Vav, a novel human oncogene derived from a locus ubiquitously expressed in hematopoietic cells

The EMBO Journal vol.8 no.8 pp.2283 - 2290, 1989 vav, a novel human oncogene derived from a locus ubiquitously expressed in hematopoietic cells Shulamit Katzav, Dionisio Martin-Zanca and Mariano Barbacid1 Departmental Oncology Section, BRI-Basic Research Program, Frederick Cancer Research Facility, PO Box B, Frederick, MD 21701, USA 'Present address: Department of Molecular Biology, Squibb Institute for Medical Research, PO Box 4000, Princeton, NJ 08543, USA Communicated by J.Schlessinger A novel human oncogene, designated vav, was generated by a genetic rearrangement during gene transfer assays. The vav oncogene directs the synthesis of a 3.0 kb mRNA from which we isolated a 2.8 kb-long complementary DNA copy. Nucleotide sequence analysis of this vav oncogene cDNA clone revealed that its 5' 167 bp were derived from pSV2neo DNA cotransfected as a selectable marker during gene transfer. The remaining 2597 bp were unrelated to genes included in current data banks, indicating that the vav oncogene is likely to be derived from a novel human locus. The vav oncogene cDNA clone encompasses a 2391 bp long open reading frame (ORF) capable of directing the synthesis of a 797 amino acid long polypeptide. The predicted vav oncogene protein sequence exhibits several motifs reminiscent of transcriptional factors. They include a highly acidic amino-terminal region separated from two putative nuclear localization signals by a proline-rich sequence, presumably a hinge region. In addition, we identified two zinc-finger-like domains, one of which conforms to the canonical pattern Cys-X2-Cys-X13-Cys-X2-Cys previously found to confer trans-activating activity to the adenovirus ElA protein. Transcription of its normal allele, the vav proto-oncogene, has been exclusively observed in cells of hematopoietic origin, including those of erythroid, lymphoid and myeloid lineages. These findings raise the possibility that this novel locus might play an important role in hematopoiesis. Key words: vav/oncogene/hematopoietic cells/zinc finger Introduction Malignant transformation has been used as a genetic marker to identify vertebrate loci that play active roles in the regulation of cellular proliferation and/or differentiation. These loci harbor the normal alleles of transforming genes present in acute transforming retroviruses, retroviral-induced tumors and spontaneous and carcinogen-induced malignan- cies (Bishop, 1987). In addition to those oncogenes present in naturally occurring tumours, transforming genes can be generated by rearrangements that occur during gene transfer assays. Some of these in vitro activated oncogenes are homologous to those previously found in human tumors such as trk (Martin-Zanca et al., 1986; Kozma et al., 1988) or in certain retroviruses such as raf(Rapp et al., 1983; Muller and Muller, 1984; Fukui et al., 1985; Shimizu et al., 1985). Other oncogenes including mas (Young et al., 1986), dbl/mcf-2 (Fasano et al., 1984; Eva and Aaronson, 1985), ret (Takahashi et al., 1985), hstlK-fgf (Delli Bovi et al., 1987; Taira et al., 1987), fgf-5 (Zhan et al., 1988), B-raf (Ikawa et al., 1988) and tre (Nakamura et al., 1988) represent previously unidentified loci. Characterization of these transforming genes and their normal alleles will expand the number of loci available for the study of molecular pathways involved in the control of cell proliferation and differentiation. We describe here the isolation and molecular characteriza- tion of vav, a new human oncogene that became activated during the course of gene transfer assays aimed at determining the presence of transforming genes in human esophageal carcinomas suspected of having a chemical etiology. Molecular characterization of this oncogene revealed a novel locus that encodes a zinc finger-containing protein whose expression is restricted to cells of hematopoietic origin. Results Identification of the vav oncogene DNAs isolated from several esophageal carcinomas kindly provided by Dr R.Montesano (WHO, Lyon) were tested for the ability to transform NIH3T3 mouse fibroblasts. Although none of the DNAs was able to induce foci of morphologically transformed cells, one DNA sample was positive in the in vivo nude mouse tumorigenicity assay (Blair et al., 1982; Fasano et al., 1984). NIH3T3 cells transfected with this tumor DNA in the presence of the selectable marker pSV2neo (Southern and Berg, 1982) were expanded in vitro, pooled and inoculated into mice. These cells elicited the appearance of tumors in 9 out of the 10 mice injected in < 4 