Inhibition of tumor necrosis factor signal transduction in endothelial cells by dimethylaminopurine.

Tumor necrosis factor (TNF) promotes diverse responses in endothelial cells that are important to the host response to infections and malignancies; however, less is known of the postreceptor events important to TNF action in endothelial cells than in many other cell types. Since phosphorylation cascades are implicated in cytokine signaling, the effects of the protein kinase inhibitor dimethylaminopurine (DMAP) on TNF action in bovine aortic endothelial cells (BAEC) were investigated. In BAEC, TNF promotes phosphorylation of eukaryotic initiation factor 4E (eIF-4E), c-Jun N-terminal kinase (JNK) and ceramide-activated protein kinase activities, Jun-b expression, prostacyclin production, and, when protein synthesis is inhibited, cytotoxicity. DMAP abrogated or significantly attenuated each of these responses to TNF, without affecting the specific binding of TNF to its receptors. Histamine, another agent active in the endothelium, promotes phosphorylation of elongation factor-2 (EF-2) and prostacyclin production, but not phosphorylation of eIF-4E in BAEC. Histamine-stimulated EF-2 phosphorylation was not inhibited and prostacyclin production was unaffected by DMAP. These observations demonstrate that a distinct signal transduction cascade, which can be selectively inhibited by DMAP, promotes the response of BAEC to TNF. Thus, we have identified a reagent, DMAP, that may be useful for characterizing the TNF signal transduction pathway.

Tumor necrosis factor-␣ (TNF) 1 is a potent multifunctional hormone produced predominantly by activated macrophages (1,2). TNF induces the hemorrhagic necrosis and regression of tumors in animals, is cytotoxic to transformed cells, and also promotes immunity, inflammation, insulin resistance, hypotension, shock, and, in some chronic diseases, the syndrome of wasting and malnutrition known as cachexia (1,2). The endothelium is an important target for TNF action (reviewed in Refs. 3 and 4). The oncolytic activity of TNF is largely mediated indirectly, through the induction of an inflammatory state and procoagulant activity in tumor capillary beds. TNF increases the expression of major histocompatibility class I antigens and stimulates the production of chemotactic factors and other cytokines by the endothelium, thereby promoting immunity. Additionally, TNF induces expression of cell adhesion molecules, such as ICAM-1 and E-selectin, on the surface of endothelial cells. The acquisition of surface adhesive properties by neutrophils and lymphocytes is important for inflammatory processes dependent on transendothelial cell migration. TNF also alters the barrier function of the endothelium and increases the permeability of endothelial cell monolayers to macromolecules.
TNF initiates its diverse actions by binding to either of two receptors (5,6). The extracellular domains of the receptors share homologies with one another, whereas the intracellular domains do not; thus, it is not surprising that the receptors induce distinct responses (7)(8)(9)(10)(11)(12)(13)(14). Neither TNF receptor contains intrinsic protein tyrosine kinase activity or any recognizable motif, which suggests a biochemical activity (5,6). Receptors that do not contain enzyme activity interact with accessory proteins, which couple the receptor to signaling pathways (15,16). For this reason, recent studies from a number of laboratories, including our own, have been directed at identifying proximal elements in the TNF signal transduction cascade, most particularly proteins that bind directly to either TNF receptor and may promote cellular responses (17)(18)(19)(20)(21)(22)(23)(24).
As important as understanding how TNF receptors couple to effector mechanisms is characterization of downstream elements in the signaling cascade that promote responses. As with other hormones, the phosphorylation of specific target proteins resulting in the production of mediators, alterations of gene expression, and changes of cell function and/or viability is associated with TNF action in some cell types (25), although the endothelial cell has not been particularly well investigated. In the present study, the protein kinase inhibitor dimethylaminopurine (DMAP) (26 -29) was used to study TNF signaling in bovine aortic endothelial cells (BAEC), as it displays a unique scope of action that distinguishes it from other protein kinase inhibitors. DMAP induces premature mitosis in mammalian cells (28) and inhibits a protein kinase that supports the maturation of starfish oocytes (29). An analog of DMAP, 2-aminopurine, blocks the induction of human ␤-interferon and c-Fos and c-Myc expression by virus in various cell lines (30).
