Identification and characterization of a novel related adhesion focal tyrosine kinase (RAFTK) from megakaryocytes and brain.

We have isolated a cDNA encoding a novel human intracytoplasmic tyrosine kinase, termed RAFTK (for a related adhesion focal tyrosine kinase). In addition, we have cloned and characterized the murine homolog of the human RAFTK cDNA. Comparison of the deduced amino acid sequences of human RAFTK and murine Raftk cDNAs revealed 95% homology, indicating that RAFTK is highly conserved between these species. The RAFTK cDNA clone, encoding a polypeptide of 1009 amino acids, has closest homology (48% identity, 65% similarity) to the focal adhesion kinase (pp125FAK). Comparison of the deduced amino acid sequences also indicates that RAFTK, like pp125FAK, lacks a transmembrane region, myristylation sites, and SH2 and SH3 domains. In addition, like pp125FAK, RAFTK contains a kinase domain flanked by large N-terminal (426 residues) and C-terminal (331 residues) domains, and the C-terminal region contains a predicted proline-rich stretch of residues. In fetal tissues, RAFTK expression was abundant in brain, and low levels were observed in lung and liver. In adult tissues, it was less restricted, indicating that RAFTK expression is developmentally up-regulated. Expression of RAFTK was also observed in human CD34+ marrow cells, primary bone marrow megakaryocytes, platelets, and various areas of brain. The human RAFTK gene was assigned to human chromosome 8 using genomic DNAs from human/rodent somatic cell hybrid lines. The mouse Raftk gene was mapped to chromosome 14, closely linked to gonadotropin-releasing hormone. Using specific antibodies for RAFTK, a approximately 123-kDa protein from the human megakaryocytic CMK cell line was immunoprecipitated. Treatment of the megakaryocytic CMK cells with thrombin caused a rapid induction of tyrosine phosphorylation of RAFTK protein. The structural features of RAFTK suggest that it is a member of the focal adhesion kinase gene family and may participate in signal transduction in human megakaryocytes and brain as well as in other cell types.

Protein-tyrosine kinases participate in a variety of signal transduction pathways that modulate cell growth and differen-tiation (1)(2)(3). Signal transduction is triggered by stimulation of a cell-surface receptor that either has kinase activity itself or is physically and/or functionally linked to an intracellular protein-tyrosine kinase (4 -6). The integrins are also capable of transducing cytoplasmic signals (7)(8)(9), and activation of this pathway is linked to one or more protein-tyrosine kinases (10,11). A nonreceptor cytosolic tyrosine kinase, termed the focal adhesion kinase (pp125 FAK ), has been identified as one of the cellular proteins that is phosphorylated in response to ␤ 1 -and ␤ 3 -integrin-mediated cell adhesion (8,(12)(13)(14)(15). Induction of the kinase activity and tyrosine phosphorylation of pp125 FAK were observed following the adherence of fibroblasts to fibronectin (10 -16); the adherence of epidermal carcinoma cells to fibronectin, laminin, or collagen type IV (17); and the aggregation of platelets in the presence of fibrinogen, a ligand for ␣ IIb ␤ 3 -integrin (glycoprotein IIb/IIIa) (18). Phosphorylated pp125 FAK is localized in focal adhesion contacts. pp125 FAK has been cloned from Xenopus laevis, avian, rodent, and human species and is expressed in a wide range of cell types (13,14,19,20). In addition to activation of pp125 FAK following stimulation of ␤ 1 -integrins (11) and glycoprotein IIb/ IIIa integrin (21), it has been observed that pp125 FAK is activated in signaling mediated via high affinity IgE receptors (22,23) and neuropeptide receptors (24) and upon oncogenic transformation (16) in adherent cells. In cells expressing both pp125 FAK and src, pp60 src formed a stable complex with pp125 FAK (25,26). Functional studies of pp125 FAK protein in various systems showed that bradykinin stimulated tyrosine phosphorylation of pp125 FAK , paxillin, Ras GTPase-activating protein-associated p125, and src transformation-associated p130 (27). In addition, stable association of pp125 FAK with phosphatidylinositol 3-kinase in NIH 3T3 mouse fibroblasts was observed (28). pp125 FAK expressed by nerve cell lines manifested increased tyrosine phosphorylation in response to Alzheimer's A␤ peptide (24). Recently, a novel pp125 FAK -related protein that is a substrate of tyrosine kinases in T and B lymphocytes was reported (29).
