TAB1β (Transforming Growth Factor-β-activated Protein Kinase 1-binding Protein 1β), a Novel Splicing Variant of TAB1 That Interacts with p38α but Not TAK1

The mitogen-activated protein kinases (MAPKs) play an important role in a variety of biological processes. Activation of MAPKs is mediated by phosphorylation on specific regulatory tyrosine and threonine sites. We have recently found that activation of p38α MAPK can be carried out not only by its upstream MAPK kinases (MKKs) but also by p38α autophosphorylation. p38α autoactivation requires an interaction of p38α with TAB1 (transforming growth factor-β-activated protein kinase 1-binding protein 1). The autoactivation mechanism of p38α has been found to be important in cellular responses to a number of physiologically relevant stimuli. Here, we report the characterization of a splicing variant of TAB1, TAB1β. TAB1 and TAB1β share the first 10 exons. The 11th and 12th exons of TAB1 were spliced out in TAB1β, and an extra exon, termed exon β, downstream of exons 11 and 12 in the genome was used as the last exon inTAB1β. The mRNA of TAB1β was expressed in all cell lines examined. The TAB1β mRNA encodes a protein with an identical sequence to TAB1 except the C-terminal 69 amino acids were replaced with an unrelated 27-amino acid sequence. Similar to TAB1, TAB1β interacts with p38α but not other MAPKs and stimulates p38α autoactivation. Different from TAB1, TAB1β does not bind or activate TAK1. Inhibition of TAB1β expression with RNA interference in MDA231 breast cancer cells resulted in the reduction of basal activity of p38α and invasiveness of MDA231 cells, suggesting that ΤΑΒ1β is involved in regulating p38α activity in physiological conditions.

Activation of MAPK is mediated through phosphorylation of specific tyrosine and threonine residues by upstream MAPK kinase (MAPKK), which is activated through phosphorylation of serine/threonine residues by MAPKKK further upstream (4, 16 -18). ERK family kinases are activated by the MEK1 and MEK2, JNK by MKK4 and MKK7, and p38 by MKK3 and MKK6 (3,4,10,19). Further upstream, Raf-1, A-Raf, B-Raf, and MOS function as MAPKKKs in the ERK activation pathway (20,21). A subgroup of MAPKKK, including MEKK1-4, MTK1, MLK1-3, DLK/MUK, ASK1, Tpl-2/Cot, and TAK1, activates the MAPKKs that phosphorylate JNK and perhaps p38 as well (4,7,17,22). Biochemical and genetic evidence have supported the above three-kinase module of MAPK activation (1,7,18). In addition, the MAPK cascades are evolutionarily conserved (1,18). However, we have recently demonstrated an alternative activation pathway of p38␣ MAPK that did not require upstream kinases. We found that TAB1 interacts with p38␣ and leads to the activation of p38␣ (23). TAB1 is a protein that was initially described as an activator of a member of MAPKKK, TAK1, in response to stimulation of transforming growth factor-␤ (24). When ectopically expressed together with TAK1, TAB1 interacts with TAK1 and leads to enhancement of TAK1 activation (24 -29). The C-terminal 68-amino acid portion of TAB1 is sufficient for binding to and activation of TAK1 (24,25,28). Activation of TAK1 by TAB1 is mediated via autophosphorylation of Ser-192 within its kinase activation loop (26). We found that TAB1-mediated p38␣ activation also occurs through an autoactivation mechanism (23). However, the portion of the TAB1 protein that is responsible for p38␣ interaction and activation resides within amino acids 373-418, which is N-terminal to the TAK1 binding site (23,24).
In this report we describe the characterization of a splicing variant of TAB1, TAB1␤. TAB1␤ is an alternative splicing product of the TAB1 gene. TAB1␤ and TAB1 have identical amino acid sequences with the exception of the C terminus. Similar to TAB1, TAB1␤ interacts with p38␣ and induces p38␣ autoactivation. Different from TAB1, TAB1␤ did not bind with TAK1 and had no effect on TAK1 activation. The present findings suggest that the alternative splicing of TAB1 has a role in intracellular signaling transduction.

EXPERIMENTAL PROCEDURES
cDNA of TAB1␤-We completely sequenced two EST clones that showed homologues to TAB1 (clone identification numbers: 50210 and 3049027). One of the clones showed 100% homology with the 5Ј-end portion of TAB1 cDNA but different 3Ј-end sequence. This cDNA was termed as TAB1␤, and the cDNA sequence has been deposited in GenBank TM (AF425640).
Transfection of Cells-HEK293 or MDA231 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 5% fetal bovine serum, 2 mM glutamine, and 100 g/ml penicillin and streptomycin. Cells on six-well plates were transiently transfected with 1 g (total) of plasmid DNA using LipofectAMINE. After 36 h, the cells were treated with stimuli as described in the text. The plasmid pCMV␤ (Clontech, Palo Alto, CA) was co-transfected, and transfection efficiency was normalized by quantifying ␤-galactosidase activity.