weeks. Southern transfer analysis ofDNA isolated from these nude mouse tumors revealed the presence of multiple human Alu+ repetitive sequences. When the DNA isolated from the nude mouse tumors was used in additional cycles of transfection, tumors were induced in 100% of the mice inoculated. DNA isolated from second- and third-cycle nude mouse tumors exhibited a consistent pattern of human Alu+ sequences (EcoRI DNA fragments of 30, 9 and 5 kb), a result indicative of the presence of a human oncogene (Figure IA). The human sequences present in these human Alu+ DNA fragments did not hybridize, even under low-stringency conditions, to any of 20 oncogene-specific probes available at the time these experiments were conducted (data not shown). These results indicated that these nude tumors might contain a novel human oncogene. Since this transforming gene represented the sixth distinct oncogene identified in our laboratory, it was designated by the acronym vav, the sixth letter of the Hebrew alphabet, the native language of one of us (S.K.). 2283 S.Katzav, D.Martin-Zanca and M.Barbacid In view of the complexity of the vav locus (>235 kb) we wished to isolate vav cDNA clones. To identify such clones, we generated a genomic probe containing vav exonic sequences. DNA isolated from a third-cycle nude mouse tumor was partially digested with Sau3AI and used to create a genomic library in XEMBL4. One million recombinant phages were screened with a probe consisting of 32P-labeled nick-translated human genomic DNA. One hundred phages were found to be positive. Five plaque-purified phages carrying inserts of 15-20 kb were grown up, digested with restriction endonucleases and several DNA fragments free of human repetitive sequences were isolated by preparative gel electrophoresis. These DNA fragments were screened for the possible presence of exon sequences by hybridizing them to genomic DNAs isolated from several mammalian species. This test was based on the assumption that exonic sequences are more likely to be evolutionarily conserved than intron sequences. Among those DNA fragments tested, a 800 bp BamHI-EcoRI probe efficiently hybridized with single-copy DNA sequences present in the hamster, mouse, mink, macaque and human genomes (data not shown). This probe was subsequently used in Northern transfer analysis to identify a single 3.0 kb mRNA species specifically expressed in vav-induced nude mouse tumors (Figure iB). This transcript is likely to represent the transcriptional product of the vav oncogene. Isolation of cDNA clones Poly(A)-selected RNA isolated from a vav-induced third- cycle nude mouse tumor was used to prepare a 106-member EcoRl cDNA library in the XgtlO cloning vector. When this library was screened with the 800 bp BamHI-EcoRI Alu- probe, 103 phages were positive. Twenty recombinant phages were picked at random from those exhibiting the strongest signal. The longest insert found in these recombinant phages was a 2.3 kb EcoRI fragment, likely to represent an incomplete cDNA clone. To isolate a more representative vav cDNA clone, a 300 bp PstI DNA fragment corresponding to the 5' end of this insert was purified and used to rescreen the cDNA library. Eleven phages carrying inserts of 2.5 -2.8 kb were isolated. The one exhibiting the longest insert was selected for further studies. Transforming activity of the vav oncogene To test whether the vav cDNA clone (2.8 kb) was biologically active, we subcloned it into the unique EcoRI site of pMEX, a mammalian expression vector that carries a multiple cloning site flanked by a Moloney murine sarcoma virus (MSV) long terminal repeat (LTR) and polyadenyla- tion signal from SV40. The resulting expression plasmid, pSK27, induced 103-104 foci per itg of DNA when transfected into NIH3T3 cells (Graham and van der Eb, 1973). A similar construction in which the cDNA clone was inserted in the opposite orientation did not have detectable biological activity (data not shown). vav-induced foci consist of dense, non-refractile cells that closely resemble those that spontaneously appear in dense cultures of NIH3T3 cells (Figure 2A). In addition, the vav-induced foci often exhibit formation of syncytia resulting in giant multinucleated cells (Figure 2B). Cells derived from either NIH3T3 foci or from nude mouse tumors retain a rather normal morphology in subconfluent cultures. However, the vav-transformed NIH3T3 cells grow in semisolid media with relatively good efficiency (0.2-1 %) and readily induce tumors when injected into nude mice. Nucleotide sequence analysis The nucleotide sequence of the biologically active 2.8 kb cDNA clone of the vav oncogene