The kinase activation loop is the key to mixed lineage kinase-3 activation via both autophosphorylation and hematopoietic progenitor kinase 1 phosphorylation.

We have demonstrated previously that Cdc42 induced MLK-3 homodimerization leads to both autophosphorylation and activation of MLK-3 and postulated that autophosphorylation is an intermediate step of MLK-3 activation following its dimerization. In this report we sought to refine further the mechanism of MLK-3 activation and study the role of the putative kinase activation loop in MLK-3 activation. First we mutated the three potential phosphorylation sites in MLK-3 putative activation loop to alanine in an effort to abrogate MLK-3 autophosphorylation. Mutant T277A displayed almost no autophosphorylation activity and was nearly nonfunctional; mutant S281A, that displayed a low level of autophosphorylation, only slightly activated its downstream targets, whereas the T278A mutant, that exhibited autophosphorylation comparable to that of the wild type, was almost fully functional. Thus, these residues within the activation loop are critical for MLK-3 autophosphorylation and activation. In addition, when the Thr277 and Ser281 residues were mutated to negatively charged glutamic acid to mimic phosphorylated serine/threonine residues, the resulting mutants were fully functional, implying that these two residues may serve as the autophosphorylation sites. Interestingly, HPK1 also phosphorylated MLK-3 activation loop in vitro, and Ser281 was found to be the major phosphorylation site, indicating that HPK1 also activates MLK-3 via phosphorylation of the kinase activation loop.

We have demonstrated previously that Cdc42 induced MLK-3 homodimerization leads to both autophosphorylation and activation of MLK-3 and postulated that autophosphorylation is an intermediate step of MLK-3 activation following its dimerization. In this report we sought to refine further the mechanism of MLK-3 activation and study the role of the putative kinase activation loop in MLK-3 activation. First we mutated the three potential phosphorylation sites in MLK-3 putative activation loop to alanine in an effort to abrogate MLK-3 autophosphorylation. Mutant T277A displayed almost no autophosphorylation activity and was nearly nonfunctional; mutant S281A, that displayed a low level of autophosphorylation, only slightly activated its downstream targets, whereas the T278A mutant, that exhibited autophosphorylation comparable to that of the wild type, was almost fully functional. Thus, these residues within the activation loop are critical for MLK-3 autophosphorylation and activation. In addition, when the Thr 277 and Ser 281 residues were mutated to negatively charged glutamic acid to mimic phosphorylated serine/ threonine residues, the resulting mutants were fully functional, implying that these two residues may serve as the autophosphorylation sites. Interestingly, HPK1 also phosphorylated MLK-3 activation loop in vitro, and Ser 281 was found to be the major phosphorylation site, indicating that HPK1 also activates MLK-3 via phosphorylation of the kinase activation loop.
Many kinases are phosphorylated in the kinase activation loop located between the conserved sequence DFG of subdomain VII and APE of subdomain VIII (28). The activation loop plays a crucial role in substrate recognition, and in many cases; phosphorylation in this segment is required to allow correct alignment of the substrates to the catalytic site (28 -30). This segment can be regulated either through autophosphorylation or phosphorylation by other kinases. For example, cyclic AMPdependent protein kinase autophosphorylates the Thr 197 located in its activation loop (31), whereas ERK2 is phosphorylated on Thr 183 and Tyr 185 by MEK in the corresponding activation loop (32). In both cases, phosphorylation in the activation loop is essential for the activation of these kinases. The putative kinase activation loop of MLK-3 contains three potential phosphorylation sites at residues Thr 277 , Thr 278 , and Ser 281 that may play a role in MLK-3 activation.
