Multiple Activation Mechanisms of p38α Mitogen-activated Protein Kinase*

The p38α MAPK participates in a variety of biological processes. Activation of p38α is mediated by phosphorylation on specific regulatory tyrosine and threonine sites, and the three dual kinases, MAPK kinase 3 (MKK3), MKK4, and MKK6, are known to be the upstream activators of p38α. In addition to activation by upstream kinases, p38α can autoactivate when interacting with transforming growth factor-β-activated protein kinase 1-binding protein 1 (TAB1). Here we used MKK3 and MKK6 double knock-out (MKK3/6 DKO) and MKK4/7 DKO mouse embryonic fibroblast (MEF) cells to examine activation mechanisms of p38α. We confirmed that the MKK3/6 pathway is a primary mechanism for p38α phosphorylation in MEF cells, and we also showed the presence of other p38α activation pathways. We show that TAB1-mediated p38α phosphorylation in MEF cells did not need MKK3/4/6, and it accounted for a small portion of the total p38α phosphorylation that was induced by hyperosmolarity and anisomycin. We observed that a portion of peroxynitrite-induced phospho-p38α is associated with an ∼85-kDa disulfide complex in wild-type MEF cells. Peroxynitrite-induced phosphorylation of p38α in the ∼85-kDa complex is independent from MKK3/6 because only phospho-p38α not associated with the disulfide complex was diminished in MKK3/6 DKO cells. In addition, our data suggest interference among different pathways because TAB1 had an inhibitory effect on p38α phosphorylation in the peroxynitrite-induced ∼85-kDa complex. Mutagenesis analysis of the cysteines in p38α revealed that no disulfide bond forms between p38α and other proteins in the ∼85-kDa complex, suggesting it is a p38α binding partner(s) that forms disulfide bonds, which enable it to bind to p38α. Therefore, multiple mechanisms of p38α activation exist that can influence each other, be simultaneously activated by a given stimulus, and/or be selectively used by different stimuli in a cell type-specific manner.

The p38␣ MAPK participates in a variety of biological processes. Activation of p38␣ is mediated by phosphorylation on specific regulatory tyrosine and threonine sites, and the three dual kinases, MAPK kinase 3 (MKK3), MKK4, and MKK6, are known to be the upstream activators of p38␣. In addition to activation by upstream kinases, p38␣ can autoactivate when interacting with transforming growth factor-␤-activated protein kinase 1-binding protein 1 (TAB1). Here we used MKK3 and MKK6 double knock-out (MKK3/6 DKO) and MKK4/7 DKO mouse embryonic fibroblast (MEF) cells to examine activation mechanisms of p38␣. We confirmed that the MKK3/6 pathway is a primary mechanism for p38␣ phosphorylation in MEF cells, and we also showed the presence of other p38␣ activation pathways. We show that TAB1-mediated p38␣ phosphorylation in MEF cells did not need MKK3/4/6, and it accounted for a small portion of the total p38␣ phosphorylation that was induced by hyperosmolarity and anisomycin. We observed that a portion of peroxynitrite-induced phospho-p38␣ is associated with an ϳ85-kDa disulfide complex in wild-type MEF cells. Peroxynitrite-induced phosphorylation of p38␣ in the ϳ85-kDa complex is independent from MKK3/6 because only phospho-p38␣ not associated with the disulfide complex was diminished in MKK3/6 DKO cells. In addition, our data suggest interference among different pathways because TAB1 had an inhibitory effect on p38␣ phosphorylation in the peroxynitrite-induced ϳ85-kDa complex. Mutagenesis analysis of the cysteines in p38␣ revealed that no disulfide bond forms between p38␣ and other proteins in the ϳ85-kDa complex, suggesting it is a p38␣ binding partner(s) that forms disulfide bonds, which enable it to bind to p38␣. Therefore, multiple mechanisms of p38␣ activation exist that can influence each other, be simultaneously activated by a given stimulus, and/or be selectively used by different stimuli in a cell type-specific manner.
Cellular responses to extracellular stimuli are mediated through intracellular signaling pathways such as the MAPK 2 pathways. MAPKs are members of discrete signaling cascades and serve as focal points in response to a variety of extracellular stimuli (1)(2)(3)(4). Several distinct groups of MAPKs have been characterized in mammals, and each group is composed of a set of three evolutionarily conserved, sequentially acting kinases: a MAPK, a MAPK kinase (MKK), and an MKK kinase (MAP3K) (5,6). The MAP3Ks are serine/threonine kinases and are often activated through phosphorylation. MAP3K activation leads to the phosphorylation and activation of an MKK, which then stimulates MAPK activity through dual phosphorylation on threonine and tyrosine residues located in the activation loop of kinase subdomain VIII. Once activated, MAPKs phosphorylate target substrates on serine or threonine residues. It is believed that the substrate selectivity of MAP3K, MKK, and MAPK is conferred by specific interaction motifs located on the kinases and substrates (7,8). In addition, MAPK cascade specificity is also mediated by scaffolding proteins that organize pathways in specific modules through simultaneous binding of several components (9). Although the kinase cascade is the primary activation mechanism of MAPKs, autoactivation of the MAPK p38␣ is promoted by its interaction with transforming growth factor-␤-activated protein kinase 1 (TAK1)-binding protein 1 (TAB1) or by phosphorylation on Tyr-323 with the tyrosine kinase Zap70 (10 -12). To date, however, there is little information regarding whether the multiple mechanisms are simultaneously or individually used by a given stimulus or regarding the contribution of the different pathways.