was determined by an automated chain-termination method using a difluores- ceinated primer and laser excitation as described (Brumbaugh et al., 1988). Single-stranded DNA purified from 5' and 3' deletion mutants generated by unidirectional deletions using the Exo III/Mung bean nuclease method were used as templates for sequencing. Figure 3 depicts the entire nucleo- tide sequence of the 2765-bp vav oncogene cDNA clone and the deduced amino acid sequence of its putative gene product. Several biologically relevant features could be deduced from the nucleotide sequence: (i) nucleotides 1-167 were identical to the pSV2neo sequences corresponding to the end of the SV40 early promoter region (nucleotides 1-17) and the beginning of the bacterial Tn5 sequences harboring the neo gene (nucleotides 18-167); (ii) no significant homology could be detected among sequences located between nucleotides 168 and 2765 of the vav cDNA clone and those stored in gene data banks; (iii) the cDNA clone contains a 2391 bp-long open reading frame (nucleotides 111-2501) capable of coding for a polypeptide of 797 amino acid residues with a calculated relative molecular mass of 91 641 daltons; (iv) this open reading frame (ORF) is followed by Fig. 1. (A) Detection of human Alu+ sequences in nude mouse tumors induced by the vav oncogene. DNAs (20 Ag) isolated from (a) NIH3T3 cells, (b) second and (c) third-cycle vav-induced nude mouse tumors were digested with EcoRI and submitted to Southern transfer analysis (Southern, 1975). Hybridization was conducted for 48 h under stringent conditions (42°C in 5 x SSC, 50% formamide, 1 x Denhardt's solution) using 5 x 107 c.p.m. of 32P-labeled nick- translated high mol. wt human DNA. Filters were exposed to Kodak XAR-5 film at -70°C for 20 h in the presence of intensifier screens. X HindHI DNA fragments were used as mol. wt markers. The migration of the three major Alu+ DNA fragments is indicated by arrows. (B) Identification of vav oncogene transcripts. Poly(A)-selected RNA (3 jig) was isolated from (a) NIH3T3 cells and (c) a third-cycle vav-induced nude mouse tumor and submitted to Northern transfer analysis (Lehrach et al., 1977). Nitrocellulose filters were hybridized for 48 h under stringent conditions (50% v/v formrnamide, 42°C) to 5 x 107 c.p.m. of a 32P-labeled nick-translated 800 bp BamHI-EcoRI DNA derived from an Alu- genomic fragment of the vav oncogene. The filter was exposed to Kodak XAR-5 film for 30 min. 28S and 18S RNA of Saccharomyces cerevisae and 23S and 16S RNA of Escherichia coli R-13 were used as mol. wt markers. The migration of the 3.0 kb vav oncogene transcript is indicated by an arrow. 2284 vav, a novel human oncogene .4:_ -Es... 0.' - I t I . VP e~ / . ^l ~ ~~S'* tr t R sA; ^ tr., I.I ...' . ..> lip / w> g '+& 31bi34',Jj)..s z ; 1. i ,! t~~~ . 00 Fig. 2. Morphology of NIH3T3 cells transfected with the vax oncogene cDNA clone under the transcriptional control of retroviral regulatory elements. (A) A representative focus. (B) occasional syncytia resulting in the formation of giant multinucleated cells. 264 bp of 3' non-coding sequences that contain a consensus polyadenylation signal (AATAAA, nucleotides 2738 -2743) followed 14 residues downstream by a short (8 bp) poly(A) tail. Predicted amino acid sequences The nucleotide sequence of the cDNA clone of the vav oncogene predicts the synthesis of a 797 amino acid long polypeptide with the following relevant features (Figure 3): 2285 A aft .... 1^ .. f .-_ N _. Jt e I _ *, O. ,. i.. i `. :- V i i x nf;, ..Idir .kl f .t f.-: 11 0 .,P I "N -.4. *. i Ili, ..i -1 . ... .1 .. .v v : S.Katzav, D.Martin-Zanca and M.Barbacid 4! '..**** .*. -4 -t+ 2286 vav, a novel human oncogene (i) the 19 amino-terminal residues are coded by sequences derived from the bacterial Tn5 sequences harboring the neo gene, albeit in a different reading frame than the bacterial enzyme; (ii) there are six potential N-glycosylation sites (Asn-X-Ser/Thr), two of which are located within the rearranged sequences contributed by the bacterial Tn5 gene; (iii) a 45 amino acid long domain (residues 84-128) composed of mostly acidic (Glu/Asp) amino acid residues (23 residues or 51 %); (iv) two proline-rich sequences including Pro-Pro-Ser-Pro (residues 289-292) and Pro-Pro- Pro-Pro (residues 558-561) that may represent hinge regions needed for the appropriate folding of the vav oncogene product; (v) an Arg-Arg-Gly-Asp-Ser-Tyr motif (residues 387-392) that may be a phosphorylation site for protein kinase A (Creighton, 1984); (vi) two putative nuclear localization signals, Lys-Thr-Arg-Glu-Leu-Lys-Lys-Lys (residues 438-445) and