Cloning and characterization of MST4, a novel Ste20-like kinase.

MST4, a novel member of the germinal center kinase subfamily of human Ste20-like kinases, was cloned and characterized. Composed of a C-terminal regulatory domain and an N-terminal kinase domain, MST4 is most closely related to mammalian Ste20 kinase family member MST3. Both the kinase and C-terminal regulatory domains of MST4 are required for full activation of the kinase. Northern blot analysis indicates that MST4 is ubiquitously distributed, and the MST4 gene is localized to chromosome Xq26, a disease-rich region, by fluorescence in situ hybridization. Although some members of the MST4 family function as upstream regulators of mitogen-activated protein kinase cascades, expression of MST4 in 293 cells was not sufficient to activate or potentiate extracellular signal-regulated kinase, c-Jun N-terminal kinase, or p38 kinase. An alternatively spliced isoform of MST4 (MST4a) was isolated by yeast two-hybrid interaction with the catalytic domain of Raf from a human fetal brain cDNA library and also found in a variety of human fetal and adult tissues. MST4a lacks an exon encoding kinase subdomains IX-XI that stabilizes substrate binding. The existence of both MST4 isoforms suggests that the MST4 kinase activity is highly regulated, and MST4a may function as a dominant-negative regulator of the MST4 kinase.

MST4, a novel member of the germinal center kinase subfamily of human Ste20-like kinases, was cloned and characterized. Composed of a C-terminal regulatory domain and an N-terminal kinase domain, MST4 is most closely related to mammalian Ste20 kinase family member MST3. Both the kinase and C-terminal regulatory domains of MST4 are required for full activation of the kinase. Northern blot analysis indicates that MST4 is ubiquitously distributed, and the MST4 gene is localized to chromosome Xq26, a disease-rich region, by fluorescence in situ hybridization. Although some members of the MST4 family function as upstream regulators of mitogen-activated protein kinase cascades, expression of MST4 in 293 cells was not sufficient to activate or potentiate extracellular signal-regulated kinase, c-Jun N-terminal kinase, or p38 kinase. An alternatively spliced isoform of MST4 (MST4a) was isolated by yeast twohybrid interaction with the catalytic domain of Raf from a human fetal brain cDNA library and also found in a variety of human fetal and adult tissues. MST4a lacks an exon encoding kinase subdomains IX-XI that stabilizes substrate binding. The existence of both MST4 isoforms suggests that the MST4 kinase activity is highly regulated, and MST4a may function as a dominant-negative regulator of the MST4 kinase.
Mitogen-activated protein kinases (MAPKs) 1 are highly conserved mediators of signal transduction that are present in all eukaryotes and play essential roles in regulating cell differentiation, cell proliferation, and cell death. The core of the cascade consists of three sequentially acting protein kinases: a MAPK, a MAPK kinase, and a MAPK kinase kinase (1). In yeast, there are five MAPK cascades that have pivotal roles in regulating sporulation, cell wall remodeling, osmolyte synthesis, filamentation, and mating (2,3). In mammalian systems, the most thoroughly studied MAPKs are ERK1/2, JNK, and p38 MAPK modules. The ERK1/2 cascade consists of Raf as the MAPK kinase kinase, MEK as the MAPK kinase, and ERK1/2 as the MAPK. This cascade is involved in regulation of cell growth and differentiation, whereas the JNK and p38 cascades have been implicated in several stress responses including apoptosis and inflammation (4 -9).
In budding yeast, the serine/threonine kinase Ste20 activates the MAPK kinase kinase Ste11, an upstream activator of the MAPK kinase Ste7. Ste7 in turn activates two MAPKs, FUS and KSS1, in the mating pathway of budding yeast (10,11). The homologous family of mammalian Ste20-like kinases can be divided into two subfamilies based on their regulation and structure and is rapidly growing. p21 Rac/Cdc42-activated kinases (PAKs) (12) represent the first subfamily of kinases, containing a C-terminal Ste20-like kinase domain and a Nterminal regulatory domain with Cdc42 and Rac binding regions. Like Ste20, PAKs can be activated by binding to the GTP-bound form of Rac/Cdc42 (12)(13)(14)(15). In cotransfection experiments, constitutively active mutants of PAK1 or PAK3 can activate the JNK and p38 pathways (14,16). PAKs have also recently been shown to directly phosphorylate c-Raf at serine 338, enhancing Raf activity (17), and to potentiate activation of ERKs (18). Thus, PAKs can act as upstream activators of the MAPK cascades in mammalian cells.
