The p38β Mitogen-activated Protein Kinase Possesses an Intrinsic Autophosphorylation Activity, Generated by a Short Region Composed of the α-G Helix and MAPK Insert*

Background: Kinases, involved in all processes of life, are enzymes capable of self-activation. MAP kinases, a kinase subfamily, do not self-activate. Results: We found that p38β is a unique MAP kinase that can self-activate. Significance: Through this we disclosed the element responsible for self-activation. Conclusion: Minor differences between kinases impose dramatic differences on their self-activation property. Protein kinases are regulated by a large number of mechanisms that vary from one kinase to another. However, a fundamental activation mechanism shared by all protein kinases is phosphorylation of a conserved activation loop threonine residue. This is achieved in many cases via autophosphorylation. The mechanism and structural basis for autophosphorylation are not clear and are in fact enigmatic because this phosphorylation occurs when the kinase is in its inactive conformation. Unlike most protein kinases, MAP kinases are not commonly activated by autophosphorylation but rather by MEK-dependent phosphorylation. Here we show that p38β, a p38 isoform that is almost identical to p38α, is exceptional and spontaneously autoactivates by autophosphorylation. We identified a 13-residue-long region composed of part of the αG-helix and the MAPK insert that triggers the intrinsic autophosphorylation activity of p38β. When inserted into p38α, this fragment renders it spontaneously active in vitro and in mammalian cells. We further found that an interaction between the N terminus and a particular region of the C-terminal extension suppresses the intrinsic autophosphorylation of p38β in mammalian cells. Thus, this study identified the structural motif responsible for the unique autophosphorylation capability of p38β and the motif inhibiting this activity in living cells. It shows that the MAPK insert and C-terminal extension, structural motifs that are unique to MAPKs, play a critical role in controlling autophosphorylation.

Protein phosphorylation is a central biochemical mechanism for controlling all biological processes. 518 genes, ϳ1.7% of all genes in the human genome, encode protein kinases that catalyze these phospho-group transfers (1). All eukaryotic protein kinases (EPKs) 2 adopt a similar topology that enables ATP binding and subsequent group transfer (2). This shared and highly conserved fold serves as a scaffold for additional loops and subdomains that appeared during the course of evolution in different kinases (2)(3)(4)(5) and play various roles, mainly in regulation and substrate recognition. Some of these elements, including the activation loop and the GHI helices subdomain (5,6), are common to all EPKs (2)(3)(4)(5). Other elements are familyspecific. Kinases of the MAPK family, for example, contain two motifs known as the MAPK insert and the C-terminal extension (1,5,7). These motifs impose dramatic effects on the proteins structure, but their roles in catalysis and regulation are not clear (see "Discussion").
EPK activity is regulated through a multitude of mechanisms, unique to each subfamily, or even to a single kinase. However, a regulatory mechanism common to almost all EPKs is activation via phosphorylation of a conserved activation loop threonine (5). In many protein kinases, this phosphorylation/activation event occurs by autophosphorylation, making activation loop autophosphorylation a fundamental reaction of living cells. The mechanism of the autophosphorylation reaction is poorly understood, but because it occurs when the kinase is in its inactive form, it must be different from that of substrate phosphorylation by active kinases (8,9).
The mammalian p38 family of MAPKs is composed of four isoforms; p38␣, p38␤, p38␥, and p38␦. This MAPK family is critical in mediating the cellular response to a wide variety of stresses, including DNA damage, oncogenic activity, and proinflammatory cytokines. Like all MAPKs, p38s have, in addition to the conserved activation loop threonine, a proximal tyrosine phospho-acceptor, composing a unique TGY phosphorylation motif. Phosphorylation of both sites is required for full catalytic ability and is mediated mainly by the MAPK kinases (MAPKK) MKK3 and MKK6 (10,11). Unlike many other EPKs, MAPKs do not manifest spontaneous autophosphorylation. However, it is now clear that p38s are capable of TGY autophosphorylation, but this reaction is tightly regulated and can occur only under particular situations (12)(13)(14)(15)(16)(17)(18). For example, p38␣ autophosphorylation is induced when it is in complex with TAB1 (17,19) or when phosphorylated on Tyr 323 by Zap70 (15). Furthermore, specific point mutations, D176A and F327(S‫گ‬L), also render p38␣ capable of autophosphorylation and autoactivation (13,14,20). p38␣ autophosphorylation has been shown to be physiologically relevant in the immune system (21,22) and in ischemic preconditioning (17,19) and to be sufficient to induce cellular senescence (23). The finding that MAPKs can autophosphorylate suggests that they are similar to most EPKs in this respect, but evolution provided them with inherent inhibition that prevents spontaneous autophosphorylation. Little is known about the structural basis that prevents MAPK spontaneous autophosphorylation and activates it when necessary.