Western Blot and Immunoprecipitation Analysis-Total cell lysates were prepared using a lysis buffer A: 20 mM Tris-HCl, pH 7.5, 120 mM NaCl, 10% glycerol, 1 mM Na 3 VO 4 , 2 mM EDTA, 1 mM phenylmethanesulfonyl fluoride, 1% Triton X-100. Equal protein loading of cell extracts in SDS-PAGE was determined by Bio-Rad protein assay solution (Bio-Rad, Hercules, CA) and by staining the transferred nitrocellulose membrane with Ponceau S solution (Sigma, St. Louis, MO). Standard Western blot methods were used (35). Rabbit polyclonal antibodies raised against bacterially expressed recombinant His-p38 protein, rabbit polyclonal antibodies raised against bacterially expressed recombinant His-TAB1␤ protein, polyclonal goat anti-TAB1 (N-19) (Santa Cruz Biotechnology, Santa Cruz, CA), anti-FLAG M2 mAb (Sigma, St. Louis, MO), anti-HA mAb (12CA5) (Roche Molecular Biochemicals, Indianapolis, IN), and anti-phospho p38 (New England BioLabs, Beverley, MA) were used in immunoblotting. Selective viewing of ectopically overexpressed TAB1 and TAB1␤ by Western blotting was achieved by adjusting exposure time. For co-immunoprecipitation, cell lysates prepared as described above were incubated with anti-FLAG M2 beads (Sigma) and gently shaken for 4 h at 4°C. The beads were washed three times with lysis buffer and 1 time with 50 mM Tris, pH 6.8. Then, the SDS sample buffer (50 l) was added and heated for 5 min at 100°C. The supernatant was applied to SDS-PAGE and detected by immunoblotting.
Reporter Gene Assay-Cells were grown on 35-mm-diameter multiwell plates and transiently transfected with GAL4-responsive lucifer-ase plasmid or the NF-B-dependent luciferase reporter plasmid (36,37). A ␤-galactosidase expression plasmid (pCMV-␤-gal, Clontech, Palo Alto, CA) was used to control for transfection efficiency. The total amount of DNA for each transfection was kept constant by using the empty vector pcDNA3. Cell extracts were prepared 36 h later, and ␤-galactosidase and luciferase activities were measured.
Preparation of Recombinant Proteins-GST fusion proteins of TAB1, TAB1␤, and MKK4 were expressed in Escherichia coli strain BL21 and purified using glutathione-Sepharose 4B beads (Amersham Biosciences, Uppsala, Sweden). All of the His 6 -tagged recombinant proteins were expressed in the BL21(DE3) strain and purified using the nickel-nitrilotriacetic acid purification system (Qiagen, Valencia, CA). Myelin basic protein (MBP) was purchased from Sigma.
Protein Kinase Assays-In vitro kinase assays were carried out at 37°C for 30 min using ϳ0.2 g of recombinant kinase or recombinant protein, 5 g of kinase substrate, 250 M ATP, and 10 Ci of [␥-32 P]ATP in 20 l of kinase reaction buffer. Reactions were terminated by the addition of Laemmli sample buffer. Reaction products were resolved by 12% SDS-PAGE, and the extent of protein phosphorylation was visualized by autoradiography. For the transiently expressed protein, FLAG-p38␣ protein in 293 cells was immunoprecipitated with anti-M2 beads. After washing three times with lysis buffer and one time with kinase reaction buffer, the beads were further used in the analysis of kinase activity.
RNAi-RNAi constructs were constructed using pSuper vector (38). The hairpin RNA encoded by R1 and R1(M) constructs is shown in Fig.  2C. The sequences of other RNAi constructs, which cannot decrease in endogenous genes, are available upon request. Each of the pSuper constructs was co-transfected with pSVNeo into breast cancer cell line MDA231, and stable lines were selected by G418 at 1 mg/ml concentration.
Matrigel Invasion Assay-Cell invasion was analyzed using BIO-COAT Matrigel invasion chambers (BD Biosciences) according to the manufacturer's instructions. Briefly, 2 ϫ 10 5 cells in 300 l were added to each chamber and allowed to invade Matrigel for 20 h at 37°C and 5% CO 2 atmosphere. The non-invading cells on the upper surface of the membrane were removed from the chambers, and the invading cells on the lower surface of membrane were stained with a Diff-Quick stain kit (BD Biosciences). After two washes with water, the chambers were allowed to air dry. The number of invading cells was then counted using a phase-contrast microscope.