It is clear that MLK-3 plays a central role in integrating and * This work was supported in part by the National Cancer Institute of Canada. 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  processing inputs into SAPK, p38, and NF-B signaling pathways, however, the activation mechanism of MLK-3 has not been elucidated fully. We have demonstrated previously that dimerization via the tandem leucine zippers is required both for MLK-3 autophosphorylation and the activation of SAPK, suggesting that autophosphorylation might be an intermediate step of MLK-3 activation (27). In this report, we aimed to further define the mechanism of MLK-3 activation and examine the role of the kinase activation loop in such a mechanism.

EXPERIMENTAL PROCEDURES
Expression Constructs-The pcDNA3 HA-SEK1 and HA-HPK1 eukaryotic expression constructs were obtained from Dr. J. Woodgett and Dr. F. Kiefer, respectively. The pcDNA3 Flag-MLK-3 and pEBG GST-HA-MLK-3 expression constructs have been described previously (27). MLK-3 expression constructs bearing mutation in the activation loop were generated using a megaprimer PCR based mutagenesis method. First, 3Ј primers containing single/double mutations (T277A, GGCA-CAAAGCCACACAAATG; T278A, GGCACAAAACCGCACAAATG; S281A, AAATGGCTGCCGCGGGCACC; T277D, GGCAAAGACACAC-AAATG; S281D, AAATGGATGCCGCGGGCACC; T227D/S281D, GGC-ACAAAGACACACAAATGG) were used independently with the 5Ј primer (Mut1: TGAGGAAGGGTGACCGT) for PCR amplification. The resulting DNA fragments containing the mutations (5Ј megaprimer) together with the 3Ј primer D8AA (GATGGGCAGTGTGAGCTTGT) were then used in the second round of PCR. The second PCR products were restriction digested with HpaI and BsrGI and replaced the same DNA fragment in the pcDNA-3 Flag MLK-3/pEBG GST-HA MLK-3 wild type constructs. To generate the K144R/S281A mutant, the Pflm I-Pflm I DNA fragment of the pEBG GST-HA MLK-3 K144R was replaced with the Pflm I-Pflm I DNA fragment containing the S281A mutation. The prokaryotic expression constructs bearing the MLK-3 activation loop were generated as follows. The 5Ј primer (BamHI-DFG, TTTGGATCCCTGGCCCGAGTGGCAC) and 3Ј primer (NotI-APE, TT-TGCGGCCGCCAGGCGTAGG) bearing restriction sites were used to amplify the DNA sequence corresponding to the MLK-3 activation loop region from the wild type and the double mutants described above. The resulting DNA fragments were digested with BamHI and NotI and inserted into the NotI and BamHI sites of the pGEX 4T-2 vector (Amersham Pharmacia Biotech).
Expression and Purification of GST Fusion Protein and Peptide-The pGEX bacterial expression constructs were transformed into BL21 cells, a protease-deficient strain of Escherichia coli. The bacterial cultures were incubated overnight at 37°C, diluted 50 times into a total volume of 200 ml, and incubated at 30°C. When the O.D. reached 0.7 to 1.0, the cultures were induced with 0.2 mM isopropyl-␤-D-thiogalactoside at 30°C for 2.5 h. The bacteria were then resuspended in 10 ml of phosphate-buffered saline supplemented with 1% Triton, 10 mM dithiothreitol, 50 mM EDTA, and protease inhibitors (50 g/ml aminoethylbenzenesulfonyl fluoride, 1 g/ml antiPAIN, 1 g/ml leupeptin, 1 g/ml pepstatin, and 1 g/ml aprotinin) (ICN Pharmaceuticals), lysed on ice with 10 g/ml lysozyme and centrifuged to remove the cellular debris. The fusion proteins were then affinity purified by incubating the lysates with 200 l of the 50% glutathione-Sepharose 4B for 1 h at 4°C with gentle rotation. After washing 3 times with 10 ml of phosphate-buffered saline supplemented with 1% Triton, 10 mM dithiothreitol, and 50 mM EDTA, the glutathione-Sepharose bounded fusion proteins were either subjected to elution or thrombin digestion.