We show here that when purified as a recombinant protein, p38␤ manifests spontaneous catalytic activity, which is a consequence of autophosphorylation. This observation is particularly surprising because of the high sequence identity between p38␤ and p38␣, which does not autophosphorylate spontaneously. We took advantage of the high sequence similarity on one hand and the dramatic difference in activity between p38␤ and p38␣ on the other to map the motif responsible for autophosphorylation and autoactivation. We have identified a C-terminal region, Arg 233 -Val 246 , which composes part of the G-helix and the MAPK insert (MKI), to be responsible for triggering the intrinsic autophosphorylation activity of p38␤. Inserting this region into p38␣ rendered it intrinsically active both in vitro and in HEK293 cells. We further show that the autophosphorylation activity of p38␤ is regulated in mammalian cells. Its regulation is mediated by part of the MKI and via an interaction between the N-lobe and C-tail. Thus, the autophosphorylation activity of p38␤ is regulated and executed by a number of C-terminal elements that were so far unnoticed.

EXPERIMENTAL PROCEDURES
Mammalian Cell Culture-HEK293 were grown in DMEM supplemented with 10% fetal bovine serum, sodium pyruvate, and antibiotics. Cells were grown at 37°C and 5% CO 2 . The cells were transfected with the calcium phosphate method.
Protein Expression and Plasmid Construction-For bacterial expression, the pET15b and pET28b vectors were used, with the ORFs hexahistidine-tagged N-terminally, as described previously (14). For expression, the Rosetta strain (Novagene) was used. For mammalian expression, the pcDNA3 (Stratagene) and pCEFL (Addgene) vectors were used with the ORFs tagged N-terminally with 3ϫ HA or Flag tag. Site-directed mutagenesis, using the Stratagene QuikChange kit, was performed according to the manufacturer's instructions. Truncation mutants were obtained by inserting stop codons by PCR with specific primers to p38␤ cDNAs. p38␣‫گ‬p38␤ chimeras were obtained without changing the native sequence by using five restriction sites: (i) the Kpn1 site in p38␤; (ii) the Bsu36i site, p38␤; (iii) the Sal1 site, p38␣; (iv) the Xho1 site, which was generated by to p38␤ by introducing a silent point mutation (CTGGAG 3 CTCGAG, CTC and CTG both encode Ile); and (v) the Mfe1 site, which was generated in p38␣ by introducing a silent mutation (CAGTTG 3 CAATTG). The corresponding restriction sites were generated in the other isoform by sitedirected mutagenesis. Chimeras that could not be obtained by restriction‫گ‬ligation were generated by PCR with the Phusion site-directed mutagenesis kit (thermo-scientific) according to the manufacturer's instructions.
Protein Purification-Protein purification from Escherichia coli cells was performed using nickel-nitrilotriacetic acid beads (Hadar Biotech, Rehovot, Israel) as previously described (14). Protein concentrations were determined by the Bradford method.
In Vitro Kinase Assay-All reactions with GST-ATF2 as a substrate were conducted in 96-well plates in triplicate. To initialize the reaction, 45 l of the reaction mixture were added to 5 l of p38 enzyme (0.2 g, 100 nM, of purified hexahistidine tag p38). Final reaction conditions were 25 mM Hepes, pH 7.5, 20 mM MgCl 2 , 20 mM 2-glycerolphosphate, 0.1 mM Na 3 VO 4 , 1 mM dithiothreitol, 40 g of GST-ATF2, 50 M ATP (kinase buffer), and 0.5 Ci of [␥-32 P]ATP. The reactions proceeded for 10 min at 30°C with agitation and were terminated with 50 l of 0.5 M EDTA, pH 8 (final concentration, 250 mM). Following reaction termination, 15 l from each set of samples were subjected to SDS-PAGE, stained with Coomassie, dried, and exposed to film, and aliquots of 85 l from each well were spotted onto 3X3-cm Whatman 3MM paper squares and air-dried. Each square was rinsed three times with 10% trichloroacetic acid and 3% sodium pyrophosphate (10 ml/square) for 1.5 h (each time) with gentle agitation, and a fourth wash for overnight was given without shaking. The following day, the squares were rinsed twice with 100% ethanol (4 ml/square) for 20 min each time and air-dried. The radioactivity of each square was counted using a scintillation counter running a 32 P Cherenkov program. Autophosphorylation reactions of proteins purified from E. coli were conducted under similar conditions, with 1 g of the tested p38 protein and without GST-ATF2. These reactions were analyzed by SDS-PAGE. For activation of p38 with MKK6, an active mutant of MKK6 in which Ser 207 and Thr 211 were both mutated to Glu (termed MKK6 EE ) was used under similar conditions. Immune complex kinase assay was preformed as previously described (24).