RESULTS
Identification of TAB1␤-In our study of TAB1, we have sequenced a couple of EST clones that have sequence homology with TAB1. The cDNA of one of the EST clones appeared to be a splicing variant of TAB1; its sequence was 100% identical to TAB1 in the 5Ј-portion but different at the 3Ј-end. The protein encoded by this cDNA was termed TAB1␤ and shown is in Fig.  1A in comparison with TAB1. The N-terminal sequence of amino acids 1-435 of TAB1␤ is identical to TAB1. Different from ⌻〈〉1, ⌻〈〉1␤ lacks the C-terminal 69 amino acids found in TAB1 and has an unrelated 29-amino acid sequence instead (Fig. 1A). Because the C-terminal 68-amino acid portion of TAB1 was mapped as a TAK1 binding domain (25, 28), TAB1␤ does not have the TAK1 binding domain.
To determine whether TAB1 and TAB1␤ are splicing vari- ants, we analyzed the gene structure in the TAB1 locus. We found that the TAB1 gene is composed of 12 exons. TAB1␤ shares 1-10 exons with TAB1. Exons 11 and 12 of TAB1 were spliced out in TAB1␤, and an exon further downstream termed exon ␤ was the 11th exon of TAB1␤ (Fig. 1B). Thus, the difference between TAB1 and TAB1␤ is indeed generated by alternative splicing.
Expression of TAB1␤-To investigate the expression of TAB1␤ in different cells, total RNAs were extracted from HEK293, HPEG2, MDA-231, and RPMI 8226 cells. The existence of TAB1␤ and TAB1 transcripts was examined by RT-PCR using their specific primers. Both TAB1 and TAB1␤ transcripts were found in all cell lines tested ( Fig. 2A), suggesting that TAB1 and TAB1␤ are concurrently expressed.
To examine the protein expression of TAB1 and TAB1␤, total cell lysates were prepared from HEK293, HEPG2, MDA231, RPMI8226, and MCF-7 cells. The expression of TAB1 or TAB1␤ was analyzed by Western blotting using anti-TAB1 antibodies. The antibody against a peptide, found in the N terminus of both TAB1 and TAB1␤, detected two major protein bands in Western blotting analysis (Fig. 2B, left panel). The same result was obtained with an antibody that was raised by using recombinant TAB1␤ protein (data not shown). We overexpressed TAB1 and TAB1␤ in 293 cells and detected the ectopically expressed TAB1 and TAB1␤ by Western blotting (Fig. 2B, right panel). Comparison of the migrations of endogenous proteins detected by the anti-TAB1/TAB1␤ antibodies with the ectopically expressed TAB1 and TAB1␤ gives no indication which protein band or bands are TAB1 or TAB1␤, because there was a higher molecular weight band of endogenous protein that could be post-translationally modified TAB1 or TAB1␤ or could even be another TAB1 isoform. We have tried to raise TAB1-and TAB1␤-specific antibodies using C-terminal peptides from TAB1 and TAB1␤. Because the difference between TAB1 and TAB1␤ is very limited (29 and 69 amino acids), there were not many peptide sequences that could be chosen to generate antibodies. As a result, we have not generated any antibodies that can selectively detect TAB1 or TAB1␤. We cannot formally conclude whether TAB1 and/or TAB1␤ were expressed as protein in cells by the Western blotting analysis shown in Fig. 2B, but the protein expression of at least one of TAB1 isoforms was confirmed.
RNA interference (RNAi) is the process of gene silencing whereby double-stranded RNA induces the homology-dependent degradation of its cognate mRNA. We designed RNAi constructs to target TAB1 and TAB1␤ using the method of Brummelkamp et al. (38) and stably transfected these constructs into MDA231 cells. Stable transfection of the RNAi constructs, which target the common region of TAB1 and TAB1␤, and the specific region of TAB1 did not affect the expression of TAB1s in MDA231 cells (data not shown). Transfection of R1 (Fig. 2C), one of the RNAi constructs that were designed to target TAB1␤, resulted in reduction of the expression of the upperband protein detected by anti-TAB1 antibody (Fig. 2D). In contrast, transfection of R1(M), which had a single point mutant in comparison with R1 ( Fig. 2C), did not produce any effect, indicating the specificity of the RNAi effect. The other two RNAi constructs that were designed to target TAB1␤ also had no effect on TAB1␤ expression. These results indicated that the upper band is TAB1␤. Because the apparent molecular weight of endogenous TAB1␤ is higher than ectopically expressed TAB1␤ (Fig. 2, B and D), the endogenous TAB1␤ is most likely being post-translationally modified. The reduction of TAB1␤ protein by R1 is specific, because the protein level of the lower band, which is most likely to be TAB1, and p38␣ were not affected by R1 (Fig. 2D). To further confirm that R1 selectively targeted TAB1␤ mRNA, we examined mRNA expression of TAB1 and TAB1␤ by competitive quantitative RT-PCR using their specific primers. The cells stably transfected with R1, but not with R1(M) or empty vector, had reduced TAB1␤ mRNA (Fig. 2E), confirming that R1 specifically mediated a reduction of TAB1␤ mRNA. Thus, TAB1␤ protein is expressed in cells.