The p38 MAPK group is composed of p38␣, p38␤, p38␥, and p38␦ (13)(14)(15). Among the four members, p38␣ is the most studied, and it has been implicated in cellular processes such as proliferation, differentiation, apoptosis, senescence, and inflammation (13)(14)(15). p38␣ activation has been observed in response to a variety of extracellular stimuli, including proinflammatory cytokines, bacterial components, and stress (16,17). MKK3 and MKK6 are the two main MKKs that are known to activate p38␣ (18,19). MKK4, an upstream kinase of JNK, can aid in the activation of p38␣ in cells exposed to various stimuli (18,20). In addition, p38␣ can also be activated by interaction with TAB1 (10). TAB1 was originally identified as an interacting protein of TAK1 (21). The interaction between TAB1 and p38␣ was later independently found by two research groups (10,22). A splicing variant of TAB1, TAB1␤, has also been found to interact only with p38␣ and not TAK1 (11). In vitro and co-expression experiments have indicated that the interaction of TAB1 and p38␣ leads to p38␣ autophosphorylation on the dual phosphorylation sites in the activation loop (10,11). TAB1-dependent p38␣ activation appears to play a role in injury response during myocardial ischemia (23), monocyte-derived dendritic cell maturation (24), and peripheral T-cell anergy maintenance (25). On one hand, phosphorylation of TAB1 by p38␣ has been observed, and negative feedback on TAK1 activation by phosphorylated TAB1 was proposed as a major function of TAB1-p38␣ interaction (22). On the other hand, a study using MEF cells generated from MKK3/6 DKO mice ruled out the role of TAB1 in tumor necrosis factor-␣ (TNF)-and UV radiationinduced p38␣ phosphorylation in MEF cells (20). Therefore, differing opinions exist as to whether MKK-independent p38␣ phosphorylation occurs.
To better understand the relevance of MKK-independent p38␣ activation, we examined p38␣ activation in MKK3/6 DKO and MKK4/7 DKO MEF cells. The MKK independence of TAB1-mediated p38␣ phosphorylation is supported by our data from MKK3/6 and MKK4/7 DKO MEF cells. We further show that MKK-independent p38␣ phosphorylation accounts for a small portion of the total p38␣ phosphorylation induced by some, but not all, of the different ligands in MEF cells. In addition to the role in promoting p38␣ phosphorylation, we found that TAB1 has an inhibitory effect on peroxynitrite-induced p38␣ phosphorylation that is associated with a disulfide complex. This p38␣ phosphorylation is likely to be independent from MKK3/6 because MKK3/6 double knock-out mostly eliminated the phosphorylation of p38␣ that is not associated with the disulfide complex. Our data indicate that, although the kinase cascade is a major mechanism in p38␣ activation, multiple mechanisms exist and can be simultaneously activated by a single stimulus, and they may coordinately regulate p38␣ phosphorylation.

MATERIALS AND METHODS
Reagents-Lipopolysaccharide (Escherichia coli 0111:B4) was purchased from List Biological Laboratories (Campbell, CA), and CpG DNA was purchased from Invivogen (San Diego, CA). Peroxynitrite, thapsigargin, and SB203580 were purchased from Calbiochem. Anisomycin, sorbitol, and anti-FLAG M2 antibodies were purchased from Sigma. Mouse TNF was purchased from R&D Systems (Minneapolis, MN). Antibodies against phospho-p38, phospho-JNK, and phospho-ERK were purchased from Cell Signaling Technology (Beverly, MA), and polyclonal antibodies against p38␣, TAB1, and MKK4 were generated from New Zealand White rabbits by stepwise subcutaneous and muscle injections of each of the affinity-purified recombinant proteins.
Vectors and RNA Interference-FLAG-tagged p38␣, TAB1, or TAB1␤ was cloned into the vector pcDNA3. The point mutations in p38␣ were generated with a QuikChange sitedirected mutagenesis kit (Stratagene) and confirmed by DNA sequencing. Oligonucleotides were cloned into the vector pSuper to express short hairpin RNA. The following sequences were targeted: 5Ј-AGCAGTCCTTCTCAACAGCAAG-3Ј for siTAB#1 and 5Ј-AGGCCCTTCTGTGCAAATCTAC-3Ј for siTAB#2. Double-stranded RNA targeting the sequence 5Ј-AATGCGGAGTAGTGATTGCCCAT-3Ј of MKK4 was purchased from Dharmacon Inc. (Chicago, IL), and control double-stranded RNA targeting the sequence 5Ј-ATGATT-GATTTGTCAATGGTCCACC-3Ј of green fluorescent protein was obtained from Invitrogen.
Cell Culture and Transfection-Wild-type, MKK3/6 DKO, or MKK4/7 DKO MEF cells were maintained in high glucose Dulbecco's modified Eagle's medium (Invitrogen) plus 10% fetal bovine serum at 37°C in a humidified 5% CO 2 atmosphere. The transfection of expression vectors was done using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. Briefly MEF cells were freshly seeded to 70% confluence 1 day before transfection. For the selection of TAB1 knockdown MEF cells, cells were incubated in a medium containing 100 g/ml hygromycin and were further cultured for 2 weeks and then pooled. MEF cells were transfected with siRNA oligonucleotides using Lipofectamine 2000, and then they were incubated for 48 h and subjected to Western blot analysis.
