Direct activation of the stress-activated protein kinase (SAPK) and extracellular signal-regulated protein kinase (ERK) pathways by an inducible mitogen-activated protein Kinase/ERK kinase kinase 3 (MEKK) derivative.

The extracellular signal-regulated kinase (ERK) pathway, the stress-activated protein kinase (SAPK) pathway, and the p38 pathway are three major mitogen-activated protein kinase (MAPK) cascades known to participate in the regulation of cellular responses to a variety of extracellular signals. Upstream regulatory components of these kinase cascades, the MAPK/ERK kinase kinases (MEKK), have been described in several systems. We have isolated a cDNA encoding human MEKK3. Transfected MEKK3 has the ability to activate both SAPK and ERK pathways, but does not induce p38 activity, in agreement with a previous report on murine MEKK3 (Blank, J. L., Gerwins, P., Elliott, E. M., Sather, S., and Johnson, G. L. (1996) J. Biol. Chem. 271, 5361-5368). We now demonstrate that MEKK3 activates SEK and MEK, the known kinases targeting SAPK and ERK, respectively. Utilizing an estrogen ligand-activated MEKK3 derivative, we furthermore demonstrate that MEKK3 regulates the SAPK and the ERK pathway directly. Consistent with the fact that several SAPK-inducing agents activate the transcription factor NFκB, we now show that MEKK3 also enhances transcription from an NFκB-dependent reporter gene in cotransfection assays. The ability of MEKK3 to simultaneously activate the SAPK and ERK pathways is remarkable, given that they have divergent roles in cellular homeostasis.

The extracellular signal-regulated kinase (ERK) pathway, the stress-activated protein kinase (SAPK) pathway, and the p38 pathway are three major mitogenactivated protein kinase (MAPK) cascades known to participate in the regulation of cellular responses to a variety of extracellular signals. Upstream regulatory components of these kinase cascades, the MAPK/ERK kinase kinases (MEKK), have been described in several systems. We have isolated a cDNA encoding human MEKK3. Transfected MEKK3 has the ability to activate both SAPK and ERK pathways, but does not induce p38 activity, in agreement with a previous report on murine MEKK3 (Blank, J. L., Gerwins, P., Elliott, E. M., Sather, S., and Johnson, G. L. (1996) J. Biol. Chem. 271, 5361-5368). We now demonstrate that MEKK3 activates SEK and MEK, the known kinases targeting SAPK and ERK, respectively. Utilizing an estrogen ligand-activated MEKK3 derivative, we furthermore demonstrate that MEKK3 regulates the SAPK and the ERK pathway directly. Consistent with the fact that several SAPK-inducing agents activate the transcription factor NFB, we now show that MEKK3 also enhances transcription from an NFB-dependent reporter gene in cotransfection assays. The ability of MEKK3 to simultaneously activate the SAPK and ERK pathways is remarkable, given that they have divergent roles in cellular homeostasis.
One of the major signaling systems by which cells transduce extracellular signals into intracellular responses are the mitogen-activated protein kinase (MAPK) 1 cascades. These have been characterized in a variety of organisms including yeast, the nematode Caenorhabidits elegans, Drosophila, and mammalian cells, revealing substantial evolutionary conservation (for reviews, see Refs. [1][2][3][4][5][6]. The core components of these cascades are three sequential kinases. The MAPKs are activated by dual phosphorylation on threonine (T) and tyrosine (Y) by upstream MAPK kinases (MAPKK). MAPKKs themselves are phosphorylated and thereby activated by serine/ threonine-specific MAPKK kinases (MAPKKK), a subset of which is often referred to as MEK kinases (MEKK). In yeast several parallel MAPK cascades have been characterized, mostly by genetic screens (1,2,4,6). Each kinase cascade in yeast appears to function as a linear unit which transduces a distinct signal.
