TACC1 Is Down-Regulated in Human Cancers and Associates with Mrna Regulators

Carcinogenesis and translational controls: TACC1 is down-regulated in human cancers and associates with mRNA regulators

Nathalie Conte1, Emmanuelle Charafe-Jauﬀret2,Be´ ne´ dicte Delaval1, Jose´ Ade´ laı¨ de3, Christophe Ginestier2, Jeannine Geneix2, Daniel Isnardon4, Jocelyne Jacquemier2 and Daniel Birnbaum*,1,2,3

1De´partement d’Oncologie Molećulaire, U119 Inserm, 27 Bd. Leı¨ Roure, 13009, Marseille, France; 2Laboratoire de Pathologie Molećulaire, De´partement d’Oncologie Molećulaire, Institut Paoli-Calmettes, Marseille, France; 3Laboratoire de Biologie des Tumeurs, De´partement d’Oncologie Molećulaire, Institut Paoli-Calmettes, Marseille, France; 4Atelier de Bio-Imagerie, Institut Paoli-Calmettes, Marseille, France

The three human TACC genes encode a family of The short arm of chromosome 8 is a site of frequent proteins that are suspected to play a role in carcinogen- genetic alterations in several types of human cancers, esis. Their function is not precisely known; a Xenopus including breast and other gynecologic carcinomas. TACC protein called Maskin is involved in translational These alterations include amplification of oncogenes control, while the Drosophila D-TACC associates with such as FGFR1 in around 15% of cases (Ugolini et al., microtubules and centrosomes. We have characterized 1999), deletions of tumor suppressor genes (TSG)of the human TACC1 gene and its products. The TACC1 unknown identity (Ade´ laı¨ de et al., 1998; Ugolini et al., gene is located in region p12 of chromosome 8; its 1999; Pribill et al., 2001), and in one case translocation mRNA is ubiquitously expressed and encodes a protein resulting in the fusion of the NRG1 gene on 8p to the with an apparent molecular mass of 125 kDa, which is DOC4 gene from 11q13 (Wang et al., 1999). We cytoplasmic and mainly perinuclear. We show that previously conducted a series of expression studies using TACC1 mRNA gene expression is down-regulated in various cDNA arrays. Based on the differential expres- various types of tumors. Using immunohistochemistry of sion assay of chromosome 8p11 – 21 genes, we identified tumor tissue-microarrays and sections, we confirm that SFRP1 and TACC1 genes, whose mRNA are reduced or the level of TACC1 protein is down-regulated in breast absent in breast carcinomas (Ugolini et al., 1999, 2001). cancer. Finally, using the two-hybrid screen in yeast, TACC1 (Still et al., 1999a) belongs to a family of GST pull-downs and co-immunoprecipitations, we identi- three paralogous genes (Popovici et al., 2001) that also fied two potential binding partners for TACC1, LSM7 includes TACC2/AZU1/ECTACC (Chen et al., 2000; and SmG. They constitute a conserved subfamily of Sm- Pu et al., 2001) and TACC3 (Still et al., 1999b), and like small proteins that associate with U6 snRNPs and has an ortholog in fly, tacc, located on chromosome play a role in several aspects of mRNA processing. We 3R of the Drosophila genome (Gergely et al., 2000b). speculate that down-regulation of TACC1 may alter the TACC1 was first identified as a potential oncogene; it control of mRNA homeostasis in polarized cells and is sometimes included in the amplification of the 8p12 participates in the oncogenic processes. region in breast cancers and can transform fibroblasts Oncogene (2002) 21, 5619 – 5630. doi:10.1038/sj.onc. (Still et al., 1999a). TACC2 was identified by 1205658 differential display strategies used to compare different stages of transformation of mammary epithelial cells Keywords: microtubules; mitosis; breast cancer; mRNA (Chen et al., 2000) or to isolate upregulated genes from processing; Sm proteins; TACC proteins erythropoietin-stimulated endothelial cells (Pu et al., 2001). In the former study, TACC2 appears as a suppressor of growth. TACC3 was initially identified in Introduction the databases (Still et al., 1999b). From these early analyses, it is difficult to conclude on the exact role of Genes and mechanisms that participate in mammary TACC genes in carcinogenesis. carcinogenesis are numerous and not well identified or Clues about the function of the TACC proteins in characterized. Several genes from the chromosomal physiological conditions have come from different region 8p11 – 21 seem to play a role in breast cancer. studies. The tacc gene in Drosophila encodes D-TACC, a centrosomal protein required for normal spindle function in the early embryo (Gergely et al., 2000b); it interacts with the microtubule-associated msps/ *Correspondence: D Birnbaum; E-mail: [email protected] XMAP215/ch-TOG protein to regulate microtubule Received 11 March 2002; revised 19 April 2002; accepted 10 May behavior (Gergely et al., 2000a,b; Cullen and Ohkura, 2002 2001; Lee et al., 2001; Theurkauf, 2001). In Xenopus,a TACC1 in breast cancer N Conte et al 5620 TACC protein called Maskin associates with the a series of human mammary carcinoma cell lines for 3’UTR-CPE binding protein to regulate eIF-4E control the presence of TACC1 mRNA. Expression of the of translation and oocyte maturation (Stebbins-Boaz et 7.7 kb transcript was found at various levels in the al., 1999); CPEB and Maskin are present on the different cell lines (Figure 2b). An additional transcript mitotic apparatus of animal pole blastomeres in of 5 kb was also detected in some cell lines. It may embryos (Groisman et al., 2000). Thus, TACC proteins correspond to TACC1s; the presence of the TACC1s are microtubule-associated proteins that could regulate spliced form in some cell lines was confirmed by RT – RNAs specifically located near spindles and centro- PCR assays of mRNA from mammary gland HME-1 somes. In the mouse, TACC3 associates with ARNT, cells (not shown) using specific oligonucleotide primers an arylhydrocarbon receptor-nuclear translocator (see Figure 1b). protein (Sadek et al., 2000). ARNT is a transcriptional Expression of the TACC1 proteins was studied in regulator of the bHLH-PAS family that dimerizes with mammary carcinoma cell lines by Western blot analysis a number of partners, thus playing a central role in (Figure 2c). Anti-TACC1-N and anti-SPAZ1 antibo- development and in several cellular processes such as dies were obtained by immunization of rabbits with detoxification and hypoxia-induced responses. peptides from N-terminus or SPAZ1 motif, respectively While it is too early to propose a single, integrative (see Materials and methods). A commercial anti- view of the various studies on TACC proteins, it may TACC1 was also used. Proteins of 125 and 130 kDa nevertheless be thought that alterations of TACC genes were detected by the three antibodies in most cell lines, may disturb important regulations and participate to except in HeLa where only the 130 kDa band was seen carcinogenesis. In this study, we have investigated (Figure 2c). The anti-SPAZ1 and commercial anti- TACC1 expression in breast cancer and gathered clues bodies detected only the 130 and 125 kDa bands, about its function. We have shown that breast cancers whereas the anti-TACC1-N detected an additional frequently express low levels of TACC1 protein, we band of 85 kDa. This band was detected in cell lysates have characterized TACC1 localization in human cells obtained with RIPA, a strong detergent which extracts and identified binding partners of TACC1 that are proteins from the nucleus, but weakly or not when lysis highly suggestive of its function. was done with NP40, a milder detergent. As shown by the reactivity with three different antibodies, the 125 kDa band corresponds to the TACC1l protein; Results the 130 kDa is likely to correspond to a post- translationally modified form, possibly a phosphory- Characterization of the TACC1 locus and mRNA lated form of TACC1l (see Figure 6d). The 85 kDa The exon – intron organization of the human TACC1 band was observed by other authors (Gergely et al., locus was assembled by searching for TACC1 genomic 2000b). We detected it only with the anti-TACC-N sequences in databases (http://ensembl.ebi.ac.uk/) and antibody which is the only antibody to be directed is shown in Figure 1a. The TACC1 gene contains 13 against a peptide outside the SPAZ region. The p85 putative exons of various length spanning ca. 50 kb of TACC1 protein thus does not contain the SPAZ region genomic DNA. A full-length cDNA of 7758 bp and could be coded by TACC1s, or by another, encoding human TACC1 was found in databases unidentified type of spliced isoform. As already noted (GenBank: AF049910) (Figure 1b). The deduced by others, the apparent molecular mass of the TACC amino-acid sequence of the TACC1 protein (TACC1l, proteins, including the fly protein (Gergely et al., see below) contains an open-reading frame of 805 2000b) is much higher than the predicted one. It could amino-acid residues. TACC1l contains two types of be due to conformational structures created by the motifs, a coiled-coil motif at the C-terminus, and two SPAZ and coiled-coil motifs. This would be true also so-called SPAZ (serine and proline rich AZU-1) motifs for TACC1s. (Chen et al., 2000) at the N-terminus (Figure 1c). Two potential nuclear localization signals are also present. Subcellular localization of TACC1 proteins TACC1 coding sequence was also subjected to BLASTN searches with dbEST database (http:// To further characterize TACC1, we determined its www.ncbi.nlm.nih.gov/BLAST/). Most of the clones subcellular localization. Affinity-purified TACC1-N retrieved were located in the 3’ non coding region of antibody was used in indirect immunofluorescence the gene. Four cDNA clones were located in the 5’ studies on transfected and non-transfected HeLa cells. coding region of TACC1 cDNA (qh26c10, ql62e12, Preimmune rabbit serum showed no specific reactivity zw32e05, zb77b03) and represented a spliced form of in these studies (data not shown). In all the cells the TACC1 transcript (we named TACC1s), which examined, TACC1 reactivity was observed within the differs from the full length cDNA by the lack of exons cell cytoplasm; it was especially concentrated in the 2 and 3 that codes for the SPAZ region of the TACC1l perinuclear area (Figure 3). Endogenous TACC1 protein (Figure 1b,c). The putative TACC1s protein (Figure 3a – c, c’) was also found in the nucleus, in contains 418 amino-acid residues. large subnuclear dots. These dots may correspond to Hybridization of a TACC1 probe to human adult nucleolar sites as seen with differential interference multiple-tissue Northern blots identified a ubiquitously contrast image (data not shown). In HeLa cells expressed transcript of 7.7 kb (Figure 2a). We screened transfected by either TACC1l (Figure 3d – f) or

