Multiple chromosome 3p tumor suppressor genes (TSG) have been proposed in the pathogenesis of ovarian cancer based on complex patterns of 3p loss. To attain functional evidence in support of TSGs and identify candidate regions, we applied a chromosome transfer method involving cell fusions of the tumorigenic OV90 human ovarian cancer cell line, monoallelic for 3p and an irradiated mouse cell line containing a human chromosome 3 in order to derive OV90 hybrids containing normal 3p fragments. The resulting hybrids showed complete or incomplete suppression of tumorigenicity in nude mouse xenograft assays, and varied in their ability to form colonies in soft agarose and three-dimensional spheroids in a manner consistent with alteration of their in vivo tumorigenic phenotypes. Expression microarray analysis identified a set of common differentially expressed genes, such as SPARC, DAB2 and VEGF, some of which have been shown implicated in ovarian cancer. Genotyping assays revealed that they harbored normal 3p fragments, some of which overlapped candidate TSG regions (3p25–p26, 3p24 and 3p14-pcen) identified previously in loss of heterozygosity analyses of ovarian cancers. However, only the 3p12-pcen region was acquired in common by all hybrids where expression microarray analysis identified differentially expressed genes. The correlation of 3p12-pcen transfer and tumor suppression with a concerted re-programming of the cellular transcriptome suggest that the putative TSG may have affected key underlying events in ovarian cancer.
Molecular genetic analysis of epithelial ovarian cancer has suggested a role for tumor suppressor genes (TSG) located on the short arm of chromosome 3 (3p) in the pathogenesis of this disease. Cytogenetic studies have detected recurrent numerical and structural anomalies of chromosome 3 in ovarian cancer that include a nonrandom loss of all or regions of the 3p arm (Whang-Peng et al., 1984; Pejovic et al., 1992; Yonescu et al., 1996; Mertens et al., 1997). A temporal order of chromosomal imbalances, where whole or partial chromosome 3 loss occurs as an early event in ovarian tumorigenesis in a subset of ovarian cancers, has been proposed (Hoglund et al., 2001, 2003). It has also been proposed that ovarian cancers develop through ‘phases of karyotypic evolution’ where an initial phase of increased instability and complexity of chromosome anomalies is followed by their stabilization (Hoglund et al., 2003). As observed with most chromosome arms, there appears to be a higher frequency of deletions at telomeric ends of chromosomes relative to centromeric ends. However, 3p losses are distinctly different from that of other chromosomes with a high frequency of 3p14/3p21 breaks and 3p14–p21 interstitial losses (Hoglund et al., 2003). The fragile site, FRA3A at 3p14, which is prone to breakage does not readily explain these results as breakage often occurs proximal to this region, suggesting that another mechanism is at play (Hoglund et al., 2001, 2003). In an independent study, 3p cytogenetic events were associated with a subset of ovarian cancer, particularly cases with poor prognosis (Taetle et al., 1999a, 1999b; Simon et al., 2000). Loss of heterozygosity analysis (LOH) has identified candidate TSG loci on 3p (Ehlen and Dubeau, 1990; Zheng et al., 1991; Dodson et al., 1993; Lounis et al., 1998; Fullwood et al., 1999; Petursdottir et al., 2004). We have reported two minimal regions of LOH associated with 3p25–p26, 3p24 and a third region of deletion proximal to a marker at 3p14, in the analysis of a large panel of ovarian tumors (Lounis et al., 1998). Transcriptome analysis of four spontaneously immortalized ovarian cancer cell lines showed decreased expression of 3p genes in association with reduction to hemizygosity of the 3p arm (Manderson et al., 2002a). Recent evidence from parallel analyses of sporadic primary ovarian carcinomas by spectral karyotyping, comparative genomic hybridization and expression microarrays also has revealed anomalies involving chromosome 3 (Bayani et al., 2002). Although the cytogenetic models did not take into consideration the underlying mechanism of 3p gene dysregulation, the results combined with LOH analyses suggest that 3p harbors multiple TSGs important in the pathogenesis of ovarian cancer.
