Oncogene (1999) 18, 4032 ± 4038 ã 1999 Stockton Press All rights reserved 0950 ± 9232/99 $12.00 http://www.stockton-press.co.uk/onc

SHORT REPORT Cloning of TACC1, an embryonically expressed, potentially transforming coiled coil containing gene, from the 8p11 breast cancer amplicon

Ivan H Still*,1, Mark Hamilton2, Pauline Vince1, Alan Wolfman2 and John K Cowell1

1Center for Molecular Genetics, The Lerner Research Institute, Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, Ohio, OH 44195, USA; 2Department of Cell Biology, The Lerner Research Institute, Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, Ohio, OH 44195, USA

Ampli®cation of several chromosomal regions have been The involvement of these genes in tumorigenesis has observed in human breast carcinomas. One such region, now been well established, and demonstrates the 8p11, is ampli®ed in 10 ± 15% of tumor samples. importance of characterising genes which are ampli- Although the FGFR1 gene is located close to this region, ®ed in breast cancer. In a search for other regions of and is often included within the amplicon, the observation the genome which are ampli®ed in breast cancer, an that tumors exhibiting 8p11 ampli®cation do not always additional amplicon in 8p11 has been identi®ed in 10 ± overexpress FGFR1 suggests that another gene located 15% of tumors (Adnane et al., 1991, Theillet et al., close to FGFR1 is involved in the tumorigenic process. 1993). The presence of ampli®cation in 8p11 has been We now report the precise location of four expressed associated with estrogen receptor-positive, lobular sequence tags (ESTs) within this region and the cloning carcinomas, which metastasize to axillary lymph nodes of a novel gene, designated TACC1 (transforming acidic (Adnane et al., 1991, Courjal et al., 1997). Although coiled coil gene 1), which encodes an 8 kb transcript and this region includes the gene, which encodes the which is expressed at high levels during early embry- ®broblast growth factor receptor, FGFR1, the exact ogenesis. Constitutive expression of this gene under the involvement of this receptor in the progression of the control of the cytomegalovirus (CMV) promoter in cancer is unclear because, in some tumors, it is not mouse ®broblasts, results in cellular transformation and included in the ampli®ed region (Dib et al., 1995). In anchorage independent growth, suggesting that inap- addition, the expression level of FGFR1 mRNA in propriate expression can impart a proliferative advan- most tumors is at comparable levels to normal breast tage. This observation raises the possibility that tissue (Theillet et al., 1993). Furthermore, in one breast ampli®cation of TACC1 could promote malignant tumor cell line, MDA-MB-134, which overexpresses growth, thereby making TACC1 an attractive candidate FGFR1, treatment with FGF1 or FGF2 inhibits cell for the gene promoting tumorigenicity as a result of the growth, suggesting that, in this cell line, overexpression 8p11 ampli®cation in human breast cancers. of FGFR1 could be detrimental to increased tumor growth (McLeskey et al., 1994). In fact, detailed Keywords: TACC1; embryonic; transforming; breast mapping of the 8p11 amplicon suggests the potential cancer existence of two amplicons, amplicon 1, located close to FGFR1 and a second amplicon centred around PLAT (Adelaide et al., 1998), although concomitant The accumulation of genetic changes, such as ampli®cation of these two regions is also noted (Dib et structural chromosomal abnormalities, is associated al., 1995, Adelaide et al., 1998). However, in those with the increase in the malignant potential of cancer tumors carrying amplicon 1 only, the minimal region cells. Very often this increased malignancy is accom- implicated has been re®ned to a 600 kb region panied by ampli®cation of large regions of genomic extending from the FGFR1 locus to D8S1416E (Dib DNA encompassing several genes. Once ampli®cation et al., 1995, Adelaide et al., 1998). Thus, the mapping has occurred, it is believed that the retention of the and characterisation of genes within this 600 kb region ampli®ed region is selected for by the expression of a will be critical to our understanding of the mechanism critical `driver' gene, which further promotes tumor by which 8p11 ampli®cation a€ects tumor growth and growth and/or metastasis. Ampli®cations, which are metastasis in human breast cancer. We now report the frequently observed in breast carcinomas, involve identi®cation and characterization of a novel, embry- regions (amplicons) in 8q24, 11q13 or 17q12, each onically expressed gene from the 8p11 amplicon 1, observed in 20% of all breast tumors. The genes which, upon constitutive expression in mouse fibro- located within these particular amplicons have been blasts, is capable of cellular transformation. identi®ed and encode the known proto-oncogenes, Prior to the latest publication of the human gene MYC (Escot et al., 1986), CCND1 (Theillet et al., expression maps, FGFR1 was the only gene assigned to 1990) and ERBB2 (Slamon et al., 1987), respectively. the 8p11 breast cancer amplicon. However, with the new gene mapping data, a number of additional expressed gene markers (ESTs) have been broadly assigned to the pericentromeric region of the short arm *Correspondence: IH Still Received 28 September 1998; revised 2 March 1999; accepted 2 of chromosome 8 (Schuler et al., 1996). Using primers March 1999 sequences obtained from dbEST, we assigned ®ve of Cloning a novel transforming gene, TACC1 IH Still et al 4033

