Molecular Cloning and Functional Expression of a cDNA Encoding a New Member of Mixed Lineage Protein Kinase from Human Brain*

We have cloned a novel protein kinase from human cerebellum and named it LZK (leucine zipper-bearing kinase). The LZK cDNA encoded a 966-amino acid polypeptide that contains a kinase catalytic domain and double leucine/isoleucine zippers separated by a short spacer region. The amino acid sequence of the kinase catalytic domain was a hybrid between those in serine/threonine and tyrosine protein kinases, indicating that LZK belongs to the subfamily of the mixed lineage kinase (MLK) family. The kinase catalytic domain of LZK was most similar to DLK (Holtzman, L. B., Merritt, S.E., and Fan, G. (1994) J. Biol. Chem. 269, 30808–30817), MUK (Hirai, S., Izawa, M., Osada, S., Spyrou, G., and Ohno, S. (1996)Oncogene 12, 641–650), and ZPK (Reddy, U. R., and Presure, D. (1994) Biochem. Biophys. Res. Commun. 202, 613–620), which belong to the same subfamily of the MLK family. However, besides the kinase catalytic domain and double leucine/isoleucine zippers, there was no significant homology with known proteins. The recombinant LZK autophosphorylated in the presence of ATP and divalent cations, and exhibited serine/threonine kinase catalytic activity. Northern blot analysis revealed that LZK is expressed most strongly in the pancreas, with a pattern that differs from other MLKs. Expression of LZK in COS7 cells induced phosphorylation of c-Jun and activation of JNK-1, indicating the association of LZK in the c-Jun amino-terminal kinase/stress-activated protein kinase pathway. The expressed LZK was detected primarily in the membrane fraction, suggesting that LZK interacts with other cellular components in vivo.

Protein kinases play critical roles in the regulation of many cellular processes (1), such as the transmission of signals from growth factor (2,3), control of cell growth and division (4), regulation of cytoskeletal changes (5), gene expression and differentiation (6), translation (7), and metabolism (1). The protein kinases can be divided into two groups based on their sequence similarities and their specificity for the acceptor amino acid (1,8,9). Most protein kinases phosphorylate either serine/threonine or tyrosine, although protein kinases that modify histidine have been found. However, a small number of dual-specificity kinases can phosphorylate both serine/threonine and tyrosine residues (10), although they are structurally related to the serine/threonine-specific group. Protein kinases can also be grouped as receptor protein kinases and non-receptor protein kinases. Receptor protein kinases have an intracellular catalytic domain, transmembrane region, and extracellular ligand-binding domain. Protein kinases share, besides the protein kinase catalytic domain, some structural features reflecting their particular roles in protein-protein interactions. For example, the SH3 1 domains are found not only in tyrosine kinases and serine/threonine kinases but also in receptor-type and non-receptor protein kinases (11,12). The leucine/isoleucine zipper sequence is found in some protein kinases (13). Recently, many new closely related intracellular kinases have been identified. One of these groups, mixed lineage kinases (MLKs), contains a unique double leucine/isoleucine zipper (14). MLK has a characteristic kinase catalytic domain with a sequence hybrid between those in serine/threonine and tyrosine protein kinases. These kinases include MLK1 (15), MLK2 (16,17), MLK3/SPRK/PTK1 (18 -20), and DLK/ZPK/MUK (21)(22)(23). Of these, DLK/ZPK/MUK are considered a secondary subfamily of MLK because of their characteristic sequences. However, little is known about the overall biochemical features and functional roles of MLKs.
We examined the cloning, expression, and preliminary characteristics of the novel intracellular protein kinase. The LZK cDNA encoded a protein with an apparent molecular mass of 135-150 kDa on reducing SDS-PAGE. Sequence analysis revealed that LZK belongs to the MLK family, containing the kinase catalytic domain and leucine/isoleucine zippers. However, LZK had no other strong homologies with any known proteins. The recombinant LZK protein exhibited serine/threonine protein kinase activities in vitro. Expression of LZK in COS7 cells induced phosphorylation of c-Jun and activation of JNK-1, suggesting the association of LZK in the JNK/SAPK pathway.