In the present study, using a screen for Raf-interacting proteins, we cloned and characterized MST4, a new member of the mammalian Ste20 family that has a strong homology to mammalian Ste20-like kinases MST1, MST2, MST3, and SOK1. The existence of a new member of the MST family, as well as an alternatively spliced, inactive isoform, suggests that these enzymes may play discrete and specific roles in regulating MAPK signaling pathways.

MATERIALS AND METHODS
Mutagenesis of PEG202-c-Raf-The catalytic domain of Raf was amplified by PCR with primer 1 (5Ј-gacggatccgttcacagccgaaaacccccgtgcc-3Ј) and primer 2 (5Ј-cgaccatggctagaagacaggcagcctcgg-3Ј), digested with XhoI and BamHI, and subcloned into PEG202 to construct PEG202-c-Raf. MEK1 was amplified by PCR with primer 3 (5Ј-gcctcgagcccaagaagaagccgacg-3Ј) and primer 4 (5Ј-gcctcgagtcagatgctggcagcgtg-3Ј), digested with XhoI, and subcloned into PJG4 -5 to construct PJG-4 -5-MEK. The PEG202-Raf was transformed into Epicurian coli XL1-red cells (Stratagene), which are deficient in three of the primary DNA repair pathways. After recovery in SOC medium (GIBLO BRL) at 37°C for 1 h, the culture was diluted to 10 ml of SOC for further incubation overnight. The next day, 200 l of the culture was diluted with a fresh 10 ml of SOC and grown overnight, while the remaining culture was used for plasmid DNA preparation. The second overnight culture was further diluted for another overnight incubation and plasmid DNA preparation. A third overnight culture was also used for preparation of plasmid DNA. These three plasmid DNA preparations were pooled together as the mutated PEG202-c-Raf library, which was transformed into yeast EGY48 along with PJG-MEK in order to screen for a mutated PEG202-Raf protein that has a high affinity for MEK. A mutant construct was isolated (PEG202-Raf-M), and we confirmed that the mutant Raf protein had an enhanced affinity for MEK by retransforming the PEG202-c-Raf-M construct back into yeast EGY48 with PJG4 -5-MEK and testing its interaction with MEK. Sequencing the c-Raf insertion in PEG202 revealed a mutation at residue Ser 610 that introduced a new stop codon at this site, resulting in a truncated C-terminal Raf protein.
Yeast Two-hybrid Screen-The LexA yeast two-hybrid system was kindly provided by Dr. Roger Brent. A human fetal library, made from a 22-week-old human fetal frontal cortex, was used to search for proteins interacting with the kinase domain of mutant PEG202-Raf-M (see above). Two clones showing strong interaction with Raf were obtained. Routine yeast work and yeast transformation were performed as described (36).
PCR Analysis of cDNA Libraries-cDNA from two human fetal brain libraries, a human testes library, and multiple human tissue cDNA panels from CLONTECH were analyzed by polymerase chain reaction using primer MST-761 (5Ј-ctactaagattccgaatcagagccctc-3Ј) and primer MST-1 (5Ј-atggcccactcgccggtggctgtc-3Ј).
Tissue Northern Blot-A ϳ1-kilobase pair DNA fragment was amplified by PCR, purified using a Qiagen PCR purification kit, and labeled with 32 P using the Megaprime DNA labeling kit (Amersham Pharmacia Biotech). The probe was hybridized to a poly(A) ϩ RNA human multiple tissue Northern blot (CLONTECH) following the user manual.
Site-directed Mutagenesis-A primer (5Ј-gtcgttgctattcgaatcatagac-3Ј) was used for unique site elimination site-directed mutagenesis of MST4a to generate the kinase inactive mutant MST4a-K53R. The critical lysine in the kinase domain (VAIKIID) was mutated to arginine. The mutagenesis was performed according to the manual for the Amersham Pharmacia Biotech U.S.E. mutagenesis kit. The smaller BamHI/BsmI DNA fragment from PCR3.1-MST4a-K53R was ligated into the larger BamHI/BsmI DNA fragment from PCR3.1-MST4 to construct PCR3.1-MST4-K53R.
