Proviral Integrations at the Evi5 Locus Disrupt a Novel 90 Kda Protein with Homology to the Tre2 Oncogene and Cell-Cycle Regulatory Proteins

Home , Locus (genetics)

Proviral integrations at the Evi5 locus disrupt a novel 90 kDa protein with homology to the Tre2 oncogene and cell-cycle regulatory proteins

Xiaobei Liao1, Yubin Du1, Herbert C Morse III1, Nancy A Jenkins2 and Neal G Copeland2

1Laboratory of Immunopathology, National Institute of Allergy and Infectious Diseases, Bethesda, Maryland 20892; 2Mammalian Genetics Laboratory, ABL-Basic Research Program, NCI-Frederick Cancer Research and Development Center, Frederick, Maryland 21702, USA

Evi5 is a common site of retroviral integration in T-cell AKR/J parent (Gilbert et al., 1993). In contrast, seven lymphomas of AKXD mice. Mapping studies have strains die predominantly of B-cell tumors, one dies localized Evi5 to a region approximately 18 kb upstream predominantly of myeloid tumors and seven die of a of another common viral integration site, G®1, on mouse mixture of T- and B-cell tumors (Mucenski et al., chromosome 5 (Liao X, Jenkins NA and Copeland NG, 1986; Gilbert et al., 1993). (1995a). J. Virol., 69, 7132 ± 7137). G®1 encodes a zinc A common viral integration site that has been ®nger transcription factor involved in interleukin-2 identi®ed in AKXD lymphomas is Evi5 (Liao et al., signaling. To determine if Evi5 encodes a gene separate 1995a). Among 283 AKXD T-cell, B-cell, and myeloid from G®1 that might also be involved in T-cell disease, tumors screened for Evi5 rearrangements, ten tumors we have searched within the Evi5 locus for novel showed rearrangements. All but one of these tumors transcripts. A 6.0 kb transcript was identi®ed in these was a T-cell lymphoma. These results suggest that studies that spans the Evi5 locus and is disrupted by viral Evi5 encodes a gene primarily involved in T-cell integration at Evi5. This transcript is expressed in all disease. Subsequent mapping studies localized Evi5 to embryonic and adult mouse tissues examined. While mouse chromosome 5 approximately 18 kb upstream blast searches indicated that Evi5 is a novel gene, of another common viral integration site, G®1 (Liao et homologies were detected between Evi5 and a known al., 1995a). G®1 was identi®ed as a common viral oncogene, Tre2, as well as mammalian and yeast cell integration site during progression of Moloney murine cycle regulatory proteins. Evi5 thus encodes a gene leukemia virus (MuLV)-induced rat interleukin-2 (IL- separate from G®1 that may also be involved in T-cell 2)-dependent T-cell lymphoma lines to IL-2-indepen- disease. dent growth (Gilks et al., 1993). Proviral integration at G®1 upregulates the expression of a novel zinc Keywords: Evi5; G®1; leukemia; oncogene; Tre2 ®nger protein that is expressed at low levels in IL-2- dependent T-cell lymphoma lines. Studies in mitogen- stimulated splenocytes suggests that G®1 is a down- stream target of the IL-2 receptor and functions

