Isolation of the Protein Kinase TAO2 and Identification of Its Mitogen-activated Protein Kinase/Extracellular Signal-regulated Kinase Kinase Binding Domain*

We previously reported the cloning of the thousand and one-amino acid protein kinase 1 (TAO1), a rat homolog of the Saccharomyces cerevisiae protein kinase sterile 20 protein. Here we report the complete sequence and properties of a related rat protein kinase TAO2. Like TAO1, recombinant TAO2 selectively activated mitogen-activated protein/extracellular signal-regulated kinase kinases (MEKs) 3, 4, and 6 of the stress-responsive mitogen-activated protein kinase pathways in vitro and copurified with MEK3 endogenous to Sf9 cells. To examine TAO2 interactions with MEKs, the MEK binding domain of TAO2 was localized to an ∼135-residue sequence just C-terminal to the TAO2 catalytic domain. In vitro this MEK binding domain associated with MEKs 3 and 6 but not MEKs 1, 2, or 4. Using chimeric MEK proteins, we found that the MEK N terminus was sufficient for binding to TAO2. Catalytic activity of full-length TAO2 enhanced its binding to MEKs. However, neither the autophosphorylation of the MEK binding domain of TAO2 nor the activity of MEK itself was required for MEK binding. These results suggest that TAO proteins lie in stress-sensitive kinase cascades and define a mechanism by which these kinases may organize downstream targets.

Ste20p 1 was originally isolated as a gene of budding yeast whose product functioned downstream of the ␤␥ subunits of a heterotrimeric G protein but upstream of enzymes in the MAP kinase module of the pheromone response pathway (1,2). Several mammalian protein kinases related to Ste20p have been identified that phosphorylate MAP/ERK kinase (MEK) family members in stress-activated MAP kinase cascades. These in-clude mixed lineage kinases, TGF-␤-activated protein kinase, and TAO1 (3)(4)(5)(6)(7)(8). In the yeast protein kinase family tree the Ste20p branch is closest to the MEK kinases (MEKKs) (9). Thus, it is not surprising that several mammalian Ste20prelated kinases are MEKKs. Some have selectivity for MEKs in the c-Jun N-terminal kinase/stress-activated protein kinase (JNK/SAPK) pathway and others for the p38 stress-sensitive pathway, whereas most phosphorylate both groups of MEKs in vitro (3, 7, 8, 10 -13). The plethora of Ste20p-like kinases with effects on stress pathways and their overlapping biochemical activities have made it difficult to define their roles in the physiological regulation of these kinase cascades. MEK3 and MEKK1 are almost certainly important for regulation of JNK/ SAPKs because they bind to JNK/SAPK and other cascade components either through a scaffold protein, with a function believed to be analogous to the yeast scaffold protein Ste5p (14,15), or directly (16,17). The association of kinases in complexes provides compelling evidence for their interrelated or dependent functions even in the absence of information regarding physiological roles.
To identify novel components of MAP kinase cascades, we isolated several PCR products and cDNAs encoding homologs of Ste20p from Saccharomyces pombe and mammals (8,18,19). Among the mammalian cDNAs, we isolated one that encoded the protein kinase TAO1, named for its one thousand and one amino acids. TAO1 is like certain other relatives of Ste20p in that it phosphorylates and activates MEKs from the stressresponsive MAP kinase cascades. Copurification experiments indicated that TAO1 interacted with MEK3, a p38 activator, although direct binding was not demonstrated. These findings suggested that TAO1 forms complexes with components of p38 MAP kinase cascades and may, therefore, be an important regulator of p38-dependent events.
Here we report the isolation of cDNA clones encoding the complete sequence of TAO2, a close relative of TAO1. Both TAO1 and TAO2 are expressed most highly in brain cells, suggesting their tissue-restricted function (8). The in vitro substrate specificities of TAO1 and TAO2 are also similar. Importantly, TAO2, like TAO1, copurifies with MEK3 endogenous to Sf9 cells. This suggests that the intracellular specificity of TAO proteins may be determined by their ability to bind stably to a subset of potential MEK substrates. To define the mechanism by which TAO proteins associate with MEKs, we determined that they interacted directly, identified the MEK binding domain of TAO2, and examined the MEK specificity of this domain.

