MEKK4 Stimulation of p38 and JNK Activity Is Negatively Regulated by GSK3β*

The MAPK kinase kinase MEKK4 is required for neurulation and skeletal patterning during mouse development. MEKK4 phosphorylates and activates MKK4/MKK7 and MKK3/MKK6 leading to the activation of JNK and p38, respectively. MEKK4 is believed to be auto-inhibited, and its interaction with other proteins controls its dimerization and activation. TRAF4, GADD45, and Axin each bind and activate MEKK4, with TRAF4 and Axin binding to the kinase domain and GADD45 binding within the N-terminal regulatory domain. Here we show that similar to the interaction with TRAF4 and Axin, the kinase domain of MEKK4 interacts with the multifunctional serine/threonine kinase GSK3β. GSK3β binding to MEKK4 blocks MEKK4 dimerization that is required for MEKK4 activation, effectively inhibiting MEKK4 stimulation of the JNK and p38 MAPK pathways. Inhibition of GSK3β kinase activity with SB216763 results in enhanced MEKK4 kinase activity and increased JNK and p38 activation, indicating that an active state of GSK3β is required for binding and inhibition of MEKK4 dimerization. Furthermore, GSK3β phosphorylates specific serines and threonines in the N terminus of MEKK4. Together, these findings demonstrate that GSK3β binds to the kinase domain of MEKK4 and regulates MEKK4 dimerization. However, unlike TRAF4, Axin, and GADD45, GSK3β inhibits MEKK4 activity and prevents its activation of JNK and p38. Thus, control of MEKK4 dimerization is regulated both positively and negatively by its interaction with specific proteins.

The MAPK kinase kinase MEKK4 is required for neurulation and skeletal patterning during mouse development. MEKK4 phosphorylates and activates MKK4/MKK7 and MKK3/MKK6 leading to the activation of JNK and p38, respectively. MEKK4 is believed to be auto-inhibited, and its interaction with other proteins controls its dimerization and activation. TRAF4, GADD45, and Axin each bind and activate MEKK4, with TRAF4 and Axin binding to the kinase domain and GADD45 binding within the N-terminal regulatory domain. Here we show that similar to the interaction with TRAF4 and Axin, the kinase domain of MEKK4 interacts with the multifunctional serine/ threonine kinase GSK3␤. GSK3␤ binding to MEKK4 blocks MEKK4 dimerization that is required for MEKK4 activation, effectively inhibiting MEKK4 stimulation of the JNK and p38 MAPK pathways. Inhibition of GSK3␤ kinase activity with SB216763 results in enhanced MEKK4 kinase activity and increased JNK and p38 activation, indicating that an active state of GSK3␤ is required for binding and inhibition of MEKK4 dimerization. Furthermore, GSK3␤ phosphorylates specific serines and threonines in the N terminus of MEKK4. Together, these findings demonstrate that GSK3␤ binds to the kinase domain of MEKK4 and regulates MEKK4 dimerization. However, unlike TRAF4, Axin, and GADD45, GSK3␤ inhibits MEKK4 activity and prevents its activation of JNK and p38. Thus, control of MEKK4 dimerization is regulated both positively and negatively by its interaction with specific proteins.
MEKK4 is a 180-kDa mitogen-activated protein kinase kinase kinase (MAP3K) 2 that functions in neurulation and skeletal patterning in the developing mouse embryo (1,2). Mice harboring a kinase-inactive MEKK4 generated by knock-in mutation of the active site lysine K1361R are similar in phenotype to the knock-outs of TRAF4 and dishevelled-2 (1,3,4). TRAF4 and dishevelled-2 knock-out embryos and MEKK4 kinase-inactive knock-in embryos have significant defects in neural tube closure and skeletal patterning, including vertebral/rib malformations and scoliosis (3,4). TRAF4 directly binds MEKK4 inducing MEKK4 dimerization and activation of MEKK4 kinase activity toward MKK4/7, resulting in the activation of JNK (5). Although the phenotypes of the dishevelled-2 knock-out and MEKK4 kinase-inactive knock-in are similar, there is no evidence that dishevelled-2 interacts with MEKK4. MEKK4 and dishevelled-2 are both binding partners of Axin, a scaffold protein involved in Wnt and Notch signaling (6). The binding of MEKK4 to Axin has been shown to regulate JNK activity (7), and here we also show that MEKK4 controls Axin-dependent activation of JNK and p38. Thus, dishevelled-2 and MEKK4 are both binding partners for Axin and genetically are in a common pathway based on their strongly overlapping phenotypes when deleted or mutated in the developing mouse embryo (1,3,6).
