and Characterization - Molecular and Cellular Biology - American ...

MOLECULAR AND CELLULAR BIOLOGY, Nov. 1988, p. 48964903 0270-7306/88/114896-08$02.00/0 Copyright © 1988, American Society for Microbiology

c-kit Protein,

a

Vol. 8, No. 11

Transmembrane Kinase: Identification in Tissues and Characterization

SADHAN MAJUMDER, KAREN BROWN, FEI-HUA QIU, AND PETER BESMER*

Laboratory of Molecular Oncology, Graduate Program in Molecular Biology, Memorial Sloan-Kettering Cancer Center and Cornell University Graduate School of Medical Sciences, New York, New York 10021 Received 18 April 1988/Accepted 1 August 1988

The proto-oncogene c-kit encodes a transmembrane kinase which is related to the receptors for colonystimulating factor type 1 and platelet-derived growth factor, as well as to the immunoglobulin superfamily. Antibodies specific for the kinase domain of the P80 gag-kit protein of the Hardy-Zuckerman 4 feline sarcoma virus were prepared. These kit-specific antibodies were used to identify and characterize the c-kit protein in cat brain tissue. The c-kit protein product displays an autophosphorylating activity in immune complex kinase assays, and, in turn, this activity was used to identify the c-kit protein in different tissues. In cat brain, a single 145-kilodalton (kDa) glycoprotein was detected. Its N-linked carbohydrates were found to be sensitive to digestion with the endoglycosidases (neuraminidase, endoglycosidase F, and endoglycosidase H), indicating hybrid and/or complex and high-mannose structures. A partial purification of the c-kit protein was achieved by wheat germ agglutinin affinity chromatography, and the autophosphorylating activity of the partially purified c-kit protein was characterized and found to be specific for tyrosine. The kit antibodies cross-react with the murine c-kit protein product, and variant c-kit proteins in different mouse tissues were identified, with sizes of about 145 kDa (brain), 160 kDa (spleen), and 150 kDa (testis).

Factors that control cell proliferation, differentiation, and specific cellular functions often are small proteins. To initiate a mitogenic signal, a growth factor binds to a specific receptor which is located on the cell surface. Several cell surface receptors of growth factors and hormones have been characterized in recent years; they include the receptors for insulin, epidermal growth factor, low-density lipoprotein, nerve growth factor, insulinlike growth factor, colony-stimulating factor type 1 (CSF-1) and platelet-derived growth factor (PDGF) (9, 16, 33, 35). The receptors of these hormones and factors have common characteristic topological features: they contain an extracellular domain which often is glycosylated and binds the ligand, a single transmembrane segment (20 to 30 amino acid residues), and an intracellular domain, which may contain a tyrosine-specific protein kinase. Upon binding of the ligand to the extracellular domain of a transmembrane receptor kinase, the kinase becomes activated; this leads to autophosphorylation as well as the phosphorylation of cellular substrates (5, 15, 28). Experiments with mutant receptors with inactive kinases indicate that this activity is crucial for the transmission of a signal from the cell surface to the nucleus; however, the detailed steps, i.e., specific substrates and second messenger systems, involved in this process are less well understood (6, 7, 14, 28). Ligand binding is also known to initiate downmodulation of cell surface receptors by endocytosis via clathrin-coated pits (4, 5). The down-modulation of ligandbound receptor is an important mechanism for controlling the incoming signal. Elucidation of the mechanism of how signals are transmitted from the cell surface to the nucleus, as well as identification of new signal receptor systems and elucidation of their biological function, is of great importance. Proto-oncogenes are thought of as elements of signal transduction pathways, and their oncogenic activation is believed to be due to the constitutive expression of their *

signaling mechanism. Transformation of fibroblasts by simian sarcoma virus and Parodi-Irgens feline sarcoma virus (PI-FeSV) is facilitated by the expression of v-sis (PDGF) in an autocrine mechanism (17, 22). Oncogenic versions of growth factor receptors, which are obtained by various mutations of the receptor gene, have a constitutively activated kinase and appear to function in the absence of a ligand. In avian leukosis virus-induced erythroblastosis, the epidermal growth factor receptor (c-erbB) is activated by N-terminal truncation (24). In the two transduced versions of v-fms in the McDonough strain (SM) of FeSV (SM-FeSV) and the Hardy-Zuckerman 5-FeSV (HZ5-FeSV), C-terminal truncation of the CSF-1 receptor (c-fms) appears to play a contributing role in the oncogenic activation of this receptor (2, 9, 30). In addition, overexpression of tyrosine kinase receptors such as c-erbB2 as seen in mammary carcinoma or NIH 3T3 cells may provide a sufficient amplification of a basal kinase activity to function in signal transduction (18, 21, 32). In the recent past, several new receptor kinase oncogenes have been identified and characterized. Although the ligands of the respective proto-oncogenes are not known, these receptors provide the opportunity to characterize new signaling mechanisms. The HZ4-FeSV is an acute transforming feline retrovirus which was isolated from a feline leukemia virus (FeLV)-associated feline fibrosarcoma (2). The HZ4FeSV originated by transduction of feline c-kit sequences with FeLV. The v-kit oncogene of the HZ4-FeSV encodes a 370-amino-acid polypeptide chain which displays homology with the enzymatic domain of tyrosine-specific protein kinases, and recently the P80 gag-kit transforming protein found in HZ4-FeSV-infected cells was shown to display a tyrosine-specific protein kinase activity (S. Majumder and P. Besmer, manuscript in preparation). The kit kinase is most closely related to the kinases of the transmembrane receptors for CSF-1 (CSF-1R) and PDGF (PDGFR) (2, 35). The kinase domains of v-kit CSF-1R and PDGFR display a unique tripartite structure not known from other kinases,

