ONCOGENE, POLYPEPTIDES ENCODED THEREFROM,

AND DIAGNOSTIC AND OTHER METHODS BASED THEREON

BACKGROUND OF THE INVENTION This work was supported by grants from the American

Cancer Society, American Business for Cancer Research Founda¬ tion and the National Institutes of Health. The U.S. Govern¬ ment has certain rights in this invention.

Throughout this application various publications are referenced by number within parentheses. Full citations for these publications may be found at the end of the speci¬ fication immediately preceding the claims. The disclosures of these publications in their entireties are hereby incor¬ porated by reference into this application in order to more fully describe the state of the art to which the invention pertains.

Cellular oncogenes, which are genes capable of altering the growth properties of normal cells, were first detected as the homologs of transforming genes of RNA tumor viruses (3). Additional cellular oncogenes have been dis¬ covered by their amplification in certain tumor cells (12,22) and by DNA transfer techniques (24,14,7,1). The most commonly used assay for detecting oncogenes following DNA transfer has been focus formation in NIH3T3 cells (24). An alternate assay for oncogenes based on tumorigenicity in nude mice is descri¬ bed herein (4,8). The details of this procedure have been published previously (8). In brief, DNA isolated from a tumor is cotransfected into NIH3T3 cells in the presence of a G418 antibiotic resistance gene (33,26). Cells which have taken up foreign DNA are selected by their resistance to G418, grown into colonies, pooled and injected into animals.

Studies of cotransfor ation with DNA from the human mammary carcinoma cell line MCF-7 have previously been des¬ cribed (8). After cotransforming NIH3T3 cells with this _t DNA and injecting cotransfor ed cells into nude mice, three "pri-

mary" tumors, called MCF-7-1, MCF-7-2 and MCF-7-3 were obtained. DNA from each of these tumors was capable of effi¬ ciently inducing "secondary" tumors following another round of cotransformation into NIH3T3 cells and tumorigenicity assays. The MCF-7-1 tumor, and subsequent tumors derived from MCF-7-1 DNA, _. were shown to contain the human N-ras gene. DNA from all tumors derived from MCF-7-2 DNA contained a com¬ mon gene which was called mcf2; and DNA from MCF-7-3 and its derived tumors all contained a gene which was called cf3. The isolation of portions of mcf2 and mcf3 was previously described. The molecular characterization of mcf3 is herein described. mcf3 derives in part from the closest human homo- log of the avian v-ros oncogene which is called rosl. The human rosl gene appears to- have been activated during gene transfer. The nucleotide sequence of the activated gene is _ also described. Based on the predicted amino acid sequence, the human rosl gene, like the v-ros gene from the avian sar¬ coma virus UR2 (17), encodes a^ ^' putative transmembrane protein kinase, possibly a growth factor receptor. Experiments des- cribed herein have determined that the human v-rosl gene is oncogenic, and that the polypeptide encoded by the human v-rosl gene, which functions as a tyrosine-activated protein kinase and has a truncated extracellular segment, is expres¬ sed in human tumor cells.

SUMMARY OF THE INVENTION

The invention concerns a DNA sequence comprising an activated oncogene which encodes a polypeptide capable of transforming NIH3T3 cells and of inducing a tumor when injec- ted into nude mice, and further said polypeptide having tyrosine-specific protein kinase activity, said DNA sequence having a nucleotide sequence substantially as shown in Figure 1.

This invention also concerns methods for detecting tumor cells which comprise isolating genomic DNA or RNA from

a cell, contacting the DNA or RNA so isolated -with a detect¬ able marker which binds specifically to at least a portion of the sequence encoding an activated oncogene of this invention, or at least a portion of the RNA sequence encoded by an activated oncogene of this invention, an detecting the marker so bound, the presence of bound marker indicating a predisposition of the subject to the disease.

This invention also concerns a method of determin¬ ing the predisposition of a subject to a disease, which co - prises isolating the genomic DNA or RNA from a cell from the subject, contacting the DNA or RNA so isolated with a detect¬ able marker which specifically binds to at least a portion of an activated oncogene of this invention, or to a portion of the RNA encoded by an activated oncogene of this invention, and detecting the marker so bound, the presence of bound mar¬ ker indicating a predisposition of the subject to the disease.

The invention also concerns a polypeptide molecule encoded by an activated oncogene, said polypeptide having the properties of transforming NIH3T3 cells, inducing a tumor when injected into nude mice, acting as a tyrosine-speci ic protein kinase, and further said polypeptide having an a ino acid sequence substantially as shown in Figure 1.

Tumor cells and tumors expressing the polypeptide of this invention may be detected with a detectable marker which specifically binds to at least a portion of the poly- peptides of this invention. Further, subjects predisposed to diseases associated with the polypeptides of this inven¬ tion may be identified with a detectable marker which speci¬ fically binds to at least a portion .of the polypeptide. Methods for detecting a tumor or tumor cells com¬ prise isolating serum from a subject, contacting the serum

< with a detectable marker which binds specifically to at least a portion of a polypeptide of this invention to form a marker-polypeptide complex and detecting the marker so bound, the presence of bound marker indicating the presence of a

tumor.

Finally, the invention concerns a method for treat¬ ing in a subject a tumor induced by an activated rosl oncog¬ ene which comprises isolating an immunoglobulin molecule which specifically binds at least a portion of a polypeptide encoded by the activated rosl oncogene, attaching to the immunoglobulin molecule so isolated a substance which sub¬ stantially limits the growth of a tumor or which destroys tumors to produce an antitumor immunoglobulin molecule, and contacting the tumor with an effective amount of the anti¬ tumor immunoglobulin molecule so produced, thereby limiting tumor growth or destroying the tumor.