Lys-Lys-Asp-Lys-Leu-His-Arg-Arg (residues 527 -534) (Dingwall and Kaskey, 1986); and (vii) two putative zinc-finger-like motifs (Sunderman and Barber, 1988). The first of these motifs, Cys-X2-Cys-X13-Cys- X2-Cys (residues 480-500), conforms to the canonical pattern Cys-X2-4-Cys-X2 15-Cys-X2-4-Cys (Berg, 1986). Moreover, this type of motif has been found to confer trans- activating activity to the larger (289 amino acids) of the two proteins coded for by the adenovirus ElA oncogene (Culp et al., 1988). The second motif, His-X2-Cys-X6-Cys- X2-His (residues 505-518), exhibits two histidine residues in the flanking positions. Although histidine residues often replace cysteines in the Cys-X2-4-Cys domain, this particular structure has not been previously described. These features, taken together, suggest that the vav oncogene might encode a DNA-binding phosphoprotein, perhaps a transcriptional factor (Ptashne, 1988). Mechanism of activation The above nucleotide sequence suggests that the vav oncogene became activated by a rearrangement that placed SV40 regulatory sequences present in the cotransfecting pSV2neo plasmid DNA in front of a novel cellular proto- oncogene. Since nucleotide 1 of our vav cDNA clone is located 46 bp downstream from the major site used in vivo for the initiation of SV40 early mRNA synthesis (SV40 nucleotide 5237), it is likely that transcription of vav oncogene sequences is directed by SV40 regulatory elements. Transcriptional activation of the vav oncogene by the SV40 early promoter raised the possibility that this transforming gene might have been generated during the course of gene transfer assays. To prove this hypothesis, DNA isolated from the original esophageal carcinoma was tested by Southern transfer analysis along with DNA prepared from white blood cells of healthy donors and from vav-induced nude mouse tumors. When the DNAs were probed with a 900 bp PstI DNA fragment derived from the 5' end of the vav oncogene cDNA clone, only the mouse tumor DNAs induced by this oncogene contained rearranged sequences (data not shown). These results indicate that the vav oncogene became activated during in vitro manipulations and is unlikely to play a role in the genesis of human esophageal tumors. To determine whether additional mutations were required to activate the vav oncogene, we isolated its normal allele from a cDNA library of K562 cells (Shtivelman et al., 1985). K562 is a hematopoietic cell line derived from a chronic myelogenous leukemia which expresses significant levels of a 2.9 kb-long vav-related transcript. Two million recombinant XgtlO phages were screened with the 5' 900 bp PstI DNA fragment of the vav oncogene. Five phages were found to be positive, and the one containing the longest insert (2.8 kb) was subcloned into Bluescript KS and used in sub- sequent studies. Comparison of the restriction endonuclease map of this normal vav cDNA clone with that of its transforming allele revealed complete identity except at their respective 5' termini (data not shown). Nucleotide sequence analysis of the 5' sequences of the normal vav gene transcript confirmed that they were replaced by those derived from pSV2neo during the generation of the vav oncogene (Figure 4). Next, we prepared a chimeric vav gene construct in which the 2.6 kb Sac -EcoRI DNA fragment of the vav oncogene was replaced by the corresponding sequences derived from the normal vav proto-oncogene (Figure 5). The resulting plasmid, pSK77, exhibited a transforming activity (10 f.f.u./tg) comparable to that of the vav oncogene (Figure 5). These findings demonstrate that insertion of the pSV2neo sequences in the 5' domain of the normal vav proto-oncogene was sufficient for its malignant activation. vav proto-oncogene expression In order to gain information regarding the physiological role of the normal human vav locus, we screened a battery of human cell lines from distinct lineages for expression of proto-vav gene sequences. Representative results are depicted in Figure 6. None of the epithelial, mesenchymal or neuro- ectodermal cell lines tested expressed detectable levels of the normal 2.9 kb proto-vav gene transcript (Table I). In contrast, this transcript could be readily observed in mRNAs prepared from cell lines of hematopoietic origin (Figure 6). Cell lines representative of each of the three major hematopoietic differentiation lineages-lymphoid, myeloid and erythroid-were found to express approximately equivalent levels of the 2.9 kb proto-vav gene transcript Fig. 3. Nucleotide sequence of a vav oncogene cDNA clone. (A) Schematic diagram. Untranslated sequences are depicted by a thin bar. Coding