Introduction during the transition from the G1 to the S phase of the cell cycle. Consistent with this hypothesis, over- Retroviruses induce disease in susceptible hosts via expression of G®1 does not induce expression of IL-2 insertional mutagenesis of cellular proto-oncogenes or but does contribute to the emergence of the IL-2- tumor suppressor genes (Ihle et al., 1990; Peters, 1990; independent phenotype (Gilks et al., 1993). van Lohuizen and Berns, 1990; Berns, 1991). The Like Evi5, G®1 is also a common viral integration disease genes that are aected by viral integration can site in AKXD lymphomas. Among 283 tumors thus be identi®ed as common sites of retroviral analysed, 14 had G®1 rearrangements and 12 were integration in tumor DNAs. One set of mouse strains T-cell lymphomas (Liao et al., 1995a). G®1 thus also that has proven particularly valuable for the study of appears to encode a gene that predisposes to T-cell retrovirally induced leukemias is the AKXD recombi- disease in the AKXD strains. Among the tumors nant inbred (RI) strains (Mucenski et al., 1986; with G®1 rearrangements, one also carried a Gilbert et al., 1993). The 21 AKXD strains were rearrangement at Evi5, although it was not possible derived by crossing mice from two inbred mouse to tell if these rearrangements were in the same or strains that dier in their lymphoma incidences, AKR/ dierent subpopulations of tumor cells (Liao et al., J and DBA/2J. AKR/J is the prototypic highly 1995a). lymphomatous inbred mouse strain; nearly all of An important question arising from these studies is these animals develop T-cell lymphomas by 7 to 16 whether the Evi5 locus encodes a gene separate from months of age. DBA/2J is a strain with a low G®1 that also predisposes to T-cell disease or whether incidence of lymphoma. Among 21 highly lymphoma- viral integrations at Evi5 merely serve to upregulate tous AKXD strains available for study, only six expression of G®1. In studies described here, we show strains die predominantly of T-cell tumors, like the that the Evi5 locus does in fact encode a gene that is separate and distinct from G®1. While blast searches indicated that Evi5 is a novel gene, homologies were detected between Evi5 and a known oncogene, Tre2, Correspondence: NG Copeland as well as mammalian and yeast cell cycle regulatory Received 7 August 1996; revised 24 October 1996; accepted 31 proteins. Evi5 thus encodes a gene separate from G®1 October 1996 that may also be involved in T-cell disease. Evi5 homology to cell cycle regulators XLiaoet al 1024 were identi®ed 18 nucleotides upstream of the polyA Results tail (Figure 2, bottom). The 3' UTR contains 12 copies of the mRNA destabilizing sequence AUUUA (Figure Evi5 transcripts 2, Shaw and Kamen, 1986), suggesting that Evi5 To search for transcripts from the Evi5 locus Northern mRNA expression may be tightly regulated. blots of normal adult mouse tissues were probed with genomic probes from the Evi5 locus. One probe (probe Evi5 encodes a novel coiled-coil protein with homology S) derived from cellular DNA located to the left of to a known oncogene and cell cycle regulatory proteins provirus 82-6 (Figure 1; Liao et al., 1995a), identi®ed a 6 kb transcript in all tissues tested with the highest BLAST searches (Devereux et al., 1984) indicated that level of expression detected in liver (data not shown). the Evi5 gene encodes a novel protein. Searches for Probe S was then used to screen a mouse liver cDNA known protein coding motifs identi®ed four potential N- library (Stratagene) in order to isolate cDNA clones linked glycosylation sites (aa 107, 126, 495, 685), a from the Evi5 gene. A number of overlapping cDNA potential tyrosine kinase phosphorylation site (aa 415), a clones were obtained. Sequence analysis of the longest potential serine protein kinase C phosphorylation site clone (CP7) indicated that it was approximately 4.5 kb (aa 113) and several potential casein kinase II in length and was derived from sequences located at phosphorylation sites. The N-terminus of Evi5 is also the 3' end of the Evi5 gene (Figure 2, top). To isolate rich in serine and proline residues. The region of Evi5 more 5' clones a new cDNA library was constructed from amino acids 50 to 150 contains 27% serine and the using an antisense Evi5 oligonucleotide primer for ®rst region from amino acids 50 to 100 contains 10% proline strand cDNA synthesis that was homologous to residues, which represents nearly a threefold increase in sequences located near the 5' end of CP7. One clone the frequency of such residues when compared to the obtained from this library (S5A-1; Figure 2, top) protein as a whole. A second proline rich region (13.6% extended the Evi5 sequence to within *0.5 kb from the proline) was also identi®ed in the Evi5 C-terminus (aa expected 5' end of the gene. The Evi5 5' end was then 750 ± 809). Secondary structure predictions indicated obtained by anchor-ligation mediated 5' RACE (see that Evi5 is a coiled-coil protein (Figure 3). Materials and methods). The 5' RACE reproducibly yielded a PCR product of *450 bp, which was cloned into PCR-script (R6-1, Figure 2, top).

The Evi5 sequence The Evi5 sequence is shown at the bottom of Figure 2. The sequence is 5870 nucleotides in length and predicts a short 68 bp 5' UTR, a 2427 bp open-reading-frame (ORF) and a relatively long 3375 bp 3' UTR. The ®rst ATG of the predicted ORF is located within a good Kozak translation initiation sequence (Kozak, 1986) and is preceded by an in-frame TAG stop codon at bp 51 ± 53. A canonical polyadenylation signal sequence (AAUAAA, Proudfoot and Brownlee, 1976) was not found near the 3' end of the gene, however, two overlapping alternative polyadenylation signals (AA- TAAT and AATTAA, Wickens and Stefenson, 1984)

Figure 1 Restriction map of the Evi5 common viral integration site. Abbreviations for restriction endonucleases: B, BamHI; V, EcoRV; H, HindIII. The positions and orientations (5' to 3')of proviruses integrated at Evi5 are shown above the restriction map. The type of integrated provirus, when known, is also indicated: E, Figure 2 Evi5 sequence. The rectangle at the top of the ®gure ecotropic; M, mink cell focus-forming. The locations of Evi5 depicts the overall structure of the Evi5 protein. The coding exons are indicated by rectangles on the restriction map with region is hatched while the untranslated regions are open. The tbc hatched regions indicating coding sequences and open regions box (aa 169 ± 348) is shown in black. Three Evi5 cDNA clones indicating untranslated sequences. The direction of Evi5 used for sequencing are shown below this depiction as thickened transcription is from right to left. The location of genomic lines with their name designations on the right. The Evi5 sequence probe S is indicated by a ®lled rectangle. The location of two is shown at the bottom of the ®gure. Twelve ATTTA 3'UTR mouse genomic lambda clones (lg2 and lg4) are shown below the destabilizing sequences are boxed. Two potential polyA addition restriction map signals are underlined Evi5 homology to cell cycle regulators XLiaoet al 1025