MEKK4 is predicted to exist in an auto-inhibited state with the N terminus folded over the C-terminal kinase domain (8). Binding to MEKK4 by activators, including TRAF4, GADD45␣, -␤, and -␥, and possibly Axin is thought to release MEKK4 from its auto-inhibited state and to promote MEKK4 homodimerization (5,8,9). The requirement of MEKK4 dimerization in promoting activation was demonstrated using a chemical inducer of dimerization to activate MEKK4 (5). The fact that MEKK4 and dishevelled-2 both bind Axin and appear genetically to be in the same pathway led us to search for other Axin-interacting proteins that would potentially regulate MEKK4 activity. Analysis of the MEKK4 primary sequence identified several putative GSK3 phosphorylation sites.
GSK3 is a serine/threonine kinase that regulates multiple signaling pathways. There are two highly homologous isoforms, GSK3␣ and GSK3␤. GSK3␤ is a constitutively active kinase that is regulated by phosphorylation, localization, and GSK3␤-binding proteins (10,11). GSK3␤ is activated by phosphorylation of tyrosine 216 and negatively regulated by phosphorylation of serine 9 (10,11). GSK3␤ is a negative regulator of several signaling networks in cells. For example, GSK3 inhibits the canonical Wnt signaling pathway by promoting ␤-catenin degradation (10 -12). In addition, GSK3␤ has been shown to negatively regulate JNK activation, but the mechanism of this regulation is undefined (13). Here we show that GSK3␤ binds the kinase domain of MEKK4, inhibiting dimerization and MEKK4 activity. The findings define MEKK4 as a MAP3K regulated by GSK3␤ for the negative regulation of JNK and p38 signaling.

MATERIALS AND METHODS
Cell Lines, Culture Conditions, and Transfections-COS-7 cells were cultured in Dulbecco's modified Eagle's high glucose media supplemented with 10% fetal bovine serum, 1% penicillin and streptomycin. Transfection of COS-7 cells was performed in 35-and 60-mm dishes for 24 h using Lipofectamine Plus (Invitrogen) according to the manufacturer's specifications. Trophoblast stem (TS) cells were isolated from blastocysts as described previously (14). Blastocysts were isolated at E3.5 from timed matings of 129 SvEv mice according to university and federal guidelines for the use of animals. TS cells were cultured without feeders in 30% TS media (RPMI 1640, 20% heat-inactivated fetal bovine serum, 1% sodium pyruvate, 1% penicillin and streptomycin, 1% glutamine, and 100 M ␤-mercaptoethanol) and 70% conditioned media isolated from primary mouse embryonic fibroblasts (MEFs) cultured in TS media. Isolation of MEFs from littermate wild-type and homozygous MEKK4 K1361R MEFs was described previously (1). For treatment with GSK3 inhibitor, cells were incubated with 30 M SB216763 (Sigma).
Plasmids-Wild-type and K1361R FLAG-tagged MEKK4 were constructed as described previously (5). Full-length HAtagged MEKK4 and HA-tagged MEKK4 kinase domain were as described previously (15). HA-tagged GSK3␤ wild-type and K85M constructs were kind gifts of Xianjun Fang (Virginia Commonwealth University). Potential serine/threonine phosphorylation sites were mutated to alanines using the QuikChange multisite-directed mutagenesis kit (Stratagene). The resulting four mutated fragments were verified by sequencing and reinserted into full-length FLAG-tagged MEKK4.