sequences (flanked by the initiator ATG and terminator TGA codons) are represented by the thicker box. Highlighted domains include sequences derived from pSV2neo (17), a region rich in acidic amino acid residues (-), two proline-rich stretches (M), a potential target for protein kinase A phosphorylation (a), two putative nuclear localization signals (3) and two zinc-finger-like domains (EO). Cysteine residues (0) and consensus glycosylation sites (v) are also shown. (B) Nucleotide and predicted amino acid sequence of the 2765 bp insert of the vav oncogene cDNA clone. The sequences of the flanking EcoRI linkers have been omitted. Nucleotide numbers are indicated in the right column. Amino acid numbers are indicated underneath the corresponding residue. Nucleotides I - 17 (SV40 early promoter) and 18- 167 (TnS bacterial gene) are derived from pSV2neo. Nucleotide 168-2765 are derived from the vav proto-oncogene. The vertical arrow (+) indicates the separation between pSV2neo and vav proto-oncogene-derived sequences. The presumed initiator ATG codon is boxed and the terminator TGA codon is marked with asterisks. Other in- frame termination codons are underlined. A consensus polyadenylation signal (AATAAA) is underlined by a wavy line. Sequences encompassing each of the domains highlighted in the above schematic diagram are indicated as follows: cysteine residues are shaded; consensus glycosylation sites (Asn-X-Ser/Thr) are underlined by broken lines; the highly acidic region is indicated by solid black underlining; proline rich sequences are underlined by open bars; the putative nuclear localization signals are indicated by a hatched underlining; the potential site for protein kinase A phosphorylation is underlined by a crossed bar: and the two zinc-finger-like sequences are boxed. 2287 S.Katzav, D.Martin-Zanca and M.Barbacid Hindill Hin TGGCCCAGGCCCTCCGGGATGGTGTCCTTCTGTGTCAGCTGCrTAACAACCTGCTACCCCATGCCATCAACCTGCGTGAG5TC7 11 11 II I I I 11IAGCGGAACACGTAGAAAGCCAGTCCGCAGAAACGGTGCTGACCCCG ATGTCAGCTACTGGGCTATCTGGACAAGGGAA, CCCCCAGATGTCCCAGTTCCTGTGCCTTAAGAACArrAGAACCTTCCTGTCCACCTGCTGTGAGAAGrTCGGC CTCAAGCGGAC 11 1111111111111111111111111111111111111111111111111111111111111111111CGCAAAGAGAAAGCAGTTCCTGTGCCTTAAGAACATTAGAACCTTCCTGTCCACCTGCTGTGAGsAGrTCGGCCTCAAGCGGAC GTGTGAAC 93 (Table I). Similar results were obtained when normal human CTAAGGA 63 B-cells, T-cells (either untreated or PHA-stimulated), CI 184 monocytes or platelets were used instead of cell lines (Table ,IIIGCAAG 154 I). These results indicate that the vav proto-oncogene is Sacl specifically expressed in cells of the hematopoietic system. ,GCrAW"T 275 Sacl a 245 Discussion Fig. 4. Mechanism of activation of the vav oncogene. Comparative nucleotide sequence analysis of the 5' domains of the vav proto- oncogene and oncogene cDNA clones. Vertical lines indicate identical nucleotides. The putative initiator codon (ATG) of the vav oncogene is boxed. In-frame terminator codons are underlined. The black vertical arrow indicates the breakpoint between pSV2neo and vav gene sequences. Diagnostic restriction endonuclease cleavage sites are also indicated. Ec7 pSK 27 MSV-LTR _H 103 Fig. 5. Transforming activity of vav cDNA clones. Expression plasmids including pSK27, which contains the entire vav oncogene cDNA sequences (hatched box), and pSK77, which encompasses a chimeric vav gene containing 5' vav oncogene (hatched box) and 3' vav proto-oncogene (open box) sequences, were tested for their ability to transform NIH3T3 cells in gene transfer assays (Graham and van der Eb, 1975). Sequences derived from pSV2neo are indicated by a solid black box. Coding cDNA sequences are represented by the thick boxes. Non-coding sequences are indicated by thin boxes. Transcriptional regulatory elements (MSV-LTR) and polyadenylation signal (PA) present in the pMEX expression plasmid are also indicated. ... Fig. 6. vav proto-oncogene expression in human cells. Poly(A)-selected RNAs were isolated from (a) K562; (b) MOLT-4; (c) RPMI-6666; (d) CMS; (e) U937; (f) KG-1; (g) IM-9; (h) CCRF-CEM; (i) A431; (j) HOS; (k) T98G; (1) IMR32; (m) Y79; (n) A172; (o) M413; and (p) SK-N-SH human cell lines. The developmental lineages of each of these cell lines are described in Table I. RNAs (3 ig) were submitted to Northern transfer analysis (Lehrach et al. 1977) and hybridized for 48 h under stringent conditions (42°C in 5 x SSC, 50% formamide, 1 x Denhardt's solution) to 32P-labeled nick-translated probes specific for the vav proto-oncogene cDNA clone (2.8 kb EcoRI DNA fragment) and chicken ,B-actin (2 kb HindIII DNA fragment of plasmid B2000, Cleveland et al., 1980). In