Relatively little is known about the repertoire of signal transduction molecules in human megakaryocytes (30,31). Platelets, the progeny of megakaryocytes, contain pp125 FAK that is phosphorylated on tyrosine following platelet activation (18,32). We have identified and characterized a novel intracytoplasmic kinase isolated from human megakaryocytic and brain cells with 48% identity (65% similarity) to pp125 FAK . We have also cloned the murine homolog of this cDNA. Given its homology to pp125 FAK , we have termed this new gene a related adhesion focal tyrosine kinase (RAFTK). Based on its molecular structure, its pattern of expression, and the induction of tyrosine phosphorylation of RAFTK proteins by thrombin in megakaryocytic cells, it is likely that RAFTK participates in signal transduction and may have a role in cell growth and differentiation.

EXPERIMENTAL PROCEDURES
Materials-Chemical reagents were purchased from Sigma. Restriction endonucleases, modifying enzymes, and terminal deoxynucleotidyltransferase were purchased from Pharmacia Biotech Inc. and New England Biolabs Inc. (Beverly, MA). The primers for polymerase chain reaction (PCR), 1 RNA PCR, and sequencing were synthesized by an automated DNA synthesizer (Applied Biosystems Inc., Model 394). The PCR and RNA PCR reagents were obtained from Perkin-Elmer, and random-primed labeling kits were obtained from Stratagene (La Jolla, CA). Manual and automated sequencing kits were obtained from U. S. Biochemical Corp. and Pharmacia Biotech Inc., respectively. Automated sequencing was performed using Pharmacia's automated laser fluorescent sequencer. Monoclonal antibody 2A7 against pp125 FAK protein was kindly obtained from Dr. J. Thomas Parsons (Department of Microbiology, University of Virginia School of Medicine, Charlottesville, VA). Monoclonal antibody PY-20 directed against Tyr(P) was obtained from ICN (Costa Mesa, CA).
Cells-Human marrow megakaryocytes were isolated by a method employing immunomagnetic beads using anti-human glycoprotein IIIa monoclonal antibody as described previously (33,34). CD34-bearing marrow progenitor cells were purified from heparinized bone marrow aspirates using immunomagnetic beads coated with an anti-CD34 monoclonal antibody as described previously (34). The CMK cell line, provided by Dr. T. Sato and derived from a child with megakaryoblastic leukemia, has properties of cells of the megakaryocytic lineage (35). The CMK cell line was cultured in RPMI 1640 medium with 10% fetal calf serum. Additional permanent human megakaryocytic cell lines were studied. DAMI cells were obtained from Dr. S. Greenberg (Brigham and Women's Hospital, Boston, MA); Mo7e and erythroid megakaryocytic HEL cells were obtained from Dr. L. Zon (Children's Hospital, Boston, MA). Each cell line was cultured as described previously (34,36,37). Other permanent human cell lines such as Ramos (human B-cells) were obtained from the American Type Culture Collection and maintained in liquid culture according to the specifications in the catalog. Human platelets were isolated by gel filtration from freshly drawn blood anticoagulated with 0.15 volume of NIH formula A acid/citrate/dextrose solution supplemented with 1 M prostaglandin E 1 as described previously (18).
DNA Amplification and Cloning-Total RNA derived from CMK cells was prepared by a standard protocol of lysis in guanidinium isothiocyanate followed by cesium chloride gradient centrifugation (38). Proteintyrosine kinase sequences were amplified with degenerate oligonucleotide primers as described previously (39). Briefly, total RNA (10 g) was used as a template for synthesis of cDNA. The PTK3 oligonucleotide SDVWS(F/Y)G (5Ј-(C/G)(T/A)(A/G)TC(A/C/G/T)ACCCA(A/C/G/T)(C/G)-(T/A)(A/G)(T/A)A(A/C/G/T)CC-3Ј) was designed in our laboratory and was used as a primer. PCR was performed on one-fourth of the cDNA synthesis reaction mixture (original volume of 20 l) using the PTK1 DLAARN (5Ј-TCGACGA(T/C)CT(A/C/G/T)GC(A/C/G/T)(A/G)C(A/C/G/ T)AA-3Ј) and PTK2 WMAPE (5Ј-GGTACC(T/C)TC(G/C/A)GG(A/C/G/ T)GCCATCCA-3Ј) oligonucleotides (50 pmol each) (39). The mixture was then subjected to PCR amplification using a Perkin-Elmer thermal cycler set for 30 cycles as follows: denaturing at 95°C for 2 min, primer annealing at 37°C for 1.5 min, and primer extension at 72°C for 2.30 min. 1-min ramp times were used between these temperatures. PCR products of the amplified tyrosine kinase domains were purified from the agarose gel, digested with EcoRI and BamHI, ligated into pUC19, and transformed into Escherichia coli DH5␣. Sequencing was carried out by the dideoxy chain termination method using a Version 2.0 Sequenase kit (U. S. Biochemical Corp.). Sequences were compared with known sequences in GenBank TM and EMBL Data Banks using the Autosearch computer program. A novel clone was identified. This 160base pair (bp) PCR product, designated JJ3, was radiolabeled using the Prime It II random priming protocol (Stratagene) and used as a probe to screen human cDNA libraries.