Oncogene TACC1 in breast cancer N Conte et al 5621

Figure 1 The TACC1 locus and putative TACC1 proteins. (a) Exon – intron organization of the human TACC1 gene. (b) TACC1 cDNAs: the long (TACC1l) and short (TACC1s) isoforms are depicted, with the corresponding sequences found in the dbEST database. Arrows indicate the position of the oligonucleotide primers used to differentiate between the two forms in RT – PCR experiments. (c) The resulting putative TACC1 proteins are shown with their motifs (NLS: nuclear localization signals). The SPAZ (serine and proline-rich AZU-1) motif has been deﬁned in TACC2/AZU-1 by Chen et al. (2000). Numbers refer to amino acid positions

TACC1s (Figure 3g – i, i’), the pattern of TACC1 To further explore the status of TACC1 in human reactivity in the perinuclear region was similar to that cancer, we studied the expression of TACC1 mRNA in of non transfected cells but no protein was detected in a series of 61 consecutive sporadic breast carcinomas the nucleus. by Northern blot (Figure 4b). Of these, 39 showed an altered TACC1 expression as compared to normal breast: TACC1 was significantly underexpressed (more Alteration of TACC1 mRNA expression profiles in than a log) in 18 tumors and undetectable in 21 tumors human cancers (Figure 4c). We explored the status of TACC1 mRNA in human cancers. We first hybridized a commercial ‘cancer Alteration of TACC1 protein expression in breast profiling array’ with a TACC1 probe. A total of 241 carcinomas couples of various tumors and corresponding normal tissues were surveyed. As seen in Figure 4a, variations in We next studied the expression of the TACC1 protein mRNA signals were observed in tumors of breast, in a series of 57 breast tumor and 10 normal breast uterus, colon, ovary, lung and rectum but much less in samples deposited in a tissue microarray (Richter et al., stomach, kidney and thyroid cancers. Completely 2000; Kallioniemi et al., 2001). To allow internal different variations were observed with a different, non comparison, we selected a series that contained TACC1 probe (not shown). After quantification, in 53 relatively similar tumors of poor prognosis from breast cancers, TACC1 expression was found higher women treated with anthracyclin-based adjuvant (more than 1.5 times) in the normal sample as compared chemotherapy. Figure 5a shows an example of the to the tumoral counterpart in 42 cases (79.2%), results. The expression of TACC1 was observed in all including the three cases with metastases; it was higher normal breast tissue samples. It was found in 38 out of in the tumoral sample in seven cases (13.2%), including 57 tumors (69%) and absent in 19. the single case of medullary carcinoma; in four cases To determine more precisely the level and distribu- there was no difference. The same increase in the normal tion of the TACC1 protein in breast tumor cells, vs tumoral ratios were observed for the 21 lung (85.7%), samples from normal breast tissue and 48 sporadic 15 ovary (93.3%) and 44 uterus (75%) cases. primary breast carcinomas were subjected to immu-

Oncogene TACC1 in breast cancer N Conte et al 5622

Figure 2 Expression of the TACC1 gene and TACC1 protein in human tissues. Northern blot analysis of TACC1 gene expression in various normal human tissues (a) and cancer cell lines (b). TACC1 mRNA is ubiquitously expressed as a major transcript of 7.7 kb. (c) Western blot analysis of TACC1 proteins: proteins from lysates of Cos-1, HME1, 293T, and HeLa cells were prepared in various conditions, separated on polyacrylamide gel, transferred onto nitrocellulose, and detected with three different anti- TACC1 antibodies (anti-SPAZ1, anti-TACC1-N, and commercial anti-TACC1). TACC1 proteins of 125 and 130 kDa (arrowheads) were detected with the three antibodies. They represent TACC1l; TACC1l*, detected in Cos-1, 293T and HME1 cells but not in HeLa cells, is probably a phosphorylated form of TACC1l. The anti-TACC1-N detected an additional band of 85 kDa (TACC1i, insoluble)