Epitope Tagging-Primer FLAG-MST-5Ј (5Ј-gccgccatggactacaaggacgacgatgacaaggcccactcgccggtggctg-3Ј) and primer MST-3Ј-1165 (5Ј-ccatcgatcatggagctcatgggttaag-3Ј) were used to clone the MST4a from PJG-MST4 by PCR and clone MST4 from Marathon-ready TM cDNA by PCR with a CLONTECH PCR Advantage 2 kit. The PCR products were purified by the Qiagen PCR purification kit and subcloned into the TA cloning vector PCR3.1 (Invitrogen) to construct PCR3.1-MST4a and PCR3.1-MST4. Primer FLAG-MST-5Ј and primer MST-761 (5Ј-ctactaagattccgaatcagagccctc-3Ј) were used to clone the MST4-NT from Placenta Marathon-ready cDNA by PCR. The PCR products were ligated into the TA cloning vector PCR3.1 (Invitrogen) to construct the Nterminal construct PCR3.1-MST4-NT. The FLAG-MST-5Ј primer contained the FLAG epitope right after the ATG code and the 19 nucleotides in the 5Ј-end of MST4. The plasmids were sequenced by the interdisciplinary center for biotechnology research sequencing core laboratory at the University of Florida or by the University of Chicago Cancer Research Center DNA sequencing facility.
Immunoprecipitation and in Vitro Kinase Assay-For MST4, FLAG-MST4a, FLAG-MST4, FLAG-MST4-K53R, or FLAG-MST4-NT were transfected into cells, grown in regular medium for 24 h, and starved in serum-free medium or grown in regular medium for 48 h. Following lysis in 1% Triton X-100 buffer, the cell extracts were incubated with protein G beads coated with anti-FLAG M5 antibody for 3 h. The immune complex was washed three times with lysis buffer and two times with kinase buffer (20 mM Hepes, pH 7.4., 10 mM MgCl 2, 1 mM MnCl 2 , 1 mM dithiothreitol, 0.2 mM sodium vanadate, 10 mM -nitrophenyl phosphate). 5 g of MBP substrate was used per reaction in kinase buffer containing 5 Ci of [␥-32 P]ATP. The kinase reaction was incubated for 25 min at 30°C and then boiled with sample buffer for 5 min. The reaction products were separated on 12% SDS-polyacrylamide gels and transferred to a nitrocellulose membrane. Western blot analysis was performed as described previously (37). For MAPK assays, either HA-ERK2, HA-JNK, or HA-p38 was expressed in cells along with FLAG-MST4 expression vectors and/or Myc-Raf-1 expression vectors as indicated and immunoprecipitated with anti-HA antibodies as above. Kinase activities were monitored by immunoblotting with the appropriate anti-phospho-pTXpY antibodies (New England Biolabs, Boston).
Fluorescence in Situ Chromosomal Hybridization-Human metaphase cells were prepared from phytohemagglutinin-stimulated peripheral blood lymphocytes. The MST4 probe was a cDNA probe containing full-length MST4a. Fluorescence in situ hybridization was performed as described previously (38). Biotin-labeled probes were prepared by nick translation using Bio-16-dUTP (Enzo Diagnostics). Hybridization was detected with fluorescein-conjugated avidin (Vector Laboratories), and chromosomes were identified by staining with 4,6diamidino-2-phenylindole dihydrochloride.

Molecular Cloning of MST4 and MST4a, New Members of the
Mammalian Ste20-like Kinase Family-A human fetal brain library was employed in a LexA-based yeast two-hybrid screen to identify proteins that interact with Raf. A modified Raf bait was created by mutagenizing the yeast plasmid PEG202 containing the catalytic domain of Raf and selecting for enhanced interaction with MEK1 (see "Materials and Methods"). Using the modified Raf bait plasmid, we obtained two cDNA clones that coded for the same previously uncharacterized cDNA. Sequence analysis showed that both clones encoded a polypeptide of 354 amino acids, which was later termed MST4a (Fig. 1).