EXPERIMENTAL PROCEDURES
cDNA Library Screening and Sequence Determination of LZK-A 826-bp rat cDNA clone with unknown functions, which had been iso-* This work was supported in part by the Special Coordination Funds of the Japanese Science and Technology Agency for Promoting Science and Technology and also by a grant-in-aid for Scientific Research on Priority Areas from the Japanese Ministry of Education, Science and Culture. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AB001872.
Analysis of LZK Transcript Expression-Multiple human tissue Northern blot (CLONTECH) was hybridized to radiolabeled human LZK cDNA fragment (corresponding to nucleotides 1895-3174), which had been amplified by polymerase chain reaction and then labeled with [␣-32 P]dCTP by a random primer method. Hybridization was performed as described above for cDNA screening. The filter was finally washed at 65°C in 0.1 ϫ SSC and 0.1% SDS, and analyzed by BAS 2000 image analyzer. To ensure the integrity and the quantity of RNA per lane, the blot was rehybridized to radiolabeled ␤-actin cDNA.
Construction of Epitope-tagged LZK-The cDNA fragment encoding the LZK open reading frame was engineered with XbaI restriction sites, and the product was amplified by long and accurate (LA)-polymerase chain reaction (oligonucleotides: 5Ј-GCTCTAGAATGGCCAACTT-TCAGGAGCACCTGAGCTGCTCCT-3Ј; 5Ј-GCTCTAGATCATTACCA-GGTAGCAGAGCTGTAGTGTTTATTGGT-3Ј). The digested fragment and a double-strand oligonucleotide linker (oligonucleotides: 5Ј-AGCT-TCCACCATGAGAGGATCGCACCACCATCATCACCACT-3Ј; 5Ј-CTA-GAGTGGTGATGATGGTGGTGCGATCCTCTCATGGTGGA-3Ј) were inserted into the cytomegalovirus promotor-based eukaryotic expression vector pCDM8, which had been double-digested with HindIII and XbaI. The linker presented above was engineered with HindIII (5Ј) and XbaI (3Ј) restriction sites, and contained the typical Kozak's consensus sequence and coding sequence for the "MRGSHis 6 " epitope (Met-Arg-Gly-Ser-His 6 ). The MRGSHis 6 tag was inserted into the amino terminus of the LZK coding sequence. The construct was sequenced to confirm Taq polymerase fidelity and maintenance of the appropriate reading frame.
Expression of the LZK Constructs-COS7 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and kanamycin. Cells (2 ϫ 10 6 ) plated onto a 10-cm tissue culture dish were grown overnight and transiently transfected with 10 g of the eukaryotic expression plasmid using LipofectAMINE™ (Life Technologies, Inc.) according to the manufacturer's protocol. After 48 h, cells were washed twice in ice-cold phosphate-buffered saline and then lysed by adding 1 ml of lysis buffer (50 mM Hepes, pH 7.5, 150 mM NaCl, 1.5 mM MgCl 2 , 1 mM EGTA, 10% glycerol, 0.2% Triton X-100, and protease inhibitors). The lysate was sonicated on ice and then centrifuged for 20 min at 105,000 ϫ g at 4°C.
For immunoprecipitation, 2 g of anti-MRGSHis 6 antibody (QIA-GEN) and 40 l of anti-mouse IgG-Sepharose (Sigma) (50% v/v) were added to the supernatant of the cell lysate, and the mixture was incubated at 4°C overnight. Beads were washed five times in HNTG buffer (20 mM Hepes, pH 7.5, 150 mM NaCl, 0.1% Triton X-100, and 10% glycerol). For cell fractionation study, the cells were homogenized in the lysis buffer without Triton X-100, and the homogenate was centrifuged as above.
