBP), microcystin-LR, and 3,3'diaminobenzidine tetrahydrochloride were from Sigma, and cell culture media and antibiotics were from GIBCO/Life Technologies (Paisley, U. K.). Peroxidase-conjugated secondary antibodies were obtained from Dako (Glostrup, Denmark), mouse anti-MAP kinase (extracellular signal-regulated kinase) monoclonal antibodies were from Zymed Laboratories (San Francisco), and polyvinylidene difluoride membranes were from Millipore. The PCS phosphatase C (protein phosphatase type 2AC) and the ATP,Mg-dependent protein phosphatase (protein phosphatase type 1C) catalytic subunits were purified from rabbit skeletal muscle as described in (21). The protein phosphatase activity is expressed as units of phosphorylase phosphatase, 1 unit catalyzing the release of 1 nmol of [32P]phosphate/min at 30 "C from 10 p M [32P]phosphorylase a.

Cell Culture and Preparation of Extracts
Swiss 3T3 cells were cultured as described in (22), minimizing the risks for spontaneous transformation. L929 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated calf serum, 2 mM L-glutamine, 50 units/ml nystatin, 100 units/ml penicillin, and 100 pg/ml streptomycin. When the cells reached confluence, they were starved in Dulbecco's modified Eagle's medium supplemented with 0.1% heat-inactivated newborn calf serum for 48 h. Incubation was started by the addition of 10 p1 of TNF dissolved in phosphate-buffered saline or 10 pl of phosphatebuffered saline (controls) to 10 ml of the old medium which was essentially depleted of growth promoting activity. Incubations were stopped by rinsing twice with 5 ml of ice-cold phosphate-buffered saline and once with 4 ml of ice-cold extraction buffer containing 20 mM Tris/HCl, pH 7.4, 10 mM NaF, 10 mM p-nitrophenyl phosphate, 0.2 mM Na3V04, 10 p M ammonium molybdate, 0.05% Triton X-100, 20 mM 0-glycerophosphate, 2 mM EGTA, 10 mM MgCl,, 1 mM phenylmethylsulfonyl fluoride, 2 pg/ml aprotinin, 0.5 pg/ml leupeptin, 0.7 pg/ml pepstatin, 100 nM microcystin, and 150 mM NaCl (buffer A). Cells were scraped in 0.5 ml of buffer A, homogenized in a Dounce homogenizer (pestle B), and centrifuged at 100,000 X g for 30 min (4 "C). Supernatants (0.5-0.7 mgof protein/ml) were aliquoted and immediately frozen in liquid nitrogen.

Chromatography
Anion exchange chromatography was carried out on a Mono Q HR 5/5 using a Pharmacia LKB Biotechnology Inc. FPLC system. Cytosols (6 mg of protein, 12 150-mm dishes) for chromatographic analysis were prepared essentially as described above with the exception that in buffer A the 150 mM NaCl was omitted. Cytosols were loaded at a rate of 0.5 ml/min, and after a 10-ml wash the elution of protein kinases was performed at a rate of 1 ml/min with a 50-ml gradient from 0 to 0.5 M NaCl in a buffer containing 20 mM Tris/ HCl, pH 7.4,lO mM MgClZ, 20 mM 0-glycerophosphate, 10 mM NaF, 0.2 mM Na3V04, 10 pM ammonium molybdate, 2 mM EGTA, 2% glycerol, 1 mM benzamidine, and 2 pg/ml aprotinin (buffer B). After withdrawing a 200-pl aliquot for subsequent phosphatase treatment, 8 pl of a protease/phosphatase inhibitor mixture was added to the remaining 800 p1 of each fraction producing a final concentration of 0.5 pg/ml leupeptin, 0.7 pg/ml pepstatin, 1 mM phenylmethylsulfonyl fluoride, and 100 nM microcystin. In the presence of the protease/ phosphatase inhibitor mixture, kinase activities were stable for about 1 week. All chromatographic steps were performed at 4 "C.

