Isolation and Characterization of a Novel Receptor-type Protein ...

THEJOURNALOF BIOLOGICAL CHEMISTRY

Vol. 267. No. 5, Issue of February 15,pp. 3262-3267 1992 Printed in ir.S.A.

0 1992 by The American Society for Biochemistry and Molecular Biologv, Inc.

Isolation and Characterizationof a Novel Receptor-type Protein Tyrosine Kinase (hek)from aHuman Pre-B Cell Line* (Received for publication, July 2, 1991)

Andrew W. BoydSQIl, Larry D. Ward((**,Ian P. Wicks$, RichardJ. SimpsonII**,Evelyn SalvarisS, Andrew Wilks**, Karen Welch$, Maureen Loudovaris$, Steven Rockman$, and Inese Busmanis$$ From the $Lions Clinical Cancer Research Laboratory and the )IJoint Protein Sequence Laboratory, the Walter and Eliza Hall Institute of Medical Research, the **Ludwig Institute for Cancer Research (Melbourne Branch), and the §Department of Diagnostic Hematologyand the $$Department of Anatomical Pathology, Royal MelbourneHospital, Victoria 3050, Australia

In this report we describetheidentificationand of the kinase family (7). Abnormalities of the c-kit gene locus characterization of a novel tumor-associated receptor-lead to the defects in the congenitally anemic W/W mouse type tyrosine kinase (hek).We produced a monoclonal (8-10). The c-kit ligand has now been identified (11-14) and on shown to be encoded by the S1 locus which is abnormal inthe antibody (111.A4) that detected a novel glycoprotein the immunizing pre-B cell acute lymphoblastic leuke- steel mouse (15). The steel mouse has developmental defects mia cell line (LK63). This antigen was shown to be expressed sporadically on hemopoietic tumor cell lines identical to the W/W mouse but possesses a normal c-kit stain- gene. and on ex vivo tumors. However, using antibody The other line of evidence for a critical role of tyrosine ing, the molecule was undetectable on normal tissues. Further biochemical characterization showed this mol- kinase proteins ingrowth control came from the study of viral ecule (hek)to be a phosphoroprotein.This observation oncogenes (16, 17). These genes were shown to be directly of hek involved in growth dysregulation by the introduction of DNA taken togetherwith the tumor-associated nature expression suggested that hek might be a receptor-typeencoding these genes into murine fibroblasts. These oncoproteintyrosine kinase. This was demonstratedby genes have been shown to have close cellular homologues affinity purification of hek. In in vitro kinase experi- (proto-oncogenes). One of the first identified oncogenes was hek protein was autophosphorylated v-src, the cellular homologue of which (c-src) is the prototypments the purified on tyrosine and also mediated tyrosine phosphoryla- ical representative of the family of cytoplasmic tyrosine tion of casein. Purified hek was subjected to N-terminal kinases which, after myristylation, become associated with amino acid sequence analysis which showed that hek the inner leaf of the cell membrane (18).Within the hemohad a unique N terminus. Amino acid sequence deter- poietic system a number of lineage-restricted src-like kinases mination of peptides from a VS protease digest of hek yielded one 21-amino acid stretch of sequence which have been defined (19). For example the T cell-associated srcshowed close homologywith the eph subfamily of pro- like kinase, Ick, has been shown to associate independently tein tyrosine kinases. These studies show hek to be a with both the CD4 and CD8 transmembrane glycoproteins to novel human tumor-associated protein tyrosine kinase, form a signaling complex (20, 21). In contrast, v-erbB and vwhich by analogy with previously characterized pro- fms, like their cellular homologues the epidermal growth factein tyrosine kinase proto-oncogenes, may have a role tor receptor and colony-stimulating factor-1 receptor, respecin tumorigenesis. tively, are transmembrane molecules encoding the entire signal transduction machinery in a single polypeptide (1, 17). Analysis of the amino acid sequences of these proteins has revealed conserved structural motifs within the catalytic doProtein tyrosine kinases are an important class of molecules mains (5). Both tyrosine and serine-threonine kinases have a involved in the regulation of growth and differentiation (1). consensus GXGXXG sequence which is found in many nuOne line of evidence came from the identification of receptors cleotide-binding proteins (5). Some conserved sequence motifs that bind certain soluble growth factors. The receptors for are sharedby both typesof kinase whereas others are specific epidermal growth factor (2), platelet-derived growth factor for the tyrosine or the threonine-serine kinase subgroups (5). (3), and colony-stimulating factor-1 (4) were all shown to be The tyrosine kinases, although having regions of sequence transmembrane molecules with the cytoplasmic regions enconservation specific to thisfamily, can be subdivided further coding a protein tyrosine kinase catalyticdomain. The CSF- according to the structural features of the regions 5’ to the 1receptor is homologous to theplatelet-derived growth factor catalytic domain (1, 4-7). receptor in both the catalytic and extracellular domains (1, In this report we describe a novel cell surface glycoprotein 5). The extracellular domain of these proteinsis distinguished which is readily detected by the III.A4 monoclonal antibody from other members of the protein tyrosine kinase family by on sporadic cell lines and tumorspecimens but noton normal the presence of immunoglobulin-like repeats (1, 6). More cells. We show that the III.A4 antibody identifies a novel recently the c-kit protein was identified as another member humanreceptor-typetyrosine kinase. Purification of this * The costs of publication of this article were defrayed in part by protein from a pre-B acute lymphoblastic leukemia cell line, the payment of page charges. This article must therefore be hereby LK63, and amino acid sequencing identified this molecule as marked “aduertisement” in accordance with 18 U.S.C. Section 1734 a member of the ephlelk family of tyrosine kinases (22-24). solely to indicate this fact. li To whom correspondence should be sent: The Walter and Eliza We assigned this molecule the provisional name hek (human Hall Institute of Medical Research, Post Office, Royal Melbourne ephlelk-like kinase).The possible role of hek in the neoplastic process is discussed. Hospital, Victoria 3050, Australia. Tel.: 61 3 345 2555.