TAB1␤ Interacts with p38␣ but Not with TAK1-Our previous studies have demonstrated that the TAB1 protein interacts with p38␣ (23). To find out if the TAB1␤ interacts with p38␣, expression plasmids encoding TAB1␤, TAB1, and FLAG-p38␣ were co-transfected into 293 cells. The FLAG-p38␣ was precipitated with agarose beads conjugated with anti-FLAG antibody (M2), and the existence of TAB1 in the immunoprecipitation complexes was detected by immunoblotting with anti-TAB1 antibody. The results showed that both TAB1␤ and TAB1 bind with p38␣ (Fig. 3A).
To investigate the specificity of the interaction between TAB1␤ and p38␣, we first tested if TAB1␤ could interact with three other p38 isoforms. As shown in Fig. 3B, a small amount of TAB1␤ was pulled down by p38␥ but not by p38␤ or p38␦. The expression level of p38␥ was higher than any other p38 subfamily members (Fig. 3B, middle panel, and data not shown), but the amount of TAB1␤ pulled down with p38␥ was very little in comparison with p38␣ (Fig. 3, A and B), so the affinity of TAB1␤ and p38␥ should be much smaller than TAB1␤ and p38␣. We next evaluated whether there is interaction between TAB1␤ with the other two MAPKs, JNK1 and ERK2. The results demonstrated that TAB1 bound strongly with p38␣ but not with ERK2 or JNK1 (Fig. 3C).
TAB1 was originally identified as a binding protein of TAK1. It promotes the TAK1 kinase activation through the binding with TAK1. The sequence data shown in Fig. 1A suggest that TAB1␤ does not have a TAK1 binding domain, however, whether TAB1␤ can interact with TAK1 needs to be examined.
We transfected the HA-TAK1, TAB1, and TAB1␤ expression plasmids into 293 cells. Immunoprecipitation was performed with agarose beads conjugated with anti-HA antibody. The immunoprecipitates were immunoblotted with anti-TAB1 antibody. We found that immunoprecipitation of TAK1 pulled down TAB1 but not TAB1␤ (Fig. 3D), indicating that there was no interaction between TAB1␤ and TAK1. These data agree with domain mapping of TAB1 reported by other groups and indicate that TAB1␤ only interacts with p38␣ and not TAK1.
Activation of p38␣ by TAB1␤-It has been shown that interaction of TAB1 with p38␣ resulted in autoactivation of p38␣ (23). We next evaluated the ability of TAB1␤ to activate p38␣. The expression plasmids of TAB1␤, TAB1, and p38␣ were co-transformed into 293 cells in different combinations, and the extent of p38␣ dual phosphorylation was examined by immunoblotting using anti-phospho-p38 antibody. As we reported previously, co-expression of TAB1 with p38␣ in cells led to p38␣ phosphorylation; expression of TAB1␤ also led to the phosphorylation of p38␣ (Fig. 4A). To ensure that the phosphorylation of p38␣ detected by anti-phospho-p38 antibody indeed repre-FIG. 3. TAB1␤ interacts with p38␣ but not TAK1. A, FLAG-p38␣ was coexpressed with TAB1 or TAB1␤ in 293 cells. Cells were lysed 24 h after transfection, and immunoprecipitation was performed with ant-FLAG antibody. The immunoprecipitates and cell lysates were analyzed by immunoblotting with antibody against TAB1 or FLAG as indicated. TAB1␤ was immunoprecipitated with p38␣. B, the interaction of TAB1␤ with p38␤, p38␥, or p38␦ was examined as in A. These three p38 family members had no or very weak interaction with TAB1␤. C, the interaction of TAB1␤ with JNK1 or ERK2 was examined as in A. No interaction was detected. D, TAB1 or TAB1␤ was co-expressed with HA-TAK1. TAK1 was immunoprecipitated with anti-HA antibody. The immunoprecipitates and cell lysates were analyzed by immunoblotting with antibody to TAB1 or HA as indicated. TAB1␤ was not co-immunoprecipitated with TAK1. The experiments were performed three times with comparable results.