Characterization of Protein Kinases
Peptide Substrates Used-The s6 peptide, RRLSSLRA, which corresponds to a C-terminal amino acid sequence (232-239) of the eukaryotic ribosomal protein s 6 (23), has been used for the detection and characterization of s6 kinase activities in a variety of cellular systems (22,(24)(25)(26)(27)(28). Kemptide (LRRASLG) was also used as an alternative s 6 peptide kinase substrate. The G peptide, SsPQPSgRRGS'3ES'5SEE, which corresponds to the primary sequence surrounding the protein kinase A phosphorylation site 1 (Ser13) in the glycogen-binding subunit of the type 1 phosphatase (29,30), also contains a consensus sequence for phosphorylation by casein kinase-2 (Ser15 site). Prior phosphorylation of Ser13 would be required for the recognition of this peptide by kinase FA/glycogen synthase kinase-3 at the Ser' and Serg sites (29). The Ser13 site is also phosphorylated by an insulin-stimulated protein kinase (30), which has been identified as the 90-kDa s6 kinase-2 (28). RRREEESEEE is a synthetic peptide which is specifically phosphorylated by casein kinase-2 (31). MBP was used as a protein substrate to detect the MBP/ MAP2 kinases (22,25).
One should note that to prevent contributions of protein kinase A, protein kinase C, and Caz+/calmodulin-dependent kinases, all phosphorylation assays were performed in the presence of the protein kinase inhibitor peptide (32), the calcium-chelating agent EGTA ,and the calmodulin inhibitor calmidazolium.
Phosphatase treatment of the activated kinases separated by Mono Q FPLC column chromatography was performed as described in (22).

Preparations of Antibodies
Preparation of Anti-phosphotyrosine and Anti-phosphothreonine Antibodies-Affinity purified anti-phosphotyrosine antibodies and anti-phosphothreonine antibodies were prepared as described in (34) using as antigen keyhole limpet hemocyanin polymerized with an equimolar mixture of L-phophotyrosine, L-alanine, and L-threonine (for the anti-phosphotyrosine antibodies) or with a mixture of Lphosphothreonine, L-alanine, and L-tyrosine (for the anti-phosphothreonine antibodies) using l-ethyl-3-(3-dimethylaminopro-py1)carbodiimide as coupling agent. Rabbits were immunized with these conjugates and the antibodies purified on affinity columns of either phosphotyrosine-Sepharose or phosphothreonine-Sepharose; their specificity was confirmed in enzyme-linked immunosorbent assays using conjugates of phosphotyrosine, phosphothreonine, and phosphoserine with bovine serum albumin.
Anti-casein Kinase-2 Antibodies-Casein kinase-2 was purified to homogeneity from porcine spleen essentially as in (35) but with an additional Mono Q chromatography step to improve the purity of the enzyme. Chickens were immunized with the whole casein kinase-2 molecule, and the immunoglobulin fractions of the egg yolk were isolated as described in (36). The anti-casein kinase-2 immunoglobulins recognized all three potential subunits (a-, a'-, and P-subunits) of the casein kinase-2 enzyme.

Time-and Dose-dependent Effect of TNF on Soluble Kinase
Actiuities-Cytosols of either control Swiss 3T3 cells or cells treated for various time periods with TNF (20 ng/ml) were TNF-stimulated SerlThr Protein Kinases assayed for their ability to phosphorylate a number of peptide and protein substrates such as S6 peptide, Kemptide, G peptide, RRREEESEEE peptide, and MBP. Fig. 1 shows that kinase activities toward all but one of these substrates started t o rise as early as 2 min after the addition of TNF and peaked after 10 min of incubation with the cytokine. Similar kinetics of kinase activation were seen with L929 cells for the same set of substrates, over the 40-min time course (not shown).
To characterize the dose dependence of the protein kinase stimulations, Swiss 3T3 cells were treated for 10 min with various concentrations of TNF. As shown in Fig. 2, a maximal response was observed with concentrations of T N F around 2-20 ng/ml, corresponding to the concentration range which elicits a mitogenic response in Swiss 3T3 fibroblasts or other biological responses in different cell lines (3,9,10,15,18,20).
A similar dose dependence was observed for the same kinase activations in L929 cells (not shown).
As we have reported recently for bombesin-stimulated Swiss 3T3 cells (22), the TNF-mediated stimulation of the RRREEESEEE kinase activity (casein kinase-2) could not be measured in the cytosol of either cell line, although a clearcut stimulation of the casein kinase-2 activity using any of its substrates was revealed after the Mono Q chromatography step (see below). The reason for this discrepancy probably lies in the still unknown mechanism of activation of the casein kinase-2 enzyme as mentioned in (22). Separation of TNF-activated Protein Kinase Activities by Mono Q FPLC Chromatography-The kinase activities from control cells and cells treated for 10 min with 20 ng/ml T N F were resolved by Mono Q FPLC chromatography.   Fig. 3A shows that the S6 peptide kinase activity in the TNF-treated Swiss 3T3 cells was separated into several peaks. A major part of the activity came in the breakthrough of the column, and four overlapping activity peaks eluted early in the gradient: at 0.11,0.13,0.16, and 0.23 M NaCl, respectively. A similar pattern was seen with Kemptide as substrate (not shown).
Kinases phosphorylating the G peptide eluted in the breakthrough and at 0.13, 0.16, and 0.35 M NaCl (Fig. 3C).
As shown in Fig. 3 0 , a peak of RRREEESEEE peptide phosphorylation was observed exactly coeluting with the G peptide kinase activity at 0.35 M NaC1, so that this activity is likely due to casein kinase-2 which is known to phosphorylate both peptides. This was confirmed by the observation that the kinase activity toward both substrates was totally inhibited by heparin (not shown). Moreover, anti-casein kinase-2 antibodies recognized proteins with the exact molecular weight of the casein kinase-2 subunits in SDS-polyacrylamide gel electrophoresis blots of the Mono Q fractions of the 0.35 M NaCl kinase peak (not shown).
A very similar pattern of TNF-mediated kinase activations was seen in L929 cells (not illustrated), and the TNF-stimulated MBP kinase activities in this cell line were analyzed further. Characterization of the TNF-stimulated MBP Kinases from L929 Cells-The addition of microcystin, a powerful inhibitor of protein phosphatases, to the column fractions was absolutely required to maintain the activation of all protein kinases. This strongly suggests that a phosphorylation step is involved in the mechanism of kinase activations triggered by TNF. This idea was confirmed for the two stimulated MBP kinases from L929 cells.