In-Solution Tryptic Protein Digest-5 g of p38␣ WT , p38␣ D176AϩF327S , p38␤ WT , and p38␤ D176A proteins were diluted to 8 M urea and 100 mM ammonium bicarbonate. Before tryptic digestion, cysteine residues were reduced with 2.8 mM DTT (Sigma; D9163-5G) for 30 min at 60°C and carboxymethylated with 8.8 mM iodoacetamide (Sigma; I6125-100G) for 30 min at room temperature in the dark. Trypsin (sequencing grade modified; Promega V511A) was added, after the samples were diluted in 2 M urea, in a ratio of 1:50 and incubated overnight at 37°C. A second dose of trypsin was added the following morning for 4 h at 37°C. The samples were then acidified with 0.1% TFA and desalted with micro TipColumn C-18 (Harvard Apparatus 74 -5201).
Analysis by Liquid Chromatography Tandem MS and Interpretation of MS Data-Peptides were resolved by reversed phase chromatography on 0.075 ID 200-mm fused silica capillaries (Agilent 160-2644-10) packed with Reprosil-Aqua C 18 reversed phase material (Dr. Maisch, GmbH, Germany). The peptides were eluted with linear gradient, 7-45% acetonitrile in 0.1% formic acid, at a flow rate of 0.25 l/min, using an Eksigent Nano capillary HPLC. Mass spectrometry was performed with an OrbitrapXP mass spectrometer (Thermo-Fisher) using the positive mode and repetitively full MS scan, followed by collision induced dissociation with multistage activation of the 10 most dominant ion selected from the full MS scan. The mass spectrometry data were analyzed using MaxQuant version 1.3.0.5 (25) with Andromeda (26) search engines, searching against the E. coli section of the UniProt database from September 2012 including human p38 proteins and their mutated variants with mass tolerance of 6 ppm for the precursor masses and 0.5 Da for the fragments. Identifications of the proteins and phosphorylation sites were filtered to 1% false discovery rate. The resulting identified phospho-sites table was filtered to include only phospho-sites on p38 proteins and their mutated variants.
Western Blot Analysis-For recombinant proteins, 100 ng were mixed with X4 Laemmli buffer and boiled at 100°C for 4 min. Mammalian cell protein lysates were prepared by decanting the growth medium, washing cells with PBS, and adding Laemmli buffer directly to the plate. Lysates were then collected and boiled for 10 min. The different samples were separated via a 10% SDS-PAGE and transferred to a nitrocellulose paper. Antibodies used in this study were: anti-phospho-p38 (Cell Signaling; #9211), anti-HA antibodies (obtained from the 12CA5 hybridoma cell line), anti-MAPKAPK2 (Cell Signaling; #3042) anti-phospho MAPKAPK2 (Cell Signaling; #3007), anti-Flag (Sigma; A2220), and anti-polyhistidine (Sigma; H1029).

p38␤, but Not p38␣, Possesses an Intrinsic Catalytic Activity
Caused by Autophosphorylation on Thr 180 -MAPKs do not manifest basal intrinsic activity but require MEK-mediated phosphorylation for activation. Surprisingly, we found that purified p38␤ WT is spontaneously active and able to phosphorylate GST-ATF2 in an in vitro kinase assay (Fig. 1A). To analyze and quantify this activity, we compared the ability of purified WT, kinase-dead (KD), and intrinsically active variants of p38␣ and p38␤ to phosphorylate GST-ATF2 in an in vitro kinase assay. p38␤ WT , but not p38␣ WT , was purified in an active state, demonstrated by its ability to phosphorylate ATF2 (Fig. 1A). The intrinsic activity of p38␤ WT is very significant, reaching a level of 35% of that of MKK6-phosphorylated p38␤ WT (Fig. 1A). The activity of the active mutant p38␤ D176A (12) is just slightly higher (Fig. 1A). Following incubation with MKK6 EE , the active variants of the p38 MAPKK, both p38␣ WT and p38␤ WT , were able to efficiently phosphorylate ATF2, indicating that both proteins can acquire a catalytically active conformation. The high intrinsic activity of p38␤ WT is unexpected because this protein is very similar to p38␣ ( Fig. 2A), which shows very low intrinsic activity, ϳ0.5% of the activity shown by MKK6-activated p38␣ WT (Fig. 1A).
The catalytic activity of the p38 proteins is expected to be correlated with phosphorylation of their TGY motif. Indeed, Western blot analysis showed that p38␤ WT reacted more efficiently with anti-phospho-p38 (␣-pp38) antibody than did p38␣ WT (Fig. 1A). On the other hand, the kinase dead (p38␤ KD ) variant, p38␤ K53R , reacted with the ␣-pp38 antibody only after MKK6 EE phosphorylation. It implies that the efficient phosphorylation of purified p38␤ WT is a result of an autophosphorylation activity of the TGY motif that is more efficient than that of p38␣ WT .
