Phosphorylation of Serine 526 Is Required for MEKK3 Activity, and Association with 14-3-3 Blocks Dephosphorylation*

MAPK/ERK kinase kinase 3 (MEKK3) is a mitogen-activated protein kinase kinase kinase (MAP3K) that functions upstream of the MAP kinases and IκB kinase. Phosphorylation is believed to be a critical component for MEKK3-dependent signal transduction, but little is known about the phosphorylation sites of this MAP3K. To address this question, point mutations were introduced in the activation loop (T-loop), substituting alanine for serine or threonine, and the mutants were transfected into HEK293 Epstein-Barr virus nuclear antigen cells. MEKK3-dependent activation of an NF-κB reporter gene as well as ERK, JNK, and p38 MAP kinases correlated with a requirement for serine at position 526. Constitutively active mutants of MEKK3, consisting of S526D and S526E, were capable of activating a NF-κB luciferase reporter gene as well as ERK and MEK, suggesting that a negative charge at Ser526 was necessary for MEKK3 activity and implicating Ser526 as a phosphorylation site. An antibody was developed that specifically recognized phospho-Ser526 of MEKK3 but did not recognize the S526A point mutant. The catalytically inactive (K391M) mutant of MEKK3 was not phosphorylated at Ser526, indicating that phosphorylation of Ser526 occurs via autophosphorylation. Endogenous MEKK3 was phosphorylated on Ser526 in response to osmotic stress. In addition, phosphorylation of Ser526 was required for MKK6 phosphorylation in vitro, whereas dephosphorylation of Ser526 was mediated by protein phosphatase 2A and sensitive to okadaic acid and sodium fluoride. Finally, the association between MEKK3 and 14-3-3 was dependent on Ser526 and prevented dephosphorylation of Ser526. In summary, Ser526 of MEKK3 is an autophosphorylation site within the T-loop that is regulated by PP2A and 14-3-3 proteins.

The mitogen-activated protein kinase (MAPK) 2 pathways are divided into four separate groups known as MAPK ERK , MAPK JNK , MAPK p38 , and MAPK ERK5 (reviewed in Ref. 1). Within each group, there is a highly conserved three-kinase signaling module consisting of a MAPK, a MAPK kinase (MAP2K), and a MAPK kinase kinase (MAP3K). Regulation of the MAP3K, presumably by phosphorylation, provides the impetus for activation of the three-kinase module. Once activated, the MAP3Ks activate MAP2Ks by phosphorylation of two residues within the activation loop. Phosphorylation of MAP2Ks activates these dual specificity kinases to phosphorylate MAPKs on a conserved threonine and tyrosine motif, TXY, also within the activation loop. Once phosphorylated, the MAPKs phosphorylate protein substrates and regulate cellular processes like growth, protein synthesis, gene expression, and nucleotide synthesis (2).
Over the last decade, a large body of work has characterized events that occur downstream of the MAPKs. However, little is known regarding the regulatory mechanisms that modulate the MEKK proteins to ultimately regulate the MAPKs. For example, it is known that overexpression of MEKK3 activates the ERK (3,4), JNK (3)(4)(5), p38 (5,6), ERK5 (7), and NF-B pathways (8 -10). Typically, MEKK3-dependent regulation of these pathways is studied by using transfection studies, and activation of these pathways rarely requires an agonist. Therefore, it appears that some process intrinsic to MEKK3 is critical and sufficient for activation of these pathways.
The activation loop of some protein kinase families, such as the arginine-aspartate family, is positioned between subdomains VII and VIII and is phosphorylated by other protein kinases or through autophosphorylation of the kinase itself (11). Phosphorylation within the activation loop of protein kinases results in conformational changes in the protein structure that (i) enhance substrate binding, (ii) correctly position amino acids involved in catalysis, and (iii) relieve steric hindrance within the catalytic domain. Regardless of how the activation loop is phosphorylated, regulation of catalytic activity frequently correlates with phosphorylation of the activation loop.
Given that phosphorylation plays a critical role in regulating the MAPKs and the MAP2Ks, we investigated how phosphorylation of MEKK3 might affect its catalytic activity. Since phosphorylation sites within the activation loop of MEKK3 have not been reported, we systematically mutated serine and threonine residues within the activation loop to alanine and monitored MEKK3-dependent activities in HEK293 EBNA cells. Two key amino acids were identified at positions 526 and 530 using a luciferase-based reporter gene assay as well as assays that measure the ERK, JNK, and p38 MAP kinases. We demonstrate that Ser 526 is phosphorylated by using a specific antibody that recognizes phospho-Ser 526 . In contrast, Thr 530 is not phosphorylated but is required for MEKK3 catalytic activity. In addition, phosphorylation of endogenous MEKK3 occurs on Ser 526 in response to osmotic stress. Finally, dephosphorylation of phospho-Ser 526 is mediated by PP2A, and Transfections-HEK293 EBNA cells were transiently transfected with an empty vector as a control, MEKK3, K391M, or point mutants in the pcDNA3.1 HisA plasmid using 9 l of Fugene 6 (Roche Applied Science) and 3 g of DNA. Cells were transfected in OPTI-MEM serum-free medium (Invitrogen) and then replaced with complete medium 5 h post-transfection.
Immunoprecipitation and Immunoblotting-Cells were washed twice with ice-cold PBS and then lysed in buffer containing 20 mM Tris-HCl, pH 7.6, 0.5% Nonidet P-40, 0.25 M NaCl, 1 mM phenylmethylsulfonyl fluoride, 2 mM Na 3 VO 4 , 0.25 g/ml leupeptin, 1 g/ml aprotinin. Cellular extracts were then cleared by centrifugation at 14,000 ϫ g for 10 min at 4°C. The protein concentration was determined using bovine serum albumin as a standard (15). For immunoprecipitations, 1 mg of cell extract was incubated with 10 l of anti-FLAG affinity beads and rotated at 4°C for 2 h. Samples were washed twice with 1 ml of PAN-Nonidet P-40 (10 mM Pipes, pH 7.0, 20 g/ml aprotinin, 100 mM NaCl (PAN), and 0.5% Nonidet P-40) and once with 1 ml of PAN. Immune complexes were collected by centrifugation at 2000 ϫ g for 2 min, and the protein-bound beads were heated at 90°C for 5 min in Laemmli sample buffer. Samples were resolved by 8% SDS-PAGE and probed with appropriate antibodies. For Western blots, 100 g of cellular extract was resolved by 8% SDS-PAGE, transferred to nitrocellulose, and probed with appropriate antibodies.