Corresponding author. 4896

VOL. 8, 1988 A

Sac

Sal

Xho I v-kit

F777M ____

___

__

kit 1 pE _

J

cloning site

/~~~~~~~~~~~~~ pATH

kit2 pE

45

c-kit PROTEIN IDENTIFICATION AND CHARACTERIZATION

B

K-2 kd

i UUKU pkdEK1pEK2

-

U

-'3

33 i6-

k -W

TH1pE-kit

FIG. 1. Construction and expression of E. co fusion proteins. (A) Schematic representation of the pA' t expresson vector and the v-kit segments used for the constru rial expression of trpE-kit fusion proteins. E. coli the expression plasmids pEKl and pEK2, as well aas pATH11, were grown in the presence (lanes I) and absence (lanies U) of the trp inducer 3p-indoleacrylic acid, and total cell lysattes were analyzed by SDS-PAGE. The Coomassie blue-stained ge1 is shown. The trpE-kit fusion proteins are indicated by arrows, and protein size markers are indicated in kilodaltons (kd).

icHBn1.0(B Barcreg

i.e., a 70- to 100-amino-acid insert separ,ates the ATPbinding domain from the rest of the enzyme irn these kinases The recent characterization of cDNAs of the lproto-oncogene c-kit of mouse and human origin predicts tha t c-kit encodes a transmembrane kinase similar to CSF-1R a] nd PDGFR (25 36). The outer cellular domain of c-kit, sinnilar to that of CSF-1R and PDGFR, was found to be m,ade up of five immunoglobulinlike repeats, indicating that c-kit is a member of the immunoglobulin superfamily. The xunique relationship of c-kit, CSF-1R, and PDGFR in thie extracellular domain, as well as in the kinase domain, sugg'ests a common evolutionary origin for these molecules. We now have prepared v-kit-specific antibodies and have ch aracterized the c-kit protein products in cat and mouse tissu es. MATERIALS AND METHOD'S Cells and animals. Normal CCL64 mink li Ling cells, HZ4FeSV-infected mink cells (HZ4 Cl 2 cells), and SM-FeSV infected mink cells (mink SM-FeSV Cl 1!5-1 cells) were grown in Dulbecco modified Eagle medium supplemented with 10o calf serum (2, 27). Cats were purchased from Quaker Labs, and NFS mice were obtained from our own colony (P. O'Donnell, Sloan-Kettering Instit:ute). Construction and expression of trpE-kit vecltors. Growth of bacteria, plasmid preparations, subcloning porocedures, and transformations were done as described 4 (23). The trpE expression vector pATH11 was kindly proovided by T. J. Korner and A. Tzagalof, Columbia Universiity, New York, N.Y. In the pATHll vector the pUC12 polIylinker cloning site is inserted 13 amino acids before the C Iterminus of the trpE gene (T. J. Korner and A. Tzagalof, pe rsonal communication). The HZ4-FeSV v-kit Sac-Xho (1.7 kilobases) and Sal-Xho (1.0 kilobase) restriction fragments Nwere subcloned into the polylinker sequence of the pATHIll vector and characterized (2) (Fig. 1). For expression of the fusion proteins, the method of Kleid et al. was use(d (19). Cultures of Escherichia coli HB101 containing the e xpression plas-

4897

mids were grown to stationary phase in M9 medium containing 1% Casamino Acids (Difco Laboratories), 10 p.g of thiamine per ml, 50 ,ug of tryptophan per ml, and 50 jig of ampicillin per ml, and then diluted 1:25 into the above medium without tryptophan and grown to an optical density at 600 nm of 0.4. Indoleacrylic acid was added (5 and incubation was continued for 8 h. Whole-cell extracts were obtained by solubilization of cell pellets in 10 mM sodium phosphate-8 M urea-1% mercaptoethanol. For preparative partial purification of the fusion proteins, an insoluble fraction was prepared. Cell pellets were suspended in one-fifth of the original volume in 50 mM Tris (pH 7.5)-S5 mM EDTA-3 mg of lysozyme per ml and let sit on ice for 2 h. Sodium chloride and Nonidet P-40 then were added to final concentrations of 0.3 M and 0.7%, respectively, and the mixture was kept on ice for 30 min. After sonication, the insoluble protein was collected by centrifugation at 15,000 x g for 10 min. Protein pellets were washed once in 1 M NaCl-10 mM Tris (pH 7.5) and once in 10 mM Tris (pH 7.5) and then solubilized in sodium dodecyl sulfate (SDS)-gel sample buffer (62.5 mM Tris [pH 6.8], 2% SDS, 2% mercaptoethanol, 10%o glycerol). The proteins were subjected to SDSpolyacrylamide gel electrophoresis (PAGE) (10% polyacrylamide gels), and protein bands were then visualized by staining with Coomassie blue (analytical gels) or by the sodium acetate shadowing technique (preparative gels) (13). The fusion proteins from preparative gels were excised and used for immunization of rabbits by standard procedures. Antibody production in rabbits was monitored by immunoprecipitation of the P80 gag-kit protein from extracts of [35S]methionine-labeled mink HZ4-Cl2 cells. Metabolic labeling of cells and immunoprecipitation analysis. For labeling, subconfluent cultures were rinsed with phosphate-buffered-saline and incubated for 1 h at 37°C with methionine-free Dulbecco modified Eagle medium supplemented with 10% dialyzed calf serum. [35S]methionine (200 p.Ci/ml) was added, and incubation was continued for 3 h. The cells were then rinsed twice with phosphate-buffered saline and lysed in lysis buffer (50 mM Tris [pH 7.4], 150 mM NaCl, 20 mM EDTA, 1% Triton X-100, 1% deoxycholate, 0.1% SDS, 1 mM phenylmethylsulfonyl chloride, 2% Trasylol), and the extract was clarified at 100,000 x g for 1 h. For preclearing, extracts containing equal amounts of trichloroacetic acid-precipitable radioactivity in 300 ,ul of lysis buffer were incubated for 3 h at 4°C, with 20 ,ul of normal rabbit or preimmune serum and 7.5 ,ul of preswollen protein A-Sepharose beads (Pharmacia, Inc.), and then centrifuged at 5,000 x g. For immunoprecipitation, the appropriate amounts of antiserum (10 ,ul for a-kit sera, 1.5 p.l for a-FeLV serum [obtained from W. D. Hardy, Jr.], 0.5 p.1 for a-FeLV p15 monoclonal antibody, and 20 ,ul of the corresponding preimmune rabbit and normal rabbit sera) were added to precleared extract and incubated at 4°C. After 1 h, 7.5 ,u1 of preswollen protein A-Sepharose beads was added to the reactions with rabbit sera, and 7.5 p.l of rabbit a-rat immunoglobulin G-coated protein A-Sepharose was added to reactions with rat monoclonal antibody, and incubation was continued for another 2 h with gentle shaking. Immunoprecipitates were collected by centrifugation, washed five times in lysis buffer, suspended in SDS-gel sample buffer (62.5 mM Tris [ph 6.8], 2% SDS, 2% mercaptoethanol, 10% glycerol), boiled for 3 min, and analyzed by SDS-PAGE and autoradi-