3262

hk,a Novel Human Receptor-type Protein Tyrosine Kinase TABLEI Cell lines tested for expression of the hek antigen Hemopoietic cell lines were analyzed by indirect immunofluorescence and flow cytometric analysis. Adherent cell lines were cultured on glass coverslips and analyzed by staining of the cultured cells in situ using indirect fluorescence. Hemopoietic cell lines Pre-B LK63. Lila-1. Reh. Nalm-1. FAKEM B cell Raji, Daudi, RAMOS, U266, BALL-1 T cell HSB2, MOLT4, HPB-ALL, PEER, Jurkat, JM Myelomonocytic HL60, U937, RC2a, THP-1, KG-1 Erythroid K562 Megakaryocytic HEL Epithelial cell lines A431, HT29, LIM1215, ST6, LIM1899, COL0205, LIM1863 Other Embryonic fibroblasts, bone marrow stromal cells, umbilical vein endothelial cell. svnovial cells. melanoma cell lines "96 and C32 MATERIALS ANDMETHODS

Cell Lines The LK63 and Lila-1 cell lines were established in culture from primary isolates of bone marrow and peripheral blood of two patients with acute lymphoblastic leukemia. Both had pre-B cell phenotypes and cytogenetic features characteristic of pre-B cell acute lymphoblastic leukemia, including the tl:19 translocation (25). Details of these studies will be published elsewhere.' Other hemopoietic lines used in this study are listed in Table I. Nonhemopoietic cell lines included two melanoma lines and a number of epithelial lines A431, COLOHT29, COLOSTG, LIM1899, LIM1863, and COL0205, which were the gift of Dr. R. Whitehead, Ludwig Institute, Melbourne Branch. Normal human skin fibroblasts, umbilical vein endothelial cells, and synovial cells were also tested. Monoclonal Antibodies The III.A4 monoclonal antibody (IgG1, K ) was prepared by intraperitoneal immunization of a BALB/c mouse with LK63 cells. Spleen cells from this mouse were fused with NS-1 cells to form hybridomas. Supernatants of the resulting hybridomas were screened on LK63 cells by indirect immunofluoresence (see below) and positive clones isolated. The III.A4 antibody clone reacted strongly with LK63 but not with other hemopoietic cell lines. Other monoclonals used included FMC63 (CD19), B1 (CD20), IA7 (CD36), W6/32 (major histocompatibility complex class I), OKT8 (CD8) and PHM6(CD10/ CALLA). Tumor Samples by Indirect Immunofluorescence Freshtumor material was obtained from biopsy specimens in accord with procedures approved by Institutional Ethical Review Boards. Single cell suspensions were prepared and thesamples applied to Ficoll-Hypaque gradients. The buoyant density cells were isolated and cell surface phenotypic analysis performed. Cells (lo6) were mixed with appropriate dilutions of monoclonal antibodies for 30 min on ice. The cells were washed and mixed with fluorescein-labeled F(ab')' fragments of sheep anti-mouse immunoglobulin antibody and held for 20 min on ice. The samples were washed, fixed in 1% formalin in phosphate-buffered saline, and analyzed on a FACS I1 cell sorter (Becton Dickinson, Mountain View, CA). Immunoperoxidme Staining of Tissue Sections Frozen sections were cut at 3 pm, placed on gelatin-coated slides, and fixed in acetone for 15 min and air dried. Sections were stained with III.A4 and control antibodies using the peroxidase-anti-peroxidase immunoperoxidase method (26). Cell Labeling and Immunoprecipitation surface proteins were labeled with "'I using a modification of the lactoperoxidase-catalyzed iodination technique as described previously (27). "'1 Labeling-Cell