FIG. 4. TAB1␤ promotes p38␣ activation. A, FLAG-p38␣ was co-expressed with TAB1␤ or TAB1 in 293 cells. FLAG-p38␣ was immunoprecipitated with anti-FLAG antibody. The immunoprecipitates and cell lysates were analyzed by immunoblotting with anti-phospho-p38, anti-FLAG, or anti-TAB1 as indicated. Co-expression of TAB1␤ or TAB1 increased p38␣ phosphorylation. B, the experiments were performed as in A, and the immunoprecipitates were used as kinase in the in vitro kinase assay using MBP as a substrate. TAB1␤ and TAB1 increased p38␣ kinase activity. C, expression vectors of a luciferase reporter gene under the control of 5xGal4 binding site (5xGAL), GAL4 binding domain fused with ATF2 activation domain (GAL-ATF2), TAB1␤, TAB1, and FLAG-p38␣ were transfected into 293 cells in different combinations as indicated. Luciferase activity was measured 24 h later. TAB1 and TAB1␤ enhanced ATF2-dependent luciferase expression when co-expressed with p38␣. Experiments were performed three times with comparable results. sented p38␣ activity, FLAG-p38␣ protein was immunoprecipitated by using anti-FLAG M2 beads and further analyzed by in vitro kinase assay using MBP as a substrate. The increase of p38␣ dual phosphorylation by TAB1 or TAB1␤ was correlated with the increase of p38␣ activity (Fig. 4B).
It is known that p38␣ activation has an effect on expression of a number of genes. Previous studies showed that activation of p38␣ by MKK6 or MKK3 augments the transcriptional activity of ATF2 in transient transfection assays (19). To determine whether activation of p38␣ by TAB1␤ interaction had a similar effect on ATF2, we used fusion proteins containing the transactivation domain of ATF2 fused with GAL4 DNA binding domain. Whether TAB1␤-mediated p38␣ phosphorylation can transduce signaling downstream was determined by whether TAB1␤-mediated activation of p38␣ could enhance the transcriptional activity of GAL4-ATF2 protein. The GAL4-driven luciferase reporter construct, the expression plasmid of GAL-ATF2 fusion protein, and the expression plasmids of p38␣, TAB1, or TAB1␤ were transfected into 293 cells in different combinations. Luciferase activity was examined 48 h after transfection. Luciferase expression was significantly increased in the cells co-expressing p38␣ and TAB1 or TAB1␤ (Fig. 4C). Thus, TAB1␤-mediated p38␣ activation can transduce signaling downstream.
TAB1 but Not TAB1␤ Activates TAK1-Because TAB1 activates TAK1 (24), we investigated whether the TAB1␤ had effect on TAK1. The expression vectors of TAB1␤ or TAB1 were transfected into 293 cells with HA-TAK1. HA-TAK1 was immunoprecipitated by anti-HA beads, and kinase activity of TAK1 was examined by a coupled in vitro kinase assay using the GST-MKK4 and His-p38␣(⌴) as substrates (Fig. 5A). The increase in kinase activity of TAK1 can be determined by activation of MKK4, which in turn phosphorylates p38␣(M). As reported by others, co-expression of TAB1 increased TAK1 activity. In contrast, expression of TAB1␤ had no effect TAK1 activity (Fig. 5A). These data are consistent with the data in Fig. 3D, showing that TAB1␤ did interact with TAK1.
It was reported that TAB1-mediated TAK1 activation could dramatically induce NF-B reporter gene expression. We were able to reproduce the result of NF-B activation by TAK1 and TAB1 co-expression (Fig. 5B). In contrast to TAB1 co-expression, TAB1␤ co-expression did not have any effect on TAK1mediated NF-B-dependent reporter gene expression (Fig. 5B). These data further support the notion that TAB1␤ does not have any effect on TAK1.
TAB1␤-induced p38␣ Activation Is Dependent on p38␣ Autophosphorylation-We uncovered recently that TAB1-mediated p38␣ activation is through p38␣ autophosphorylation (23). The same approaches were used here to determine whether TAB1␤mediated p38␣ phosphorylation is also autophosphorylation. First, we examined whether TAB1␤-mediated p38␣ phosphorylation can be inhibited by dominant negative MKK6 or TAK1, the known upstream kinases of p38␣. Mutation of regulatory phosphorylation sites in MKK6 to alanine (MKK6(A)) resulted in the loss of kinase activity, and this mutant has been successfully used as a dominant negative mutant in a number of studies (19,31,39). Mutation in the ATP binding site of TAK1 (TAK1(K63W)) leads to kinase death, and this mutant can be used as a dominant negative mutant (24,27). We cotransfected MKK6(A) or TAK1(K63W) with TAB1␤ and p38␣ and compared the level of p38␣ phosphorylation. Expression of MKK6(A) or TAK1(K63W) did not interfere with TAB1␤-induced p38␣ activation (Fig. 6A). Similar results were obtained when MKK3(A) or MKK4(A) was expressed (data not shown). These results agreed with the report that TAB1-mediated p38␣ phosphorylation is independent from the known upstream kinases of p38␣ and indicated that TAB1␤-mediated p38␣ occurs most likely through a similar mechanism. Second, we investigated whether TAB1␤-induced p38␣ activation is also dependent on p38␣ intrinsic enzymatic activity. Two methods were used to diminish the p38␣ intrinsic activity. One was to use kinase-dead p38␣ mutants. A p38␣ mutant with an impaired ATP binding site (Lys-53 to Met mutant termed p38␣(M)) was used in the experiments. The other method was to treat the cells with p38 inhibitor SB203580 (40,41). As shown in Fig. 6B, phosphorylation of the kinase-dead mutant of p38␣ was observed in the cells expressing MKK6(E) but not in the cells expressing TAB1␤; activation of p38␣ by TAB1␤ was inhibited by treatment of cells with SB203580. These data indicate that TAB1␤-mediated p38␣ phosphorylation requires intrinsic kinase activity of p38␣. Third, we examined TAB1␤-mediated p38␣ activation in vitro. p38␣ and TAB1␤ were synthesized in and purified from bacteria as His-tagged and GST fusion proteins, respectively. GST-TAB1␤ was incubated with His-p38␣ in kinase reaction buffer with cold ATP. The activity of p38␣ was examined by using myelin basic protein (MBP) as a substrate. TAB1␤ significantly increased p38␣ activity toward MBP. Moreover, TAB1␤-mediated p38␣ activity was reduced by SB203580 (Fig. 6C). Collectively, our data demonstrate that TAB1␤-mediated p38␣ activation is p38␣ autoactivation.
