MRK, a mixed lineage kinase-related molecule that plays a role in gamma-radiation-induced cell cycle arrest.

Mitogen-activated protein (MAP) kinase pathways are three-kinase modules that mediate diverse cellular processes and have been highly conserved among eukaryotes. By using a functional complementation screen in yeast, we have identified a human MAP kinase kinase kinase (MAPKKK) that shares homology with members of the mixed lineage kinase (MLK) family and therefore was called MRK (MLK-related kinase). We report the structure of the MRK gene, from which are generated two splice forms of MRK, MRK-alpha and MRK-beta, encoding for proteins of 800 and 456 amino acids, respectively. By using a combination of solid phase protein kinase assays, transient transfections in cells, and analysis of endogenous proteins in stably transfected Madin-Darby canine kidney cells, we found that MRK-beta preferentially activates ERK6/p38gamma via MKK3/MKK6 and JNK through MKK4/MKK7. We also show that expression of wild type MRK increases the cell population in the G(2)/M phase of the cell cycle, whereas dominant negative MRK attenuates the G(2) arrest caused by gamma-radiation. In addition, exposure of cells to gamma-radiation induces MRK activity. These data suggest that MRK may mediate gamma-radiation signaling leading to cell cycle arrest and that MRK activity is necessary for the cell cycle checkpoint regulation in cells.

Mitogen-activated protein (MAP) kinase pathways are three-kinase modules that mediate diverse cellular processes and have been highly conserved among eukaryotes. By using a functional complementation screen in yeast, we have identified a human MAP kinase kinase kinase (MAPKKK) that shares homology with members of the mixed lineage kinase (MLK) family and therefore was called MRK (MLK-related kinase). We report the structure of the MRK gene, from which are generated two splice forms of MRK, MRK-␣ and MRK-␤, encoding for proteins of 800 and 456 amino acids, respectively. By using a combination of solid phase protein kinase assays, transient transfections in cells, and analysis of endogenous proteins in stably transfected Madin-Darby canine kidney cells, we found that MRK-␤ preferentially activates ERK6/p38␥ via MKK3/MKK6 and JNK through MKK4/MKK7. We also show that expression of wild type MRK increases the cell population in the G 2 /M phase of the cell cycle, whereas dominant negative MRK attenuates the G 2 arrest caused by ␥-radiation. In addition, exposure of cells to ␥-radiation induces MRK activity. These data suggest that MRK may mediate ␥-radiation signaling leading to cell cycle arrest and that MRK activity is necessary for the cell cycle checkpoint regulation in cells.
In a wide range of organisms, from yeast to mammals, mitogen-activated protein kinase (MAPK) 1 pathways mediate a variety of signals that regulate multiple physiological pro-cesses, including cell proliferation, cell differentiation, and cell death as well as stress-induced responses (1)(2)(3). These MAPK modules consist of distinct cascades of kinases, beginning with a serine/threonine kinase, MAPKKK, which phosphorylates and activates a dual specificity kinase, MAPKK or MEK, that in turn transfers phosphates onto threonine and tyrosine residues of a third enzyme, MAP kinase. The MAP kinase subsequently phosphorylates and activates various transcription factors, among other substrates. In mammals, the best characterized MAPK pathways are defined by the four main classes of MAPK they activate: extracellular signal-regulated protein kinases (ERK-1 and -2), Jun amino-terminal kinases (JNK-1, -2, and -3), p38 proteins (p38␣, -␤, -␥, and -␦), and ERK5 (4). The MAPKKK family consists of at least 14 members that include the MEKK group (MEKK1-4), the mixed lineage kinase group (MLK1-3, DLK, and LZK), the ASK proteins (ASK1 and -2), TAK1, TAO, and Tpl2/Cot. Although members within each group are highly homologous, with identity ranging between 50 and more than 90%, the homology between groups is significantly reduced and is restricted to the kinase domain. The large number of structurally diverse MAPKKKs may reflect tissue specificity or stimulus-specific signaling. Although substantial progress has been made in linking each of the known MAPKKK proteins to specific MAP kinase pathways, their precise contribution has not been clearly defined. For instance, MEKK1-3, DLK, MLK, and Tpl2 have been reported to activate preferentially JNK or ERK, rather than the p38 MAPK (5)(6)(7)(8)(9)(10). Conversely, TAK1, MEKK4, TAO, and ASK1 more effectively activate the p38 pathway (11)(12)(13)(14). The link between individual MAPKKKs and specific upstream control molecules has only been identified for some family members and remains to be firmly established for most. Despite our growing knowledge of the signaling elements involved in each cascade, no upstream MAPKKKs have yet been described for some MAP kinases, such as ERK3 (15) and ERK6/p38␥ (16).