A 340-bp probe was prepared from the 5Ј-end of one of the CMK cDNA clones (termed 2-1) and used to screen the human brain (hippocampus) cDNA library. Twelve clones were isolated, and two clones were sequenced on both strands. In addition, a 248-bp probe was prepared from the 5Ј-end of one of the clones (termed 4C), and the human hippocampus cDNA library was rescreened. Twelve clones were identified and isolated, and of these, one clone (termed 3B) was sequenced on both strands.
The mouse brain cDNA library (catalog No. ML1042b, CLONTECH, Palo Alto, CA) in gt11 vector was screened (ϳ5 ϫ 10 5 recombinants/ screening) using the 381-bp 5Ј 3 KpnI fragment or the 764-bp ApaI 3 3Ј-fragment of human RAFTK cDNA as a probe, and the filters were hybridized and washed under high stringency conditions. Six clones were isolated. The DNA was isolated as described previously (38), subcloned into pBSK, and thereafter sequenced. Nucleotide sequences were determined by the automated laser fluorescent DNA sequencer using Autoread (Pharmacia Biotech Inc.) and by manual sequencing using the Sequenase kit.
Chromosomal Localization of the Human RAFTK Gene-Genomic DNAs from the NIGMS hybrid mapping panels 1 and 2 were obtained from the NIGMS Genetic Mutant Cell Repository (Coriel Cell Institute for Medical Research, Camden, NJ). In addition, both mapping panels included DNA samples isolated from human and rodent parental cell lines (mouse and Chinese hamster). Approximately 5 g of human, hamster, and mouse genomic DNAs were digested with BamHI, HindIII, and PstI to find a suitable restriction fragment length polymorphism (RFLP) or unique genomic fragment for use in mapping. Subsequently, genomic DNAs from each panel were cut with BamHI. Southern blots were probed with a 1.4-kb human RAFTK cDNA, and hybridizations were carried out as described previously (40,41). Hybrids were scored for the appropriate human-specific restriction endonuclease fragment on the autoradiographs. The results were compared with the chromosome contents of the hybrid cell lines, and the concordance between restriction fragments and specific chromosome content was used to establish the localization of human RAFTK.
Backcross Mapping of the Mouse Raftk Gene-Genomic DNAs from C57BL/6J, Mus spretus, and a (C57BL/6J ϫ M. spretus) M. spretus BSS-type backcross DNA panel were obtained from The Jackson Laboratory (Bar Harbor, Maine) (40). Southern blots and hybridizations were performed as described previously (41). Approximately 5 g of C57BL/6J and M. spretus genomic DNAs were digested with 29 different restriction enzymes to identify a potential RFLP genetic marker. The Southern blots were probed with a 1.4-kb human RAFTK cDNA fragment labeled with 32 P using a Decaprime II kit (Ambion Inc., Austin, TX). Digestion of the backcross DNA panel with BamHI, Southern blotting, and hybridizations were carried out as described previously (41).
Recombinant Inbred (RI) Line Mapping of the Mouse Raftk Gene-Genomic DNAs isolated from the progenitors of BxD RI lines (C57BL/6J and DBA/2J) were digested with 29 different restriction enzymes to identify a RFLP genetic marker for mapping. Subsequently, genomic DNAs isolated from the BxD RI lines were digested with SacI. Conditions for Southern blotting and hybridizations were the same as described previously (41), and the 1.4-kb human RAFTK cDNA was used as a probe. Data were compared with strain distribution patterns recorded in GBASE (42). 2 Northern Blot Analysis-Total RNA was prepared by a standard protocol of lysis in guanidinium isothiocyanate followed by cesium chloride gradient centrifugation (38). The human adult and fetal tissue Northern blots, the brain regions, and the human tissue II blots were obtained from CLONTECH. Hybridization was carried out according to the manufacturer's instructions. Each RNA blot was probed with a 146-bp 3Ј-gene-specific RAFTK cDNA radiolabeled to a high specific activity (10 8 -10 9 cpm/g) with [␣-32 P]dCTP. The level of expression for each mRNA was also determined densitometrically (E-C Apparatus densitometer). The radioactivity associated with each band was also assayed with a Betascope 603 blot analyzer (Betagen, Mountain View, CA). The same blot was assessed for the presence of the actin-or glyceraldehyde-3-phosphate dehydrogenase-specific probes.