pg/ml),

ography. Immune complex kinase assays with tissue samples. Immune complex kinase assays with tissue samples were done essentially as described by Cooper et al. (9). A 300- to

4898

MAJUMDER ET AL.

400-mg samples of tissue was homogenized in a polytron homogenizer in 2 ml of modified lysis buffer (50 mM Tris [pH 7.4], 150 mM NaCl, 20 mM EDTA, 1% Triton X-100, 1% deoxycholate, 0.1% SDS, 1 mM phenylmethylsulfonyl fluoride, 20 ,ug of leupeptin per ml, 1 mM sodium vanadate), filtered through a 200-gauge mesh, and clarified at 100,000 x g for 45 min. For immunoprecipitation, 1 ml of extract (containing equal amounts of protein) was first incubated with 50 ,ul of preimmune serum and 25 ,ul of preswollen protein A-Sepharose for 45 min at 4°C and then cleared in an Eppendorf centrifuge. The supernatant was incubated for 45 min with 50 ,ul of anti-kit antibody and 25 ,ul of preswollen protein A-Sepharose. The protein A-Sepharose-bound immune complexes were then washed three times with wash 1 (phosphate-buffered saline containing 0.5% Triton X-100, 1 mM sodium vanadate, 1 mM phenylmethylsulfonyl fluoride, and 20 ,ug of leupeptin per ml), twice with wash 2 (20 mM Tris [pH 7.5], 0.5 M NaCl, 5 mM EDTA, 1 mM sodium vanadate, 1 mM phenylmethylsulfonyl fluoride, 20 ,ug of leupeptin per ml), and once with 20 mM piperazine-N, N'-bis(2-ethanesulfonic acid) (PIPES; pH 7.2). The pellet was suspended in 200 ,ul of 20 mM PIPES (pH 7.2)-10 mM MnCl2-2.5 ,uM ATP containing 320 ,uCi of [_y-32P]ATP (3,000 Ci/mmol). The reaction was incubated for 10 min at 30°C and terminated by adding 300 RI of 2 x SDS sample buffer added, and the mixture was boiled and then analyzed by SDSPAGE and autoradiography. For reprecipitation, 100 ,ul of the denatured reaction mixture was diluted into 1 ml of lysis buffer and incubated with 50 1±1 of preimmune or anti-kit antiserum-bound protein A-Sepharose for 2 h at 4°C, and the immunocomplexes were washed and analyzed by SDSPAGE as described above. Characterization of N-linked oligosaccharide chains. Immune complexes containing 32P-labeled c-kit protein were prepared by precipitation of cat brain extracts (or wheat germ affinity-purified material; see below) with a-kit2 antibody, in vitro immune complex kinase reaction, and reprecipitation as described above. The immune complexes were washed three times with wash 1 and twice with wash 2 and then, alternatively, with 50 mM sodium acetate (pH 5.5) (buffer A), 150 mM sodium citrate (pH 5.3) (buffer B), or 100 mM sodium phosphate (pH 6.7)-50 mM EDTA-1% Triton X-100-0.1% SDS-1% P-mercaptoethanol (buffer C). For digestion with neuraminidase, immune complexes were suspended in buffer A containing 0.2 U of neuraminidase from Clostridium perfringens (Sigma Chemical Co.) (EC 3.2.1.18), for digestion with endoglycosidase H, they were suspended in buffer B containing 0.004 U of endoglycosidase H from Streptomyces griseus (Sigma) (EC 3.2.1.96) and for digestion with endoglycosidase F, they were suspended in buffer C containing 0.4 U of endoglycosidase F from Flavobacterium meningosepticum (Boehringer Mannheim Biochemicals) (EC 3.2.1.96), respectively (26, 34). The reactions were incubated for 18 h at 37°C, stopped by the addition of SDS-sample buffer, and analyzed by SDS-PAGE. Samples without the addition of enzymes were treated

identically.