Protein Phosphatase Treatment-2.5 g of His-MEKK4 purified from Sf9 cells or immunoprecipitated HA-MEKK4 from transfected COS-7 cells was incubated for 45 min at 30°C in the absence or presence of 800 units of protein phosphatase (New England Biolabs). Phosphatase reactions were terminated by washing twice with cold buffer A containing phosphatase FIGURE1.PhosphorylationofHis-MEKK4byHis-GSK3␤invitro.A,theNterminus of MEKK4 contains three putative GSK3␤ consensus sites. Consensus sequences were identified with in silico analysis using Scansite software. An underline indicates the serines and/or threonines predicted to be phosphorylated by GSK3␤. inhibitors and once with kinase buffer described above. Reactions were incubated with or without His-GSK3␤ in radioactive kinase buffer as described above.

RESULTS
GSK3␤ Phosphorylates MEKK4-The consensus sequence for GSK3␤ phosphorylation is (S/T)XXX(S/T)P ( Fig. 1A) (11). The first serine or threonine in the consensus sequence is predicted to be phosphorylated by GSK3␤. Although not absolutely required, GSK3␤ prefers substrates that are already phosphorylated on the second serine or threonine in the consensus sequence. This priming phosphorylation located four residues C-terminal to the GSK3␤ phosphorylation site is thought to increase the efficiency of GSK3␤ toward its substrates (11). Scanning of the MEKK4 sequence identified three consensus GSK3␤ phosphorylation sequences (Fig. 1A). All of the consensus sequences were located in the N terminus of MEKK4 (amino acids 1-1326), whereas none were found in the catalytic domain (amino acids 1327-1597). Two of the potential sites, Ser 77 and Thr 112 , are located approximately 50 residues upstream of the GADD45 binding domain. The third site, Ser 1237 , is located within a putative dimerization domain located 90 residues upstream of the kinase domain (9). To test the hypothesis that MEKK4 was a substrate for GSK3␤, purified recombinant GSK3␤ and MEKK4 were used. MEKK4 purified from Sf9 cells was incubated with increasing units of recombinant GSK3␤. In vitro kinase assays were performed, and phosphorylation of MEKK4 was quantitated (Fig. 1B). Addition of 2 units of GSK3␤ resulted in a 31-fold increase in MEKK4 phosphorylation demonstrating that MEKK4 is a direct phosphorylation substrate for GSK3␤. Similar results were obtained using immunoprecipitated HA-MEKK4 as a substrate for purified GSK3␤ (Fig. 1C). Kinase-inactive MEKK4 (HA-MEKK4 K1361M ) was also a phosphorylation substrate for GSK3␤ (Fig.  1C). However, the autophosphorylation of GSK3␤ was not affected by the expression of either wild-type or kinase-inactive MEKK4 K1361M (Fig. 1C).
As described above, preferred substrates for GSK3␤ are generally primed by phosphorylation of a second site by another kinase (11). To test for a role of a second site priming phosphorylation on MEKK4, purified MEKK4 was treated with -phosphatase, a protein with activity toward phosphorylated serine, threonine, and tyrosine (Fig. 1D). Subsequent GSK3␤ in vitro kinase assays with control or phosphatase-treated MEKK4 demonstrated that phosphatase-treated MEKK4 was no longer a substrate for GSK3␤, consistent with the known properties of previously defined GSK3␤ substrates (11). Identical results were obtained using -phosphatase-treated immunoprecipitated HA-MEKK4 (data not shown). Cumulatively, the findings in Fig. 1 demonstrate that MEKK4 is a prototypical in vitro GSK3␤ phosphorylation substrate.