E. Salvaris, J . R. Novotny, K. Welch, L. J. Campbell, and A. W. Boyd, manuscript in preparation.

3263

r'S/Methionine Labeling-Metabolic labeling was performed with [35S]methionine.Cells were starved in methionine-free RPMI 1640, 5% dialyzed fetal calf serum for 2 h. In some experiments the cells were labeled for 4 h in RPMI 1640, 5% fetal calf serum containing 100 pg/ml [35S]methionine.In other experiments cells were pulsed for 10 min in RPMI 1640, 5% fetal calf serum plus [3'S]methionine at 500 pCi/ml. The cells were washed in ice-cold medium, resuspended in normal RPMI 1640 with 10% fetal calf serum, and recultured for differing "chase" intervals. 32PLabeling-Labeling with 32Pwas carried out in phosphate-free medium (30 mM Hepes? pH 7.4, 110 mM NaC1, 10 mMKC1, 10 mM glucose, 1 mM CaCl,, 1 mM M&12, 2 mM glutamine, 2 mg/ml bovine serum albumin). LK63 cells were resuspended at 4 X 106/ml and incubated for 1h with 32Pat 500 pCi/ml). Labeling was quenched by washing twice in ice-cold phosphate-buffered saline. Preparation of Cell Lysates-Cells were lysed in 1%Triton X-100 in 25 mM Tris-HC1, pH 7.4, with 150 mM NaCl. Leupeptin, antipain, chymostatin, and pepstatin (each at 1pg/ml) were added to prevent proteolysis. After 45 min on ice the lysates were clarified by ultracentrifugation. Immunoprecipitations-The lysates were precleared overnight with 50 pl of protein A-Sepharose 4B beads and 10 pg of rabbit anti-mouse IgG antibody. Supernatants were transferred to fresh tubes, mixed with monoclonal antibodies, and held on ice for at least 4 h. The immune complexes were precipitated on rabbit anti-mouse Ig and protein A-Sepharose 4B (as above). In some cases antigen was absorbed directly to III.A4 antibody conjugated to Trisacryl beads (see below). In thiscase the beads were mixed with the supernatant and mixed at 4 "C on an end-over-end stirrer for 2 h. In each case beads were washed twice with 1% Triton X-100, 0.2% deoxycholate, 25 mM Tris, pH 7.4, 0.5 M NaC1, 1 mM EDTA, 1mM EGTA and twice with the same buffer except that NaCl was 0.15 M and 0.01% SDS was added. The samples were eluted into running buffer and analyzed by SDS-PAGE. Glycosylation Studies Glycosylation Inhibitors-Cells were incubated for 6 h with biosynthetic labeling medium a t 5 X 106/ml in the presence of 10 pg/ml tunicamycin (Sigma), 2 mM deoxymannojiromycin (Boehringer Mannheim), 100 pl/ml castanospermine (Boehringer Mannheim), or with medium alone. The radiolabeled hek antigen was analyzed as described above. Endoglycosidase Treatment-In other experiments cells were labeled with "'1 as described above. Immunoprecipitates with III.A4 and control antibodies were treated with neurominidase, endoglycosidase F, and sequentially with 0-glycanase (Endo-a-N-acetylgalactosaminidase) according to the manufacturer's conditions. Deglycosylated and control precipitates were analyzed by SDS-PAGE. In Vitro Kinase Reactions Cell lysates of unlabeled LK63 cells were prepared and lysates incubated with III.A4-conjugated Trisacryl beads. Pure III.A4 antibody wasisolated on proteinA-Sepharose coupled to glutaraldehydeactivated Trisacryl at 2.5 mg of III.A4/ml according to the manufacturer's protocol (IBF, France). Beads were washed as described above followed by two washes with kinase buffer (50 mM Hepes, pH 7.4,lO mM MnC12, 1%Triton X-100). The washed beads were incubated in kinase buffer containing 20 pCi of [y3'P]ATP for 15 min at room temperature. In some experiments 0.1 pg of dephosphorylated casein (Sigma) was added as a substrate in these reactions. The reaction was stopped by the addition of an equal volume of 2 X SDS sample buffer. The samples were analyzed on 7.5% SDS-PAGE gels. Analysis of Phosphamino Acids Dried slab gels from the in vitro kinase experiments were aligned with autoradiographs and bands containing 32P-labeledphosphoproteins excised with a scalpel blade. Gel slices were rehydrated and washed twice in 20 mM NH4HC03(1 h/wash). Protein was eluted by treatment with 100 p1 of proteinase K (0.5 mg/ml) in 20 mM NH4HC03. The eluates were pooled and lyophilized and thentreated with 100 p1 of 6 N HCl for 1 h at 110 "C. After relyophilization the 'The abbreviations used are: Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid; EGTA, [ethylenebis(oxyethylenenitrilo)] tetraacetic acid; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis.