Western Blot Analysis-Cells were washed twice with icecold phosphate-buffered saline and were lysed in a lysis buffer containing 50 mM Tris-Cl (pH 7.5), 0.15 M NaCl, 1% Nonidet P-40, 0.5% deoxycholate, 1 mM EDTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 2 g/ml aprotinin, and 5 g/ml pepstatin. After centrifugation at 12,500 rpm for 5 min, the supernatant was collected, and the protein concentration was determined by the Bradford method. Nonreducing cell lysate samples were prepared in lysis buffer without dithiothreitol and were mixed with the SDS-PAGE sample buffer without ␤-mercaptoethanol. Cell lysates were separated by SDS-PAGE and transferred onto a polyvinylidene difluoride membrane (Amersham Biosciences) and were then immunoblotted with antibodies against FLAG, phospho-p38, phospho-JNK, phospho-ERK, p38␣, TAB1, and MKK4.
In Vitro Kinase Assay-In vitro kinase assays were performed using recombinant His-tagged p38␣ as a kinase and myelin basic protein as a substrate as described previously (16).

TAB1-or TAB1␤-mediated Phosphorylation of p38␣ Occurs in Both
Wild-type and MKK3/6 DKO or MKK4/7 DKO MEF Cells-TAB1-mediated p38␣ phosphorylation in cells can be seen by co-expression of TAB1 and p38␣ (10). The MKK independence of TAB1-mediated p38␣ phosphorylation has been established by the observation that TAB1-mediated p38␣ phosphorylation is not inhibited by dominant negative MKKs, and it is dependent on the intrinsic activity of p38␣ (10). To determine whether TAB1-mediated p38␣ phosphorylation in cells is unequivocally independent of MKK3 and MKK6, we co-expressed p38␣ with TAB1 or TAB1␤ in MKK3/6 DKO and control wild-type MEF cells. The phosphorylation of p38␣ was measured by Western blotting with anti-phospho-p38 antibodies. As shown in Fig. 1A, TAB1 and TAB1␤ enhanced phosphorylation of p38␣ in both wildtype and MKK3/6 DKO cells. The phosphorylation level of p38␣ is comparable in wild-type and MKK3/6 DKO cells. To further demonstrate the dependence of TAB1 on p38␣ phosphorylation in a co-expression assay, we co-transfected p38␣ expression vector with increased amounts of TAB1␤ expression vector. We found that the higher the TAB1 dose the higher the level of p38␣ phosphorylation in MKK3/6 DKO cells (Fig. 1B). Therefore, both MKK3 and MKK6 are not involved in TAB1-mediated p38␣ phosphorylation.
Although TAB1-mediated p38␣ phosphorylation was not affected by MKK3/6 or MKK4 knock-out ( Fig. 1, A, B, and C), it is still possible that MKK3/6 and MKK4 compensate for each other in these knock-out cells, and thus TAB1-mediated p38␣ phosphorylation still requires MKK. To evaluate this possibility, we used siRNA to knock down MKK4 in MKK3/6 DKO cells. Because knockdown of MKK4 in MEF cells by siRNA has been successfully done before (20), the same experimental pro-tocol was used here. As shown in Fig. 1D, transfection of siRNA targeting MKK4 (siMKK4) gene expression reduced MKK4 protein levels in the MKK3/6 DKO cells. Reduction of MKK4 in MKK3/6 DKO cells did not affect TAB1-mediated p38␣ phosphorylation as determined by co-expression of TAB1␤ and p38␣. Therefore, TAB1-mediated p38␣ phosphorylation is independent from MKK3, MKK6, and MKK4.

MKK3/6/4-independent p38␣ Phosphorylation Contributes a Small Portion of the Extracellular Stimuli-induced p38␣
Phosphorylation in MEF Cells-Because TAB1-mediated p38␣ phosphorylation requires the intrinsic activity of p38␣, inhibition of p38␣ phosphorylation by the p38 inhibitor SB203580 has been used to dissect MKK-dependent and MKK-independent p38␣ phosphorylation (10,(23)(24)(25). However, there are drawbacks in the drug-based inhibition approach that include a lack of absolute target specificity. To conclusively determine whether MKK-independent p38␣ phosphorylation is involved in extracellular stimuli-induced p38␣ activation, we examined p38␣ phosphorylation in wild-type and MKK3/6 DKO cells in response to different stimuli. As shown in Fig. 2, wild-type MEF cells responded to bacterial lipopolysaccharides and CpG oligonucleotides, to the proinflammatory cytokine TNF, to the endoplasmic reticulum Ca 2ϩ -ATPase inhibitor thapsigargin, to the protein synthesis inhibitor anisomycin, to the nitrogen monoxide (NO) producer peroxynitrite, and to hyperosmolarity (0.5 M sorbitol). The level of response in p38␣ phosphorylation varied considerably among the different stimuli. Lipopolysaccharide-and CpG-induced p38␣ phosphorylation was  Wild-type and MKK3/6 DKO MEF cells were treated with lipopolysaccharide (LPS) (1 g/ml), CpG DNA (5 g/ml), TNF (100 ng/ml), thapsigargin (10 nM), anisomycin (50 ng/ml), peroxynitrite (500 M), or sorbitol (0.5 M) for the indicated time periods. Cell lysates were prepared and analyzed by immunoblotting with anti-phospho-p38 (*p-p38␣) and anti-p38␣ antibodies. The relative phosphorylation levels shown on the right were obtained by measuring the pixels of each band of phospho-p38␣ using AlphaEase software and then normalizing that with the total p38␣. Comparable results were obtained in two to three experiments.