Protein Analysis-Metabolic labeling, immunoprecipitation, and Western blot analysis were performed in CMK cells as described previously (44 -47). For immunoblot analysis, total cell lysates of CMK cells untreated or stimulated with ␣-thrombin (1 or 2 units/ml as indicated; ChromoLog Corp., Havertown, PA) for 5 min were prepared as described previously (45). Relative protein concentrations were determined with a colorimetric assay kit (Bio-Rad) with bovine serum albumin as the standard. A portion of lysate containing ϳ0.05 mg of protein was mixed with an equal volume of 2 ϫ SDS sample buffer containing ␤-mercaptoethanol, boiled for 5 min, fractionated on SDS-8% polyacrylamide gels (44), and transferred to Immobilon polyvinylidene difluoride filters (Millipore Corp., Bedford, MA). Protein blots were treated with RAFTK-specific antibodies (R-4250) (see below). Primary binding of the RAFTK antibodies (see below) was detected using anti-IgG second antibodies conjugated to horseradish peroxidase and subsequent chemiluminescence development using the ECL Western blotting system (Amersham Corp.).
For metabolic labeling, 10 6 cells were labeled with 100 Ci of [ 35 S]methionine in 1 ml of Dulbecco's modified Eagle's medium minus methionine (Amersham Corp.) for 16 h. Immunoprecipitation of RAFTK protein from labeled cells with RAFTK antiserum or with normal rabbit serum was performed as described previously (31,45). For immunoprecipitation of Tyr(P) proteins, cold soluble extracts were first incubated with RAFTK antibodies (R-4250) overnight at 4°C. The extracts were then incubated with protein G-Sepharose beads precoupled to goat anti-rabbit IgG for 1.5 h at 4°C. Proteins were eluted from the beads by heating the samples at 100°C for 5 min in SDS-polyacrylamide gel electrophoresis buffer. Proteins were separated by SDS-polyacrylamide gel electrophoresis, transferred, and immunoblotted with PY-20 (diluted 1:5000). The immunoreactive bands were visualized using the ECL system.
Antibodies-Anti-RAFTK antiserum was obtained from New Zealand White rabbits immunized with a bacterially expressed fusion protein consisting of GST and the C terminus (amino acids 681-1009) of human RAFTK cDNA subcloned into the pGEX-2T expression vector. The sera were titered against the GST-RAFTK C terminus fusion protein by enzyme-linked immunosorbent assay (48,49), and the serum (R-4250) exhibiting the highest titer (1:256,000) was used in subsequent experiments.

RESULTS
Isolation and Characterization of RAFTK cDNAs-To identify tyrosine kinases in human megakaryocytes, PCR primers based on conserved sequences of protein-tyrosine kinases were used (39). RNA from the human megakaryocytic CMK cell line was used as a template to synthesize CMK cDNA. The cDNA was amplified by using the protein-tyrosine kinase primers. Fragments of the expected size (ϳ160 bp) were isolated and subcloned for sequence analysis. One clone that appeared to represent a novel tyrosine kinase (termed JJ3) was used as a probe to screen the human hippocampus cDNA library. A partial cDNA clone (termed S2-3) containing an ϳ2.0-kb insert (Fig. 1A) was isolated. A homology analysis of this clone with human pp125 FAK was performed, and regions were chosen to design specific primers to generate an RAFTK gene-specific probe. The JJ3 fragment was used to screen the human hippocampus cDNA library to obtain overlapping cDNAs. The 5Ј-end of each of these clones was in turn used as a probe to obtain the full-length RAFTK cDNA. Eight different overlapping sequences were obtained of the coding region of RAFTK. Fig. 1A is a schematic representation along with a restriction map of the sequence showing the pattern of overlapping cDNAs. The structure of the composite cDNA is represented in Fig. 1B. The 3.6-kb length of the RAFTK cDNA contains an open reading frame with the first in-frame ATG codon located at nucleotides 294 -296, followed by a stop codon at positions 3260 -3262. This open reading frame encodes a predicted protein of 1009 amino acid residues with a calculated molecular mass of ϳ123 kDa and has been given the name RAFTK (for a related adhesion focal tyrosine kinase). Analysis of the hydrophobicity of the