Figure 3 Evi5 secondary structure predictions. Secondary structure predictions were made using the MacVector 5.0.2 software using algorithms developed by Chou and Fasman (1974a,b)

While Evi5 encodes a novel protein, BLAST searches identi®ed a number of interesting regions of homology between Evi5 and known proteins. For example, the N- terminal region of Evi5 was found to contain a recently identi®ed 180 ± 200 amino acid domain called the tbc domain (Richardson and Zon, 1995; Figure 4). While the function of this domain is not well de®ned, it is likely to be involved in regulating protein-protein Figure 4 Amino acid sequence comparison of the Evi5 tbc box interactions. Other genes that contain a tbc domain with other tbc box-containing proteins. Alignment of the Evi5 (aa include Tbc1, Tre2, BUB2 and cdc16 (Figure 4). Tbc1 169 ± 348) and Tbc1 tbc boxes was carried out using the PILEUP was identi®ed in a screen for genes that are dierentially program provided by GCG (Devereux et al., 1984). All other tbc boxes were aligned to the Tbc1 tbc box as described previously regulated during mast cell development (Richardson (Richardson and Zon, 1995). Amino acid residues that were and Zon, 1995). Tbc1 encodes a nuclear protein and is identical in three or more tbc box proteins are indicated in bold expressed at highest levels in hematopoietic cells, testis and kidney. Within these tissues, expression of Tbc1 is cell- and stage-speci®c suggesting that Tbc1 may play a role in the cell cycle and dierentiation of various Evi5 is a member of a well conserved multigene family tissues. Tre2 is a human oncogene isolated during transfection of NIH3T3 cells with human Ewing's Chromosome mapping studies using a 1 kb Evi5 3' sarcoma DNA (Nakamura et al., 1988, 1992). BUB2 UTR cDNA probe (bps 3421 ± 4437) or an Evi5 5' (Hoyt et al., 1991; Wang and Burke, 1995) and cdc16 coding region probe (S5A-1 (bps 348 ± 1446), Figure 2) (Chang and Nurse, 1993; Fankhauser et al., 1993) are detected a single locus that mapped to mouse yeast genes involved in the regulation of mitosis during chromosome 5. These results are consistent with the cell cycle. previous mapping studies which localized the Evi5 The C-terminal region of Evi5 shows signi®cant common viral integration site to chromosome 5 (Liao homology to a number of coiled-coil proteins including et al., 1995a). However, when a more 3' coding region several myosin heavy chain proteins and mitosin (also probe (bps 1370 ± 3548) was used for mapping two called CENP-F), a novel *350-kilodalton nuclear additional loci, which mapped to chromosomes 10 and phosphoprotein involved in regulating mitotic-phase 18 were detected (Figure 5). cell cycle progression (Zhu et al., 1995; Liao et al., Zoo blot analysis suggests that the chromosome 10 1995b). and 18 loci represent Evi5-related genes and not pseudogenes. Hybridization of the 3' Evi5 cDNA probe to DNAs from multiple dierent species Proviral integrations at the Evi5 locus are located within identi®ed two, and often times three, Evi5-hybridizing the coding region of the Evi5 gene fragments in every species tested, indicating that all Two mouse genomic lambda clones (lg2 and lg4, three genes are evolutionarily well conserved. (Figure Figure 1) that span the entire Evi5 common integration 6). site (Liao et al., 1995a) were used to determine the location and orientation of proviral integrations with Evi5 is widely expressed respect to the Evi5 gene. An oligonucleotide probe spanning bps 2191 ± 2211 of the Evi5 cDNA hybridized Hybridization of an adult mouse tissue Northern blot to both lambda clones and mapped to the left of the with an Evi5 3' UTR probe identi®ed an *6kb proviral integrations at Evi5. Genomic sequence transcript in all tissues analysed (Figure 7). This analysis indicated that these sequences were contained hybridization pattern was similar to that obtained within a single exon which included the entire Evi5 3' previously with the mouse genomic probe S (Figure 1; UTR and the last 104 amino acids of the Evi5 protein data not shown). Evi5 is also expressed in the embryo (bps 2185 ± 2496; Figure 1). The direction of Evi5 where Evi5 transcripts could be detected in all transcription was from right to left and in the same embryonic stages examined (E7-E17, Figure 7). transcriptional orientation as G®1 (Figure 1). Two Evi5 is thus widely expressed, at least at the other oligonucleotide probes spanning bps 2166 ± 2184 RNA level, in both embryonic and adult mouse and 2162 ± 2178 of the Evi5 cDNA failed to hybridize tissues. In vitro transcription and translation of a to either lambda clone indicating that the proviral 2.9 kb Evi5 construct containing the entire Evi5 open integrations are located within a single 3' intron of the reading frame produced a 90 kDa protein, which is Evi5 gene. The proviral integrations at Evi5 are thus the predicted size of the full-length Evi5 protein likely to disrupt normal Evi5 expression. (Figure 8). Evi5 homology to cell cycle regulators XLiaoet al 1026