Extracellular Stimuli Affects TAB1␤ and p38␣ Association-We reported previously that TAB1-dependent activation of p38␣ was used differently by different extracellular stimuli (23). Due to the identification of the similar effects of TAB1␤ on p38␣ activation to those of TAB1, we test whether TAB1␤ could selectively be involved in p38␣ activation induced with different stimuli. The 293 cells were transfected with TAB1␤ and stimulated with different stimuli. As shown in Fig. 7A, overexpression of TAB1␤ significantly enhanced peroxynitrite and TNF-induced p38␣ phosphorylation. The effects of TAB1␤ on anisomycin-and arsonite-induced p38␣ phosphorylation were modest. Overexpression of TAB1␤ did not effect high osmolar sorbital-induced p38␣ phosphorylation. The selective involvement of TAB1␤ in enhancing p38␣ phosphorylation induced by different stimuli fit well with the profiles of kinase cascadeindependent p38␣ activation reported previously (23). Because the association of TAB1␤ with p38␣ can induce p38␣ activation, we next investigated whether the extracellular stimuli selectively affected the association of TAB1␤ with p38␣. 293 cells were transfected with expression plasmids of TAB1 and p38␣ for 24 h, and then were stimulated with several extracellular FIG. 5. TAB1␤ does not activate TAK1. A, TAB1 or TAB1␤ were co-expressed with HA-TAK1 in 293 cells. TAK1 was immunoprecipitated, and in vitro coupled kinase assay was performed using immunoprecipitates as kinase, GST-MKK4 as a substrate for TAK1, and His-p38␣(M) as a substrate for MKK4. Only TAB1 increased TAK1 activity. B, NF-B reporter was co-expressed with TAK1, TAB1, or TAB1␤ in 293 cells as indicated. Luciferase activity was measured 24 h later. TAK1 and TAB1␤ co-expression did not enhance NF-B-dependent gene expression. The experiments were performed two times with comparable results.
stimuli. The association of TAB1␤ with p38␣ was examined by using a co-immunoprecipitation assay. We observed that TNF and peroxynitrite significantly increased the binding of TAB1␤ with p38␣ and anisomycin or arsonite also induced this association but to a lesser extent. Sorbitol failed to induce this association (Fig. 7B). These data correlated with the enhancement of p38␣ phosphorylation by TAB1␤ overexpression in cells treated with different stimuli (Fig. 7, A and B). The time course of TAB1␤-p38␣ interactions induced by TNF is shown in Fig. 7C. These results suggest that enhanced p38␣ and TAB1 interactions are differentially involved in different extracellular stimuli-induced cellular activation.