Because MKK4 participates in the activation of p38␣ under certain conditions, we sought to determine whether MKK4 is involved in anisomycin-, peroxynitrite-, and hyperosmolarity (sorbitol)-induced p38␣ phosphorylation using MKK4/7 DKO MEF cells. As shown in Fig. 3, MKK4 deletion had no detectable effect on anisomycin-, peroxynitrite-, and hyperosmolarity (sorbitol)-induced p38␣ phosphorylation in MEF cells, indicating that MKK4 is not involved in or only has a little contribution to these stimuli-induced p38␣ phosphorylation events.
TAB1 Can Both Positively and Negatively Regulate p38␣ Phosphorylation in a Stimulus-dependent Manner-We next examined whether the MKK-independent p38␣ phosphorylation shown in Fig. 2 is TAB1-dependent. Because TAB1 knockout cells were not available to us, we used siRNA to knock down TAB1 expression in MEF cells. After testing a number of TAB1targeting siRNAs, we found two (named siTAB#1 and siTAB#2) that were able to effectively knock down TAB1 expression in MEF cells. DNA sequence analysis (Blastn) indicated that siTAB#1 and siTAB#2 only target TAB1. We stably expressed siTAB#1 or siTAB#2 in MKK3/6 DKO cells and measured TAB1 expression by Western blot analysis. A significant reduction in TAB1 was seen in both siTAB#1-and siTAB#2-expressing MKK3/6 DKO cells (Fig. 4A). We treated control (vector-transfected) and either siTAB#1-or siTAB#2expressing MKK3/6 DKO cells with anisomycin, peroxynitrite, or sorbitol and then measured p38␣ phosphorylation by Western blot analysis. More cell lysates and longer exposure times were needed in the Western blot analysis due to the low levels of p38␣ phosphorylation in MKK3/6 DKO cells. As shown in Fig.  4B, anisomycin-and hyperosmolarity (sorbitol)-induced p38␣ phosphorylation was reduced in cells expressing TAB1-targeting siRNA, suggesting that anisomycin-and hyperosmolarity (sorbitol)-induced MKK-independent p38␣ phosphorylation in MEF cells is TAB1-dependent.
To our surprise, peroxynitrite-induced p38␣ phosphorylation was not reduced by siTAB#1 and siTAB#2 siRNA, but on the contrary, it was significantly enhanced in both cell lines (Fig.  4B). The same results were obtained when pools of independently transfected cells were used in the experiments, indicating that this is unlikely an observation resulting from variation among individual cell clones or different batches of stably transfected cell lines. Therefore, we concluded that TAB1 has an inhibitory effect on peroxynitrite-induced p38␣ phosphorylation in MKK3/6 DKO cells.
Previous studies by us and by others have suggested that peroxynitrite-induced p38␣ phosphorylation in human embryonic kidney 293 cells and heart cells was mediated in part by TAB1 because SB203580 significantly inhibits peroxynitrite-induced p38␣ phosphorylation (10,23). Unlike peroxynitrite, sorbitolinduced p38␣ phosphorylation in human embryonic kidney 293 cells and heart cells cannot be inhibited by SB203580, and anisomycin-induced p38␣ phosphorylation is only weekly  inhibited by SB203580. Like the data obtained in human embryonic kidney 293 cells and heart cells, anisomycin-induced MKK-independent p38␣ phosphorylation in MEF cells was found to be dependent on TAB1 (Fig. 4B). However, although peroxynitrite-induced p38␣ phosphorylation was not dependent on TAB1 in MKK3/6 DKO MEF cells, sorbitol-induced p38␣ phosphorylation was (Fig. 4B). To determine whether these differences are due to the different types of cells or the experimental methods used, we used SB203580 to inhibit autophosphorylation-dependent p38␣ phosphorylation. As shown in Fig. 5, SB203580 partially inhibited anisomycin-and sorbitol-induced p38␣ phosphorylation, but it did not inhibit peroxynitrite-induced p38␣ phosphorylation in either wildtype (Fig. 5A) or MKK3/6 DKO MEF cells (Fig. 5B). Therefore, the results obtained through siRNA knockdown of TAB1 and through the inhibition of p38␣ autophosphorylation are consistent in MEF cells. Based on published data (10,23) and the data shown in Fig. 5, it can be concluded that it is the cell type that determines whether a TAB1-dependent or -inde-pendent mechanism is used by peroxynitrite to induce p38␣ phosphorylation.