EXPERIMENTAL PROCEDURES
Isolation of cDNA Clones Encoding TAO2-A 420-base pair PCR product was obtained as described (8) 1 The abbreviations used are: Ste20p, sterile 20 protein; MAP, mitogen-activated protein; ERK, extracellular signal-regulated protein kinase; MEK, MAP/ERK kinase or MAP kinase kinase; MEKK, MEK kinase; TAO, thousand and one amino acid protein kinase; GST, glutathione S-transferase; MBP, myelin basic protein; JNK, c-Jun N-terminal kinase; SAPK, stress-activated protein kinase; PCR, polymerase chain reaction; DTT, dithiothreitol. of a random-primed adult rat forebrain cDNA library and approximately 0.6 ϫ 10 6 plaques of an oligo(dT)-primed rat brain cDNA library (both provided by Jim Boulter, UCLA). cDNA clones encoding TAO1 and TAO2 were obtained. Subsequent rounds of screening yielded the full-length TAO2 cDNA, which was assembled into pBluescript from 3 of over 50 positive clones. The complete sequence of the assembled cDNA was deposited in GenBank TM with the accession number AF140556.
Plasmid Construction-pBluescript-TAO2- , containing the catalytic domain of TAO2, and a catalytically defective mutant pBluescript-TAO2D169A were generated by PCR. Wild-type TAO2, TAO2D169A, and TAO2-(1-320) were cloned into pRSETB (Invitrogen) to incorporate a MRGSH 6 tag and subsequently transferred into the baculoviral shuttle vector pVL1393. Recombinant viruses were selected as described (8). For expression in mammalian cells, the cDNAs encoding these TAO2 proteins were also cloned into pCMV5 that had been modified to place a Myc epitope tag at the N terminus of the encoded protein. A truncated, catalytically defective TAO2 in pRSETB was created by changing lysine 57, in the conserved VAIK motif, to alanine (K57A) by PCR.
For binding assays, fragments of TAO2 were subcloned into pGEX-KG by PCR. TAO2-(314 -451) was subsequently transferred into pRSETA utilizing the BamHI and EcoRI restriction sites. Catalytically defective MEK3 was created in pNPT7-5 by changing lysine 64 to methionine (K64M). A MEK1/6 chimera, which contains the N-terminal domain of MEK1 and the C-terminal domain of MEK6, and a MEK6/1 chimera with the reciprocal domains (see Fig. 4B) were transferred into pRSETA or -C, respectively, from the original pGEX-KG-MEK1/6 and MEK6/1 plasmids (generously provided by Lori Christerson) utilizing the BamHI and HindIII restriction sites.
Immunoprecipitation and Affinity Purification from Transfected 293 Cells and Sf9 Cells-pCMV5-Myc-TAO2 constructs were transfected into 293 cells using calcium phosphate (21). After 48 h, cells were lysed (22), and transfected proteins were detected by anti-Myc Western blotting. Lysate volumes containing equal amounts of expressed protein were used for subsequent immunoprecipitation with anti-Myc antibodies for kinase assays. Sf9 lysates containing His 6 -TAO2 proteins were incubated with Ni 2ϩ -nitrilotriacetic acid-agarose in buffer containing 0.15 M NaCl and 0.5 mM DTT and washed with 0.3 M NaCl, 0.5 mM DTT, and 10 mM imidazole. Bound proteins were eluted with 250 mM imidazole in buffer and subjected to Western blotting with an anti-MEK3 antibody (23).