DESCRIPTION OF THE FIGURE Figure 1 Sequence of the common region of the mcf3 cDNA clones

The nucleotide sequence of the common region of the mcf3 cDNA clones is depicted. Below the nucleotide sequence, the predicted a ino acid sequence is shown. The numbers at the end of each line refer to the position of the. predicted amino acid residues, with position + 1 defining the first amino acid of the mcf3 cDNA which is encoded by the rosl derived part of the locus. The potential membrane spanning domain of 21 hydrophobic amino acids is boxed. The position of two splice junctions close to the point of rearrangement are indicted by arrowheads.

DETAILED DESCRIPTION OF THE INVENTION A DNA sequence comprising an activated oncogene has been isolated which encodes a polypeptide capable of transforming NIH3T3 cells and of inducing a tumor when injec¬ ted into nude mice. 'The polypeptide also has tyrosine- specific protein kinase activity. The DNA sequence has a nucleotide sequence substantially as shown in Figure 4. > The DNA sequence of this invention encodes a ros

oncogene, preferably a rosl gene. The sequence may be iso¬ lated from a variety of sources, although the presently pre¬ ferred sequence encodes the human rosl gene. The polypeptide produced by the transcription of the gene and the translation of the gene product will vary with the initial DNA sequence.

A method of detecting a tumor cell which contains the DNA sequence of this invention is described. The method comprises isolating genomic DNA from a cell, contacting the DNA isolated from the cell with a detectable marker which " binds specifically to at least a portion of the DNA sequence of this invention which encodes an activated oncogene and detecting the marker so bound. The presence of bound marker indicates the presence of a tumor cell. Specifically men¬ tioned tumor cells which may be detected by the method of the present invention are those arising from astrocytσma and glioblastoma cells.

The detectable marker may be a labelled DNA sequence, including a labelled cDNA sequ-e ce, having a nucleotide seq¬ uence complementary to at least a portion of the DNA sequence of this invention, produced by methods known to those skilled in the subject art.

The detectable marker may also be a labelled ribo- nucleotide (RNA) sequence having a nucleotide sequence com¬ plementary to at least a portion of the DNA sequence of this invention, and may be isolated by methods known to those skilled in the art.

Detectable markers of this invention will be label- led with commonly employed radioactive labels, i.e. 32P, although other labels may be employed. The markers will be detected by autoradiographic, spectrophotometric or other means known in the art.

A method of detecting a tumor cell which contains

RNA encoded by a DNA sequence of this invention is described.

The method comprises isolating RNA from a cell, contacting the RNA isolated from the cell with a detectable marker which

binds specifically to at least a portion of the RNA encoded by an activated oncogene and detecting the marker so bound. The presence of bound marker indicates the presence of a tumor cell. Specifically mentioned tumor cells which may be detected by the method of the present invention are those arising from astrocytoma and glioblastoma cells.

The detectable marker may be a labelled DNA sequence, including a labelled cDNA sequence, having a nucleotide seq¬ uence complementary to at least a portion of the DNA sequence of this invention, produced by methods known to those skilled in the subject art.

The detectable marker may also be a labelled ribo- nucleotide (RNA) sequence haying a nucleotide sequence com¬ plementary to at least a portion of the DNA sequence of this invention, and may be isolated by methods known to those skilled in the art.

Detectable markers of this invention will be labelle with commonly employed radioa&tive labels, i.e. 32P, although other labels may be employed. The markers will be detected by auto adiographic,- spectrophotometric or other means known in the art. Methods to be used for the isolation of DNA or RNA for the practice of this invention are well known in the subject art (cf. ref. 13).

A method of determining the predisposition of a subject to a disease associated with a DNA sequence of this invention is presented. The method involves isolating the genomic DNA of a cell from the subject, contacting the DNA so isolated with a detectable marker which specifically binds to at least a portion of an activated oncogene of this inven- tion and detecting the marker so bound. The presence of bound marker indicates a predisposition of the subject to the disease. Specifically mentioned diseases to which a predis¬ position of a subject may be determined by the method of the present invention are those which arise from human tumor ^' cell lines, especially those of astrocytoma and glioblastoma cells.

The detectable marker may be a labelled DNA sequence, including a labelled cDNA sequence, having a nucleotide seq- uence complementary to at least a portion of the DNA sequence of this invention, produced by methods known to those skilled in the subject art.

The detectable marker may also be a labelled ribo- nucleotide (RNA) sequence having a nucleotide sequence com¬ plementary to at least a portion of the DNA sequence of this invention, and may be isolated by methods known to those skilled in the art.

Detectable markers of this invention will be label- led with commonly employed radioactive labels, i.e. 32P, although other labels may be. employed. The markers will be detected ^' by autoradiographic, spectrophotometric or other means known in the art.

A method of determining the predisposition of a subject to a disease is disclosed which comprises isolating the RNA from a cell from the s-abject, contacting the RNA so isolated with a detectable marker which specifically binds to at least a portion of the RNA encoded by an activated on¬ cogene of this invention and detecting the marker so bound, the presence of bound marker indicating a predisposition of the subject to the disease. Specifically mentioned diseases to which a predisposition of a subject may be determined by the method of the present invention are those which arise from human tumor cell lines, especially those of astrocytoma and glioblastoma cells.

The detectable marker may be a labelled DNA sequence, including a labelled cDNA sequence, having a nucleotide seq- uence complementary to at least a portion of the DNA sequence of this invention, produced by methods known to those skilled in the subject art.

The detectable marker may also be a labelled ribo- nucleotide (RNA) sequence having a nucleotide sequence com- ple entary to at least a portion of the DNA sequence of this

invention, and may be isolated by methods known to those skilled in the art.

Detectable markers of this invention will be label- led with commonly employed radioactive labels, i.e. 32P, although other labels may be employed. The markers will be detected by autoradiographic, spectrophotometric or other means known in the art.