In Saccharomyces cerevisiae, there are five MAPK modules, one of which is the well characterized mating pheromones pathway (17). In this system, Ste11 is the MAPKKK that activates Ste7, the MEK counterpart, which in turn activates the MAPK, Fus3 (18). We and others (19,20) have shown that loss of Ste11 by gene knock out can be functionally complemented in this system by an active mammalian Raf protein and its substrate MEK. In the present study we conducted a functional screen in this system to identify novel components of MAPK pathways, and we discovered a gene that encodes a serine/threonine kinase, designated MRK for MLK-related kinase. Here we describe the characterization of the structure of the MRK gene and the effect of the MRK protein on the known mammalian MAP kinase cascades. We found that MRK expression preferentially activated the ERK6/p38␥ and JNK path-ways, both in transiently transfected and stable cell lines, whereas it had a marginal effect on p38␣ and no significant effect on ERK. The activation of these pathways is accompanied by stimulation of their respective MKKs, in particular MKK3/MKK6 and MKK4. We also report that expression of wild type MRK induces an increase in the G 2 /M cell population. Conversely, ␥-radiation-mediated G 1 and G 2 arrest are decreased in cells expressing the dominant negative allele of MRK. The effect of ␥-radiation is accompanied by activation of endogenous MRK. These findings suggest a role for MRK in the regulation of cell cycle checkpoints.

EXPERIMENTAL PROCEDURES
Plasmids, cDNA Library Constructions, and Genomic DNA Analysis-The 2-m based plasmid pAB23BXN2, containing the URA3 selectable marker, was derived from pAB23-BXN (21) by inserting a new polylinker (SalI, SacI, AatII, and XhoI) between BstXI and NotI and used as the library vector. cDNA synthesis was performed on 5 g of poly(A ϩ ) RNA, isolated from 2 ϫ 10 8 Jurkat cells using the Invitrogen Fast track kit. The first strand was primed with a linker-primer from the ZAP-cDNA synthesis kit (Stratagene) that contains an XhoI site. Protection of this 3Ј-cloning site allowed for unidirectional cloning of the finished cDNA. The 5Ј-cloning site was provided by ligation of a BstXI adaptor. The cDNA library was size-fractionated to collect the portion containing inserts above 500 bp and ligated to the prepared vector. Transformation of Escherichia coli DH10B (Invitrogen) yielded approximately 1 ϫ 10 6 total transformants of which vector religation represented a 3% background.
Catalytically inactive MRK mutants were generated using the Quick-Change, site-directed mutagenesis kit from Stratagene. The following primer was used to convert Lys-45 to Ala: 5Ј-GAGGTGGCTGTCGCGAAG-CTCCTCA-3Ј, generating MRK-␤-K45A. Mutagenesis was conducted as suggested by the manufacturer. pBS-MRK-␤-KT3 was constructed by inserting the KT3 tag after the last codon of MRK, using the following 3Ј end oligo in a PCR: GGATCCAACACCACCACCAGAACCAGAAACATG-AGCGGCCGC.
A human whole blood genomic library (Stratagene) was screened with an MRK kinase domain probe, generated using the following primers: 5Ј-GAGATGTCGTCTCTCGGTGC-3Ј and 5Ј-GTGATC-CATATCCATCTCCTC-3Ј. A High Density CITB Human BAC colony DNA membrane (Research Genetics) was screened with the same probe used for the library and with an upstream probe derived from intron sequences 5Ј of exon 1 of the MRK gene.