JNK Activity Assay-This assay was performed as previously described (16). Briefly, cell extract (40 g of protein) was incubated with 10 g of GST-c-Jun bound to glutathione-Sepharose and rotated for 2 h at 4°C. After washing with PAN and PAN-Nonidet P-40 as described above, precipitated JNK was subjected to an in vitro kinase assay with [␥-32 P]ATP and then resolved by SDS-PAGE and visualized by autoradiography.
NF-B Luciferase Assay-HEK293 EBNA cells were transiently cotransfected with 3 g of MEKK3, K391M, or point mutants and 3 g of NF-B luciferase reporter construct (Promega, Madison, WI). Cellular extracts were prepared 48 h post-transfection using passive lysis buffer (Promega) and cleared by centrifugation at 14,000 ϫ g for 10 min, and 10 g of protein was used to measure luminescence with a TD luminometer (Turner Designs, Sunnyvale, CA).
p38 MAP Kinase Assay-HEK293 EBNA cells were transiently transfected as previously described above. Cell extracts were prepared by lysis in buffer containing 20 mM Tris-HCl, pH 7.5, 1% Triton X-100, 0.5% Nonidet P-40, 150 mM NaCl, 20 mM NaF, 0.2 mM Na 3 VO 4 , 1 mM EDTA, and 1 mM EGTA. The p38 MAP kinase was immunoprecipitated as described previously (17). Briefly, 1 mg of cellular extract was incubated with polyclonal anti-p38 antibody (provided by Gary Johnson) and then rotated at 4°C for 1 h with Protein A conjugated to Sepharose (Sigma). Protein complexes were collected by centrifugation at 2000 ϫ Nucleotides in italic type denote the codon that generates the mutation. In some cases, the mutant nucleotide sequence introduced a novel endonuclease site, whereas other nucleotide sequences produced a silent mutation that did not change the amino acid sequence of MEKK3. Endonuclease sites are indicated as underlined nucleotides.

Site
Sense primer sequence  Endonuclease site   S511A  TTT GGG GCC GCC AAA CGC CTG CAG ACC  PstI  T516A  CGC CTA CAG GCC ATA TGC ATG  NdeI  S520A  TGC ATG GCA GGG ACA GGC ATT CGA TCG GTC ACT  PvuI  T522A  TCA GGG GCA GGC ATT AGA TCT GTC ACT  BglII  S526A  GGG ACA GGA ATT CGC GCT GTC ACT GGC  EcoRI  S526D  TCA GGT ACC GGC ATT CGC GAT GTC ACT GGC  KpnI  S526E  GGG ACA GGG ATT CGC GAA GTC ACT GGC  BamHI  T528A  CGC TCT GTC GCC GGC ACA CCC  NaeI  T530A  GTC ACT GGC GCC CCC TAC TGG  NarI  T530D  GGC ATT AGA TCT GTC ACT GGC GAT CCC TAC  BglII  T530E GGC ATT CGA TCG GTC ACT GGC GAA CCC TAC TGG PvuI g for 5 min. Samples were washed three times in lysis buffer and twice in kinase buffer containing 25 mM Hepes, pH 7.2, 25 mM ␤-glycerophosphate, 25 mM MgCl 2 , 2 mM DTT, 0.1 mM Na 3 VO 4 . After the final wash, the buffer was removed from the beads, and a kinase reaction mix consisting of kinase buffer, 10 Ci of [␥-32 P]ATP (PerkinElmer Life Sciences), and 10 g of GST-ATF-2 was added to the beads. Samples were incubated for 20 min at 30°C. The kinase reaction was stopped by the addition of 40 l of 2ϫ Laemmli sample buffer. Proteins were heated for 5 min at 90°C and resolved by SDS-PAGE, and then the gel was dried and processed for autoradiography. Production of Phosphoantibody Ser 526 -Phosphorylated and nonphosphorylated peptides, CMSGTGIRpSVTGTPY and CMSGTGIRS-VTGTPY (where pS denotes phosphoserine), were synthesized corresponding to amino acids 519 -532 of MEKK3 (New England Peptide, Inc.). The amino-terminal cysteine was introduced to provide a sulfhydryl moiety for conjugating to keyhole limpet hemocyanin (KLH) and a solid support. The phosphopeptide was dissolved at 1 mg/ml in PBS, and the pH was adjusted to 7.5 with 0.2 N NaOH for conjugation to KLH. The phosphopeptide (5 mg) was combined with 5 mg of KLH and incubated at 25°C for 2 h, and the conjugated peptide was dialyzed overnight in PBS. Protein concentration was determined using the Bradford method, and the peptide-KLH conjugate was aliquoted (250 or 500 g) for injection into New Zealand White rabbits (Covance, Denver, PA).
Both peptides were dissolved in 50 mM Tris-HCl, pH 8.5, and 5 mM EDTA before coupling to a solid support. Sulfolink coupling gel (Pierce) was brought to room temperature, and ϳ5 ml of gel slurry (2.5-ml column volume) was packed into a column (Econo columns; Fisher). The column was equilibrated with 6 volumes of 50 mM Tris, pH 8.0, and 5 mM EDTA prior to the addition of 4 mg of peptide. Gel and peptide were incubated for 15 min at room temperature without mixing. Subsequently, the peptide and column mixture were incubated for 30 min at room temperature without mixing. The resin was washed with 3 volumes of 50 mM Tris-HCl, pH 8.5, and 5 mM EDTA, and then 50 mM cysteine (prepared in 50 mM Tris-HCl, pH 8.5, and 5 mM EDTA) was added to block nonspecific sites on the gel by applying 1 ml of cysteine for every 1 ml of gel. This reaction proceeded for 15 min with no mixing and 30 min of mixing. The resin was washed with 16 volumes of 1 M NaCl and 16 volumes of 0.05% sodium azide/PBS and then stored at 4°C prior to use.