When expressed alone, transfected MEKK4 exhibits a modest basal level of phosphorylation and kinase activity ( Fig. 2A), as predicted from studies suggesting that the N terminus of MEKK4 is auto-inhibitory (9). In vitro kinase assay of coimmunoprecipitated GSK3␤ and MEKK4 from cell lysates resulted in a 21-fold increase in the GSK3␤-dependent phosphorylation of MEKK4 compared with immunoprecipitation of MEKK4 in the absence of GSK3␤ (Fig. 2A). The addition of purified, recombinant MKK6, a MEKK4 substrate (16), to the in vitro kinase assay resulted in the expected MEKK4-dependent phosphorylation of MKK6. Strikingly, coimmunoprecipitation of GSK3␤ with MEKK4 resulted in a marked inhibition of MKK6 phosphorylation ( Fig. 2A). This phosphorylation of MKK6 is MEKK4-dependent as kinase-inactive MEKK4 K1361M that is phosphorylated by GSK3␤ was unable to phosphorylate MKK6 ( Fig. 2A).
Full-length MEKK4, the N terminus of MEKK4 upstream of the kinase domain, or the kinase domain of MEKK4 was expressed in the absence or presence of GSK3␤ (Fig. 2B). Phosphorylation of full-length MEKK4 and the N terminus of MEKK4 in the presence of GSK3␤ was similar, suggesting that GSK3␤ phosphorylates the N terminus of MEKK4 (Fig. 2B). The kinase domain of MEKK4 is weakly autophosphorylated (Fig. 2, B and C), whereas the autophosphorylation of the kinase-inactive MEKK4 K1361R kinase domain is completely inhibited (Fig. 2C). In the presence of GSK3␤, autophosphorylation of the kinase domain of MEKK4 was significantly reduced (Fig. 2, B and C), even though the MEKK4 kinase domain does not appear to be a phosphorylation substrate for GSK3␤ (Fig. 2, B and C). Consistent with this conclusion, a downward mobility shift during SDS-PAGE of the kinase domain of MEKK4 was observed in the presence of GSK3␤, suggesting that GSK3␤ inhibited the autophosphorylation of MEKK4. Fig. 2D shows that kinase-inactive GSK3␤ (GSK3␤ K85M ) does not phosphorylate MEKK4, confirming that GSK3␤ kinase activity is essential for phosphorylation of MEKK4.
Identification of MEKK4 Phosphorylation Sites for GSK3␤- Fig. 1A identified three consensus sequences in MEKK4 as potential sites for GSK3␤ phosphorylation. Phosphatase treatment of MEKK4 inhibited GSK3␤ phosphorylation of MEKK4, suggesting a priming phosphorylation is required for GSK3␤ phosphorylation of MEKK4 (Fig. 1D). Both the consensus GSK3␤ phosphorylation site and the priming site were mutated in each of the three consensus sequences (Fig. 3A). Mutation of each GSK3␤ phosphorylation consensus sequence individually resulted in a partial loss of GSK3␤-catalyzed phosphorylation of MEKK4 (Fig. 3B). Mutation of serine 1237 and 1241 in the third consensus sequence resulted in the greatest inhibition of MEKK4 phosphorylation (Fig. 3B). Mutation of multiple consensus sequences was generally additive with simultaneous mutation of all three sequences resulting in a very dramatic loss of MEKK4 phosphorylation catalyzed by GSK3␤ (Fig. 3C). These findings define three different sites of MEKK4 phosphorylation catalyzed by GSK3␤.
Endogenous GSK3␤ and MEKK4 Are Binding Partners-Cumulatively, the findings in Figs. 1-3 clearly demonstrate that MEKK4 is a GSK3␤ phosphorylation substrate. The findings in Fig. 2, A-C, and Fig. 3 are striking but seemed at first to be somewhat contradictory. First, GSK3␤ phosphorylates three sites N-terminal to the kinase domain and inhibits MEKK4 phosphorylation of MKK6. Second, GSK3␤ appears to inhibit the autophosphorylation activity of the MEKK4 kinase domain even though the MEKK4 kinase domain is not phosphorylated by GSK3␤. For these reasons, the interaction of endogenous MEKK4 and GSK3␤ was examined by coimmunoprecipitation (Fig. 4A). Unlike GSK3␤ that is expressed ubiquitously and strongly in most cell types, MEKK4 is expressed rather weakly in most adult cell types (1,17). However, MEKK4 is expressed strongly during development in both embryonic and extraembryonic cells (1,2). Therefore, the association of endogenous GSK3␤ and MEKK4 was examined using trophoblast stem (TS) cells, which are epithelial extraembryonic cells that differentiate to form the tissues of the placenta (14). Immunoprecipitation of TS cell lysates with anti-GSK3␤ antibody resulted in the specific coimmunoprecipitation of MEKK4 with GSK3␤ (Fig.  4A). MEKK4 did not coimmunoprecipitate with a nonspecific anti-rabbit antibody (Fig. 4A). These data show the specific association of endogenous GSK3␤ and MEKK4 in TS cells.