There remains a paucity of functional evidence to support 3p TSGs in ovarian cancer. Microcell-mediated chromosome transfer (MMCT) of whole chromosome 3 was shown to suppress tumorigenicity in the HEY cell line (Rimessi et al., 1994). Although HEY harbors a 3p12 deletion based on karyotype analysis, the ‘revertant’ hybrids that developed tumors each harbored loss of the same three regions, two at 3p23–p24.2 and one at 3p21.1–p21.2, suggesting that they contain TSGs (Rimessi et al., 1994). Transfection of either SEMA3F or SEMA3B in the HEY cell line resulted in suppression of tumorigenicity (Tse et al., 2002; Xiang et al., 2002). These genes map to 3p21.3, one of the candidate TSG regions identified by MMCT assays of the HEY cell line (Rimessi et al., 1994). However, the mechanism of inactivation of these genes or their role in ovarian cancer has not been determined. At least one study has shown decreased expression and a missense mutation in DRR1 (also known as TU3A) at 3p14.2 in at least one ovarian cancer cell line (Wang et al., 2000). Alteration of promoter methylation of RASSF1A at 3p21.31 has been shown in ovarian cancers (Agathanggelou et al., 2001; Teodoridis et al., 2005). Hence, as has been proposed for the pathogenesis of lung cancer where 3p deletions are frequent (Huebner, 2001; Xian et al., 2001; Dallol et al., 2002; Zabarovsky et al., 2002), a coordinate inactivation of a group of 3p genes by genetic (deletion and/or mutation) and epigenetic (promoter methylation and/or haploinsufficiency) means may be important for the pathogenesis of this disease.
We have applied a variation of the chromosome transfer technique to identify 3p regions important in suppressing tumorigenicity of ovarian cancer to complement and guide the positional cloning approach based on LOH analysis. The chromosome transfer method was based on the fusion of the spontaneously immortalized and tumorigenic OV90 ovarian cancer cell line, monoallelic for the 3p arm (Lounis et al., 1998; Provencher et al., 2000), with an irradiated mouse cell line that contained a normal human chromosome 3 (Speevak et al., 1995; Speevak and Chevrette, 1996). The OV90 cell line was also previously shown to exhibit decreased expression of a number of 3p genes by microarray analysis (Manderson et al., 2002a). We turned to this alternative method of gene transfer due to failed attempts to generate stable hybrids through MMCT of whole chromosome 3 (unpublished data). Here, we present functional evidence supporting the presence of 3p TSGs by this alternative method of chromosome 3 gene transfer. By transcriptome analysis we demonstrate that genes reportedly altered in their expression profile in ovarian cancers were affected in hybrids that had acquired normal chromosome 3 fragments.
Transfer of chromosome 3 fragments
Several attempts of MMCT assays of whole human chromosome 3 from the B78MC166 mouse cell containing a normal human chromosome 3, into OV90 neor, a derivative of OV90 cancer cell line, failed to generate stable hybrids (unpublished data). To overcome such limitations, we modified the gene transfer approach by fusing OV90 neor cells with B78MC166 cells that were irradiated with different doses of ionizing radiation in order to fragment normal chromosome 3. Using this approach, 12 ‘radiation-derived’ hybrids (RH) were generated where eight were propagated in culture. Hybrids RH-1, RH-3 and RH-7 were obtained from B78MC166 cells irradiated with 2000 rads, RH-5, RH-6 and RH-12 from 1500 rads, and RH-9 and RH-10 from 1000 rads. Fluorescent in situ hybridization (FISH) analysis revealed that RH-9 and RH-12 contained a large number of mouse chromosomes with about five to 15 human chromosomes (Supplementary Figure S1). These hybrids could have been generated from a whole-cell fusion between the two parental cell lines and were not characterized further. In contrast, using a mouse painting probe, no mouse chromosomal elements were detected in RH-3, RH-6 and RH-10 (Supplementary Figure S1). These results were confirmed upon hybridization of all chromosomes with a whole human painting probe. A small proportion of metaphase chromosome spreads from RH-1 (2/38 spreads) and RH-7 (3/25 spreads) indicated they each harbored one small mouse chromosome (data not shown). These results are consistent with the transfer of a mouse chromosome during cell fusion, and its progressive loss with subsequent cell division. FISH analysis also indicated the presence of a translocated mouse/human chromosome in all metaphase spreads of RH-5.
To test for evidence of chromosome 3 transfer, the earliest passages of the hybrids were genotyped with polymorphic microsatellite repeat markers, D3S2426, D3S1611, D3S1300, D3S1600 and D3S1274. These markers confirmed the successful transfer chromosome 3 fragments in RH-3, RH-5, RH-6, RH-7 and RH-10 (Supplementary Figure S2). Using this limited number of markers, there was no evidence that RH-1 acquired a transferred fragment (data not shown). RH-5, RH-6 and RH-10 were passaged once every 4–7 days. RH-1, RH-3, RH-7 and OV90 neor were passaged every 2–3 days and thus were most similar to OV90, which has a doubling time of about 1.4 days (Provencher et al., 2000).
Cell morphology of the hybrids
The cell morphology of the earliest passages of the hybrids was compared with that of OV90 neor (Figure 1). RH-5, RH-6 and RH-10 cells were similar, exhibiting features of elongated and flattened cells. In contrast, RH-3 and RH-7 exhibited a mixture of cell morphologies such as small, rounded, refractile cells with ruffled membranes similar to OV90 neor, and larger, flattened, elongated cells similar to RH-5, RH-6 and RH-10.