To elute the fusion proteins, the Sepharose was treated 3 times with 150 l of 10 mM reduced glutathione (dissolved in 50 mM Tris-HCl, pH 8.0) at 37°C for 5 min. The elutions were combined and spun through a Centriplus concentrator (Amicon), with a molecular weight cut-off of 10,000 daltons, to remove the reduced glutathione and concentrate the fusion proteins. The columns were subsequently washed 3 times with 500 l of phosphate-buffered saline containing protease inhibitors and spun at 4°C for 30 min and the concentrated GST fusion proteins stored at Ϫ20°C with 50% glycerol at a concentration of 5 g/l.
To release the peptide, the Sepharose was further washed 4 times with 10 ml of 50 mM Tris, pH 7.4, and 10 mM MgCl 2, then incubated with 95 l of 50 mM Tris, pH 7.4, and 5 l of 0.5 unit/l thrombin at room temperature for 75 min. The Sepharose was then centrifuged and the supernatants containing the thrombin and the peptide were spun through a Microcon centrifugal filter device (Millipore) with a molecular cut-off at 30,000 daltons to separate the thrombin and the peptide. The flow-through containing the peptide was stored at Ϫ20°C and subjected to mass spectrometry analysis to check the purity and size.
Antibodies-Anti-phospho-SAPK, anti-phospho-p38, and anti-p38 antibodies were gifts from Dr. J. McGlade. Anti-SAPK and anti-phospho-SEK1 antibodies were purchased from New England Biolabs. Mouse monoclonal M2 anti-Flag antibody was obtained from Eastman Kodak, Rochester, NY, and ascitic fluid containing anti-HA antibody was prepared from 12CA5 hybridoma cells using standard methods.
Cell Culture and Transient Transfections-HeLa cells were cultured in Dulbecco's modified Eagle's medium supplemented with 8% fetal bovine serum (Life Technologies, Inc.), 250 ng/ml Fungizone (Bristol-Myers Squibb), 200 unit/ml penicillin, and 100 unit/ml streptomycin (ICN Pharmaceuticals). In most of the experiments, HeLa cells were transfected using a vaccinia virus-based transfection method as described in Tibbles et al. (13). For the experiments involving the endogenous SAPK and p38, HeLa cells were transfected using Lipofectin (Life Technologies, Inc.) according to the manufacturer's instructions. 293 cells were cultured and transfected as described previously (27).
Immunoprecipitation and Affinity Purification-Transfected HeLa and 293 cells were harvested and purified as described previously (27) with the following modifications. Pre-chilled Nonidet P-40 lysis buffer (20 mM Tris, pH 8.0, 137 mM NaCl, 10% glycerol, and 1% Nonidet P-40) supplemented with proteinase inhibitors together with phosphatase inhibitors (1 mM sodium orthovanadate, 10 mM sodium fluoride, and 10 mM ␤-glycerophosphate) was used to lyse the transfected cells. For immunoprecipitation, antibody together with 0.1 mg/ml protein A-Sepharose (Amersham Pharmacia Biotech), was added simultaneously to the lysates for 1 h at 4°C. For affinity purification, GSTtagged MLK-3 proteins were incubated with 25-50 l of 50% glutathione 4B-Sepharose (Amersham Pharmacia Biotech) for 1 h at 4°C. Finally, the proteins bound either to protein A or glutathione-Sepharose were then washed 3 times with the Nonidet P-40 lysis buffer.
