A cDNA was cloned that encodes human stress‐activated protein kinase‐4 (SAPK4), a novel MAP kinase family member whose amino acid sequence is ∼60% identical to that of the other three SAP kinases which contain a TGY motif in their activation domain. The mRNA encoding SAPK4 was found to be widely distributed in human tissues. When expressed in KB cells, SAPK4 was activated in response to cellular stresses and pro‐inflammatory cytokines, in a manner similar to other SAPKs. SAPK4 was activated in vitro by SKK3 (also called MKK6) or when co‐transfected with SKK3 into COS cells. SKK3 was the only activator of SAPK4 that was induced when KB cells were exposed to a cellular stress or stimulated with interleukin–1. These findings indicate that SKK3 mediates the activation of SAPK4. The substrate specificity of SAPK4 in vitro was similar to that of SAPK3. Both enzymes phosphorylated the transcription factors ATF2, Elk‐1 and SAP‐1 at similar rates, but were far less effective than SAPK2a (also called RK/p38) or SAPK2b (also called p38β) in activating MAPKAP kinase‐2 and MAPKAP kinase‐3. Unlike SAPK1 (also called JNK), SAPK3 and SAPK4 did not phosphorylate the activation domain of c‐Jun. Unlike SAPK2a and SAPK2b, SAPK4 and SAPK3 were not inhibited by the drugs SB 203580 and SB 202190. Our results suggest that cellular functions previously attributed to SAPK1 and/or SAPK2 may be mediated by SAPK3 or SAPK4.
Four mitogen‐activated protein (MAP) kinase family members are activated by cellular stresses (chemical, heat and osmotic shock, UV radiation, inhibitors of protein synthesis), bacterial lipopolysaccharide (LPS), and the cytokines interleukin‐1 (IL1) and tumour necrosis factor (TNF), and have therefore been termed stress‐activated protein kinases or SAPKs (reviewed in Cohen, 1997). Isoforms of SAPK1 [also called c‐Jun N‐terminal kinases (JNKs)] phosphorylate Ser63 and Ser73 in the activation domain of c‐Jun (Pulverer et al., 1991), thereby increasing its transcriptional activity. The same sites in c‐Jun also become phosphorylated when cells are exposed to the stresses and cytokines that activate SAPK1 (Pulverer et al., 1991; Hibi et al., 1993; Dérijard et al., 1994; Kyriakis et al., 1994), suggesting that c‐Jun is a physiological substrate for SAPK1.
SAPK2a [also termed p38 (Han et al., 1994), p40 (Freshney et al., 1994), RK (Rouse et al., 1994), CSBP (Lee et al., 1994) and Mxi2 (Zervos et al., 1995)] is inhibited very specifically by the pyridinyl imidazoles SB 203580 and SB 202190 (Lee et al., 1994; Cuenda et al., 1995; reviewed in Cohen, 1997) which have been exploited to identify several physiological substrates. These include four protein kinases, namely MAP kinase‐activated protein kinase‐2 (MAPKAP‐K2, Rouse et al., 1994) and the closely related MAPKAP‐K3 (Clifton et al., 1996; McLaughlin et al., 1996; Sithanandam et al., 1996), as well as MAP kinase interacting protein kinases‐1 and ‐2 (Mnk1 and Mnk2) (Fukunaga and Hunter, 1997; Waskiewicz et al., 1997).
Physiological substrates of MAPKAP‐K2/K3 include heat shock protein (HSP) 27 (Cuenda et al., 1995; Huot et al., 1995) and the transcription factor CREB (Tan et al., 1996), whereas transcription factor eIF4E is a physiological substrate of Mnk1/2 (Waskiewicz et al., 1997). SAPK2a also mediates the stress‐induced phosphorylation and activation of the CEBPβ‐related transcription factor CHOP (Wang and Ron, 1996) and the ternary complex factor Elk‐1 (Price et al., 1996).