Wheat germ agglutinin affinity purification. Cleared tissue (500 ,ul) was applied twice to 1 ml of a wheat germ agglutinin agarose column (Vector Labs, Inc.), which was previously equilibrated in lysis buffer, as described (12). The column was washed with 25 ml of lysis buffer, and the glycoproteins were eluted with 2 ml of lysis buffer containing 0.3 M N-acetylglucosamine and 5 mM EDTA. Immune complex kinase assays were then carried out as described extract

MOL. CELL. BIOL.

above. The glycoprotein fraction was stored in liquid N2 for more than 1 month without detectable loss of kinase activity. Phosphoamino acid analysis. The phosphoamino acid analysis was carried out essentially as described by Cooper et al. (9). Protein bands were cut out from the dried gel and hydrolyzed in the presence of 5.5 N HCI at 110°C for 1 h. Amino acid hydrolysates were lyophilized, rehydrated with water, and subjected to Dowex AG7-X8 (Bio-Rad Laboratories) chromatography, along with unlabeled phosphoamino acids as markers. The phosphoamino acids were eluted with 0.5 N HCl, lyophilized, suspended in water, and analyzed by two-dimensional electrophoresis on thin-layer cellulose plates. Separation in the first dimension was done in 88% formic acid-glacial acetic acid-water (ratio, 25:78:897) (pH 1.9) at 1,600 V for 20 min, and separation in the second dimension was done in pyridine-glacial acetic acid-water (ratio, 1:10: 189) (pH 3.5) at 1,300 V for 16 min. The cellulose plates were then dried, developed with ninhydrin for visualization of the phosphoamino acid markers, and autoradiographed. RESULTS Generation of v-kit specific antibodies. To identify and characterize the protein product of the proto-oncogene c-kit, we needed kit-specific antibodies at hand. To prepare kitspecific antibodies, we constructed bacterial expression vectors in which v-kit sequences were fused with the E. coli trpE gene as outlined in Fig. 1. In the pATH11 vector, which was used for this purpose, the trpE gene is inducible and contains the trp operator, promoter, attenuator, and trpL leader sequences. Two restriction fragments, Sac-Xho (1.7 kb) and Sal-Xho (1.0 kb), encoding 298 and 63 amino acids of C-terminal v-kit sequences, respectively, were inserted into the cloning site near the C terminus of the E. coli trpE coding sequences in the vector pATH11. In Fig. 1B an SDS-PAGE analysis of total cellular proteins of bacteria containing the plasmids pTEK1 and pTEK2 is shown. trpE fusion proteins of 44 and 65 kilodaltons (kDa) are apparent. The two fusion proteins were partially purified, and antibodies were raised in rabbits. The antisera then were characterized by immunoprecipitation analysis (Fig. 2). The P80 gag-kit protein was precipitated from extracts of [35S]methionine-labeled cells infected with HZ4-FeSV by anti-FeLV gag antibody and a monoclonal antibody specific for the FeLV gag p15 protein. A protein of identical size was identified with a-kitl and a-kit2 sera. In extracts made from uninfected CCL 64 mink cells, the P80 protein was not seen. Furthermore, preimmune sera did not precipitate this protein. These results indicated that both a-kit sera, a-kitl and a-kit2, precipitate the P80 gag-kit transforming protein and, because of the way they were generated, recognize v-kit-specific epitopes. The kinases of the oncogenes kit and fms are closely related on the basis of amino acid homology. Therefore, it was of interest to determine whether the two a-kit sera cross-react with the SM-FeSV v-fms protein product. In the experiments shown in Fig. 2B, the v-fms protein products in extracts made from SM-FeSV-infected cells were not detected with the a-kit sera, thus indicating that the a-kit sera do not cross-react with the v-fms protein products. Identification of the c-kit protein product in cat brain. Our previous investigation of c-kit RNA expression in cat tissues indicated low levels of c-kit expression in several tissues, particularly the brain (25). It seemed most reasonable to assume that brain tissue would express the c-kit protein product as well. In recent experiments, we had obtained


VOL. 8, 1988

a- kit2 a-FeLV MAB pre- imm. pre - immr. T N T N T N T NT N T N

A.

a-p5a-Bkpie t 1

4899

B. a-fins a-kit 1 a-kit 2 a b c PI "P I1

kd 200 200

-

116 92-

km I

_O

66 -

116

-

92 45-

66 FIG. 2. Specificity analysis of anti-v-kit antibodies by immunoprecipitation. (A) Extracts of mink HZ4-C1 2 cells (T) and mink CCL64 cells (N), metabolically labeled with [35S]methionine, were immunoprecipitated with anti-FeLV antiserum, anti-p15 monoclonal antibody (lanes MAB), kit-1 preimmune and immune serum, and kit-2 preimmune and immune serum. The immunoprecipitates were separated in an SDS10%o polyacrylamide gel, and bands were visualized by fluorography. (B) Extracts of SM-FeSV mink cells, metabolically labeled with [35S]methionine, were immunoprecipitated with threefms sera [a-fms MAB (lane a), a-fms(R1B6) (lane b), and a-fms(R2BA) (lane c)] and with the a-kitl and a-kit2 immune sera (I) and the respective preimmune sera (P) and analyzed by SDS-PAGE (7% acrylamide). Protein size markers are indicated in kilodaltons (kd).

evidence for immune complex tyrosine kinase activity of the HZ4-FeSV P80 gag-kit protein by using FeLV gag and v-kit antibodies, and it was very likely that a similar tyrosinespecific protein kinase activity could be ascribed to the c-kit protein product as well. In vitro protein kinase reactions provide an efficient way to introduce high-specific-activity label into unlabeled proteins (26). We therefore attempted to identify the c-kit product in brain tissue by using the immune complex kinase procedure. Immune complexes were formed by using cat brain extracts and the a-kit2 serum, and a kinase reaction was carried out adding [y-32P]ATP and manganese chloride, after extensive washing of the protein A-Sepharose-bound immune complexes. The SDS-PAGE analysis of the products of the in vitro kinase reaction is shown in Fig. 3a. A high background of labeled proteins was apparent in the sample with the immune and the preimmune sera, although a specific band representing a protein of 140 to 150 kDa could be discerned in the sample with the immune serum. To achieve a partial purification of the presumptive c-kit protein product, the reaction mixture of the kinase reaction was dissociated by boiling in sample buffer and subjected to a second precipitation with the two a-kit antibodies. SDS-PAGE analysis of the samples reprecipitated with a-kitl, as well as with a-kit2, revealed a single band corresponding to a protein of 145 kDa (Fig. 3b). Carbohydrate modifications of the c-kit protein. The nucleotide sequence of a murine c-kit cDNA predicts a c-kit protein product with a molecular mass of 110 kDa and with features characteristic of single polypeptide chain transmembrane kinases. In the predicted extracellular domain there are nine sites for N-linked glycosylation, and it seemed possible that the difference between the predicted molecular mass of the c-kit protein and the observed molecular mass of the 145-kDa protein we had identified was in part due to carbohydrate modifications. To investigate this possibility, immune complexes containing 32P-labeled c-kit protein obtained from autophosphorylation reactions were digested

1 kd

b

2

1

C.