For immunoblot analysis, 20 l of cell lysate or the immunoprecipitates were separated under reducing conditions on a 7% SDS-polyacrylamide gel according to Laemmli (24). Proteins were electrically transferred onto nitrocellulose membranes, blocked for 2 h in Tris-buffered saline (TBS, pH 7.5) containing 3% nonfat dry milk, followed by incubation with the MRGSHis 6 antibody or the rabbit anti-LZK immune serum diluted 1:2000 in TBS containing 3% nonfat dry milk and 0.05% Tween 20, and then probed with appropriate horseradish peroxidaseconjugated second antibodies. Blots were developed using the chemiluminescent reagent (Pierce) and subjected to autoradiography.
Phosphoamino Acid Analysis-Phosphoamino acid analysis was carried out as described by Zheng and Guan (25). Following the in vitro kinase assay, the radioactive band of 135-150 kDa was excised from the PVDF membrane. The strip was incubated in 1 ml of 6 M HCl at 105°C for 2 h. After removing of the strip, the sample was dried in a SpeedVac, then washed and dried twice in 1 ml of H 2 O. The resulting amino acids were separated on a cellulose plate by one-dimensional electrophoresis. Phosphoamino acid standards were visualized by ninhydrin staining, and radioactivity was detected by a BAS 2000 image analyzer.
Immunoblot Analysis of c-Jun-Cells (2 ϫ 10 6 ) were transiently transfected with the expression vector harboring epitope-tagged LZK. After 48 h, cells were washed with ice-cold phosphate-buffered saline three times and then lysed in situ with 1 ml of Laemmli sample buffer. For control experiment, cells were stimulated by UV radiation (100 J/m 2 ). After 1 h, they were used for the experiment. Cell lysate (20 l) was subjected to SDS-PAGE, and separated proteins were transferred on to the nitrocellulose membrane. The membrane was blocked by soaking in TBS containing 3% nonfat dry milk, incubated with diluted anti-c-Jun antibodies (0.1 g/ml in TBS containing 3% nonfat dry milk and 0.05% Tween 20), and subsequently with appropriately diluted horseradish peroxidase-conjugated secondary antibodies. The resulting membrane was developed using the chemiluminescent reagent and subjected to autoradiography.

RESULTS
Isolation of a LZK cDNA and Its Deduced Amino Acid Sequence-A 826-bp rat cDNA fragment with unknown function was used as a probe to screen a human cerebellum cDNA library. Three independent clones were isolated, and their inserts were sequenced. The nucleotide sequence of the longest insert is shown in Fig. 1. The cDNA extends over 3450 nucleotide bases and contains 272 bp of 5Ј-untranslated nucleotides, a continuous open reading frame of 2898 bp, and 399 bp of 3Ј-untranslated nucleotides. The putative initiation codon was assigned at nucleotide 273. This methionine codon is located within a sequence context favorable for the Kozak's rule and is preceded by an in-frame stop codon beginning at base 234. Within the 3Ј-untranslated region, putative polyadenylation signals are found at 3318 bp (AATAA), at 3354 bp (AATAA), at 3370 bp (AATTAA), and at 3517 bp (AATAA) upstream from the poly(A) tract. The longest open reading frame of the cDNA encodes a putative polypeptide of 966 amino acids, with a calculated molecular mass of 108 kDa. Hydrophobicity analysis revealed that the protein contains no obvious signal sequence or transmembrane domain (data not shown). Comparison of the sequence with other known proteins revealed that the protein can be divided into several structural domains: a kinase catalytic domain, a double leucine/isoleucine zipper separated by a short spacer region, and an acidic domain at its carboxyl-terminal end (Fig. 2). Because of this characteristic amino acid sequence, the novel protein was designated as H-LZK (human leucine zipper-bearing kinase).