TAB1 Knockdown in MEF Cells Non-selectively Enhances the Phosphorylation of MAP Kinases in a Ligand-dependent
Manner-Next we sought to determine whether the TAB1 knockdown-mediated enhancement of p38␣ phosphorylation is specific for p38␣. MKK3/6 DKO MEF cells expressing siTAB#1 or siTAB#2 were stimulated with peroxynitrite or sorbitol, and the phosphorylation levels of p38␣ or JNK1/2 were determined by Western blot analysis using anti-phospho-p38 and anti-phospho-JNK antibodies, respectively. Like the data shown in Fig. 4B, peroxynitrite-induced p38␣ phosphorylation was found to be much stronger in cells expressing TAB1-targeting siRNA (Fig. 6A). It is important to note that peroxynitrite-induced JNK phosphorylation was also enhanced by TAB1 knockdown (Fig. 6A). In contrast, sorbitol-induced JNK phosphorylation was not affected by TAB1 knockdown. As previously seen in Fig. 4B, sorbitol-induced p38␣ phosphorylation was inhibited by TAB1 knockdown.
We next examined whether TAB1 knockdown affected p38␣ phosphorylation in wild-type MEF cells. Consistent with the data obtained with MKK3/6 DKO MEF cells, TAB1 knockdown enhanced peroxynitrite-induced p38␣ phosphorylation (Fig.  6B). Peroxynitrite-induced ERK phosphorylation was also enhanced by TAB1 knockdown. In a manner consistent with the data shown in Fig. 5A, sorbitol-induced p38␣ phosphorylation was reduced in TAB1 knockdown cells, and sorbitol-induced ERK phosphorylation was not affected by TAB1 knockdown. Therefore, TAB1 inhibits p38␣ phosphorylation when cells are stimulated with peroxynitrite but not other stimuli such as hyperosmolarity (sorbitol). Furthermore the enhancement of peroxynitrite-induced phosphorylation by TAB1 knockdown is not restricted to p38␣.
A recent report showed that reactive oxygen species sustain JNK activation by inactivating MAP kinase phosphatases (MKPs) (26). The inactivation of MKP-1 by reactive oxygen species is achieved by converting the catalytic cysteine of MKP to sulfenic acid. The increase in oxidized MKP-1 results in a reduction of non-oxidized MKP-1, which can be detected by performing a Western blot using non-reducing and reducing gels. Because peroxynitrite is an NO producer, we examined the peroxynitrite-induced oxidization of MKP-1 in MEF cells and were unable to detect oxidized MKP-1 (data not shown). Also we did not detect a reduction in MKP-1 protein in either sorbitol-or peroxynitrite-treated samples when we reprobed the Western blot membranes shown in Fig. 6A (reprobed results are shown in Fig. 6C). Therefore, modification of MKP-1 is unlikely to be the mechanism used by TAB1 to enhance peroxynitrite-induced p38␣ phosphorylation.
The Formation of an ϳ85-kDa Disulfide Complex Is Involved in Peroxynitrite-induced p38␣ Phosphorylation in MEF Cells-In Schizosaccharomyces pombe, the peroxideinduced activation of the p38 homolog Sty1 requires the formation of a peroxide-induced disulfide complex between 2-Cys peroxiredoxin (Tpx1) and Sty1 (27). We analyzed whether peroxynitrite induces any disulfide complex that is associated with p38␣ by Western blot analysis using reducing and non-reducing gels. We found that when peroxyni-FIGURE 5. Inhibition of stimuli-induced p38␣ autophosphorylation by the p38␣ inhibitor SB203580 in MEF cells. A, wild-type MEF cells were preincubated with SB203580 (2.5 M) for 1 h and were then treated with either nothing (Ϫ), anisomycin, sorbitol, or peroxynitrite for 30 min. Cell lysates were prepared, and p38 phosphorylation was analyzed by immunoblotting with anti-phopho-p38 (*p-p38␣) antibodies. Equal loading was determined by reprobing the blot with anti-p38␣ antibodies. The relative phosphorylation levels were obtained by measuring the pixels of each band of phospho-p38␣ using AlphaEase software and then normalizing that with the total p38␣. B, the same as A except MKK3/6 DKO MEF cells were used.
trite-treated MEF cells were analyzed on a non-denaturing gel, a small portion of phospho-p38␣ was associated with a complex of ϳ85 kDa (Fig. 7A). This phospho-p38␣ complex is specific to peroxynitrite treatment because it was not detected when the cells were treated with TNF or sorbitol (data not shown). The association of p38␣ with this complex requires the formation of a disulfide bond(s) because the high molecular weight phospho-p38␣ was not detected when Western blotting was done using a reducing gel (Fig.  7B). However, the disulfide bond could be between p38␣ molecules, between p38␣ and its binding partner, between p38␣ binding partners, or even within a molecule in that complex. The membranes shown in Fig. 7, A and B, were stripped and then reprobed with anti-p38␣ antibodies to show equal sample loading (Fig. 7, C and D). It should be noted that only a very small amount of p38␣ is associated with the peroxynitrite-induced disulfide complex (Fig. 7A) because anti-p38␣ antibodies only detected p38␣ with an apparent molecular mass of ϳ40 kDa (Fig. 7C). The fact that we could not detect p38␣ at ϳ85 kDa with anti-p38␣ antibodies in peroxynitrite-treated cells is probably because our anti-p38␣ antibodies were not sensitive enough in comparison with anti-phospho-p38 antibodies. We noticed that the levels of p38␣ phosphorylation detected by non-reducing and reducing gels were not quite the same. We know that Western blotting with a non-reducing gel is less sensitive and often has a higher background than when using a reducing gel, but we do not know the reason why at this moment.