panel A, the vav proto-oncogene probe was stripped from the filter prior to hybridization of the (3-actin probe. In panel B, both probes (7 x 107 c.p.m. of each) were mixed in the hybridization reaction. Hybridized filters were exposed to Kodak XAR-5 film at -700C for 4 h in the presence of intensifer screens. Molecular weight markers are those described in the legend to Figure 1. The migration of the vav proto-oncogene and 3-actin transcripts is indicated by arrows. The vav oncogene was generated by a genetic rearrangement that replaced its 5' domain by sequences derived from the bacterial Tn5 gene present in the cotransfecting pSV2neo DNA used as a selectable marker in gene transfer assays. This rearrangement also resulted in the transcriptional activation of vav gene sequences presumably by the neighboring SV40 early promoter contained within pSV2neo. At the present time, we do not know whether the trans- forming properties of the vav oncogene are due to its ectopic expression in NIH3T3 cells, the modification of its amino- terminal domain or a combination of both. Isolation of full- length cDNA clones of the vav proto-oncogene should provide conclusive information regarding the mechanism of activation of this human gene. Analysis of the predicted amino acid sequence of the vav gene product reveals a particularly hydrophilic molecule rich in both acidic and basic residues. Glutamic acid is the most abundant amino acid residue (64 residues or 8.4%). Other highly charged amino acid residues are lysine (53 residues), arginine (52 residues) and aspartic acid (46 residues). In total, charged amino acids represent 31.1 % of all residues. Acidic residues are concentrated between positions 84 and 128 (23 out of these 45 residues are either glutamic acid or aspartic acid with a net charge of -23) with total exclusion of basic residues. Acidic domains are believed to be involved in protein-protein interactions (Sigler, 1988), thus suggesting that the vav gene product might be part of a functional protein complex. In addition, the predicted sequence of the vav protein contains two clusters of basic residues that could represent nuclear localization signals (Dingwall and Kaskey, 1986) and a putative site for protein kinase A phosphoryla- tion (Creighton, 1984). These features suggest that the vav gene might encode a nuclear phosphoprotein. The predicted sequence of the vav gene product exhibits two zinc-finger-like domains. Putative zinc finger domains have been identified in transcriptional factors such as the yeast GAL4 and the Xenopus TFIIIA proteins; several morphogenic gene products of Drosophila; the mammalian transcriptional activator SPI; and the steroid and retinoic acid receptor families (Berg, 1986; Evans, 1988; Sunderman and Barber, 1988). However, it has been recently pointed out that zinc fingers might not be exclusive to DNA-binding proteins (Frankel and Pabo, 1988). In the vav protein the two zinc-finger-like motifs and the N-terminal acidic region are separated by at least a cluster of prolines (residues 289-292) that are likely to serve as a hinge region separating two domains presumably involved in DNA binding and transcriptional activation. This structural feature has been found in well-characterized transcriptional activators such as the yeast GCN4 and GAL4 proteins, the jun oncogene and the steroid receptor family, independently of whether they mediate their DNA interactions with zinc finger motifs or not (Ptashne, 1988). Analysis of vav proto-oncogene transcripts in cells of various developmental lineages has revealed a truly 2288 proto-vav proto-vav vav proto-vav vav 11111111 GCCGAGCTC -------------------------- PA.............. :oRl Sacl ,j--4 EcoRl EcoRl Sac pSK 77 MSV-LTR vav, a novel human oncogene Table I. vav proto-oncogene expression in human cells Hematopoietic cells vav Non-hematopoietic cells vav proto-oncogene proto-oncogene expression expression Lymphoid Epithelial Raji + HeLA IM-9 + A431 - Normal B-cells + MCF-7 CCRF-CEM + PA- I - MOLT-4 + TERA2 - RPMI-6666 + M413 - Normal T-cells + Monocytic Mesenchymal CMS + HOS - KG-I + HT-1080 - U-937 + Normal monocytes + Multipotential Neuroectodermal K562 + U-373 MG - HEL + HS 683 - HEL + hemina + SK-N-SH HEL + TPAb + A172 - Y79 _ IMR 32 - T98G - Other Other Normal platelets + HuACIlI aErythroid lineage. bMonocytic lineage. remarkable pattern of expression. Whereas no vav mRNA could be detected in cell lines of epithelial, mesenchymal or neuroectodermal origin, high levels of vav proto-oncogene expression were observed in each of the hematopoietic cells tested, including those of lymphoid, myeloid and