2 3 4

1

2

VR

2v 000:V :js

,

aw

-*-

92 t5

reprecip.

wga

FIG. 3. Identification of a 145-kDa c-kit glycoprotein in cat brain tissue and partial purification. (a) Identification of the c-kit protein by immune complex kinase assay. Extracts from cat brain were precipitated with preimmune serum and a-kit2 antiserum (lanes 1 and 2), the protein A-Sepharose-bound immune complexes were incubated with [y-32P]ATP, and the reaction products were analyzed by SDS-PAGE (7.5% acrylamide). Reprecipitation analysis. Samples were denatured in sample buffer, diluted in lysis buffer (1:10), immunoprecipitated with a-kitl serum (lane 2), a-kit2 serum (lane 4), and the respective preimmune sera (lanes 1 and 3), and analyzed by SDS-PAGE. (c) Wheat germ agglutinin affinity purification. Cat brain tissue extract was subjected to wheat germ agglutinin (wga) chromatography, immunoprecipitated with preimmune (lane 1) and a-kit2 (lane 2) sera, incubated with [y-32P]ATP in an in vitro kinase reaction, and analyzed by SDS-PAGE.

4900

MOL. CELL. BIOL.

MAJUMDER ET AL. end OF ..

..

endoH;

....-

n euld _

+

k,t

200,,-.

I 92

. a

--

2

3 4 ..7F

FIG. 4. Characterization of N-linked carbohydrate modifications of the c-kit protein. c-kit protein from cat brain tissue was immunoprecipitated, 32P-labeled in an in vitro kinase reaction, reprecipitated, and then digested with endoglycosidase F (lane 2), endoglycosidase H (lane 4), and neuraminidase (lane 6) (respective enzyme-minus controls are in lanes 1, 3, and 5) and analyzed by SDS-PAGE. Analysis of the c-kit protein under nonreducing and reducing conditions, respectively, is shown in lanes 7 and 8. Protein size markers are indicated in kilodaltons (kd).

with the endoglycosidases neuraminidase, endoglycosidase H, and endoglycosidase F and analyzed by SDS-PAGE (Fig. 4). These three enzymes are known to hydrolyze glycosidic bonds of sialic acid residues of complex carbohydrates and N-acetylglucosamine residues of high-mannose and of highmannose and complex N-linked oligosaccharides, respectively (20, 34). The 145-kDa c-kit protein was sensitive to all three enzymes. These results suggested that the 145-kDa c-kit protein is a glycoprotein and contains hybrid and/or complex and high-mannose N-linked oligosaccharide structures. With the knowledge that the 145-kDa c-kit protein is

a

glycoprotein, we attempted a partial purification of the c-kit protein product by using lectin affinity chromatography (12). Cat brain extracts prepared in lysis buffer were applied to a wheat germ Sepharose column, and the glycoproteins were eluted with N-acetylglucosamine. An immune-complex kinase assay was then performed with the glycoprotein fraction, and the reaction products were analyzed by SDSPAGE. A single major band of 145 kDa was obtained, in the absence of a reprecipitation step, indicating a partial purification of the c-kit protein product (Fig. 3c). In the extracellular domain, the c-kit protein contains 12 cysteines. We attempted to determine whether any of these cysteines form intermolecular disulfide bonds. For this purpose 32P-labeled c-kit protein obtained by an in vitro kinase reaction of immunoprecipitates from lectin-purified cat brain extracts was analyzed by SDS-PAGE in the absence or the presence of the reducing agent P-mercaptoethanol (Fig. 4, lanes 7 and 8). The reduced and the nonreduced form of the c-kit protein both had similar mobilities, indicating the absence of intermolecular disulfide bonds. Characterization of the c-kit protein kinase. To determine the specificity of the c-kit-associated protein kinase activity, protein bands from acrylamide gels were hydrolyzed in 5.5 N HCl and the hydrolysates were analyzed by two-dimensional