TNF-stimulated SerlThr
Column fractions containing the peaks of the activated MBP kinases were subjected to PCSc phosphatase (protein phosphatase 2Ac) or ATP,Mg-dependent protein phosphatase (protein phosphatase IC) action. After 30 min of incubation in the presence of the PCSC phosphatase catalytic subunit, both MBP kinases lost their activity, and microcystin could prevent this inactivation (Fig. 4). The type 1 phosphatase catalytic subunit had no influence on these kinase activities (not shown). It can thus be concluded that activation af both MBP kinases depends upon a Ser/Thr phosphorylation reaction which is reversed by PCS phosphatase action. Identical results were obtained with the activated MBP kinases from Swiss 3T3 cells (not shown).
To characterize these two kinases further, fractions of the Mono Q comprising the activated MBP kinase-1 and -2 peaks were concentrated, subjected to SDS-polyacrylamide gel electrophoresis, blotted onto hydrated polyvinylidene difluoride membranes, and reacted with either anti-phosphothreonine, anti-phosphotyrosine, or anti-extracellular signal-regulated kinase antibodies. As shown in Fig. 5, the MBP kinase-1 activity correlated with the presence of a 42-kDa protein which was recognized by all three antibodies. Likewise, a 44-kDa protein specifically eluting with the MBP kinase-2 activity was detected by all three antibodies. This would relate MBP kinase-1 to the extracellular signal-regulated kinase-2 enzymes and MBP kinase-2 to the extracellular signal-regulated kinase-1 family (40). Western blots of the TNF-activated MBP kinase-1 and -2 mono Q elution fractions. After SDS-polyacrylamide gel electrophoresis (10% gels) and electroblotting onto polyvinylidene difluoride membranes of the concentrated column fractions, blots were incubated with panel A, 4 pg/ml of anti-phosphothreonine antibodies; panel B, 4 pg/ml of anti-phosphotyrosine antibodies; and panel C, 0.2 pg/ml of anti-extracellular signal-regulated kinase antibodies. The immunocomplexes were visualized with peroxidase-conjugated secondary antibodies and 3,3'-diaminobenzidine tetrahydrochloride as color reagent. The elution profile of MBP kinase-1 and -2 is illustrated.