To test this notion and to see whether this autophosphorylation activity is unique to p38␤, we conducted an in vitro autophosphorylation assay with all p38 isoforms. Indeed, of the four isoforms tested, only p38␤ WT was phosphorylated on the TGY motif (Fig. 1B). Curiously, p38␦ WT manifested some autophosphorylation activity, as indicated by the radiogram (Fig. 1B). However, this phosphorylation does not occur on the TGY motif, as shown by Western blot (Fig. 1B). Thus, p38␤ is the only p38 isoform capable of spontaneous autophosphorylation of the TGY motif, explaining the observation that it is the only p38 isofom that manifests spontaneous activity.
The ␣-pp38 antibody, raised against dual phosphorylation of the TGY motif, also detects threonine 180-monophosphorylated p38 (24). Therefore, to unequivocally determine which residues are phosphorylated by autophosphorylation, we preformed an MS/MS analysis. p38␤ WT was found to be monophosphorylated on Thr 180 (Thr of the TGY motif) at very high levels compared with p38␣ WT . ϳ80% of the peptides derived from p38␤ WT that contain the TGY motif had a phosphate group on Thr 180 , compared with only ϳ7% of the corresponding peptides derived from p38␣ WT (Fig. 1C). Thr 180 was found to be phosphorylated at higher levels in p38␤ D176A and p38␣ D176AϩF327S than in their WT counterparts (Ͼ98% and ϳ40%, respectively). Interestingly, although no phosphorylation was detected on Tyr 182 of p38␤ WT or p38␤ D176A , ϳ5% phosphorylation on this residue was found in p38␣ D176AϩF327S , indicating that this protein possesses some tyrosine kinase activity. Although Thr ϩ Tyr dual phosphorylation is considered an absolute requirement for MAPK activity, we and others have shown that p38 and ERK molecules are active to various degrees even when monophosphorylated on the threonine phospho-acceptor (16,24,27). The intrinsic activity of monophosphorylated p38␤ (35% of maximum) seems exceptional because the activity of monophosphorylated p38␣, or ERK2 is much lower (16,24).
p38 autophosphorylation of the regulatory threonine, Thr 180 in p38␤ and p38␣, has been proposed to occur both by cis (19) and by trans (28) mechanisms. To determine by which mechanism the autophosphorylation of p38␤ WT occurs, we co-incubated purified His-tagged p38␤ WT with GST-tagged kinase dead p38␤ variant (p38␤ KD ) and monitored the phosphorylation status of these proteins via Western blot with the ␣-pp38 antibody. We found that although GST-p38␤ KD was phosphorylated by MKK6 EE , showing that it is a legitimate substrate, it was not phosphorylated by p38␤ WT , which was efficiently p38␤ MAPK Possesses Intrinsic Autophosphorylation Activity phosphorylated in this assay (Fig. 1D). Thus, the autophosphorylation of Thr 180 of p38␤ occurs in cis.
A Region Containing Part of the ␣-G Helix and the MKI Is Responsible for the Intrinsic Activity of p38␤-p38␣ and p38␤ are 75% identical and 88% similar ( Fig. 2A). Therefore, to identify the region responsible for the intrinsic activity of p38␤, we constructed a series of p38␣‫␤گ‬ chimeras (chimera swapping points are presented in Fig. 2A as dashed arrows numbered 1-6. A schematic representation and chimera numeration is presented in Fig. 2B). We first constructed two chimeras: p38␣‫,␤گ‬ composed of the N terminus of p38␣ fused to the 164 C-terminal amino acids of p38␤ and a reciprocal p38␤‫␣گ‬ chimera (Fig. 2B, chimeras 1 and 2). When tested for their ability to phosphorylate ATF2 in vitro, the p38␣‫␤گ‬ chimera manifested a high basal activity that was 9.4% of MKK6-activated p38␣ WT . No basal activity was observed for the reciprocal chimera, p38␤‫␣گ‬ (Fig. 2C). Correspondingly, Western blot analysis showed that p38␣‫␤گ‬ was phosphorylated to higher levels than p38␣ WT (Fig. 2C). Importantly, both chimeras were fully activated by MKK6, indicating that they preserved regulatory and catalytic properties of their parental proteins. These chimeras disclosed that the intrinsic activity of p38␤ resides somewhere in its C terminus. Two additional p38␣‫␤گ‬ chimeras, composed of the N-terminal part of p38␣ fused to the 83 and 31 C-terminal amino acids of p38␤ (Fig. 2B, chimeras 3 and 4) were both inactive unless pre-incubated with MKK6 (Fig. 2D), strongly suggesting that the region responsible for the autoactivity of p38␤ resides between Gln 218 -Ala 281 . Indeed, a ‫␣گ␤گ␣83‪p‬‬ chimera containing the Gln 218 -Ala 281 region of p38␤ (Fig. 2B, chimera 5) was active prior to MKK6 activation (Fig. 2E). Supporting this notion, a ‫␤گ␣گ␤83‪p‬‬ chimera protein containing the corresponding Thr 218 -Ala 281 region of p38␣ (Fig. 2B, chimera  8) was active only in a MKK6-dependent manner (Fig. 2F). To determine whether this entire region (Gln 218 -Ala 281 ) is FIGURE 1. p38␤, but not p38␣, is intrinsically active. A, in vitro kinase assay was preformed with the indicated p38 proteins with GST-ATF2 as a substrate. Activity is shown in graphs, expressed as percentages of the activity of MKK6-activated wild type p38␣ or p38␤ (100%), and by autoradiograms. 100 ng of each protein was also analyzed by Western blot with the indicated antibodies. B, in vitro autophosphorylation assay was preformed with the indicated p38 isoforms. Activity is shown by autoradiograms. 100 ng of each protein was also analyzed by Western blot with the indicated antibody. C, 5 g of p38␣ WT , p38␣ D761AϩF327S , p38␤ WT , and p38␤ D176A were incubated with ATP for 1 h in a kinase reaction mixture with no other substrate (autophosphorylation). Proteins were then cleaved into tryptic peptides and analyzed for phosphorylation sites by tandem MS. The relative intensities of the TGY motif-containing peptides in the different phosphorylation states: nonphosphorylated (Non Phos), Thr 180 -phosphorylated, and Thr 180 ϩ Tyr 182 -phosphorylated (0P, 1P, and 2P) are displayed as the percentages of all TGY peptides intensities (0P ϩ 1P ϩ 2P). D, 250 nM of purified His-tagged p38␤ WT or of MKK6 EE were incubated with 250 nM GST-tagged p38␤ KD in a kinase reaction mixture for 1 h. Samples of each reaction were analyzed by Western blot with the indicated antibodies. AUGUST 22, 2014 • VOLUME 289 • NUMBER 34 required for activity, we constructed two additional ‫␣گ␤گ␣83‪p‬‬ chimeras: the first containing Gln 218 -Val 246 of p38␤ and the second containing Leu 247 -Ala 281 (Fig. 2B, chimeras 6 and 7).

p38␤ MAPK Possesses Intrinsic Autophosphorylation Activity
vated p38␣ WT and was purified as a phosphorylated protein (Fig. 2E). However, in the reciprocal experiment conducted with ‫␤گ␣گ␤83‪p‬‬ chimeras, only the chimera with the 64-amino acid region of p38␣, Thr 218 -Ala 281 , did not possess MKK6independent activity, whereas the chimeras that contained part of this region maintained some MKK6-independent activity (Fig. 2, B, chimeras 9 and 10, and F). Therefore, the Thr 218 -Leu 246 region of p38␣ can replace that of p38␤ in the context of the p38␤ protein. We next tested whether the p38␤ fragment Gln 218 -Val 246 is the minimal region sufficient to render an autophosphorylating p38␣ protein. For this, we generated two additional chimeras, each containing approximately half of the Gln 218 -Val 246 region (Fig. 2B, chimeras 11 and 12). Interestingly, only chimera 12, containing the Arg 234 -Val 246 region, possessed intrinsic activity, ϳ8% of MKK6-activated p38␣ WT (Fig. 2G), whereas chimera 11, containing Gln 218 -Tyr 228 of p38␤, was active only in an MKK6-dependent manner (Fig. 2G). Finally, two ‫␣گ␤گ␣83‪p‬‬ chimeras, each containing part of the Arg 234 -Val 246 region (Fig. 2B, chimeras 13 and 14), were not intrinsically active (Fig. 2H). To show that the activity rendered on p38␣ by the small fragment of p38␤ is correlated with an increase in autophosphorylation activity, we conducted an in vitro autophosphorylation assay and found that chimera 12, a p38␣ protein containing the Arg 234 -Val 246 fragment of p38␤ indeed manifested a very high autophosphorylation activity (Fig. 2I). Thus, the Arg 234 -Val 246 region termed hereafter G-helix-MKI motif, is the minimal region of p38␤ required for autophosphorylation.