The kinase catalytic domain extends 249 amino acids from amino acids 166 through 414 (Fig. 3). All 11 subdomains (8) heptad repeats of nonaromatic hydrophobic amino acids separated by a 25-amino acid spacer. By Chou and Fasman analysis (32), this amino acid sequence formed an ␣-helix structure, indicating that these regions of LZK are composed of two leucine/isoleucine zipper motifs (Figs. 2 and 3), which may promote homo-or heterodimerization of proteins through hydrophobic interactions. As shown in Fig. 4, hydrophobic residues are conserved at the d position in zipper 1 and 2, forming a hydrophobic stripe on the face of the helix. Except for the d position, these regions are comparatively rich in charged amino acids. In particular, position b (EETE) and position f (KSRR) in zipper 1, and position g (IRRK) in position 2 were primarily composed of negatively or positively charged amino acids, suggesting that they are involved in intra-or intermolecular electrostatic interactions (33,34).
The regions containing the kinase catalytic domain and leucine zipper domain of this protein have 86.4% and 86.4% identity, respectively, to previously reported proteins DLK (dual leucine zipper-bearing kinase) (21) and ZPK (leucine zipper protein kinase) (22) (see Fig. 3). In addition, the sequence of this region was homologous to MLK1 (15), MLK2 (16,17), and MLK3 (18 -20) by 40.2%, 40.4%, and 39.5%, respectively (Figs. 3 and 5), suggesting that LZK, together with DLK/ZPK, belongs to the MLK (mixed lineage kinase) family, although no strong similarity was found outside this region. However, in contrast to the other of MLKs, which have a SH3 domain at their amino-terminal ends, LZK (as well as DLK/ ZPK) did not contain such a structure (Fig. 5). In addition, LZK and DLK/ZPK have a single invariant Glu at 7 amino acid residues downstream from the invariant Lys in subdomain II, but this is not the case with ordinary MLKs. This Glu residue is believed to play an important role in stabilizing ATP in the ATP-binding site from the crystallographic study (35). These results suggest that LZK, together with DLK and ZPK, belongs to the secondary subgroup of MLK. In addition, LZK and DLK/ ZPK share a unique sequence, Ser-Ser-Glu-Glu-Glu-Glu-Gly-Glu-Val-Asp-Ser-Glu-Val-Glu (Ser 815 -Glu 828 in LZK) (Figs. 5 and 6). However, the glycine/proline-rich region present in DLK/ZPK at the carboxyl-and amino-terminal ends was not detected in LZK (Fig. 5). It should be noted that the sequence of the LZK kinase catalytic domain is 94.6% identical with that of a partial putative serine/threonine protein kinase (36), implying that these proteins are identical or closely related (Fig. 3).
Tissue Distribution of LZK mRNA-Expression of LZK mRNA was examined by Northern blotting mRNA from several human tissues. The probe used for this analysis was corresponded to nucleotides 1895-3174 (See Fig. 1). Three bands at about Ͼ9.5, 8.7, and 6.5 kb were found with pancreas mRNA at the highest level. These bands were also markedly detected in the brain, liver, and placenta, and no positive signal was detected in the heart, lung, skeletal muscle or kidney (Fig. 7A). The expression levels of these three transcripts varied among the tissues. The 8.7-kb band was detected only in mRNA from pancreas. Similarly, the Ͼ9.5-kb band was detected only with pancreas and brain. After initial probing with LZK cDNA, the blot was rehybridized with ␤-actin cDNA to confirm the integrity of the RNA from different tissues (Fig. 7B).

Expression of LZK cDNA in COS 7 Cells and in Vitro Phosphorylation of the Recombinant LZK-
To facilitate the detection and immunoprecipitation of the LZK, MRGSHis 6 epitope was incorporated at the amino terminus of LZK (see "Experimental Procedures"). The epitope-tagged full-length LZK cDNA was incorporated into the eukaryotic expression vector pCDM8, and the resulting plasmid was transfected into COS 7 cells. Upon immunoblot analysis of LZK transfectants following the SDS-PAGE under reducing conditions, a protein with a molecular mass of 135-150 kDa, which is in good agreement with the predicted mass of the epitope-tagged LZK, was detected, while no band was detected for the non-transfectant (Fig. 8A). In addition, a protein of 135-150 kDa was specifically immunoprecipitated with a MRGSHis 6 antibody from the lysate of the transfectant (data not shown).