GSK3␤ Binds the Kinase Domain of MEKK4- Fig. 4B shows that FLAG-MEKK4 coimmunoprecipitated with HA-GSK3␤. Although GSK3␤ phosphorylates the N terminus of MEKK4, the recognition of the MEKK4 phosphorylation sites by GSK3␤ within the N terminus of MEKK4 is not of sufficient affinity to allow coimmunoprecipitation of the two proteins (Fig. 4C). Instead, GSK3␤ binds to the kinase domain of MEKK4, as the kinase domain of MEKK4 strongly coimmunoprecipitated with GSK3␤ (Fig. 4D). Interestingly, the MEKK4 M5 mutant protein with alanine substitutions in each of the defined GSK3␤ consensus sequences (Fig.  3) retained the ability to coimmunoprecipitate with GSK3␤ (Fig. 4E). These findings demonstrate that endogenous MEKK4 and GSK3␤ are in a coimmunoprecipitating complex and that the kinase domain of MEKK4 is the primary site for the stable interaction of GSK3␤ and MEKK4.
GSK3␤ Inhibits the Kinase Activity of MEKK4-Coexpression of GSK3␤ with full-length MEKK4 inhibits MEKK4 kinase activity ( Fig. 2A and Fig. 6A). Although the phosphorylation sites for GSK3␤ lie in the N terminus of MEKK4, GSK3␤ retains the ability to inhibit the kinase activity of the MEKK4 kinase domain in the absence of the MEKK4 N terminus (Fig. 6A). These findings demonstrate that GSK3␤ inhibits MEKK4 kinase activity through the direct binding to the kinase domain of MEKK4.
Activation of MEKK4 results in the activation of both the JNK and p38 MAPK pathways (1,15). Consistent with the inhibition of MEKK4 kinase activity in the presence of GSK3␤, MEKK4 activation of both p38 (Fig. 6B) and JNK (Fig. 6C) is inhibited by GSK3␤. The MEKK4 M5 mutant with all three GSK3␤ phosphorylation sites mutated retains the ability to activate both JNK and p38, and this activity is inhibited by GSK3␤ (Fig. 6, D and E). This finding is consistent with the retention of GSK3␤ binding to the MEKK4 M5 phosphorylation mutant (Fig. 4E).
Inhibition of GSK3 by SB216763 Enhances MEKK4 Activation of JNK and p38-For reasons that are unclear, mutation of the GSK3␤ active site lysine (GSK3␤ K85M ) impedes the binding of GSK3␤ with MEKK4 (not shown). For this reason, we tested whether an active site inhibitor of GSK3 (GSK␣ and -␤), SB216763 (18), would alter GSK3␤ regulation of MEKK4 activity. Cells coexpressing HA-MEKK4 and HA-JNK that were treated with SB216763 exhibited enhanced JNK phosphorylation compared with cells treated with carrier alone (Fig. 7A). SB216763 had no effect on JNK activation by the related MAP3K MEKK1 (Fig. 7B). SB216763 treatment had a more modest effect on MEKK4 activation of p38, as compared with MEKK4 activation of JNK. MEKK4 activation of p38 was not enhanced with 2.5 h of SB216763 treatment. However, prolonged treatment for 16 h resulted in a 50% increase in MEKK4 activation of p38 (Fig. 7C).