Here, we report that TNF induces a distinct signal transduction cascade in BAEC that can be distinguished from that employed by histamine and is inhibited selectively by DMAP.

EXPERIMENTAL PROCEDURES
Materials-Recombinant human TNF-␣ (5 ϫ 10 7 units/mg) was a gift from Genentech. [␣-32 P]dCTP (3000 Ci/mmol) and carrier-free Na 125 I were from Amersham. Sephadex G-25 (fine) and oligolabeling kits were from Pharmacia. BA-S supported nitrocellulose membranes were from Schleicher & Schuell. DMAP and cycloheximide were from Sigma. Other materials were from standard suppliers.
Cell Culture-Cells were grown on gelatin-coated dishes. To prepare these, a 0.2% gelatin solution was added to dishes, which were allowed to stand at room temperature for 1 h. The gelatin solution was then removed by aspiration. BAEC were maintained in Medium 199 supplemented with 20% heat-inactivated fetal bovine serum, 20 mM Hepes, pH 7.4, L-glutamine, and 100 units/ml penicillin and streptomycin (31). HUVECs obtained by the method of Jaffe et al. (32) were maintained in Medium 199, 20 mM Hepes, pH 7.4, supplemented with 16% fetal bovine serum and 4% human serum, 1 ng/ml fibroblast growth factor 1␣, 15 units/ml heparin, 1.6 mM L-glutamine, and 80 units/ml penicillin and streptomycin. Cells were maintained at 37°C in a humidified 5% CO 2 , 95% air incubator.
Phosphorylation and Purification of eIF-4E-BAEC were washed twice with phosphate-free Dulbecco's modified Eagle's medium supplemented with bovine serum albumin (10 mg/ml), bacitracin (0.8 mg/ml), and gentamicin (50 g/ml) (buffer A). The cells were equilibrated in buffer A containing 15 mM glucose and carrier-free [ 32 P](ortho)phosphate (0.2-0.3 mCi) for 45 min at 37°C, treated with vehicle or 0 to 10 mM DMAP for 15 min at 37°C, and then incubated in the absence or presence of TNF (1 nM, 10 min). eIF-4E from control or TNF-treated cells was purified by affinity chromatography on m 7 GTP-Sepharose at 4°C as described previously (33). In brief, cells were solubilized into 1 ml of KHEGM (100 mM KCL, 20 mM Hepes, pH 7.5, 0.2 mM EDTA, 10% glycerol (v/v), 7 mM 2-mercaptoethanol, and 0.5% Triton X-100). Nuclei and cellular debris were pelleted by centrifugation (10,000 ϫ g, 5 min), and supernatants were incubated with 50 l of m 7 GTP-Sepharose in 1 ml of KHEGM (1 h). After centrifugation (10,000 ϫ g, 1 min), the supernatant was aspirated, the Sepharose was resuspended with KHEGM (1 ml), shaken for 15 min, and again pelleted. This wash was repeated three times and was followed by incubation of the m 7 GTP-Sepharose with GTP (1 mM, 1 h) to further dissociate nonspecifically bound proteins. Specifically bound proteins were eluted by incubation with 50 l of m 7 GTP (75 M, 1 h) and fractionated by SDS-PAGE.
Gel Electrophoresis and Autoradiography-Equal amounts of protein from incubates were fractionated by SDS-PAGE on 11% linear gels as described (33). After electrophoresis, dried gels were exposed to Kodak XAR5 film for 3-7 days at Ϫ80°C in cassettes containing Dupont Cronex Lightning Plus enhancing screens. To quantitate the effect of TNF on phosphorylation of eIF-4E, autoradiograms were scanned using an LKB Ultroscan XL laser densitometer. Results are representative of three experiments.