Phosphospecific antibody was purified via a two-step affinity column procedure. Briefly, ϳ7 ml of rabbit antisera was applied through a 0.45-m filter, the serum was delipidated with octanoic acid, and the IgG fraction was collected by ammonium sulfate precipitation (18). IgG was applied twice to an affinity column containing the phosphopeptide, and bound IgG was eluted with 0.1 M glycine, pH 2.7. Affinity-purified antibody was dialyzed in PBS and then applied to an affinity column containing the nonphosphopeptide, and the antibody that passed through the column (anti-phospho-Ser 526 antibody) was collected, dialyzed in PBS containing 50% glycerol, and stored at 4°C until use.
Dephosphorylation of MEKK3-HEK293 EBNA cells were transfected with FLAG-MEKK3 or an empty vector, and cell extracts were prepared in buffer containing 20 mM Tris-HCl, pH 7.6, 0.5% Nonidet P-40, 250 mM NaCl, 2 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, 0.25 g/ml leupeptin, 1 g/ml aprotinin, which was also supplemented with one tablet of serine and cysteine protease inhibitor (Roche Applied Science) per 10 ml of buffer. The buffer lacked serine/threonine phosphatase inhibitors, and the cell extracts were maintained on ice. Extracts were cleared of cellular debris by centrifugation at 14,000 ϫ g for 5 min at 4°C. Cell extracts containing MEKK3 (100 g) were supplemented with 1.5 l of 12 mM Tris-HCl, pH 7.4, 0.5 mM EDTA, 0.5 mM EGTA, 0.5 mM DTT, and 0.12 mg/ml bovine serum albumin. The reactions were incubated in a volume of 20 l at 30°C, and after 1 h, 2ϫ Laemmli sample buffer was added to the cell extracts, and the proteins were resolved by SDS-PAGE and then immunoblotted with anti-phospho-Ser 526 antibody.
Phosphorylation of MKK6-FLAG-MEKK3 was transfected into HEK293 EBNA cells and immunoprecipitated as described above. Preparation of the cDNA that encodes GST-KA-MKK6 was previously described (19), and the GST portion of the fusion protein was removed by proteolysis with Factor Xa. The kinase assay was performed as follows in buffer containing 25 mM Tris-HCl, pH 7.4, 25 mM MgCl 2 , 1 mM EGTA, 2 mM DTT, 25 mM ␤-glycerophosphate, 50 M ATP, 20 Ci of [␥-32 P]ATP (10 Ci/mmol), and 10 g of recombinant KA-MKK6, incubated at 30°C for 20 min, and stopped by the addition of 40 l of 2ϫ Laemmli-DTT buffer. The proteins were denatured by heating for 5 min at 95°C, resolved by 8% SDS-PAGE, and stained with Coomassie Blue, and the gel was dried prior to autoradiography.
Isolation of Proteins with GST-14-3-3⑀-Cell extracts were prepared by lysis in buffer containing 20 mM Tris-HCl, pH 7.6, 0.5% Nonidet P-40, 0.25 M NaCl, 1 mM phenylmethylsulfonyl fluoride, 2 mM Na 3 VO 4 , 0.25 g/ml leupeptin, 1 mM DTT, and 1 g/ml aprotinin. Extracts were cleared of cellular debris by centrifugation at 14,000 ϫ g for 5 min at 4°C. Approximately 500 g of cellular extract was incubated with 20 g of recombinant GST-lamin (amino acids 66 -206) or GST-14-3-3⑀ fusion proteins bound to Sepharose beads for 2 h at 4°C. Samples were then washed twice with PAN-Nonidet P-40 and once with PAN, proteins were collected at 2000 ϫ g in a microcentrifuge for 1 min, and then the isolated proteins were incubated with cell extract containing endogenous protein phosphatases. The proteins were separated by SDS-PAGE, and after transfer to nitrocellulose, the membrane was probed with an amino-terminal antibody that recognizes total MEKK3 or phospho-Ser 526 , followed by goat anti-rabbit horseradish peroxidase antibody.

RESULTS
Ser 526 and Thr 530 Are Required for MEKK3-dependent NF-B Luciferase Activity-MEKK3 has previously been shown to play a critical role in the NF-B signal transduction pathway (9,20), and therefore a NF-B luciferase assay was chosen as one method to screen point mutants in the activation loop of MEKK3. To investigate which amino acids might be essential in the regulation of NF-B-dependent transcription, each serine and threonine within the activation loop of MEKK3 was mutated to alanine by site-directed mutagenesis (Fig. 1A). Each point mutant of MEKK3 was transfected into HEK293 EBNA cells, along with a luciferase reporter gene containing the NF-B recognition sequence. After 48 h, cell extracts were prepared and assayed for luciferase activity. Expression of MEKK3 resulted in agonist-independent transcription of the NF-B luciferase reporter gene, whereas only the (S526A) and (T530S) point mutants failed to activate the reporter gene (data not shown). This result suggested that Ser 526 and Thr 530 are required for MEKK3-dependent regulation of the NF-B promoter. Moreover, this result strongly suggested that Ser 526 and Thr 530 might be phosphorylation sites within the activation loop of MEKK3.
Ser 526 or Thr 530 were mutated to aspartic or glutamic acid, since such mutations have been shown to produce constitutively active kinases (21,22). Experiments were performed in which S526A, S526D, S526E, T530A, T530D, and T530E MEKK3 were transfected into HEK293 EBNA cells, along with the NF-B luciferase reporter gene. The S526A and T530A MEKK3 mutants were unable to activate transcription of the NF-B luciferase reporter gene (Fig. 1B). However, the aspartic and glutamic acid mutants of Ser 526 restored ϳ50% of wild-type MEKK3 activity, suggesting that the negative charge of the carboxyl group might mimic the negative charge of a phosphate moiety at this position. In contrast, the aspartic and glutamic acid mutants of Thr 530 were unable to restore NF-B luciferase activity, indicating that a negative charge at position 530 was not tolerated within the activation loop. In addition, mutation of the active site lysine to methionine (K391M) abrogated luciferase activity, indicating that the catalytic activity of MEKK3 was required for NF-B-dependent luciferase activity. Thus, S526A and T530A MEKK3 were unable to activate transcription through the NF-B promoter, although the active site lysine was retained, demonstrating that serine 526 and threonine 530 are important for MEKK3dependent transcription.