5Ј-RACE on full-length mRNA prepared from human SKM cells (Clonetics) was performed with the First Choice RNA ligase-mediated RACE kit (Ambion) according to the manufacturer's protocol. For 3Ј-RACE, total RNA from SKM cells was used to perform first strand synthesis with the Thermoscript RT-PCR System (Invitrogen) at 60°C for 1 h. After ligation of the 3Ј-RACE adapter primer (Invitrogen), RT-PCR was conducted using Expand High Fidelity DNA polymerase (Roche Molecular Biochemicals) in the presence of 1 M betaine (Sigma).
RNA and Protein Analysis-Poly(A ϩ ) RNA from HT1080 cells was prepared with the Invitrogen Fast Track kit. Human Multiple Tissue Northern blots I and II were purchased from CLONTECH. Monoclonal (4-23) and polyclonal  anti-MRK antibodies were generated against recombinant MRK expressed in E. coli. These antibodies were typically used at the 1:1000 dilution. Mammalian cells were lysed in lysis buffer (20 mM Tris-HCl, pH 8.0, 137 mM NaCl, 1 mM EGTA, 1% Triton X-100, 10% glycerol, 1.5 mM MgCl 2 ), supplemented prior to use with 1 mM sodium orthovanadate, 50 mM NaF, protease inhibitor mixture tablets (1 tablet/7 ml) (Roche Molecular Biochemicals). Cells were lysed at 4°C for 20 min, and debris was collected at 14,000 rpm at 4°C. Typically, 500 g of proteins were immunoprecipitated with the appropriate antibodies, either coupled to protein G-Sepharose beads (KT3 antibodies) or together with protein A beads, for 2 h at 4°C. Immune complexes were washed three times with lysis buffer, and the proteins were eluted in Laemmli sample buffer heated for 5 min at 95°C. Samples were processed for Western blot analysis. Kinase reactions were performed on the immune complexes in the presence of 30 mM Tris-HCl, pH 8.0, 20 mM MgCl 2 , 1 mM EDTA, 75 mM NaCl, 1 mM dithiothreitol, 1 mM sodium orthovanadate, 10 M cold ATP, 5 Ci of [␥-32 P]ATP, and the appropriate substrate as described in the figure legends. MKK4, MKK7, and MKK6 recombinant proteins were purchased from Upstate Biotechnology, Inc., GST-Jun from Stratagene, and ATF-2 from Cell Signaling Technologies.
Cell Culture and Transfections-COS-1 cells were transiently transfected by electroporation essentially as described in Porfiri and McCormick (24).
The parental MDCK T23 clone (25), which expresses the tetracyclinerepressible transactivator (26), was used to generate stable lines expressing wild type (pTRE-MRK-␤), catalytically inactive MRK-␤ (pTRE-MRK-␤-K45A), and pTK-Hyg (CLONTECH) which carries the hygromycin resistance gene. Cells were transfected with the Effectene method from Qiagen, following the manufacturer's protocol. Clone selection was carried out in the presence of 200 g/ml hygromycin B (Invitrogen) and 20 ng/ml doxycycline (Sigma). Drug-resistant clones were further tested for expression of the transgenes after removal of doxycycline to induce expression of the recombinant MRK-␤ genes. Generally, maximum expression of the transgenes occurred 48 h after removal of doxycycline from the culture medium. Expression of the recombinant proteins was attenuated at high cell density; therefore, cells were typically cultured at low density prior to analysis. The cells were routinely cultured as described in Jou et al. (27).
Cell Cycle Analysis-Cells were grown for 24 h in the absence of doxycycline to induce the ectopic genes. They were then plated to 40% confluency (2.5-5 ϫ 10 6 /plate) on 60-mm plates. Twenty four hours later, the cells were exposed to 20 Gy ionizing radiation from a 137 Cs source. Eight hours post-irradiation, the cells were trypsinized, centrifuged, and washed twice with phosphate-buffered saline (PBS) supplemented with 1% fetal bovine serum (FBS) and fixed by stepwise addition of 10 volumes of ice-cold 95% ethanol and incubated for 1 h at 4°C. After two washes in PBS, 1% FBS, 1 ϫ 10 6 cells were incubated in 0.5 ml of PBS, 1% FBS containing 1 mg/ml RNase A for 30 min at 37°C and then stained with propidium iodide (0.05 mg/ml). Cells were filtered to remove cell aggregates (Falcon filter top tubes) and analyzed for DNA content by fluorescence-activated cell sorting (FACS) analysis with a FACSCalibur flow cytometer (BD PharMingen). Data were analyzed using the Cell Quest (BD PharMingen) and ModFit (Verity) analysis software.