A diverse array of signals and factors activating mammalian MAPK cascades have been described (10,(13)(14)(15). ERKs are often activated by growth signals, such as epidermal growth factor (EGF) or platelet-derived growth factor (29), and the molecular steps by which these factors activate ERKs have been well worked out (6, 30 -32). The growth factors effect tyrosine phosphorylation of their receptors, which in turn initiates the formation of well defined signaling complexes at the membrane. One component of this complex is the small G protein Ras, which upon GDP/GTP exchange recruits Raf-1. Raf-1 then becomes activated through phosphorylations to effect the mitogenic response to growth factor stimulation via the ERK pathway. ERK activity, however, can also be induced through receptors with seven transmembrane helices coupled to trimeric G proteins (33). Depending on the cell type, ERK activation results in either proliferation or differentiation (34). The SAPK and p38 pathways are both regulated by stress signals such as UV light or by proinflammatory cytokines, such as tumor necrosis factor-␣ (TNF-␣) or interleukin-1 (IL-1). MEKK1 appears to mediate activation of SAPK in response to * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The some of the SAPK-inducing signals (7,23,25,26). For other SAPK activity-inducing signals and for the p38 pathway, the upstream MAPKKKs are unknown. To transduce the multiplicity of MAPK-activating signals into appropriate responses, it is reasonable to postulate additional MEKKs.
Using a PCR strategy to identify mammalian homologues of a yeast MEKK, a strategy which first led to the identification of MEKK1 (7,23,35), Blank et al. (36) recently cloned additional murine MEKKs, namely MEKK2 and MEKK3. We independently cloned and characterized human (h) MEKK3 and we report here on novel functional aspects of this kinase. Expression of hMEKK3 in transfected cells potently activates the SAPK and the ERK pathway, in agreement with Blank et al. (36). Extending these observations, we show that hMEKK3 regulates SAPKs and ERKs via SEK and MEK1, respectively. Furthermore, we demonstrate that hMEKK3 directly activates these pathways. This rules out the possibility that hMEKK3 activates the SAPK and/or ERK pathways via induction of a factor, such as a cytokine. This critical result was obtained with the use of an inducible derivative of hMEKK3, which allowed for a kinetic analysis of MAPK activation. Finally we provide evidence that MEKK3 (in addition to MEKK1) is able to induce NFB-dependent transcription. This is consistent with the fact that many of the agents which activate SAPKs and/or ERKs, such as TNF-␣, IL-1, or serum, also activate the transcription factor NFB (37), and suggests that multiple MEKKs can participate in activation of NFB.

MATERIALS AND METHODS
Plasmids-A cDNA encoding amino acids 12-626 of human MEKK3 was cloned as a fusion with an unrelated cDNA during screening of a ZAPII library of activated peripheral blood T-lymphocytes (38) with probes selected via a differential display screen. Further 5Ј sequences were cloned with a 5Ј-rapid amplification of of cDNA ends kit (Life Technologies, Inc.) and fused to the original cDNA via overlap extension PCR (39). This extended cDNA encoded an in-frame methionine in the context of a Kozak sequence (40) and an in-frame stop codon located further 5Ј. It was cloned into pBluescript SK (Stratagene) to yield BSK hMEKK3-F. For expression in eukaryotic cells, the complete open reading frame was cut out from BSK hMEKK3-F as a XbaI/XhoI fragment, the ends were filled in with Klenow fragment and ligated into the Klenow-filled EcoRI site of PMT2TMCS, a modified PMT2T vector (41), to yield PMT2T hMEKK3-F. To construct PMT2T hMEKK3-CD, the C-terminal part of the catalytic domain (the ScaI/EcoRI fragment from BSK hMEKK3-F) was ligated to the N-terminal