Enhanced MEKK4 Kinase Activity with Inhibition of GSK3 by SB216763- Fig. 7D shows that continuous exposure of cells expressing wild-type HA-MEKK4 to SB216763 over a 16-h time period resulted in enhanced MEKK4 kinase activity. HA-MEKK4 immunoprecipitated from cells treated for increasing time with SB216763 was subjected to in vitro kinase assays using kinase-inactive MKK6 as a substrate. A time-dependent increase in MEKK4 kinase activity was observed with SB216763 treatment demonstrating that inhibition of GSK3 enhanced MEKK4 kinase activity (Fig. 7D). The activity measured in the in vitro kinase assay was dependent on MEKK4, because kinase-inactive HA-MEKK4 K1361M did not have any kinase activity toward MKK6 in the assay (Fig. 7D). In addition to enhancing the kinase activity of overexpressed MEKK4, treatment with GSK3 inhibitor also increased the kinase activity of endogenous MEKK4. Endogenous MEKK4 immunoprecipitated from MEFs treated with SB216763 exhibited a 34 Ϯ 0.02% increase in kinase activity relative to treatment with carrier alone. The activation of MEKK4 resulting from inhibition of GSK3 kinase activity further demonstrates that GSK3 is a negative regulator of MEKK4 signaling. The loss of GSK3␤mediated inhibition of MEKK4 kinase activity with SB216763 treatment is consistent with the observation that kinase-inactive GSK3␤ K85M binds poorly to MEKK4, suggesting that the interaction of GSK3␤ with MEKK4 involves residues within the kinase domain of GSK3␤.
MEKK4 Mediates Axin-dependent Activation of JNK and p38-It has been demonstrated that TRAF4 and GADD45 proteins both activate MEKK4 (5,9). Fig. 8A and the work of others (7) show that the scaffolding protein, Axin, binds the kinase domain of MEKK4. Kinase-inactive MEKK4 (MEKK4 K1361R ) strongly abrogates the activation of both JNK and p38 in response to Axin expression (Fig. 8, B and C). These results show that Axin binds the kinase domain of MEKK4 and are the first to show that a kinase-inactive full-length MEKK4 inhibits Axin-mediated activation of both JNK and p38, suggesting that MEKK4 regulates Axin-dependent control of JNK and p38. GSK3␤ also binds Axin (19), demonstrating that MEKK4 and GSK3␤ are not only in a complex in cells but also interact with a common scaffold protein.

DISCUSSION
We have shown previously that chemical induction of MEKK4 dimerization activates MEKK4 kinase activity (5). In addition, we have shown that through its TRAF domain TRAF4 also induces oligomerization of MEKK4 and activates its kinase activity (5). Activation of MEKK4 by two independent mechanisms, the FKBP dimerization system and TRAF4-induced oligomerization, clearly shows that like other protein kinases MEKK4 is activated by oligomerization and transphosphorylation in the kinase domain (5). It has also been proposed that GADD45 proteins induce dimerization required for activation of MEKK4 (9). It is unclear exactly how Axin activates MEKK4, but Axin is a dimer and could potentially activate MEKK4 in a manner similar to TRAF4 and GADD45 proteins (7,20). Whereas Axin and TRAF4 bind to the C-terminal kinase domain of MEKK4, GADD45 proteins bind to an N-terminal region; thus there are multiple mechanisms to regulate MEKK4 dimerization. Our current findings demonstrate that endogenous MEKK4 and GSK3␤ are binding partners in cells, and like TRAF4 and Axin, GSK3␤ binds to the C-terminal region that encodes the kinase domain of MEKK4. The binding of GSK3␤ to the MEKK4 C-terminal kinase domain inhibits MEKK4 kinase activity by inhibiting MEKK4 dimerization. Fig. 9A shows a schematic model of the known effectors that control MEKK4 activity. It is interesting that multiple proteins that directly bind MEKK4 regulate its activity by controlling its dimerization.