In Vitro Kinase Assays-The immunopurified Flag-MLK-3 proteins were incubated with 30 l of kinase buffer I (50 mM Tris, pH 7.4, 10 mM MgCl 2 , 1 mM EGTA) and 5 Ci of [␥-32 P]ATP for 30 min at 30°C. The immunopurified HA-HPK1 proteins together with 2.5 g of the purified GST fusion proteins were incubated with kinase buffer II (25 mM Tris, pH 7.5, 5 mM ␤-glycerophosphate, 2 mM dithiothreitol, 0.1 mM Na 3 VO 4 , and 10 mM MgCl 2 ) and 5 Ci of [␥-32 P]ATP for 30 min at 30°C. The kinase reactions were terminated by adding 2 ϫ SDS sample buffer containing 20 mM EDTA. The samples were then boiled, separated by SDS-polyacrylamide gel electrophoresis, the polyacrylamide gels were dried, and exposed to x-ray film. For the nonradioactive labeled in vitro kinase assay of the purified peptides, HA-HPK1 immunoprecipitates were incubated with the peptide in 20 l of kinase buffer II and 2 mM ATP for 35 min at 37°C. The sample was then centrifuged and the buffer containing the peptide was removed and subjected to mass spectrometry analysis.
Mass Spectrometry-Matrix-assisted laser desorption ionization mass spectrometry time of flight analysis was performed by the Mass Spectrometry Laboratory of the Molecular Medicine Research Center at the University of Toronto. In brief, 1 l of the peptide sample and 1 l of the matrix (saturated ␦-cyano-4-hydroxycinnamic acid in 50% acetonitrile, 0.1% trifluoroacetic acid) were spotted on the matrix-assisted laser desorption ionization and air-dried. The sample was then analyzed using the positive mode on a Voyager-DE STR matrix-assisted laser desorption ionization-time of flight mass spectrometer (Perseptive Biosystems, Inc.), equipped with a pulsed nitrogen laser (337 nm, 3-ns pulse).

Mutations in the Kinase Activation Loop of MLK-3 Affect
MLK-3 Autophosphorylation Activity-Many kinases are phosphorylated in the activation loop between catalytic subdomains VII and VIII. For example, mitogen-activated kinases such as the ERK2 and SAPK are phosphorylated and activated by a mitogen-activated kinase kinase in the TXY motif (31,33). The MAPKKK C-Raf is phosphorylated by PKC, whereas MEKK1 is activated by autophosphorylation in the activation loop (34 -36) (Fig. 1A). Within the putative MLK-3 activation loop between the DGF (subdomain VII) and APE (subdomain VIII) residues, there are 3 serine/threonine residues (Thr 277 , Thr 278 , and Ser 281 ) that might serve as potential phosphorylation sites.
To investigate whether any of these amino acids play a role in MLK-3 autophosphorylation, each one was independently mutated to alanine (T277A, T278A, and S281A) using PCRbased site-directed mutagenesis. The mutant kinases were assayed to determine whether they retained autophosphorylation activity. In these experiments, GST-HA-tagged MLK-3 expression constructs containing the above mutations were transfected into cells, purified using GS4B-Sepharose, and the autophosphorylation compared with the wild type and kinasedead (K144R) forms of MLK-3 via in vitro kinase assays. As shown in Fig. 1B, when the various forms of MLK-3 were expressed at equivalent levels, mutant T277A displayed autophosphorylation activity comparable to that of the kinase-dead form of MLK-3; mutant T278A retained close to normal autophosphorylation activity, whereas mutant S281A exhibited a low level of autophosphorylation activity. These results suggest that the activation loop plays a key role in MLK-3 autophosphorylation and in particular, that the Thr 277 (and Ser 281 to lesser extent) residue is crucial for MLK-3 autophosphorylation.