Peroxynitrite-induced p38␣ Phosphorylation in MKK3/6 DKO MEF Cells Is Associated with the Disulfide Complex-To address the role of MKKs in p38␣ phosphorylation in the ϳ85-kDa complex, we examined peroxynitrite-induced p38␣ phosphorylation in MKK3/6 DKO cells using non-reducing and reducing gels. We found that peroxynitrite-induced p38␣ phosphorylation in MKK3/6 DKO cells is mainly associated with the disulfide complex (Fig. 8A, top panel). The role of disulfide was confirmed by analysis using a reducing gel (Fig. 8A, middle  panel). Equal sample amount was confirmed by reprobing the membrane with anti-p38␣ antibodies (Fig. 8A, bottom panel). Because the ϳ40-kDa phospho-p38␣ can be seen in the nonreducing gel analysis of wild-type cells (Fig. 7A) but not MKK3/6 DKO MEF cells (Fig. 8A), MKK3 and/or MKK6 must be responsible for the phosphorylation of p38␣ that is not associated with the disulfide complex in peroxynitrite-treated MEF cells.
To examine the role of MKK4, we used MKK4/7 DKO cells. As shown in Fig. 8B, top panel, phospho-p38␣ with both high FIGURE 6. TAB1 knockdown non-selectively enhances peroxynitrite-induced MAP kinase phosphorylation in MEF cells. A, control-, siTAB#1-, or siTAB#2-expressing MKK3/6 DKO MEF cells were treated with or without sorbitol or peroxynitrite for 30 min. Cell lysates were prepared and analyzed by immunoblotting using anti-phospho-p38 (*p-p38␣), anti-phospho-JNK (*p-JNK), and anti-p38␣ antibodies. B, control-, siTAB#1-, or siTAB#2-expressing wild-type MEF cells were treated with or without sorbitol or peroxynitrite for 30 min. Cell lysates were prepared and analyzed by immunoblotting using anti-phospho-p38, anti-phospho-ERK (*p-ERK ), and anti-p38␣ antibodies. C, the samples described in A were analyzed by immunoblotting with anti-MKP-1 antibodies. and low molecular weights was detected in peroxynitritetreated MKK4/7 DKO cells. The roles of disulfide and equal sample preparation were confirmed by analysis using a reducing gel (Fig. 8B, middle and bottom panels). This data confirmed that MKK3 or MKK6 is critical for the phosphorylation of p38␣ not associated with the disulfide complex.
TAB1 Knockdown-enhanced p38␣ Phosphorylation Is Associated with the Disulfide Complex and Requires Intrinsic p38␣ Activity-Because TAB1 knockdown did not inhibit but rather enhanced peroxynitrite-induced p38␣ phosphorylation (Fig. 4), we examined whether it enhances p38␣ phosphorylation in the ϳ85-kDa complex. Because peroxynitrite-induced p38␣ phosphorylation is associated with the disulfide complex in MKK3/6 DKO cells, we compared the levels of phosphorylation in MKK3/6 DKO cells expressing and not expressing siRNA against TAB1. TAB1 knockdown enhanced peroxynitrite-induced p38␣ phosphorylation (Fig. 9A). As shown earlier, phospho-p38␣ was detected as being ϳ85 and ϳ40 kDa in nonreducing and reducing gels, respectively (Fig. 9A), indicating an association between phospho-p38␣ and the disulfide complex.
The phospho-p38␣ in the ϳ85-kDa complex was detected in both MKK3/6 DKO and MKK4/7 DKO cells (Fig. 8, A and B), suggesting that MKK3, MKK6, and MKK4 are not necessary for the p38␣ phosphorylation occurring in the ϳ85-kDa complex. However, this cannot exclude the possibility that all three contribute to p38␣ phosphorylation in the ϳ85-kDa complex and that neither a knock-out of MKK3/6 nor a knock-out of MKK4 is sufficient to block this pathway. To address this possibility, we used siRNA to knock down MKK4 expression in MKK3/6 DKO cells. Because peroxynitrite-induced p38␣ phosphorylation is stronger in MKK3/6 DKO cells expressing siTAB#1 or siTAB#2 than in those expressing neither, we used MKK3/6 DKO cells that stably expressed siTAB#1 or siTAB1#2 for this experiment. As shown in Fig. 9B, knockdown of MKK4 in MKK3/6 DKO cells did not have a significant effect on per-oxynitrite-induced p38␣ phosphorylation. This suggests that the phosphorylation of p38␣ in the ϳ85-kDa complex is MKKindependent. However, a definite conclusion cannot be reached because knockdown of MKK4 cannot be completed. Therefore, we examined whether the phosphorylation of p38␣ in the ϳ85-kDa complex depends on the intrinsic activity of p38␣. In MEF cells we transiently expressed either FLAGtagged p38␣(WT) or the FLAG-tagged catalytically inactive p38␣ mutant p38␣(KM), and then stimulated the cells with peroxynitrite. Phospho-p38␣ and FLAG-p38␣ were analyzed using non-reducing gels (Fig. 9C). The association of phospho-FLAG-p38␣ with a disulfide complex was found in cells Cell lysates were prepared, resolved using non-reducing or reducing SDS-PAGE, and then analyzed by immunoblotting using antiphospho-p38 (*p-p38␣) and anti-p38␣antibodies. * indicates a nonspecific band. FIGURE 9. Peroxynitrite-induced p38␣ phosphorylation in the disulfide complex is enhanced by TAB1 knockdown and requires intrinsic p38␣ kinase activity. A, MKK3/6 DKO cells stably transfected with control (C ), siTAB#1, or siTAB#2 expression vectors were treated with or without peroxynitrite for 30 min. Cells lysates were prepared, resolved using nonreducing or reducing SDS-PAGE, and then analyzed by immunoblotting with anti-phospho-p38 (*p-p38␣) and anti-p38␣antibodies. B, siTAB#1-and siTAB#2-expressing MKK3/6 DKO MEF cells were transiently transfected with control siRNA (Control) or siRNA targeting MKK4 (siMKK4). 