p38␤-specific Structural Elements Are Responsible for Regulating Its Intrinsic Activity in Mammalian Cells-To test whether p38␤ WT might be spontaneously active not only as recombinant purified protein, but also in mammalian cells, we overexpressed it in HEK293 cells. Like other p38 isoforms (20), p38␤ WT was not phosphorylated in unprovoked cells (Fig. 3A). Accordingly, p38␣ WT and p38␤ WT showed a very low ability to phosphorylate ATF2 when immunoprecipitated from these  WT . I, p38␣ WT and ‫␣گ␤گ␣83‪p‬‬ chimera 12 were tested in an autophosphorylation assay. Samples were collected at the indicated times, separated by SDS-PAGE, and exposed to film. cells, unless the cells were exposed to UV (Fig. 3B). Elevating expression levels of p38␣ WT and p38␤ WT resulted in only a slight increase in their apparent phosphorylation (Fig. 3C) that was not sufficient to induce phosphorylation of the p38 substrate MAPKAPK2 (MK2). Thus, the intrinsic activity of p38␤ WT is regulated in mammalian cells. On the other hand, intrinsically active variants p38␣ D176AϩF327S and p38␤ D176A were found to be spontaneously phosphorylated and active toward MK2 under the same conditions (Fig. 3, A and C) and manifested elevated catalytic activity when immunoprecipitated from cells not exposed to any activating signal (Fig. 3B).
Several possible explanations can account for the incapability of p38␤ to express its intrinsic activity in vivo. A general model would be that the suppression involves cellular components (because in vitro the activity is not suppressed) that may bind to particular domains in p38␤ and lock the G-helix-MKI motif. If so, the ‫␣گ␤گ␣83‪p‬‬ chimera that contains only the "activating fragment" (G helix-MKI) of p38␤ should not be suppressed in mammalian cells. Indeed, when expressed in HEK293 cells, the chimera containing Gln 218 -Val 246 , which possessed the highest intrinsic activity as a recombinant purified protein, was found to be spontaneously phosphorylated at high levels in HEK293 cells (Fig. 3D). Importantly, expression of this ‫␣گ␤گ␣83‪p‬‬ chimera led to a spontaneous, signal-independent phosphorylation of MK2. Thus, the G helix-MKI region of p38␤ is sufficient to activate p38␣ not only as a recombinant purified protein, but also in mammalian cell culture. These results support the notion that in addition to the region responsible for autoactivation found in p38␤, an additional region exists that is responsible for suppressing this activity in mammalian cells.

A p38␤ Molecule Lacking the C-terminal Tail Has Increased Spontaneous Activity Both in Mammalian Cells and in Vitro-
We next tried to identify the region responsible for suppressing the activity of p38␤ in mammalian cell culture. In previous studies, we found that the C termini of the yeast MAPKs Hog1 and Mpk1 suppress autophosphorylation activity (29,30). The MAPK extension, a MAPK-unique C-terminal region, is therefore an attractive region for a potential regulatory domain. The function of this region is still unknown. To test whether an autophosphorylation inhibitory domain may reside in the C terminus of p38␤, we generated and expressed in HEK293 cells a series of four C-terminal truncated p38␤ proteins (truncation points are marked by solid arrows 7-10 in Fig. 2A). The region selected for this analysis was Thr 341 -Gln 364 , composing the C-tail of p38␤. Of this region, the L16 helix (Tyr 342 -Phe 348 , marked by a solid box in Fig. 2A) was shown to be necessary for autophosphorylation in p38␣ and Hog1 (30). Also, this region contains the only p38␤ unique insertion Glu 352 -Lys 355 ( Fig. 2A,  dashed box). Of the series of C-terminally truncated p38␤ proteins, p38␤ K355 and p38␤ V345 were not found to be spontaneously phosphorylated in HEK293 cells but were phosphorylated when co-expressed with MKK6 EE (Fig. 4A). p38␤ F348 , however, showed a mild but clear elevation in its spontaneous phosphorylation levels (Fig. 4A). Thus, Lys 349 -Gln 364 , which composes an unstructured region C-terminal to the L16 helix, might be involved in suppressing autophosphorylation.
The proteins lacking the putative inhibitory C terminus were also tested as recombinant purified proteins. The activity of p38␤ K355 (Fig. 4B) was similar to that of p38␤ WT . However, p38␤ F348 displayed an elevated spontaneous (MKK6-independent) activity, ϳ160% over the spontaneous, MKK6-idependent, activity of p38␤ WT (Fig. 4B); this result is reminiscent of the increased spontaneous activity shown by p38␤ F348 in mammalian cells (Fig. 4A). This supports the notion that the Phe 348 -Lys 355 region of the C terminus bears inhibitory properties on the autophosphorylation of p38␤.
The C-terminal Part of p38␤ MAPK Insert Suppresses the Autophosphorylation Activity in Mammalian Cells-The data so far shows that: (i) replacing the G-helix and part of the MKI of p38␣ with that of p38␤ rendered this protein intrinsically active in vitro and in HEK293 cells (Figs. 2G and 3D) and (ii) FIGURE 4. The C tail of p38␤ inhibits autophosphorylation. A and B, premature stop codons were inserted into p38␤ generating p38␤ C-terminal truncation mutants p38␤ T341 , p38␤ V345 , p38␤ F348 , and p38␤ K355 . A, these proteins and p38␤WT or an empty vector (emp) were transiently expressed in HEK293 cells with or without plasmids expressing MKK6 EE . Cell lysates were prepared, and 30 g of protein from each sample were separated by SDS-PAGE, blotted, and probed with the indicated antibodies. A summary of all chimeras tested is provided in Fig. 2B. B, 200 ng of recombinant p38 proteins were subjected to a kinase assay as described in the legend to Fig. 1.