Cloning of the MRK Gene and Characterization of Its
Genomic Structure-To identify novel signal transduction molecules affecting MAP kinase pathways, we made use of a yeast system in which mammalian Raf and MEK proteins were expressed to complement the yeast Ste11 protein (19,23). Strain SY1984, expressing Raf and MEK (SY1984 R-L M-T), responds to Raf activation by signaling through the yeast MAP kinase, Fus3. In this strain, activation of Fus3 induces transcription of the HIS3 gene off the FUS1 promoter, which leads to growth in the absence of exogenous histidine. Cells were transformed with a human cDNA library prepared from Jurkat cells, and colonies that grew in the absence of histidine were characterized. This screen could in principle identify activators of each member of the MAPK cascade regulating the expression of the HIS gene, as well as genes that could indirectly affect growth on selection media. Therefore, to identify the functional target of each cloned library gene, cDNAs from each colony were isolated and re-tested in four strains as follows: the original strain used in the primary screen (SY1984 R-L M-T); a strain that lacked Raf (SY1984 M-T); one that lacked MEK (SY1984 R-L); and a Ste7 null (SY1943-L21). The screen yielded several genes that activated Raf, MEK, or the yeast MEK, Ste7. Different Ras and 14-3-3 clones were isolated as Raf activators in that they were capable of stimulating growth in the absence of histidine only when Raf was present. Among the MEK activators, MEKK1 and MEKK2 partial clones, encompassing the kinase domains of these genes, were obtained. In addition, two novel genes were isolated as activators of Ste7. They allowed growth on media lacking histidine in the wild type STE7 strain SY1984 but not in the ste7 null strain SY1493-L21. These clones, designated J42 and J207, encoded identical kinase domains but diverged in their 3Ј sequences, suggesting that they might represent splice variants of the same gene. Northern blot analysis using a cDNA probe from the common kinase domain identified a predominant band of 7.5 kb and less abundant forms of 3.8 and 1.6 kb. A specific probe from the 3Ј end of J42 identified the longest transcript as J42 mRNA, whereas a probe from the 3Ј end of J207 recognized the 3.8-kb message (Fig. 1A). The identity of the 1.6-kb species remains to be elucidated. Analysis of mRNA distribution indicated that J42 and J207 are ubiquitously expressed in normal tissues but are most abundant in skeletal muscle and heart (Fig. 1B). Although the signal from brain and kidney tissues is very weak, protein analysis in cells derived from such tissues (see below and data not shown) indicated expression of these genes. As the original J42 cDNA clone was only 2.2 kb long, we screened several cDNA libraries and performed 5Ј-RACE analysis on poly(A) ϩ RNA as well as on 5Ј-capped mRNA (which represents a population of full-length mRNAs), to identify the additional sequences corresponding to the 7.5-kb mRNA. These approaches identified the 5Ј start site of the two major RNA forms located 195 bp upstream of the ATG start codon. Analysis of 2 genomic and 10 BAC clones indicated the presence of a putative promoter within the sequences preceding the 5Ј start site. A DNA fragment encompassing about 500 bp of this region was sufficient to direct expression of a luciferase reporter gene. 2 The identification of the 5Ј end sequences, however, did not account for the difference in size between the mRNA and the isolated cDNA. Analysis of the genomic sequence 3Ј of the J42 STOP codon, obtained from the BAC clones and from partial sequences deposited in the Washington University Genome Sequencing Center Data base, indicated the presence of a polyadenylation site about 5 kb downstream of the 3Ј end of the initial J42 cDNA. This sequence was confirmed to be present in the mRNA by RT-PCR (data not shown). These results indicated that the J42 mRNA has a 3Ј-untranslated region (3Ј-UTR) of about 5.6 kb. A schematic representation of the genomic structure corresponding to the two clones is shown in Fig. 2. The common domains of the two splice forms are encoded by 11 exons, followed by the nearly 6-kb exon encoding the unique J42 carboxyl-terminal region and nine exons encoding the unique J207 carboxyl terminus.