48 h after transfection, cells were stimulated with nothing (Ϫ) or peroxynitrite for 30 min and then lysed. Cell lysates were analyzed by immunoblotting with anti-phospho-p38, anti-MKK4, and anti-p38␣ antibodies. C, MEF cells were transfected with either the empty vector (Ϫ), FLAG-p38␣(WT) expression vector, or FLAG-p38␣(KM) expression vector. 24 h after transfection, the cells were treated with or without peroxynitrite for 30 min. Cell lysates were prepared, resolved using non-reducing or reducing SDS-PAGE, and then analyzed by immunoblotting with anti-phospho-p38 and anti-FLAG antibodies. *p-flag-p38␣, phospho-FLAG p38␣. D, in vitro kinase assays were performed using recombinant murine p38␣ (0.5 g) as a kinase and myelin basic protein (MBP) (20 g) as a substrate at 30°C for 45 min, with nothing (None), 1 mM GSH, 0.5 mM GSSG plus 0.5 mM GSH, or 1 mM GSSG included in the reaction mixture. *p-MBP, phospho-MBP. expressing FLAG-p38␣(WT) but not FLAG-p38␣(⌲⌴). This suggests that p38␣ phosphorylation in the ϳ85-kDa complex requires intrinsic p38␣ activity (e.g. autophosphorylation). We noticed that ectopic overexpression of FLAG-p38␣ (either WT or KM) reduced the phosphorylation of endogenous p38␣ in the ϳ85-kDa complex; this could have resulted from competition.
The data in Fig. 5 show that the p38 inhibitor SB203580 did not inhibit peroxynitrite-induced p38␣ phosphorylation, whereas the data in Fig. 9C suggest that this p38␣ phosphorylation requires intrinsic p38␣ activity. To address this confusion, we examined whether SB203580 would function differently under reduced and oxidized environments because the disulfide complex should be in an oxidized environment. An in vitro kinase assay with recombinant p38␣ and myelin basic protein (as a substrate) was used to determine the inhibitory effect of SB203580 on p38␣ activity in the presence of GSH, GSH and GSSG (1:1), or GSSG. p38␣ is insensitive to inhibition by SB203580 in an oxidized environment (Fig. 9D). This may explain why peroxynitrite-induced p38␣ phosphorylation requires its intrinsic kinase activity but is insensitive to SB203580.
Unlike the Tpx1-Sty1 Complex in Yeast, p38␣ Does Not Associate with Its Partner in the ϳ85-kDa Complex through Disulfide Bonding-Because yeast Tpx1 forms a disulfide complex with Sty1 in co-expression experiments (27), we sought to determine whether the same complex can be formed in mammalian cells. There are five members of 2-Cys peroxiredoxin in mammals, and we co-expressed peroxiredoxin-1, a counterpart of Tpx1, with p38␣. We were unable to detect a disulfide complex formation between peroxiredoxin-1 and p38␣ both before and after peroxynitrite or H 2 O 2 stimulation (data not shown). We used 1-Cys peroxiredoxin (peroxiredoxin-6) and obtained the same result (data not shown). It is possible that another form of peroxiredoxin is responsible for forming the disulfide complex with p38␣ in mammals. However, it is also possible that the complex we observed is different from the Tpx1-Sty1 complex in yeast.
Because Cys-35 in Sty1 forms a disulfide bond with Tpx1 in yeast, one way to address this question is to determine whether p38␣ utilizes Cys-39, which corresponds to Cys-35 in Sty1, to form a disulfide bond in the ϳ85-kDa complex. Because ectopically overexpressed FLAG-p38␣ apparently can replace the endogenous p38␣ in the ϳ85-kDa complex (Fig. 9C), we overexpressed mutants of p38␣ to determine the role of cysteine residues. There are four Cys residues in murine (and human) p38␣. We mutated each of the four Cys residues to Ser, expressed the mutants in MKK3/6 DKO cells, and examined whether any of the mutants is not associated with the ϳ85-kDa complex when treated with peroxynitrite. Mutating Cys-39, Cys-119, Cys-162, or Cys-211 did not eliminate peroxynitriteinduced p38␣ phosphorylation and the association of p38␣ with the ϳ85-kDa complex (Fig. 10A), suggesting two possibilities: either none of the four Cys residues is required for p38␣ to bind with the ϳ85-kDa complex or more than one Cys is involved in the binding. We then generated several mutants, including various combinations of double mutants, a triple mutant of Cys-39, Cys-119, and Cys-211, and a quadruple mutant of all four residues. As shown in Fig. 10, B, C, and D, all mutants were capable of being associated with the ϳ85-kDa complex and phosphorylated in MKK3/6 DKO cells after peroxynitrite treatment. This indicates that the association of p38␣ with the ϳ85-kDa complex is not through disulfide bonds. Therefore, the disulfide bond required for p38␣ to bind to the ϳ85-kDa complex is most likely formed between p38␣ binding partners or within a p38␣ binding partner.