electrophoresis (Fig. SD). The analysis of the c-kit protein obtained by double precipitation (Fig. 3b) revealed phosphorylations on serine, threonine, and tyrosine (Fig. SD, panel a). In contrast, the c-kit protein obtained by wheat germ affinity chromatography (Fig. 3c) contained only phosphotyrosine (Fig. 5D, panel b). These results indicated that the c-kit protein product has a tyrosine-specific protein kinase activity. The serine and threonine phosphorylations seen in the protein obtained by direct immunoprecipitation can be ascribed to contaminating serine-threonine protein kinases present in the immunoprecipitate. The a-kit-i serum had been shown to precipitate the 145-kDa c-kit protein (Fig. 3b). However, this antibody did not facilitate the autophosphorylation of the c-kit protein in brain extracts (data not shown) and the wheat germ affinitypurified material (Fig. 5C), suggesting that the a-kit-i antibody is interfering with the autophosphorylating activity of the c-kit protein. An investigation of the cation requirements of the c-kit kinase is shown in Fig. 5A. Maximal activity was obtained with Mn2"; in contrast, Mg2+ gave a 5- to 10-fold-lower activity and Ca2+ was ineffective as a cation in this reaction. Analysis of the kinetics of in vitro phosphorylation of the c-kit protein is shown in Fig. 5B and indicated a linear relationship between the extent of phosphorylation and time. Variant c-kit proteins in mouse tissues. The analysis of c-kit RNA in cat and mouse tissues indicated expression in the brain and, to a somewhat lesser extent, in hematopoietic tissues and testis (25; data not shown). To determine whether the kit antibodies cross-react with the murine c-kit protein and to analyze c-kit proteins in mouse brain, spleen, thymus, and testis, lectin-purified extracts were prepared for in vitro kinase reaction and analyzed by SDS-PAGE. The results (Fig. 6) indicate a distinct glycoprotein of 145 kDa in mouse brain and faint bands indicating proteins of 150 and 160 kDa in testis and spleen, respectively. These results show that the kit antibodies cross-react with the murine brain c-kit protein, and they revealed less abundant variant c-kit proteins in the testis and spleen. For reasons unknown at present, the level of expression of the variant c-kit proteins in the testis and spleen, determined by using the kinase assay, is significantly lower than that expected from the RNA data. Attempts to identify the c-kit protein product in tissues by Western immunoblotting were unsuccessful, presumably owing to the low abundance of this protein. DISCUSSION The c-kit protein product, predicted from the nucleotide sequence of c-kit cDNAs, has the features of a transmembrane kinase (25, 36). We have prepared two antibodies which are specific for the viral oncogene v-kit to characterize the protein product of the proto-oncogene c-kit. The basis of our identification of the c-kit protein products in extracts from normal tissues was an autophosphorylating activity detected in immune complexes that is associated with the 145-kDa c-kit protein. The c-kit kinase is specific for tyrosine residues and displays optimal activity with manganese as a divalent cation. Although both of the a-kit sera precipitated the c-kit protein, only one facilitated autophosphorylation in the immune-complex reaction. The kinase-positive serum a-kit2 was made against epitopes contained within c-kit amino acids 639 to 925 (2, 25). The kinase-inhibiting serum, on the other hand, recognizes a subset of the former epitopes at the C terminus of the kinase (c-kit amino acids 873 to 925). The neutralization of kinase activity by a-kitl antiserum was


VOL. 8, 1988

A.

2

+ + +t 4- + + + rC F a C

B.

C. Uf)

C.~

a -kit 1

Cd0

P I ;P

4901

D. K 2 I

a.

kL

S_ _

a

145 kd--w

9-2

*

a

b. -0

s

FIG. 5. Characteristics of the in vitro c-kit kinase activity. (A) Divalent cation requirements. Wheat germ agglutinin affinity-purified brain extract was immunoprecipitated and incubated with [y-32PIATP in an in vitro kinase reaction in the presence of Mn2" (10 mM), Mg2+ (10 mM), Ca2+ (10 mM), and Mn2+-Mg2+ (10 mM, 10 mM) and analyzed by SDS-PAGE. Protein size markers are indicated in kilodaltons (kd). (B) Kinetics of the c-kit autophosphorylation. Wheat germ agglutinin affinity-purified brain extract was immunoprecipitated, and kinase reaction mixtures were incubated for the indicated times (minutes). (C) Inhibition of c-kit autophosphorylation by a-kitl antiserum. Wheat germ agglutinin affinity-purified brain extract was precipitated with a-kitl and x-kit2 immune sera (I) and the respective preimmune sera (P), an in vitro kinase reaction was performed, and the products were analyzed by SDS-PAGE. (D) Phosphoamino acid analysis of 32P-labeled c-kit protein. Amino acid hydrolysates were analyzed by two-dimensional electrophoresis on cellulose plates. In the first dimension, migration was from right to left, and in the second dimension, it was from bottom to top. Migration of phosphoserine (S), phosphothreonine (T), and phosphotyrosine (Y) markers is indicated. The 32P-labeled phosphoamino acids were visualized by autoradiography. Analysis was peiformed with (a) c-kit protein obtained by double precipitation and (b) c-kit protein obtained by wheat germ agglutinin affinity purification.

0

D

c, )

cn

co)

E

0

-r-

-o

:

0

U.r

pr

kd 200-

1164

92-

66-

rW.6

FIG. 6. Identification of variant c-kit protein products in different tissues. Wheat germ agglutinin affinity-purified extracts obtained from brain, testis, spleen, and thymus tissues of NFS mice were precipitated with a-kit2 serum, and the immune complexes were processed for in vitro phosphorylation and analysis by SDSPAGE. Protein size markers are indicated in kilodaltons (kd). The arrows indicate the relative mobilities of the c-kit proteins. mouse

not unexpected, since site-specific antibodies directed against the homologous C-terminal domain of the kinases src and fes display the same property (11, 31). Transmembrane kinases such as the insulin receptor and PDGFR display enhanced kinase activity in vitro in response to ligand binding (3, 29). With insulin, this stimulation is three- to fourfold, and with PDGFR it is about twofold. The c-kit transmembrane kinase most probably functions as a receptor for an as yet unknown ligand. The c-kit-associated kinase activity we have identified by in vitro immunecomplex autophosphorylation could be due to the basal activity of the c-kit receptor or, alternatively, to receptor aggregation by noninhibiting antibodies. The extracellular segment of the c-kit protein contains five immunoglobulinlike domains (25). Four of these domains contain pairs of cysteines which are spaced by 40, 50, 59, and 60 amino acids. These cysteine pairs are an important feature of immunoglobulin structure and form intramolecular disulfide bonds to secure the immunoglobulin folds. Interestingly, a fifth immunoglobulin-related segment (amino acids 320 to 415) lacks these cysteines (a conserved feature in c-kit, PDGFR, and CSF-1R). In domains II and V, two additional cysteines are predicted which could be involved in inter- or intramolecular disulfide bond formation. The electrophoretic behavior of the c-kit protein under nonreducing conditions indicates that there are no intermolecular disulfide bonds which would give rise to c-kit dimers or oligomers. This result then may imply that the four additional cysteines in domains II and V form intramolecular disulfide bonds as well (Fig. 7). c-kit was shown to contain carbohydrate modifications, and we presume that they reside in the N-terminal extracel-

4902

MAJUMDER ET AL.