Based on the effects of SB 203580, the activation of SAPK2a is rate‐limiting in the LPS‐induced production of IL1 and TNF in monocytes (Lee et al., 1994), in the TNF‐stimulated transcription of IL6 and GM‐CSF in fibroblasts (Beyaert et al., 1996), in the IL1‐induced stimulation of glucose uptake in epithelial cells (Gould et al., 1995), in collagen‐induced platelet aggregation (Saklatvala et al., 1996) and in the stress‐induced transcription of c‐Jun and c‐Fos in fibroblasts (Hazzalin et al., 1996; Price et al., 1996). The SAPK2a catalysed phosphorylation of Elk‐1 (Price et al., 1996) and the MAPKAP‐K2 catalysed phosphorylation of CREB (Tan et al., 1996) are both likely to contribute to the stress‐induced transcription of c‐Fos (Ginty et al., 1994).
Recently, two additional SAP kinases were identified, called SAPK2b [or p38β (Jiang et al., 1996)] and SAPK3 (Mertens et al., 1996) [also called ERK6 (Lechner et al., 1996) and p38γ (Li et al., 1996)]. The amino acid sequence of SAPK2b is 73% identical to SAPK2a and it is inhibited by SB 202190 at similar concentrations to SAPK2a. In contrast, the amino acid sequence of SAPK3 is only 60% identical to SAPK2a and SAPK2b, and SAPK3 is not inhibited by SB 203580 (Cuenda et al., 1997). SAPK2b and SAPK3 have been introduced into mammalian cells by transient transfection and shown to be activated in response to pro‐inflammatory cytokines and stressful stimuli in a manner similar to SAPK1 and SAPK2a. The physiological roles of SAPK2b and SAPK3 are unknown. The mRNAs encoding these enzymes are present in all mammalian tissues examined (Jiang et al., 1996; Mertens et al., 1996; Goedert et al., 1997), with the mRNA encoding SAPK3 being most abundant in skeletal muscle. Expression of wild‐type SAPK3 and an inactive mutant in the muscle cell line C2C12 respectively enhanced and inhibited differentiation into myotubes (Lechner et al., 1996). In vitro, SAPK2b and SAPK3 phosphorylated several proteins that are also substrates for SAPK2a. SAPK2b was reported to phosphorylate the transcription factor ATF2 more efficiently than SAPK2a (Jiang et al., 1996) but, since the stress‐ and cytokine‐ induced phosphorylation of ATF2 in fibroblasts is unaffected by SB 203580 (Beyaert et al., 1996; Hazzalin et al., 1996), neither SAPK2a nor SAPK2b appears to be rate‐limiting for ATF2 phosphorylation in vivo, in contrast to earlier studies using transfection‐based approaches (Gupta et al., 1995). The substrate specificity of SAPK3 in vitro was similar to that of SAPK2a, except that it was much less effective in activating MAPKAP‐K2/K3 and (like SAPK1, but unlike SAPK2a) phosphorylated ATF2 at Ser90, as well as at Thr69 and Thr71 (Cuenda et al., 1997). However, whether SAPK1 and/or SAPK3 are rate‐limiting for ATF2 phosphorylation in vivo is unknown.
Five chromatographically distinct SAP kinase kinases (SKKs) have been identified in mammalian cells (Meier et al., 1996; Cuenda et al., 1996). In vitro, SKK1 [also termed MKK4 (Dérijard et al., 1995), SEK1 (Sanchez et al., 1994) and XMEK2 (Yashar et al., 1993)] activates all four SAPKs (Sanchez et al., 1994; Dérijard et al., 1995; Doza et al., 1995; Jiang et al., 1996; Cuenda et al., 1997), although SAPK2b and SAPK3 are phosphorylated less efficiently. SKK2 [also termed MKK3 (Dérijard et al., 1995)] and SKK3 (Cuenda et al., 1996) [also called MKK6 (Han et al., 1996; Moriguchi et al., 1996; Raingeaud et al., 1996) and MEK6 (Stein et al., 1996)] activate SAPK2a but not SAPK1, while SKK3 was the only detectable activator of SAPK3 induced by pro‐inflammatory cytokines and stressful stimuli in human epithelial KB cells or human embryonic kidney 293 cells (Cuenda et al., 1997). SKK3 was also the most efficient activator of SAPK2b in co‐transfection experiments (Jiang et al., 1996). SKK4 and SKK5 activate SAPK1 but not SAPK2a (Meier et al., 1996) or SAPK3 (Cuenda et al., 1997). SKK4/SKK5 have not been purified or cloned and their amino acid sequences are thus still unknown.