part obtained as a PCR fragment with an artificial XhoI site at the 5Ј-end and a restricted ScaI site at the 3Ј-end; the ligation product was then cloned into PMT2TXSE opened at the XhoI and EcoRI sites. PMT2TXSE is a modified PMT2T vector that supplies an N-terminal in-frame methionine (in the context of a Kozak sequence), which, upon insertion of the catalytic domain, results in the addition of amino acids MASTN N-terminal to glutamic acid at position 341 in the hMEKK3 amino acid sequence. cDNA fragments encoding murine MEKK1 and the catalytic domain of cRaf-1 (RafBXB) were recloned into PMT2T upon excision from pCEV29MEKK and RSV-RafBXB (kindly provided by S. Gutkind and U. Rapp, respectively). pCEV MEKK3CD-hERLBD encodes a fusion protein of the catalytic domain of hMEKK3 (amino acids 340 -626) and the human estrogen receptor ligand binding domain (hERLBD, amino acids 281-595). cDNA fragments encoding MEKK3-CD and hERLBD were amplified by PCR, which allowed the introduction of a BamHI site at the 5Ј-end and an XhoI site at the 3Ј-end of MEKK3-CD, and an XhoI site and a EcoRI site at the 5Ј-and 3Ј-ends of hERLBD. Fragments were cloned into pBluescript SK, and correct ligation was confirmed by sequencing. pCEV MEKK3CD-hERLBD was obtained by recloning of a BamHI/HincII fragment from BSK hMEKK3CD-hERLBD into the BamHI and Klenow-filled NotI site of pCEV29 (42), which supplies an in-frame initiator methionine. PMT2T HAp54␤, pcDNAI HA-ERK2, PMT2T HAp38, and pcDNAI HA-MEK encode the respective kinases with hemagglutinin epitope (HA) tags at their N termini. These were kindly provided by J. R. Woodgett (PMT2T HAp54␤) and S. Gutkind. pEBG-SEK, encoding SEK/JNKK fused to the C terminus of glutathione S-transferase (GST), was kindly provided by L. I. Zon. The prokaryotic expression plasmids GST c-Jun (amino acids 1-79 of c-Jun), GST ATF2 (amino acids 1-96 of ATF2), and GST HA-JNK1 (full-length JNK1 with an HA-tag at the N terminus), were used to express recombinant GST fusion proteins that were purified by glutathione-Sepharose affinity chromatography (43).
Cell Culture and Transfection-Cos7 and NIH3T3 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 1 mM glutamine, 10% fetal bovine serum and antibiotics. Lipofectamine-mediated transfections were performed according to the manufacturer's instructions (Life Technologies, Inc.), with the indicated amounts of plasmid DNA per 100-mm plate. One day after transfection, cells were serum-starved overnight in Dulbecco's modified Eagle's medium containing 0.5% fetal bovine serum. Two days after transfection, cells were stimulated as indicated and harvested. For stimulation with estrogen (Sigma), cells were not serum-starved. For CAT reporter assays, NIH3T3 cells were transfected by a calcium phosphate precipitation protocol as described previously (41), except that the cells were serum-starved as indicated, and then either left untreated or stimulated for an additional 7 h with 2000 units/ml murine TNF-␣ (Boehringer Mannheim).
Kinase Assays-Immunocomplex kinase assays using the anti-HA immunoprecipitates or GST-SEK bound to glutathione-Sepharose beads were performed at 30°C for 10 min (SAPK, ERK) or 30 min (p38) using 2 g of substrate, 10 Ci of [␥ 32 P]ATP and 50 M ATP (SAPK, ERK) or 5 M ATP (p38) in total 20 l of kinase buffer. The substrates used were GST c-Jun (1-79) for SAPK, myelin basic protein (Life Technologies, Inc.) for ERK, GST ATF2 (1-96) for p38, GST HA-JNK1 for SEK, and GST-ERK[K71A] (Upstate Biotechnology Inc.) for MEK. The reactions were terminated by boiling in SDS sample buffer and the products resolved by 12.5% SDS-polyacrylamide gel electrophoresis. After staining of the gel with Comassie stain and drying, the phosphorylated proteins were visualized by autoradiography. Phosphorylation signals were quantitated by PhosphorImager analysis (Molecular Dynamics, Inc.).