Autophosphorylation Is Required for MLK-3 Activation-We have proposed that MLK-3 dimerization leads to its autophosphorylation, its subsequent activation, and ultimately leads to the phosphorylation of MLK-3 substrates such as SEK1 and MKK3/6 (13). To establish whether autophosphorylation is re-quired for MLK-3 activation, we first tested the ability of the various MLK-3 activation loop mutants to phosphorylate SEK1 using a gel mobility shift assay and Western blotting using phospho-specific SEK1 (Thr 223 ) antibody. If autophosphorylation is required for MLK-3 activation, a mutant lacking autophosphorylation activity will be inactive and will not phosphorylate SEK1 (and thus will not induce an electrophoretic shift in SEK1). For these studies, HA-SEK1 was transfected either alone or together with the Flag-tagged wild type T277A, T278A, or S281A forms of MLK-3. Half of the resulting cell lysates were immunoprecipitated and Western blotted using anti-HA Ab to detect the HA-SEK1 proteins. The same blot was then probed with anti-phospho-SEK1 (Thr 223 ) antibody to detect the phosphorylated form of SEK1. To detect the various forms of MLK-3, the balance of the lysates were subjected to immunoprecipitation and Western blotting using anti-Flag Ab. As shown in Fig. 2, panel A, in the presence of the wild type MLK-3, the electrophoretic mobility of SEK1 was retarded; whereas mutation of MLK-3 Thr 277 to alanine completely abolished the SEK1 mobility shift. The MLK-3 T278A mutant induced a SEK1 mobility shift similar to that of the wild type MLK-3, whereas the S281A mutant induced an intermediate shift. Reprobing of panel A with anti-phospho-SEK1 antibody revealed that T278A (which exhibited relatively high level of autophosphorylation activity) phosphorylated SEK1 to the highest level among the three mutants. Notably, however, T278A activity was lower than wild type in this experiment. While this may due to experimental variation, we cannot rule out that mutation of this residue has an effect on MLK-3 activity toward SEK1. Mutant T277A that had very low level of autophosphorylation activity only weakly phosphorylated SEK1. Although S281A caused an intermediate shift of SEK1, the level of phospho-SEK1 detected was lower than anticipated suggesting that the SEK1 mobility shift was due to phosphorylation at a site other than the Thr 233 residue. These observations suggest that autophosphorylation is required for MLK-3 function and a correlation between the level of MLK-3 autophosphorylation activity and its ability to phosphorylate SEK1.
Autophosphorylation Is Required for MLK-3-mediated Activation of the Stress Signaling Pathways-To corroborate these findings, we next asked if activation of endogenous SAPK and p38 is also depended on MLK-3 autophosphorylation activity. For these experiments, GST-HA-tagged WT, K144R, T277A, T278A, and S281A MLK-3 expression constructs were transfected into HeLa cells. Part of the resulting cell lysates were separated on SDS-polyacrylamide gels and Western blotted with anti-phospho-SAPK, anti-SAPK, anti-phospho-p38, and anti-p38 antibodies to detect the endogenous phospho-SAPK, total SAPK, phospho-p38, and total p38 proteins, respectively. The rest of the lysates were affinity purified using glutathione-Sepharose 4B and Western blotted with anti-MLK-3 Ab. As shown in Fig. 3, similar levels of phospho-SAPK were detected following the expression of the wild type or T278A mutant. No phosphorylation of SAPK was observed with either the kinasedead or the T277A mutant, whereas only low levels of phospho-SAPK were detected in the presence of the S281A mutant. Very similar results were observed with p38. We consistently observed the highest levels of phospho-p38 in the presence of the wild type and T278A relative to the S281A and T277A mutants (although a basal level of the phospho-p38 was detected in the nontransfected HeLa cells). Notably, T278A mutant, which displayed a reduced ability to phosphorylate SEK1 (Fig. 2), appeared to activate SAPK as efficiently as wild type. The fact that MLK-3 can activate SAPK via other signaling molecules, for example, SEK4, may account for this difference. These data suggest that the amount of activated SAPK and p38 is dependent on the level of MLK-3 autophosphorylation activity. Collectively these data indicate that autophosphorylation is crucial for MLK-3 activation and the subsequent phosphorylation and activation of two separate stress signaling pathways.