In this paper, we report the cloning and characterization of a novel MAP kinase family member that we call SAPK4. This enzyme, which also contains a TGY sequence in the activation domain, shows ∼60% identity to SAPK2a, SAPK2b and SAPK3, and its mRNA is widely expressed in human tissues. We show that SAPK4 is activated by the same stimuli that activate other SAP kinases and that SKK3 is likely to be the major upstream activator of SAPK4 in vivo. The substrate specificity of SAPK4 in vitro is similar to that of SAPK3 and, like SAPK3, SAPK4 is not inhibited by SB 203580 or SB 202190.
Molecular cloning of human SAPK4
To identify novel members of the SAPK family, we used EST clone 156272 (GenBank accession number R72662), which encodes a portion of SAPK2b (p38β), as the probe to screen a human pituitary gland cDNA library. Sequencing of positive clones showed three distinct sequences. The first set of clones encoded portions of ERK5 (Zhou et al., 1995) [also called BMK1 (Lee et al., 1995)], whereas the second set encoded a novel form of p38β (see Discussion). The third set of clones encoded a protein whose sequence was similar to, but differed from that of known SAPKs and which was consequently named stress‐activated protein kinase‐4 (SAPK4). The nucleotide and deduced amino acid sequence of human SAPK4 is shown in Figure 1. An open reading frame encodes a protein of 365 amino acids, with a predicted molecular mass of 42 kDa. It possesses the conserved amino acid domains (I–XI) characteristic of protein kinases and shows 64% sequence identity with SAPK3, 59% identity with SAPK2a, 58% identity with SAPK2b, 42% identity with HOG1 from Saccharomyces cerevisiae (Figure 2), 45% identity with SAPK1 and 41% identity with p42 MAP kinase. Residues Thr180 and Tyr182 in subdomain VIII are in an equivalent position to the TEY, TPY and TGY sequences in known MAP kinases and SAP kinases, phosphorylation of which is required for enzymatic activity. SAPK4 shares a TGY sequence with SAPK2a, SAPK2b, SAPK3 and HOG1 (Figure 2). Moreover, as in SAPK2a, SAPK2b, SAPK3 and HOG1, subdomain VII is separated by only six amino acids from the activation region in subdomain VIII (Figure 2), whereas this gap is eight residues in SAPK1 and over 12 residues in MAP kinases. The tissue distribution of SAPK4 mRNA in human tissues was assessed by RNA blotting (Figure 3). Hybridization of 32P‐labelled SAPK4 cDNA to multiple tissue Northern blots showed a transcript of ∼2.3 kb which was present in most of the 16 tissues examined, albeit at variable levels (Figure 3). Highest levels were detected in pancreas, testis, small intestine and prostate gland. Hybridization of the blots with a probe for β‐actin showed approximately equal loading of RNA (data not shown).
SAPK4 is activated by cellular stresses and cytokines
SAPK4 is most closely related to SAPK3, SAPK2a and SAPK2b, enzymes that are activated by cellular stresses and the cytokines IL1 and TNF (see Introduction). We therefore investigated whether the same stimuli would activate SAPK4. Human epithelial KB cells were transiently transfected with a myc epitope‐tagged SAPK4 and, after exposure to cellular stresses or cytokines, the enzyme was immunoprecipitated and assayed. These experiments showed that the stimuli which trigger the activation of SAPK2a, SAPK2b or SAPK3 also activate SAPK4 (Figure 4), while stimuli that do not activate SAPK2a, SAPK2b or SAPK3 [such as insulin‐like growth factor‐1 (IGF–1) and phorbol esters] also failed to activate SAPK4 (Figure 4).