Soft agarose, spheroid formation and tumorigenicity assays of the hybrids
The hybrids were tested for their ability to grow in anchorage independent conditions by assaying their ability to form colonies in soft agarose (Figure 1 and Table 1). OV90 neor was able to form large dense colonies throughout the agarose plate as previously demonstrated in a similar assay of the parental OV90 cell line (Provencher et al., 2000). RH-3 formed a variable number of colonies per experiment some of which were comparable to OV-90 neor, whereas RH-7 had approximately half the number of colonies as OV90 neor. In contrast, colony formation was not detectable in assays of RH-5, RH-6 and RH-10.
The capacity to form three-dimensional spheroids in culture was also investigated. Spheroid forming capacity was evident for OV90 neor, where the cells appeared in a tight cluster with clearly delineated margins (Figure 1). In contrast, this capacity was altered in RH-5, RH-6 and RH-10 where a tight margin of the three-dimensional structure was not evident. This capacity also appeared altered in the analysis of RH-3 and RH-7, which exhibited an intermediate phenotype showing maintenance in the aggregation of cells with a loose margin of cells at the periphery of the three-dimensional structure.
The tumorigenic potential of the hybrids was tested in nude mouse assays as OV90 has previously been shown to form tumors at both subcutaneous (s.c.) and intraperitoneal (i.p.) injection sites with accompanying ascites (Provencher et al., 2000) (Table 1). OV90 neor formed s.c. (with masses detectable at 7 days) and i.p. tumors accompanied by ascites formation in all mice assayed. Hence, the tumor forming capacity of OV90 neor was similar to the OV90 parental cell line (Provencher et al., 2000). In stark contrast, tumor formation was not observed in any of the assays with RH-5, RH-6 or RH-10. Although with RH-3 and RH-7 tumor formation was accompanied with ascites at both s.c. and i.p. injection sites, the capacity to do so varied from OV90 neor. Most notable was that not all mice formed tumors and in some cases s.c. masses detectable with the same latency period as that of OV90 neor were not detectable at the conclusion of the assay. Interestingly, the 0.6 cm s.c. tumor mass measured in RH-3 assays was detected as a 0.8 cm mass at day 49 and was reduced to a 0.6 cm mass at day 99; this tumor remained similar in size on the day of killing (180 days). In another s.c. RH-3 assay, a tumor mass was detectable at seven days post injection, which measured about 0.7 cm in diameter at day 56 but was no longer detectable at day 86 or at killing. Similar observations were made of s.c. assays of RH-7, where all mice had palpable masses at day 7 but only two had a tumor mass at killing. In addition, masses of about 0.6 cm in diameter were no longer detectable at day 14 (two cases) and day 54 (two cases) postinjection.
Chromosome 3 genotype of hybrids
Polymorphic microsatellite repeat markers were used to characterize the chromosome 3 content in the hybrids (Figure 2 and Supplementary Figure S3). Chromosome 3 is about 200 Mb with proportionate 3p and 3q arm lengths. There were 78 of 121 markers tested that enabled the distinction of endogenous OV90 from transferred chromosome 3 alleles, where 54 markers represented 3p loci and 24 markers represented 3q loci. The informative markers were located within 0.5–5 Mb intervals on 3p and genotyping assays suggested that all hybrids had acquired more than one 3p fragment in the transfer experiments. RH-6 and RH-7 acquired a fragment containing the marker D3S1300, which was not detectable in OV90 neor and RH-3, RH-5 and RH-10 (Supplementary Figure S2). This marker is located within intron 5 of FHIT. We have previously shown that OV90 harbors a deletion within intron 5, including D3S1300 locus, which does not affect transcription of this gene (Manning et al., 1999). With the exception of D3S2448, the markers tested for the 3p arm were monoallelic for OV90 neor. D3S2448 is located immediately proximal to the centromere. These results reaffirm initial LOH studies showing a reduction to hemizygosity of the majority of the 3p arm in OV90 (Lounis et al., 1998). Due to heterozygosity of 3q in OV90, identifying informative 3q markers was more challenging. However, genotyping suggested RH-3, RH-5, RH-7 and RH-10 (but not RH-6) also had acquired normal 3q fragments, which included markers flanking the centromeric region (Supplementary Figure S3). With the same panel of genetic markers, there was no evidence that RH-1 acquired fragments overlapping any of the tested regions (data not shown).