GSK3␤ has been shown previously to be a negative regulator of JNK signaling (13). However, the mechanisms whereby GSK3␤ decreases JNK signaling were unknown. Here we show that GSK3␤ inhibits MEKK4 kinase activity and MEKK4 activation of JNK. The exact mechanism of how GSK3␤ inhibits MEKK4 activity is unclear, but several findings suggest that the active site of GSK3␤ might bind MEKK4. The GSK3␤ inhibitor, SB216763, binds to the active site of GSK3␤ and blocks GSK3␤ inhibition of MEKK4. Mutation of the kinase domain active site lysine in GSK3␤ (K85M) causes a marked decrease in the stable interaction of GSK3␤ and MEKK4, suggesting a functional GSK3␤ active site is required for MEKK4 binding. This hypothesis is consistent with the finding that GSK3␤ binding to the MEKK4 kinase domain inhibits MEKK4 kinase activity. However, the kinase domain of MEKK4 does not appear to be a substrate for GSK3␤-catalyzed phosphorylation, so the inhibition of MEKK4 kinase activity by GSK3␤ appears to be directly due to protein-protein interaction and not phosphorylation. The phosphorylation of the MEKK4 N-terminal residues defined in Figs. 1-3 does not have a measurable function in controlling MEKK4 inhibition by GSK3␤, and so far we have been unable to define a function for GSK3␤ phosphorylation of these sites in the MEKK4 N terminus. It is possible that phosphorylation of these sites controls the interaction of MEKK4 with other proteins such as GADD45 or Axin, but we have no evidence of this, and additional studies will be required to answer this difficult question. The MEKK4 M5 mutant with substitution of all three GSK3␤ consensus phosphorylation sites still retains a low level of GSK3␤-dependent phosphorylation. The presence of this residual phosphorylation suggests that there may be additional GSK3␤ phosphorylation sites in the N terminus of MEKK4. In addition, this residual phosphorylation of MEKK4 by GSK3␤ may be sufficient for GSK3␤-dependent regulation of MEKK4 function by phosphorylation.
The finding that GSK3␤ is a negative regulator of MEKK4 is of great importance in defining the function of MEKK4 in mouse development. MEKK4 binds Axin, TRAF4, and GSK3␤, and GSK3␤ also binds Axin (5,7,19). Our observations indicate that TRAF4 and Axin are not binding partners. 3   Immunoprecipitates were subjected to a nonradioactive in vitro kinase activity with the substrate His-MKK6. Activity was measured as described in Fig. 2. The percent inhibition of MEKK4 full-length and kinase domain only activity by GSK3␤ expression is 54 Ϯ 15 and 68 Ϯ 3, respectively. B, GSK3␤ inhibits MEKK4 activation of p38. COS-7 cells were transfected with the indicated constructs. p38 activity was monitored using a phospho-specific anti-p38 antibody. Anti-HA and anti-FLAG antibodies were used to monitor protein expression levels. The percent inhibition of MEKK4 activation of p38 by GSK3␤ expression is 38 Ϯ 10. C, GSK3␤ inhibits MEKK4 activation of JNK. COS-7 cells were transfected with the indicated constructs. JNK activity was monitored using a phospho-specific anti-JNK antibody. Anti-HA antibody was used to monitor protein expression levels. The percent inhibition of MEKK4 activation of JNK by GSK3␤ expression is 48 Ϯ 12. D, GSK3␤ retains the ability to inhibit MEKK4 phosphorylation mutant activation of JNK. Cells were treated as in C. The percent inhibition of wild-type MEKK4 and M5 mutant activation of JNK by GSK3␤ expression is 57 Ϯ 15 and 52 Ϯ 3.3, respectively. E, GSK3␤ retains the ability to inhibit MEKK4 phosphorylation mutant activation of p38. Cells were treated as in B. The percent inhibition of wild-type MEKK4 and M5 mutant activation of p38 by GSK3␤ expression is 28 Ϯ 0.5 and 16 Ϯ 24, respectively. A-E, a representative experiment from two independent experiments with similar results is shown.