Identification of SKK3 as the major activator of SAPK4 in epithelial cells
We have shown previously that SKK3 accounts for 95% and SKK2 for 5% of the SAPK2a activator detected after Mono S chromatography of lysates from KB cells that have been stressed in several ways or stimulated with IL1 (Cuenda et al., 1996; Meier et al., 1996). As shown in Figure 5, the single peak of SAPK4 activator detected after subjecting KB cells to IL1 or the protein synthesis inhibitor anisomycin co‐migrated with SKK3 on Mono S and was immunoprecipitated quantitatively and specifically by anti‐MKK6 antibodies. This experiment also demonstrated that SKK4 and SKK5, which are activated in KB cells by these stimuli, and which elute from Mono S at a higher NaCl concentration than SKK3 (Meier et al., 1996), do not activate SAPK4. SKK3 was also the only activator of SAPK4 after subjecting KB cells to osmotic stress using sorbitol (data not shown). Further evidence that SKK3 can activate SAPK4 in vivo was obtained by co‐transfection into COS cells. SAPK4 activity was elevated 12.5 ± 0.63‐fold by co‐expression with SKK3 (n = 3). In contrast, SAPK4 was not activated significantly by co‐transfection with MEK kinase, under conditions where MEK kinase activated the endogenous SKK1 in COS cells (data not shown).
Activation of SAPK4 by SKK3 in vitro
SAPK4 was activated in vitro by a highly purified preparation of SKK3 from skeletal muscle (Figure 6), but could not be activated by MKK1 under conditions where p42 MAP kinase was activated maximally (data not shown). The activation of SAPK4 by SKK3 occurred 2‐ to 3‐fold more slowly than that of SAPK2a, although both enzymes attained the same specific activity towards myelin basic protein (MBP) after 2 h (Figure 6A). The activity of SAPK2a and SAPK4 towards MBP is 25‐ to 50‐ fold lower than the activity of p42 MAP kinase towards this substrate (Stokoe et al., 1992). The activation of SAPK4, like that of SAPK3 (Cuenda et al., 1997), reached a plateau at ∼1.6 mol/mol subunit (data not shown) and was accompanied by the appearance of phosphotyrosine and phosphothreonine in similar amounts (Figure 6B). Interestingly, SAPK4, like SAPK3 but unlike SAPK2a, also became phosphorylated at a serine residue(s). This did not occur when wild‐type SAPK4 was replaced by the catalytically inactive SAPK4‐D168A mutant (Figure 6B), indicating that serine phosphorylation is catalysed by SAPK4 itself after it has been activated. Interestingly, the inactive SAPK4 mutant could only be phosphorylated on tyrosine and not on threonine, indicating that this mutation induces a conformational change that affects the recognition of SAPK4 by SKK3.
Activation of SAPK4 by SKK1 (MKK4) in vitro
SAPK2 is not only phosphorylated by SKK2 and SKK3 in vitro, but also by SKK1 (MKK4) (see Introduction). SKK1 is activated by cytokines and cellular stresses and it was therefore of interest to investigate whether this enzyme also had the potential to activate SAPK4 in vitro. Figure 7 shows that SAPK4 is phosphorylated and activated by SKK1 in vitro, but at a much slower rate than SAPK2a under identical conditions. Consistent with this slow rate of activation, SAPK4 was not activated after co‐transfection into COS1 cells with MEK kinase (data not shown), an upstream activator of SKK1. This is consistent with the negligible contribution of SKK1 to the SAPK4‐activating activity in KB cell extracts (Figure 5).
Comparison of the substrate specificity of SAPK4 with that of SAPK2a, SAPK2b and SAPK3
SAPK4 was more active than SAPK2a or SAPK2b in phosphorylating fusion proteins with glutathione S‐transferase (GST) linked to the activation domains of the transcription factors Elk‐1, ATF2, SAP‐1, SAP‐2 and p53, while c‐Jun was only phosphorylated poorly by the three enzymes (Table I). SAPK4 was far less effective than SAPK2a in activating GST–MAPKAP‐K2(5–400) and full‐length GST‐MAPKAP‐K3, the initial rate of activation of MAPKAP‐K2 and MAPKAP‐K3 and the half‐time for maximal activation being 20 times slower (Figure 8). The specificity of SAPK4 in vitro was similar to that of SAPK3 (Table I; Cuenda et al., 1997). Consistent with their similar amino acid sequences (Figure 2), the substrate specificities of SAPK2a and SAPK2b were virtually indistinguishable (Table I and Figure 8).