Autoradiograms from genotyping assays suggested the majority of introduced alleles were present in copy numbers similar to endogenous OV90 neor chromosome 3 alleles, in all hybrids with the notable exception of RH-7 (Figure 2). For example, D3S1600 exhibited allelic imbalance reminiscent of LOH in RH-7, implying that loss of this allele occurred in later passages (Supplementary Figure S2). A number of markers exhibited this feature. Moreover, allelic imbalance was only observed for the transferred normal allele.
Global analysis of gene expression of hybrids
Expression microarray analysis of the hybrids was performed to assess the effect of normal chromosome 3 fragments on the overall transcriptome relative to the parental OV90 cell line. Expression values generated by normalized MAS5 data sets of OV90 neor and each of the hybrids (RH-3, RH-5, RH-6, RH-7 and RH-10) were highly correlated (>91%) based on all possible two-way comparative analyses of the entire data set (54 613 probe sets). The non-tumorigenic hybrids, RH-5, RH-6 and RH-10, exhibited the least similarity (91–95%) in all two-way comparative analyses with OV90 neor, RH-3 and RH-7. RH-3 and RH-7 exhibited the highest similarity in two-way comparative analyses with each other and OV90 neor (95–97%). Thus, although the hybrids varied significantly in their tumorigenic potential, these results imply that modest differences in their transcriptome had occurred.
The expression analysis was also investigated using an alternative statistical method based on R and Bioconductor to further assess the similarity in expression profiles. Clustering analysis was performed using probe sets that passed a nonspecific filtering step (5932 probe sets). The dissimilarity matrix corresponding to the final level of the hopach hierarchical tree, revealed two distinct groups (Supplementary Figure S4), which appeared to relate to tumorigenic potential, as OV90 neor, RH-3 and RH-7 clustered in one group, and RH-5, RH-6 and RH-10, the hybrids exhibiting complete suppression of tumorigenecity, in another. The relevance of the alterated transcriptome in RH-5, RH-6 and RH-10 was assessed by identifying those genes that were uniquely and similarly differentially expressed relative to OV90 neor. These genes were identified by individual two-way comparative analyses of OV90 neor and each hybrid using the MAS5 microarray expression profiles that were normalized, rescaled, and contained only probe sets exhibiting at least one high reliability score or P call (n=30 726 probe sets). This analysis revealed 511 (1.7%) differentially expressed probe sets exhibiting at least a three-fold difference in RH-5, RH-6 and RH-10, but less than a three-fold difference in RH-3 and RH-7. These probe sets exhibited differences consistent with upregulation (n=236 probe sets or 46%) or downregulation (n=275 probe sets or 54%) of gene expression in RH-5, RH-6 and RH-10 relative to OV90 neor. The 511 probe sets represented 409 known/hypothetical genes or ESTs. To begin to assess the biological relevance of these differentially expressed genes, we performed a literature search on the genes/ESTs represented by 324 of the 409 probe sets that exhibited expression values of at least 100 in any of the hybrids or OV90 neor. This cutoff was chosen in order to not overestimate the significance of differences that occur in the low range of gene expression. The 324 probe sets represented 263 known/hypothetical genes or ESTs that were either upregulated (n=156 probe sets or 48%) or downregulated (n=168 probe sets or 52%) in RH-5, RH-6 and RH-10 relative to OV90 neor, RH-3 and RH-7. A literature review revealed 35 genes represented by these probe sets were implicated in ovarian cancer and some examples are shown in Table 2 (Supplementary Table S1). Reverse transcriptase–polymerase chain reaction (RT–PCR) analysis of VEGF and SPARC (featured in Table 2) and NEDD9 (implicated in carcinogenesis but not previously studied in ovarian cancer) was comparable to expression patterns generated by microarray analysis (Figure 3).
A statistical analysis of the raw expression data by Bioconductor methods was also used to compare OV90 neor, RH-3 and RH-7 with RH-5, RH-6 and RH-10. The samples were similarly grouped based on their capacity to form tumors. Using this approach, 283 probe sets were identified (5% false discovery rate) representing expression data that differed between the two groups. The 283 probe sets represented 224 known/hypothetical genes and ESTs. A comparison with differentially expressed probe sets using fold-differences in two-way comparison analysis with one value greater than 100 as described above, revealed 168 probe sets in common using the two methods of analyses of expression data sets. These 168 probe sets represent 134 known/hypothetical genes and ESTs, and included known ovarian cancer genes such as DAB2, VEGF and SPARC (refer to Table 2).