p38␤ MAPK Possesses Intrinsic Autophosphorylation Activity
truncation of the C-terminal extension of p38␤ up to the L16 helix (generating p38␤ F348 ) partially relieved p38␤ autoactivation from inhibition in HEK293 cells (Fig. 4A). We postulated that if the C-terminal extension has an inhibitory role, perhaps replacing the C-terminal extension in the active ‫␣گ␤گ␣83‪p‬‬ chimera, which originates from p38␣, with that of p38␤ would eliminate the spontaneous activity of this protein in HEK293 cells. Therefore, we generated a ‫␤گ␣گ␤گ␣83‪p‬‬ chimera that has, in addition to the G-helix-MKI (Gln 218 -Val 246 ) region from p38␤, also the C terminus of p38␤; Leu 334 -Gln 364 (Fig. 2B, chimera 15). However, this ‫␤گ␣گ␤گ␣83‪p‬‬ chimera not only did not lose spontaneous activity, it seemed to be more active than the "parental" chimera. ‫␤گ␣گ␤گ␣83‪p‬‬ was phosphorylated at higher levels and led to higher phosphorylation of MK2 in HEK293 cells than ‫␣گ␤گ␣83‪p‬‬ (Fig. 5A). The data combined suggest that the C terminus of p38␤ is an inhibitor of the autophosphorylation activity but cannot function alone. To determine which elements cooperate with the C-terminal inhibitory region, we tested the phosphorylation of a p38␣‫␤گ‬ chimera, composed of the N terminus of p38␣ and the C terminus of p38␤ and a p38␤‫␣گ‬ reciprocal chimera in HEK293 cells (Fig. 2B, chimeras 1  and 16). Of note, both chimeras contained the Gln 218 -Val 246 activating region. Interestingly, both the p38␣‫␤گ‬ chimera and the p38␤‫␣گ‬ chimera were found to be phosphorylated at similar levels like the ‫␣گ␤گ␣83‪p‬‬ chimera (Fig. 5B). Thus, when the N terminus and the C terminus are of different isoforms, inhibition of autophosphorylation is not achieved. In MAPKs, the C-terminal extension protrudes into the N-lobe in a unique manner (5). The function of this unique interaction is not known, but now we propose that disruption of this interaction, either by truncation of the C-terminal extension or by generating a chimera that contains an N terminus of one isoform and a C terminus of another, leads to an increase in spontaneous activity (Figs. 4A and 5A). To further test this notion, we expressed two additional ‫␣گ␤گ␣83‪p‬‬ chimeras that contain the activating fragment in HEK293 cells. The first chimera contains a p38␤ fragment (Gln 218 -Ala 281 ) that, in addition to the activating G-helix-MKI region, based on the crystal structure, is not in the interface of the C-to N-terminal interaction, thus maintaining the N-to C-terminal interaction of p38␣. The second chimera contains a larger p38␤ region (Gln 218 -Thr 333 ), enforcing C-tail to N-lobe interaction from the two different isoforms (Fig. 2B, chimeras 5 and 17). Indeed, although chimera 17 was found to be active, chimera 5 was not (Fig. 5C). Based on this, we suggest the following model: p38␤ spontaneous activity depends on a fragment composed of the G-helix and part of the MKI (Arg 234 -Val 246 ). This activity is suppressed in cells via a mechanism involving the remaining part of the MKI (Leu 247 -Ala 281 ), as well as the interaction between the C-terminal extension and the N-lobe (Fig. 6).

DISCUSSION
During the course of evolution, MAPKs acquired some properties that distinguish them from other EPKs. A prominent property is the requirement of an unusual, MEK-dependent, dual phosphorylation for activation. Evolution also equipped MAPKs with two unique structural motifs: the MAPK insert (also found in CDKs) and a C-terminal extension. In this study, we show that these MAPK-specific motifs are critical players in controlling MAPK autophosphorylation. Unlike the spontaneous autophosphorylation manifested by most EPKs, such activity is rarely monitored in MAPKs. However, there are strong indications that MAPKs are capable of efficient autophosphorylation that is manifested only under very specific conditions (12, 13, 15-17, 31). Consequently, structural elements involved in executing autophosphorylation, or rather, elements blocking spontaneous autophosphorylation and thereby differing MAPKs from most EPKs, were not revealed. We were able to propose such elements through the unusual spontaneous autophosphorylation property of p38␤.