SAPK4 is not inhibited by SB 203580 or SB 202190
Like SAPK3, bacterially expressed SAPK4 that had been activated by SKK3 in vitro was not inhibited by SB 203580 or SB 202190 (Figure 9). In contrast, SAPK2a and SAPK2b were inhibited by both drugs with similar IC50 values of 0.3–0.6 μM (Figure 9). SB 203580 and SB 202190 also failed to inhibit the dephosphorylated forms of SAPK3 and SAPK4 which have <0.1% of the activity of the phosphorylated forms (data not shown). In addition, SB 203580 did not prevent the activation of transfected SAPK3 by osmotic shock in KB cells.
In this paper we report the cloning of the cDNA encoding human SAPK4, a novel MAP kinase family member whose mRNA is widely expressed and whose amino acid sequence is ∼60% identical to SAPK2a, SAPK2b and SAPK3. Moreover, SAPK4, like SAPK2a, SAPK2b and SAPK3, contains the dual TGY phosphorylation motif and a six amino acid insertion between subdomain VII and the activation loop in subdomain VIII. SAPK4 is less similar to SAPK1 (45% identity) and p42/p44 MAP kinases (41% identity). During the course of this study, we have also sequenced several cDNA clones for SAPK2b. They were found to encode a protein of 364 amino acids that differs from the published p38β sequence (Jiang et al., 1996) in two respects. It lacks an eight amino acid sequence between kinase subdomains V and VI and it shows two amino acid differences in this region. As a result, our SAPK2b sequence aligns with the other SAP kinase sequences without requiring any gaps.
Consistent with its amino acid sequence similarity to SAPK2a, SAPK2b and SAPK3, SAPK4 is activated in response to the same cellular stresses and cytokines. The only activator of SAPK4 that could be detected in extracts prepared from epithelial KB cells exposed to a cellular stress or IL1 was SKK3, the product of the MKK6 gene. SAPK4 also became activated when co‐transfected with SKK3 DNA into COS cells. In contrast, SAPK4 was activated poorly by SKK1 (MKK4) in vitro and was not activated in co‐transfection experiments with MEK kinase, an upstream activator of SKK1 (Yan et al., 1994). SAPK4 was also not activated by MKK1, SKK2 (MKK3), SKK4 or SKK5. Thus, SKK3 which is the dominant activator of SAPK2a (Cuenda et al., 1996), SAPK2b (Jiang et al., 1996) and SAPK3 (Cuenda et al., 1997) in several mammalian cell lines, also appears to be the major activator of SAPK4.
In contrast to SAPK2a and SAPK2b, which are inhibited by SB 203580 or SB 202190 at submicromolar concentrations, SAPK3 and SAPK4 were not affected by these drugs; moreover, SB 203580 did not inhibit the activation of SAPK3. The failure of SB 202190 to inhibit SAPK3 is in disagreement with the work of Li et al. (1996) who reported that the basal activity of bacterially expressed SAPK3 was inhibited by this drug. The reason for this discrepancy is unclear, because we failed to find any effect of SB 203580 or SB 202190 on either the basal activity of expressed SAPK3 or on SAPK3 that had been maximally activated by phosphorylation with SKK3. Identical results were obtained using bacterially expressed rat SAPK3 (Cuenda et al., 1997) and human SAPK3.