Ceramide-activated Protein Kinase (CAP Kinase) Activity-BAEC incubated in serum-free Dulbecco's modified Eagle's medium/1% bovine serum albumin for 4 h were treated with vehicle, 1 nM TNF, 1 nM TNF together with 1 mM DMAP, or 1 mM DMAP by itself for 15 min at 37°C. Cells were scraped from the culture plates into 25 mM Hepes, pH 7.4, 5 mM EGTA, 50 mM NaF, 10 g/ml leupeptin, 10 g/ml soybean trypsin inhibitor and homogenized with a tight fitting Dounce homogenizer. The homogenate was centrifuged at 1,000 ϫ g for 10 min, and the supernatant was centrifuged at 250,000 ϫ g for 30 min. To eliminate cytosolic contaminants, the membrane pellet was washed with Hepes buffer and recentrifuged at 250,000 ϫ g for 30 min. The microsomal membranes (150 g/sample) thus isolated were taken into Laemmli sample medium and fractionated on 7.5% polyacrylamide gels. Renaturation of the CAP kinase activity was performed as described previously (34). Briefly, the gel was fixed in buffer A (50 mM Tris, pH 7.4, 5 mM 2-mercaptoethanol) containing 20% isopropyl alcohol at room temperature for 1 h, denatured in buffer A containing 6 M guanidine HCl for 1 h, and renatured in buffer A containing 0.04% Tween 20 at 4°C. This buffer was changed once 1 h before the autophosphorylation reaction. At that time the gel was equilibrated for 10 min at room temperature in kinase reaction mixture (25 mM Hepes, pH 7.4, 10 mM MgCl 2 , 0.5 mM EGTA, and 5 mM NaF) and then incubated for 1 h in 4 ml of kinase reaction mixture that contained 0.175 mCi/ml [ 32 P]ATP and 25 M ATP. Autophosphorylation was terminated by removal of the reaction mix-ture; the gel was then washed once with water for 10 min and three times with 5% trichloroacetic acid, 1% sodium pyrophosphate (20 min each wash). The gel was then autoradiographed, and the autophosphorylated CAP kinase was excised and subjected to Cerenkov counting.
Prostacyclin Production-Confluent BAEC in 35-mm Petri dishes were washed three times with Hank's balanced salt solution. Cells were treated in the absence or presence of 1 mM DMAP for 15 min before addition of 1 nM TNF or 100 M histamine. After incubation at 37°C for 16 h with TNF or for 5 min with histamine, supernatants were collected and assayed by enzyme immunoassay for 6-keto-prostaglandin F 1␣ , which is the stable degradation product of prostaglandin I 2 as described (36,37).
Cellular Cytotoxicity-BAEC were incubated with TNF, cycloheximide, or TNF plus cycloheximide in the absence or presence of DMAP for 6 h at 37°C. Cells were washed twice with phosphate-buffered saline, detached from the culture plates using 0.05% trypsin, 0.5 mM EDTA, and counted in a Coulter counter. Cell viability was verified by Trypan blue exclusion.
Binding of 125 I-TNF to BAEC-125 I was coupled to TNF by using IODO-GEN. 125 I-TNF (specific activity 60 Ci/g) was separated from Na 125 I by chromatography on Sephadex G-25 (fine). 125 I-TNF was greater than 95% precipitable in 5% (w/v) trichloroacetic acid and retained greater than 80% of the antiproliferative activity of unlabeled TNF on the ME-180 line of human cervical carcinoma cells. Confluent BAEC in six-well plates were incubated for 16 h at 4°C in Hank's balanced salt solution, 1% bovine serum albumin, 1 mM phenylmethylsulfonyl fluoride, and 1 nM 125 I-TNF in the absence or presence of 500 nM TNF. Cells were washed twice with ice-cold Hank's balanced salt solution, solubilized into 1 M NaOH/1% SDS and assayed for specific binding, the difference in radioactive uptake without or with excess unlabeled TNF, in a ␥ counter.