predicted protein revealed lack of a transmembrane region and no recognizable sites for acylation. The kinase domain is flanked by large N-terminal (426 residues) and Cterminal (331 residues) domains. Comparison of the nucleotide sequence and the deduced amino acid sequence of the encoded protein with the National Biomedical Research Foundation and GenBank TM Data Banks revealed that this cDNA encoded a tyrosine kinase related to pp125 FAK . The predicted amino acid sequence of pp120 RAFTK contains the structural motifs common to all protein kinases, including the putative ATPbinding site (Gly 432 -Xaa-Gly 434 -Xaa-Xaa-Gly 437 ) and 3 residues that are predicted to interact with the ␥-phosphate group of the bound ATP molecule (at amino acids 402, 529, and 655). In addition, RAFTK contains two peptide sequences that are highly conserved among protein-tyrosine kinases (Asp 549 -Ile 550 -Ala 551 -Val 552 -Arg 553 -Asn 554 and Pro 588 -Ile 589 -Lys 590 -Trp 591 -Met 592 ). Interestingly, like chicken pp125 FAK , the Cterminal region of RAFTK contains a proline-rich stretch (residues 690 -767) in which the proline content exceeds 20%. A unique domain is found at the N terminus of RAFTK (amino acids 1-39) (Fig. 2 and Fig. 3). This region is the most divergent among various protein-tyrosine kinases and may be involved in cellular localization and/or interaction with other cellular proteins. Like pp125 FAK , RAFTK does not contain SH2 or SH3 domains. The kinase domain (amino acids 427-679) of RAFTK shares 60% identical homology with mouse pp125 FAK , 54% with human pp125 FAK , and 36% with Src (Fig. 3). The kinase domain consists primarily of the catalytic domain including the putative ATP-binding site (amino acids 432-437). RAFTK shares 42% homology in the N-terminal domain and ϳ39% in the C-terminal domain with mouse pp125 FAK . The overall amino acid homology of RAFTK is 48% identity (65% similarity) to mouse pp125 FAK .
Molecular Cloning of the Full-length Murine Raftk cDNA-Southern blot analysis of human and mouse genomic DNAs digested with EcoRI, HindIII, BamHI, XbaI, and PstI and probed under conditions of high stringency with the 3Јfragment of RAFTK cDNA from bp 1595-2974 (ϳ1.4 kb) as a probe revealed a single band in each lane, indicating that the human RAFTK gene and the mouse Raftk gene (data not shown) are highly homologous and are single genes. Therefore, a random-and oligo(dT)-primed mouse adult brain cDNA library was screened under conditions of high stringency for the full-length mouse Raftk cDNA using the 5Ј-and 3Ј-fragments of human RAFTK cDNA as probes. Four clones were isolated; two of these clones were sequenced in both directions, and additional clones were partially sequenced. Sequence analysis of these clones revealed identical sequences. residues and possesses 95.6% identical homology to the human RAFTK gene (Fig. 2).
Chromosomal Localization of the Human RAFTK Gene-Hamster, human, and mouse DNAs were digested with BamHI, HindIII, and PstI to identify a specific RFLP pattern for the RAFTK gene in each species. Southern blots were probed with a 1.4-kb human RAFTK cDNA. Unique 16.5-and 14.5-kb BamHI fragments for RAFTK were identified in human DNA from the parental cell lines used to prepare human/ rodent cell hybrids (Fig. 4). DNAs from the parental and somatic hybrid cell lines in mapping panel 2 were digested with BamHI, Southern-blotted, and probed. Analysis indicated that the human-specific BamHI pattern was observed in cell line 8, which contains human chromosome 8 (Fig. 4). A fainter signal was also observed for the human-specific BamHI pattern in hybrid cell line 20 (Fig. 4), which contains an intact human chromosome 20, but also carries a gene from human chromosome 8 (NEFL (neurofilament light polypeptide), 8p21) as determined by Southern blot hybridization (Coriel Cell Institute for Medical Research). All other hybrid cell lines were negative for the human-specific BamHI RFLP. Additionally, when the 1.4-kb human RAFTK cDNA was used to probe Coriel mapping panel 1, the human-specific fragment was detected in all hybrids containing Ͼ4% of human chromosome 8 and was absent in every hybrid that lacked chromosome 8 (Table I).