The substrate specificity of SAPK4 in vitro resembled that of SAPK3 in that, while both enzymes phosphorylated a number of proteins (including the activation domains of several transcription factors) at similar rates to SAPK2a and SAPK2b, they were far less effective in activating MAPKAP‐K2 and MAPKAP‐K3 than either SAPK2a or SAPK2b. The last mentioned result is consistent with the finding that SB 203580 suppresses the stress‐ and cytokine‐induced activation of MAPKAP‐K2 and MAPKAP‐K3 by 80–95% in every mammalian cell so far examined (Cuenda et al., 1995; Beyaert et al., 1996; Clifton et al., 1996; McLaughlin et al., 1996). Moreover, although SAPK3 and SAPK4 phosphorylate the activation domain of the transcription factor Elk‐1 efficiently in vitro, neither enzyme appears to be rate‐limiting for Elk‐1 phosphorylation in vivo, because Elk‐1 phosphorylation induced by cellular stresses can be prevented by SB 203580 in fibroblast cell lines (Hazzalin et al., 1996) or by a combination of SB 203580 and PD 98059 in HeLa cells (Price et al., 1996). PD 098059, a specific inhibitor of the activation of MKK1 (Alessi et al., 1995), prevents the activation of p42/p44 MAP kinases. Candidates as physiological substrates for SAPK3 and SAPK4 are proteins whose phosphorylation/activation triggered by cellular stresses and/or pro‐inflammatory cytokines is not prevented by SB 203580. Such proteins include the transcription factors c‐Jun, ATF2 and NFκB (Beyaert et al., 1996; Hazzalin et al., 1996). However, c‐Jun is phosphorylated very poorly by SAPK3 and SAPK4 and isoforms of SAPK1 are likely to phosphorylate this protein in vivo (see Introduction).
In summary, the number of MAP kinase family members which are activated by cellular stresses and/or pro‐inflammatory cytokines is much greater than was realized previously (shown schematically in Figure 10). In addition to SAPK1, SAPK2a, SAPK2b, SAPK3 and SAPK4, the p42/p44 MAP kinases, which are strongly activated by growth factors, are also activated by stressful stimuli and pro‐inflammatory cytokines in some cellular backgrounds, albeit more weakly. Moreover, the MAP kinase family member ERK5 (also called BMK1) is activated by osmotic and oxidative stresses (Abe et al., 1996). The development of specific inhibitors for each of these MAP kinase family members would greatly facilitate the elucidation of their physiological roles.
Materials and methods
SB 203580 and SB 202190 were generous gifts from Dr John Lee (SmithKline Beecham, King of Prussia, PA, USA) and dissolved in DMSO. Anti‐MKK6 antibodies (raised against the peptide CNPGLKEAFEQPQTS corresponding to a sequence near the N‐terminus of human MKK6) were generated and purified as described previously (Cuenda et al., 1996; Meier et al., 1996). MalE–Mpk2, the Xenopus homologue of SAPK2a (Rouse et al., 1994) and rat GST–SAPK3 (Cuenda et al., 1997) were expressed in Escherichia coli and purified as described previously. Sources of all other materials, enzymes, fusion proteins and methods are given in Cuenda et al. (1997).
cDNA cloning and sequencing
The NotI–EcoRI insert from EST clone 156272 (DDBJ/EMBL/GenBank accession number R72662), which encodes a portion of SAPK2b (p38β, Jiang et al., 1996), was used as the probe to screen a human pituitary gland cDNA library (Clontech, Palo Alto, CA) at high‐stringency. Several partial ERK5 clones, a number of SAPK2b clones, some of which were full‐length, and one partial SAPK4 clone were obtained after screening 2×106 phage. Following sequencing, the SAPK4 insert was used as the probe to screen the pituitary gland cDNA library under high‐stringency conditions. A total of 16 hybridization‐positive clones were obtained after screening 3×106 plaques; they were isolated, the EcoRI inserts subcloned into M13mp18 and sequenced. Three of the 16 clones were full‐length. Sequencing was performed both manually using synthetic oligonucleotides as primers and on an Applied Biosystems 377 DNA sequencer with fluorescent primers. Full‐length sequence was compiled from both strands of cDNA clone hSAPK412. The NCBI sequence databases were searched using the blast algorithm ( ). A multiple alignment of SAPK4, SAPK3, SAPK2a, SAPK2b and HOG1 was built up by eye.