Activation Loop Mutants S526A and T530A MEKK3 Is Unable to Activate the MEK and ERK Pathway-Overexpression of MEKK3 results in the activation of several MAP kinases that include the ERK, JNK, p38, and ERK5 pathways. Due to the intriguing biological activity of the S526A and T530A activation loop point mutants in the NF-B luciferase assay, we chose to investigate the role of these point mutants in activating the ERK, JNK, and p38 pathways. First, we investigated the ERK MAPK pathway. HEK293 EBNA cells were transiently transfected with MEKK3, catalytically inactive MEKK3, and the seven activation loop point mutants to determine whether the activation loop point mutants were capable of activating the ERK pathway. After 48 h, cellular extracts were prepared and resolved by SDS-PAGE, transferred to nitrocellulose, and immunoblotted for phosphorylated ERK. Consistent with the luciferase assays described above, expression of (S526A) and (T530A), as well as catalytically inactive MEKK3, did not result in phosphorylation of ERK1 and -2 ( Fig. 2, top). In contrast, expression of MEKK3, as well as the S511A, T516A, T520A, T522A, and the T528A point mutants resulted in ERK activation. In addition, the constitutively active mutants of Ser 526 , S526D and S526E, also yielded a robust phosphorylation of the ERKs, whereas the constitutively active mutants of Thr 530 , T530D and T530E, were not able to activate the ERKs (data not shown), which was consistent with the luciferase assay previously shown. The nitrocellulose membrane was subsequently reprobed for ERK1 and -2 as a loading control, thereby demonstrating that differences in ERK phosphorylation were due to MEKK3 activity and not ERK expression.
Previous work has shown that ERK1 and -2 are phosphorylated and activated by MEK1 and -2 (23). To determine whether expression of MEKK3 affected MEK phosphorylation, the experiment described above was also probed for phosphorylated MEK1 and -2. Consistent with the ERK data, cells transfected with the S526A and T530A point The mouse and human isoforms of MEKK3 express serine and threonine at identical positions within the activation loop, which have been mutated to alanine for this study and are numbered according to the mouse sequence. The amino-terminal domain consisting of amino acids 1-350 is referred to as the regulatory domain, although no specific function has been ascribed to this domain at this time. B, constitutively active mutants of Ser 526 regulate NF-B luciferase activity. HEK293 EBNA cells were transfected with cDNAs that encode FLAG-tagged MEKK3 or the indicated point mutants with a cDNA that encodes the firefly luciferase gene downstream of the NF-B promoter element. Cell extracts were prepared 48 h after transfection, and 10 g was used to measure luciferase activity. C, in addition, 50 g of cell extract was resolved by SDS-PAGE and immunoblotted (IB) using the monoclonal antibody that recognizes the FLAG epitope. The results are from a representative experiment. mutants did not result in phosphorylation of MEK1 and -2 ( Fig. 2, middle). The membrane was subsequently reprobed for MEK1 and -2 as a loading control. In addition, the membrane was also probed for MEKK3 expression by using the monoclonal antibody that recognizes the FLAG epitope, which demonstrated similar transfection efficiencies for each of the mutants. It is interesting to note that K391M, S526A, and T530A MEKK3, which are the three mutants that were unable to activate ERK and MEK phosphorylation, appeared as discreet polypeptides on the MEKK3 immunoblot, which typically indicates a lack of phosphorylation (Fig. 2, bottom, lanes c, e, and i). The reduced width of the immunoblotted protein bands probably reflects decreased phosphorylation of MEKK3 and not decreased expression (also see Fig. 5, lane g).
Activation Loop Mutants S526A and T530A MEKK3 Are Unable to Activate the p38 MAP Kinase-Since MEKK3 has been shown to activate the p38 MAP kinase (5) and function in an osmosensing scaffold with Rac, MKK3, and OSM (osmosensing scaffold for MEKK3) (6), we wanted to determine whether Ser 526 and Thr 530 were also involved in regulating the p38 MAP kinase. HEK293 EBNA cells were transfected with MEKK3, (S526A), and (T530A) MEKK3, and after 48 h, p38 MAP kinase was immunoprecipitated from cell extracts. An in vitro kinase assay was performed using GST-ATF2 as a substrate and [␥-32 P]ATP. Proteins were resolved by SDS-PAGE, and phosphorylated GST-ATF2 was identified by autoradiography. Consistent with previous studies, expression of MEKK3 resulted in activation of p38 MAP kinase (Fig. 3 c and d). These results demonstrate that Ser 526 and Thr 530 are critical for MEKK3-dependent p38 MAP kinase activation.
Activation Loop Mutants S526A and T530A MEKK3 Are Unable to Activate JNK-MEKK3 has been shown to activate JNK (3,5), and therefore we wanted to determine whether Ser 526 and Thr 530 were required to activate JNK. HEK293 EBNA cells were transfected with an empty vector as a control or constructs encoding various forms of MEKK3, as indicated in Fig. 3B. After 48 h, 40 g of cell extract was incubated with GST-c-Jun to precipitate endogenous JNK. After washing the precipitated proteins with buffer, the GST-c-Jun⅐JNK protein complex was incubated with [␥-32 P]ATP for 20 min, and then the proteins were resolved by SDS-PAGE. Phosphorylated GST-c-Jun was visualized by autoradiography. Expression of MEKK3 resulted in a robust activation of JNK (Fig. 3, lane b), whereas expression of an empty vector as a control (lane a), catalytically inactive MEKK3 (lane c), or the activation loop point mutants (lanes d and e) resulted in very little JNK activity. In summary, we demonstrate that the catalytic activity of MEKK3 is required to activate the ERK, p38, and JNK MAP kinases. In addition, mutation of Ser 526 and Thr 530 to alanine, while retaining the active site lysine at position 391, also produced catalytically inactive mutants of MEKK3. Thus, Ser 526 and Thr 530 are critical amino acids within the activation loop of MEKK3.