BLAST analysis of the DNA sequences indicated that the common kinase domain is homologous to proteins of the MAP-KKK family members. In particular, it shares 52% similarity with the mixed lineage kinases, MLK2 and MLK1 (28,29), and 47% with TAK1 (14) (Fig. 3). Both of these classes of proteins have been implicated in activation of the stress pathways, JNK and p38 (9, 14, 28, 30 -32). Similarly to the MLK proteins, the J42 and J207 kinase domains are followed by a leucine zipper region, that, although double in the MLK proteins, is a single domain in the J42 and J207 clones. The unique J42 3Ј region 2 E. Gross and R. Ruggieri, unpublished data. encodes a highly acidic domain. The carboxyl terminus of J207, on the other hand, has a sterile ␣-motif domain, which has been implicated in protein-protein interactions (33). The isolated J42 and J207 cDNAs have a coding potential for proteins of calculated molecular masses of 51.5 and 91.1 kDa, respectively. Interestingly, the long mRNA encodes for the shorter of the two proteins. We named the J207 and J42 genes (and their corresponding products) MRK-␣ and -␤, respectively, for MLK Related Kinases. In this paper we focus primarily on the characterization of MRK-␤.
Characterization of the MRK Proteins-To characterize the function of the MRK proteins, we tagged the MRK-␣ and -␤ cDNAs with FLAG and KT3 (34) tags, respectively. When expressed in COS-1 cells, the two cDNAs yielded proteins with SDS-PAGE mobility of about 98 and 55 kDa (Fig. 4A), in agreement with the coding potential of the two cDNA clones. To verify that these clones represented the full-length endogenous proteins, we generated a monoclonal antibody, 4-23, against the MRK common kinase domain. We used this reagent to identify endogenous MRK polypeptides in the human epithelial cell line, MCF-10A. Fig. 4B shows that the anti-MRK antibody identified two bands with SDS-PAGE mobility similar to that of the respective recombinant proteins. The slight electrophoretic retardation observed for recombinant MRK-␤ (see Fig.  4, A and B) when compared with the endogenous protein was attributed to the KT3 tag, as demonstrated by removal of the tag (Fig. 4C).
To verify that MRK-␤ possessed kinase activity, we tested recombinant MRK-␤ for kinase activity when expressed in COS-1 cells. Fig. 4D shows that recombinant MRK-␤, in addition to autophosphorylation, exhibits kinase activity toward the generic substrate, MBP. Both auto-and exogenous substrate phosphorylation depend on the integrity of the MRK kinase domain, as a substitution of alanine for a critical lysine in the active site in the mutant MRK-␤-K45A abolishes these activities (Fig. 4D, lane 3). Thus, the MRK-␤ protein is a functional kinase.
Effect of MRK-␤ on MAPK Pathways-The functional screen used to clone MRK, as well as the kinase domain sequence, suggested that the MRK proteins belong to the MAPKKK family. To test whether MRK affects the known MAP kinase path-ways, we transiently co-expressed MRK-␤ with ERK, JNK, or p38␣ in COS-1 cells. The MAP kinases were immunoprecipitated and tested for kinase activity in vitro against their respective substrates. To assess the relative stimulation of the different MAP kinases by MRK-␤, in each group we compared the effect of known activators and stimuli of each pathway with that of MRK-␤. Fig. 5 shows that MRK-␤ stimulated ERK2 activity by about 2.5-fold over background, whereas Ras and serum activated ERK2 6-and 11-fold, respectively. A slightly greater stimulation of p38␣ was observed, although its extent was still about a third of that obtained with MKK6 or with osmolarity treatment. On the other hand, we found that MRK-␤ activated JNK 10-fold above background, and this level was comparable with that caused by UV treatment and MEKK1.