Figure 5 Evi5 maps to mouse chromosome 5 while two Evi5-related loci map to chromosomes 10 and 18. The chromosomal location of Evi5 and two Evi5-related loci (Evi5-rs1 and Evi5-rs2) were determined by interspeci®c backcross analysis. The segregation patterns of the Evi5 loci and ¯anking genes are shown at the top. For individual pairs of loci more animals than are shown in the haplotype analysis were often typed. These additional animals were used in calculating the recombination distances. Partial chromosome linkage maps showing the location of the Evi5 loci in relation to linked genes are shown at the bottom. Recombination distances between loci in centiMorgans (+ the standard error) are indicated to the left of each map. Where no recombinant animals were identi®ed the upper 95% con®dence limit of the distance between the loci is given in parentheses. References for human map positions can be obtained from GDB (Genome Data Base), a computerized database of human linkage information maintained by the William H Welch Medical Library of the Johns Hopkins University (Baltimore, MD) e y n Heart Brain Splee Lung Liver Muscl Kidne Testes E7 E11 E15 E17 Hamster Guinea pig Rat Mink Dog Cat Chicken Turkey Sheep Human Mouse Evi5 —23 —9.4 —6.6 —4.4 GAPDH

Figure 7 Northern blot analysis of Evi5 expression. A mouse multiple tissue Northern blot and a mouse embryo Northern blot —2.3 (Clontech) was hybridized with an Evi5 probe. The 6.0 kb Evi5 —2.0 transcript was detected in all embryonic and adult tissues examined. The same blot was hybridized with a GAPDH probe to control for loading

whether viral integration at Evi5 merely serves to Figure 6 Zoo blot analysis suggests a family of highly conserved activate expression of the G®1 zinc ®nger gene or Evi5-related genes. DNAs from a number of dierent species were whether Evi5 encodes a separate disease-causing gene. hybridized at low stringency with an Evi5 coding region probe. The numbers to the right indicate the sizes in kb of HindIII- In studies described here, we show that Evi5 does in digested lambda DNA fragments run in a parallel lane of the fact encode a separate gene that is oriented in the same same gel transcriptional direction as G®1 and is disrupted by viral integration at Evi5. Evi5 is widely expressed in both embryonic and Discussion adult mouse tissues and is highly conserved throughout evolution suggesting that Evi5 serves an important The Evi5 locus is a common site of viral integration in function in the cell. Two Evi5-related loci that map to AKXD T-cell lymphomas (Liao et al., 1995a). mouse chromosomes 10 and 18 were also identi®ed by Mapping studies have localized this common site to a chromosome mapping. Like Evi5, zoo blot analysis region approximately 18 kb upstream of another indicates that these two Evi5-related genes are well common viral integration site, G®1 (Liao et al., conserved throughout evolution indicating that they 1995a). G®1 encodes a zinc ®nger transcription factor too may serve important functions in the cell. thought to be involved in IL-2 signaling (Gilks et al., Genomic analysis showed that the Evi5 common 1993). An important question raised by these studies is viral integration site is located within an intron in the Evi5 homology to cell cycle regulators XLiaoet al 1027 The gene with the highest homology to Evi5 is Tbc1. l While little is known about this protein, it has been shown to encode a nuclear protein that is dierentially

Contro Evi5 expressed during mast cell dierentiation (Richardson and Zon, 1995). Tbc1 may thus play a role in the mast cell dierentiation. The Tre2 oncogene was generated