RESULTS
We previously demonstrated that TNF rapidly (within minutes) promotes the phosphorylation of an mRNA cap-binding protein, eIF-4E, in several cell types that are important in vitro models of TNF action (38). Phosphorylation promotes interac-tion of eIF-4E with the 7-methylguanosine cap found on the 5Ј end of eukaryotic mRNAs and with other initiation factors (39). The complexes produced interact with ribosomes where translation initiates. Thus, by activating eIF-4E, TNF may alter the pattern of protein synthesis and the function of cells. In the present study, we evaluated the effects of DMAP on this early cellular response to TNF. To accomplish this, the ATP pools of BAEC were equilibrated with 32 P during a 45-min incubation at 37°C and then stimulated with 1 nM TNF in the absence or presence of various concentrations of DMAP. eIF-4E was isolated from incubates by affinity chromatography on M 7 GTP-Sepharose, fractionated from other proteins by SDS-PAGE, and detected after autoradiography. As illustrated in Fig. 1, DMAP suppressed TNF-promoted phosphorylation of eIF-4E in a concentration-dependent manner. Inhibition of TNF-promoted phosphorylation was manifested between 0.5 and 10 mM DMAP; these concentrations of DMAP diminished phosphorylation of eIF-4E such that it was lower than the basal level observed in control cells. A very low concentration of DMAP (0.1 mM) had no effect on TNF-promoted phosphorylation of eIF-4E.
The sphingomyelin pathway is a signal transduction pathway that mediates some cellular responses to TNF (40,41). In this system, TNF induces hydrolysis of plasma membrane sphingomyelin to ceramide by the action of a neutral sphingomyelinase. Ceramide then acts as a second messenger, stimulating a membrane-bound serine/threonine kinase, CAP kinase, which plays a central role in signal transduction. Recent studies suggest that CAP kinase phosphorylates and plays a role in activating Raf (42). These observations led us to test whether TNF activates CAP kinase in BAEC and whether this process could be inhibited by DMAP. As shown in Fig. 2, stimulation of BAEC with 1 nM TNF activates CAP kinase. Such activation was inhibited by pretreatment of BAEC with 1 mM DMAP for 10 min.
The p46/p54 serine/threonine kinases, JNKs, are emerging members of the MAP kinase family which promote phosphorylation of the N-terminal activation domain of the transcription factor c-Jun (43). These kinases are activated by UV radiation and proinflammatory cytokines, including TNF (43,44). To determine whether TNF activates such kinases in BAEC and whether such activation could be inhibited by DMAP, these cells were treated with vehicle or DMAP and then stimulated with 10 nM TNF. JNK activity was then measured by a solid phase kinase assay using glutathione S-transferase-c-Jun as substrate. As shown in Fig. 3, TNF augmented JNK kinase activity in BAEC and this effect was substantially inhibited by 1 mM DMAP.
We (45) and others (see Ref. 46 and references therein) have shown that TNF promotes expression of early response genes in diverse cell lines. Experiments were performed to determine whether TNF promotes such gene expression in BAEC and whether DMAP might suppress such a response. Fig. 4 shows that incubation of BAEC with TNF augmented expression of Jun-b within 30 -60 min. Pretreatment of cells with 1 mM DMAP for 15 min abrogated the ability of TNF to promote Jun-b expression.
Prostacyclin is a labile prostaglandin, which inhibits platelet aggregation and serotonin release and also promotes vasodilation (47,48). One of the important responses induced by TNF in the endothelium is production of prostacyclin (49,50). For this reason, we tested whether DMAP affects this event. As shown in Fig. 5A, treatment with TNF for 16 h induces a 4-fold increase in prostaglandin production by BAEC. By itself, incubation with DMAP had no effect on the production of prostacyclin. However, coincubation with DMAP (1 mM) abrogated the ability of 1 nM TNF to elicit this response. In a separate experiment, histamine (100 M) promoted a 4-fold increase in prostaglandin synthesis by BAEC, but this process was not inhibited by DMAP (Fig. 5B). The ability of DMAP to affect signaling induced by TNF, but not by histamine, demonstrates that the effects of DMAP on responses to TNF are specific. These observations also suggest that TNF and histamine do not act through identical postreceptor signaling pathways in BAEC. Consistent with this suggestion are observations show- ing that prostacyclin induction by histamine reaches a maximal level within minutes (51), whereas induction following treatment of endothelial cells with TNF initiates within minutes (52), but then continues to increase for at least 16 h (49). Additionally, in contrast to TNF, histamine is unable to promote eIF-4E phosphorylation (data not shown), but does induce phosphorylation of elongation factor-2 (EF-2) (53). This latter observation led us to test whether DMAP could suppress the ability of histamine to promote EF-2 phosphorylation. To accomplish this, BAEC were treated for various times with histamine (0.1 mM), DMAP (1 mM), or histamine together with DMAP. Proteins in cell lysates were then fractionated by SDS-PAGE and transferred to nitrocellulose which was probed with an antibody that recognizes phosphorylated, but not nonphosphorylated, EF-2. 2 This led to the surprising observation that DMAP as well as histamine promoted a time-dependent increase of EF-2 phosphorylation (Fig. 6, bottom). To ensure that this observation was not unique to BAEC, the experiment was repeated with HUVEC with similar results (Fig. 6, top). In neither BAEC nor HUVEC did DMAP attenuate EF-2 phosphorylation induced by histamine; rather, the combination of the two reagents further increased EF-2 phosphorylation. These observations reinforce the view that DMAP selectively acts on components of the TNF signal transduction pathway and also indicate that this reagent may be useful in promoting some cellular responses to histamine by augmenting postreceptor events.