erythroid lineages. The ubiquitous expression of the proto-vav gene in hematopoietic cells, along with the resemblance of its gene product to transcriptional factors, raise the possibility that this locus might be involved in the transcriptional machinery required for the proliferative maintenance of the hematopoietic system. Materials and methods Cell lines and gene transfer assays Human cell lines were obtained from the American Type Culture Collection (ATCC), except for CMS (Monaco et al., 1982) and HuACI1 (Peebles et al., 1973). Platelets, and B and T lymphocytes were isolated as described (Timmen and Saksela, 1980; Abe et al., 1986). Isolation of monocytes was performed by using counterflow centrifugal elutriation (Wahl et al., 1983). HEL cells were treated with the phorbol ester TPA and with hemin accor- ding to published procedures (Papayannopoulou et al., 1983; Larson and Papayannopoulou, 1985). NIH3T3 mouse cells were transfected with 20 Ag of high mol. wt cellular DNA or with 1 jig of plasmid DNA by the calcium phosphate precipitation technique (Graham and van der Eb, 1973). Foci of transformed cells were scored after 10-14 days. Tumorigenicity assays were performed as described (Blair et al., 1982; Fasano et al., 1984). Tumor appearance was followed up to 2 months after inoculation. Isolation of genomic and cDNA clones A vav genomic library was made in XEMBL-4. DNA isolated from a third- cycle vav-induced nude mouse tumor was partially digested with Sau3AI and fractionated on sucrose gradients. DNA of 15-20 kb was ligated to BamHI-digested XEMBL-4 DNA. 106 phage clones were screened by hybridization with 32P-labeled nick-translated high mol. wt human DNA under stringent conditions (42°C in 5 x SSC, 50% formamide, 1 x Denhardt's solution). Filters were washed three times at room temperature with 2 x SSC, 0.1 % SDS and twice at 55°C with 0. Ix SSC, 0.1 % SDS. The vav cDNA library in XgtIO was prepared from poly(A)-selected RNA of a third-cycle vav-induced nude mouse tumor using a cDNA cloning kit (Amersham). One million phages were screened under stringent conditions using a 32 P-labeled nick-translated 800 bp BamHI-EcoRi Ala- vav genomic DNA fragment as a probe. Among those recombinant XgtlO phages isolated, that containing the longest (2.8 kb) insert was selected for further studies. Its insert was subcloned into the EcoRI site of Bluescript KS in both orientations to generate pSK33 and pSK47. A XgtlO cDNA library (2 x 106 clones) prepared from human K562 cells (Shtivelman et al., 1985) was hybridized under stringent conditions to a 32P-labeled nick- translated 2.8 kb vav cDNA probe. A recombinant phage carrying a 2.9 kb vav proto-oncogene insert was isolated and subcloned in Bluescript KS in both orientations to generate pSK65 and pSK66. Southern and Northern transfer analysis High mol. wt DNA was digested to completion with appropriate restriction endonucleases, electrophoresed in 0.8% agarose gel and submitted to Southern transfer analysis as described (Southern, 1975). Total cellular RNA was extracted by the guanidium thiocyanate method (Chirgwin et al., 1979) and purified by centrifugation through cesium chloride. Poly(A)-containing RNA was isolated by retention on oligo(dT) columns (Collaborative Research). Total RNA (10 jig) or poly(A)-selected RNA (3 jig) were submitted to Northern transfer analysis (Lehrach et al., 1977). The nitrocellulose filters were hybridized to various 32P-labeled nick-translated probes for 48 h under stringent conditions (42°C in 5 x SSC, 50% formamide, 1 x Denhardt's solution). Nucleotide sequencing A series of nested deletions were generated from pSK33 and pSK47 by the combined use of Exonuclease Ill and Mung bean nuclease (Stratagene). Escherichia coli MV 1193 cells were transformed with the corresponding deletion mutants and single-strand phages rescued by subsequent infection with the helper M13 K07 phage. Single-stranded DNAs were prepared from 2289 S.Katzav, D.Martin-Zanca and M.Barbacid these phages and submitted to nucleotide sequence analysis by an automated chain termination method using primers with multiple fluorophores (Brumbaugh et al., 1988). To dctzrmine the sequence of the 5' domain of the normal vav gene, a 270 bp SacI fragment from pSK65 was subcloned in both orientations in Bluescript KS. Single-stranded DNA was obtained from these plasmids and used as a template for sequencing by the dideoxynucleotide chain termination technique (Sanger et al., 1977) using the Sequenase kit (USB). Expression plasmids pSK27 was generated by subcloning the entire 2.8 kb EcoRI vav oncogene cDNA insert of pSK33 into the unique EcoRI