To study the subcellular localization of LZK, COS7 cells expressing LZK were homogenized in the absence of detergent. The homogenate was fractionated into the soluble and the membrane fractions, and the respective fractions were subjected to SDS-PAGE and followed by immunoblot analysis. Strong immunoreactive bands were detected in the membrane fraction, while only weak bands were found in the soluble fraction (Fig. 8B), suggesting that LZK protein binds to some membrane components probably through interaction with some other cellular components such as lipid and/or anchor protein.
To confirm that LZK is an active protein kinase, MRGSHis 6 antibody immunoprecipitates of the LZK transfectants were incubated with [␥-32 P]ATP in the presence of Mn 2ϩ , Mg 2ϩ , and Na 3 VO 4 (protein-tyrosine phosphatase inhibitor), and then the proteins were separated by SDS-PAGE under reducing conditions followed by transfer onto PVDF membranes. Upon autoradiography, immunoprecipitates from the transfectants revealed radioactive bands of 135-150 and 50 kDa, but no detectable bands in non-transfectants (Fig. 9B). The radioactive band of 50 kDa comigrated with the band of heavy chain of IgG, indicating that LZK not only autophosphorylated itself but also phosphorylated heavy chain of IgG.
The radioactive 135-150-kDa band of LZK from the in vitro kinase assay was excised and subjected to partial acid hydrolysis. The resulting materials were separated by one-dimensional electrophoresis on a cellulose plate (25). Analysis by autoradiography and comparison to ninhydrin-stained phosphoamino acid standards revealed only phosphoserine and phosphothreonine (Fig. 9C), indicating that LZK has a serine/ threonine kinase activity. However, the present experiment cannot completely exclude the possibility that LZK has a tyrosine kinase activity.
Activation of JNK Pathway by LZK-Recent studies show that some MLKs activate JNK pathway (23,26,27). JNK pathway is believed to be predominantly activated by cellular stresses such as UV radiation, inflammatory cytokines, and osmotic shock (28,29), which results in the activation of transcriptional factors such as c-Jun and ATF2 (30,31). Because the amino acid sequence of LZK showed high homology to DLK/MUK, which were known to activate JNK pathway, we tested whether or not LZK activates the phosphorylation of c-Jun. COS7 cells were transiently transfected with the expression vector harboring an epitope-tagged LZK, after which the mobility delay of endogenous c-Jun was monitored by immunoblot analysis with anti-c-Jun antibodies. As shown in Fig. 10, expression of LZK induced the mobility delay of c-Jun as much as was observed with UV radiation. Because the mobility delay is caused by the phosphorylation of c-Jun, these results suggested that expressed LZK activates the endogenous JNK pathway (28). Then to confirm that the phosphorylation of c-Jun observed was really caused by activation of JNK, endogenous JNK1 was immunoprecipitated from the cell lysate and JNK1 activity was determined by in vitro kinase assay using soluble glutathione S-transferase-c-Jun as substrate. As shown in Fig.  11, expression of LZK elevated the JNK1 kinase activity. The extent of JNK1 activation by expression of LZK was comparable to that caused by UV radiation. These results taken together indicated that LZK can effectively activate JNK pathway. DISCUSSION We examined the cDNA cloning, expression, and characteristics of a novel protein kinase, which is expressed in a spatially regulated fashion in adult human tissues. This protein kinase contains a kinase catalytic domain, followed by two leucine/ isoleucine zipper motifs, which are separated by a short spacer region. We designated this protein kinase as LZK. The LZK cDNA encodes a protein with an apparent molecular mass of 135-150 kDa, and has serine/threonine kinase activity.