TNF is variously reported to have no effect on the viability of endothelial cells or to induce cytostasis or apoptosis (54 -56). The response to TNF may depend on the number of times that endothelial cell cultures have been passed (56), or other as yet unidentified variables. In our hands, TNF, at concentrations up to 1 nM and over a period of 6 h, did not elicit cytotoxicity or inhibit the proliferation of BAEC (data not shown). This observation was not unexpected as many non-transformed cells contain salvage proteins that protect them from TNF-induced cytotoxicity. Consistent with this is the demonstration that inhibition of protein synthesis sensitizes cells to TNF-induced cytotoxicity, presumptively through loss of protective proteins (54,57). We now report that whereas TNF or cycloheximide, an inhibitor of protein synthesis, by themselves did not induce a cytotoxic response from BAEC, coincubation of cells with both 2 A. C. Nairn, manuscript in preparation.  1 and 2) or presence (lanes 3 and 4) of 1 mM DMAP for 15 min at 37°C and then treated with vehicle (lanes 1 and 3) or 10 nM TNF (lanes 2 and 4) for 5 min at 37°C. JNK kinase in cell lysates was precipitated using glutathione S-transferase-c-Jun, which served as substrate, as described under "Experimental Procedures." After addition of Laemmli sample buffer, protein fractionation on 12% polyacrylamide gels and autoradiography, JNK kinase activity was defined by phosphoimage analysis. Proteins from cells lysed in SDS were fractionated by SDS-PAGE and transferred to nitrocellulose. For HUVEC and BAEC, all samples were run on the same gel and transferred to the same blot to overcome blot variability with ECL. Two separate controls representing the high and low response observed in the absence of any treatment are also illustrated to account for sample variation. The Western blots were then probed with an antibody raised to the major phosphorylation site in EF-2, threonine 56. agents did induce cytotoxicity. This effect was manifest within 1 h, and a near complete loss of cell viability was observed within 6 h (90% cell kill). These observations permitted us to determine whether DMAP was capable of inhibiting the cytotoxic response elicited from BAEC by TNF and cycloheximide. As illustrated in Fig. 7, DMAP substantially attenuated TNF/ cycloheximide-promoted cytotoxicity in BAEC and augmented cell viability from 14% to about 50%.
The type 1 TNF receptor associates with a protein serine kinase, the TNF receptor-associated kinase, in various cells (17,18). In additional experiments (data not shown), we found that a protein kinase associates with type 1 TNF receptor in BAEC and that DMAP inhibited this activity. However, we were unable to demonstrate that TNF could stimulate the TNF receptor associated kinase in BAEC, apparently due to high basal activity.