RNA blot analysis
RNA blots were performed using human multiple tissue Northern blots from Clontech with 2 μg poly(A)+ RNA per lane. Probes were labelled with [32P]dCTP by random priming and hybridized under high‐stringency conditions. The SAPK4 probe was prepared from the gel‐purified insert of cDNA clone hSAPK412. The human β‐actin probe was purchased from Clontech.
SAPK2b, SAPK3 and SAPK4 expression plasmids
For bacterial expression, the open reading frames of human cDNA clones hSAPK2B2, hSAPK32 and hSAPK412 were amplified by PCR and subcloned as EcoRI fragments into expression vector pGEX4T‐1 (Pharmacia), followed by transformation into E.coli strain BL21(DE3). The transformed bacteria were grown to an absorbance of 0.6 at 600 nm and induced with 0.4 mM isopropyl‐1‐thio‐β‐galactopyranoside (IPTG). Human GST–SAPK2b, GST–SAPK3 and GST–SAPK4 were purified by affinity chromatography on glutathione‐agarose. For expression of c–myc epitope‐tagged SAPK4, PCR was used to introduce the nucleotide sequence encoding the amino acid sequence MEQKLISEEDLN at the carboxy‐terminus of SAPK4, followed by a stop codon. The resulting EcoRI–NotI fragment was subcloned into the mammalian expression vector pcDNA3.1 (Invitrogen). Substitution of Asp168 by Ala in SAPK4 to produce a kinase‐inactive mutant was performed by site‐directed mutagenesis. PCR fragments were verified by DNA sequencing. Transfections into COS1 cells were carried out as described (Cuenda et al., 1997).
Protein kinase assays
SAPK4, SAPK2a and SAPK2b were assayed routinely by the phosphorylation of MBP, exactly as described previously for SAPK3 (Cuenda et al., 1997). One unit of activity was the amount of enzyme which incorporated 1 nmol of phosphate into MBP in 1 min. SAPK4 activators were assayed by their ability to activate GST–SAPK4, as described previously for GST–SAPK3 (Cuenda et al., 1997). Control experiments were carried out in which GST–SAPK4 was omitted. One unit of SAPK4 activator was the amount which increased SAPK4 activity by 1 U/min. MAPKAP‐K2 and MAPKAP‐K3 were assayed using the peptide KKLNRTLSVA as substrate (Rouse et al., 1994) and one unit of activity was the amount which catalysed the phosphorylation of 1 nmol of peptide substrate in 1 min. SAPK2a activators were assayed by their ability to activate the Xenopus laevis homologue of SAPK2a (Meier et al., 1996). Activated SAPK2a itself was measured by the activation of GST–MAPKAP‐K2(46–400). 0.1% DMSO was present in all experiments where the effects of SB 203580 and SB 202190 were being studied. This concentration of DMSO inhibited each SAPK by ∼10%.
Immunoprecipitation of SAPK4
Lysates of cells transfected with myc epitope‐tagged SAPK4 were centrifuged at 4°C for 10 min at 13 000 g. Aliquots of the supernatant (100 μg protein) were incubated for 2 h on a shaking platform with 5 μl of protein G–Sepharose coupled to 3 μg of monoclonal antibody 9E10 which recognizes the c‐myc epitope. The immunoprecipitates were washed and assayed as described (Cuenda et al., 1997).
We are grateful to Dr C.Marshall for providing GST–MKK1, Dr J.Woodgett for GST–MKK4, Dr A.Nebreda for MalE‐MEKK and a bacterial expression construct encoding human SKK3 and Dr A.Ashworth for the DNA expressing the C‐terminal kinase domain of MEK kinase. We thank J.Hasegawa for help in cDNA library screening and R.MacKintosh and N.Morrice for help in purifying GST‐fusion proteins. This work was supported by the UK Medical Research Council (to M.G. and P.C.) and by the Royal Society (to P.C.).
↵† M.Goedert and A.Cuenda contributed equally to this work
- Copyright © 1997 European Molecular Biology Organization