3264

hek, a Novel Human Receptor-type Protein Tyrosine Kinase

material was taken up in 5 pl of pyridine/acetic acid/water at 10:1001890 ratio (pH3.5). Samples and phosphamino acid standards (tyrosine, serine and threonine, Sigma) were analyzed on precoated cellulose thin layer chromatography plates (Merck 5757) by twodimensional chromatography performed as described previously (28) using an HTLE-7000 chromatography apparatus (C.B.S. Scientific CO.,Del Mar, CA). The TLC plates were stained with 1%ninhydrin, dried, and autoradiographed. Antigen Purification LK63 cells were prepared in batches of 1O'O cells and lysed in 100 ml of lysis buffer (25 mM Tris, pH7.4,0.5 M NaC1, 1%Triton X-100, plus enzyme inhibitors as described above). The lysate was centrifuged at 2,000 X g to remove unlysed nuclei. The supernatant was centrifuged at 35,000 X g in aSorvall RC4B centrifuge for 1h at 4 "C. The lysate was incubated with Sepharose 4B beads (2% v/v) overnight. The absorbed lysate was incubated with Trisacryl beads coupled with FMC63 (CD19) antibody for >4 h. The FMC63-absorbed lysate was incubated with 2 ml of 111.A4-coupledTrisacryl beads for 4 h. The slurry was transferred to a small column and the lysate removed. The III.A4 beads were washed with 20 volumes of lysis buffer and with 10 volumes of 10 mM Tris, pH 7.4, 0.5% Triton X100. The beads were washed sequentially with 4 volumes of 50 mM ethanolamine with 0.1% Triton X-100 at pH 9 and again at pH 10. The beads were then eluted sequentially with pH 11, 12, and 12.5 50 mM diethylamine, 0.1% Triton X-100 buffers. Samples were removed from each eluate, neutralized with HCl, lyophilized, and analyzed by 7.5% SDS-PAGE. The gels were stained with a modified silver stain protocol (Bio-Rad). The hek protein eluted at pH 11-12. These fractions were pooled and bound to 1II.A.l-Trisacrylonce more. The hek protein was reeluted with pH 12 10 mM diethylamine, 0.05% Triton X-100 and lyophilized for sequence analysis. Amino AcidSequence Analysis Electrophoresis and Electrotransfer-The N-terminal sequence of hek was determined as described previously (29). Briefly, hek was separated using a 7% polyacrylamide gel employing the Laemmli buffer system and theBio-Rad protean I1 apparatus. The separating gel waspreelectrophoresed a t 60 V for 4 h in375 mM Tris-HC1 buffer, pH 8.8, containing 1 mM thioglycollic acid prior to pouring the stacking gel. Electrophoresis was performed at 35 mA constant current, the temperature being maintained a t 4 "C by circulation of water from an external cooling system. Thioglycollic acid (1 mM) was included in the upper buffer reservoir during electrophoresis. After electrophoresis, the gel was equilibrated in 10 mM CAPS transfer buffer, pH 11, containing 15% (v/v) methanoland 1mM thioglycollic acid for 30 min prior to electroblotting onto polyvinylidene difluoride membrane. The transfer was performed at 90 V for 2 h at 4 "C using a Bio-Rad Transblotapparatus.Protein was stained using 0.1% Coomassie Blue in 50% methanol and destained using 50% methanol, 10% acetic acid and then washed extensively with distilled water to remove contaminating amino acids which otherwise interfere with interpretation of the initial Edman degradation cycles (29). Peptide Mapping: In SituProteolytic Digestionof Proteins in Acrylamide Gel Matrix-The eluate of the 111.A4-Trisacryl column was reduced with dithiothreitol and alkylated with vinylpyridine using the following protocol. The lyophilized eluate was dissolved in 125 mM Tris-HC1 buffer, 3% SDS, pH 6.8, containing 8.45 mM dithiothreitol and incubated at 37 "C for 4 h. Vinylpyridine (2 p1/200 p1 of reaction mixture) was added and incubated for an additional 25 min. The reaction was stopped by the addition of 8-mercaptoethanol. The hek protein was subjected to in situ enzymatic cleavage as described by Cleveland (30) with modifications (31). After separation on a 7% polyacrylamide gel, pyridylethylated hek was visualized by staining with 0.1% (w/v) Coomassie Blue R-250 in methanol/acetic acid/H20 (50:10:40) (v/v) for 20 min and then destained in methanol/acetic acid/H20 (12:7:81). Protein bands were excised, soaked for 1 h in equilibration buffer (125 mM Tris-HC1 buffer, pH 6.8, containing 10% (v/v) glycerol, 50 mM 8-mercaptoethanol, 1 mM EDTA, 0.1% (w/v) SDS), and positioned in the sample wellof a second 15% polyacrylamide gel. Gel slices were overlayed with 50 pl of 125 mM Tris-HC1 buffer, pH 6.8, containing 20% glycerol, 1 mM EDTA, 50 mM fbmercaptoethanol, and 0.1% (w/v) SDS followed by 2 pgof Staphylococcus aureus strain V8 protease in 10 pl of equilibration buffer. A modified Laemmli buffer system (32) (0.75 M Tris in separating gel, 0.05 M Tris in reservoir buffer) was used to achieve optimal resolution of low M. peptides. Electrophoresis was performed

at 60 V until the dye front passed through the stacking gel (4 h) and thereafter a t 35-mA constantcurrent. Blotting of peptides onto polyvinylidene difluoride and protein visualization were performed as described above. Amino Acid Sequence Determination-Sequence analysis of electroblotted proteins and peptides was performed using an Applied Biosystems (model 470A, Foster City, CA) Sequencer equipped with an on-line phenylthiohydantion-derivative analyzer (model 120A). To facilitate total injection of the PTH-derivatives, an improved transfer device (33) was fitted to the system. To obtain optimal sequencer performance, polyvinylidene difluoride strips containing protein or peptide were partially sliced and positioned upon a Polybrene (34)-treated glass fiber disc in the sample cartridge of the Sequencer in a manner allowingoptimal flow of reagents and solvents. RESULTS