Identification of TACC1-interacting partners and nohistochemical analysis of whole sections, using the phosphorylation anti-TACC1-N antibody (Figure 5b,d). The epithelial layer of the normal ducts showed immunoreactivity to In an effort to define the molecular complex anti-TACC1 with a cytoplasmic predominance as seen associated with TACC1, we used the two-hybrid in the normal sample (Figure 5b). Weak staining was system in yeast to isolate potential TACC1 partners. found in the stromal cells. No staining was observed We screened a human mammary gland cDNA in the myoepithelial cells (Figure 5b). Thirty-one out expression library with the 805 amino-acid residues of 46 measurable tumors were positively stained of human TACC1l fused to the Gal4-binding domain. (Figure 5c), and 15 were negative (Figure 5d). Various We isolated 10 cDNA clones. Eight encoded the 103 staining patterns were observed in the tumoral tissues, amino-acids of the LSM7 protein and two encoded either cytoplasmic only, or both cytoplasmic and the 76 amino-acids of the snRNP Sm protein G. The nuclear. In the cytoplasm, TACC1 immunoreactivity LSM7 gene maps on chromosome arm 19p and SmG was mostly found around the nucleus, in agreement (Herman et al., 1995) is encoded by the SNRPG gene with the pattern observed in immunofluorescence (OMIM603542) on chromosome arm 1q. Both LSM7 (Figure 5c). These experiments show that the TACC1 and SmG are Sm-like proteins. They are related to the protein is down-regulated in around one-third of yeast Sm small nuclear ribonucleoproteins (snRNPs). breast tumors. No correlation between TACC1 The Sm/LSM proteins associate with small nuclear immunoreactivity and any of the histoclinical para- RNA to form the core of snRNPs required in a large meters was observed. variety of RNA maturation processes, most notably in

Oncogene TACC1 in breast cancer N Conte et al 5623

Figure 3 Subcellular localization of the TACC1 protein. Subcellular localization of TACC1 by indirect immunofluorescence studies of non transfected fixed HeLa cells (a – c, c’) and on cells transfected with TACC1l-HA (d–f) or with TACC1s-myc (g–i,i’). Non transfected HeLa cells were stained with anti-TACC1-N antibody (a, in red) and transfected cells with anti-HA antibody (d, in red) or with anti-myc antibody (g, in red). All the cells were counterstained with SYTO-16 (b, e, h, in green). Merged images are shown in c, f and i. Three-dimensional views of staining of TACC1 and transfected TACC1s are respectively shown in c’ and i’. These views are available at http://u119.marseille.inserm.fr/Db/image-tacc.html pre-mRNA splicing and mRNA degradation, and in Finally we wanted to determine if TACC1 proteins telomere formation (He and Parker, 2000; Tharun et were phosphorylated and performed an in vitro kinase al., 2000). assay (Figure 6d). Immunoprecipitation of TACC1 To further document the interaction of TACC1 proteins showed that TACC1l was heavily phosphory- with these potential partners, we produced GST lated whereas TACC1s was not. fusion proteins that were used in pull-down experiments (Figure 6a). Both GST-LSM7 and GST-SmG fusion proteins were able to precipitate TACC1l Discussion (Figure 6a) and TACC1s (data not shown) proteins. Next, proteins from Cos-1 cells cotransfected by HA- The TACC are regulators of cell functions which are tagged TACC1l with myc-tagged LSM7 or myc- beginning to be unravelled. Their activities have been tagged SmG were immunoprecipitated with either investigated from at least three aspects, their role in cell anti-myc or anti-TACC1-C antibody and Western transformation, their function in RNA processing, blotted. In both cases, blotting with the anti-HA evidenced in Xenopus, and their role as centrosomal antibody revealed TACC1l protein and blotting with proteins, particularly studied in Drosophila. In this the myc antibody revealed LSM7 or SmG (Figure 6b) study, we have examined the status of human TACC1 confirming the interactions. Proteins from Cos-1 cells in carcinomas, especially breast cancer and looked for cotransfected with myc-tagged TACC1s and myc- a lead to its function. tagged LSM7 or myc-tagged SmG were immunoprecipitated by anti-TACC1-C antibody and Western The potential role of TACC1 in human breast cancer blotted. Blotting with anti-myc antibody revealed LSM7 or SmG (Figure 6c). Both GST pull-down We showed that the TACC1 gene is ubiquitously and co-immunoprecipitation results were highly expressed and codes for two putative protein isoforms reproducible. with coiled-coil motifs, i.e. an a superhelix (Lupas,

Oncogene TACC1 in breast cancer N Conte et al 5624 degradation product. We studied the expression of the TACC1 gene in human cancer. We found that TACC1 mRNA is underexpressed in several types of cancers. We further showed that TACC1 mRNA and TACC1 protein are underexpressed in a good proportion of breast tumors. When present, the TACC1 protein is localized largely in the cytoplasm of tumor cells. We and others (Dhanasekaran et al., 2001) have also observed downregulation of TACC1 mRNA respectively in breast and prostate tumors using cDNA arrays (Dhanasekaran et al., 2001; Bertucci et al., 2002). There have been conflicting results regarding the status of TACC genes in human cancers. The first member of the family was attributed a cell transforming capacity (Still et al., 1999a), TACC2 was in contrast viewed as a tumor suppressor gene (Chen et al., 2000) and TACC3 was suspected to act as an oncogene (Still et al., 1999b). It is frequent that different members of families of genes have contrasting activities, and may even antagonize each other. Another explanation has been proposed by Gergely et al. (2000b) who investigated the function of the D-TACC protein in the regulation of the interactions between centrosomes and microtubules: it has been proposed that, because an exquisite control is required in this sensitive area, either an increase or a decrease of TACC activity could have deleterious effects on genetic stability. This would explain how TACC1 may be transforming when overexpressed and in the same time down-regulated in human cancers. The TACC1 gene is located in a region where several cancer genes are suspected to be located (Ade´ laı¨ de et al., 1998; Ugolini et al., 1999). TACC1 may be one of the suspected TSG. Further work will be necessary to determine the mechanism leading to the alteration of TACC1 expression. As a preliminary analysis, we did a search for loss of heterozygosity (LOH) at microsatellite (CA)n markers of the 8p11 – 21 chromosomal region in microdissected breast tumor Figure 4 TACC1 mRNA expression in human cancer. (a) Dot samples. The frequency of LOH at D8S1821 (5’ of blot analysis of TACC1 mRNA in a variety of human cancers TACC1) and D8S255 (3’ of TACC1) was high but not using the commercial ‘cancer profiling array’. Arrayed SMART- amplified cDNAs from tissues were hybridized with a 32P-labeled the highest (not shown). This suggests that deletion TACC1 probe. For each type of tumor (Sm. Int.: small intestine), may be one mechanism of TACC1 inactivation; other couple of normal samples (N) are arrayed to the right, tumor genes in the vicinity of TACC1 may be more frequently samples (T) are arrayed to the left, in one or two rows. In few altered in breast cancer. cases, an additional spot represents a metastasis sample (boxed, lower spot). cDNA derived from cell lines and negative controls (yeast RNA, E. Coli DNA, polyA stretches, Cot-1 DNA – The potential normal function of TACC1: the RNA boxed) are spotted to the right. The control used to determine equal amount of cDNA in each spot is done by the manufacturer, connection by hybridization with an ubiquitin probe (not shown). (b) North- To better understand the potential role of TACC1 in ern blot analysis of TACC1 mRNA in breast carcinomas. (c) Graph representation in log scale of the expression levels of cancer, we initiated a study of its normal cell function. TACC1l mRNA as determined by Northern blot as shown in We searched for potential protein partners of TACC1 (b) and compared to the level (arbitrarily set to 1) in normal using the two-hybrid system in yeast and isolated the breast tissue LSM7 and SmG proteins. The interaction between TACC1 and LSM7 was confirmed by biochemical tests. LSM7 and SmG belong to a class of small proteins with Sm motifs that are highly conserved from 1996). The major form has an observed molecular mass yeast to mammals and associate with the U1-U6 of 125 – 130 kDa. A shorter form is observed in more snRNAs to form the snRNPs. There are eight stringent conditions of cell lysis and may represent a ubiquitous Sm proteins (SmB, B2, D1-D3, E, F, G); protein strongly associated with a cell component or a they associate with U1, U2, U4 and U5. There are 10