Because the activation of the p38␣ MAP kinase was weak but reproducible, we asked whether MRK-␤ affected a less well characterized member of the p38 subfamily, ERK6/p38␥. These members of the p38 subfamily have been reported to be differentially activated by several stimuli, such as hypoxia and ␥-radiation (35,36), and respond differently to the inhibitory compound, SB203580 (37). In contrast to p38␣, we observed a 17-fold activation of ERK6/p38␥ by MRK-␤, as reflected by the phosphorylation of ATF-2. This effect was comparable with that obtained with MKK6, the direct activator of ERK6/p38␥, and it depended on MRK kinase activity, as expression of the kinase inactive mutant, MRK-␤-K45A, did not result in activation. This level of activation was also observed when the translation factor PHAS-1 (38, 39) was used as substrate in the in vitro kinase reaction (data not shown). These results indicate that MRK-␤ is upstream of the stress kinases JNK and ERK6/p38␥.
To assess whether MRK-␤ could activate endogenous MAP kinases, we generated stable Madin-Darby canine kidney (MDCK) (25) cell lines that expressed wild type or catalytically inactive MRK-␤-K45A under the control of a tetracycline-repressible transactivator (26). In this system, removal of the tetracycline analog doxycycline induced expression of the recombinant proteins. This expression system offered the advantage of controlling the recombinant protein expression levels by titration of doxycycline. Vector-transfected clones were used as control to monitor the effect of the induction treatment. Fig. 6 shows that, upon induction of the MRK-␤ wild type protein, both endogenous JNK and ERK6/p38␥ are activated. Although the anti-phospho-ERK6/p38␥ antibodies cross-react with phospho-p38␣, a parallel set of samples blotted with the anti-phospho-p38␣-specific antibodies showed a very weak signal. These results confirmed our observations with transient transfection that JNK and ERK6/p38␥ are strongly activated by MRK-␤, whereas no significant activation of p38␣ or ERK1 and -2 was detected. Activation of these pathways depended on the kinase activity of MRK-␤, as stimulation was not observed after expressing similar levels of the recombinant catalytically inactive mutant, MRK-␤-K45A (Fig. 6).
To confirm that MRK-␤ activated JNK and ERK6/p38␥ via their respective MKK proteins, we tested the activation state of the endogenous MKK4 and MKK3/MKK6 proteins in the MDCK clones using antibodies specific for these phosphoki-nases. Fig. 6 shows that, upon induction of MRK-␤ expression, these MKK proteins became phosphorylated at their active sites. Interestingly, both the MKK3/MKK6 and ERK6/p38␥ proteins are activated as early as 8 h after induction of MRK-␤ expression, when the levels of the recombinant protein are relatively low. In contrast, the JNK pathway is activated at later time points. Therefore, in this system MRK-␤ strongly stimulates the ERK6/p38␥ pathway.
MRK-␤ Phosphorylates MKK4, -6, and -7 in Vitro-The sequence homology between MRK and members of the MAPKKK family suggests that it might phosphorylate and activate members of the MAPKK family of proteins. We therefore tested some of the known members of this family as substrates to be phosphorylated directly by MRK-␤ in vitro. We transiently transfected COS-1 cells with MRK-␤, MEKK1 as positive control for MEK activation, or the empty vector as negative control. The recombinant proteins were immunoprecipitated with  Fig. 7 shows that MRK-␤ did not phosphorylate a kinase-inactive version of recombinant MEK, which, however, was a good substrate for MEKK1. This result is in agreement with the lack of substantial activation of the ERK pathway by MRK-␤ and suggests that the ERK activation observed in COS-1 cells is indirect and a possible result of production of autocrine growth factors. On the other hand, we found good incorporation of [␥-32 P]ATP in the other three MKK proteins.
These in vitro results, together with the activation observed in cells, suggest that MRK-␤ preferentially activates the JNK pathway by phosphorylating MKK4 and MKK7 and the ERK6/ p38␥ pathway via MKK3/MKK6. A distinction between the latter two proteins is not possible because the anti-phospho-MKK3/6 antibodies cannot discriminate between these two kinases.

Constitutive Activation of MRK Increases the G 2 /M Cell Population-
The activation of ERK6/p38␥ by MRK prompted us to investigate the role of MRK in the checkpoint regulation in response to DNA damage, a pathway that has been reported to be mediated by ERK6/p38␥ (35). We first analyzed the cell cycle profile of the MDCK cell line expressing wild type MRK. Upon removal of doxycycline for 48 h, both vector control cells and the MRK-expressing cells were subjected to FACS analysis. Fig. 8 shows that the MRK-expressing cells had a significantly (p Ͻ 0.02) higher percentage of cells in the G 2 /M phase of the cell cycle (28%) compared with the control cells (17%), whereas the other phases were not significantly affected. Because overexpression of MRK results in activation of its kinase activity, this observation suggested that activation of MRK might be involved in the regulation of the G 2 checkpoint control.