Figure 1 Physical map of the 380 kb region surrounding the TACC1 gene. 8p11 amplicon 1 is de®ned by the FGFR1 gene (telomeric marker) and EST D8S1416E (centromeric marker). The YACs used to map the four EST markers D8S1943, STSG33341, D8S1416E and D14665 (corresponding to the MDC-9 gene) are shown at the top of the ®gure. These markers map to BACs 311J16, 341J6 and 286P2/280P14 respectively. The presence of three CpG islands centred on NotI restriction sites (N1-3) within BAC 311J16 and 341J6 suggests the presence of at least three genes within this region. Two of these CpG islands are located close to the EST marker D8S1943 in BAC311J16. The centromeric CpG island within this BAC, however, is associated with the TACC1 gene which spans approximately 125 kb of this critical region. MDC-9 gene is located centromeric to amplicon 1, thus excluding the MDC-9 gene as a candidate driver gene

Figure 2 Northern blot analysis of TACC1. (a) cDNA clone 295786 hybridized to 2 mg of polyadenylated mRNA from adult tissues (Clontech). (b) cDNA clone 309581 hybridized to polyadenylated mRNA from di€erent murine developmental stages (Clontech). This clone detects a transcript expressed highest at E7, prior to down regulation between E7 and E11. The developmental multiple tissue northern (MTN) was exposed for 4 h. The adult MTN was exposed for 2 days with intensi®cation. Longer exposure of the developmental MTN revealed transcript levels comparable to those seen in adult tissues Cloning a novel transforming gene, TACC1 IH Still et al 4034 these ESTs, D8S1943, D8S1461E, WI-17448, independent, overlapping clones were identi®ed and STSG33341, and D14665, corresponding to the sequenced, which allowed a 7760 bp cDNA contig to MDC-9 gene (Weskamp et al., 1996) to the 1.2 Mb be assembled (Figure 3; GenBank accession: YAC 959A4, which is known to contain the critical AF049910). This sequence also joined ®ve UNIGENE region of amplicon 1 (Dib et al., 1995). Based on their presence in YAC 959A4, but their absence in YACs 676A7 and 782H11, four of these ESTs (D8S1943, D8S1461E, STSG33341 and D14665) sublocalized to the 600 kb region, which encompasses the 8p11 amplicon 1 (Figure 1). As a ®rst stage in the construction of a contig spanning this region, these four EST markers were then used to screen a BAC library (Research Genetics). To con®rm overlap between these clones, the end-sequences of each BAC were obtained either by using the vectorette PCR method described by Hawthorn et al. (1995), or by direct sequencing. To ®ll the gap between 311J16 and 341J6, primers were designed from the end-sequences of both of these BACs and used to screen the BAC library. This resulted in the isolation of the 70 kb BAC 364I12, which overlapped both BAC clones. The ®nal BAC contig spans approximately 380 kb of the critical region of amplicon 1, and is summarized in Figure 1. A 33 kb fragment between NotI site N2 and 6SP from 311J16 (Figure 1) was also subcloned and the ends of this fragment (311N33) were sequenced. The sequence from the two ends of 311N33 matched two di€erent UNIGENE cDNA clusters, HS30256 and HS125093, which also included IMAGE cDNA 80982, the source of EST marker STSG33341 (Figure 1), which is located in BAC 341J6. This demonstrated that a gene, which we have named TACC1 ± for transforming acid coiled coil gene 1 (see below) ± spans a region of approximately 125 kb, which extends from NotI Site N2 to BAC341J6, and contains both the endclone 6SP and EST STSG33341. In order to determine the expression pro®le and transcript size of TACC1, two cDNA clones, 295786 and 309581, were selected from the corresponding human UNIGENE clusters (HS30256 and HS125093). Hybridization of IMAGE clone 295786, which contains 1.3 kb of the 3' untranslated region, to multiple tissues Northern blots (Clontech) identi®ed a single major transcript of 8 kb (Figure 2a). In adult heart, di€erential splicing may account for the di€erence of approximately 300 bp in the transcript size (Figure 2). The 8 kb transcript was detected in all tissues tested, although comparatively lower levels of expression were observed in adult liver and lung. Sequence analysis of IMAGE cDNA clone 309581 indicated that this clone contained part of the TACC1 open reading frame. When this clone was