Oncogene TACC1 in breast cancer N Conte et al 5625

Figure 5 Immunodetection of TACC1 in breast tissues. (a) Example of immunohistochemistry analysis of TACC1 protein in breast tumor samples arrayed in a speciﬁc tissue microarray; left: negative tumor, right: positive tumor. Insert: hematoxylin-eosin staining of a parafﬁn block section (25630 mm) from the tissue microarray containing 435 samples (5657 tumor cylinders and controls). (b – d) Immunohistochemistry analysis of TACC1 protein in breast tumor sections. (b) TACC1 staining of a normal duct: the lumi- nal but not the myoepithelial cells are positive (arrow). (c) TACC1 staining of a positive tumor; note the strong staining in the perinuclear region (arrows); (d) Absence of TACC1 staining in a negative tumor

LSM proteins (LSM1 – 10) in humans; they specifically metabolism. This role has been demonstrated in associate with U6 snRNA and with mRNA; they thus Xenopus. In this animal, Maskin, the TACC3 ortholog, play a role in several aspects of mRNA processing; in associates with the 3’UTR-CPE binding protein particular they activate mRNA decapping and degra- (CPEB) to repress eIF-4E control of translation and dation, and function in pre-mRNA splicing (He and oocyte maturation (Stebbins-Boaz et al., 1999; Mendez Parker, 2000; Mitchell and Tollervey, 2000, for and Richter, 2001). CPEB and Maskin are present on reviews). The Sm and LSM related families assemble the mitotic apparatus of animal pole blastomeres in in two distinct seven-member ring complexes (Salgado- embryos (Groisman et al., 2000). Our results show that Garrido et al., 1999). The Sm and LSM proteins human TACC1 also associates with regulators of shuttle from the cytoplasm to the nucleus to ensure mRNA processings. The role of TACC1 in this their functions. A phylogenetic tree (Figure 7) of the regulation will need to be better precised. Our study LSM/Sm protein families showed that LSM7 and suggests that alteration of TACC1 function in mRNA SNRPG/SmG, and their respective orthologs metabolism may have important consequences in cell CG13277 and CG9742, constitutes a subfamily and behavior and may participate in carcinogenesis events. share a common ancestor (7/G) that separated from Indeed, an Sm-like protein, CaSm/LSM1, has been other ancestors (i.e. 2/D1, 3/D2, 4/D3, 5/E, 6/F). There shown to contribute to the transformed state of are no recorded mutant alleles of CG13277 and pancreatic cells (Schweinfest et al., 1997), and CaSM CG9742 (http://flybase.bio.indiana.edu:82/). It is possi- antisense therapy has been shown to reduce tumor ble that different TACC proteins associate with pancreatic growth (Kelley et al., 2000). members of distinct Sm subfamilies. We cannot exclude Several important studies have shown that TACC that TACC1 does not interact with other Sm or Sm- proteins interact with centrosomes and microtubules in like proteins. It will also be interesting to see if D- various ways (Gergely et al., 2000a,b; Cullen and TACC interacts with any of the Drosophila Sm Ohkura, 2001); TACC1 proteins are only weakly orthologs. concentrated at centrosomes, and only during mitosis; The interaction of TACC1 with LSM7 and SmG when overexpressed TACC1 proteins are uniformly points to a role of TACC1 in the control of mRNA distributed in the cytoplasm during interphase and

Oncogene TACC1 in breast cancer N Conte et al 5626

Figure 6 Identiﬁcation of TACC1 partners and phosphorylation of TACC1. (a) GST pull-down experiment. HA-tagged TACC1 (TACC1l-HA) was transiently expressed in Cos-1 cells and pulled down with GST, GST-LSM7 or GST-SmG fusion proteins. After washing, bound proteins were probed with anti-HA antibody. One tenth of the lysate used for the precipitation was run for control (right lane). Comparable amount of GST fusion proteins were present in the lanes (data not shown). (b) Co-immunoprecipitation of LSM7 and SmG with TACC1l protein. HA-tagged TACC1 (TACC1l-HA) was coexpressed with mock vector (pRK5-myc), myc- tagged LSM7 or myc-tagged SmG in Cos-1 cells. Proteins were immunoprecipited with either anti-myc or anti-TACC1-C antibody and bound proteins were respectively detected with anti-HA or anti-myc antibody. TACC1l was co-immunoprecipitated with both LSM7 and SmG protein. Lysates were probed with anti-HA antibody (upper panel) and anti-myc antibody (lower panel). One tenth of the lysate used for the precipitation was run for control (TL: total lysate). (c) Co-immunoprecipitation of LSM7 and SmG with TACC1s protein. Myc-tagged TACC1s (TACC1s-myc) was coexpressed with mock vector (pRK5-myc), myc-tagged LSM7 or myc- tagged SmG in Cos-1 cells. Proteins were immunoprecipitated with anti-myc antibody and bound proteins were detected with anti- myc antibody. TACC1s was co-immunoprecipitated with both LSM7 and SmG protein. Lysates were probed with anti-myc antibody. One tenth of the lysate used for the precipitation was run for control (TL: total lysate). (d) In vitro kinase assay showing phosphorylation of TACC1l. Lysates of Cos-1 cells transfected either by TACC1l-HA or TACC1s-myc were immunoprecipitated with anti-TACC1-C antibody before assay in the presence of [g32P]ATP. Proteins were separated on 7.5% gel and the dried gel was submitted to autoradiography