Autophosphorylation of MEKK3 at Ser 526 -The Ser 526 and Thr 530 point mutants of MEKK3 clearly demonstrated that these two amino acids are critical for activity. However, it is unclear whether these amino acids are required to maintain the structural integrity of the activation loop or whether Ser 526 and Thr 530 are phosphorylation sites. In an attempt to address this question, we generated phosphospecific antibodies using peptide sequences that correspond to phospho-Ser 526 and phospho-Thr 530 of MEKK3. To evaluate the specificity of the phosphospecific antibodies, HEK293 EBNA cells were transfected with MEKK3, catalytically inactive MEKK3, S526A MEKK3, and T530A  MEKK3. After 48 h, cell extracts were prepared, and 50 g of each extract was resolved by SDS-PAGE and immunoblotted with the phosphospecific antibodies. The anti-phospho-Ser 526 antibody recognized MEKK3 (Fig. 4A, lane b), whereas cells transfected with an empty vector showed no immunoreactivity, except a nonspecific protein of ϳ90 kDa (lanes a-e). Importantly, the phospho-Ser 526 antibody showed no immunoreactivity toward cell extracts expressing S526A MEKK3 (lane d), indicating that the antibody was specific for phospho-Ser 526 . In addition, catalytically inactive MEKK3 was not detected by the phospho-Ser 526 antibody (lane c), strongly suggesting that phosphorylation of Ser 526 occurs via autophosphorylation of MEKK3 and not by an upstream kinase. If phosphorylation of MEKK3 occurred via an upstream kinase that is endogenously expressed in HEK293 cells, one would expect phosphorylation of catalytically inactive MEKK3 at Ser 526 . It should be emphasized that the catalytically inactive mutant of MEKK3 can be phosphorylated at Ser 166 , which demonstrates that this mutant form of MEKK3 retains its structure for phosphorylation by another kinase (12), and presumably the structurally features surrounding Ser 526 in the catalytically inactive mutant resemble wild-type MEKK3. This result demonstrates that the catalytically inactive mutant of MEKK3 retains the structural features needed for phosphorylation by another kinase and indicates that phosphorylation of Ser 526 occurs by autophosphorylation. Finally, it is interesting to note that T530A MEKK3 was recognized by the anti-phospho-Ser 526 antibody (lane e), albeit to a lesser extent than wild type MEKK3. This result suggests that Thr 530 is not essential for phosphorylation of Ser 526 , although Thr 530 is required for signaling to downstream substrates. Reprobing the immunoblot with a monoclonal antibody that recognizes the FLAG epitope of transfected MEKK3 indicated that similar levels of the MEKK3 proteins were expressed in the cells (Fig. 4A, bottom).
Similar experiments were performed using affinity-purified antisera generated against a peptide containing phospho-Thr 530 of MEKK3 (data not shown). Antisera from three different rabbits were affinitypurified and tested for immunoreactivity against phospho-Thr 530 . Although the sera tested positive against the antigen that was used to immunize the rabbits, all of the antisera tested negative for phosphorylation at Thr 530 when MEKK3 was overexpressed in HEK293 EBNA cells. Thus, based on these data and the fact that the (T530D) and (T530E) mutants were unable to activate MEKK3-dependent signaling pathways, we conclude that Thr 530 is not phosphorylated in the activation loop of MEKK3.
Phosphorylation of Endogenous MEKK3 at Ser 526 -Having developed an antibody that recognized phospho-Ser 526 , we could determine whether Ser 526 is phosphorylated in vivo. Since MEKK3 functions in the p38 MAPK pathway (6), osmotic stress was used to activate endogenous MEKK3. NIH3T3 cells were deprived of serum for 18 h and then treated with 0.4 M sorbitol for 30 min to induce osmotic stress. Cell extracts were prepared, and 50 g was resolved by SDS-PAGE, transferred to nitrocellulose, and immunoblotted with anti-phospho-Ser 526 antibody. Sorbitol treatment of NIH3T3 cells induced phosphorylation of Ser 526 (Fig. 4B, top, lane c). In addition, NIH3T3 cells were pretreated with 100 nM okadaic acid for 15 min to reduce endogenous serine/threonine phosphatase activity and to enhance phosphorylation of Ser 526 (Fig. 4B,  top, lanes b and d), and although okadaic acid caused a shift in the electrophoretic mobility of MEKK3 (Fig. 4B, bottom, lane b), there was no change in phosphorylation of Ser 526 in the presence or absence of okadaic acid (top). In summary, these data demonstrate that endogenous MEKK3 is phosphorylated on Ser 526 in the response to exogenous stimuli like osmotic stress.
Dephosphorylation of Ser 526 -By developing an antibody that recognized Ser 526 when phosphorylated, we could determine whether Ser 526 was sensitive to dephosphorylation. Since expression of MEKK3 in HEK293 EBNA cells resulted in phosphorylation of Ser 526 , we postulated that protein phosphatase(s) expressed in this cell line would also dephosphorylate the site. To test this hypothesis, HEK293 EBNA cells were transfected with a cDNA that encodes MEKK3 (Fig. 5, lanes f-j) or an empty vector (lanes a-e), and after 48 h, cell extracts were prepared at 4°C in buffer lacking serine/threonine phosphatase inhibitors. After cell debris was removed by centrifugation, 100 g of cell extract in 20 l  c and d). Cell extracts (50 g) were resolved by SDS-PAGE, transferred to nitrocellulose, and immunoblotted with antibody as described in A. was supplemented with 1.5 l of 12 mM Tris, pH 7.4, 0.5 mM EDTA, 0.5 mM EGTA, 0.5 mM DTT, and 0.12 mg/ml bovine serum albumin. Then the cell extracts were incubated at 30°C for 1 h. Where indicated, phosphatase reactions also included 10 mM NaF or 100 nM okadaic acid. As we have previously shown, expression of MEKK3 in HEK293 EBNA cells resulted in phosphorylation of MEKK3 at Ser 526 (Fig. 5, top, lane f ). In this experiment, we demonstrate that incubation of MEKK3 in the HEK293 EBNA cell extract resulted in complete dephosphorylation of MEKK3 (Fig. 5, top, lane g). Immunoblotting the same nitrocellulose membrane with antibody that recognizes the amino terminus of MEKK3 demonstrated that the lack of immunoreactivity in lane g was not due to degradation of MEKK3. In fact, the loss of phosphate from MEKK3 is reflected by the increased mobility and decreased molecular mass of MEKK3 by SDS-PAGE (Fig. 5, bottom, compare lane g with lane  f ). Moreover, serine/threonine phosphatase inhibitors such as NaF and okadaic acid were included in the phosphatase reactions and were highly effective at inhibiting the phosphatase activity in the HEK293 EBNA cell extracts (Fig. 5, top, lanes h-j). In summary, these data demonstrate that Ser 526 of MEKK3, when incubated in cell extracts, is susceptible to dephosphorylation and that the phosphatase(s) responsible for this activity is sensitive to NaF and okadaic acid.