Reduction of TAB1␤ by RNAi Reduced Basal Activity of p38␣ and Invasiveness of MDA231 Cells-MDA231 is an invasive breast cancer cell line. We showed previously that the invasiveness of MDA231 is dependent on, at least partially, the high basal activity of p38␣ (42). Because the RNAi construct R1 can specifically inhibit TAB1␤ expression in MDA231 cells (Fig.   FIG. 6. TAB1␤-mediated p38␣ dual phosphorylation is a p38␣ autophosphorylation. A, 293 cells were co-transfected with different expression vectors as indicated. 2-and 4-fold plasmid DNA of MKK6(A) and TAK1(K63W) were used in some samples to increase the expression of these dominant negative proteins. Immunoprecipitation and Western blotting were performed as in Figs. 2 and 3. Overexpression of dominant negative MKK6 (MKK6(A)) or TAK1 (TAK1(K63W)) had no effect on TAB1␤-mediated p38␣ dual phosphorylation. B, co-expression of FLAG-p38␣ or FLAG-p38␣(M) with TAB1 or MKK6(E) was performed as in A. SB203580 (5 M) was added into cell culture medium 4 h after transfection in the sample indicated. Immunoprecipitation and Western blotting were performed as in A. TAB1-mediated p38␣ dual phosphorylation requires intrinsic p38␣ activity. C, TAB1-medtated p38␣ phosphorylation in vitro requires intrinsic p38␣ kinase activity. GST-TAB1␤, His-p38␣, and MBP were incubated in kinase reaction buffer containing [ 32 P]ATP in the presence of different concentrations of SB203580. The kinase reaction was stopped by SDS-sample buffer, and phosphorylated proteins were resolved on SDS-PAGE and exposed to x-ray film. SB203580 inhibited both autophosphorylation and activity of p38␣. Data shown are representative of two to three independent experiments.  Figs. 2 and 3. Expression of TAB1␤ enhanced p38␣ phosphorylation induced by some stimuli but not others. B, TAB1␤ and FLAG-p38␣ were expressed in 293 cells, and the cells were treated with different stimuli as indicated. FLAG-p38␣ was immunoprecipitated, and the co-precipitated TAB1␤ was determined. The amount of TAB1␤ co-precipitated varied when the cells were treated with different stimuli. C, TAB1␤ and FLAG-p38␣ were expressed in 293 cells, and the cells were treated with TNF for different periods of time. FLAG-p38␣ was immunoprecipitated, and co-precipitated TAB1␤ was determined. A time-dependent co-precipitation of TAB1␤ was observed in TNF-treated cells. Data shown are representative of two to three independent experiments. 2D), we evaluated the role of TAB1␤ in p38␣ activity and invasion of MDA231 cells. As shown in Fig. 8A, the basal level of p38␣ phosphorylation was significantly reduced in R1 transfected cells in comparison with non-transfected, vector-transfected, and R1(M)-transfected cells. Therefore, TAB1␤ has a role in regulating basal activity of p38␣. Stimulation of MDA231 cells with TNF can lead to a 2-to 3-fold increase in p38␣ phosphorylation in vector-and R1(M)-transfected cells (Fig. 8B). The p38␣ phosphorylation in R1-transfected cells was significantly lower in comparison with controls before and after TNF stimulation (Fig. 8B), and the -fold induction of p38␣ phosphorylation by TNF was also lower. We used Matrigel invasion assays to evaluate whether reduction of TAB1␤ by RNAi had an effect on invasion of MDA231 cells. As shown in Fig. 8C, invasion of MDA231 cells was inhibited ϳ50% when TAB1␤ was knocked down by RNAi.

DISCUSSION
TAB1␤ is a splicing variant of TAB1 based on its gene structure (Fig. 1B). The expression of TAB1␤ was demonstrated by combined results from RT-PCR, Western blotting, and RNAi (Fig. 2). TAB1␤ has a different C-terminal sequence from that of TAB1 and does not have a TAK1 interacting domain that is harbored at the C terminus of TAB1. TAB1␤ retained the p38␣ interacting domain located in the N-terminal side to the TAK1 interacting domain in TAB1. In addition to the interacting domains for p38␣ and TAK1, computer analysis indicated that there is a protein phosphatase 2C (PP2C)-like domain in the N-terminal of TAB1 and TAB1␤. The domain structure of TAB1 and TAB1␤ is shown in Fig. 9. Like TAB1, TAB1␤ interacts with p38␣, and this interaction leads to autoactivation of p38␣. Unlike TAB1, TAB1␤ cannot interact with TAK1 and has no effect on TAK1 activity. Specific inhibition of TAB1␤ expression by RNAi reduced basal activity of p38␣ in MDA231 cells and decreased the invasiveness of MDA231 cells, indicating that TAB1␤ is involved in the regulation of p38␣ activity in a physiological setting.