A polypeptide molecule encoded by an activated on¬ cogene is also provided by this invention. The polypeptide has the properties of transforming NIH3T3 cells, inducing a tumor when injected into nude mice and of acting as a tyrosine- specific protein kinase. The polypeptide has an amino acid sequence substantially as shown in Figure 1.

The presently preferred polypeptide will be encoded by a ros oncogene, preferably by the rosl oncogene, and will be expressed in a human, although the polypeptide may be expressed in a variety of other organisms.

The polypeptide of -fe is invention may be obtained by synthetic means, i.e. chemical synthesis of the polypep- tide from its component amino acids, by methods known to those skilled in the art. In the presently preferred embodi¬ ment, the polypeptide may be obtained by isolating it from cells expressing the rosl gene, i.e. a cloned rosl gene in a bacterial cell, or by _irι vitro translation of the mRNA encoded by the rosl gene to produce the polypeptide of this invention. Techniques for the isolation of polypeptides by these means are well known to those skilled in the art.

A method of detecting a tumor cell is also provided. The method involves isolating a cell, contacting the cell with a detectable marker which binds specifically to at least a portion of a polypeptide of this invention and detecting the presence of marker bound to the cell. The presence of marker so bound indicates that the cell may be a tumor cell. Specifically mentioned tumor cells which may be detected ^' by the method of the present invention are those arising from

astrocytoraa and glioblastoma cells.

The detectable marker will preferably be a labelled immunoglobulin molecule, although other markers known in the art may be employed. The immunoglobulin molecule may be an antibody produced by contacting the immune system of an ani¬ mal with at least a portion of the polypeptide of this inven¬ tion, or with a synthetic amino acid sequence substantially similar to a portion of a polypeptide of this invention. The antibody molecule may also be produced by a combination of recombinant DNA techniques with other techniques known in the art.

Detectable markers of this invention will be commonly employed markers such as heavy metals, radioactive, e.g. 35S, fluorescent, e.g. fluoresceln, or enzymatic, e.g. peroxidase. Detection of the labelled markers may be carried out by auto- radiographic, spectrophotometric, or colorimetric techniques or by other methods known in the art.

A method for detecting ^' a tumor in. subject is also presented. This method comprises contacting the tumor with a detectable marker which specifically binds at least a por¬ tion of a polypeptide of this invention and detecting the marker so bound. The presence of bound marker indicates .the presence of a tumor. Specifically mentioned tumors which may be detected by the method of the present invention are those which arise from astrocytoma and glioblastoma cells.

A presently preferred method comprises introducing into the bloodstream of the patient a detectable amount of the marker such that the marker contacts and binds to a tumor expressing the polypeptide encoded by an activated on- cogene, the tumor being detectable thereby.

The presently preferred marker is a labelled immuno¬ globulin molecule although other markers known in the art may be employed. The immunoglobulin molecule may be an anti¬ body produced by contacting the immune system of an animal with at least a portion of the polypeptide of this invention, or

with a synthetic amino acid sequence substantially similar to a portion of a polypeptide of this invention. The antibody may also be produced by a combination of recombinant DNA techniques with other techniques known in the art. Presently preferred labels for the antibody mole¬ cules are radioopaque labels, i.e. heavy metals, or enzymatic markers, known in the art.

A serum-based method for detecting a tumor in a subject is also presented. The method comprises isolating serum from the subject, contacting the serum with a first detectable marker which binds specifically to at least a por¬ tion of a polypeptide of this invention and detecting the marker so bound. The presence of bound marker indicates the presence of a tumor. The first detectable marker is preferably a labelled antibody which may be free, i.e. in a solution, or may be bound to a matrix, i.e. a matrix such as polystyrene beads or the wall of a tube.

Presently preferred labels for the antibody mole- cule are radiolabels, i.e. 35S, heavy metals, or enzymatic labels. The bound antibodies will be detectable by auto- radiography, scintillation counting or by colorimetry or by various other means.

The second detectable marker of this invention will also be labelled antibody molecule and may specifically bind to the polypeptide or to the first marker or to a com¬ bination. The second marker will be labelled by radiolabels, heavy metals or enzymes, and detectable by means similar to those used to detect the first marker. The second marker may also be detectable by eye if it causes precipitation of the first marker-polypeptide complex.

Finally, a method for treating in a subject a tumor induced by an activated rosl oncogene is disclosed. The method comprises isolating an immunoglobulin molecule which specifically binds to at least a portion of a polypep-

tide encoded by the activated rosl oncogene, attaching to the immunoglobulin molecule so isolated a substance which substantially limits the growth of a tumor or which destroys tumors to produce an antitumor immunoglobulin molecule, and contacting the tumor with an effective amount of the anti¬ tumor immunoglobulin molecule so produced, thereby limiting tumor growth or destroying the tumor. Specifically mentioned tumors in a subject which may be treated according to the method of the present invention are those arising from astro- cytoma and glioblastoma cells.

The immunoglobulin molecule is preferably obtained by contacting the immune system of an animal with at least a portion of a polypeptide encoded by the associated rosl on¬ cogene, or with a synthetic ^* amino acid sequence substantially similar to a portion of the polypeptide, to produce a speci¬ fic antibody molecule, -'and isolating the antibody therefrom. The immunoglobulin molecule may also be produ-ced by a-combin¬ ation of recombinant DNA techniques and ^' other, techniques known in the art. Substances which substantially limit the growth of a tumor or which destroy tumors are known in the art, and may be molecules such as interferon, tumor necrosis factors, or radioactively labelled amino acids.