In Vitro Binding Assays-For binding assays involving GST-tagged TAO2 fragments and His 6 -tagged MEK proteins, 3 g of each GST fusion protein or GST alone was incubated with glutathione-agarose beads at 4°C in the presence of 0.1 mg/ml bovine serum albumin for 30 min and washed with 0.1 M NaCl in 50 mM Tris, pH 7.4. 5 g of His 6 -tagged protein were incubated with the beads in the presence of 0.1 mg/ml bovine serum albumin and 0.1 M NaCl at 4°C for 1 h. The beads were washed with 0.3 M NaCl, 0.1% Triton X-100, and 50 mM Tris, pH 7.4. Bound proteins were released with 1ϫ SDS electrophoresis sample buffer and subjected to anti-His 6 Western blotting. Similar binding assays were performed for His 6 -TAO2-(314 -451) and GSTtagged MEK proteins.

RESULTS
Isolation of TAO2 cDNAs-Degenerate oligonucleotide primers designed from the sequence of the Saccharomyces cerevisiae Ste20p kinase were used in PCR to amplify fragments of related protein kinases from rat cDNAs. One PCR product was used in isolating overlapping cDNAs from two rat brain cDNA libraries that encoded two protein kinases, TAO1 (8) and the related kinase TAO2, described here. The assembled TAO2 cDNA predicted an open reading frame of 993 amino acids (Fig.  1A). The presumed start codon is located at base 193 and is preceded by an in-frame stop codon at base 145. The longest 3Ј-untranslated region was 1317 base pairs in length, including a poly(A) track at its end, ϳ1.3 kilobase pairs 3Ј to the stop codon (not shown).
Amino Acid Sequence of TAO2-The deduced TAO2 protein has a calculated molecular mass of 114 kDa. The serine/threonine protein kinase catalytic domain is at its N terminus. In its 690 C-terminal residues, TAO2 contains a possible nucleotide binding site, a serine-rich region, and a proline and leucine-rich region, all shared with TAO1, and an unbroken stretch of 17 glutamic acid residues unique to TAO2. Like TAO1, TAO2 does not appear to contain a small G protein binding consensus motif found in several other Ste20p relatives (14). The TAO2 protein kinase domain displays 90 and 63% identity to TAO1 and the Caenorhabditis elegans TAO ortholog (CeTAO, accession number U32275), respectively (not shown). TAO2 displays marked similarities to TAO1 and the C. elegans kinase outside the catalytic domain (Fig. 1B).
Expression and Activity of TAO2-Truncated, recombinant TAO2-(1-320) purified from Sf9 cells phosphorylated MBP with a specific activity of 0.6 mol⅐min Ϫ1 ⅐mg Ϫ1 . The full-length protein purified on MonoQ had lower intrinsic activity, about 10% of the truncated enzyme (not shown). Kinase-deficient mutants, His 6 -TAO2D169A expressed in Sf9 cells and purified on MonoQ or His 6 -TAO2K57A expressed in bacteria, were inactive toward MBP in vitro. TAO2 and TAO2-(1-320) expressed in either Sf9 or mammalian cells autophosphorylated extensively on serine and threonine residues (data not shown).
TAO2 Activates MEK3, MEK4, and MEK6 in Vitro-TAO1 was previously shown to activate MEKs 3, 4, and 6 in vitro. We therefore examined the ability of TAO2 to activate MEK family members. TAO2-(1-320) produced in Sf9 cells was subjected to a linked kinase assay by incubating it with recombinant MEKs produced in bacteria in the presence of ATP. Aliquots of the first stage reactions were transferred to second reactions to measure the phosphorylation of appropriate MAP kinase substrates by the recombinant MEKs ( Fig. 2A). TAO2-(1-320) activated MEK3 and MEK6 40-and 20-fold, respectively, toward their substrate p38 (Fig. 2B). TAO2 also increased the ability of MEK4 to phosphorylate its substrate SAPK by 7-fold. TAO2-(1-320) was unable to increase the activity of MEK1 or MEK2 toward their substrate K52R ERK2. Full-length TAO2 displayed about 20% of the MEK3-activating ability of TAO2-(1-320), consistent with its lower activity toward MBP. Neither TAO2 mutants D169A nor K57A activated any of the MEKs (data not shown). TAO2-(1-320) expressed in 293 cells also enhanced the ability of MEK3 and MEK4 to phosphorylate their substrates (not shown).