The MRK-mediated Pathway Is Needed for Cell Cycle Arrest in Response to ␥-Irradiation-Because MRK activation increases the cell population in the G 2 /M phase, we tested whether the cell cycle arrest induced by ␥-irradiation is affected by expression of the MRK dominant negative allele. MDCK clones expressing MRK-K45A or vector control were induced to express the recombinant gene and treated 24 h later with 20 Gy of ␥-radiation. The cell cycle profile was analyzed for the treated and untreated cells. Fig. 9A shows that the untreated population of control cells and MRK-K45A clones have a similar cell cycle profile. However, after irradiation the clones expressing dominant negative MRK had a reduced percentage of cells arrested in G 2 compared with the control population (52% versus 67%, p Ͻ 0.02). In addition, whereas control cells had a very low percentage in S phase, indicating a G 1 block, this fraction was more than doubled in the MRK-K45A population (13 and 32%, respectively). The presence of more cells in S phase indicates that the G 1 arrest is also reduced and is consistent with the lower percentage of cells in G 1 (16 versus 20%, p Ͻ 0.001). These data collectively indicate that expression of dominant negative MRK attenuated the ␥-radiationinduced G 1 and G 2 arrest.
␥-Radiation Induces MRK Activation-To assess whether the endogenous MRK pathway is involved in the cellular response to DNA damage, we asked whether its kinase activity is induced in response to ␥-radiation. We used the MDCK vector control clones and subjected them to the same treatment as described above, 20 Gy of ␥-radiation, and at different times after irradiation, we tested the kinase activity of endogenous MRK after immunoprecipitation with MRK-specific antibodies. Fig. 10 shows that MRK activity, as measured by autophosphorylation and by MBP phosphorylation, is increased 2-fold over background 15 min after irradiation, and it is still above background 4 h post-treatment. Thus, ␥-irradiation stimulates a pathway that leads to MRK activation, which contributes to cell cycle arrest. DISCUSSION In this study, we describe the identification of a human serine/threonine kinase, MRK, discovered as an activator of S. cerevisiae Ste7, the yeast MEK homolog that mediates the mating pheromone response. We also characterize MRK-␤ as a member of the MAPKKK family and show that MRK-␤ preferentially activates the ERK6/p38␥ and the JNK MAP kinase pathways. In addition, we provide evidence that the MRK-␤mediated pathway is activated by ␥-radiation and is necessary for the G 1 and G 2 arrest induced by DNA damage.
We identified two splice variants of the MRK gene, as supported by the characterization of the genomic structure of the MRK locus. The gene is spread over more than 200 kb, a rather long stretch of genomic sequence. Interestingly, the MRK-␤ mRNA has an unusually long 5.6-kb 3Ј-UTR that could be involved in post-transcriptional regulation. Although rare, a long 3Ј-UTR has been reported for other mRNAs, such as one of the FGF-2 mRNA species (40) where it has been implicated in modulating translation (41). The role of this region in MRK mRNA stability or in protein expression remains to be investigated. It is also possible that the transcript length may control splicing, yielding a much less abundant mRNA encoding the alternative splice form, MRK-␣. This form is, in fact, expressed at much lower levels than MRK-␤ in all tissues with the exception of liver, where it appears to be the major species.
The MRK proteins share significant homology in the kinase domain with proteins of the MAPKKK family. The most closely related members are those in the MLK subfamily. However, the homology is restricted to the kinase domain and remains in the 50% similarity range. There is a single leucine zipper domain in the MRK proteins, whereas this is found as a double domain in the MLK family members.