Dephosphorylation of Ser 526 by PP2A Isoforms-Since we had established dephosphorylation of Ser 526 when MEKK3 was incubated in cell extracts, we tested whether the protein serine/threonine phosphatase, PP2A, could dephosphorylate Ser 526 . To obtain two different PP2A holoenzymes, HEK293 cells were transfected with either the FLAGtagged ␣ or ␦ isoform of the regulatory B subunit of PP2A. The A subunit is a scaffolding protein, whereas the C subunit is the catalytic protein of PP2A. The PP2A holoenzymes, consisting of either AB␣C or AB␦C subunits, were isolated by immunoprecipitation using the monoclonal antibody that recognizes the FLAG epitope. The PP2A holoenzymes were eluted from the Sepharose beads using the FLAG peptide at a concentration of 0.1 mg/ml as previously described (24). In a separate transfection, HEK293 EBNA cells were transfected with FLAG-tagged MEKK3, and after 48 h, cell extracts were prepared, and tagged MEKK3 was immunoprecipitated using the FLAG monoclonal antibody. Immunoprecipitated FLAG MEKK3 was incubated with 9 ng of PP2A holoenzymes, AB␣C (Fig. 6, lanes b and c) or AB␦C (Fig. 6, lanes d and e) for 30 or 60 min at 30°C. Within 30 min, phospho-Ser 526 was completely dephosphorylated by either PP2A holoenzyme. These results suggest that the protein phosphatase that dephosphorylates Ser 526 from HEK293 EBNA cell extracts is probably an isoform of PP2A.
Dephosphorylation of Ser 526 Affects MEKK3 Catalytic Activity toward MKK6-Since MEKK3 is a MAP3K that functions upstream of the p38 MAP kinase and we had shown the phospho-Ser 526 was required for MEKK3-dependent activation of p38 (Fig. 3A), we tested whether phosphorylation of Ser 526 was required for MEKK3 to directly phosphorylate a substrate like MKK6. The cDNA that encodes FLAG-MEKK3 was transfected into HEK293 EBNA cells, and after 48 h, Sepharose beads conjugated to the monoclonal antibody that recognizes the FLAG epitope were used for immunoprecipitation. Precipitated proteins were washed twice with 1 ml of PAN Nonidet P-40 buffer, once with 1 ml of PAN buffer, and aliquoted into two tubes. After centrifugation to collect the protein-bound beads, 100 g of nontransfected HEK293 EBNA cell extract that was prepared with 10 mM NaF and 100 nM okadaic acid (Fig.  7, A (lanes a and b), B (lane b), and C (lane a)) or without phosphatase inhibitors (Fig. 7, A (lanes c and d), B (lane b), and C (lane b)) was added to each tube and incubated for 1 h at 30°C. After the dephosphorylation reaction, 20% of the sample was removed to assess dephosphorylation of FLAG-MEKK3 by immunoblotting with antibody that recognizes phos-  (lanes a-e). In a separate transfection, HEK293 cells were transfected with the FLAG-tagged B␣ or B␦ subunit of PP2A. The PP2A holoenzymes consisting of AB␣C or AB␦C were eluted from the Sepharose beads with the FLAG peptide, and then 2 ng were incubated with immunoprecipitated MEKK3 for the indicated times at 30°C (lanes b-e). The reactions were terminated with Laemmli sample buffer, the proteins were resolved by SDS-PAGE and immunoblotted (IB) with anti-phospho-Ser 526 antibody (top), and then the nitrocellulose membrane was reprobed with antibody that recognizes the amino terminus of MEKK3 (bottom). FIGURE 7. Dephosphorylation of Ser 526 decreases MEKK3 catalytic activity toward MKK6. HEK293 EBNA cells were transfected with FLAG-tagged MEKK3, and immunoprecipitation was performed using Sepharose beads conjugated to the monoclonal antibody that recognizes the FLAG epitope (A). After the immunoprecipitation, the proteinbound beads were washed with buffer and aliquoted into two separate Eppendorf tubes, and the beads were collected by centrifugation. One tube was incubated with 100 g of HEK293 EBNA cell extract prepared with (A, lanes a and b; B (lane a), and C (lane a)) or without phosphatase inhibitors for 1 h at 30°C (A (lanes c and d), B (lane b), and C (lane  b)). To assess the dephosphorylation of FLAG-MEKK3, 20% of each sample was collected, and Laemmli sample buffer was added to terminate the dephosphorylation reaction. These samples were resolved by SDS-PAGE and first immunoblotted (IB) with antibody that recognizes phospho-Ser 526 (B), and then the membrane was reprobed with antibody that recognizes the amino terminus of MEKK3 (C). To the remaining samples, cell extract was removed after centrifugation of the beads, and the FLAG MEKK3-bound beads were washed twice with PAN buffer and aliquoted into two separate tubes again (A, lanes a and b and lanes c and d). An in vitro kinase assay was performed with (lanes b  and d) or without (lanes a and c) 10 g of KA-MKK6 as substrate and [␥-32 P]ATP. The kinase assays were terminated by the addition of Laemmli sample buffer, and the proteins were resolved by SDS-PAGE and stained with Coomassie Blue (data not shown), followed by autoradiography (A). This is a representative figure of four separate experiments.
pho-Ser 526 (Fig. 7B), whereas the remaining sample was washed twice with PAN buffer and aliquoted again into two tubes to perform a kinase assay. An in vitro kinase assay was performed with KA-MKK6 as substrate.
Incubation of FLAG-MEKK3 with cell extract prepared in the absence of phosphatase inhibitors resulted in dephosphorylation of Ser 526 (Fig. 7B, lane b). When FLAG-MEKK3 was dephosphorylated at Ser 526 (Fig. 7B, lane b), FLAG-MEKK3 was not capable of phosphorylating recombinant KA-MKK6 (Fig. 7A, lane d), although the same amount of KA-MKK6 was present in the kinase assay as determined by Coomassie staining of the gel (data not shown). In contrast, FLAG-MEKK3 was capable of phosphorylating KA-MKK6 when phosphorylated at Ser 526 (Fig. 7A, lane b). Thus, the catalytic activity of MEKK3 requires phosphorylation of Ser 526 to phosphorylate a substrate like MKK6.