To determine the domain in TAO2 that mediates the interaction with MEK3, the series of fragments that span the noncatalytic domains of TAO2 were expressed as GST fusion proteins and tested for their abilities to bind His 6 -MEK3 in vitro (Fig. 3D). The MEK3 binding domain was localized to an ϳ135residue region, residues 314 -451, just C-terminal to the TAO2 catalytic domain. This region was further subdivided, but all of the shorter fragments containing residues 395-451 were degraded. TAO2-(314 -377), which precedes the polyglutamic acid region, was insufficient for MEK3 binding.
TAO2 Binds MEKs 3 and 6 in Vitro but Not MEKs 1, 2, or 4-To investigate the binding specificity of the TAO2 MEK binding domain, His 6 -tagged MEK proteins were compared for their capacity to bind to TAO2-(314 -451). The TAO2 fragment bound MEK6 in addition to MEK3 but not MEK1, MEK2, or MEK4 (Fig. 4, A and D). Binding to both MEK3 and MEK6 is 321-993 demonstrates significant similarity outside their kinase domains. Identical residues are boxed in black and conserved residues are shaded. consistent with their significant sequence similarity compared with the other MEK family members. Chimeric proteins generated from MEK6 and MEK1 (Fig. 4B) were used to determine the portion of the MEK that binds to the TAO2 domain. His 6 -MEK1/6 was unable to bind to TAO2-(314 -451), whereas GST-MEK6/1 is as efficient as GST-MEK6 in binding to the TAO2 fragment (Fig. 4, C and D).
As noted earlier, catalytically defective TAO2 was deficient in MEK3 binding. To explore the underlying reason, we asked whether autophosphorylation of TAO2 might have an effect on its ability to bind to MEK3. The MEK3 binding fragment of TAO2 was autophosphorylated by the catalytic domain of TAO2 on both serine and threonine residues (Fig. 5, A and B). We, thus, first phosphorylated TAO2-(314 -451) with TAO-(1-320) for different lengths of time to determine whether phosphorylation would alter its binding activity. Different concentrations of ATP and Mg 2ϩ were also tested in the binding assay. Little or no effect of the autophosphorylation state or [ATP⅐Mg 2ϩ ] on MEK3 binding activity was observed (Fig. 5C).
To determine whether MEK3 kinase activity was necessary for binding to TAO2, the binding of kinase-inactive MEK3 (K64M) was tested (Fig. 5D). This defective mutant binds to TAO2 as well as wild-type MEK3, suggesting that MEK3 kinase activity is dispensable for interaction with TAO2. DISCUSSION We isolated cDNAs encoding TAO2, a homolog of the previously reported TAO1 (8). We found that TAO2, like TAO1, activated MEKs in the stress-responsive MAP kinase pathways and displayed stable binding to MEK3 endogenous to Sf9 cells. In examining TAO2 expressed in Sf9 cells, we found that the full-length enzyme was significantly less active than the truncated kinase. Thus, the full-length protein was inhibited relative to its truncated forms. Subsequent work indicated that full-length TAO1 is also less active than proteins with C-terminal domain truncations. The inherently higher activity of fragments of TAO1 and -2 suggested that we may have removed an autoinhibitory or pseudosubstrate domain. However, we have not yet identified such a domain, as none of the recombinant fragments from the putative regulatory domain of TAO2 inhibited the activity of its catalytic domain (not shown).
Because TAO2 was purified in a stable complex with MEK3

FIG. 3. Identification of the TAO2 MEK binding domain.