Phenotypic Analysis of hek Antigen Expression Using the III.A4 Antibody-The expression of the hek antigen detected by the IILA4 monoclonal antibody was initially analyzed on human cell lines. The expression of the antigen on the immunizing cell line (LK63) is depicted in Fig. 1. It is evident that III.A4 antibody detected significant levels of its target hek antigen on this cell line but failed to detect the antigen on two other hemopoietic cell lines.Analysis of the extensive group of lines depicted in Table I revealed only one other IILA4-positive line (JM). The Raji cell line showedweak expression, but all other lines tested were consistently negative. In particular, four other pre-B cell lines showedno detectable expression. Normal and tumortissue isolated from patients undergoing biopsy procedures allowed us to analyze expression in vivo. As shown in Table 11, hek was not detectable on normal tissues by the methods employed. Analysis of tumors showed that a very low proportion of hemopoietic tumors expressed the hek antigen. Moreover, analysis of early passage LK63 cells showed that hek antigen was detectable within a few passages of initial in vitro culture, suggesting that elevated hek antigen expression on the LK63 tumor might be an in vivo event. However, as no direct ex vivo tumor tissue could be examined, this could not be confirmed. Molecular Properties of the hek Antigen-The results of phenotypic analysis suggested that the hek antigen was expressed at an abnormally high level on LK63, perhaps indicating a role for this antigen in the oncogenic process. We used the III.A4 antibody to analyze the molecular properties of the hek antigen. As shown in Fig. 2, cell surface labeling and immunoprecipitation revealed a protein of 135 kDa. To determine the glycosylation status of the protein two experiments were performed. The first was a biosynthetic labeling of LK63 cells treated with inhibitors of glycosylation. Tunicamycin was extremely toxic to thecells, but a faintband was detected at 95 kDa (Fig. 3). Castanospermine and deoxymannojiromycin treatments resulted in intermediate sized glyco-

FLUORESCENCE INTENSITY

+

FIG. 1. Expression of hek on cell lines. LK63 (panel A ) , RC2a (panel B ) and K562 (panel C) cell lines were incubated with III.A4 antibody (solid line) or a control isotype-matched antibody (broken line) and stained by indirect immunofluorescence. The cells were analyzed by flow cytometry. Histograms depict cell number (vertical ais)versus fluorescence intensity.

hek, Novel a

Human Receptor-type Protein Tyrosine Kinase

3265

TABLE I1

r " " "

Detection of hek on freshly isolated h u m n tissues Method of analysis

Tissue

Flow cvtometrv'

IMPOXb

Normal spleen 0/3 0/2 0/5 Normal lymph node 0/4 Normal bone marrow 0/5 NT' Normal tonsil 0/2 NT Normal breast NT 0/3 Normal brain NT 0/2 1/28 NT Chronic lymphocytic leukemia 0/85 0/9 Lymphoma Acute myeloid leukemia 2/39 NT Acute lymphoblastic leukemia 0/19 NT 0/8 Breast carcinoma NT 1/6 NT Other carcinomasd 'Primary hemopoietic tumors and tissues were prepared as single cell suspensions and stained by indirect immunofluorescence. *Tissue sections were analyzed by the immunoperoxidase (IMPOX) technique. NT = not tested. Specimens included one cervical, one prostate, one ovarian, and one renal carcinoma (positive) and two carcinomas of unknown origin.

A

220 92-

- 220

-

I

FIG.4. Deglycosylation of hek antigen by N- and O-glycanase. LK63 cells were labeled with '?'I and thecell lysate immunoprecipitated with III.A4. Shown area control precipitate (-1 anda precipitate treated sequentially with N- and 0-glycanase (+ENZ).

61

45 A B

m-I).

-92

-67 1 2 3

&

z

N

B

67-

45-

-92 - 61

-45

1 2 FIG. 2. Immunoprecipitationof hek antigen by III.A4 antibody. A, autoradiograph of immunoprecipitates as from '2sI-labeled LK63 cell lysates (lane I , immunoprecipitation with 111.A4-Trisacryl beads. Lane 2, immunoprecipitation with control antibody (LA7 antiCD36) coupled to Trisacryl beads. B , silver staining of immunoprecipitation of LK63 cell lysate. Lane I, elution at pH 11 from III.A4Trisacryl beads. Lane 2, pH 11 eluate of reprecipitation of runthrough cell lysate on 11.A4-Trisacryl.Lane 3, eluate of 1.A7-Trisacryl column.

FIG.5. 32Plabeling of LK63 cells. Autoradiograph of immunoprecipitation of LK63 cells using A, IILA4-Trisacryl beads; B, Bl(CD2O) antibody bound to protein A-Sepharose 4B. p32 1125 " .

.

4

9261 -

45

A

B C D E

67

-

FIG. 3. Biosynthetic labeling of he&antigen in the presence of inhibitors of glycosylation. [%]Methionine labeling was performed in the presence of tunicamycin (TUN),deoxymannojiromycin ( D J M ) ,castanospermine ( C S P ) ,and medium alone. The lysates were immunoprecipitated with III.A4-Trisacyl beads and thesamples analyzed by SDS-PAGE. The arrow indicates the faint band seen after labeling in the presence of tunicamycin.

sylated forms. The nonglycosylated size was confirmed by enzymatic digestion of '2sI-labeled, immunoprecipitated hek with 0- and N-glycanase. As shown in Fig. 4, the size of the nonglycosylated protein was 95 kDa. The final labeling procedure used was 32Plabeling of LK63 cells. The immunoprecipitate with III.A.1 antibody consistently revealed a faint band not seen in controlimmunoprecipitates (Fig. 5). The faintness of this band suggested a very low level of phosphorylation. Treatment of LK63 cells with phorbol ester and with cytokines (interleukins-1, -4, -6, and

F

FIG.6. hek antigen is labeled in an in vitro kinase reaction. Immunoprecipitates of hek and surface IgMwere prepared using 111.A4-Trisacryl and S.ADA.14 (anti-p chain) antibodies from unlabeled LK63 (A, B, E, and F) or '251-labeledcells. In vitro kinase reactions were analyzed. Lanes A and E, two exposures of the precipitates with IILA4. Lanes C and D show, respectively, parallel autoradiographs of '251-labeledhek ( l a n e C) and p chain ( l a n e D).