by DNA recombination during transfection of NIH3T3 ¨ 90 kDa cells with human Ewing's sarcoma DNA (Nakamura et al., 1988). Tre2 has been proposed to function as a tumor suppressor by inhibiting protein degradation mediated by the ubiquitin/deubiquitin system (Natka- mura et al., 1988, 1992; Papa and Hochstrasser, 1993). A number of genes involved in growth control are short-lived and are regulated by protein degradation involving the ubiquitin/deubiquitin system. Tre2 could cause tumors by interfering with the degradation of these growth control proteins. cdc16, isolated from S. pombe (Frankhauser et al., 1993; Frankhauser and Figure 8 In vitro-translated Evi5 protein analysed by SDS ± PAGE. A full-length Evi5 protein product of approximately Simanis, 1994), and BUB2, its functional homologue in 90 kDa can be seen in lane 2 along with a number of minor S. cerevisiae (Hoyt et al., 1991; Wang and Burke, smaller products. A vector control is shown in lane 1 1995), are implicated in mitotic spindle checkpoint control. These genes are thought to achieve this control by checking on the state of mitotic spindle formation at the end of mitosis and, if not complete, arresting the 3' coding region of the Evi5 gene. Viral integrations cell in mitosis until spindle formation is complete. How into 3' coding regions of genes are fairly common and this arrest is caused is not entirely clear but have been reported for a number of other genes experiments suggest that arrest occurs by preventing including Myb (Nazarov and Wol, 1995), Tpl2 the inactivation of the cdc2 mitotic kinase that (Makris et al., 1993), Tiam1 (Habets et al., 1994), normally is required for exit from mitosis (Hoyt et Int3 (Robbins et al., 1992) and Int6 (Marchetti et al., al., 1991; Frankhauser et al., 1993). 1995). Such integrations could inactivate the gene or The C-terminal half of Evi5 shows signi®cant produce truncated protein products with altered homology to a number of coiled-coil proteins biological activity. In the case of Myb and Tpl2, viral including mitosin, a protein involved in cell cycle integrations are in the same transcriptional orientation control (Liao et al., 1995a; Zhu et al., 1995). Mitosin is as the gene and produce a truncated protein. These a *350 kDa nuclear phosphoprotein that is expressed proteins are likely to be oncogenic as they lack only during the S and G2/M phases of the cell cycle. negative regulatory domains located in the truncated Mitosin undergoes a dynamic spatial reorganization to sequences (Introna et al., 1994; Patriotis et al., 1994). several components of the mitotic apparatus (e.g. the Viral integrations in Tiam1 and Int3 are also oriented centromere and mitotic spindle) during M-phase in the same direction as the gene. However, instead of progression and this redistribution coincides closely producing truncated proteins, novel 3' end products, with changes in phosphorylation. Overexpression of N- which are initiated in viral 3' long terminal repeat terminally truncated mutants block cell cycle progres-

