Oncogene (2005) 24, 4741–4753 & 2005 Nature Publishing Group All rights reserved 0950-9232/05 $30.00 www.nature.com/onc ORIGINAL PAPERS KIF14 is a candidate oncogene in the 1q minimal region of genomic gain in multiple cancers Timothy W Corson1,2, Annie Huang3, Ming-Sound Tsao4,5,6 and Brenda L Gallie*,1,2,5,7 1Division of Cancer Informatics, Ontario Cancer Institute/Princess Margaret Hospital, University Health Network, Toronto, ON, Canada M5G 2M9; 2Department of Molecular & Medical Genetics, University of Toronto, Toronto, ON, Canada M5S 1A8; 3Labatt Brain Tumour Research Centre, Cancer Research Program, Hospital for Sick Children, Toronto, ON, Canada M5G 1X8; 4Division of Cellular & Molecular Biology, Ontario Cancer Institute/Princess Margaret Hospital, University Health Network, Toronto, ON, Canada M5G 2M9; 5Department of Medical Biophysics, University of Toronto, Toronto, ON, Canada M5G 2M9; 6Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, ON, Canada M5G 1L5; 7Department of Ophthalmology, University of Toronto, Toronto, ON, Canada M5G 1X5 Gain of chromosome 1q31–1q32is seen in >50% of in retinoblastoma development is the loss of function of retinoblastoma and is common in other tumors. To define both alleles of the prototypic tumor suppressor gene, the minimal 1q region of gain, we determined genomic RB1, which encodes a key cell-cycle negative regulatory copy number by quantitative multiplex PCR of 14 transcription factor, pRB (Classon and Harlow, 2002). sequence tagged sites (STSs) spanning 1q25.3–1q41. The These initiating ‘M1’ and ‘M2’ mutations of Knudson’s most frequently gained STS at 1q32.1 (71%; 39 of 55 classic two-hit model for oncogenesis (Knudson, 1971) retinoblastoma) defined a 3.06 Mbp minimal region of are necessary for retinoblastoma initiation, but not gain between flanking markers, containing 14 genes. Of sufficient for malignancy. All retinoblastoma show these, only KIF14, a putative chromokinesin, was over- further mutational events (M3 to Mn), which perhaps expressed in various cancers by real-time RT–PCR. drive the transition to malignancy from the intermediate KIF14 mRNA was expressed in 20/22 retinoblastoma of benign retinoma (Gallie et al., 1999a; DiCiommo samples 100–1000-fold higher than in retina (t-test et al., 2000). P ¼ 0.00002); cell lines (n ¼ 10) had higher levels than Genomic aberrations may reflect M3 to Mn events. tumors (n ¼ 12) (P ¼ 0.009). KIF14 protein was over- The most prevalent karyotypic abnormality in retino- expressed in retinoblastoma tumors and breast cancer cell blastoma is gain of portions of the long arm of lines by immunoblot. KIF14 was expressed in 4/4 breast chromosome 1, seen in 21 of 27 tumors (Squire et al., cancer cell lines 31–92-fold higher than in normal breast 1985). At the higher resolution of comparative genomic tissue, in 5/5 medulloblastoma cell lines 22–79-fold higher hybridization (CGH), gain of a minimal region spanning than in fetal brain, and in 10/22 primary lung tumors 1q31 is the most common change seen in retinoblastoma 3–34-fold higher than in normal lung. Patients with lung in our study (Chen et al., 2001), and was seen tumors that overexpress KIF14 showed a trend toward cumulatively in 76/162 (47%) tumors across five decreased survival. KIF14 may thus be important in published studies (Figure 1a) (Mairal et al., 2000; Chen oncogenesis, and has promise as a prognostic indicator et al., 2001; Herzog et al., 2001; Lillington et al., 2002; and therapeutic target. van der Wal et al., 2003). Oncogene (2005) 24, 4741–4753. doi:10.1038/sj.onc.1208641; The long arm of chromosome 1 is particularly published online 9 May 2005 interesting in cancer biology, since gain of regions spanning 1q31–1q32 is reported in a wide variety of Keywords: retinoblastoma; breast cancer; lung cancer; cancers (Figure 1b). The Progenetix database of over medulloblastoma; kinesin; chromosome 1 10 000 tumor karyotypes, CGH, and array CGH profiles (Baudis and Cleary, 2001) indicates gain or amplifi- cation of this region in 43% of 414 breast cancers, 47% Introduction of 355 liver tumors, 22% of 322 cervical tumors, and 24% of 295 non-brain central nervous system tumors, Study of the rare pediatric eye cancer, retinoblastoma all of which show a pattern of 1q31–1q32 gain consistent (MIM# 180200), has revealed fundamental mechanisms with a ‘hotspot’ in this region (Baudis and Cleary, 2001) of cancer (DiCiommo et al., 2000). The initiating event (Figure 1b). Gain of 1q31–1q32 has also been seen in smaller fractions of cancers from over 60 other sites *Correspondence: BL Gallie, Division of Cancer Informatics, Room listed in the Progenetix database (Baudis and Cleary, 8-415, Ontario Cancer Institute/Princess Margaret Hospital, Univer- 2001), including 58/280 lung