Eukaryotic protein kinases have a common catalytic core structure in their kinase domain, which typically contains 11 conserved subdomains (39). Comparison of the MST4a protein sequence with that of MST family members revealed that MST4a lacked part of the kinase domain. In order to see if MST4a might be a splicing variant of a full-length MST cDNA, the kinase subdomains I-XI in MST4 cDNAs from three separate libraries were examined by PCR analysis. As shown in Fig.  2 (left panel), two PCR products with different sizes were amplified from a fetal human brain (CLONTECH) cDNA library that was different from the human brain cDNA library used in the original yeast two-hybrid screen as well as from a testes and a placenta library. Sequence analysis indicated that the smaller PCR product had the same sequence as the kinase domain of MST4a, and the larger PCR product had an additional cDNA insert. Subcloning the full-length fragments into an expression vector and subsequent sequencing revealed that the larger PCR product corresponded to the full-length MST cDNA containing the missing IX, X, and XI kinase domains (MST4, Fig. 1). The deduced protein sequence of MST4, consisting of 416 amino acids (Fig. 3A), has a kinase domain at the N terminus and a regulatory domain at the C terminus. Comparison of the protein sequence of MST4 with other enzyme sequences (Fig. 3B) indicated that the kinase domain is most closely related to the catalytic domains of Mammalian Ste20like kinases MST1 (51% identity), MST2 (54% identity), MST3 (65% identity), and SOK1 (63% identity). The C-terminal regulatory domain of MST4 is most similar to the C-terminal domains of MST3 and SOK1.
Most of the genomic DNA sequence of human MST4 has been mapped. Interestingly, there is a putative 186-base pair exon corresponding to kinase subdomains IX-XI that, if spliced out of the MST4 mRNA transcript, could account for the sequence found in the truncated MST4a variant. These results suggest that MST4a is likely to be an alternatively spliced form of MST4. When we did a more comprehensive analysis of the tissue distribution of MST4 and MST4a by PCR, we were able to detect both forms in cDNAs from a variety of other adult and fetal human tissues (CLONTECH) including kidney, lung, liver, and pancreas as well as fetal thymus, spleen, muscle, liver, and kidney (Fig. 2, right panel). In most tissues, the major expressed form is MST4, but it appears that MST4a is more highly expressed in the brain.
Tissue Distribution and Chromosomal Localization of MST4 -A human multiple tissue Northern blot was used to identify the expression pattern of MST4. Only one mRNA band with a size of 3.6 kilobases was detected. The results showed that MST4 is ubiquitously expressed, with strong expression in placenta, weak expression in skeletal muscle and pancreas, and moderate expression in brain, heart, lung, liver, muscle, and kidney. It was not possible to determine the difference in expression of MST4 versus MST4a by this approach, since the difference in size between the two mRNAs is too small to be detected by Northern analysis (Fig. 4). To map the MST4 gene, we performed fluorescence in situ hybridization using a biotinlabeled MST4 probe on normal human metaphase chromosomes. Hybridization of the MST4 cDNA probe resulted in specific labeling only of the X chromosome (Fig. 5). Labeling of Xq25-27 was observed on four (eight cells), three (16 cells), or two (one cell) chromatids of the X chromosome homologues in 25 cells examined from mitogen-stimulated lymphocytes isolated from a healthy female. Of 82 signals observed, one signal (1.2%) was located at Xq25, 67 signals (82%) were located at Xq26, and 14 signals (17%) were located at Xq27. No background signals were observed at other chromosomal sites. We also observed a specific signal at Xq26 in an additional hybridization experiment using this probe (data not shown). These results indicate that the MST4 gene is localized to Xq26. This result was later verified by examination of the relative location of the MST4 gene in the X chromosome genomic map. Catalytic Activity of MST4 and MST4a-In order to determine whether MST4 or its variant is an active kinase, the N-terminal domains of MST4 and MST4a were tagged with a FLAG epitope, and the cDNAs were expressed in 293 cells using a cytomegalovirus promoter (Fig. 6A). To control for nonspecific kinase activity, a kinase-inactive mutant of MST4, MST4-K53R, was also transiently transfected into 293 cells. Following transfection, 293 cells were lysed, and the cell lysates were subjected to immunoprecipitation with anti-FLAG antibody (Fig. 6B). The immunoprecipitates were then assayed for in vitro kinase assay using myelin basic protein as a substrate. The results indicate that MST4 is an active kinase, whereas MST4a did not exhibit kinase activity (Fig. 6B). Interestingly, the kinase activity of MST4 isolated from cells grown in serum-free or serum-containing medium was comparable, indicating that MST4 possesses a high basal kinase activity (Fig. 6C). Under the same conditions, a phosphoprotein with the same size as MST4 (46 kDa) was also detected in an in vitro kinase reaction when immunoprecipitates containing MST4 but not kinase-inactive MST4 (MST4-K53R) were used, suggesting that MST4 also functions as an autophosphorylating kinase (Fig. 6C).