Both Thr 277 and Ser 281 Residues May Serve as the MLK-3 Autophosphorylation Sites-We demonstrated that substitution of the activation loop residues Thr 277 and Ser 281 with alanine either abolished or diminished the autophosphorylation activity of MLK-3. Since phosphorylation of the kinase activation loop is a prerequisite for substrate recognition of some kinases, it seemed reasonable to propose that the Thr 277 and Ser 281 residues would be essential for MLK-3 substrate recognition. To examine this possibility, we mutated these two residues into negatively charged glutamic acid residues, either independently or together (so as to mimic the phosphorylation state), and tested whether the resulting mutants were capable of activating downstream signaling leading to SAPK phosphorylation. Wild type, T277D, S281D, and T277D/S281D MLK-3 expression constructs were transfected into 293 cells and the resulting lysates analyzed for the presence of the activated endogenous SAPK using anti-phospho-SAPK antibody. As shown in Fig. 4, phospho-SAPK was detected in all of the lysates (except for the nontransfected control), suggesting that all the mutants were able to activate endogenous SAPK. These results imply that these negatively charged residues could functionally replace the Thr 277 and Ser 281 residues and allow normal substrate recognition and kinase activity. The Thr 277 and Ser 281 residues may therefore act as the MLK-3 (auto-)phosphorylation sites. If so, because the mutation of the Thr 277 residue to alanine almost completely abolished MLK-3 autophosphorylation activity, whereas the S281A mutation only reduced MLK-3 autophosphorylation activity, Thr 277 is likely to be the major autophosphorylation site of MLK-3.
HPK1 Phosphorylates the Activation Loop of MLK-3-As mentioned above, some kinases such as the mitogen-activated kinases ERK and SAPK are phosphorylated by other kinases in their activation loop thereby leading to their activation. HPK1 has also been shown to activate MLK-3 via phosphorylation (21); however, the precise site of HPK1 phosphorylation has not yet been identified. Accordingly, we investigated the possibility that HPK1 phosphorylates MLK-3 within the activation loop. We first tested whether HPK1 could phosphorylate a peptide containing the MLK-3 activation loop amino acid sequence. A DNA fragment corresponding to the activation loop was amplified via PCR and subcloned into a bacterial expression construct bearing an N-terminal GST tag. The peptide was then cleaved specifically from the purified GST fusion protein using thrombin protease and further purified. Mass spectrometry analysis indicated that the molecular mass of the resulting peptide was 2167.55 daltons, which matched the amino acid sequence "GSLAREWHKTTQMSAAGTYA," this corresponds to the expected size of the MLK-3 activation loop, except for the first 2 amino acids resulting from the thrombin cleavage site of the GST protein. HA-HPK1 was then expressed in HeLa cells, immunopurified using anti-HA Ab and incubated with the purified peptide in the presence of ATP and MgCl 2 . When the peptide was subjected to mass spectrometry analysis to detect any change in molecular weight, we observed an additional peak of 2247.71 daltons (Fig. 5). The additional 80 daltons corresponds exactly to the anticipated replacement of a hydroxyl group with a phosphate group. As a negative control, the same peptide was also incubated with a kinase-dead form of HPK1, but no additional peak was observed (data not shown). Taken together, these observations indicate that HPK1 phosphorylated one residue in the MLK-3 activation loop peptide in vitro.
HPK1 Phosphorylates the Ser 281 Residue of the MLK-3activation Loop-To determine whether Thr 277 , Thr 278 , or Ser 281 residue(s) was phosphorylated by HPK1, we next examined whether HPK1 could phosphorylate GST fusion proteins containing the same region of the MLK-3 activation loop with only one of the three potential residues unchanged. Accordingly, double mutants T277A/T278A, T277A/S281A, and T278A/ S281A were generated. (The C-terminal sequences together with that of the wild type and the vector are listed in Fig. 6A.) Affinity purified GST fusion proteins were incubated with immunopurified HA-HPK1 in an in vitro kinase. As shown in Fig.  6B, in the presence of equivalent amounts of GST fusion proteins, HPK1 specifically phosphorylated the GST fusion proteins bearing the wild type or the T277A/T278A double mutant, but not the T277A/S281A or T278A/S281A double mutants or the GST control. When the GST wild type activation loop fusion protein was incubated with purified activated SAPK and ERK2, it was not phosphorylated by either kinase (data not shown) confirming that the HPK1 phosphorylation is specific.