RESULTS
Cloning of Human MEKK3-A cDNA encoding a previously unknown human kinase was cloned from a cDNA library made from activated peripheral blood T lymphocytes, as described under "Materials and Methods." Comparison of the sequence of this clone with the GenBank TM data base (Blast Fileserver, National Center for Biotechnology Information) revealed high homology to MEKKs, especially to Byr2 of fission yeast and Ste11 of budding yeast. We therefore investigated the relationship of this putative MEKK to known mammalian MAPK pathways. During preparation of this manuscript, the cloning of two highly related murine MEKKs, MEKK2 and MEKK3, was reported by Blank et al. (36). Since our clone has highest homology to murine MEKK3 (see below) it appears to be the human homologue, and we refer to it hereafter as human (h) MEKK3.
Human and murine MEKK3 exhibit 96.5% identity at the amino acid level over the entire sequence, including the region N-terminal to the catalytic domain, a region which is quite divergent among distinct MEKKs. Alignment of the catalytic domains of MEKKs cloned in mammals, yeast, and the tobacco plant ( Fig. 1) reveals that MEKK3 is more closely related to the plant MEKK (N.t.NPK) and Ste11 of fission yeast (S.c.Ste11), than to mammalian MEKK1, except MEKK2.RT PCR and Northern blot analyses with RNA from various human and murine cell lines and murine tissues revealed an approximately 4.6-kilobase transcript, which appears to be ubiquitously expressed at low levels (data not shown).
MEKK3 Activates the SAPK and ERK Pathway, but Not the High Osmolarity Glycerol/p38 Pathway-To investigate which of the known MAPK cascades are regulated by MEKK3, we cotransfected NIH3T3 (Fig. 2), Cos7, and HeLa cells (data not shown) with expression vectors for MEKK3 together with HA epitope-tagged versions of SAPK p54␤, ERK2, or p38. The epitope-tagged MAPKs were then immunoprecipitated using the anti-HA antibody and tested for activity in an immunocomplex kinase assay with their preferred substrates (Fig. 2, A-C). For comparison, MEKK1 and a constitutively active form of human Raf-1, RafBXB (44), both cloned into the same expression vector as MEKK3, were also transfected with the different MAPKs. Initially we expressed either the MEKK3 catalytic domain (CD) or the full-length protein, considering the possibility that full-length MEKK3 might be inactive as observed for Raf-1 (44). But like MEKK1 (23), both the N-terminal truncation and the full-length forms of MEKK3 behaved similarly ( Fig. 2, and data not shown). MEKK3 activated SAPK activity as strongly as did MEKK1 (shown in Fig. 2A with two different concentrations of expression vectors). An additional titration from 25 to 1000 ng/plate of the MEKK3 or MEKK1 expression plasmids in Cos7 cells further supported this conclusion (data not shown). Surprisingly, both MEKK3 and MEKK1 are more efficient activators in this experiment than the known SAPK inducers anisomycin or TNF (compare lanes 4 -7 with lanes 2 and 3). As expected, RafBXB did not effect SAPK p54␤, even with high amounts of expression vector ( Fig. 2A, and data not  shown).
In contrast to the very similar activities of MEKK1 and MEKK3 toward SAPK, ERK is highly activated by MEKK3, but only marginally by MEKK1 (Fig. 2C). This difference with respect to ERK is unlikely to be due to a significant difference in expression of the two MEKKs, since both have equally high activity with respect to SAPK. By comparison, the constitutively active form of Raf-1 (RafBXB), a well known activator of ERK in response to growth factors (6, 30 -32), was a weaker activator of ERK than was MEKK3 in these transfection experiments; nevertheless, it still appeared to be 5-10 times more potent than MEKK1 (Fig. 2C). It was not possible to determine precise levels of activity of these kinases, since quantitation of the amounts of expressed protein was not feasible. p38, which can be induced by various stress signals such as exposure to arsenite (Fig. 2B, lane 2), was not significantly activated after transfection of MEKK3. As expected, RafBXB had no influence on p38 activity, whereas MEKK1 induced it, as observed by others (45).