Our data show that TAB1-and TAB1␤-mediated p38␣ activation is an autophosphorylation of p38␣. However, how an interaction with TAB1 or TAB1␤ leads to p38␣ autoactivation remains to be elucidated. It is well established that activation of MAPKs by their corresponding MAPKKs is through dual phosphorylation on specific threonine and tyrosine residues (4,18), and we have shown in our previous work (23) that p38␣ autoactivation is also achieved by dual phosphorylation on the regulatory sites. MAPKs are serine/threonine kinases, but low level phosphorylation on their regulatory tyrosine residue has been observed in ERK1, ERK2, and p38␣ in vitro (23,43,44). It was demonstrated quite a long time ago that this phosphorylation is mediated by autophosphorylation (43)(44)(45). These early works suggested that autophosphorylation on one of the regulatory sites could occur in MAPKs, although the efficiency was low. It is possible that the binding of TAB1 or TAB1␤ to p38␣ changes the confirmation of p38␣ and facilitates autophosphorylation of p38␣ on its regulatory sites. p38␣ has a similar overall structure to the ERK2 MAPK (46). Analysis of crystal structures of inactive and active forms of the ERK2 has indicated that phosphorylation of ERK2 results in a conformational change to an "energetically unfavorable state" (47). It is possible that autophosphorylation of p38␣ can be enhanced by stabilization of p38␣ in an "energetically unfavorable transient confirmation" by the binding of an "activator" protein like TAB1 or TAB1␤ (48,49). A recent study by Emrick et al. (50) shows that a double mutation of leucine 73 to proline and serine 151 to aspartic acid in ERK2 leads to ERK2 autophosphorylation and the mutant is constitutively active. By analyzing the three-dimensional structure of ERK2, authors of this work suggested that mutations of L73P and S151D permit a regulatory phosphorylation site Y185 to move within hydrogen bonding distance of the catalytic aspartate Asp-147, facilitating the phosphoryl transfer. This work supports the idea that conformational changes can lead to MAPK autophosphorylation and activation. The activation mechanism of p38␣, after binding to TAB1 or TAB1␤, should be similar to the constitutive active ERK2 mutant, because both are activated by autophosphorylation. It is thus very likely that TAB1 and p38␣ interaction leads to conformational changes that are similar to that in ERK2(L73P,S151D) and that such changes relieve structural constraints in the catalytic site that suppress autophosphorylation and autoactivation.
Autophosphorylation is a common mechanism for the activation of a number of different kinases such as receptor tyrosine kinases (51,52) and some of the MAPKKKs (53)(54)(55)(56). TAB1mediated TAK1 activation was shown to be a result of autoactivation (26,29). Because TAK1 and p38␣ bind to different sites of TAB1 and both are activated by autophosphorylation after binding to TAB1, one would assume that TAB1 has a structural feature that triggers autophosphorylation of these two different kinases. Sequence analysis revealed that the N-terminal two-thirds of TAB1 and TAB1␤ protein had low level sequence homology with phosphatase PP2C. However, phosphatase ac- TAB1␤ Interacts with p38␣ but Not TAK1 tivity has not been detected in TAB1 or TAB1␤ so far. Whether this PP2C-like domain has a function in regulating p38␣ and/or TAK1 activity is a subject for future studies. Because the Cterminal 68-amino acid peptide of TAB1 is sufficient to bind and activate TAK1, the N-terminal PP2C-like domain is dispensable for TAK1 activation. To date we were unable to obtain a protein that only contains the p38␣ interacting domain, so we were unable to determine whether this domain alone can lead to p38␣ activation. A common docking domain has been identified in MAPKs as a common site for binding to upstream MKK, downstream substrates, and dual phosphatase (57,58). A gain-of-function mutant of a Drosophila melanogaster MAPK was found to be resulting from disrupted interactions between MAPKs and phosphatase by a mutation in a common docking domain (D334N) (59). It is possible that TAB1 prevents p38␣ and phosphatase interactions to enhance p38␣ activity. However, this possibility was excluded, because purified TAB1 can cause p38␣ autophosphorylation in the absence of any phosphatase in vitro (23). Furthermore, TAB1 did not compete with PP2C-mediated inactivation of p38␣ in co-expression experiments (data not shown). Genetic screening has isolated a gainof-function mutant of p38␣ (49). This mutant (F327L or S) can gain kinase activity in the absence of tyrosine phosphorylation, suggesting the driven force for converting p38␣ to active conformation in this mutant is different from that of TAB1-mediated p38␣ activation.
The p38␣ MAPK is activated in response to a variety of extracellular stimuli, including pro-inflammatory cytokines and environmental stresses (4,60). How these different stimuli via different receptors or other molecular sensors activate p38␣ is still not fully understood. Previous work by a number of investigators, including us, showed that there are at least two different mechanisms immediately upstream of p38␣ in regulating p38␣ activity (19,23). One is dependent on upstream kinase MKK3 or MKK6; another is dependent on TAB1-mediated p38␣ autophosphorylation. The data presented in this report added TAB1␤ into the regulation network of p38␣. To date we were unable to generate isoform-specific antibodies for TAB1␤ and TAB1, however, in combination with RNAi, we were able to identify the protein band of TAB1␤ resolved on SDS-PAGE (Fig. 2D). The role of TAB1␤ in controlling the basal activity of p38␣ has been documented by our experiments (Fig. 8). Whether TAB1 and TAB1␤ have different roles in mediating p38␣ activation in cells has not been determined. Nevertheless, TAB1 and TAB1␤ are clearly different in their functions, because TAB1 is capable of regulating both TAK1 and p38␣ whereas TAB1␤ only activates p38␣. The different functions of TAB1 splicing variants may be a mechanism required for precise regulation of intracellular signaling that controls p38␣ activation.