LZK is most similar to DLK and ZPK. DLK was identified by Holzman (21) as a novel protein kinase with a unique kinase catalytic domain, the expression of which is regulated spatially and developmentally. ZPK is cloned and identified as a novel putative protein kinase, which is up-regulated in retinoic acidtreated NT2 cells (22). When the region containing the kinase catalytic domain and the leucine/isoleucine zipper domain of LZK was aligned to DLK and ZPK, homology was 86.4% and 86.4%, respectively, with no insertion and/or deletion. Like DLK and ZPK, LZK had invariant Glu at the specific location 7 amino acids downstream from invariant Lys in subdomain II. From crystallographic study and structure-function analysis of other protein kinases, the invariant Glu in subdomain III and invariant Lys in subdomain II are believed to play an important role in stabilizing ATP in the ATP-binding site.
The amino acid sequence WMAPE in subdomain VIII is often found in Raf family protein kinases, suggesting that LZK has a MAPKKK-related activity. It is interesting to note that Hirai et al. (23) recently identified MUK, which corresponds to rat homologue of DLK (mouse) and/or ZPK (human), as a MAP-KKK-related protein kinase such as c-Raf and MAPK/ERK kinase kinase (MEKK) (37). They showed that MUK phosphorylates and activates JNKK in vivo and in vitro. JNKK (38,39) can be phosphorylated and activated by the MAPKKK-related kinase, MEKK (40,41), and acts on Jun kinases, resulting in activation of c-Jun (29,42). MUK-transfected cells induced hyperphosphorylation of c-Jun, suggesting that MUK can regulate the JNK/SAPK pathway in vivo. The induction of JNK was also observed in a truncated MUK consisting of the kinase catalytic domain and leucine/isoleucine zipper motifs, the amino acid sequence of which was 86.4% identical to that of LZK. As might be expected from this high homology with MUK, LZK was in fact shown to induce phosphorylation of c-Jun and activation of JNK1, indicating that LZK stimulates the JNK/ SAPK pathway. The extent of JNK1 activation by LZK expression was comparable to that caused by UV radiation. Consid-  4. Helical wheel representation of the leucine/isoleucine zippers of LZK. The residues of putative leucine/ isoleucine zippers of LZK were arrayed on a helical wheel. The spokes of the wheel show the relative positions of the amino acids in an ␣-helix, and the positions a-d correspond to the location of the amino acid residues. In an ideal ␣-helix, amino acid residues appear on one side of the helix in every two turns. In this model, conserved hydrophobic amino acids were located at the d position.
ering the efficiency and cytotoxicity of the transfection procedure, it seems reasonable to speculate that LZK directly phosphorylates and activates the main components of JNK pathway, such as JNKK and MEKK in vivo.
When expressed in COS7, LZK was present in both cytosol and membrane fractions. Because LZK contains no obvious signal sequence or transmembrane domain, LZK should first be synthesized in cytosol and then translocated to membranes. It has been thought that subcellular compartmentalization is crucial in providing specificity in the regulation and function of protein kinases (43). Some protein kinases were targeted in a given compartment in the cell, and following various stimulations, they translocated to new sites within the cell, where they associated with anchor proteins, regulated by other protein and/or lipid, to gain access to their physiological substrates. Mata et al. (44) recently reported that DLK also exists in both cytosolic and membrane-bound form. They showed that each form of DLK has different biochemical characteristics. The membrane-bound form of DLK is not phosphorylated and forms high molecular complexes, and the cytosolic form of DLK is phosphorylated and exists as monomers. Since LZK, unlike other related protein kinases, does not contain a SH3 domain or a proline-rich region that is a presumed SH3-binding motif, it remains to be clarified what kind of interaction induces the translocation of LZK and regulates the functions of LZK. Recently, MLK-3 was shown to interact specifically with the GTPbound form of Rac and Cdc42 (45) activation (46 -48). Considering that structurally related MUK activated the JNK pathway, LZK might be associated with mitogen-activated protein kinase pathways under the regulation of a small GTP-binding protein. However, to clarify the mechanism which might regulate the function of LZK, further studies must be done on the biochemical difference between phosphorylated and non-phosphorylated forms of LZK and the signals regulating the compartmentalization of LZK.