As a further test of SAPK and ERK activation, the effect of activation on the electrophoretic mobility of these proteins was investigated. Mobility shifts induced upon activation reflect phosphorylation of threonine and tyrosine in the conserved TXY motif by the upstream MAPKKs (see Introduction). Immunoblot analysis clearly demonstrated that MEKK3 expression led to phosphorylation of ERK, much as did expression of RafBXB (Fig. 2D, right panel), whereas expression of MEKK1 did not (data not shown). In contrast, both MEKK3 and MEKK1 evoked the appearance of several hyperphosphorylated forms of the SAPK p54␤, while RafBXB did not, as expected (Fig. 2D, left panel). Interestingly, the continuous presence of MEKK3 caused an apparent decrease in the amount of SAPK p54␤ protein (Fig. 2D, lanes 4 and 6). This is evident even when p54␤ is dephosphorylated, ruling out the possibility that the apparently weaker signal might have resulted from the distribution of a stronger signal over a broader area due to the presence of multiple phosphorylated isoforms (data not shown). It is not known why the total amount of this protein appeared to decrease over time.
Thus, MEKK3 has the potential to activate both the SAPK and ERK pathways, in contrast to MEKK1, which activates SAPK strongly and p38 more weakly, and in contrast also to Raf-1, which appears to be specific for ERK. These results regarding the apparently dual activities of MEKK3 are essentially in agreement with Blank et al. (36), who also demonstrated that MEKK2 behaves similarly (MEKK3 and MEKK2 are 86.6% identical in their catalytic domains). Our work adds the direct comparisons with Raf and MEKK1 as well as the MEKK 3-induced changes in the phosphorylation of SAPK and ERK. In the following we investigate by what intermediate steps MEKK3 activates both ERK and SAPK, a question which was not previously resolved.
MEKK3 Activates the SAPK and ERK Pathways via SEK and MEK-Presently known MAPKKs for SAPKs and ERKs are SEK and MEK1/2, respectively. After coexpression of tagged SEK and MEK1 with MEKK3, MEKK1, or RafBXB, tagged SEK (Fig. 3A) and MEK1 (Fig. 3B) were immunoprecipitated and incubated in an in vitro kinase assay with their matching downstream substrates. MEKK3 activated both SEK (Fig. 3A) and MEK1 (Fig. 3B), and its strength of activation appeared similar to that of MEKK1 toward SEK and to that of RafBXB toward MEK1. Surprisingly, the MEKK1 control also activated MEK1 in these assays, despite the fact that ERK activity was only marginally increased (Fig. 2C). The reason for this latter observation is unknown but the data are in agreement with those reported by Xu et al. (46). In summary, our results establish that MEKK3 expression activates both SEK and MEK1, which in turn activate SAPK and ERK, respectively. Blank et al. (36) were unable to establish a connection between MEKK3 and SEK or MEK1, possibly because the recombinant MEKK3 protein with which they attempted to demonstrate this in vitro may not have been fully functional. An alternative hypothetical explanation of their data is that MEKK3 may regulate the SAPK and ERK pathways only indirectly, an important question addressed below.
MEKK3 Is a Direct Activator of the SAPK and ERK Pathways-In the experiments described above, the MAPKs and MAPKKs were cotransfected with constitutively active MEKK3. It is thus possible that continuous expression of MEKK3 during the relatively long time course of a transfection experiment may have induced a factor (via an unknown pathway), such as a secreted cytokine, which in turn may have been the true inducer of SAPK and/or ERK activity. An example of this scenario is the delayed activation of SAPK activity by Raf-1 in HeLa cells (23), which occurs via the Raf-1-induced synthesis of the heparin-binding EGF, the actual signal for activation of SAPK (47).