hybridized to a murine developmental Northern blot (Clontech), a single transcript of 8 kb was also detected, which showed highest expression at embryonic day 7 (Figure 2b). TACC1 transcript levels were then dramatically down- regulated by day 11, when expression levels are comparable to those seen in adult tissues (Figure 2b). Figure 3 The nucleotide sequence and deduced amino acid This Northern analysis also indicated that the sequence of TACC1. The ®rst 2880 nucleotides encompassing the approximately 2 kb of sequence provided by the 5'UTR and the open reading frame are shown. The complete UNIGENE clusters was incomplete. To obtain the cDNA sequence extends a further 4878 nucleotides and has been complete human TACC1 cDNA sequence, IMAGE deposited in the GenBank database with accession number AF049910. The 805 amino acid open reading frame contains clones 309581 and 295786 were hybridized to both a two bipartite nuclear localization signals (underlined) suggesting fetal liver/spleen cDNA library (Clontech Cat. No. that TACC1 is a nuclear protein. The predicted coiled coil HL5003a) and a fetal lung library (Stratagene). Six domain is highlighted Cloning a novel transforming gene, TACC1 IH Still et al 4035 clusters, HS30256, HS62440, HS49669, HS57124 and entire protein sequence. Indeed, two typical bipartite HS125093. nuclear localization signals (Robbins et al., 1991) are Examination of the TACC1 cDNA sequence reveals located at residues 226 ± 241 and 455 ± 471, suggesting a 320 nt 5' untranslated region (UTR) preceding a that the TACC1 protein may be localized to the 2415 nt open reading frame (Figure 4a). The nucleotide nucleus (Figure 4b). However, if TACC1 is localized to sequence surrounding the proposed initiator codon the nucleus, the absence of a de®ned DNA or RNA (CTCATGG) is in good agreement with the Kozak binding domain precludes direct interaction with sequence for protein synthesis initiation. The TACC1 nucleic acid. Secondary structure analysis of the cDNA has an extensive 3' untranslated region of TACC1 protein indicated that the 200 amino acid 4997 nt, containing a single Alu repeat between nt5195 carboxy-terminus can form extensive a-helical seg- and nt5579. Further sequence analysis of cDNA clone ments. This a-helical region is predicted to adopt a 80982 indicated that this clone represented a partially coiled-coil tertiary structure, similar to the rod like processed transcript, and that STSG33341 actually corresponds to sequence from the last intron of TACC1. The 3'UTR also contains the D8S1461E EST, which de®nes the centromeric boundary of the 8p11 breast cancer amplicon (Figure 4a). A typical polyadenylation signal is located 17 nucleotides up- stream of the polyadenylation site. The TACC1 open reading frame encodes a protein of 805 amino acids with a predicted molecular mass of 87.8 kDa (Figure 4b). The TACC1 protein is rich in serine, proline and acidic residues, the frequency of which in the ®rst 600 residues of the protein rises to 15, 8 and 17% respectively. The N-terminal 20 amino acids are particularly rich in tryptophan (25%). This domain is followed by two potential attachment sites for Figure 5 Comparison of the coiled coil domain of TACC1 and myristylic acid, which may indicate a potential the translated cDNAs corresponding to TACC2 and TACC3. Sequence identity is denoted by a dash ( ± ). The potential tyrosine association with the plasma membrane. No transmem- phosphorylation site found in human TACC1-3 and murine tacc1 brane domains or signal sequences are evident in the (Still et al., unpublished) is underlined

Figure 4 Schematic representation of the TACC1 cDNA and protein. (a) The 7760 nucleotide TACC1 cDNA sequence joins ®ve UNIGENE clusters and contains sequence from around NotI site N2, 6SP, and the EST marker D8S1461E. The sequence corresponding to STSG33341 is located within the last intron of the TACC1 gene, and is, therefore, not included in this cDNA contig (see text). (b) The 805 amino acid open reading frame consists of an N-terminal region enriched in serine, proline and acidic residues, and a 200 amino acid C-terminal coiled coil domain. NLS1 and NLS2 represent the potential bipartite nuclear localization signals Cloning a novel transforming gene, TACC1 IH Still et al 4036