-7; granulocyte-macrophage colony-stimulating factor; granulocyte colony-stimulating factor; leukemia inhibitory factor; and y interferon) failed to induce any detectable increase in hek phosphorylation. The hek Antibody Detects a Tyrosine Kinuse-The elevated expression of hek antigen in tumor cell lines, taken together with the data showing that hek is phosphorylated, suggested to us that thisprotein may be a tyrosine kinase. To explore this possibility cell lysates were prepared from unlabeled LK63 cells and immunoprecipitated on either 111.A4-Trisacryl or with anti-IgM(S.ADA.14). These precipitates were washed with kinase buffer (see "Materials and Methods") and incubated with [y3'P]ATP. As shown in Fig. 6 hek protein was strongly labeled with '*P. In the overexposed autoradiograph shown in lanes E and F several other fainter bands areseen. The most intense of these bands is at the top of the gel and may represent aggregated material. The other bands were

3266

hek, a Novel Human Receptor-type Protein Tyrosine Kinase

inconstantly seen and presumably represent contaminating phosphoproteins present as a result of incomplete removal of other cellular proteins. The major band was excised and subjected to phosphamino acid analysis which confirmed phosphorylation on tyrosine (Fig. 7).In thiscase, dephosphorylated casein was added to thereaction mixture. It is evident that the hek and casein bands are heavily phosphorylated. The phosphamino acid analysis of each band showed phosphorylation on tyrosine. Isolation of the Antigen Detected by III.A4"To confirm that the hek protein was indeed a tyrosine kinase we sought amino acid sequence information. The protein was purified by scaling up the isolation procedure using III.A4 antibody coupled to Trisacryl. Other affinity supports failed to give clean purification, but thissupport yielded hek protein in high purity as analyzed by silverstaining of SDS-PAGE gels (Fig. 2B). Ten-liter batches of cells (about 10'' cells) wereprepared for antigen isolation. Each batch yielded 5-10pg(55-110 pmol) of purified hek antigen basedoncomparison with protein standards. Initially, N-terminal sequencing was performed on twoseparate batches of material. In each case the same sequence was detected (Table 111). Internal sequence analysis was obtained by in situ digestion of hek protein (10 pg, 111pmol) in the polyacrylamide matrix with staphylococcal V8 protease. The generated fragments were separated by SDS-PAGE and electrotransferred onto polyvinylidene difluoride membranes. The peptide map obtained, after visualization by Coomassie Blue R-250 staining, is presented in Fig. 8. Peptides 1 and 2 (Fig. 8) were subjected to N-terminal

9 4 -

" " 14-

o-Pl "P2

FIG. 8. SDS-PAGE of staphylococcal V8 protease digest of hek protein. Two peptides (PIand P2) yielded useful sequence data and were found to have an identical N-terminal sequence.

sequence analysis and were found to have identical N-terminal sequences (Table 111). Other bands were also sequenced but did not yield sequencedata amenable for subsequent gene cloning studies. DISCUSSION

We have shownthat a monoclonal antibody raised against the human LK63 pre-B cell line (Fig. 1)is specific fora 135kDa glycoproteindesignated hek (Figs. 2 and 3). This protein was clearly expressedby LK63 cells and by a T cell line JM (Table I). However, many other cell lines and a number of normal tissues did not stain with this antibody by immunofluorescence and/or immunoperoxidase techniques (Tables I and 11). When fresh tumor specimens were examined, a small number of cases showed detectable hek expression (Table 11). These findings suggested sporadic overexpression of hek on human tumors. We postulated that hek may be an oncogenic ~1354 protein and, given its cell membrane localization, we explored the possibility that hek is a tyrosine kinase. We showed that hek was only weaklyphosphorylated by biosynthetic labeling with 3zP(Fig. 5). Moreover, wewere unable to increase the level of phosphorylation by exposure to a number of known growth factors and chemical inducers (data not shown). This level of labeling was insufficient to allow a direct demonstration of tyrosine phosphorylation in these experiments. However, in Vitro kinase reactions performed with purified hek protein and a kinase substrate (casein) allowed us to show that hek was autophosphorylated on tyrosine and could tyro1 [, sine phosphorylate casein (Fig. 7). To exclude the possibility & that thekinase activity was generated by a minor, copurifying 1 2 Casein protein, sequence data were obtained (Table 111). The seFIG. 7. Phosphamino acid analysis of hek antigen and casein after an invitro kinase reaction of proteins immunopre- quence of V8 protease peptides (Fig. 8) confirmed that hek is cipitated with III.A4 antibody. Immunoprecipitated hek antigen a member of the tyrosine kinase family. Several known tyrosine kinases have a molecular weight bound to 111.A4-Trisacryl was mixedwith casein in a kinase reaction ) (as above). After SDS-PAGE the reaction labeled hek ( ~ 1 3 5and similar to hek. However, the N-terminal sequence obtained casein were identified (lane I ) . The bands were removedand subjected (Table 111) is unique accordingto data base searches (EMBL, t o phosphamino acid analysis (see "Materials and Methods"). Swissprot, GenBank, Japanese Protein). However, oneinternal sequence obtained (peptides P1 and P2) exhibited signiTABLE 111 ficant sequence identity to a region of the kinase domain of Sequences obtainedfrom purified III.A4 antigen other protein tyrosine kinases (Table IV). The closest matches The sequences obtained were confirmed on two independent spec- are with elk and eph, which are members of the eph family of imens. The internal sequence was obtained from peptides P1 and P2 (Fig. 8). The yields of phenylthiohydantoin derivative observed in the protein tyrosine kinase (22, 23, 35). Another member of this Partial sefirst cycle of the Edman degradation for each sequence are given in family, eck (24), is somewhatmoredivergent. quences of eek and erk (36) show that hek is also quite parentheses. different from eek. The relevant amino acid sequencesare not N-terminal sequence published for erk, but preliminary polymerase chain reactionE L I P Q P S N E V N L X D derived nucleotide sequences of hek (data not shown) show ( S ) K X I Q (26pmol) Internal sequence that it is also distinct from erk. Comparisonwith other G Y R L P P P M D C P A A L receptor kinases shows marked divergence overthis region of Y Q L M L D C (lOpmol) sequence, strongly suggesting that hek is a member of the eph