Chromosomal man RAFTK cDNA. A BamHI RFLP was detected (Fig. 5A). The alleles for this BamHI RFLP consist of 8.6-and 5.2-kb genomic DNA bands, characteristic of C57BL/6J, and 15.5-and 6.7-kb bands, which are found in M. spretus. These alleles were characterized in 87 DNAs from the C57BL/6J ϫ M. spretus backcross panel. Results of the haplotype analysis from this mapping data indicate that the Raftk gene colocalizes with D14Bir10 (DNA segment Birkenmeier 10) and is linked to Nfl (neurofilament light polypeptide) on mouse chromosome 14 (Fig. 5B). The Raftk locus mapped between Xmv19 (xenotropic MCF leukemia virus-19) and Nfl, and the calculated map distances for these loci are as follows: Xmv19, 7.1 Ϯ 5.3 cM, Raftk, 3.5 Ϯ 2.0 cM, Nfl.
The position of Raftk on mouse chromosome 14 was confirmed by determining the segregation of a SacI RFLP for Raftk DNAs from BxD RI lines. The SacI RFLP for Raftk was indicated by the presence of a 16.5-kb genomic DNA band in C57BL/6J or a 6.2-kb fragment in DBA/2J (Fig. 6A). These alleles were characterized for 26 DNAs from the BxD RI line (Fig. 6B). The strain distribution patterns of Raftk and the locus coding for gonadotropin-releasing hormone, Gnrh (50), indicate close linkage between these two loci on chromosome 14 (Fig. 6B). Perfect concordance was observed with the BxD strain distribution pattern for the Gnrh locus, indicating linkage of Ͻ1-map unit distance from Raftk Gnrh (51). These mapping data place Raftk distal to Nfl and are a contradiction to the backcross data. However, backcross data are not as accurate as RI data since backcross mice were derived from an interspecies cross.
Expression of RAFTK in Tissues and Cell Lines-A specific RAFTK probe was designed (nucleotides 2958 -3103). This sequence is present in RAFTK and not in human pp125 FAK . This probe was used for hybridization of all Northern blots described here.
Northern blot analysis of RNA from human fetal heart, brain, lung, liver, and kidney revealed a weak single major species of mRNA of 4.5 kb in brain, and it appears to be expressed at low levels in lung and liver (Fig. 7A). Expression in human adult tissues was assessed by hybridization of the cDNA probe to a Northern blot of poly(A ϩ ) RNA from heart, brain, placenta, lung, liver, skeletal muscle, kidney, and pancreas. While heart and skeletal muscle RNA samples were negative for RAFTK, a single mRNA was observed in all other tissues, with the highest levels expressed in brain (Fig. 7B). To further characterize the distribution of RAFTK expression in other human tissues, Northern blot analysis of spleen, thymus, prostate, testis, ovary, intestine, colon, and peripheral blood leukocytes revealed a high level of expression of RAFTK in thymus, spleen, and peripheral blood leukocytes (Fig. 7C). Northern blot analysis of different human brain regions (amygdala, caudate nucleus, corpus callosum, hippocampus, hypothalamus, substantia nigra, subthalamic nucleus, and thalamus) revealed that the highest level of expression of RAFTK was in the amygdala and hippocampus (Fig. 8). A lower level of expression was observed in the other brain regions, with the exception of the corpus callosum and substantia nigra, where there was no detectable signal. These results indicate that the brain has abundant expression of RAFTK, especially in the amygdala and hippocampus.
Expression of RAFTK was observed in several megakaryocytic cell lines such as CMK, Mo7e, HEL, and DAMI (data not shown). In addition, expression of RAFTK was detected in Ramos, FHS, and HeLa cells, but a low level of expression was detected in Jurkat, Hep3B, and CCL75 cells (data not shown). Using PCR techniques, expression of RAFTK was also found in primary bone marrow megakaryocytes, blood platelets, and CD34 ϩ marrow progenitor cells. Interestingly, it appears that   the level of expression of RAFTK mRNA is similar to that of FAK in CD34 ϩ cells and is higher than that of FAK in bone marrow megakaryocytes. In platelets, the level of expression of RAFTK mRNA is lower than that of FAK, as observed by PCR under the same experimental conditions. It appears also that RAFTK mRNA expression in bone marrow megakaryocytes is higher than that in CD34 ϩ cells (Fig. 9). Taken together, these results indicate that RAFTK is abundantly expressed in brain and hematopoietic cells. The restricted expression observed in fetal versus adult tissues indicates that its expression is upregulated during development.