DISCUSSION
The present study clearly illustrates that treatment of Swiss 3T3 and L929 cells with TNF leads to the rapid stimulation of several cytosolic Ser/Thr kinases active toward a number of peptide and protein substrates. Our data show that there are other kinases besides protein kinase A and protein kinase C involved in TNF signal transduction. This has been postulated before but never actually demonstrated.
Several groups have already reported the in uiuo phosphorylation of multiple cellular proteins in response to . The kinetics of these phosphorylations are very similar to the kinetics of the TNF-induced kinase activations observed in the present report, which could suggest that the kinases identified in our study may be involved in the reported TNF-stimulated SerlThr Protein Kinases in vivo phosphorylations. One of the most common responses observed after TNF treatment of cells is the phosphorylation of the 27-kDa heat shock protein (HSP27) (16). Heat shock or treatment of HeLa cells with arsenite, phorbol esters, or T N F results in the rapid phosphorylation of a site in HSP27 that contains a consensus sequence for the s6 kinase-2 (41). Our data provide good evidence for several TNF-mediated s6 peptide and G peptide kinase activations. Both peptide substrates contain the consensus sequence for the Ss kinase-2, and the TNF stimulation of their phosphorylation is seen in the breakthrough fractions of the Mono Q column as well as early in the gradient. This corresponds to the elution pattern of the 90-kDa S g kinase-2 enzymes, which are collectively referred to as the ribosomal Ss kinase family (42).
The activation of MBP/MAPZ kinases appears to be a very general extracellular signal-regulated phenomenon, hence their classification as extracellular signal-regulated kinases (40,43). Historically, two proteins with respective molecular masses of 42 and 44 kDa were first found to be heavily phosphorylated on tyrosine and threonine after stimulation by a number of growth factors and oncogenes (44). On the basis of their enzymatic activities as kinases that phosphorylate different in vitro substrates, they have been called microtubule-associated protein-2 kinases (MAP2 kinases) (45), myelin basic protein kinases (MBP kinases) (25), or epidermal growth factor receptor-threonine kinases (46). Finally, Boulton et al. (40) proposed to label them as extracellular signalregulated kinases because of the multiple stimuli that activate these enzymes. Activation of both the 42-and the 44-kDa kinases requires phosphorylation at tyrosine as well as at threonine residues; removal of either the threonine phosphate (by the PCS or type 2A phosphatase) or the tyrosine phosphate (by the CD45 tyrosine phosphatase) results in their inactivation (40,45). The minimal consensus amino acid sequence that is phosphorylated by these kinases has been determined as Ser/Thr followed by a proline (47). Extracellular signal-regulated kinase activation may constitute an early intracellular message which primes the cell for the more specific information that comes along the signal transduction pathway to dictate what kind of response is needed. This idea is supported by the present study which shows that TNF promotes a similar activation of MBP kinases in two different cell lines while being mitogenic in one and cytotoxic in the other. A somewhat similar observation has been made in the yeast-mating signal transduction, where the activity of the FUS3 gene product, a Ser/Thr kinase with more than 50% homology to the extracellular signal-regulated kinases-1 and -2, can either promote vegetative growth or mediate pheromone-induced differentiation and cell cycle arrest (48,49).
There are several reasons to believe that the two MBP kinases activated by T N F in the present report may represent two members of the extracellular signal-regulated kinase family. First, their elution position on Mono Q FPLC chromatography is identical to the elution position of the epidermal growth factor-stimulated extracellular signal-regulated kinases-1 and -2 (25,(37)(38)(39). Second, their specific inactivation by a type 2A phosphatase is clearly reminiscent of the regulation of the extracellular signal-regulated kinases (25,(37)(38)(39). Third, Western blotting of the MBP kinase-1 and -2 Mono Q eluate fractions revealed 42-and 44-kDa proteins that were recognized by anti-phosphothreonine and antiphosphotyrosine antibodies, as well as by anti-extracellular signal-regulated kinase antibodies. This would classify the 42-kDa MBP kinase-1 enzyme as a member of the extracellular signal-regulated kinase-2 family, and the 44-kDa MBP kinase-:! as an extracellular signal-regulated kinase-1 type en-zyme. Interestingly, Guy et al. (15) recently reported a TNFmediated increase in a cytosolic kinase activity using MAP2 as a substrate, but no attempt was made to characterize this kinase activity further. In addition, Kohno et al. (20) reported that TNF, bombesin, and platelet-derived growth factor induced the tyrosine phosphorylation of 41-and 43-kDa proteins, that could well represent activated extracellular signalregulated kinases which are known to be tyrosine phosphorylated (40,50). Bird et al. (51) recently provided evidence for the interleukin-1-mediated stimulation of an MBP kinase of the extracellular signal-regulated kinase family. Since there seems to be a reasonable similarity in the biological action and signal transduction of interleukin-1 and TNF, it is quite conceivable that the two MBP kinases carry part of the common message of interleukin-1 and TNF.
The approach using defined peptide substrates should enable us to characterize further the TNF-stimulated kinases and to use them as sensors for the upstream events in early T N F signaling.