(LTR) sequences, are produced. These 3' end products sion at G2/M suggesting that mitosin plays an are also likely to be oncogenic (Jhappan et al., 1992; important role in mitotic-phase progression. Robbins et al., 1992; Habets et al., 1994). In contrast, Given the homology of Evi5 to known oncogenes Int6 integrations are in the opposite orientation as the and cell cycle regulators it is tempting to speculate that gene and prematurely terminate Int6 transcripts due to Evi5 is itself a T-cell disease gene involved in cell cycle a cryptic transcription stop signal present in the viral control. This would not rule out the possibility that LTR. The biological signi®cance of this truncated viral integration at Evi5 also serves a second purpose, transcript has not been determined. Like Int6, viral namely, to activate expression of G®1. Consistent with integrations at Evi5 are in the opposite transcriptional this possibility, ZoÈ rnig and colleagues (Schmidt et al., orientation as the gene. However, as RNA from the 1996; ZoÈ rnig et al., 1996) have recently shown that tumors with viral integrations at Evi5 is not available, viral integration at Evi5 activates G®1 expression. It is it has not yet been possible to determine whether these thus possible that the Evi5 gene cooperates with G®1 in viral integrations produce a truncated protein or T-cell disease. Future studies involving transgenic mice suppress Evi5 expression. Experiments are in progress may help resolve this question. to identify additional tumors with viral integrations at Evi5 so that the aect of these integrations on Evi5 expression can be directly determined. BLAST searches indicated that Evi5 is a novel Materials and methods protein. A number of interesting homologies were detected, however, between Evi5 and known genes. Northern blot hybridization Within the N-terminal half of Evi5 (aa 169 ± 348) a tbc RNA blots were purchased from Clontech Laboratories, box, a probable regulatory domain for protein-protein Inc. (Palo Alto, CA). Northern blot hybridizations were interactions, was identi®ed. Other tbc box-containing performed as previously described (Church and Gilbert, proteins include Tbc1, Tre2, cdc16 and BUB2. 1984). Evi5 homology to cell cycle regulators XLiaoet al 1028 Southern blot hybridization cDNA library constructed by Stratagene custom library service using random primers (Stratagene) and an Evi5 Southern blot hybridizations using 32P-labeled oligonucleo- antisense primer, 5A (bps 1514 ± 5132 of Evi5), derived tide probes were performed under standard conditions from the 5' end of sequence of Evi5 cDNA clone cp7: 5'- (Maniatis et al., 1982) at 428C and washed at 428Cwith CACTTTATCCTGCATCTCC-3'. Plaque hybridizations 46SSC, 0.1% SDS. Zoo-blot cross species hybridization were performed under the standard conditions (Maniatis was performed as previously described (Gilbert et al., 1993) et al., 1982). and the blots were washed at 658C with 46 SSC and 0.1% SDS. DNA sequencing Probes Both strands of cDNA clones were sequenced by the Probe S, isolated from the 5' proviral-cellular junction of Sanger method of dideoxynucleotide chain termination BXH-2 tumor 82-6 as described previously (Liao et al., reactions (Sanger et al., 1977), using a Sequenase Kit 1995a), was used for detecting Evi5 transcripts and for the (USB). initial cDNA library screening. cDNA clones S5A-1 (see Figure 2), S610-C and a 5' fragment, P7-U, of cDNA clone CP7 (see Figure 2), were used for chromosomal mapping Ligation-mediated 5' RACE studies. S610-C, was a 1 kb cDNA clone of the 3' UTR of The 5'-end clone of Evi5 cDNA was obtained by Evi5; P7-U was a 2.3 kb DNA fragment generated by 5' RACE (rapid ampli®cation of cDNA ends), using a EcoRI and XbaI double-digest of cDNA clone CP7. Probe 5'-AmpliFINDER RACE kit (Clontech) and two nested P400 was a 400 bp PstI fragment of CP7 (bps 1449 ± 1899 antisense Evi5 primers derived from the 5'- end sequence of of Evi5) and was used for Zoo-blot and Northern blot cDNA clone S5A-1, 90R: 5'-TGATACAAGAGAG- analysis. Three oligonucleotide probes were used for GAACC-3', (bps 413 ± 430 of Evi5 cDNA) and 70R: 5'- hybridization studies of lambda clones of the 3' genomic ACCACTGTTTCTTCTTGACC-3', (bps 396 ± 415 of Evi5 region of Evi5 to determine intron and exon regions: 5'- cDNA). PolyA-selected RNA was extracted from livers of GATGCCTAAAAGGCCAGCGGG-3' (bps 2191 ± 2211); 6 weeks old C57BL/6J mice using a mRNA isolation kit 5'-ATAGCAGAGCTGACTCATG-3' (bps 2166 ± 2184); (Stratagene). 90R was used for synthesis of the ®rst strand and 5'-TCAGCTCTGCTATCTGG-3' (antisense, bps of cDNA; primer 70R and the 5'-AmpliFINDER anchor 2162 ± 2178). primer were used for the ampli®cation of the RACE fragment of 5' end of the Evi5 cDNA. The PCR product Interspeci®c backcross mapping was cloned into PCR-script vector (Stratagene), which was designated cDNA clone R-6 (see Figure 2). Interspeci®c backcross progeny were generated by mating