there is a complex interaction of Axin, GSK3␤, MEKK4, and TRAF4, all of which are critical in early embryonic development (1, 2, 4, 21, 22), with MEKK4 acting as a signaling hub to positively control JNK and p38 activity in response to TRAF4 and Axin and negatively regulate these pathways in response to GSK3␤ (Fig. 9B). In the connections map proposed in Fig. 9B, MEKK4 is a central MAP3K for the control of signaling in the Axin-regulated Wnt signaling pathway. In addition, literature FIGURE 7. Inhibition of GSK3 with the GSK3 inhibitor SB216763 enhances MEKK4 activation of JNK. A, inhibition of GSK3 enhances MEKK4 activation of JNK. COS-7 cells transfected with the indicated constructs were treated for 2 h with carrier alone (Me 2 SO) or SB216763. Lysates were probed with either a phospho-specific anti-JNK antibody to measure JNK activity or an anti-HA antibody to measure total protein expression. A representative experiment from four independent experiments with similar results is shown. B, inhibition of GSK3 does not alter MEKK1 activation of JNK. COS-7 cells were transfected with HA-MEKK1, and cells were treated as described in A. A representative experiment from three independent experiments with similar results is shown. C, inhibition of GSK3 enhances MEKK4 activation of p38. COS-7 cells were transfected with the indicated constructs and treated as in A. Lysates were probed with a phospho-specific anti-p38 antibody to measure active p38 and an anti-HA antibody to detect HA-MEKK4 expression. A representative experiment from three independent experiments with similar results is shown. D, GSK3 inhibition with SB216763 increases MEKK4 kinase activity. COS-7 cells were transfected as indicated, and cells were treated for the indicated number of hours with either Me 2 SO or SB216763. HA-MEKK4 was immunoprecipitated with anti-HA antibody, and a radioactive kinase assay with kinase-inactive His-MKK6 as the substrate was performed. The incorporation of [␥-32 P]ATP into His-MKK6 is shown by autoradiogram and measured with a PhosphorImager. Equal immunoprecipitation of MEKK4 is shown by Western blotting with an anti-HA antibody. A representative experiment from three independent experiments with similar results is shown. The fold stimulation of MEKK4 kinase activity relative to empty vector alone expressed as the mean Ϯ S.E. for the indicated treatments is as follows: 0 h, 4.5 Ϯ 0.5; 2 h, 6.2 Ϯ 1.2; 4 h, 4.7; 16 h, 6.9 Ϯ 0.4; kinase-dead MEKK4 at 0 h, 1.1 Ϯ 0.3. searches show that DTRAF1, the Drosophila TRAF most similar to mammalian TRAF4, interacts with Misshapen, the mammalian homolog of Nck-interacting kinase (23), which indirectly may connect TRAF4 to dishevelled and Axin. Interestingly, dishevelled 1/2, Axin, TRAF4, and JNK1/2 gene knock-outs and the MEKK4 kinase-inactive gene knock-in all display similar neurulation defects during embryonic mouse development (1,3,4,21,24,25). In addition, TRAF4, dishevelled 1/2, and conditional c-Jun knock-outs and the MEKK4 kinase-inactive knock-in also have skeletal malformations (1,3,4,26). The GSK3␤ knock-out is lethal between E13.5 and E14.5 because of liver apoptosis (22). Redundant signaling by GSK3␣ is thought to compensate for the loss of GSK3␤, except in the liver (22,27). Loss of both GSK3␣ and GSK3␤ is early embry-onic lethal, so it is unclear how GSK3 function influences neurulation and vertebral/rib patterning (27). Therefore, our studies begin to provide a biochemical explanation for the genetic pathway that has been mapped by overlapping knock-out and knock-in phenotypes for MEKK4, TRAF4, disheveled 1/2, GSK3␤, and Axin. MEKK4 function is clearly critical in this signaling network, and its kinase activity is both positively and negatively regulated by proteins within the network for control of JNK and p38.