site of pMEX. pMEX is a mammalian expression vector in which the polylinker sequences of pUC 18 (minus the HindIll recognition site) are flanked by a Moloney MSV-LTR and the polyadenylation signal of SV40. pSK77 was generated by subcloning the 5' 270 bp EcoRI-Sacl DNA fragment of the vav oncogene and the 3' 2.6 kb SacI-EcoRI DNA fragment of the vav proto-oncogene into the EcoRI site of pMEX. Acknowledgements We are indebted to J.Brumbaugh and D.Conway for their help in determining the nucleotide sequence of the vav oncogene. We would also like to thank R.Montesano for providing tumor samples, E.Canaani for the K562 cDNA library, M.Ernst for help with gene transfer assays, and F.Coulier, S.Hughes and B.Stanton for helpful discussions. Research sponsored by the National Cancer Institute, DHHS, under contract no. NOI-CO-74101 with Bionetics Research, Inc. The contents of this publication do not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products or organizations imply endorsement by the US Government. Papayannopoulou,T., Nakamato,B., Yokochi,T. Chait,A. and Kannagi,R. (1983) Blood, 62, 832-845. Peebles,P.T., Fischinger,P.J., Bassin,R.H. and Papageorge,A.G. (1973) Nature New Biol.,, 242, 98-101. Ptashne,P. (1988) Nature, 335, 683-689. Rapp,U.R., Reynolds,F.H. and Stephenson,J.R. (1983) J. Virol., 45, 914-924. Sanger,F., Nicklen,S. and Coulson,A.R. (1977) Proc. Natl. Acad. Sci. USA, 74, 5463-5467. Shimizu,K., Nakatsu,Y., Sekiguchi,M., Hokamura,K., Tanaka,K., Terada,M. and Sugimura,T. (1985) Proc. Nat!. Acad. Sci. USA, 82, 5641 -5645. Shtivelman,E., Lifshitz,B., Gale,R.P. and Canaani,E. (1985) Nature, 315, 550-554. Sigler,P.B. (1988) Nature, 333, 210-211. Southern,E.M. (1975) J. Mol. Biol., 98, 503 -5 17. Southern,P.J. and Berg,P. (1982) J. Mol. Appl. Gen., 1, 327-341. Sunderman,F.W. and Barber,A.M. (1988) Ann. Clinic. Lab. Sci., 18, 267-288. Taira,M., Yoshida,T., Miyagawa,K., Sakamoto,H., Terada,M. and Sugimura,T. (1987) Proc. Natl. Acad. Sci. USA, 84, 2980-2984. Takahashi,M., Ritz,J. and Cooper,G.M. (1985) Cell, 42, 581-588. Timmen,T. and Saksela,E. (1980) J. Immunol. Methods, 36, 285-291. Wahl,L.M., Katona,I.M., Wilder,R.L., Winter,C.C., Haroui,B., Scher,I. and Wahl,S.M. (1984) Cell. Immunol., 85, 373-383. Young,D., Waitches,G., Birchmeier,C., Fasano,O. and Wigler,M. (1986) Cell, 45, 711-719. Zhan,X., Bates,B., Hu,X. and Goldfard,M. (1988) Mol. Cell. Biol., 8, 3487-3495. Received February 20, 1989; revised March 29, 1989 References Abe,H., Rossio,J.L., Ruscehi,F.W., Matsushima,K. and Oppenheim,J.J. (1986) J. Immunol. Methods, 90, 111-123. Berg,J.M. (1986) Science, 232, 485-487. Bishop,M. (1987) Science, 235, 305-310. Blair,D.G., Cooper,C.S., Oskarsson,M.K., Eader,L.A. and Vande Woude,G.F. (1982) Science, 281, 1122-1125. Brumbaugh,J.A., Midderdorf,L.R., Grone,D.L. and Ruth,J.L. (1988) Proc. Natl. Acad. Sci. USA, 85, 5610-5614. Chirgwin,J., Przybryla,A., MacDonald,R. and Rutter,W. (1979) Biochemistry, 18, 5294-5299. Cleveland,D.W., Lopata,M.A., MacDonald,R.J., Cowan,N.J., Rutter,W.J. and Kirschner,M.W. (1980) Cell, 20, 95-105. Creighton,T.F. (1984) Proteins: Structural and Molecular Principles. Freeman, New York, pp. 83-84. Culp,J.S., Webster,L.C., Friedman,D.J., Smith,C.L., Huang,W.-J., Wu,F.Y.-H., Rosenberg,M. and Ricciardi,R.P. (1988) Proc. Natl. Acad. Sci. USA, 85, 6450-6454. Delli Bovi,P., Curatola,A.M., Kern,F.G., Greco,A., Ihmarm,M. and Basilico,C. (1987) Cell, 50, 729-737. Dingwall,C. and Kaskey,R.A. (1986) Annu. Rev. Cell. Biol., 2, 367-390. Eva,A. and Aaronson,S.A. (1985) Nature, 316, 273-275. Evans,R. (1988) Science, 240, 889-895. Fasano,O., Birnbaum,D., Edlund,L., Fogh,J. and Wigler,M. (1984) Mol. Cell. Biol., 4, 1695-1705. Frankel,A.D. and Pabo,C.O. (1988) Cell, 53, 675. Fukui,M., Yamamoto,T., Kawai,S., Marvo,K. and Toyoshima,K. (1985) Proc. Natl. Acad. Sci. USA, 82, 5954-5958. Graham,F.L. and van der Eb,A.J. (1973) Virology, 52, 456-467. Ikawa,S., Fukui,M., Veyama,Y. Tamaok,N., Yamamoti,T. and Toyoshima,K. (1988) Mol. Cell. Biol., 8, 2651-2654. Kozma,S.C., Redmond,S.M.S., Xiao-Chung,F., Saurer,S.M., Groner,B. and Hynes,N.E. (1988) EMBO J., 7, 147-154. Larson,A.E. and Papayannopoulou,T. (1985) Exp. Hematol., 13, 129-135. Lehrach,H., Diamond,D., Wozney,J.M. and Boedtker,H. (1977) Biochemistry, 16, 4743-4751. Martin-Zanca,D., Hughes,S. and Barbacid,M. (1986) Nature, 319, 743-748. Monaco,G., Vigneti,E., Lancieri,M., Cornaglia-Ferraris,P., Lambertenghi- Deliliers,G. and Revoltella,R. (1982) Cancer Res., 42, 4182-4189. Muller,R. and Muller,D. (1984) EMBO J., 3, 1121-1127. Nakamura,T., Hillova,J., Mariage-Samson,R. and Hill,M. (1988) Oncogene Res., 2, 357-370. 2290