FIRST SERIES OF EXPERIMENTS

MATERIALS AND METHODS Cell culture, cotransfectiσn and tumoriσenicitv assays NIG3T3 cells at 8x10 cells/plate were cotransformed with 300 ng of pKOneo plasmid DNA (32) and 5 micrograms of each of the cosmid fragments. The selection for G418 anti¬ biotic resistance and the tumorigenicity assay were performed as previously described (8). Cell cultures were established from excised tumors following surgical removal and mincing, and were maintained in Dulbecco's medium plus 10% calf serum under our standard culture conditions. Nomenclature for

tumors and cell lines derived from them is as follows. Independent secondary tumors and cell lines derived from MCF- 7-3 DNA are called MCF-7-3n where n is a number. Independent tertiary tumors and cell lines derived from MCF-7-3n DNA are called MCF-7-3n-m where m is a number.

CONSTRUCTION OF LIBRARIES DNAs were prepared from tumors or human placentas as previously described (8). Geno ic libraries were con- structed from placental or MCF-7-3-4 tumor DNAs by partial cleavage with EcoRI and cloning into the cosmid vector pHC79 (10). Appropriate fragments from a previously isolated lambda library (8) were used. for colony filter hybridization. Additional overlapping cosmid clones were then isolated by hybridization with appropriate probes isolated from cosmid clones. cDNAs were synthesized from polyA mRNA- isolated from the nude mouse tumor derived cell line MCF-7-3-7 (13). The cDNA library was constructed in lambda gtlO (11). Phages containing mcf3 cDNAs were isolated by plaque hybridization, initially with the EcoRI fragment 1.4 kbp in length isolated from cosmid clone 101, and later with fragments isolated from cDNA clones.

DNA and DNA analysis Southern blot analyses under high and low stringency conditions were performed as previously described (25,19). DNA sequences were determined by the dideoxy method of Sanger et al. (21) as modified by Biggin et al. (2). Both strands of the coding sequences of the mcf3 cDNA were sequenced. To localize the exon sequences. Southern analysis was performed using cDNA fragments as probes. A synthetic oligonucleotide of the sequence 5'-CCAACTATAATAGTAAGTATG-3 ' , which corresponds to the noncoding strand of the sequences encoding amino acids 10 to 15, was used as probe to localize the exon encoding ^' the transmembrane domain. This oligonucleotide hybridized to the

EcoRI fragment 4.8 kbp in length, and was used as primer for sequencing double stranded DNA fragments from both placental and mcf3 cosmid clones, by a modification of the dideoxy method (6). The oligonucleotides used in this study were synthesized on an Applied Biosystems DNA Synthesizer and puri¬ fied from polyacrylamide gels as described (35).

RNA was prepared from cell lines by the guanidiniu / hot phenol method (13) and analyzed on Northern blots as des¬ cribed (28). RNA protection studies with labeled DNA as hybridization probes were performed as described (15).

RESULTS STRUCTURE OF. THE mcf3 LOCUS A set of charon 4 phage clones containing portions of the mcf3 locus was previously described (8). To obtain a complete set of clones from the mcf3 locus, we chose to use the cosmid cloning vector pHC79 (10) was chosen and a cosmid library was constructed from the DNA of a "secondary" nude mouse tumor MCF-7-3-4. DNA sequences from the charon phages were initially used as probes to isolate clones from the cos¬ mid library. Overlapping cosmids were cloned until clones which contained sequences not found in all tumors derived from cotransformation with DNA from MCF-7-3 were obtained. An EcoRI map of this 70 kbp region is depicted in Figure 1. Only EcoRI fragments to the left and including the 5.6 kbp fragment, and to the right and including the 0.4 kbp fragment were found in all mcf3 transformants.

In order to demonstrate that the entire transform¬ ing gene had been cloned and to determine the location of this gene in the 70 kb long stretch of cloned DNA, cosmid clones from different regions of the mcf3 locus were linear¬ ized at the unique Sal I site in the cosmid vector and tested for their ability to induce the tumorigenic phenotype in co- transfected NIH3T3 cells. None of the individually tested cosmids scored positive in the assay, indicating that the

transforming gene extends over a major part of the mcf3 locus. Similarly, no pairwise combination of cosmids was able to induce tumorigenicity in NIH3T3 cells. Therefore, cotrans- formation experiments were performed with three overlapping pieces of DNA: two linearized cosmids and a third cosmid fragment, which together span the whole locus. The pieces used were cosmid clones 101 and 209, linearized at their unique Sal I sites, and the Cla I fragment of cosmid clone 115. The combination of all three fragments scored positive in the cotransfection and tumorigenicity assay, whereas no combination of two fragments did (data not shown). Therefore, to form an intact mcf3 locus, two homologous recombination events had to take place. This shows that the entire mcf3 oncogene has been cloned. - It is clearly a very large gene, which may extend over nearly 70 kb of DNA.

To compare the structure of the mcf3 locus with normal cellular DNA, a cosmid library was constructed from human placental DNA. Appropriate DNA fragments from the mcf3 locus were used for screening this library, and five indep- endent cosmid clones were obtained. The DNAs present in these clones could not be aligned over their full length with the mcf3 locus. Additional cloning and Southern analysis indicated that the different structure of the placental cos¬ mid clones and the mcf3 locus was not an artifact of cosmid cloning. The mcf3 locus was created by DNA rearrangement involving the fusion of at least three separate fragments of human DNA.

The first of these is a fragment of human DNA located at what we later show to be the 5 '-end of the mcf3 oncogene and it extends from the left end of the map of the mcf3 locus to the EcoRI fragment 3.7 kbp in length. This fragment is connected by DNA of unknown origin to another fragment of the human genome which contains the majority of the coding sequences of mcf3. The second piece of human DNA extends from the EcoRI fragment 4.8 kbp in length to the

right end of the map of the mcf3 locus. The DNA rearrange¬ ments which created the mcf3 locus are probably of functional significance since they span the regions that were required in the cotransfection studies described above.