3267

Tyrosine Protein .type hek, a Novel Human Kinase Receptor-

itsu, S., Dull, T. J., Chen, E., Schlessinger, J., Francke, U., and Ullrich, A. (1987) EMBO J. 66, 3341-3351 8. Chabot, B., Stephenson, D. A., Chapman, V. M., Besner, P., and Bernstein, A. (1988) Nature 335,88-89 9. Geissler, E. M., Ryan, M. A., and Housman, E. E. (1988) Cell 55, 185-192 10. Nocka, K., Majunder, S., Chabot, B., Ray, P., Cevone, M., Bernstein, A., and Besmer, P. (1989) Genes & Deu. 3,816-826 11. Williams, D. E., Eisenman, J., Baird, A., Ranch, C., van Ness, K., March, C. J., Park, L. S., Martin, U., Mochizuki, D.Y., Boswell, H. S., Burgess, G. S., Cosman, D., and Lyman, S. D. (1990) Cell 63,167-174 12. Zsebo, K. M., Williams, D.A., Geissler, E. N., Broudy, V. C., . Martin, F. H., Atkins, H. L., Hsu, R. Y., Burkett, N. C., Okin, K. H., Langley, K. E., Smith, K. A., Takeishi, T., Cattanach, B. M., Galli, S. J., and Sugg, S. V. (1990) Cell 63,213-224 13. Huang, E., Nocka, K., Beier, D. R., Chui, T. Y., Buck, J., Lahn, H. W., Wellner, D., Leder, P., and Besner, P. (1990) Cell 63, subfamily of protein tyrosine kinase. The closest similarity is 225-233 with elk, a rat tyrosine kinase that has been shown to have 14. Copeland, N. G., Gilbert, D. J., Cho, B. C., Donovan, P. J., Jenkins, N. A., Cosman, D., Anderson, D., Lyman, S. D., and restricted expression to brain and testis (23,37). Whether hek Williams, D. E. (1990) Cell 63,175-183 is the human homologue ofelk or a different but closely 15. Bennett, D. (1956) J. Morphol. 98, 199-233 related member of the eph family remains to be determined 16. Bishop, J. M. (1983) Annu. Reu. Biockm. 52,301-354 once more extensive sequence data areavailable. However, it 17. Hunter, T., and Cooper, J . A. (1985) Annu. Rev. Biochem. 54, is notable that theN-terminal sequence of rat elk (37) is quite 897-930 unlike the N-terminal sequence of hek. Moreover, the other 18. Resh, M. (1990) Oncogene 5, 1437-1444 internal sequences, although limited, appear not to have a 19. Eiseman, E., and Bolen, J. B. (1990) Cancer Cells 2, 303-310 counterpart in the elk sequence or indeed with the sequences 20. Veillette, A., Bookman, M.A., Horak, E. M., and Bolen, J. B. (1988) Cell 55,301-308 of eph or eck (data not shown). Further comparison with 21. Rudd, C. E., Tevillyan, J. M., Dasgupta, J. D., Wong, L. L., and known eph-like kinases requires the molecular cloning of hek Schlossman, S. F. (1988) Proc. Natl. Acad. Sci. U. S. A . 85, which will allow detailed sequence comparison. 5190-5194 The overexpression of hek on some hemopoietic tumors 22. Hirai, H., Maru, Y., Hagiwara, K., Nishida, J., and Takaku, F. (1987) Science 238, 1717-1720 suggested a role in tumorigenesis. We have performed Scat23. Letwin, K., Yee, S. P., and Pawson, T. (1988) Oncogene 3 , 621chard analysis with '231-labeledIII.A4 and demonstrated that 627 both LK63 and JM have 10,000-20,000 hek sites/cell. In 24. Lindberg, R. A., and Hunter, T. (1990) Mol. Cell. Biol. 10,6316contrast, no detectable sites were found on Raji or K562 cells 6324 (data not shown). This would appear to imply that hek is not 25. Heim, S., Mitelman, F. (1987) Cancer Cytogenetics, pp. 141-174, Allan R. Liss, Inc., New York expressed at all onsome related hemopoietic tumor cell lines. One possibility is that hek is aberrantly expressed rather than 26. Sternberger, L. A. (1974) in Immunochemistry (Cliffs, N. J., ed) pp. 127-171, Prentice-Hall Inc., Englewood Cliffs, N J overexpressed on these lines. 