A scan of the MAS5 generated expression data sets of RH-3 and RH-7 revealed expression values comparable to those exhibited by RH-5, RH-6 and RH-10 or intermediate between OV90 neor and the other hybrids (Table 2). These observations implied that RH-3 and RH-7 express genes consistent with alteration in their tumorigenic potential. To further explore this notion, we re-examined the two-way comparison data sets selecting for three-fold differences in OV90 neor versus each of RH-5, RH-6 and RH-10 without regard for RH-3 and RH-7. Using this approach 1204 probe sets were identified, an increase of 693 from the previous analysis. Of these 1204 probe sets, 537 exhibited patterns of expression in RH-3 similar to those found in RH-5, RH-6 and RH-10, as did 412 for RH-7. This variation of the two-way analysis also revealed an additional 24 genes that have been described in ovarian cancer (e.g. refer to Table 2). Interestingly, a subset of the data exhibited similar three-fold differences consistent with upregulation (n=107 probe sets) or downregulation (n=149 probe sets) of gene expression in all of the hybrids relative to OV90 neor. These probe sets represent 214 known/hypothetical genes and ESTs differentially expressed at least three-fold in RH-5, RH-6 and RH-10. A literature search of some of these genes also revealed that they represented genes implicated in apoptosis or cell adhesion pathways involved in ovarian cancer, such as PTPN13, TDGF1 and KITLG.
Chromosome 3 expressed genes in hybrids
To begin to identify chromosome 3 genes modified in expression in comparison to OV90 neor, we extracted data sets with three-fold differences in gene expression in RH-5, RH-6 and RH-10. We restricted the analysis to these hybrids as both the tumorigenicity assays and transcriptome analysis of RH-3 and RH-7 highlighted their intermediate phenotype and heterogeneity. The microarray contained 3202 probe sets associated with chromosome 3 genes/ESTs. The expression values of 811-3p and 1088-3q probe sets had at least one P call in all samples tested. In two-way comparison analyses with OV90 neor, there were 22-3p probe sets which exhibited at least a three-fold difference in expression in these hybrids (Supplementary Table S2), representing nine upregulated and seven downregulated genes (Table 3). As exemplified by global analyses of gene expression, some of the 3p genes showed similar patterns of expression in all hybrids, such as HRH1, COLA7A1 and TDGF1. In addition, there were 65–3q probe sets that represented 12 upregulated and 33 downregulated genes or ESTs (Supplementary Table S2). Some of the 3q upregulated genes, such as MINA, BOC, GAP43, FSTL1 and MYLK, map to regions that appeared to have been transferred in some of the hybrids that have acquired normal 3q fragments (Supplementary Figure S3). As RH-6 also showed increased expression of these genes, these genes were unlikely modified in their expression due to gene transfer. Interestingly, a number of downregulated 3q genes, such as CP, EVI-1 and SKIL, were those shown overexpressed or implicated in ovarian cancer in independent studies (see Table 2 and Supplementary Table). It is apparent from these results that epigenetic modification of chromosome 3 genes has occurred as some of these differentially expressed genes overlap regions that appeared not to have been transferred in any of hybrids.
Of particular interest would be those 3p genes with increased expression in the hybrids that were also associated with transferred regions, as these genes could be suspected TSG candidates for further analysis. There were four genes, COL7A1, IL17RD, PRICKLE2 and VGL-3, which exhibited increased expression (at least three-fold) in the hybrids relative to OV90 neor that overlapped transferred chromosomal regions (Figure 2 and Table 3). However, only VGL-3 was upregulated greater than three-fold in RH-5, RH-6 and RH-10. This gene maps to a region defined by polymorphic markers D3S3581 and D3S2448 at 3p12–p11 that was transferred into all hybrids (Figure 2). Although the expression values were towards the low range of expression, a high reliability score was observed in expression values of these hybrids relative to OV90 neor (Table 3). RT–PCR analysis of VGL-3 was consistent with microarray expression results, where the highest value of expression was observed for RH-5 (Figure 3). Notable was that expression of VGL-3 was not detectable by RT–PCR analysis of OV90 neor, whereas it was detectable in all hybrids, including RH-3 and RH-7.
To further explore the expression patterns of genes that map within the interval shared in common by RH-5, RH-6 and RH-10, expression profiles of the probe sets that mapped within markers D3S3581 and D3S2448 were extracted. Based on the UCSC Human Genome Browser 12 known/hypothetical genes, including VGL-3, mapped within this interval, and probe sets representing these genes were identified in the microarray expression data set (Table 4). Microarray analysis indicated that ROBO1 was expressed in OV90 neor and all hybrids, but downregulated at least four-fold in RH-10. RT–PCR analysis of ROBO1 was comparable to the expression pattern generated by microarray analysis, where it was weakly detectable in RH-10 (Figure 3). Although differential expression was observed for other candidates, where some of the hybrids exhibited three-fold differences in gene expression relative to OV90 neor, there were no striking examples of differentially expressed genes comparable to that observed with VGL-3, even without regard to fold-cutoff in gene expression.