Since MST4a was isolated by association with the Raf kinase domain in a yeast two-hybrid system, we determined whether MST4 or MST4a associate with c-Raf-1 in cells. Therefore, COS cells were co-transfected with expression vectors for Myc-Raf-1 and FLAG-MST4 or FLAG-MST4a. Raf was then immunoprecipitated with an anti-Myc antibody and analyzed for MST4 association by immunoblotting with an anti-FLAG antibody. Conversely, MST4 or MST4a was immunoprecipitated with an anti-FLAG antibody and analyzed for Raf association with an anti-Myc antibody. As shown in Fig. 6D, no association of c-Raf-1 with MST4 or MST4a was observed. We also determined whether co-expression of Myc-Raf-1 with FLAG-MST4 altered MST4 kinase activity. Consistent with the co-immunoprecipitation results, no effect of Raf-1 on MST4 kinase activity was detected (data not shown). These results suggest that MST4 is not stably associated with or modulated by c-Raf-1.
C-terminal Regulatory Domain of MST4 Enhances but Is Not Essential for Its Kinase Activity-Removal of the C-terminal Left panel, the PCR products were amplified from placenta and testis Marathon ready cDNA and fetal brain cDNA. Right panel, PCR products were amplified from CLONTECH multiple human tissue cDNA panels. Samples were subjected to electrophoresis in a 1.2% agarose gel as described under "Materials and Methods." regulatory domain of MST1 and MST2 by caspases results in a significant increase in MST1 or MST2 kinase activity (35,40). To address the function of the C-terminal tail of MST4, a cDNA encoding the N-terminal kinase domain of MST4 but lacking the C terminus of the protein (FLAG-MST4-NT) was transfected into COS cells. For comparison, COS cells were also transfected with expression vectors for FLAG-MST4 or FLAG-MST4-K53R. Following immunoprecipitation with anti-FLAG antibodies, the MST4 proteins were assayed for kinase activity using MBP as a substrate. As shown in Fig. 7, FLAG-MST4-NT has reduced kinase activity compared with MST4, indicating that the C-terminal domain acts to enhance MST4 activity. Interestingly, MST4-NT is no longer autophosphorylated, suggesting that the site of autophosphorylation may be in the C-terminal tail of the protein (Fig. 7).
MST4 Does Not Activate ERK2 in 293 Cells-MST4a was isolated by interaction with the Raf kinase domain and is most mammalian Ste20-like kinases. A, cDNA and predicted protein sequence of the full-length MST4. B, sequence comparison between MST4 and other members of the MST family. The predicted amino acid sequence of MST4 was compared with that of the mammalian Ste20-like kinases MST1, MST2, MST3, and SOK1 by Geneworks. Amino acids conserved in all protein are shaded.  5. In situ hybridization of a biotin-labeled MST4 probe to human metaphase cells from phytohemagglutinin-stimulated peripheral blood lymphocytes. The X chromosome homologues are identified with arrows; specific labeling was observed at Xq26. The inset shows partial karyotypes of two X chromosome homologues illustrating specific labeling at Xq26 (arrows). Images were obtained using a Zeiss Axiophot microscope coupled to a cooled charge-coupled device (CCD) camera. Separate images of 4,6-diamidino-2-phenylindole dihydrochloride-stained chromosomes and the hybridization signal were captured and merged using image analysis software (IP LabSpectrum). closely related to MST3, an enzyme that has been reported by one group to potentiate the Raf/MEK/ERK signaling cascade.
To determine whether overexpression of MST4 might also activate ERK1 or ERK2, vectors expressing MST4, MST4-K53R or control vector were co-transfected with HA-ERK2 into 293 cells. The tagged ERK2 was immunoprecipitated from the lysates of transfected cells using an anti-HA antibody, and its kinase activity was assayed by Western blotting with antiphospho-pTEpY MAPK antibody. A sample from each lysate was assayed directly by Western blotting with anti-FLAG antibody to confirm the expression of MST4 (data not shown). As shown in Fig. 8, overexpression of MST4 does not significantly affect the activation of ERK2 in cells grown either in serum or under serum-free conditions. Furthermore, we have also conducted similar experiments involving co-expression of MST4 (or MST4a or MST4-K53R), ERK2, and Myc-Raf-1 to determine whether MST4 can potentiate or inhibit Raf activation of ERK. Again, we could not detect any effect of MST4 on Raf-mediated ERK activity (data not shown), consistent with the lack of association with c-Raf-1 in cells. Thus, despite the interaction of MST4a with the kinase domain of Raf in the yeast twohybrid system, we have no evidence to date that MST4 either stimulates or potentiates ERK2 activity in 293 cells.