Figure 6 Morphology of TACC1 transfected C3H10T1/2 mouse ®broblasts. 80% con¯uent normal mouse C3H10T1/2 ®broblasts were transfected with 100 ng TACC1 plasmid using the lipofectin-mediated DNA transfection method (Gibco-BRL) and the cells were selected with 500 mg/ml G418. Isolated G418-resistant colonies were subcloned and propagated as individual cell clones. TACC1 expression was determined by rt-PCR, according to routine protocols. The three TACC1 transfected cell lines, 10R82, 10R83 and 10R85 show signi®cant alterations in cellular morphology, as compared to normal parental C3H10T1/2 cells

domains found in proteins such as myosin. BLAST proteins. In this regard, this domain could be involved homology searches using this region revealed highest in a proposed role of TACC proteins in nuclear similarity, but low identity, with coiled coil domains of sca€olding and/or assembly of the mitotic spindle, yeast tropomyosin and other nuclear proteins involved both of which are necessary for cell division. in mitotic spindle assembly. Located within this Interestingly, the TACC2 cDNA has been mapped as domain, at amino acid 730, is a potential tyrosine part of the genome mapping project (Schuler et al., phosphorylation site, which may serve to regulate the 1996) to chromosome region 10q26, within the same conformation of the coiled coil domain. This site is interval as the FGFR2 gene, which is the site of another also conserved in the murine orthologue of TACC1 chromosome region ampli®ed in breast cancer (Adnane (Still et al., unpublished). The presence of nuclear et al., 1991, Theillet et al., 1993). localization signals and this coiled coil domain, In vitro, cell transformation can often be character- therefore, suggests that TACC1 could play a role in ized by changes in cell morphology. To investigate the nuclear sca€olding and/or assembly of the mitotic consequence of increased expression levels of TACC1, spindle. we introduced a plasmid construct containing the In order to determine whether TACC1 was part of a TACC1 ORF (nt143 ± nt2940) into murine C3H10T1/ gene family, the TBLASTN algorithm was used to 2 ®broblasts under the constitutively active Cytomega- search the dbEST database for human cDNA lovirus (CMV) promoter. The morphology of three sequences related to TACC1. The coiled coil domain TACC1 transfected cell clones, 10R82, 10R83 and identi®ed a series of partial cDNAs which could be 10R85 is shown in Figure 6. These transfected cells grouped into two distinct UNIGENE cDNA clusters, were more refractile and elongated than parental HS90415 and HS104019. This suggested the existence C3H10T1/2 cells, and failed to form complete cell of a family of TACC genes comprising at least three monolayers (Figure 6). Thus, all stable cell lines, which family members (Figure 5). We have named these genes by rt-PCR analysis expressed the introduced TACC1 TACC2 and TACC3 respectively. The TACC2 coiled transcript (data not shown), displayed morphology coil is highly related to TACC1, whilst the correspond- changes consistent with a transformed phenotype ing domain of TACC3 is more divergent. This suggests (Figure 6). that the TACC gene family evolved from a common Anchorage independent growth is a well-established ancestor and has undergone two successive duplica- characteristic of cell transformation mediated by many tions, with the TACC1/TACC2 gene duplication oncogenes, including Ras (Weyman and Stacey, 1996), arising later in evolutionary time. The conservation of Src (Scwartz, 1997) and receptor tyrosine kinases this domain between these family members also (reviewed in Schlessinger and Ullrich, 1992). Since indicates that this region may be critical to the TACC1-transfected cells exhibited a transformed function of the TACC family members, possibly cellular morphology, their oncogenic potential was through interaction between the coiled coil and other assessed by their ability to proliferate in an Cloning a novel transforming gene, TACC1 IH Still et al 4037