To determine whether the effects of DMAP on TNF signaling result from alterations of receptor expression or changes in post-receptor activities, the effect of various concentrations of DMAP on the specific binding of 125 I-TNF to BAEC was evaluated. In the experiment illustrated by Fig. 8, cells were incubated with DMAP (0.05-1 mM) for 6 h before the specific binding of 125 I-TNF to the cell cultures was assayed. DMAP did not significantly affect 125 I-TNF binding by BAEC, which suggests that its effects result from actions at postreceptor sites. DISCUSSION The mechanisms that transduce TNF binding into responses are incompletely defined. In a recent, comprehensive study of protein phosphorylation in fibroblasts, TNF was shown to induce alterations in the phosphorylation state of a diverse group of proteins, increasing the phosphorylation of some and decreasing the phosphorylation of others (58). Heat shock, epidermal growth factor, bradykinin, mellitin, and activators of protein kinase C and protein kinase A induce distinct patterns of protein phosphorylation, none of which are similar to the phosphorylations induced by TNF and interleukin-1, which share many common activities (58). These observations suggest that these cytokines can activate unique signaling mechanisms. Consistent with this conclusion are observations showing that in addition to activating protein kinase C (59), protein kinase A (60), and mitogen-activated protein kinases (61), which are implicated in the actions of many hormones, TNF also activates a group of novel kinases, including the JNK kinases (44), a ceramide-activated protein kinase (41,42), and TNF receptor-associated serine/threonine kinases (17)(18)(19).
Studies with inhibitors also point to a role for phosphorylation/dephosphorylation reactions in TNF action. Staurosporine, a protein kinase inhibitor, sensitizes tumor cell lines to TNF-mediated cytotoxicity and also to TNF-promoted activation of phospholipases C and D and the transcription factor NF-B (62). These effects are apparently mediated through actions on a unique intracellular target, as other protein kinase inhibitors do not produce similar synergy with TNF. Okadaic acid, a serine and threonine phosphatase inhibitor, mimics the changes of protein phosphorylation and gene transcription induced by TNF in fibroblasts (63), but inhibits TNF-mediated cytotoxicity in transformed cells (64), an effect also observed with the tyrosine phosphatase inhibitor orthovanadate (64). The observations summarized above suggest that a tightly regulated network of cellular kinases and phosphatases affect TNF action through both positive and negative regulation.
The present study tested whether protein kinase activity is associated with TNF signaling in endothelial cells. To accomplish this we characterized signaling in BAEC and showed that TNF enhances the phosphorylation of eIF-4E, activates JNK and CAP kinases, augments Jun-b expression, elevates the production and secretion of prostacyclin, and, when protein synthesis is inhibited, induces cytotoxicity. The protein kinase inhibitor DMAP was tested for its effect on these events and abrogated or significantly attenuated each.
The cellular response to TNF and the effects of DMAP on the cellular response to TNF were specific. A number of observations support these conclusions. First, TNF, but not histamine, induces phosphorylation of eIF-4E, indicating that there are differences in the kinase cascades induced by these reagents. Second, TNF and histamine induce prostacyclin production; however, DMAP inhibits induction of prostacyclin by TNF but not by histamine. Third, histamine, but not TNF, promotes phosphorylation of EF-2. This phosphorylation was not inhibited by DMAP; in fact, this unique reagent by itself promoted EF-2 phosphorylation. Why DMAP increases EF-2 phosphorylation is unclear. Since EF-2 is a specific substrate for a novel calcium/calmodulin-dependent protein kinase (65), one possible interpretation of this effect is that DMAP increases intracellular calcium; however, this remains to be tested as do the effects of DMAP on cellular protein synthesis.
Regardless of its mechanism of action, the observations presented in this report show that DMAP may be useful in delineating the TNF signaling pathway. The presence of unique elements in the TNF signal transduction cascade, which are FIG. 7. Effect of DMAP on cytotoxicity. BAEC were treated for 6 h at 37°C with 1 mM DMAP in the absence or presence of TNF (1 nM), cycloheximide (10 g/ml), or TNF (1 nM) ϩ cycloheximide (10 g/ml), and cell number was assayed using a Coulter counter. Data are shown as the mean Ϯ S.E., n ϭ 3. amenable to selective inhibition by reagents such as DMAP, may prove to be useful targets in developing therapies for pathologies associated with cytokine elaboration. Additionally, DMAP may prove useful in investigations of histamine action and histamine signal transduction in the endothelium.