Association of MEKK3 with 14-3-3 Prevents Dephosphorylation of Ser 526 -Serine and threonine phosphorylation of proteins is frequently accompanied by the recruitment of 14-3-3 proteins (25), which affects protein/protein interactions, subcellular localization, and activity (reviewed in Refs. 26 and 27). Since we had previously demonstrated an association between MEKK3 and 14-3-3⑀ (28), we wanted to determine whether the association between MEKK3 and 14-3-3 proteins required phosphorylation of Ser 526 . HEK293 EBNA cells were transfected with FLAG MEKK3 or point mutants, and after 48 h, cell extracts were prepared, and the MEKK3 proteins were immunoprecipitated using the FLAG epitope. Proteins were resolved by SDS-PAGE, transferred to nitrocellulose, and immunoblotted with antibody that recognizes the phospho-Ser 526 epitope. As shown previously in Fig. 4, MEKK3 was phosphorylated on Ser 526 , and the T530A point was also weakly phosphorylated on Ser 526 . In contrast, the S526A point mutant and the kinase-inactive mutant K391M were not phosphorylated on Ser 526 (Fig.  8, top). The nitrocellulose membrane was also probed for the 14-3-3 protein, and the co-precipitation of 14-3-3 protein was evident with wild-type MEKK3 and weakly with the T530A mutant, whereas mutants that were not phosphorylated on Ser 526 were also unable to associate with 14-3-3 protein. From this experiment, it is not known which isoform of 14-3-3 interacts with MEKK3, since the 14-3-3 antibody recognizes amino acids 3-21 of the human isoform of 14-3-3␤, and this sequence is highly conserved within the 14-3-3 family of proteins (␤, ␥, ⑀, , , , and ).
Since we had previously demonstrated an association between MEKK3 and the ⑀ isoform of 14-3-3 (28), we wanted to determine whether the association between MEKK3 and 14-3-3⑀ would prevent dephosphorylation of Ser 526 . HEK293 EBNA cells were transfected with FLAG MEKK3, and cell extracts were prepared in the absence of okadaic acid and sodium fluoride. Then the cell extracts were incubated at 4°C with GST-14-3-3⑀ or GST-lamin that was expressed and purified from bacteria, as previously described (28). After 1 h, the GST fusion proteins and associated proteins were collected by centrifugation and washed twice with 1 ml of PAN-Nonidet P-40 and PAN buffers to remove nonspecific proteins. Precipitated proteins were resolved by SDS-PAGE and immunoblotted using the anti-phospho-Ser 526 antibody. 14-3-3⑀ precipitated FLAG-MEKK3 while phosphorylated at Ser 526 (Fig. 9, top, lane a), whereas the lamin fusion protein was unable to precipitate FLAG-MEKK3 (lane g), as we have previously reported (28).
Since we were able to demonstrate dephosphorylation of Ser 526 as shown in Fig. 5, an experiment was designed to determine whether the association between MEKK3 and 14-3-3⑀ prevented dephosphorylation of Ser 526 . FLAG-MEKK3 was precipitated from HEK293 EBNA cell extracts with GST-14-3-3⑀, as described above, and then incubated with 100 g of HEK293 EBNA cell extract from nontransfected cells and incubated at 30°C for up to 3 h. The phosphatase reactions were terminated by the addition of Laemmli sample buffer, and the proteins were resolved by SDS-PAGE and immunoblotted with anti-phospho-Ser 526 MEKK3 antibody. The association of FLAG-MEKK3 with 14-3-3⑀ prevented the dephosphorylation of Ser526 (Fig. 9, top, lanes b-f). In contrast, incubation of cell extracts containing FLAG-MEKK3, under the same conditions that promote phosphatase activity, resulted in complete dephosphorylation of Ser 526 after 30 min (lane j), whereas the  -e), cell extracts were prepared, and an antibody that recognizes the FLAG epitope was used to immunoprecipitate the MEKK3 proteins. The samples were resolved by SDS-PAGE and immunoblotted (IB) with antibody that recognizes phospho-Ser 526 (top), and then the membrane was reprobed with antibody that recognizes the amino terminus of MEKK3 (middle). The bottom portion of the membrane was probed with 14-3-3 antibody that recognizes amino acids 3-21 of the human isoform of 14-3-3␤. This sequence is highly conserved within the 14-3-3 family of proteins (␤, ␥, ⑀, , , , and ), and therefore it is not known which isoform of 14-3-3 interacts with MEKK3.  -h), and cell extracts were incubated with GST-14-3-3⑀ (lanes a-f) or GST-lamin (lanes g and h) that were bound to Sepharose. Proteins were precipitated by centrifugation and washed several times with buffer and then incubated with 100 g of HEK293 cell extract obtained from nontransfected cells at 30°C. After the indicated times, the phosphatase reactions were terminated with Laemmli sample buffer, and the proteins were resolved by SDS-PAGE and immunoblotted (IB) with anti-phospho-Ser 526 rabbit polyclonal antibody (top). In lanes i and j, cell extracts expressing FLAG-MEKK3 were incubated at 30°C for the indicated times. The same nitrocellulose membrane was stripped and probed with antibody that recognizes the amino terminus of MEKK3 (bottom). amount of FLAG-MEKK3 that was used in each experimental condition was similar, as determined by immunoblotting with antibody that recognizes the amino terminus of MEKK3 (Fig. 9, bottom, compare lanes  a-f with lanes i-k). In summary, the association between FLAG-MEKK3 and 14-3-3⑀ inhibits dephosphorylation of Ser 526 . Thus, given the importance of phospho-Ser 526 in regulating MEKK3 activity, it appears that the interaction between MEKK3 and 14-3-3⑀ is critical for maintaining a catalytically active form of MEKK3.

DISCUSSION
MEKK3 is a serine/threonine kinase of the MAP3K family that has been implicated as an upstream regulator in the signaling pathways for a number of MAP kinases (3-7) as well as the tumor necrosis factor signaling pathway (9, 20, 29 -31). Most studies have focused on the signaling properties of MEKK3 after stable or transient introduction of this kinase into target cells or by suppression of MEKK3 expression by small interfering RNA. However, little work has been done to directly measure MEKK3 activity. To address this issue, we set out to identify phosphorylation sites in the activation loop of MEKK3, so that phosphospecific antibodies could be used to monitor MEKK3 activity.