These results indicate that Ser 281 is the HPK1 phosphorylation site since HPK1 phosphorylated the fusion protein only when the Ser 281 residue was preserved, as in the case of the wild type and the T277A/T278A double mutant. Therefore, Ser 281 not only serves as potential MLK-3 autophosphorylation site but is also a HPK1 phosphorylation site.
To determine whether HPK1 phosphorylates only the Ser 281 residue within the entire MLK-3 protein, a kinase-dead mutant containing the S281A mutation was generated and the ability of HPK1 to phosphorylate this mutant was tested using in vitro kinase assay. GST-HA-tagged K144R or K144R S281A constructs were transfected alone or together with HA-tagged HPK1. The expressed proteins were immunoprecipitated using anti-HA antibody; one-third of the precipitates were subjected to Western blotting and the balance of the samples were evaluated via in vitro kinase assay. As shown in Fig. 7, both K144R and K144R/S281A mutants exhibited undetectable levels of kinase activity. HPK1 was able to phosphorylate the K144R form of MLK-3, however, only a faint band (upon prolonged exposure) of the phosphorylated K144R/S281A form was observed. The Ser 281 residue therefore acts as the major HPK1 phosphorylation site. Notably, the autophosphorylation activity of HPK1 was greatly reduced in the presence of the K144R/ S281A MLK-3 mutant and also slightly reduced in the presence of the K144R mutant, suggesting the MLK-3 Ser 281 phospho-FIG. 5. HPK1 phosphorylates a peptide containing the MLK-3 activation loop amino acid sequence. A peptide of the amino acid sequence "GSLAREWHKTTQMSAAGTYA" was produced as described under "Experimental Procedures." The amino acid sequence of this peptide (except the first two amino acids "GS") corresponds to the putative MLK-3 activation loop amino acid sequence. The purified peptide was incubated with HA immunoprecipitates that contained the HA-HPK1 proteins, along with Mg 2ϩ and ATP. After incubation, the samples were analyzed by a mass spectrometer. An additional peak of 2247.7 daltons was observed only after the incubation with the wild type (as shown in the figure), but not the kinase-dead form of HA-HPK1 (data not shown). The box labeled GST denotes the GST moiety; the capital letters represent the amino acids encoded by the MLK-3 cDNA sequences that correspond to the activation loop, and the lower case letters represent the amino acids encoded by the vector. In addition, the underlying letters represent the mutated amino acids. Panel B, the GST fusion proteins listed in panel A were purified and incubated with the HA-HPK1 immunoprecipitates in the presence of Mg 2ϩ and ATP. The samples were then separated on a polyacrylamide gel, stained, dried, and exposed to x-ray film. The autoradiograph in the upper panel shows the phosphorylated GST fusion protein whereas the Coomassie staining in the lower panel illustrates the total amount of the GST fusion protein used in these experiments. rylation site may compete with the HPK1 autophosphorylation site(s).

DISCUSSION
The kinase activation loop, defined as the region spanning between the subdomains VII and VIII, has been shown in a number of kinases to regulate catalytic activity (28). In this study, we examined the role of kinase activation loop in MLK-3 activation. First, we mutated the three potential phosphorylation residues in the kinase activation loop to investigate whether these residues are required for MLK-3 autophosphorylation activity. We found that mutating residue Thr 277 to alanine significantly reduced MLK-3 autophosphorylation activity. This same MLK-3 mutant only had negligible phosphorylation activity toward its substrate SEK1 and did not activate either JNK/SAPK or p38/RK. The S281A mutant displayed a reduced level of autophosphorylation activity, and reduced activation of SEK1, JNK/SAPK, and p38/RK, whereas the T278A mutant, which exhibited close to normal autophosphorylation activity, can almost fully activate these proteins. These results clearly demonstrate that the activation loop is crucial for MLK-3 autophosphorylation and its subsequent activation and suggest that the MLK-3 autophosphorylation level correlates with activity.