Figure 7 Anchorage-independent growth of TACC1 transfected C3H10T1/2 mouse ®broblasts. 105 cells of the stably transfected TACC1 cell lines, 10R82, 10R83 and 10R85, were seeded in 0.3% agarose and allowed to proliferate in 100 mm dishes containing a bottom layer of 0.5% agarose. After 12 days, colonies were photographed (magni®cation6200). 11A(Ha-rasV12) cells derived from C3H10T1/2 ®broblasts by transfection with an activated (G12V)Ha-ras construct and were a kind gift from EJ Taporowsky, Purdue University (IN, USA) and represent a positive control for anchorage independent growth anchorage-independent manner. The morphologically alterations in cellular morphology and cellular transformed 10R82, 10R83 and 10R85 cells were each growth. This suggests that deregulation of TACC1 able to proliferate in soft agar (Figure 7). Multi- gene expression directly through ampli®cation of the cellular colonies were readily observed with these three entire gene, or through disruption of transcriptional TACC1 cell lines. Anchorage-independent colonies of regulatory elements could contribute to the aggressive 10R85 and 10R83 cells were similar in size to those phenotype noted for breast tumors harbouring obtained using Ha-ras transformed 11A ®broblasts. ampli®cation of 8p11. Interestingly, the second TACC Parental C3H10T1/2 cells failed to form colonies in family member, TACC2 is located close to FGFR2,ina these soft agar assays (Figure 7) although, after 14 region of chromosome 10q26 which is also ampli®ed in days, predominantly single cells, and in some cases a small percentage of breast tumors (Adnane et al., pairs of post-mitotic cells, were observed. This data, 1991, Courjal et al., 1997). TACC1 and additional therefore, suggests that deregulation of TACC1 TACC family members are, therefore, attractive expression leading to overexpression of the TACC1 candidates as potential progression factors in breast protein has the potential to induce cellular transfor- cancer. However, the exact role of these genes in mation. human breast cancer progression will require a more In summary, we have generated a BAC contig, detailed analysis of the expression of these genes in spanning 380 kb of the 8p11 amplicon 1 associated breast tumors, and the identi®cation of their normal with breast cancer. Within this region, we have cellular functions. identi®ed a gene of unknown function, TACC1, which is expressed at high levels during early embryogenesis. Acknowledgements In vitro analysis of this gene using standard cellular The ®rst two authors contributed equally to this report. transformation, and soft agar assays, suggests that This work is in part supported by NIH grant GM49652 aberrant expression of this gene could promote and AHA grant 96001110 to Alan Wolfman.

References

Adelaide J, Cha€enet M, Imbert A, Allione F, Geneix J, Courjal F, Cuny M, Simony-Lafontaine J, Louason G, Popovici C, van Alewick D, Trapman J, Zeillinger R, Speiser P, Zeillinger R, Rodriguez C and Theillet C. Borresen-Dale A-L, Lidereau R, Birnbaum D and (1997). Cancer Res., 57, 4360 ± 4367. Pebusque M-J. (1998). Genes Chrom. Cancer, 22, 186 ± Dib A, Adelaide J, Cha€anet M, Imbert A, Le Paslier D, 199. Jacquemier J, Gaudray P, Theillet C, Birnbaum D and Adnane J, Gaudray P, Dionne CA, Crumley G, Jaye M, Pebusque MJ. (1995). Oncogene, 10, 995 ± 1001. Schlessinger J, Jeanteur P, Birnbaun D and Theillet C. (1991). Oncogene, 6, 659 ± 663. Cloning a novel transforming gene, TACC1 IH Still et al 4038 Escot C, Theillet C, Lidereau R, Spyratos F, Champeme Schwartz MA. (1997). J. Cell Biol., 139, 575 ± 578. MH, Gest J and Callahan R. (1986). Proc. Natl. Acad. Sci. Slamon D, Clark G, Wong S, Levin W, Ullrich A and USA, 83, 4834 ± 4838. McGuire W. (1987). Science, 235, 177 ± 182. Hawthorn LA, Roberts T, Verlind E, Kooy RF and Cowell Theillet C, Adnane J, Szepetowski P, Simon MP, Jeanteur P, JK. (1995). Genomics, 30, 425 ± 430. Birnbaum D and Gaudray P. (1990). Oncogene, 5, 147 ± McLeskeySW,DingIYF,LippmanMEandKernFG. 149. (1994). Cancer Res., 54, 523 ± 530. Theillet C, Adelaide J, Louason G, Bonnet-Dorion F, Robbins J, Dilworth SM, Laskey RA and Dingwall C. Jacquemier J, Adnane J, Longy M, Katsaros D, Sismondi (1991). Cell, 64, 615 ± 623. P, Gaudray P and Birnbaum D. (1993). Genes Chrom. Sclessinger J and Ullrich A. (1992). Neuron, 9, 383 ± 391. Cancer, 7, 219 ± 226. Schuler GD, Boguski MS, Stewart EA, Stein LD, Gyapay G, Weskamp G, Kratzschmar J, Reid MS and Blobel CP. Rice K, White RE, Rodriguez-Tome P, Agarwal A, (1996). J. Cell Biol., 132, 717 ± 726. Bajorek E, Bentolila S, Birren BB, Butler A, Castle AB, Weyman CM and Stacey DW. (1996). Adv. Oncobiol., 1, Chiannikulchai N, Chu A, Clee C, Cowles S, Day PJ and 159 ± 177. Dibling T, et al. (1996). Science, 274, 540 ± 546.