form ordered structures that cluster around the nucleus Prigent, 1999), which phosphorylates D-TACC (Giet et (Gergely et al., 2000a). In Drosophila and in humans, al., 2002). Human Aurora A/STK6 coimmunoprecipi- msps (mini spindles) and the human ortholog of msps, tates with TACC3 (Giet et al., 2002). Some of the a microtubule-associated protein called ch-TOG (Char- interactions of TACC proteins with their partners are asse et al., 1995, 1998; Charrasse and Larroque, 2000) likely to be regulated by phosphorylations. We have have been identified as interactors of the D-TACC and shown here that TACC1l can be phosphorylated. TACC3 protein respectively (Cullen and Ohkura, 2001; Our current hypothesis is that TACC proteins Lee et al., 2001). It has been reported recently that ch- regulate mRNA processing by participating in TOG interacts with human TACC1 in a two-hybrid complexes of proteins that are localized in specialized screen (Lauffart et al., 2002). In fly, protein complexes areas (including the centrosomes) in which specific with D-TACC include the aurora kinase (Giet and proteins are being locally produced (Kloc et al., 2002).

Oncogene TACC1 in breast cancer N Conte et al 5627 Materials and methods

Oligonucleotide primers The following primers were used: sense primer: TACC1S6: 5’- CGGGGATCCATGGCGTTCAGCCCGTGGCA-3’ (nucleotides 320 – 340 in the AF049910 TACC1 sequence); antisense primers: TACC1AS6: 5’-GGATTCGGGGAGAGGTGTG- CCCTCCACAG-3’ (1078 – 1100); TACC1AS2: 5’-AAAGC- GGGACAGGCAGCAGTAG-3’ (2891 – 2867); TACC1AS1: 5’-CTAGACTATGCAACCCTCTAGGG-3’ (2963 – 2940); TACC1AS0: 5’-CAAACTCAGCAGCCTAAGGGAACCTT- CATAG-3’ (3021 – 2982).

Cloning of TACC1 cDNAs Two different experiments of 5’ RACE PCR were necessary to obtain the coding sequences of TACC1 by using a Marathon cDNA amplification kit (Clontech, Palo Alto, CA, USA). First, a 1.9 kb TACC1 cDNA fragment (nucleotides 933 to 2867 bp in the AF049910 TACC1 sequence) was amplified as follows: first-strand cDNA was synthetized from 2 mg of total RNA from the Burkitt’s lymphoma DAUDI cell line by using the specific TACC1AS0 primer; second-strand cDNA was synthetized and adapter- ligated according to the manufacturer’s protocol: a first- round of PCR using the TACC1AS1 primer and the adapter Figure 7 Phylogenetic tree of human and Drosophila Sm-like primer AP1 from the Marathon kit (35 cycles with proteins. Sm domains of Sm and LSM proteins were used for denaturation at 948C (30 s), annealing at 658C (30 s), and alignments. The tree is unrooted. Each human Sm protein and each LSM protein have an ortholog in fly but not all Drosophila elongation at 688C (4 min) using the Advantage TAQ orthologs are shown. LSM7 and SNRPG/SmG, the two identified polymerase (Clontech). One tenth of the PCR product was interactors of human TACC1, constitutes a subfamily (see also reamplified with a nested-specific TACC1 primer Salgado-Garrido et al., 1999). Bootstrap values (out of 1000 repli- (TACC1AS2) and the adapter primer AP2. PCR conditions cates) are indicated at each node were the same as for the first-round amplification. The second-round PCR product was purified by using Nucleospin extract kit (Macherey-Nagel, Du¨ ren, Germany) and subcloned into the pGMTe (Promega, Madison, USA), and sequenced by Genome Express (Meylan, France). Second, to These proteins include Cyclin B1, which controls entry obtain the 5’ part of TACC1 cDNA, the single-strand cDNA in mitosis and whose mRNA and protein are was obtained from 2 mg of PolyA+ RNA from normal breast concentrated together with Maskin/TACC3 on centro- (Clontech) using random hexanucleotides. Second-strand was somes and spindles (Groisman et al., 2000; Richter, obtained using the specific primers TACC1S6 and 2001). TACC proteins, regulated by mitotic kinases TACC1AS6. PCR conditions were as follows: 35 cycles with (Glover et al., 1998; Giet and Prigent, 1999), may help denaturation at 948C (30 s), annealing at 658C (30 s), and elongation at 728C (1 min) using the TAQ DNA polymerase coordinate microtubule-associated events involved in (Life Technologies). Subcloning of the cDNA fragment, and the control of mitosis, in particular the G2/M sequencing were done as described above. The full-length checkpoint. This critical step is highly regulated and coding cDNA (TACC1-long: TACC1l) was reconstructed loss of function of TACC3 triggers p53-mediated from the two fragments by using the StuI enzymatic site apoptosis (Piekorz et al., 2002). The role of TACC/ localized at nucleotide 939 of the TACC1 sequence. The Maskin might extend to the control of mRNAs en TACC1l sequence was then subcloned in pCDNA-3 tagged route along the microtubule tracks, to specialized areas HA in 5’ and in pBTM116 vector. A 1516-kb fragment in polarized cells (Richter, 2001), in the RNA granule corresponding to a spliced TACC1 isoform (TACC1-short: (Carson et al., 1998). Whether the precise role of TACC1s) was inserted in XbaI/HindIII sites of pRK5 tagged TACC1 is similar or is involved in a different aspect of myc in 5’. mRNA control will have to be determined. A central position at specific sites and crucial function in cell TACC1 genomic structure cycle and/or cell polarity would endow TACC proteins with potential carcinogenic capacity. In cancer, there The genomic structure of the gene was deduced based on sequence comparison of the cDNA sequence of TACC1 would be a release of TACC control on mRNAs (AF049910) with Homo sapiens genomic contig sequence coding for proteins involved in cell division and at http://www.ncbi.nlm.nih.gov/genome/seq/HsBlast.html. polarity, leading to abnormal mitotic daughter cells Briefly, in BLAST searches of genome sequence databases and/or defects in polarity. Alteration of the regulation (Altschul et al., 1997), the coding region of TACC1 identified a of mRNA metabolism represents a possible important genomic sequence from the working draft sequence fragment mechanism of carcinogenesis of which there remains AC067817 (gi:13959300 at http://www.ncbi.nlm.nih.gov:80/ much to discover (Clemens and Bommer, 1999). cgi-bin/Entrez/hum_srch?chr=hum_chr.inf&query) composed