ANALYSIS AND IDENTIFICATION OF THE mcf3 ONCOGENE

Since the mcf3 gene is large, cloning cDNAs to the transcript of the mcf3 oncogene was necessary for the further analysis of the gene. To localize the transcription unit in the mcf3 locus, polyA+ RNA from a cell line established from the- secondary nude mouse tumor MCF-7-3-7 was analysed. Various fragments from the mcf3 locus were used as probes in Northern blots. The 1.4 kbp EcoRI fragment contained in cosmid 101, detected several RNA species of 2.8 kbp to 3.3 kbp length in RNA from MCF-7-3-7 cells. These RNA species were not found in RNA from normal NIH3T3 cells. The EcoRI fragment 1.4 kbp in length was then used as probe for screening ^' a cDNA library. The library was constructed by__ ^' cloning of cDNAs synthesized from polyA+ ^' RNA of the MCF-7-3-7 cell line into the lambda gtlO cloning vector (11). The 3 '-end of one transcript was localized in cDNA clone M3.9, which contained a polyA÷ tail. This also determined the direction of transcription in the mcf3 locus which was subsequently verified by SI mapping.

Heterogeneity in the structure of the isolated cDNA clones was revealed when they were hybridized to genomic mcf3 DNA. The heterogeneity was confined to sequences derived from the 5'-end of the gene. This indicates that different splicing events occur at the 5'-end of the mcf3 transcripts. All the isolated cDNA clones contained a portion with common structure which hybridized to a common set of genomic frag¬ ments from the mcf3 locus. These common sequences are trans¬ cribed from, and completely contained within, the 3 '-portion of the locus which derives from the long contiguous piece of human DNA. To obtain information about the nature of the mcf3

transfor ing protein, the sequence of the open reading frame in the common portion of the mcf3 cDNAs was determined. These nucleotide sequences and the predicted amino acid sequences are shown in Figure 1. Initial computer analysis of the amino acid sequence indicated that the mcf3 protein sequence was similar to all oncogenes encoding tyrosine kinases, but closely related to none. When the sequence of v-ros, the transforming gene of the avian sarcoma virus UR2, was pub¬ lished (16), it was immediately obvious that the mcf3 protein was closely related to v-ros. From amino acid 51 to 370 of mcf3, 75% homology exist between the two proteins. Only at the C-terminus do they differ considerably, where the mcf3 encoded protein contains 99 additional amino acids not found in the v-ros protein. The close homology between v-ros and ^' mcf3 was surprising since in previous studies a fragment of the mcf3 gene, represented in RNA transcripts, showed no homology to a panel of cloned retroviral oncogenes which included v-ros (8). However, the fragment of mcf3 used in the earlier studies encodes the C-terminal part of the rosl protein, and has limited homology to v-ros.

The close homology between the chicken derived v-ros gene and the common part of the mcf3 cDNAs suggests that the 3' portion of the mcf3 locus derives from the human counterpart of the v-ros gene. To determine whether this is indeed the case. Southern analysis of total human DNA were performed under conditions of low stringency with two probes, a v-ros and a mcf3 cDNA fragment which encode roughly analo¬ gous sequences of the ros proteins. The probe from the mcf3 cDNA hybridized most strongly to two EcόRI fragments, 2.9 kbp and 5.0 kbp in length, which correspond to two fragments of human DNA known to contain exon sequences. The v-ros probe hybridized most strongly with an EcoRI fragment in human DNA 2.9 kbp in length, corresponding to the same fragment of the mcf3 locus. It was concluded therefore that a major portion of the mcf3 coding sequence is derived from the gene in

huraans most closely related to the v-ros gene. We call this gene rosl, since other human genes related to the v-ros or the mcf3 gene may exist.

POTENTIAL MEMBRANE SPANNING DOMAIN AND THE REARRANGEMENT OF THE mcf3 GENE

Five of the oncogenes known to encode oncogenic tyrosine kinases, v-ros (16), v-erbB (33), v-fms (9), neu (1) and trk (14), have hydrophobic potential membrane spanning domains. The membrane spanning domains of these proteins are always encoded 5'-to sequences encoding the kinase domain. Inspection of the mcf3 nucleotide sequence shows that it can encode a highly hydrophobic stretch of 21 amino acids imme¬ diately followed by a stretch rich in positively charge amino acids (see Figure 1, boxed sequence). These features are commonly found in membrane spanning domains. The hydrophobic sequences of v-ros are 30 amino acids in length, longer than in mcf3 and longer than is needed for such a domain.

The portion of the mcf3 cDNA encoding the potential transmembrane domain derived from the human rosl gene or from other parts of the mcf3 locus was also studied. By combined restriction endonuclease analysis. Southern blotting and hybridization with synthetic oligonucleotides, two coding exons were localized and sequenced within the 4.8 kbp EcoRI fragment which contains the breakpoint in the rosl derived part of the mcf3 locus. The positions of the deduced splice junctions are included in Figure 1. One exon encodes seq¬ uences for the putative intracellular domain and for one amino acid of the potential transmembrane domain. The other encodes sequences from the potential transmembrane domain, and for eight amino acids of the putative extracellular domain. This last exon is found also in placental clones containing the human rosl gene and therefore does not derive from rearranged sequences. Horaologs to the eight amino . acids of the putative extracellular domain are not found in

the avian v-ros gene.

There was no consensus structure for cDNA derived from parts of the gene 5' to the breakpoint found in the 4.8 kbp EcoRI fragment. The heterogeneity observed in the struc- ture of the cDNAs probably reflects different splicing pat¬ terns of transcripts arising from the mcf3 oncogene which do not derive from rosl.