A, His 6 -tagged TAO2, TAO2-(1-320), and TAO2D169A were expressed in separate batches of Sf9 cells, and the proteins were detected with an antibody that recognizes the N-terminal epitope. Comparable amounts of TAO2 proteins were detected in each lysate. B and C, His 6 -tagged TAO2, TAO2-(1-320), and TAO2D169A were purified from cell lysates on Ni 2ϩ -nitrilotriacetic acid-agarose and subjected to anti-MEK3 Western blotting to detect associated MEK3 that was endogenous to Sf9 cells. Lysates from Sf9 cells not expressing recombinant protein were processed as a control. One of three comparable experiments is shown. The same experiment was also performed in Sf900 cells with a similar result. D, TAO2 C-terminal fragments were expressed as GST fusion proteins in bacteria and tested for MEK3 binding activity. MEK3 binding was measured by immunoblotting the proteins bound to the beads with anti-His 6 antibodies. His 6 -MEK3 was loaded in the last lane as positive control. Binding reactions were performed from five to eight times for the various TAO2 fragments. TAO2-(314 -451). A, binding of GST-TAO2-(314 -451) to His 6 -tagged MEK family members. The MEK family members associated with bead-bound GST-TAO2-(314 -451) were detected using anti-His 6 antibodies. The last four lanes contain purified MEK1, -2, -3, and -6, respectively, to show their positions on the gel. One of three similar experiments is shown. B, chimeric proteins were derived from MEK6 and MEK1. MEK1/6 consists of the N terminus of MEK1 and the C terminus of MEK6, whereas the reciprocal chimera MEK6/1 consists of the N terminus of MEK6 and the C terminus of MEK1. C, binding of GST-TAO2-(314 -451) to His 6 -tagged MEK6 or MEK1/6. MEK1/6 chimeras were used to determine the portion of MEK6 involved in TAO2 binding as described in the legend to A. One of two similar experiments is shown. D, binding of His 6 -TAO2-(314 -451) to GST-tagged MEK1, -4, -6, or -6/1. Binding to MEK4 was tested, and the tags on the MEK family members and TAO2 proteins were reversed to confirm that the tags had no effect on their protein-protein association. Binding was detected as in A. His 6 -TAO2-(314 -451) was loaded in the last lane as a positive control. One of two similar experiments is shown. endogenous to Sf9 cells, we localized the MEK binding domain to a small, ϳ135-amino acid fragment, residues 314 -451, just C-terminal to the catalytic domain of TAO2. The N-terminal half of this fragment, residues 314 -377, did not bind to MEK3. Because TAO1 and TAO2 both bind MEKs but TAO1 has no polyglutamate stretch, it seems unlikely that these residues participate in MEK binding. Thus, residues from 395 to 451 are most likely required for the stable association with MEKs. These results are consistent with the weak binding of TAO1-(1-416) to MEK3 compared with the strong binding displayed by full-length TAO1 (8) and suggest that residues 404 -446, which are well conserved between TAO2 and TAO1, contain the MEK binding domain.

FIG. 4. Specificity of MEK binding to
Because TAO1 and -2 can activate MEKs 3, 4, and 6 in vitro, we determined the specificity of the MEK binding domain of TAO2. We found that TAO2 binds to MEK3 and MEK6, but not to MEK4, despite the fact that MEK4 is an in vitro substrate. The N terminus of the MEK is required for this binding, whereas the C terminus is dispensable. This behavior may be a general property of the organization of MAP kinase cascades. The N termini of other MEK family members contain binding domains for proteins in their cascades. MEK1 binds with high affinity to ERK2 through a basic motif N-terminal to its catalytic domain. MEK1 has been proposed to retain ERK2 in the cytoplasm of unstimulated cells through binding to this site (26), and activation of ERK2 may be impaired if this binding domain is absent. 2 MEK4 is reported to require its N-terminal extension to interact with both MEKK1, an activator, and its substrates, JNK/SAPKs (17). An inhibitory interaction between MEK4 and JNK/SAPKs has also been mapped to this N-terminal domain. 3 This suggests that the stable association of MEK3 or MEK6 with TAO proteins will link their physiological functions to p38 but not JNK/SAPK pathways by restricting their intracellular targets. Future biochemical studies will focus on determining the functions of the other domains of TAO1 and TAO2.