The transfer of normal chromosome 3 fragments into OV90, a tumorigenic ovarian cancer cell line, resulted in suppression of tumorigenicity. While only five hybrids were generated through the transfer of chromosome 3 fragments, our results are similar to that of whole chromosome 3 transfer in the HEY ovarian cancer cell line, in which only four hybrids were generated in three successive MMCT experiments (Rimessi et al., 1994). This independent study is the only known report of chromosome 3 transfer in an ovarian cancer cell line. Our assay differs from that of MMCT as normal chromosome 3 fragments were ‘transferred’ to the OV90 cell line through cell fusion with an irradiated mouse cell line containing a normal human chromosome 3 homolog. A major strength of the MMCT approach has been the ability to generate revertant hybrids, hybrids that have lost the introduced chromosome through growth in vivo tumorigenicity assays (consequently in the absence of a selectable drug to maintain the introduced chromosome) with the re-emersion of the tumorigenic phenotype (Doherty and Fisher, 2003). This poses a challenge for TSG discovery if complete loss of the introduced chromosome has occurred. However, some of these revertant hybrids have been shown to retain rearranged chromosomes and the deleted regions have been inferred to represent candidate TSG loci. While the generation of revertant hybrids may be compromised through the integration of the transferred fragments in our hybrids, our hybrids may be more amenable to long-term manipulation. Moreover, in contrast to the MMCT approach, we have associated the retention of normal 3p fragments in our hybrids with alteration of tumorigenic phenotype. Thus, the acquired normal chromosome 3 fragments are suspected candidate TSG containing regions purported to harbor TSGs affecting tumorigenic potential in OV90 through functional complementation of a dysregulated 3p gene in the host cells. Our failure to generate stable OV90 hybrids through MMCT assays of whole chromosome 3 transfer cannot be explained at the present time. As mentioned above, the MMCT studies with the HEY ovarian cancer cell line also resulted in a low frequency of hybrids (Rimessi et al., 1994). However, OV90 is hemizygous for the majority of the 3p arm, and it is notable that the RH-7 hybrid, which appears to have acquired the largest number of normal 3p alleles, was the only hybrid exhibiting extensive allelic imbalance where loss of the normal allele appeared to have occurred. Combined with our chromosome 3 fragment transfer experiments, whole chromosome 3 transfer may have exerted a strong suppressive effect on the growth of OV90 and may have contributed to the failure of generating stable hybrids.
Microarray analysis identified non-chromosome 3 genes altered in their expression, as one would predict based on a review of the literature of ovarian cancer. Since these genes were similarly differentially expressed in RH-5, RH-6 and RH-10, common molecular pathways may have been invoked in the complete suppression of tumorigenic potential. Some interesting genes previously implicated in ovarian tumorigenesis include VEGF, SPARC and DAB2. VEGF expression was associated with ascites formation in nude mice and consequently introduction of antisense VEGF into an ovarian cancer cell line that expressed this gene resulted in inhibition of ascites formation (Yoneda et al., 1998; Akutagawa et al., 2002). Overexpression of VEGF in ovarian cancer is associated with poor prognosis (Paley et al., 1997). Modulation of SPARC or DAB2 expression by transfection of tumorigenic ovarian cancer cell lines, with cDNA expressing constructs, resulted in a reduction of tumorigenic potential in nude mice assays (Mok et al., 1996, 1998; Sheng et al., 2000). Cell transfection assays of SPARC have also shown induction of apoptosis in an ovarian cancer cell line (Yiu et al., 2001). Some differentially expressed genes, such as SPARC, SLPI, VEGF, CXCL12 and BOK, are involved in molecular pathways strongly implicated in cancer such as cell growth, tumorigenesis, angiogenesis and/or apoptosis (Davidson et al., 2003; Hicklin and Ellis, 2005). There are also examples of genes that appear to play roles in similar molecular pathways, such as, VEGF, a gene involved in angiogenesis, which has been shown regulated by a variety of growth factors, signaling pathways, transcription factors, oncogenes and TSGs (McMahon, 2000; Hicklin and Ellis, 2005). Specifically, there is evidence of SPARC inhibiting the action of VEGF in vascular endothelial cells (Kupprion et al., 1998). These results support the notion that chromosome 3 transfer correlates with a concerted re-programming of the cellular transcriptome in the non-tumorigenic hybrids. These epigenetic events point to molecular networks that are characteristics of ovarian carcinogenesis and whose modulation may depend on key underlying events in ovarian cancer.