Since we do not know which, if any, of our cDNA clones reflect transcripts encoding a transforming protein, we did not realize these portions of our cDNA clones. From the position of the coding region within the cDNA clones and the length .of the longest mcf3 transcript, we estimate the maximal size of a mcf3- encoded protein to be 75 kilodaltons, of which 50 kilodaltons--derive from rosl sequences.

rosl IS NOT REARRANGED IN MCF-7 CELLS

The mcf3 gene derives from a rearrangement involving the human rosl gene, and putafcive extracellular sequences of this gene have been lost. Since similar rearrangements have occurred during the biogenesis of the v-ros (16-), v-erbB (33) and or trk (14) genes, they are probably of functional signi¬ ficance. It was thus ^' of interest to determine whether DNA from the MCF-7 cell line used in the original cotransformation studies contained the rearranged mcf3 locus, or whether this rearrangement was introduced during DNA transfer. The struc¬ ture of the locus in the MCF-7 cell line was analyzed by Southern blotting and compared to either the normal configu¬ ration of the locus in placenta or the rearranged configura¬ tion in MCF-7-3 and its derived tumors. 'The probe used for these experiments contained sequences from the 4.8 kb EcoRI fragment localized in the middle of the mcf3 locus which con¬ tains one of the breakpoints introduced during rearrangement. This probe hybridized to a BamHI fragment approximately 12 kb in length in tumor DNAs containing the rearranged mcf3 locus but not to a band approximately 10 kb in length in DNA

from MCF-7 cells and from placenta. Therefore, the rearrange¬ ment responsible for the creation of the mcf3 locus seems not to have occurred in MCF-7 DNA. The alteration is only found in DNA from MCF-7-3 and its derived tumors and must therefore have occurred during or after DNA transfer into

NIH3T3 cells. The blot hybridization studies also indicate that the mcf3 locus is very highly amplified in transformed NIH3T3 cells, a point which was made previously (8). The activation of the rosl gene was therefore probably an arti- fact of DNA trasfer. To test this hypothesis, RNA from the MCF-7 cell line was analyzed by Northern blotting and by RNA protection studies (see Materials and Methods). Expression of the rosl gene in MCF-7 cells was not detected. If expressed at all, levels of the rosl transcript in MCF-7 cells must be fifty-fold lower than levels found in cells transformed with the mcf3 locus. The oncogenic potential of the human rosl gene may have been activated by rearrangement and gene amplification occurriπag during or after gene trans¬ fer.

EXPRESSION OF rosl

As mentioned earlier, the majority of known oncog¬ enes encode proteins with tyrosine-specific protein kinase activity. However, potential membrane-spanning domains amino-terminal to kinase domains are found only in v-erbB (33), v-fms (a), neu (1), trk (14), and the v-ros and cellular ros genes (16,17). The cellular analogs of the v-erbB and the v-fms genes probably encoded the receptors for the epider¬ mal growth factor and macrophage colony-stimulating factor, respectively. Thus, the cellular ros gene very likely encodes a hormone receptor as well.

As a very first step toward the molecular charac¬ terization of this potential growth factor receptor there was analyzed the expression of rosl in human tumor cell lines ^' . There was used liquid hybridization of total cellular RNA to

Sp -produced RNA probes of high specific activities, followed by digestion of the hot hybridized probe by RNase (15). This technique proves to be highly specific and very sensitive. In a survey of 40 different cell lines (data not shown) , rosl was found to be expressed in astrocytoma and glioblastoma cells at levels ranging between 10-200 messenger molecules per cell. In contrast, rosl was found to be expressed not at all or at very low levels in the remainder of cell lines. From astrocytomas arise a variety of tumors ranging from benign to highly malignant. The malignant forms include the highly malignant glioblastoma multiforme. Interestingly, a good correlation between rosl expression and the degree of malignancy was found. This raises the possibility that the expression of rosl contributes to the malignant phenotype of this particular cell type, and may be helpful in the classi¬ fication and diagnosis of this type of cancer.

DISCS-SSION Data is presented describing the structure of an oncogene called mcf3 (8). The gene was detected by a combin¬ ation of DNA mediated gene transfer and a tumorigenicity assay in nude mice. The DNA originally used in the first transfections came from MCF-7 cells, a human mammary carci¬ noma cell line (8). The mcf3 gene was isolated by molecular cloning and its structure was compared to the structure of its normal counterpart in human placental DNA. The mcf3 on¬ cogene was a product of a major DNA rearrangement. DNA co- transfection studies indicated that this rearrangement span¬ ned functionally important domains of mcf3. The peculiar rearrangement associated with the mcf3 locus was shown to be present only in DNA isolated from transfected and tumorigenic cells, but not in the original donor DNA from the MCF-7 cell line. Since expression of the mcf3 gene was not detected in the MCF-7 cells, it seems unlikely that this gene contri- butes to the transformed phenotype of MCF-7. Thus a func-

tional mcf3 gene was created by a rearrangement introduced during or after gene transfer. A second instance of this was discovered in an independent line of experiments (34). Rearrangements and amplifications also occur in the standard NIH3T3 focus assay, and have been reported to lead to proto- oncogene activation (27). Such events appear to be rare in the focus assay but common with the cotransformation and tumorigenicity assay. Therefore, while the tumorigenicity assay may be unreliable for the detection of oncogenes in tumor DNAs, it may be a good method for searching for proto- oncogenes which can be activated by rearrangement or ampli¬ fication.