MOL. CELL. BIOL.

Cell Membrane

I

11

III

IV

V

NH2 \

\

Cytoplasm

FIG. 7. Model tor the structure of the c-kit protein. The five immunoglobulinlike domains (I to V) are shown as loops. Propo5ed intramolecular disulfide bonds are indicated by -S-S-, sites for N-linked glycosylations are shown by Y, and the split kinase is shown as two open boxes.

lular domain of the protein (Fig. 7). The brain c-kit protein is sensitive to neuraminidase, endoglycosidase H, and endoglycosidase F, indicating hybrid N-linked oligosaccharides. In contrast, the mature CSF-1 and PDGF receptors contain complex N-linked oligosaccharides (10, 26). The identification of different forms of the c-kit protein in mouse brain, spleen, and testis is of interest. The variant c-kit proteins all have a functional kinase, because of the assay that was used for their detection. The basis of their differing size may be due to different carbohydrate modifications, ubiquitin modifications, or, possibly, an altered primary amino acid sequence because of alternate RNA splicing. Alternatively, they might represent protein products of a related gene. So far, these proteins have not been characterized. The c-kit RNAs in the three tissues are of identical size as judged by RNA blot analysis, and, therefore, alternative splicing schemes should not affect the overall size of the c-kit RNAs. The distinct sizes of these c-kit proteins may reflect a differing cellular association of these proteins in the three tissues. It then seems possible that the c-kit ligand interacts with multiple cell types and displays pleiotropic functions in cell proliferation and/or differentiation processes. In a recent paper, Yarden et al. reported the identification of a 145-kDa cell surface-associated tyrosine kinase in a human glioblastoma cell line by using c-kit peptide antibody (36). ACKNOWLEDGMENTS We thank Carl Rettenmier and Charles Sherr for fms antibodies, Michelle Cervonne for providing tissues, the Viral Oncology Group and Ora Rosen for many valuable discussions, and Prabir Ray for comments on the manuscript. This work was supported by American Cancer Society grant MV246 and by Public Health Service grant CA 32926 from the National Institutes of Health. S.M. acknowledges the receipt of a postdoctoral fellowship from the Exxon Corporation. LITERATURE CITED 1. Besmer, P., E. Lader, P. C. George, P. J. Bergold, F.-H. Qiu, E. E. Zuckerman, and W. D. Hardy. 1986. A new acute transforming feline retrovirus with fms homology specifies a Cterminally truncated version of the c-fms protein that is different from SM-feline sarcoma virus v-fms protein. J. Virol. 60:194203. 2. Besmer, P., J. E. Murphy, P. C. George, F. Qiu, P. J. Bergold, L. Lederman, H. W. Snyder, D. Brodeur, E. E. Zuckerman, and W. D. Hardy. 1986. A new acute transforming feline retrovirus and relationship of its oncogene v-kit with the protein kinase gene family. Nature (London) 320:415421. 3. Bishayee, S., A. H. Ross, R. Womer, and C. Scherr. 1986. Purified human platelet derived growth factor receptor has ligand-stimulated tyrosine kinase activity. Proc. Natl. Acad.

Sci. USA 83:6756-6760. 4. Brown, M. S., R. G. W. Anderson, and J. L. Goldstein. 1983. Recycling receptors: the roundtrip itinerary of migrant membrane proteins. Cell 32:663-667. 5. Carpenter, G. 1987. Receptors for epidermal growth factor and other polypeptide mitogens. Annu. Rev. Biochem. 56:881-914. 6. Chen, W. S., C. S. Lazar, M. Poenie, R. Y. Tsien, G. N. Gill, and M. G. Rosenfeld. 1987. Requirement for intrinsic protein tyrosine kinase in the immediate and late actions of the EGF receptor. Nature (London) 328:820-823. 7. Chou, C. K., T. J. Dull, D. S. Russel, R. Gherzi, D. Lebwohl, A. Ullrich, and 0. M. Rosen. 1987. Human insulin receptors mutated at the ATP-binding site lack protein tyrosine kinase activity and fail to mediate postreceptor effects of insulin. J. Biol. Chem. 262:1842-1847. 8. Cooper, J. A., B. M. Sefton, and T. Hunter. 1983. Detection and quantification of phosphotyrosine in proteins. Methods Enzymol. 99:387-402. 9. Coussens, L., C. VanBeveren, D. Smith, E. Chen, R. L. Mitchell, C. M. Isacke, I. M. Verma, and A. Ullrich. 1986. Structural alteration of viral homologue of receptor proto-oncogenefms at carboxyl terminus. Nature (London) 320:277-280. 10. Daniel, T. O., D. F. Milfay, J. Escobedo, and L. T. Willians. 1987. Biosynthetic and glycosylation studies of cell surface platelet-derived growth factor receptors. J. Biol. Chem. 262: 9778-9784. 11. Gentry, L. A., L. R. Rohrschneider, J. E. Casnellie, and E. G. Krebs. 1983. Antibodies to a defined region of pp60 v-src neutralize the tyrosine specific kinase activity. J. Biol. Chem. 258:11219-11228. 12. Hedo, J. A., L. C. Harrison, and J. Roth. 1981. Binding of insulin receptors to lectins: evidence for common carbohydrate determinants on several membrane receptors. Biochemistry 20: 3385-3393. 13. Higgins, R. C., and M. E. Dahmus. 1979. Rapid visualization of protein bands in preparative SDS-polyacrylamide gels. Anal. Biochem. 93:257-260. 14. Honegger, A. M., D. Szapary, A. Schmidt, R. Lyali, E. Van Obberghen, T. J. DuBl, A. Ulirich, and J. Schlessinger. 1987. A mutant epidermal growth factor receptor with defective protein tyrosine kinase is unable to stimulate proto-oncogene expression and DNA synthesis. Mol. Cell. Biol. 7:4568-4571. 15. Hunter, T., and J. A. Cooper. 1985. Protein tyrosine kinases. Annu. Rev. Biochem. 54:897-930. 16. Johnson, D., A. Lanahan, C. R. Buck, A. Sehgal, C. Morgan, E. Mercer, M. Bothwell, and M. Chao. 1986. Expression and structure of the human NGF receptor. Cell 32:545-554. 17. Keating, M. T., and L. T. Williams. 1988. Autocrine stimulation of intracellular PDGF receptors in v-sis transformed cells. Science 239:914-916. 18. King, C. R., H. K. Matthias, and S. A. Aaronson. 1985. Amplification of a novel v-erbB-related gene in a human mammary carcinoma. Science 229:974-976. 19. Kleid, D. G., D. Yansura, B. Small, D. Dowbenko, D. M. Moore, M. J. Grubman, P. D. McKrecher, D. 0. Morgan, B. H. Robertson, and H. L. Bachrach. 1981. Cloned viral protein vaccine for foot-and-mouth disease: responses in cattle and swine.