Oncogene TACC1 in breast cancer N Conte et al 5628 of 28 unordered pieces. Alignment of these overlapping pieces normal breast (RNA purchased from Clontech) and the with the cDNA sequence allowed to determine the genomic tumor samples was assessed by scan of the filters in a Fujix organization of the TACC1 gene. Bas 1500 system (Fuji, Tokyo, Japan). The expression results were normalized with RNA 28S. The quantitative analyses of hybridization results were done using a Bioimage software Cell lines (TINA 2.09, Raytest, Straubenhardt, Germany). Data reports HeLa cell line and breast carcinoma cell lines BT-20, MCF7, were then exported into a microcomputer and analysed using MCF10F, MDA-MB-134, MDA-MB-157, MDA-MB-175, Excel software (Microsoft). The ‘Cancer profiling array’ from MDA-MB-231, MDA-MB-453, CAMA-1, HCC1395, Clontech (number 7841-1) was hybridized with the same HCC1569, HCC1806, HCC1937, SK-BR-7, T47D, UACC- TACC1 probe according to the manufacturer’s instructions 812, and ZR-75-1, and breast epithelial cell line HBL-100 and the measurements were done as above. were purchased from the American Type Culture Collection (http://www.atcc.org/). The HME-1 cell line was purchased Antibodies from Clontech. The BrCa-Mz-02 cell line is a gift from Dr V Mo¨ bus (Ulm, Germany) and the SUM-102, SUM-149, SUM- The GPEGDPEEEDSQAETK peptide, corresponding to 185 cell lines a gift from Dr S Ethier (Michigan, USA; http:// amino acids 35 – 50 of the TACC1 sequence, was chosen www.cancer.med.umich.edu/breast_cell/clines/clines.html). All for chemical synthesis (Neosystems, Strasbourg, France) these cells were grown as recommended. Cos-1 cells were owing to its predicted antigenicity. The peptide was coupled maintained in Dulbecco’s modified Eagle’s medium with 10% to keyhole limpet hemocyanin as hapten, suspended in PBS, newborn serum. All cells were cultured in a 5% CO2 and used to immunize rabbits by intramuscular and incubator at 378C. subcutaneous injections. The antibody generated against the peptide was designed ‘anti-TACC1-N’. It was subsequently affinity purified on a Affigel-15 column (BioRad). To obtain Patient material, clinicopathologic and statistical analyses anti-SPAZ1 and anti-TACC1-C specific antibodies, the Tumors were collected at Institut Paoli-Calmettes GDARFQKSPPDIKETPY peptide, corresponding to amino (Marseille, France) and flash frozen in liquid nitrogen. All acids 280 – 295, and the YSDKTAVLTLIREE peptide, samples were reviewed by two independent pathologists, corresponding to amino acids 601 – 612 of the TACC1 and classified according to the WHO recommendation. The sequence were respectively chosen, and the same procedure series of 61 tumors were tested in Northern blot including was used. The following polyclonal antibodies were used for 44 ductal invasive, 12 lobular invasive, two medullary immunofluorescence, immunoprecipitation and immunoblot- carcinomas, three mixed ductal and lobular carcinomas. ting: anti-HA (3F-10, Roche Diagnostics, Meylan, France) The series of 57 tumors tested in tissue microarray and the mouse monoclonal anti-myc 9E10 (Santa Cruz contained 45 ductal invasive, four lobular invasive, three Biotechnology). A commercial anti-TACC1 (Euromedex, medullary and five mixed ductal and lobular carcinomas. Mundolsheim, France) was also used for comparative The median age of the patients was 53. A series of 48 Western immunoblotting experiments. consecutive breast carcinoma samples were tested in whole section immunohistochemistry, including 38 ductal invasive Immunofluorescence analyses (31 of common type, four tubular, two with predominant intraductal component, one mucinous), seven lobular Hela or Cos-1 cells were grown on coverslips (26105 cells/ invasive, three mixed ductal and lobular carcinomas. The 60 mm plates) 1 day before transient transfection. Twelve median age of the patients was 57 years. The clinical data hours after transfection, cells were washed once in PBS and and follow-up of the women included in the study were fixed in 3.7% paraformaldehyde in 16PBS for 15 min at obtained from medical records. The following clinical and room temperature. After extensive washes, cells were histopathological parameters were analysed: menopausal permeabilized and blocked in 5% fetal calf serum/PBS, status, histological tumor size, axillary lymph node status, 0.1% Triton X-100 for 20 min. Cells were then incubated histologic grade, proliferation index (Ki-67 evaluation), P53 with either affinity-purified anti-TACC1-N (dilution, 1 : 100), and ERBB2 immunohistochemistry. Statistical analysis was which gave the best results of all anti-TACC1 antibodies in done by w2 or Fisher exact test. immunofluorescence, anti-myc (1 mg/ml), or anti-HA (dilution 1 : 1000) antibody for 1 h at room temperature, rinsed several times in 16PBS, and then incubated with 2 mg/ml of the RNA analysis Alexa-conjugated either anti-rabbit or anti-mouse or anti-rat The multiple-tissue human Northern blots (Clontech, Palo secondary antibody, respectively. Control of the specificity of Alto, CA, USA; number 7759-1, 7760-1, and 7767-1) were anti-TACC1-N was done as described above. The DNA was hybridized according to the manufacturer’s instructions. colored by incubation with Syto-16 (Molecular probes RNAs from tumor samples were isolated from frozen tissue Eugene, OR, USA). samples by the guanidine isothiocyanate method. Ten mgof each total RNA were subjected to electrophoresis in 1% Confocal analysis agarose/10% formaldehyde gels, and transferred onto biotrans membranes (ICN, CA, USA). Hybridizations were The three-dimensional immunofluorescence images were done according to Church and Gilbert (1984) with a 32P- recorded by TCS-NT confocal Microscope (LEICA micro- labeled TACC1 probe (Megaprime kit RPN-1607, system, Mannheim, Germany). Three-dimensional views of Amersham-Pharmacia Biotech, Buckinghamshire, UK) corre- cells were generated as section representation mode with the sponding to a 827-bp EcoRI/NotI fragment from zdl18D05 IMARIS software (BITPLANE AG, Zurich, Switzerland). cDNA clone (from 6911 to 7738 bp in the TACC1 sequence). Before this, three-dimensional images restoration was made The b actin probe was purchased from Clontech. The filters with an MLE (Maximum Likehood Estimation) iterative were washed at high stringency and imaged using a phospho- deconvolution method using HUYGENS2 software (SVI, imager. Differential expression of TACC1 mRNA between BITPLANE AG) on SGI o2 workstation.