Sequence analysis .indicated that mcf3 was related to the v-ros gene of the a-vian retrovirus ϋR2(16). Both mcf3 and v-ros probes hybridize most strongly to the same

DNA fragment ^' in human DNA, and the gene from which mcf3 was derived was designated as rosl. This nomenclature is used since there may be other ros-__elated genes in the human genome, In the human genome, the rosl gene is localized on chromosome 6, bands qll-q22 (20). Translocations of chromosome 6 in this ^* region have been previously observed in human tumors. Thus, although rosl is probably not activated in MCF-7 tumor cells, it is possible that chromosomal alterations affect the rosl locus and lead to an activation of its oncogenic potential in other tumor cells. Although most human tumor cell lines, like MCF-7, do not express detectable levels of rosl, several human tumor cell lines that do contain signi¬ ficant levels of rosl transcripts have been found (prelimi¬ nary results) . The majority of known oncogenes encode proteins with tyrosine specific protein kinase activity. However, a potential membrane spanning domain N-terminal to the kinase domain is found only in v-erbB (33), v-fms (9), neu (1), trk (14) and the v- and c-ros genes (16,17). The cellular aria- logues of the v-erbB and v-fms genes probably encode the

receptors for the EGF and CSF growth hormones, respectively (23,30). Thus, the cellular ros gene very likely encodes a hormone receptor as well. As has been noted before, the tyrosine kinase most closely related to ros is the insulin receptor (29). In particular, a stretch of very high (75%) homology to the insulin receptor exists between position 245 and 288 in rosl, which can be aligned with positions 178 to 215 in the cAMP dependent protein kinase (cAPK). Since cys 198 of the cAPK is protected from chemical modification by peptide substrates, this region has been implicated in sub¬ strate binding (5). The high degree of homology between the amino acid sequences of rosl and the insulin receptor in this putative substrate binding domain might indicate a similar substrate specificity for -t ^' he tyrosine kinase activities of these two proteins.

One effect of the rearrangement which created the oncogenic mcf3 gene is a deletion of all but eight amino acids of the putative extracellular domain of rosl. DNA pieces of unknown origin replaced this part of the rosl gene in the mcf3 locus. In the v-ros gene, the point of fusion between cellular and viral sequences is located in an analo¬ gous position leading to the loss of most of the putative extracellular domain (17). The deletion of the extracellu¬ lar domain may be an important event in the activation of the oncogenic potential of the rosl gene. Similar events have been observed previously for the v-erbB (33) and the c-erbB gene in ALV induced erythroblastosis (18). Insertion of the provirus into the middle of the c-erbB gene leads to the production of a truncated erbB transcript which encodes only 64 amino acids of the extracellular domain but an intact membrane spanning and intracellular domain. Similarly, trk seems to have been formed by a somatic rearrangement that replaced the extracellular domain of a putative transmembrane receptor with the first 221 amino acids of a nonmuscle tr ^' o- po yosin protein (14). The nature of the sequences in mcf3

which have replaced the extracellular domain of the rosl gene has not been analyzed. However, one should be cautious in concluding that the loss of these extracellular sequences has led to the activation of rosl. The amino acid sequence of mcf3 from position 1 to 392 is identical to the deduced cod¬ ing sequence of rosl as determined from human placental DNA, but the corresponding coding sequences C-terminal to amino acid 392 have not yet been determined from rosl. Although gross structural differences between the DNA of mcf3 and rosl coding for the C-terminal part of the protein could not be detected, any subtle changes like point mutations or small deletions would not have been detected. Moreover, the rear¬ ranged rosl gene is very highly amplified in all mcf3 trans¬ formed NIH3T3 cells we have ^" examined. It is not yet possible to assess the relative contributions of rearrangement and amplification of this gene on its oncogenicity.

REFERENCES

1. Barg ann, C.I., M.-C. Hung and R.A. Weinberg (1986). The neu oncogene encodes an epidermal growth factor receptor-related protein. Nature 319;226-230.

2. Biggin, M.D., Gibson, T.J. and Hong, G.F. (1983). Buffer gradient gels and 35S label as an aid to rapid DNA sequence determination. Proc. Natl. Acad. Sci. USA

80/.3963-3965.

3. Bishop, J.M. (1985). Viral oncogenes. Cell 42:23-38.

4. Blair, D.G., Cooper, C.S., Oskarsson, M.K., Eader, L.A. and Vande Woude, G.F. (1982). New method for detecting cellular transforming genes. Science 281:1122-1125.

5. Branson, H.N., Thomas, N.r-Mutsueda, R., Nelson, N.C., Taylor, S.S., Kaiser, E.T., (1982). Modification of the catalytic subunit of bovine heart cAMP-dependent protein kinase with affinity labels related to peptide substrates. J. Biol. Chem. 257:10575-10582.

6. Chen, E.Y., Seeburg, P.H. (1985). Supercoil sequencing: a fast and simple method for sequencing plasmid DNA. DNA 4_:165-170.

7. Cooper, C.S., M. Park, D.G. Blair, M.A. Tainsky, K. Huebner, CM. Croce and G.F. Van Woude (1984). Molecular cloning of a new transforming gene from a chemically transformed human cell line. Nature 311:29-33.

8. Fasano, O., Birnbaum, D., Edlund, L., Fogh, J. and M. Wigler (1984). New human transforming genes detected by a tumorigenicity assay. Mol. Cell. Biol. 4 _> :1695-1705.

9. Hampe, A., Gobet, M. , Sherr, C, Galibert, F. (1984). Nucleotide sequence of the feline retroviral oncogene v-fms shows unexpected homology with oncogenes encoding tyrosin-specific protein kinases. Proc. Natl. Acad. Sci. USA 81:85-89.

10. Hohn, B. and J. Collins (1980). A small cosmid for efficient cloning of large DNA fragments. Gene 11:291-298.

11. Huynh, T.V., R.A. Young and R.W. Davis (1984). Constructing and screening cDNA libraries in lambda gtlO and gtll. In DNA Cloning Techniques: A Practical Approach D. Glover, ed.,.IRL Press, Oxford.

12. Kohl, N.E., Kanda, N., Schreck, R„R., Bruns, G., Latt, S.A., Gilbert, F., and Alt, F.W. (1983). Transposition and amplification of oncogene-related sequences in human neuroblastomas. Cell 35:3-59-367.

13. Maniatis, T., Fritsch, E.F., Sambrook, J. (1982). In: Molecular Cloning, a Laboratory Manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. ^•

14. Martin-Zanca, D., Hughes, S., Barbacid, M. (1986). A human oncogene formed by the fusion of a truncated trophomyosin and tyrosine protein kinase sequences. Nature 319:743-748.