(C57BL/6J X Mus spretus)F1 females and C57BL/6J males as described (Copeland and Jenkins, 1991). A total of 205 Computer analysis N mice were used to map the Evi5 and Evi5-related loci. 2 The software package of the Genetic Computer Group DNA isolation, restriction enzyme digestion, agarose gel (GCG, Devereux and Smithies, 1984) was used for analysis electrophoresis, Southern blot transfer and hybridization of Evi5 sequences. The open reading frame of Evi5 was were performed essentially as described (Jenkins et al., identi®ed by using the program Frames. Database search + 1982). All blots were prepared with Hybond-N nylon on the level of both nucleotide and protein was performed membrane (Amersham). The probes (described above) were by the Blast program of National Center for Biological 32 labeled with [a P]dCTP by random priming (Stratagene); Information, and GCG program Fasta and Tfasta. The washing was done to a ®nal stringency of 0.16SSCP, 0.1% potential amino acid modi®cations, such as glycosylation SDS, 658C. The P7-U probe detected major fragments of and phosphorylation, of Evi5 protein was identi®ed by 14.0, 6.6, 5.0, 1.4, 1.0, 0.9 and 0.7 kb in PvuII-digested GCG program Motifs. C57BL/6J DNA and 14.0, 6.6, 5.4, 2.4 and 1.5 kb in PvuII- digested Mus. spretus DNA. The presence or absence of the 5.4, 2.4 and 1.5 kb Mus. spretus-speci®c PvuII fragments, Genbank accession number which mapped to three dierent chromosomes, were followed in backcross mice. U53586. A description of the probes and RFLPs for the loci linked to Evi5, Evi5-rs1 and Evi5-rs2 have been described. These loci Construction and in vitro expression of Evi5 vector include ®broblast growth factor 5 (Fgf5), growth factor independence 1 (G®1) and crystallin beta B2 (Crybb2)on AfulllengthEvi5 open reading frame expression vector, chromosome 5 (Benovic et al., 1991; Bell et al., 1995; Evi5p2914, was constructed by subcloning 1-2914 nt of Hulsebos et al., 1995); interferon gamma (Igf), glioma Evi5 cDNA into an eucaryotic expression vector pCR3 associated oncogene (Gli) and avian erythroblastosis onco- (Invitrogen). The Evi5 fragment was generated by a 2-step gene B3 (Erbb3) on chromosome 10 (Justice et al., 1990; PCR which linked Evi5 cDNA plasmids R-6 and S5-1A Copeland et al., 1995); and tumor progression locus 2 (Tpl2) (see Figure 2) (using PCR primer of Evi5 5' UTR: 5'- and cadherin 2 (Cdh2) on chromosome 18 (Justice et al., TCTCCTCCTCTCATCATGG-3' (bps 54 ± 72 of Evi5 1992; Miyatani et al., 1992). Recombination distances were cDNA), and an antisense primer: 5'-TTTCAATGCGC- calculated as described (Green, 1981) using the computer TGTTTTAAAAGCCTA-3' (bps 1405 ± 1431 of Evi5 program SPRETUS MADNESS. Gene order was determined cDNA) with CP7 (using the same 5' UTR primer and an by minimizing the number of recombination events required antisense primer: 5'-AGACATGCCAAGCACAGG-3' (bps to explain the allele distribution patterns. 2897 ± 2914 of Evi5 cDNA). The predicted stop codon for the Evi5 open reading resides at bps 2496 ± 2498. PCR reactions were performed for 30 cycles at the conditions of cDNA libraries and screenings 948C, 1 min; 558C, 1 min and 728C, 3 min using Taq cDNA clones of Evi5 were isolated from two cDNA polymerase (Gibco/BRL). The Evi5 expression vector was libraries. The ®rst one is a Stratagene (La Jolla, CA) pre- sequenced at regions ¯anking PCR primers to ensure made, oligo dT-primed mouse liver cDNA library accuracy. In vitro expression studies were performed by (#935302). The second one is a C57BL/6J mouse liver using the TNT Coupled Reticulocyte Lysate Systems Evi5 homology to cell cycle regulators XLiaoet al 1029 (Promega) with [35S]Methionine and T7 RNA polymerase, Acknowledgements following the manufacturer's protocol. The translation This work was supported in part by the National Cancer products were separated on a 4 ± 15% glycine gel (Bio- Institute, DHHS, under contract with ABL. We also thank Rad) and analysed by autoradiography and Western blot Jerey D Ceci, David A Largaespada and John D analysis. Shaughnessy Jr for helpful comments on the manuscript.