tumors and 13/94 medullo- sity Health Network, 610 University Avenue, Toronto, Ontario, blastomas and primitive neuroectodermal tumors Canada M5G2M9; E-mail: [email protected] Received 4 January 2005; revised 15 February 2005; accepted 17 (PNETs) (Figure 1b). Interestingly, chromosomal band February 2005; published online 9 May 2005 1E4 in the mouse genome, syntenic to 1q31, is gained in KIF14 is a 1q candidate oncogene in multiple cancers TW Corson et al 4742 Figure 1 Genomic gain at 1q31–1q32 is prominent in retinoblastoma and other cancers. (a) Gain of the long arm of chromosome 1, or even the entire chromosome, has been observed in five CGH studies of retinoblastoma. Numbers below figure indicate references: (1) Chen et al. (2001); (2) Herzog et al. (2001); (3) Mairal et al. (2000); (4) Lillington et al. (2002); (5) van der Wal et al. (2003). Vertical bars to the right of the chromosome ideogram indicate region of gain in an individual tumor; bars to the left indicate loss. Cumulatively, 83/ 162 (51%) tumors show gain of part or all of 1q, with a minimal common region of gain at 1q31–1q32 (gray boxed region) in 47%. (b) Gain at 1q31–1q32 is common in cancers at numerous sites. Fraction showing gain at 1q31–1q32 (black bars) of total tumors (white bars) at each cancer site with >10% 1q31–1q32 gain is shown. Tumor sites that display a ‘hotspot’ for gain at 1q31–1q32 are in boldface. The Progenetix database (Baudis and Cleary, 2001) was mined for gain and/or amplification at 1q31 and 1q32 for all tumor sites, with the exception of retina, where data came from our analysis of the five extant CGH studies in (a). Average percentage gain plus amplification across the region was calculated and plotted as a fraction of total tumors for each site. NHL, non-Hodgkin’s lymphoma; NOS, not otherwise specified several murine tumors, including lung adenocarcinoma compared to appropriate normal tissues. Patients whose (Sargent et al., 2002), mammary adenocarcinoma (Le primary lung cancers overexpress KIF14 tend to have Roy et al., 1993), teratocarcinomas (van Berlo et al., shorter survival. 1990), neuroblastomas (Hackett et al., 2003), and pancreatic tumors (Schreiner et al., 2003). The prevalence of 1q31 gain in the collective CGH Results and cytogenetic data suggests that one or more genes in this region are important in the genesis and/or develop- QM-PCR analysis of 1q gain in retinoblastoma ment of cancer. To test this hypothesis in the context of retinoblastoma and breast cancer, we analysed 1q To profile the genomic gain around 1q31 in retinoblast- genomic gain in 55 retinoblastoma tumors and 12 breast oma at a higher resolution than that achieved by CGH, cancer cell lines by quantitative multiplex (QM)–PCR. we scanned the region using QM-PCR in 55 retino- We identified a 3.06 Mbp minimal region of gain at blastoma tumors. An initial screen with five sequence 1q31.3–1q32.1, and found that one gene in this region, tagged sites (STSs), spaced evenly across 26.3 Mbp KIF14, an uncharacterized kinesin potentially involved spanning 1q25.3–1q41, identified a hotspot of gain, the in chromosome segregation, is highly expressed in STS SHGC-85288 at 1q31.3 (Figure 2a), gained in 62% retinoblastoma tumors, breast cancer and medulloblas- of tumors. In a second round of QM-PCR, we narrowed toma cell lines, and some lung primary tumors, the 13.2 Mbp hotspot to 3.06 Mbp centered on marker Oncogene KIF14 is a 1q candidate oncogene in multiple cancers TW Corson et al 4743 Figure 2 Minimal region of gain at 1q31.3–1q32.1 identified by QM-PCR. (a) Summary of QM-PCR analysis of copy number gains in 55 retinoblastoma. Proportion of tumors showing gain (copy number X mean of normals þ 3 s.d.) of each of 14 STSs (names indicated) spanning 1q25.3–1q41 is shown. SHGC-154194 at 1q32.1 is the hotspot STS (black), gained in 39/55 (71%) tumors. This hotspot defines a 1.03 Mbp minimal region of gain between flanking markers SHGC-133026 and SHGC-3266. A broader, 3.06 Mbp region of gain (indicated above figure) is defined by the next most commonly gained markers, SHGC-85288 and SHGC-76118, on either side of SHGC-154194. The approximate location of KIF14 is shown. (b, c) Detailed depiction of 1q gain in (b) retinoblastoma tumors and (c) breast cancer cell lines, as determined by QM-PCR. Each row is an individual tumor or cell line, while each column represents an STS along 1q (centromeric to telomeric, left to right). Black cells indicate gain and white cells indicate 2-copy. Previously published 1q CGH findings for each retinoblastoma tumor (Chen et al., 2001) or 1q cytogenetic data (available at www.atcc.org) for each breast cancer cell line, where known, are indicated to the right of the figure SHGC-154194, gained in 71% of retinoblastoma. A (Figure 2a).