15. Melton, D.A., Krieg, P.A., Rebugliati, M.R., Maniatis, T., Zinn, K., Green, M.R. (1984). Efficient in vitro synthesis of biologically active RNA and RNA hybridization probes from plas ids containing a bacteriophage SP6 promoter. Nucleic Acid Research 12:7035-7056.

16. Neckameyer, W.S. and L.H. Wang (1985). Nucleotide Sequence of avian sarcoma virus UR2 and comparison of its transforming gene with other members of the tyrosine protein kinase oncogene family. J. Virol. 53:879-884.

17. Neckameyer, W.S., Shibuya, M., Hsu, M.-T. and Wang, L.H. (1986). Proto-oncogene c-ros codes for a molecule with structural features common to those of growth factor receptors and displays tissue-specific and developmentally regulated expression. Mol. Cell. Biol. _6:1478-1486.

18. Nilsen, T., Maroney, P., Goodwin R., Rottman, F., Crittenden, L., Raines, M. and Kung, H.-J. (1985). c-erbB activation in ALV-induced erythroblastosis: novel RNA processing and promotor insertion result in expression of an amino-truncated EGF receptor. Cell 41:719-726.

19. Powers, S., Kataoka, T., Fasano, 0., Goldfarb, M. , Strathern, J., Broach, J. and Wigler, M. (1984). Genes in S. cerevisiae encoding proteins with domains homologous to the mammalian ras proteins. Cell 36_:607-612.

20. Rabin, M., Birnbaum, D., Wigler, M. and Ruddle, F. (1985). mcf3 oncogene mapped to region on chromosome 6 associated with malignant trans-formation. A er. J. Hum. Genet. 37:A36.

21. Sanger, F., Nicklen, S. and Coulson, A.R. (1977). DNA sequencing with chain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 74:5463-5467.

22. Schwab, M. , Alitalo, K., Klempnauer, K.-H., Varmus, H.E., Bishop, J.M., Gilbert, F., Brodeur, G., Goldstein, M. and Trent, J. (1983). Amplified DNA with limited homology to myc cellular oncogene is shared by human neuroblastoma cell lines and a neuroblastoma tumour. Nature 305:245-248.

23. Sherr, C, Rettenmier, C, Sacca, R. , Roussel, M. , Look, A. and Stanley, E. (1985). The c-fms proto-oncogene product is related to the receptor for the mononuclear phagocyte growth factor, CSF-1. Cell 41:665-676.

24. Shih, C, Shio, B., Goldfarb, M., Dannenberg, A. and Weinberg, R. (1979). Passage of phenotypes of chemically transformed cells via transfection of DNA and chromatin. Proc. Natl. Acad. Sci. USA 76:5714-5718.

25. Shimizu, K., Goldfarb, M.,_'Suard, Y., Perucho, M., Li, Y., Kamata, T., Feri isco, J., Stavnezer, E., Fogh, J. and Wigler, M.. (1983) ^" . Three human transforming genes are related to the viral ras oncogenes. Proc. Natl. Acad. Sci. 80:2112-2116.

26. Southern, R.J. and Berg, P. (1982). Transformation of mammalian cells to antibiotic resistance with a bacterial gene under control of the SV40 early region promoter. J. Mol. Appl. Genet. 1_:327-341.

27. Takahashi, M., Ritz, J. and Cooper, G.M. (1985). Activation of a novel human transforming gene, ret, by DNA rearrangement. Cell 42:581-588.

28. Thomas, P.S. (1980). Hybridization of denatured RNA and small DNA fragment transferred to nitrocellulose. Proc. Natl. Acad. Sci. USA 77:5201-5206.

29. Ullrich, A., Bell, J.R., Chen, E.Y., Herrera, R. , Petruzelli, L.M., Dull, T.J. ^' , Gray, A., Coussens, L., Liao, Y.-C. , Tsubokava, M. , Mason, A., Seeberg, P.H., Grunfeld, C, Rosen, O.M. and Ramachandran, J. (1985). Human insulin receptor and its relationship to the tyrosine kinase family of oncogenes. Nature 313:756-763.

30. Ullrich, A., Coussens, L., Hayflick, J., Dull, T., Gray, A., Tarn, A., Lee, J., Yarden, γ», Libermann, T., Schlesinger, J., Downward, J., Mayes, E., Whittle, N., Waterfield, M. and Seeburg, P. (1984). Human epidermal growth factor receptor cDNA sequence and aberrant expression of the amplified gene in A431 epidermoid carcinoma cells. Nature 309:418-425.

31. Van Daren, K., Hanahan, D. and Gluzman, Y. (1984). Infection of eucaryotic cells by helper-independent recombinant adenovirus: early region I not obligatory for intergration of viral DNA. J. Virol. 50:606-614.

32. Wigler, M., Sweet, R., Sim, G.K., Wold, B., Pellicer, A., Lacy, E., Maniatis, T., Silverstein, S. and Axel, R. (1979). Transformation of mammalian cells with genes from procaryotes and eucaryotes. Cell 16:777-785.

33. Yamamoto, T., Nishida, T., Miyajima, N., Kawai, S., Ooi, T. and Toyoshima, K. (1983). The erbB gene of avian erythroblastosis virus is a member of the src gene family. Cell 3_5:71-78.

34. Young, D., Waitches, G., Birchmeier, C, Fasano, O. and Wigler, M. (1986). Isolation and characterization of a new cellular oncogene encoding a protein with multiple potential transmembrane domains. Cell, in press.

35. Zoller, M. and Smith, M. (1984). Oligonucleotide directed mutageπesis: A simple method using two oligonucleotide primers and a single stranded DNA template. DNA 3_:479-488.