In this study, we demonstrate that Ser 526 is the only phosphorylation site within the activation loop of MEKK3 and that phosphorylation occurs in response to exogenous stimuli like osmotic stress. In addition, whereas mutation of other putative phosphorylation sites did not significantly alter MEKK3 activity, mutation of Thr 530 to a conserved alanine or to the phosphomimetic amino acids, aspartate or glutamate, was not tolerated, suggesting that Thr 530 might play an important role in maintaining the structural integrity of the activation loop. Although the crystal structure of the catalytic domain of MEKK3 has not been resolved, the crystal structure of the catalytic subunit of PKA provides insights as to the orientation of the activation loop of MEKK3. Two critical amino acids between subdomains VII and VIII of the catalytic subunit of PKA, Thr 197 and Thr 201 , are localized at positions equivalent to those of Ser 526 and Thr 530 of MEKK3, respectively. In PKA, Thr 197 is phosphorylated and critical for activity, whereas Thr 201 is also critical for activity but not a phosphorylation site (32). In this study, we have demonstrated that Ser 526 of MEKK3 is phosphorylated and critical for catalytic activity, like Thr 197 of PKA. We have also shown that Thr 530 , which corresponds to Thr 201 of PKA, is required for MEKK3 activity and not likely to be a phosphorylation site. This result suggests that Thr 530 is critical to maintain the structural integrity of the activation loop of MEKK3. In the crystal structure of PKA that is bound to substrate peptide, Thr 201 of the PKA catalytic subunit interacts with Asp 166 but only in a substrate-dependent manner (32). Given the similarities between the PKA and MEKK3 activation loops, it is quite possible that Thr 530 of MEKK3 interacts with Asp 489 , which corresponds to Asp 166 of PKA. The fact that alanine or negatively charged amino acids are not tolerated at position 530 supports the notion that the hydroxyl moiety of Thr 530 interacts with a negatively charged amino acid, like Asp 489 . The crystal structure of the MEKK3 catalytic domain while bound to substrate peptide should confirm this prediction.
The development of an antibody that recognizes phospho-Ser 526 provides compelling evidence that Ser 526 is a phosphorylation site within the activation loop of MEKK3. Further evidence that a negative charge at Ser 526 is critical for MEKK3 activity comes from experiments using the S526D and S526E mutants of MEKK3, which activated the ERK MAP kinases, MEK, and NF-B luciferase activity. In contrast, negatively charged amino acids at Thr 530 produced inactive mutants of MEKK3. Moreover, attempts to detect phospho-Thr 530 of MEKK3 with a phosphospecific antibody were unsuccessful (data not shown), which supports the conclusion that Thr 530 is not a phosphorylation site. However, although Thr 530 is not a phosphorylation site, Thr 530 is required for the phosphoryl transfer reaction catalyzed by MEKK3.
The catalytically inactive mutant of MEKK3, which has the active site lysine mutated to methionine (K391M), was not phosphorylated on Ser 526 . Based on this observation, our results suggest that phosphorylation of Ser 526 occurs by an autophosphorylation mechanism. If phosphorylation of Ser 526 had occurred in the K391M MEKK3 mutant, such a result would have indicated that an upstream kinase in the signaling cascade, such as a mitogen-activated protein kinase kinase kinase kinase, phosphorylates Ser 526 . But this result was never observed in these studies. It is interesting to note that although Thr 530 is required for MEKK3-dependent MAP kinase activity, autophosphorylation of Ser 526 does not require Thr 530 , albeit efficient autophosphorylation of Ser 526 may require Thr 530 . Nonetheless, our results suggest that the autophosphorylation of MEKK3 may occur via a different mechanism from phosphorylation of downstream kinases, such as the MAP2K family. Since the MKKs require dual phosphorylation for efficient catalytic activity, it is possible that Thr 530 is necessary to orient the activation loop of MEKK3 for dual phosphorylation of MKKs, whereas Thr 530 is not necessary for phosphorylation of Ser 526 . Additional mechanistic studies will be needed to delineate the difference between those phosphorylation events.
Dephosphorylation of Ser 526 was readily accomplished by using purified PP2A holoenzymes. Moreover, the presence of PP2A inhibitors such as okadaic acid or sodium fluoride in cell extracts containing transfected MEKK3 prevented dephosphorylation of Ser 526 , and the absence of phosphatase inhibitors resulted in dephosphorylation of Ser 526 . The reversible nature of Ser 526 phosphorylation further supports the fact that a post-translational modification of Ser 526 is physiologically relevant for MEKK3 activity, especially under conditions of osmotic stress. The study described herein also demonstrates that phospho-Ser 526 is also required for efficient phosphorylation of downstream substrates, such as MKK6. Importantly, dephosphorylation of Ser 526 is prevented by the interaction with 14-3-3 protein, which is consistent with the previously defined role of 14-3-3 proteins binding to phosphorylated serine or threonine (33). However, phospho-Ser 526 is not localized within a consensus 14-3-3 binding site as described by Yaffe et al. (34). Although we know that 14-3-3 binding to MEKK3 prevents dephosphorylation of MEKK3, we do not know whether phospho-Ser 526 contributes directly to the binding with 14-3-3 protein. Our previous study using the amino terminus of MEKK3 as bait in the yeast two-hybrid system suggests that Ser 526 is not directly involved in 14-3-3 binding, since we have shown that the amino terminus is sufficient to interact with 14-3-3 proteins (28). Thus, perhaps other phosphorylation sites on MEKK3, such as Ser 166 and/or Ser 337 , are needed for binding with 14-3-3 (12) and the tertiary structure of MEKK3, when bound to 14-3-3, does not allow phosphatases to access phosphorylated Ser 526 .
Future studies must now be directed toward understanding the mechanism by which MEKK3 autophosphorylation occurs and, conversely, how protein phosphatases like PP2A dephosphorylate Ser 526 . It is clear that the interaction between MEKK3 and 14-3-3 proteins helps maintain phosphorylation of Ser 526 , but what causes the dissociation of 14-3-3 proteins from phosphorylated MEKK3 is a question whose answer will probably allow us to better understand how MEKK3 functions in the cell. In summary, Ser 526 is an activation loop phosphorylation site that is critical for MEKK3 activity, much as phosphorylation of the activation loop of Thr 197 of the catalytic subunit of PKA functions as a molecular switch for enzymatic activity (35). Moreover, Ser 526 pro-vides a molecular target to readily assess MEKK3 activity under different physiological conditions.