Oncogene TACC1 in breast cancer N Conte et al 5629 Breast tumor analyses with antibodies overnight at 48C. Protein A-agarose was added and immune complexes bound to beads were recovered Tissue microarray analysis It was done as described (Richter et after 1 h, washed three times with lysis buffer described above al., 2000) with slight modifications. Briefly, for each tumor, five and boiled in 16 sample buffer. representative areas were selected from a hematoxylin-eosin stained section of a donor block; five corresponding cylinders with a diameter of 0.6 mm each were punched from these areas Immunoblotting analysis Cell extracts were boiled in reducing and deposited into a recipient paraffin block using a specific buffer for 5 min and protein were loaded onto 7.5% acrylamide arraying device (Microm, Francheville, France). Five-mm gels. Gels were transferred onto an Nitrocellulose membrane sections of the resulting microarray block were made and used (Amerscham, Life science) at 30 V for 18 h using a transblot for immunohistochemistry analysis after transfer to glass slides. apparatus (BioRad). Residual binding sites were blocked by membrane incubation in PBS-5% dry milk solution (Eurobio, France) for 2 h at room temperature. After incubation with Immunohistochemistry Five-mm sections from formalin- anti-TACC1-N, anti-SPAZ1 (dilution: 1 : 300 in TBS-5% dry embedded tissue specimens were deparaffinized in xylene and 1 milk) for 12 h, or commercial anti-TACC1 (1 : 1000) overnight, rehydrated in graded alcohol. Antigen enhancement was done filters were washed in TBS-Tween 0.05%. Peroxidase labeled by incubating the sections in Target Retrieval solution (Dako, anti rabbit serum preformed complex (Jackson Immuno- Copenhagen, Denmark) as recommended. Slides were then Research, USA) was then added (dilution: 1 : 10 000 in TBS- transferred to a Techmate 500 automater stainer (Dako). 5% dry milk) and incubation was done at room temperature Staining was performed at room temperature as follows: after for 45 min. After several washings in TBS-Tween 0.05%, filters washes in phosphate buffer, slides were first incubated with were reacted 1 min with chemiluminescent substrate luminol blocking serum (Dako) for 10 min and then with the affinity- (ECL kit; Amersham) and revealed by brief exposure using purified anti-TACC1-N (dilution, 1 : 100) for 1 h, followed by autoradiography films (Kodak, XO Mat AR). quenching of endogenous peroxidase activity by treatment with 0.1% H 2O2. After washes, the slides were incubated with GST pull-down assays GST fusion proteins were expressed biotinylated antibody against rabbit Ig for 20 min followed by R by standard techniques (Batzer et al., 1994). Lysates were streptavidin conjugated peroxydase (DAKO LSAB 2 kit), incubated with GST fusion proteins bound to glutathione counter-stained with hematoxylin, and coverslipped using agarose beads for 2 h at 48C. The beads were then washed Aquatex (Merck, Darmstadt, Germany) mounting solution. three times with lysis buffer, boiled in 16 sample buffer, and In control experiments, TACC1 immunogen peptide (10 mg/ separated by SDS – PAGE. In all experiments, the presence of ml) was coincubated with anti-TACC1-N antibody. Slides comparable amount of GST was verified by Ponceau staining were evaluated under a light microscope by two pathologists (E of the membrane. Charafe-Jauffret and J Jacquemier). In vitro kinase assay Cos-1 cells transfected with either Two-hybrid procedure TACC1l or TACC1s were rinsed twice in PBS and lysed in 500 ml NP40 lysis buffer supplemented with 1 mM PMSF and To prepare the baits, proteins were fused to the LexA-BD 1mM Na orthovanadate. Cell lysates were incubated with subunit using pBTM116 vector which carries Trp1. For anti-TACC1-C antibody for 2 h, then protein A-Sepharose library screening, an oligo-dT primed human cDNA breast was added. Immune complexes were washed three times with library cloned in pACT2 vector (Clontech) which carries lysis buffer and twice with WASH buffer (50 mm Tris pH 7.5, Leu2 as a selection marker was screened by using the LexA- 1% NP40, 1 mM Na orthovanadate). Pellets were resus- TACC1l as a bait and the yeast strain L40 following the pended in 10 ml KRB (WASH buffer supplemented with lithium-acetate protocol. Approximately 130 Trp+Leu+ 10 mM MnCl2) and incubated for 15 min at 308C with transformants were selected on plates with supplemented 0.15 mCi [g-32P]ATP (NEN, Les Ulis, France). The reaction minimum medium that lacked tryptophan, leucine, histidine was stopped with cold lysis buffer and the pellets were in the primary screening and contained 10 mM 3-aminotria- resuspended in 36 sample buffer. Precipitated proteins were zole (3-AT) and then tested for the b-galactosidase activity by resolved by 7.5% acrylamide gel. The gel was then dried and the filter method in the secondary screening. After rescue, the exposed to Kodak XAR film. DNA of selected clones was retransformed in L40 yeast containing LexA-TACC1l or LexA fused to control proteins. Specific clones were positive for growth in histidine deficient Phylogenetic analysis medium and b-galactosidase activity. LSM7 and SmG A phylogenetic tree with human (http://www.ncbi.nlm.nih.- pACT2 clones were subcloned in pRK5 tagged myc vector gov/LocusLink/) and Drosophila (http://hedgehog.lbl.gov: for biochemical studies and in PGEX-Tag vector for the 8001/cgi-bin/annot/query) sequences was constructed using production of all GST fusion proteins. the distance matrix (Blosum 30 matrix) and neighbor-joining algorithms implemented in Clustal W. A total of 1000 Biochemical procedures bootstrapped replicates were run. Protein procedures Cells were washed twice with cold PBS and lysed either in RIPA buffer (150 mM NaCl, 1% Nonidet Acknowledgments P-40 (NP40), 0.5% deoxycholate, 0.1% SDS and 50 mM Tris- We thank F Birg, C Mawas and D Maraninchi for HCl pH 8.0) or in NP40 lysis buffer (20 mM Tris, HCl discussions and support. We are especially grateful to J-P pH 7.4, 50 mM NaF, 150 mM NaCl, 1% NP40, 1 mM Borg and B Puig for discussions and help. This work was NaVO4, 1 mM EDTA) supplemented with 1 mM phenyl- supported by INSERM, Institut Paoli-Calmettes, and methylsulfonylfluoride, 10 mg.ml71 aprotinin and 10 mg.ml71 grants from the Association pour la Recherche contre le leupeptin. After centrifugation at 16 000 g for 10 min, lysate Cancer and the Ligue Nationale contre le Cancer (Label). protein content was normalized using the BioRad protein N Conte and B Delaval are recipients of a fellowship from assay kit. For immunoprecipitation, lysates were incubated MESR.