All hybrids acquired normal chromosome 3 alleles that overlapped at least one of the TSG containing regions, 3p25–p26, 3p24 and 3p14-pcen, proposed in our previous LOH analyses of epithelial ovarian cancer samples (Lounis et al., 1998). Interestingly, genes shown previously as candidate TSGs in ovarian cancer studies such as SEMA3B (Tse et al., 2002), SEMA3F (Xiang et al., 2002), RASSF1A (Agathanggelou et al., 2001) and DRR1 at 3p14.2 (Wang et al., 2000), map outside of regions transferred in all of the hybrids. While transfer of 3q fragments also occurred in some hybrids, no transferred 3q fragments were detectable in RH-6. As OV90 is monoallelic for the majority of chromosome 3p, we propose that the transfer of a 3p fragment(s) resulted in the complementation of a TSG(s) abrogated in OV90 resulting in suppression of tumorigenic potential. We propose that at least one of the TSG candidates is contained within a transferred 3p fragment that was acquired in common by all of the hybrids or was acquired in common in the hybrids that exhibited the strongest suppression in tumorigenic potential. The similarity in growth characteristics of RH-5, RH-6 and RH-10 and global patterns of gene expression by microarray analysis, suggested that complete suppression of tumorigenic potential was accompanied by alteration of similar molecular pathways. Moreover, the intermediate tumorigenic potential phenotype exhibited by RH-3 and RH-7, which also exhibited a diminished capacity to form colonies in soft agarose assays and three-dimensional spheroids in culture was mirrored in the global patterns of gene expression by microarray analysis. Hence, the intermediate phenotype exhibited by these hybrids could reflect cellular and/or genetic heterogeneity. This notion is further supported by the possible regression of tumor formation observed in tumorigenicity assays of RH-3 and RH-7.
The 3p12-pcen interval acquired by all hybrids overlaps a putative TSG containing interval reported in independent functional complementation assays, such as the nonpapillary renal cell-1 (NRC-1) locus deduced from MMCT studies of renal cell carcinomas cell lines (Sanchez et al., 1994; Lott et al., 1998) and an independently derived TSG containing interval in MMCT studies of a breast cancer cell line (Cuthbert et al., 1999). Furthermore, the studies involving renal cell carcinoma cell lines also showed strong tumor suppressor activity in MMCT experiments with a defined centric chromosome 3, which overlaps 3p12-pcen (Lovell et al., 1999). Indeed, in this later study, the authors noted that only three hybrids were obtained with KRC-7, a non-papillary renal cell carcinoma cell line. The low frequency of hybrids generated are reminiscent of our chromosome 3 transfer experiment results with OV90 and those independently performed with the HEY ovarian cancer cell line (Rimessi et al., 1994). In addition, latency in tumor formation in renal cell carcinoma derived hybrids was observed and some tumors harbored deletions of the introduced normal 3p12-pcen interval consistent with loss of normal functioning TSG located within this interval (Lovell et al., 1999). Taken together, these studies identify 3p12-pcen as a potential TSG containing region. Furthermore, the present study determined that this was the only region transferred in common to all hybrids, thus strongly implicating 3p12-pcen as a candidate TSG region in OV90.
The candidate 3p12-pcen TSG region is defined by polymorphic markers D3S3581 and D3S2248, and is approximately 16 Mb in size based on a recent draft of the UCSC Human Genome Browser. This region contains 12 known/hypothetical genes, which have not been reported previously in the context of ovarian tumorigenesis. However, ROBO1, EPHA3 and POU1F1 have been implicated in molecular studies of other cancer types. ROBO1 is the only known gene that maps within the 3p12-pcen interval shown to exhibit TSG activity in independent studies. ROBO1 was shown inactivated by promoter methylation but not by somatic mutation in a subset of breast, lung and renal cell carcinomas (Dallol et al., 2002). TSG activity was demonstrated in a hemizygous knockout mouse model exhibiting susceptibility to lung carcinomas with accompanying modification in expression of the retained allele by methylation of ROBO1 in the resulting tumors (Xian et al., 2001). These findings suggest that ROBO1 exhibits a ‘neoclassical’ TSG phenotype rather than a classical TSG phenotype exemplified by inactivation of gene function due to somatic mutation (Huebner, 2001). POUIF1 is a pituitary-specific transcription factor implicated in prolactinomas (Li et al., 1990; Lacerte et al., 2004). The tissue specificity of POU1F1 combined with the absence of detectable expression by microarray analyses in OV90 and the hybrids would not support its TSG candidacy. EPHA3 is also an unlikely TSG candidate as it belongs to the ephrin receptor subfamily of the protein-tyrosine kinase family of regulators of signaling pathways shown overexpressed in a number of cancers (Blume-Jensen and Hunter, 2001; Adams et al., 2002; Cheng et al., 2002, 2003). Recently, somatic