The functional identification and the primary sequence suggest that MRK-␤ is a member of the MAPKKK family. This was confirmed by its activation of specific MAP kinase pathways. Although activation of the three major MAP kinase pathways was observed when MRK was greatly overexpressed with the respective MAP kinases in cells, we found that the effects of relatively low levels of MRK on endogenous MAP kinases were more specific. Of the pathways tested, the ERK6/p38␥ and JNK cascades were predominantly activated, whereas the ERK and the p38␣ pathways were marginally affected. In vitro phosphorylation studies demonstrated that the effect on ERK is indirect, as shown by the inability of MRK-␤ to phosphorylate MEK directly. Therefore, the activation of MEK, when co-transfected with MRK in cells, is likely to be secondary to new gene expression of autocrine factors. In line with this interpretation, we did not observe any effect on the activation of endogenous ERK1 and -2 in MRK-expressing MDCK cells (Fig. 6). In contrast, the MKK proteins upstream of JNK and p38, MKK4 and MKK3/MKK6, respectively, were found to be good substrates in vitro as well as in cells. Remarkably, the activation of endogenous MKK3/MKK6 and ERK6/p38␥ proteins was already obvious at a time when recombinant MRK is expressed at relatively low levels, 8 h after induction, underlining the preferential stimulation of this pathway. The activation of these stressactivated kinases did not appear to be the result of cellular stress caused by overexpression of proteins in cells, because expression of similar levels of the catalytically inactive MRK kinase did not elicit any of the responses observed with the expression of the wild type protein. This observation, therefore, supports the specificity of the MRK-induced effects.
While this work was in progress, Gotoh et al. (42) reported the isolation of two mouse clones orthologous to the MRK genes, called MLTK. However, they reported indiscriminate activation of the ERK, JNK, p38, and ERK5 pathways. It is possible that the experimental approach used in their study, namely co-expression of the kinases with each of the potential substrates in cells, could explain the lack of discrimination observed among these signaling pathways. As discussed above, autocrine factors induced by the recombinant proteins may account for the observed effects on some of the pathways tested.
Members of the p38 pathway, such as MKK3, MKK6, and ERK6/p38␥, are preferentially expressed in heart or skeletal muscle (16,(43)(44)(45). Interestingly, the levels of MRK-␤ are particularly elevated in these tissues. It will be of interest to explore the possibility that MRK-␤, via the ERK6/p38␥ pathway, plays an important role in the physiology of these tissues.
This work also identifies a role for MRK in the cell cycle checkpoint regulation in response to DNA damage-inducing radiation. In the MDCK cell system, the effect on the cell cycle caused by wild type MRK suggests that this kinase mediates signals leading to G 2 arrest. This hypothesis is supported by the finding that dominant negative MRK reduces the effects of ␥-radiation on the G 1 and G 2 phases of the cell cycle. This observation is consistent with a physiological function of MRK in the regulation of cell cycle checkpoints. These data were supported by the finding that endogenous MRK is activated by ␥-irradiation shortly after treatment. How does MRK affect checkpoint regulation? The ERK6/p38␥ cascade has been implicated in ␥-radiation-induced cell cycle arrest (35). Given the prominent activation of ERK6/p38␥ by MRK, this member of the p38 family was expected to be a good candidate. Surprisingly, no activation of ERK6/p38␥ was detected in response to ␥-irradiation (data not shown). Therefore, at this time we cannot implicate ERK6/p38␥ in the observed cell cycle effects of ␥-radiation in this system. It is possible that cell type differences are responsible for this discrepancy.
In conclusion, we have shown that MRK, a member of the MAPKKK family, preferentially activates the ERK6/p38␥ and JNK pathways and plays a role in the regulation of DNA damage-induced checkpoints. Future studies will address the identification of the elements that relay the signals initiated by ␥-radiation to MRK and those that act downstream of MRK in response to DNA damage. FIG. 10. Activation of endogenous MRK by ␥ radiation. MDCK vector control cells were exposed to 20 Gy of ␥-radiation and harvested at the indicated time points. A, 800 g of proteins were immunoprecipitated with the 40-5-specific MRK antibodies and subjected to in vitro kinase assay as described under "Experimental Procedures" using MBP as substrate. MRK autophosphorylation and MBP phosphorylation are shown. MRK(WB) is shown as loading control. The incorporated 32 P was measured with the PhosphorImager, and values were normalized for the amount at time 0. B, data are means Ϯ S.E. from three independent experiments. some of the authors were at Onyx Pharmaceuticals, we thank Onyx and Bayer, Inc., for interactive support.