Mutation of Thr 277 and Ser 281 might result in the loss of MLK-3 autophosphorylation activity in two different ways; first, it may affect the MLK-3 substrate binding ability and second, it may eliminate the autophosphorylation site. Our data show that when the Thr 277 and Ser 281 residues were mutated to acidic residues, so as to mimic negatively charged phosphorylated serine or threonine residues, the ability of MLK-3 to activate the SAPK pathway was retained. These findings suggest that MLK-3 is functional when the Thr 277 and Ser 281 residues are phosphorylated and support the idea that these two residues may, in fact, serve as the MLK-3 (auto)phosphorylation sites. We found that when Thr 277 is mutated, MLK-3 autophosphorylation activity is minimal whereas when Ser 281 is mutated, there are still some levels of autophosphorylation activity indicating that Thr 277 can still be phosphorylated independent of Ser 281 . One explanation for these observations is that Thr 277 is required for the subsequent phosphorylation of the Ser 281 . An alternative possibility is that the Thr 277 is the only autophosphorylation site within the activation loop and mutation of Ser 281 simply affects the "self" recognition and subsequent autophosphorylation of Thr 277 . One way to distinguish these possibilities will be to use phospho-specific antibodies when they become available.
Our results also indicate that all three residues (Thr 277 , Thr 278 , and Ser 281 ) are important for efficient phosphorylation of SEK1 and therefore, substitution of any of these neutral hydrophilic residues to hydrophobic alanine residues may affect substrate binding specificity. However, since mutation of Thr 277 and Ser 281 to alanine have substantial effects on SEK1 phosphorylation whereas mutation of Thr 278 to alanine only resulted in relatively minor reduction of SEK1 phosphorylation seems likely that both Thr 277 and Ser 281 but not Thr 278 need to be phosphorylated for normal MLK-3 function toward SEK1.
Because MLK-3 lies downstream of HPK1 in the stress signaling pathway and is directly phosphorylated by HPK1 (21), we also asked whether HPK1 could activate MLK-3 by phosphorylating MLK-3 activation loop. We identified the Ser 281 residue as the HPK1-phosphorylation site, indicating that both autophosphorylation and HPK1 phosphorylation could activate MLK-3 via the kinase activation loop which underscores the complex regulation of MLK-3 activation. This mode of regulation, by autophosphorylation and transphosphorylation by another kinase(s) within the activation loop has also been reported for the tyrosine kinase Lck (37).
We have shown previously that MLK-3 homodimerization is required for its autophosphorylation, critical for its activation, and can be induced by Cdc42 (27). In this report, we in fact demonstrated that autophosphorylation is essential for MLK-3 activation. Taken together, these data indicate that MLK-3 employs an activation mechanism similar to that of receptor tyrosine kinases; that is, induced dimerization leads to autophosphorylation and the subsequent activation of MLK-3. Based on the fact that all MLK family members (except MLK1 which has not been reported) display autophosphorylation activity and bear tandem leucine zippers (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12), it is reasonable to speculate that all MLK family members utilize a dimerization-based activation mechanism which ultimately leads to autophosphorylation and activation. Within the MLK family, DLK has also been shown to also employ a dimerization-based activation mechanism (38). Moreover, it is noteworthy that the two residues corresponding to the MLK-3 potential autophosphorylation sites are conserved among other MLK family members. As shown in Fig. 8, the Thr 277 residue is conserved in MLK1 and MLK2 (but substituted with a serine in DLK/MUK/ ZPK and LZK), whereas the Ser 281 residue is conserved in all MLK family members. Hence it is temping to hypothesize that these two residues may also be crucial for the autophosphorylation, thus activation, of other MLK family members.