27. Boyd, A. W., Dunn, S. M., Fecondo, J. V., Culvenor, J. G., In conclusion, this study has identified a novel human Duhrsen, U., Burns, G. F., and Wawryk, S. 0. (1989) Blood 73, receptor-type tyrosine kinase. The hyperexpression of hek in 1896-1903 a number of tumors suggests that thisprotein may play a role 28. Cooper, J. A., Sefton, B. M., andHunter, T. (1983) Methods Enzymol. 99,387-402 in tumor induction. HER/mu/erb-B2 has been shown to be L. D., Hong, J., Whitehead, R. H., and Simpson, R. J . expressed aberrantly on some breast andovarian carcinomas 29. Ward, (1990) Electrophoresis 11,883-891 (38). Moreover, amplified expression of m u correlates with 30. Cleveland, D. E. (1983) Methods Enzymol. 92, 222-229 bad prognosis in human breast cancer (39). Another protein 31. Ward, L. D., Reid, G. E., Moritz, R. L., and Simpson, R. J. (1990) tyrosine kinase, ERBB3, was also shown to be overexpressed J. Chromatogr. 519,199-216 in a subset of human breast tumors (40). Within the eph 32. Fling, S. P., and Gregson, D. S. (1986) Anal. Biochem. 155, 8388 family, the eph protein itself has already been shown to be G. S., and Simpson, R. J. (1989) in Methods in Protein expressed in a number of colon carcinomas (22, 41). The 33. Begg, Sequence Analysis (Wittman-Liebold, B., ed) pp. 108-111, present findings suggest that hek, another member of the eph Springer-Verlag, Berlin family, also has oncogenic properties. 34. Klapper, D. G . . Wilde, C. E.. and CaDra. J. D. (1978) , , Anal. Biochem. 85,.126-131 REFERENCES 35. Simpson, R. J., Moritz, R.Begg, L., G. S. Rubira, M. R., and Nice, E. C. (1989) Anal. Biochem. 177, 231-236 1. Ullrich, A., and Schlessinger, J (1990) Cell 61, 203-21236. Chan,J.,andWatt, V.M. (1991) Oncogene 6, 1057-1061 2. Carpenter, G., and Cohen, S. (1990) J. B i d . Chem. 265, 7709- 37. Lhotak, V., Gree, P., Letwin, K., and Pawson, T. (1991) Mol. Cell. 7712 Bi01. 11,2496-2502 3. Williams, L. T. (1989) Science 143, 1564-1570 38. Salmon, D. J., Godolphin, W., Jones, L. A., Holt, J. A., Wong, S. 4. Yeung, Y. G., Jubinsky, P. T., Sengupta, A., Yeung, D. C. Y., and G., Keith, D. E., Levin, W. J., Stuart, S. G., Udore, J., Ullrich, Stanley, E. R. (1987) Proc. Natl. Acad. Sci. U. S. A . 84, 1268A., and Press, M. F. (1989) Science 244. 707-712 1271 39. Salmon, Clark, D. J., M., G. Wong, S. G.,'Levin, W. J., Ullrich, 5. Hanks, S. K., Quinn, A. M., and Hunter, T. (1988) Science 242, A,, and McGuire, W. L. (1987) Science 235, 177-182 42-52 40. Kraus, M. H., Issing, W., Miki, T., Popescu, C., N. and Aaronson, 6. Yarden, Y.,and Ullrich, A. (1988) Annu. Reu. Biochem. 57,443S. (1989) Proc. Natl. Acad. Sci. U. S. A . 86, 9193-9197 478 41. Maru, Y., Hirai, H., Yoshida, M. C., and Takaku, F. (1988) Mol. 7. Yarden, Y., Kuang, W-J., Yang-Feng, T., Coussens, L., MunemCell. Biol. 8 , 3770-3776

TABLE IV Comparison of hek with other protein tyrosine kinases Sequences were aligned on the determined amino acid sequence of hek. Conserved proline, methionine and cysteine residues allow correct position of the more divergent sequences. EGF-R, epidermal growthfactor receptor; PDGF-R, platelet-derived growth factor receptor. G Y R L P P P M D C P A A L Y Q L M L D C HEK . . . . . . .V....P..E...N. eph elk . . . . . . . . . . . . . . . . . . . . . eck . F . . . T . . . . . S . I . . . . M Q . EDF-R . E . . . Q . P I . T I D V . M I . V K . PDGF-R.. . M A Q . A H A S D E I . E 1 . Q K C-fmS . .QMAQ.AFA.KNI .E1 .QA. c-kit . F . M L S . E H A . . E M . D I . K T .

-

I