DISCUSSION
The activity of p38␣ is regulated by cellular mechanisms that control its phosphorylation and dephosphorylation. As for the phosphorylation mechanisms, the trikinase cascade is the major pathway. However, our study using MKK knock-out MEF cells demonstrated that other cellular events are involved in p38␣ activation. Phosphorylation of p38␣ can be mediated by interactions with TAB1 and can be modulated by a yet unknown binding partner(s) in a manner dependent upon a disulfide complex. First we show that more than one p38␣ activation pathway could be activated in the same cell by a single stimulus, second we show that the selection of p38␣ activation mechanisms was both cell type-and cell stimulus-dependent, and third we show that different pathways may target differently localized p38␣.
Using MKK knock-out MEF cells, we demonstrated that MKK-independent p38␣ phosphorylation can indeed occur in cells (Fig. 1). TAB1-dependent p38␣ phosphorylation contributed to a small portion of p38␣ phosphorylation in MEF cells treated with some stimuli (Figs. 2, 3, and 4). However, this does not exclude the possibility that TAB1-mediated p38␣ activation plays a key role in certain biological processes. There are reports that TAB1-dependent p38␣ activation plays an important role in cellular responses to ischemia in the heart (23), the maturation of monocyte-derived dendritic cells (24), the maintenance of peripheral T-cell anergy (25), and intracellular parasite-induced interleukin-12 production (28). It is interesting to note that under certain conditions TAB1 also had an inhibitory effect on p38␣ phosphorylation (Fig. 4). We show that the inhibitory effect of TAB1 on p38␣ phosphorylation was restricted to p38␣ in a peroxynitrite-induced ϳ85-kDa complex (Figs. 7, 8, and 9). Because p38␣ phosphorylation in the ϳ85-kDa complex required intrinsic p38␣ activity (Fig. 9C), the inhibitory effect of TAB1 on this p38␣ phosphorylation could be a result of competition between different p38␣ regulation pathways. Because the inhibitory effect of TAB1 on MAP kinase phosphorylation was heavily dependent on stimuli like peroxynitrite, a specific signaling mechanism must be required for TAB1 to elicit this function. Phosphorylation of TAB1 by p38␣ on Ser-423, Thr-431, and Ser-438 was originally shown by Cheung et al. (22) and was proposed to be a feedback mechanism on TAK1 activation based on the observation that p38␣ inhibition resulted in increased TAK1 activity. Negative feedback by phosphorylated TAB1 on TAK1 cannot explain why TAB1 knockdown enhances p38␣ phosphorylation because our experiments were done in cells lacking MKK3 and MKK6, the downstream kinases of TAK1. However, the involvement of other types of negative feedback control, such as phosphatase induction, is still possible.
The TAB1-enhanced p38␣ phosphorylation in peroxynitrite-treated MEF cells is likely to be independent from MKK (Fig. 8). Unfortunately an siRNA approach cannot definitely exclude the possibility that MKK3/6 and MKK4 can compensate for each other in controlling p38␣ phosphorylation in the ϳ85-kDa complex (Fig. 9B). It is also possible that a kinase other than MKK3, MKK4, or MKK6 plays a role, but no additional MKK can be found in either the human or mouse genomes. The protein kinase PDZ-binding kinase (PBK/ TOPK) was suggested to be a p38 kinase in a report (29), but because PBK has no effect on JNK or ERK phosphorylation (29), it cannot be a kinase responsible for the TAB1 knockdownenhanced phosphorylation of p38␣, JNK, and ERK in peroxynitrite-treated MEF cells. A study in yeast suggested a model in which peroxide-induced Sty1 activation requires Tpx1-Sty1 disulfide bond formation and Wis1 (MKK)-dependent phosphorylation (27). We show that the ϳ85-kDa complex in mammalian cells is different as p38␣ did not form a disulfide bond with its partner in this complex (Fig. 10). In addition, peroxideinduced Sty1 activation is primarily dependent on Tpx-1, whereas only a small portion of peroxynitrite-induced p38␣ phosphorylation was associated with the ϳ85-kDa disulfide complex. Furthermore although yeast uses a two-component system to activate the Sty1 kinase cascade, a two-component system has not been found in mammals.
The activation of p38␣ MAP kinase is involved in a variety of cellular changes (13,14). Different stimuli may share some common mechanisms, such as the well established activation cascade of p38␣ by MKK3/6. Different stimuli may also use different mechanisms to activate p38␣. Complex formation through non-covalent and disulfide bonds is involved in the regulation of p38␣ activation under different conditions. To elicit different cellular responses, different pathways may activate p38␣ in different subcellular locations or protein complexes. The same MKK-mediated p38␣ activation could occur in different places within cells or within different protein complexes. Cell type is particularly important as we found that different cell types may utilize different mechanisms to mediate MAPK activation by peroxynitrite (10,23). Taken together, we have shown that p38␣ activation can be regulated by multiple mechanisms and that different modules of p38␣ activation exist for different stimuli and different cell types. The presence of multiple p38␣ activation mechanisms may be important for cells to differentially respond to a wide range of physiological and pathological stimuli with the necessary selectivity and fidelity.