Similar experiments were done to determine whether MST4 activates or enhances JNK or p38, the other members of the MAPK superfamily. Thus, FLAG-MST4, the kinase-dead mutant FLAG-MST4, or the control vector were expressed along with HA-JNK in 293 cells. The cells were either untreated or or without serum. 293 cells were transfected with FLAG-MST4, FLAG-MST4-K53R, or control vector and grown in serum-containing medium for 24 h. Cells were then transferred to serum-free medium (SϪ) or serum-containing medium (Sϩ) for another 24 h, and the cell lysates were immunoprecipitated with anti-FLAG antibody and analyzed by Western blotting with anti-FLAG antibody. Duplicate samples were subjected to an in vitro kinase reaction using MBP as substrate. D, co-immunoprecipitation of MST4 and c-Raf-1. COS cells were transfected with expression vectors for FLAG-MST4, FLAG-MST4a, Myc-Raf-1 or green fluorescent protein as a control vector. Lysates were collected after 48 h post-transfection, and proteins were immunoprecipitated with either anti-FLAG or anti-Myc antibodies. Samples were then analyzed by Western blotting with either anti-FLAG or anti-Myc antibodies as indicated. stimulated with sorbitol, and then JNK was immunoprecipitated from the cell lysates with anti-HA antibody. As illustrated in Fig. 9, JNK was stimulated by sorbitol, and neither MST4 nor the kinase-dead MST4 mutant significantly activated or suppressed JNK activity. In a parallel experiment, 293 cells were transfected with expression vectors for FLAG-MST4, the kinase-dead mutant FLAG-MST4, or the control vector along with HA-p38. After further incubation of the cells or stimulation with sorbitol, p38 was immunoprecipitated from the cell lysates with anti-HA antibody. Like the results obtained for ERK and JNK, no effect of MST4 or the kinase-dead MST4 mutant on either p38 stimulation, potentiation, or inhibition was observed (Fig. 10). DISCUSSION A novel member of mammalian Ste20-like kinase family, MST4, was cloned and characterized. According to the primary sequence, MST4 belongs to the germinal center kinase subfamily because it has an N-terminal kinase domain and a Cterminal regulatory domain. MST4 is most highly related to MST3 (32), with 68% identity, and SOK (28), with 63% identity within the kinase domain. An inactive mutant of MST4, termed MST4a, was isolated by interaction with the Raf catalytic domain in a yeast two-hybrid screen. Although Ste20-related kinases activate the MAPK cascade in organisms ranging from yeast to mammals, MST4 does not appear to activate or potentiate the ERK, JNK, or p38 MAPK in 293 cells.
Analysis of the genomic sequence indicates that MST4a is an alternatively spliced isoform of MST4. MST4a is missing a 186-base pair exon encoding kinase subdomains IX, X, and XI. Kinase subdomain IX has been shown to be important for the kinase activity of protein kinase C␣, and it has been suggested that domains X and XI function to stabilize the kinase-bound substrate (39). Consistent with this possibility, the expressed MST4a kinase does not exhibit any kinase activity toward MBP, a potent substrate of MST4 as well as other members of the MST family. Since MST4a was independently cloned from two different brain cDNA libraries at reasonably high expression levels and MST4a transcripts can be translated into protein, it is likely that MST4a represents an alternative isoform of MST4 in the brain as well as other tissues.
The lack of kinase activity in MST4a suggests that it may function as a decoy or dominant-negative kinase, representing an alternative mechanism for regulating MAPK signaling cascades. Several other kinases and receptors appear to have similar splice variants. A 3Ј exon encoding 10 amino acids corresponding to subdomains IX and X of MEK5 can be spliced out during transcription of MEK5 (41), and an alternatively spliced exon also has been found in the comparable region of JNK/stress-activated protein kinase. (42). Unlike MST4a, the MEK5 isoform that lacks the exon encoding subdomains IX and X has equivalent in vitro protein kinase activity to the fulllength MEK5. Thus, it is possible that a large deletion of subdomains IX-XI, like the spliced exon in MTS4a, could lead to loss of activity, while a smaller deletion of subdomains IX-X, like the spliced exon in MEK5, still enables the kinase to interact with the substrate. Recently, a novel Ste20 kinase from chickens was cloned (KFC) that is also expressed as a splice variant (43). The short form of KFC has a 69-amino acid deletion outside of the kinase domain and has no impact on cell growth, but the longer version confers a growth advantage to the cell. Thus, the alternative splicing of exons may provide another mechanism for diversifying the function of protein kinases.