Oncogene TACC1 in breast cancer N Conte et al 5630 References

Ade´ laı¨ de J, Chaffanet M, Imbert A, Allione F, Geneix J, Lauffart B, Howell SJ, Tasch JE, Cowell JK and Still IH. Popovici C, Van Alewijk D, Trapman J, Zeillinger R, (2002). Biochem. J., 363, 195 – 200. Borresen-Dale A-L, Lidereau R, Birnbaum D and Lee MJ, Gergely F, Jeffers K, Peak-Chew SY and Raff JW. Pe´ busque M-J. (1998). Genes Chromosomes Cancer, 22, (2001). Nat. Cell. Biol., 3, 643 – 649. 186 – 199. Lupas A. (1996). TIBS, 21, 375 – 382. Altschul SF, Madden TL, Scha¨ ffer AA, Zhang J, Zhang Z, Mendez R and Richter JD. (2001). Nat. Rev. Mol. Cell Biol., Miller W and Lipman DJ. (1997). Nucleic Acids Res., 25, 2, 521 – 529. 3389 – 3402. Mitchell P and Tollervey D. (2000). Curr. Opin. Genet. Bertucci F, Nasser N, Granjeaud S, Eisinger F, Ade´ laı¨ de J, Develop., 10, 193 – 198. Tagett R, Loriod B, Giaconia A, Benziane A, Devilard E, Popovici C, Leveugle M, Birnbaum D and Coulier F. (2001). Jacquemier J, Viens P, Nguyen C, Birnbaum D and FEBS Lett., 491, 237 – 242. Houlgatte R. (2002). Hum. Mol. Genet., 11, 863 – 872. Piekorz RP, Hoffmeyer A, Duntsch CD, McKay C, Batzer AG, Rotin D, Urena JM, Skolnik EY and Schles- Nakajima H, Sexl V, Snyder L, Rehg J and Ihle JN. singer J. (1994). Mol. Cell. Biol., 14, 5192 – 5201. (2002). EMBO J., 21, 653 – 664. Carson JH, Kwon S and Barbarese E. (1998). Curr. Opin. Pribill I, Speiser P, Leary J, Leodolter S, Hacker N, Neurol., 9, 607 – 612. Friedlander ML, Birnbaum D, Zeillinger R and Krainer Charrasse S and Larroque C. (2000). Exp. Cell Res., 254, M. (2001). Cancer Genet. Cytogenet., 129, 23 – 29. 249 – 256. Pu JJ, Li C, Rodriguez M and Banerjee D. (2001). Cytokine, Charasse S, Mazel M, Taviaux S, Berta P, Chow T and 13, 129 – 137. Larroque C. (1995). Eur. J. Biochem., 234, 406 – 413. Richter JD. (2001). Proc.Natl.Acad.Sci.USA,98, 7069 – Charasse S, Schroeder M, Gauthier-Rouvie` re C, Ango F, 7071. Cassimeris L, Gard DL and Larroque C. (1998). J. Cell Richter JD, Wagner U, Kononen J, Fijan A, Bruderer J, Sci., 111, 1371 – 1383. Schmid U, Ackermann D, Maurer R, Alund G, Kno¨ nagel Chen H-M, Schmeichel K, Mian LS, Lelie` vre S, Petersen OW H, Rist M, Wilber K, Anabitarte M, Herring K, and Bissell MJ. (2000). Mol. Biol. Cell, 11, 1357 – 1367. Hardmeier T, Scho¨ nenberg A, Flury R, Ja¨ gerP,FehrJL, Church G and Gilbert W. (1984). Proc. Natl. Acad. Sci. USA, Schraml P, Moch H, Mihatsch MJ, Gasser T, Kallionemi 81, 1991 – 1995. OP and Sauter G. (2000). Am. J. Pathol., 157, 787 – 794. Clemens MJ and Bommer UA. (1999). Int. J. Biochem. Cell Sadek CM, Jalaguier S, Feeney E, Aitola M, Damdimopou- Biol., 31, 1 – 23. los AE, Pelto-Huikko M and Gustafsson J-A. (2000). Cullen CF and Ohkura H. (2001). Nat. Cell Biol., 31, 637 – Mech. Develop., 97, 13 – 26. 642. Salgado-Garrido J, Bragado-Nilsson E, Kandels-Lewis S Dhanasekaran SM, Barrette TR, Ghosh D, Shah R, and Se´ raphin B. (1999). EMBO J., 18, 3451 – 3462. Varambally S, Kurachi K, Pienta KJ, Rubin MA and Schweinfest CW, Graber MW, Chapman JM, Papas TS, Chinnaiyan AM. (2001). Nature, 412, 822 – 826. Baron PL and Watson DK. (1997). Cancer Res., 57, 2961 – Gergely F, Karlsson C, Still I, Cowell J, Kilmartin J and Raff 2965. JW. (2000a). Proc. Natl. Acad. Sci. USA, 97, 15352 – Stebbins-Boaz B, Cao Q, de Moor CH, Mendez R and 14357. Richter JD. (1999). Mol. Cell, 4, 1017 – 1027. GergelyF,KiddD,JeffersK,WakeﬁeldJGandRaffJW. Still IH, Hamilton M, Vince P, Wolfman A and Cowell JK. (2000b). EMBO J., 19, 241 – 252. (1999a). Oncogene, 18, 4032 – 4038. Giet R, McLean D, Descamps S, Lee MJ, Raff JW, Prigent C Still IH, Vince P and Cowell JK. (1999b). Genomics, 58, 165 – and Glover DM. (2002). J. Cell. Biol., 156, 437 – 451. 170. Giet R and Prigent C. (1999). J. Cell Sci., 112, 3591 – 3601. Tharun S, He W, Mayes AE, Lennertz P, Beggs JD and Glover DM, Hagan IM and Tavares AAM. (1998). Genes Parker R. (2000). Nature, 30, 515 – 518. Dev., 12, 3777 – 3787. Theurkauf WE. (2001). Nat. Cell Biol., 3, E159 – E161. Groisman I, Huang Y-S, Mendez R, Cao Q, Theurkauf W Ugolini F, Ade´ laı¨ de J, Charafe-Jauffret E, Nguyen C, and Richter JD. (2000). Cell, 103, 435 – 447. Jacquemier J, Jordan B, Birnbaum D and Pe´ busque MJ. He W and Parker R. (2000). Curr. Opin. Cell Biol., 12, 346 – (1999). Oncogene, 18, 1903 – 1910. 350. Ugolini F, Charafe-Jauffret E, Bardou VJ, Geneix J, Herman H, Raker FP, Hornig FK, Brahms H and Luhrmann Ade´ laı¨ de J, Labat-Moleur F, Penault-Llorca F, Longy R. (1995). EMBO J., 14, 2076 – 2088. M, Jacquemier J, Birnbaum D and Pe´ busque MJ. (2001). Kallioniemi OP, Wagner U, Kononen J and Sauter G. Oncogene, 20, 5810 – 5817. (2001). Hum. Mol. Genet., 10, 657 – 662. Wang XZ, Jolicoeur EM, Conte N, Chaffanet M, Zhang Y, Kelley JR, Brown JM, Frasier MM, Baron PL, Schweinfest Mozziconacci MJ, Feiner H, Birnbaum D, Pe´ busque MJ CW, Vournakis JN, Watson DK and Cole DJ. (2000). and Ron D. (1999). Oncogene, 18, 5718 – 5721. Surgery, 128, 353 – 360. Kloc M, Zearfoss R and Etkin LD. (2002). Cell, 108, 533 – 544.

Oncogene