Generation of Specific Antibodies for RAFTK and Detection of RAFTK Protein-The GST-RAFTK C terminus fusion protein (residues 681-1009) was chosen for rabbit immunizations in order to obtain specific antibodies for RAFTK protein. These polyclonal antibodies (R-4250) do not cross-react with pp125 FAK . The monoclonal antibody 2A7 against FAK does not cross-react with the GST-RAFTK C terminus fusion protein, 3 indicating that RAFTK might be antigenically different from FAK. Furthermore, FAK immunoprecipitated by monoclonal antibody 2A7 from megakaryocytes was not recognized by polyclonal antiserum 4250 (R-4250). Similarly, RAFTK immunoprecipitated by R-4250 also was not recognized by monoclonal antibody 2A7. 3 Taken together, these data indicate that FAK and RAFTK are distinguishable antigenically while being related members of the FAK family.
The specificity of this antiserum was examined by immunoprecipitation. The CMK cell line was metabolically labeled with [ 35 S]methionine, and extracts were immunoprecipitated with anti-RAFTK antiserum. A major protein species of ϳ123 kDa was detected in CMK cells (Fig. 10A). A similar species was observed in other human megakaryocytic cell lines such as DAMI (data not shown). This band was not observed when normal rabbit serum or preimmune rabbit serum was used for immunoprecipitation. Incubation of R-4250 with 1 or 10 g of the GST-RAFTK C terminus fusion protein abolished the appearance of the ϳ123-kDa protein, while incubation with 10 g of the GST-MATK (where MATK is megakaryocyte tyrosine kinase) SH2 domain fusion protein did not have any effects. These results indicate that R-4250 polyclonal antibodies specifically recognize RAFTK protein of ϳ123 kDa in size. Furthermore, thrombin (1 unit/ml) stimulated a rapid increase in the amount of RAFTK protein immunoreactivity in anti-Tyr(P) immunoprecipitates (Fig. 10B). These results indicate that RAFTK is a protein-tyrosine kinase and that thrombin can induce its tyrosine phosphorylation. DISCUSSION The method of PCR cloning has been successfully employed by many laboratories to identify novel members of the proteintyrosine kinase family. Using this strategy, we have identified a novel intracytoplasmic tyrosine kinase in human megakaryocytic cells that we have termed RAFTK. Sequence analysis of RAFTK revealed ϳ48% identity (65% similarity) to pp125 FAK , suggesting that RAFTK belongs to this subfamily of cytoplasmic tyrosine kinases. RAFTK does not appear to be the recently described FAKB protein (29), also related to pp125 FAK , since the specific amino acid sequence used to make antisera that recognized FAKB protein is missing in the predicted amino acid sequence of RAFTK protein (Fig. 2). Furthermore, unlike FAKB, RAFTK protein did not form stable complexes with the T-cell antigen receptor/CD3-linked tyrosine kinase ZAP 70 in T-cells, 3 indicating that RAFTK and FAKB are different proteins.
The chicken, human, and mouse focal adhesion kinases have been recently implicated as playing key roles in signal transduction pathways associated with extracellular adhesion molecules and with receptors for neuropeptide growth factors (14,15,52,53). Thus, based on its homology to pp125 FAK , one would expect RAFTK to participate in signaling pathways as well. The deduced 1009-amino acid sequence of RAFTK (with a calculated molecular mass of 120 kDa) contains a kinase domain and lacks a transmembrane region, myristylation sites, and SH2 and SH3 domains (Fig. 1B). To identify conserved regions within RAFTK between species that may have impor-3 J. Li, personal communication. tant functions, we have cloned the murine homolog of the human RAFTK cDNA. The sequence identity between the human RAFTK and murine Raftk cDNAs is 90% at the nucleotide level and 95.6% at the predicted amino acid level. In the kinase domain, 98.5% of the amino acids are identical (Fig. 2). Therefore, the RAFTK gene is highly conserved in human and rodent, again suggesting an important role in cell signaling func-tions. RAFTK has an insertion of an additional 4 amino acids between positions 76 and 81 (Gly 76 -Arg 77 -Ile 78 -Gly 79 ) compared with chicken, murine, and human pp125 FAK sequences (Fig. 2) (13,14,19). Amino acids corresponding to positions 292-320 of human pp125 FAK and amino acids corresponding to positions 850 -864 and 901-926 of chicken pp125 FAK are absent in the predicted RAFTK protein. Interestingly, like chicken pp125 FAK , the C terminus region of human RAFTK and mouse Raftk contains a proline-rich stretch (residues 690 -767). It has been shown that proteins containing proline-rich peptide motifs (such as Shc, p62, and ribonucleoprotein K) could serve as SH3 domain ligands and that the binding of these proteins to the Src SH3 domain was inhibited with a proline-rich peptide ligand (54). Furthermore, the predicted RAFTK protein, like the pp125 FAK protein, displays several unique features among the known tyrosine kinases. The primary sequence of RAFTK does not contain a signal peptide or a membrane-spanning region, and the protein is therefore presumed to be located in the cytoplasm. RAFTK lacks SH2 and SH3 domains, which are structural elements involved in protein-protein interactions (2, 47, 55-57), and does not exhibit significant homology to any known protein-tyrosine kinase beyond pp125 FAK outside of the catalytic domain (Figs. 2 and 3).  