There are other examples of nonfunctional, decoy proteins that have important physiological functions. Recently, a novel phosphoserine/threonine-binding protein (STYX), related to dual specificity protein-tyrosine phosphatase, was shown to contain a naturally occurring Gly instead of Cys, the residue that is required and conserved in dual specificity proteintyrosine phosphatase catalytic loops. The naturally occurring STYX protein does not have phosphatase activity toward FIG. 8. Overexpression of MST4 does not activate, potentiate or inhibit the ERK pathway in 293 cells. 293 cells were co-transfected with either control vector or HA-ERK and either control vector, FLAG-MST4, or FLAG-MST4-K53R. Cells were growth in serum-containing medium for 24 h and then incubated in serum-free medium for another 24 h. The cells were then left untreated (ϪSer) or stimulated with 20% serum (ϩSer). Following treatment, cell lysates were immunoprecipitated with anti-HA antibody, resolved by SDS-polyacrylamide gel electrophoresis (10%), and immunoblotted with anti-HA antibody. The blot was then stripped and reprobed with anti-phospho-pTEpY-ERK antibody. Tyr(P)-containing substrates; however, the substitution of Gly to Cys in recombinant STYX protein confers phosphatase activity, indicating that STYX is a naturally occurring "dominant negative" phosphotyrosine/serine/threonine-binding protein (44). Another example is the TNF receptor superfamily that has two main subgroups of receptors. The first group (DR) possess the death domain, which couples the receptor to caspase cascade, whereas the second group consists of a decoy receptor (DcR), which is structurally related to DR but lacks the death domain and functions as an inhibitor (45). Taken together, these studies suggest that MST4a could be a naturally occurring dominant-negative kinase that can act as an inhibitor of the signaling pathway regulated by MST4.
Although other MSTs have been implicated as mediators of apoptosis, it appears unlikely that MST4 plays a similar role. Interestingly, MST4 has a sequence (DESDS) that is very similar to the caspase 3 recognition sequence in MST1 (DEMDS) and MST2 (DELDS) at the junction of the N-terminal catalytic and C-terminal domains. However, unlike MST1 and MST2 (35,40), MST4 was not cleaved in vivo or in vitro by caspase 3 (data not shown). Loss of the C-terminal regulatory domain in MST1 and MST2 results in an activation of the catalytic activity, consistent with a positive feedback role in apoptosis. In contrast, the C-terminal tail of MST4 has an activating regulatory role, and loss of this domain by caspase cleavage would result in decreased kinase activity. Thus, even if MST4 were a target of cleavage by other caspases, the truncated enzyme would not function as a direct downstream mediator of caspaseinitiated cell death. Further studies will be required to elucidate its true physiological role.
The role of MST4 in the activation or regulation of MAPK is also unclear at this time. Unlike other MSTs, MST4 was not sufficient by itself to activate ERKs, JNKs, or p38s or to potentiate or inhibit serum or sorbitol stimulation of these enzymes in 293 cells. However, it is possible that MST4 can modulate MAPK signaling in a tissue-or cell-specific manner. Our isolation of MST4 via a Raf kinase domain-based two-hybrid screen and the relationship of MST4 to MST3, which has been implicated in ERK signaling in one report (33), suggest a role for MST4 in the Raf-activated MAPK cascade. However, in a different report, no ERK activation or potentiation by MST3 was observed (32), and we have similarly failed to observe an interaction of MST3 with ERKs. Thus, the relationship of both MST3 and MST4 to the ERK kinase cascade must be considered unresolved. Although much remains to be learned, the identification of yet another member of the mammalian Ste20 family suggests that these enzymes may play an important role in signaling specificity.
One potential approach toward elucidating MST4 function is genetic analysis. MST4 has been mapped to a region of the chromosome implicated in many disease states, particularly those involved in mental retardation. With the complete elucidation of the human genomic map, it should now be possible to determine whether MST4 plays a role in the etiology of any of these brain-related diseases.