VOL. 8, 1988


Science 241:1125-1128. 20. Kobuta, A. 1979. Use of endo- and exoglycosidases for structural studies of glycoconjugates. Anal. Biochem. 100:1-14. 21. Kraus, M. H., N. C. Popescu, S. Z. Amsbaugh, and C. R. King. 1987. Overexpression of the EGF receptor-related protooncogene erbB2 in human mammary tumor cell lines by different molecular mechanisms. EMBO J. 6:605-610. 22. Leal, F., L. T. Williams, K. C. Robbins, and S. A. Aaronson. 1985. Evidence that the v-sis gene product transforms by interaction with the receptor for platelet-derived growth factor. Science 230:327-330. 23. Maniatis, T., E. F. Fritsch, and J. Sambrook. 1982. Molecular cloning: a laboratory manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. 24. Nilsen, T. W., P. A. Maroney, R. G. Goodwin, F. M. Rottman, L. B. Crittenden, M. A. Raines, and H. J. Kung. 1985. C-erbB activation in ALV-induced erythroblastosis: novel RNA processing and promoter insertion result in expression of an aminotruncated EGF receptor. Cell 41:719-726. 25. Qui, F., P. Ray, K. Brown, P. E. Barker, S. Jhanwar, F. H. Ruddle, and P. Besmer. 1988. Primary structure of c-kit: relationship with the CSF-1/PDGF receptor kinase family-oncogenic activation of v-kit involves deletion of extracellular domain and C-terminus. EMBO J. 7:1003-1011. 26. Rettenmier, C. W., J. H. Chen, M. F. Roussel, and C. J. Sherr. 1985. The product of the c-fms proto-oncogene: a glycoprotein with associated tyrosine kinase activity. Science 228:320-322. 27. Reynolds, F. H., W. J. M. VandeVen, J. Blomberg, and J. R. Stephenson. 1981. Differences in mechanism of transformation by independent feline sarcoma virus isolates. J. Virol. 38:10841089. 28. Rosen, 0. M. 1987. After insulin binds. Science 237:1452-1458.

4903

29. Rosen, 0. M., R. Herrera, Y. Olowe, L. M. Petruzzelli, and M. H. Cobb. 1983. Phosphorylation activates the insulin receptor tyrosine protein kinase. Proc. Natl. Acad. Sci. USA 80: 3237-3240. 30. Roussel, M. F., T. J. Dull, C. W. Rettemneier, P. Ralph, A. Ullrich, and C. Scherr. 1987. Transforming potential of the c-fms proto-oncogene (CSF-1 receptor). Nature (London) 325:549552. 31. Sen, S., R. A. Houghten, C. J. Sherr, and A. Sen. 1983. Antibodies of predetermined specificity detect two retroviral oncogene products and inhibit their kinase activities. Proc. Natl. Acad. Sci. USA 80:1246-1250. 32. Slamon, D. J., G. M. Clark, S. G. Wong, W. J. Levin, A. Ullrich, and W. L. McGuire. 1987. Human breast cancer: correlation of relapse and survival with amplification of the HER-2/neu oncogene. Science 235:177-182. 33. Sudhof, T. C., J. L. Goldstein, M. S. Brown, and D. W. Russell. 1885. The LDL receptor gene: a mosaic of exons shared with different proteins. Science 228:815-822. 34. Thoakura, N. R., and 0. P. Bahl. 1987. Enzymatic deglycosylation of glycoproteins. Methods Enzymol. 138:350-359. 35. Yarden, Y., J. A. Escobedo, W.-J. Kuang, T. L. Yang-Feng, T. 0. Daniel, P. M. Tremble, E. Y. Chen, M. E. Ando, R. W. Harkins, U. Francke, V. A. Fried, A. Ullrich, and L. T. Williams. 1986. Structure of the receptor for platelet-derived growth factor helps define a family of closely related growth factor receptors. Nature (London) 323:226-232. 36. Yarden, Y., W. J. Kuang, T. Yang-Feng, L. Coussens, S. Munemitsu, T. J. Dull, E. Chen, J. Schlessinger, U. Francke, and A. Ullrich. 1987. Human proto-oncogene c-kit: a new cell surface receptor tyrosine kinase for an unidentified ligand. EMBO J. 6:3341-3351.