To resolve how MEKK3 activates the MAPK and SAPK pathways, we generated a MEKK3 whose activity could be regulated so that immediate responses to its induced activation could be investigated. We constructed an inducible MEKK3  derivative by fusing the estrogen receptor ligand binding domain (ERLBD) to the MEKK3 catalytic domain. This approach has been previously applied to reversibly induce the activity of a variety of proteins (48 -52). HA-p54␤ or HA-ERK2 were transiently transfected into NIH3T3 cells together with the MEKK3 CD-ERLBD chimera. Before harvest, cells were treated for periodsup to one hour with estrogen. Both SAPK and ERK activity were induced by 15 min post treatment, as demonstrated by an in vitro kinase assay (Fig. 4A), and also by Western blot analysis in the case of SAPK (Fig. 4B). In the absence of estrogen, MAPK activity was hardly detectable (Fig.  4A, lanes 1 and 5), although the MEKK3 chimera was expressed in all samples, as verified by Western analysis with an antibody against the estrogen receptor (data not shown). This demonstrates the tight negative control on kinase acitivity exerted by the estrogen ligand domain. The hyperphosphorylated forms of SAPK were also clearly evident by 15 min upon addition of estrogen (Fig. 4B, lane 2). These data demonstrate that MEKK3 is a direct activator of both the SAPK and ERK pathways. In contrast to what was observed in cells cotransfected with constitutively active MEKK3 (see Fig. 2D), the total amount of SAPK did not visibly decrease during 1 hour of stimulation with the inducible MEKK3 derivative. Therefore the decrease in the level of SAPK appears to be triggered by long term activation of either MEKK3 or SAPK itself.
Both MEKK1 and MEKK3 Activate NFB-Since many of the agents which induce SAPK and ERK activity are also known to activate the transcription factor NFB (37), we investigated whether MEKK3 or MEKK1 could induce transcription of an NFB-dependent reporter gene. NIH3T3 cells were transiently cotransfected with a B site-dependent reporter construct (HIV-B-CAT) and an MEKK expression vector. Raf-BXB was transfected for comparison, as it has been shown previously to increase NFB activity (53). Both MEKK3-CD and MEKK1 (and to a slightly lesser degree full-length MEKK3) induced NFB activity about 10-fold, 4-fold more than RafBXB in these transfections (Fig. 5). During preparation of this manuscript, MEKK1-induced activation of NFB was also reported by two other groups (54,55). Our findings suggest that MEKK3-and MEKK1-activated MAPK cascades may participate in transducing signals leading to activation of NFB, although actual intervening molecular steps are unknown since none of the kinases in these cascades are likely to phosphorylate IB␣ directly (56). DISCUSSION The biochemical characterization of human MEKK3 presented here clearly demonstrates the ability of MEKK3 to directly regulate two of the three currently identified mammalian MAPK cascades, namely the SAPK and the ERK pathways. In contrast, MEKK3 does not regulate the p38/high osmolarity glycerol pathway. MEKK3 activates SAPK via SEK and it activates ERK via MEK1/2. SEK is also the target for MEKK1-dependent SAPK activation and MEK1/2 is also the target for Raf-1-dependent ERK activation. In agreement with our results, Blank et al. (36) recently demonstrated the ability of MEKK3 (as well as MEKK2) to activate the two MAPK pathways upon exogenous expression, but the activation of neither SEK nor MEK1/2 by MEKK3 was demonstrated, nor was the question of a direct versus an indirect activation addressed. An indirect activation might have involved MEKK3induced synthesis of a factor, possibly a secreted cytokine, which in turn could have been the true activator of one or both of the MAPK pathways. A precedent for this exists in the case of a transfected Raf-1 kinase, which can activate SAPK via an induced cytokine in HeLa cells (23), but such a scenario was ruled out here by the rapid activation of the two MAPK pathways when MEKK3 kinase activity was selectively induced in cells. To generate an inducible MEKK3, the estrogen receptor ligand binding domain was fused to the catalytic domain of MEKK3, converting the constitutively active MEKK3 into an estrogen-dependent kinase. We thus conclude that MEKK3 can directly activate the MAPKs ERK and SAPK via MEK1/2 and SEK, respectively.