References

Bell DW, Taguchi T, Jenkins NA, Gilbert DJ, Copeland NG, Justice MJ, Siracusa LD, Gilbert DJ, Heisterkamp N, Gilks CB, Zweidler-McKay P, Grimes HL, Tsichlis PN Groen J, Chada K, Silan CM, Copeland NG and and Testa JR. (1995). Cytogenet. Cell. Genet., 70, 263 ± Jenkins NA. (1990). Genetics, 125, 855 ± 866. 267. Kozak M. (1986). Cell, 44, 283 ± 292. Benovic JL, Onorato JJ, Arriza JL, Stone WC, Lohse M, Liao X, Buchberg AM, Jenkins NA and Copeland NG. Jenkins NA, Gilbert DJ, Copeland NG, Caron MG and (1995a). J. Virol., 69, 7132 ± 7137. Lefkowitz RJ. (1991). J. Biol. Chem., 266, 14939 ± 14946. Liao H, Winkfein RJ, Mack G, Rattner JB and Yen TJ. Berns A. (1991). J. Cell. Biochem., 47, 130 ± 135. (1995b). J. Cell. Biol., 130, 507 ± 518. Chang F and Nurse P. (1993). Trends Genet., 9, 333 ± 335. Makris A, Patriotis C, Bear SE and Tsichlis PN. (1993). J. Chou PY and Fasman GD. (1974a) Biochemistry, 13, 211 ± Virol., 67, 4283 ± 4289. 222. Maniatis T, Fritsch EF and Sambrook J. (1982). Molecular Chou PY and Fasman GD. (1974b). Biochemistry, 13, 222 ± cloning: a laboratory manual, Cold Spring Harbor: New 245. York. Church GM and Gilbert W. (1984). Proc. Natl. Acad. Sci. Marchetti A, Buttitta F, Miyazaki S, Gallahan D, Smith GH USA, 81, 1991 ± 1995. and Callahan R. (1995). J. Virol., 69, 1932 ± 1938. Copeland NG, Gilbert DJ, Schindler C, Zhong Z, Wen Z, Miyatani S, Copeland NG, Gilbert DJ, Jenkins NA and Darnell Jr JE, Mui AL-F, Miyajima A, Quelle FW, Ihle Takeichi M. (1992). Proc. Natl. Acad. Sci. USA, 89, 8443 ± JN and Jenkins NA. (1995). Genomics, 29, 225 ± 228. 8447. Copeland NG and Jenkins NA. (1991). Trends Genet., 7, Mucenski ML, Taylor BA, Jenkins NA and Copeland NG. 113 ± 118. (1986). Mol. Cell. Biol., 6, 4236 ± 4243. Devereux J, Haeberli P and Smithies O. (1984). Nucleic Nakamura T, Hillova J, Mariage-Samson R and Hill M. Acids. Res., 12, 387 ± 395. (1988). Oncogene Res., 2, 357 ± 370. Frankhauser C, Marks J, Reymond A and Simanis V. (1993). Nakamura T, Hillova J, Mariage-Samson R, Onno M, EMBO J., 12, 2697 ± 2704. Huebner K, Cannizzaro LA, Boghosian-Sell L, Croce CM Frankhauser C and Simanis V. (1994). EMBO J., 13, 3011 ± and Hill M. (1992). Oncogene, 7, 733 ± 741. 3019. Nazarov V and Wol L. (1995). J. Virol., 69, 3885 ± 3888. Gilbert DJ, Neumann PE, Taylor BA, Jenkins NA and Papa FR and Hochstrasser M. (1993). Nature, 366, 313 ± 319. Copeland NG. (1993). J.Virol., 67, 2083 ± 2090. Patriotis C, Makris A, Cherno J and Tsichlis PN. (1994). Gilks CB, Bear SE, Grimes HL and Tsichlis PN. (1993). Mol. Proc. Natl. Acad. Sci. USA, 91, 9755 ± 9759. Cell. Biol., 13, 1759 ± 1768. Peters G. (1990). Cell Growth Dier., 1, 503 ± 510. Green EL. (1981). Genetics and Probability in Animal Proudfoot NJ and Brownlee GG. (1976). Nature, 263, 211 ± Breeding Experiments, Oxford University Press: New 214. York. Richardson PM and Zon LI. (1995). Oncogene, 11, 1139 ± HabetsGGM,ScholtesEHM,ZuydgeestD,vander 1148. Kammen RA, Stam JC, Berns A and Collard JG. (1994). Robbins J, Blondel BJ, Gallahan D and Callahan R. (1992). Cell, 77, 537 ± 549. J. Virol., 66, 2594 ± 2599. Hoyt MA, Totis L and Roberts BT. (1991). Cell, 66, 507 ± Sanger F, Nicklen S and Coulson AR. (1977). Proc. Natl. 517. Acad. Sci. USA, 74, 5463 ± 5467. Hulsebos TJM, Gilbert DJ, Delattre O, Smink LJ, Dunham Schmidt T, ZoÈ rnig M, Beneke R and MoÈ roÈ y T. (1996). Nucl. I, Westerveld A, Thomas G, Jenkins NA and Copeland Acids. Res., 24, 2528 ± 2534. NG. (1995). Genomics, 29, 712 ± 718. Shaw G and Kamen R. (1986). Cell, 46, 659 ± 667. Ihle JN, Morishita K, Matsugi T and Bartholomew C. van Lohuizen M and Berns A. (1990). Biochim. Biophys. (1990). Prog. Clin. Biol. Res., 352, 329 ± 337. Acta., 1032, 213 ± 235. Introna M, Luchetti M, Castellano M, Arsura M and Golay Wang Y and Burke DJ. (1995). Mol. Cell. Biol., 15, 6838 ± J. (1994). Semin. Cancer Biol., 5, 113 ± 124. 6844. Jenkins NA, Copeland NG, Taylor BA and Lee BK. (1982). Wickens M and Stephenson P. (1984). Science, 226, 1045 ± J. Virol., 43, 26 ± 36. 1051. Jhappan C, Gallahan D, Stahle C, Chu E, Smith GH, Zhu X, Mancini MA, Chang K-H, Liu C-Y, Chen C-F, Shan Merlino G and Callahan R. (1992). Genes Dev., 6, 345 ± B, Jones D, Yang-Feng TL and Lee W-H. (1995). Mol. 355. Cell. Biol., 15, 5017 ± 5029. Justice MJ, Gilbert DJ, Kinzler KW, Vogelstein B, Buchberg ZoÈ rnig M, Schmidt T, Karsunky H, Grzeschizcek A and AM, Ceci JD, Matsuda Y, Chapman VM, Patriotis C, MoÈ roÈ y T. (1996). Oncogene, 12, 1789 ± 1801. Makris A, Tsichlis PN, Jenkins NA and Copeland NG. (1992). Genomics, 13, 1281 ± 1288.