Fig. 6  Lack of SH2 and SH3 domains suggests that other regions within RAFTK protein are important for protein interaction and targeting. In the case of the pp125 FAK protein, it has been demonstrated by structural-functional analysis that 159 amino acids within the C terminus are essential as a "focal adhesion targeting" sequence (58). The homology between RAFTK and pp125 FAK within this region is 52%. The overall structure of RAFTK is characteristic of the pp125 FAK gene, with the catalytic domain flanked by large N-and C-terminal domains (Fig.  2). It has recently been reported that deletions of the N-or C-terminal noncatalytic domain of pp125 FAK including Tyr 397 did not abolish the kinase activity of pp125 FAK (59). Moreover, there is conservation of several tyrosine residues between RAFTK and pp125 FAK (Figs. 2 and 3), including Tyr 397 , which has been shown to be the major site of tyrosine phosphorylation in pp125 FAK protein (25).
RAFTK-specific mRNA expression was observed in human fetal tissues, being most abundant in brain (predominantly in the amygdala and hippocampus regions), and appeared to be developmentally up-regulated as demonstrated in the pattern of adult tissue expression (Figs. 7 (A-C) and 8). Within the hematopoietic system, in addition to peripheral blood leukocytes, a high level of specific mRNA expression of RAFTK was detected in B-cells and various megakaryocytic cell lines (data not shown). By using PCR, the specific mRNA expression of RAFTK was also detected in CD34 ϩ primary bone marrow progenitor cells, primary bone marrow megakaryocytes, and platelets (Fig. 9).
RAFTK is phosphorylated after thrombin treatment of CMK cells. Interestingly, FAK protein was also found phosphorylated on tyrosine after thrombin or collagen treatment of platelets (18). There is considerable homology in the thrombin receptors and considerable signal similarities in transduction mechanisms between platelets and megakaryocytes (60). Furthermore, bone marrow megakaryocytes in liquid culture stimulated with thrombin for 5 min revealed dramatic morphological changes reminiscent of those found in platelets, including shape change and organelle centralization that involved immature as well as mature cells (61). Megakaryocytes were also able to secrete ␣-granule proteins in the dilated cisternae of the demarcation membrane system (61).
The human RAFTK gene was found on chromosome 8 using DNAs from the somatic cell hybrid lines (Fig. 4). The signal observed in cell line 20 in mapping panel 2 suggested that a fragment of chromosome 8 is in the chromosome 20 cell line. Indeed, although cell line 20 contained the human NEFL gene, there was no evidence of chromosome 20 or a fragment of chromosome 20 in cell line 8 (Coriel Cell Institute for Medical Research). Indeed, the localization of RAFTK to chromosome 8 was confirmed using mapping panel 1. The human NEFL gene has been localized to chromosome 8p21 (62). Nfl, the murine homolog of human NEFL, has been mapped to mouse chromosome 14 and is within 3 cM of the Gnrh locus (GBASE). The close linkage of the mouse Raftk gene to Nfl (whose NEFL homolog is on human chromosome 8p21) suggested that the human RAFTK gene may be mapped to chromosome 8 based on homology between human and mouse chromosomes (62). Therefore, we predict that the human RAFTK gene will be localized to chromosome 8p21. We have mapped the mouse Raftk gene to chromosome 14 using a (C57BL/6J ϫ M. spretus)F 1 ϫ M. spretus backcross. The position of mouse Raftk was confirmed by RI line mapping using the BxD RI lines. The Raftk gene was also shown to be closely linked to Gnrh, whose human homolog (LHRH (luteinizing hormone-releasing hormone)) has been mapped to human chromosome 8p21-11.2 (42).
There has been considerable interest in the role of pp125 FAK in signaling pathways in a variety of normal and transformed cells in response to different soluble and cell-surface stimuli. Future studies will aim to gain insights into the function of RAFTK in these signal transduction pathways, particularly distinguishing a role for RAFTK versus pp125 FAK in these systems.