The demonstration that both pathways are direct targets of one kinase is particularly significant in light of the fact that the ERK and SAPK pathways have been generally associated with quite divergent biologic responses. The ERK pathway is clearly associated with growth and differentiation responses, while the SAPK pathway is associated with stress responses and apoptosis (3,23,34). Physiologic situations in which both pathways are activated include stimulation of T-cells through both the T-cell antigen receptor and the CD28 auxiliary receptor (see Ref. 64 and references therein). However, this particular case may not involve a single kinase like MEKK3. While ERK activation appears to be mediated by Raf-1 in response to stimulation through the T-cell antigen receptor (65); SAPK activation and indeed full T-cell responses depend also on the CD28 costimulus. The ability of an individual kinase (MEKK3) to simultaneously target two MAPK modules which are so distinct in their apparent biologic effects suggests a complexity in signal transduction which has not previously been recognized and for which there is as yet no precedent in yeast. MEKKs identified in yeast to date appear to function in a linear cascade in which each is generally responsive to a specific signal and in which each is dedicated to one MAPK module (5). Indeed, the kinases which channel a signal to activate a particular MAPK may be held together in a physical complex, as exemplified by the pheromone response pathway in budding yeast. In this pathway, the Ste5 protein functions as a scaffold for a multikinase complex by simultaneously binding Ste11 (a MEKK), Ste7 (a MAPKK) and Fus3 or Kss1 (MAPKs) (57). Although similar arrangements are likely in mammals, previous data as well as the data presented here show that in mammals the same MEKK can be used for different MAPKs, suggesting more versatility and a more complex regulation pattern.
Added complexity in mammalian signaling systems may also be indicated by the diversity of stimuli which regulate a particular MAPK, sometimes in a cell type-dependent manner. Agents which can activate SAPK include proinflammatory cytokines such as IL-1 and TNF-␣ (11)(12)(13), stress conditions such as UV irradiation and treatment with protein synthesis inhibitors (10,11,13), and growth factors such as EGF and nerve growth factor (13,58). Whereas TNF-␣ and UV irradiation activate SAPK in a variety of cell types, EGF activates SAPK potently in HeLa cells (23,58), but only marginally in certain fibroblast cell lines (13, 58) (data not shown). ERK activity is primarily induced by growth factors via a Ras-dependent pathway (23,59,60), but can also be induced by TNF-␣ in certain cell types (13,58). Known or suggested components upstream of the MEKKs comprise a growing family of STE20-like kinases (4, 61), low molecular weight GTP-binding proteins including Ras, Rac and Cdc42Hs (8), and trimeric G proteins (33,62,63). It is thus possible that MEKK3 (and possibly MEKK2) could be involved when a broader spectrum of activities is called for, although the exact connection to upstream signals may differ with the cell type. Alternatively, cells may be able to regulate MEKK2 and MEKK3 in such a way that they activate either the SAPK or the ERK pathway, but not both. Depending on the signal, multikinase complexes containing either the SAPK or the ERK pathway components may be formed, possibly mediated by accessory proteins (57). Which extracellular signals and intermediates ultimately regulate MEKK3 remain to be shown. The transcription factor NFB appears to be at least one possible downstream target of MEKK3. NFB plays a pivotal role in regulating genes which encode functions relevant to the immune and stress responses, as well as growth and differentiation (37). Raf-1, MEKK1 and now also MEKK3 may be intermediates in at least some of the situations in which NFB becomes activated. Elucidation of the steps linking these MEKKs and NFB activation may reveal potential targets for drug intervention in the case of pathological activation of this transcription factor in inflammatory diseases.