Although this activity is very specific to p38␤, the regions found to be responsible for it turned out to be very similar not only in p38␣ (Fig. 6A), but also in other p38 isoforms and even in other MAPKs. The most critical region identified, Arg 234 -Val 246 , which contains the capability to endow p38␣ with intrinsic autophosphorylation, is composed of the C-terminal part of the G-helix (Fig. 6B), which is a conserved structural motif in all EPKs, and the N-terminal part of the MKI, including the ␣1L14 helix, which is conserved in all MAPKs (1,2,7). This region is highly conserved within the p38 family so that bioinformatic and structural analyses could not point at it as p38␤specific. Notably, of the 13 residues that compose the G-helix-MKI fragment responsible for p38␤ intrinsic activity, six are identical and two are conserved in p38␣ (Fig. 6A), explaining why the domain was not disclosed in simple sequence analysis.
The identification of the G-helix-MKI fragment as the trigger of autophosphorylation was also unexpected because the G-helix and MKI were so far assigned roles in nuclear localization (32) and not in catalysis or in its regulation. This region also does not seem to be related to the R or C spines that regulate  red in B) and depends on an interaction between the N-lobe and C-tail. The C-tail, composed of L16 and the L16 helix, the N-lobe elements that interact with the C-tail, the C helix, and the ␤4-␤5 loop, is shown in orange.

p38␤ MAPK Possesses Intrinsic Autophosphorylation Activity
activity of protein kinases (5,6,33,34). However, in addition to its role in nuclear translocation, the MKI was reported to serve as an allosteric site for binding of regulators, activators, substrates, and lipids (35,36). The GHI helices subdomain, including the MKI in MAPKs, has been shown to be more flexible than other regions in a number of different kinases (5) and may also play a role in allosteric regulation of kinase activity (36). However, structure-function studies of MAPKs have mainly addressed the mechanisms and structural requirements of catalysis toward substrates and not of autophosphorylation, which must have specific requirements because it occurs when kinases are in their inactive conformation (8,9).
There are many crystal structures available of p38␣, including nonphosphorylated (37), dually phosphorylated (38), and induced to autophosphorylate, either by point mutation (28) or via interaction with TAB1 (19). Only two crystal structures of p38␤ (39) were reported, both in complex with a small inhibitor molecule. Therefore, structural comparison is limited. Nevertheless, comparing the structures of p38␣ and p38␤ shows an almost complete overlap of the folds of the regions identified here (Fig. 6B). Furthermore, the G-helix-MKI region does not seem to adopt an alternative conformation in the structures of p38␣ in complex with the TAB1 peptide or harboring activating point mutations that render it an autophosphorylating protein. Therefore, just like the sequence alignment, structural analysis could not disclose the role of this region in autophosphorylation.
Acquisition of an active conformation in p38␣ is accompanied with reduced mobility of various regions throughout the kinase domain (40). The Arg 234 -Val 246 region may project on other regions in the molecule and perhaps affect their mobility, allowing ATP binding and conferring correct alignment of the catalytic spines. The activation loop is a candidate to be affected by this region because it is proximal to the G helix in the threedimensional structure (28). Interestingly, the region responsible for the spontaneous activity of JNK2␣2, the only other mammalian MAPK with reported spontaneous activity (41), does not overlap with the region we have identified in this study.
Finally, the notion that an N-terminal-C-terminal tail interaction inhibits autophosphorylation in p38␤ may explain results obtained with Hog1 and Mpk1 in which C-terminal truncations evoke intrinsic activity (29,30). Perhaps in those and other MAPKs, N-terminal-C-terminal interactions suppress autophosphorylation. The C terminus proximity to the N-lobe is a unique feature of MAPKs. In most EPKs that undergo spontaneous autophosphorylation, the termini reside away from each other, supporting the notion that the role of the C-terminal protrusion into the N-lobe in MAPKs is to inhibit the autophosphorylation of MAPKs (5). This notion at least partially explains why autophosphorylation is so rarely manifested by this group of kinases.
Because p38 activity, including p38␤ (42), has dramatic effects on cell fate, their activation, either by MEKs or by autophosphorylation, must be tightly controlled. Indeed, we show that the autophosphorylation-dependent activation of p38␤ is regulated in vivo by cellular components. These components could either be an interacting protein physically inhibiting the conformation required for autophosphorylation or constitutive down-regulation by phosphatases. A candidate phosphatase could be MKP1, which is likely to bind the CD domain (43).
p38 autophosphorylation is of high physiological and clinical importance (19,44). Because mammalian MAPK families are composed of a number of isoforms that are regulated in vivo by the same MAPKKs, autophosphorylation provides an elegant mechanism for isoform-specific activation. Given the efficient intrinsic autophosphorylation of p38␤, specific activation of this isoform would require just desuppression of its intrinsic activity.