mutations in EPHA3 predicted to affect the tyrosine kinase domain has been reported in two of 147 (1.4%) colorectal cancers (Bardelli et al., 2003). Although the effect of these mutations on kinase function has not been demonstrated, constitutive activation of tyrosine kinase activity has been reported in other family members in association with affecting various cellular processes important in tumorigenesis such as differentiation, transcription, cell cycle progression, apoptosis, motility and invasion as they appear to play a central role in regulating these pathways (Blume-Jensen and Hunter, 2001). The purported function of EPHA3 combined with preliminary data indicating absence of gene expression in OV90 and the hybrids would not be consistent with TSG activity. The remaining candidates that map within the region (see Table 4) have not directly been implicated in cancer. However, VGL-3 is an interesting candidate based on detectable expression in the hybrids and absence of detectable expression in OV90. This recently annotated gene is one of three vestigal-like genes so-named because it is related to the Drosophila protein Vestigal and like other members of this family contains a consensus TEF-1 and MEF2 (TDU) interaction domain (Maeda et al., 2002). The overall low level of VGL-3 expression is consistent with independent reports showing abundant expression (which contains an unusually large noncoding 3′ UTR) in placenta but barely detectable expression in other human tissues (Maeda et al., 2002). While some of the members of the VGL family, such as VGL-2 and VGL-4, have been implicated in skeletal and cardiac muscle differentiation (Maeda et al., 2002; Chen et al., 2004), function of VGL-3 is not known. Although expression analysis of all the 3p12-pcen genes warrants further confirmation using alternative methods, differential expression of VGL-3 in the hybrids places this gene as a top TSG candidate for further studies.
The putative 3p12-pcen TSG may function independently of TP53 and CDKN2A, as OV90 harbors somatic mutations in these known TSGs (Provencher et al., 2000). In part this is consistent with our previous study showing that LOH of any 3p region was independent of TP53 somatic mutation status in ovarian cancer samples (Manderson et al., 2002b). In renal cell carcinomas, NRC-1 may function independently of VHL another known 3p TSG, and has been proposed to play a role in regulating angiogenesis and apoptosis (Lovell et al., 1999). Further support for this observation comes from the analysis of global gene expression patterns in the hybrids, which identified genes implicated in various molecular pathways, including those that affect cell growth, tumorigenesis, angiogenesis and apoptosis. The further molecular genetic characterization of the novel OV90 hybrids generated in the present study, particularly those exhibiting complete suppression of tumorigenic potential, afford the opportunity of elucidating the underlying TSG at 3p12-pcen as well as pathways affected in ovarian tumorigenesis.
Materials and methods
Cell lines and culture conditions
A neomycin clone of OV90, a previously described ovarian cancer cell line (Provencher et al., 2000), was generated to select hybrids in cell fusion assays. OV90 was derived from the ovarian malignant ascites from a chemotherapy naïve patient with a grade 3 and stage IIIc adenocarcinoma (Provencher et al., 2000). OV90 exhibits LOH of 3p arm (Lounis et al., 1998), contains somatic mutations of TP53 and CDKN2A (Provencher et al., 2000), and is tumorigenic in nude mice xenografts with ascites formation (Provencher et al., 2000). The aminoglycoside phosphotransferase 3′ (II) gene (in pSV2NEO), conferring resistance to Geneticin® (Canadian Life Technologies, Burlington, ON, Canada), was transfected into OV90 using lipofectAMINE PLUS™ reagent (Canadian Life Technologies), according to the manufacturer's instructions. Neor clones were selected in the presence of 400 μg/ml of Geneticin®. OV90 neor was maintained in ovarian surface epithelium (OSE) complete medium consisting of 50:50 medium 199:105 (Sigma, Oakville, ON, Canada) supplemented with 10% fetal bovine serum (FBS), 0.25 μg/ml of amphotericin B (Wisent, St. Bruno, QC, Canada), 50 μg/ml of gentamicin (Canadian Life Technologies) (Kruk et al., 1990) and 400 μg/ml of Geneticin®.
The B78MC166 mouse cell line was source of normal human chromosome 3 which is sensitive to Geneticin® and contains an intact human chromosome 3 tagged with a HyTK dual selectable marker that confers resistance to hygromycin B and sensitivity to ganciclovir (Speevak et al., 1995). B78MC166 is monochromosomal only for human chromosome 3 (Speevak and Chevrette, 1996) and is one of a series of B78 mouse cell lines containing various human chromosomes that was initially derived by the fusion of the B78 murine melanoma cell line and microcells generated from normal human fibroblasts (Speevak et al., 1995). B78MC166 was grown in Dulbecco's modified Eagle's medium-high glucose (Canadian Life Technologies, Burlington, Canada) supplemented with 10% FBS and 400 μg/ml of hygromycin B (Roche Diagnostics, Laval, QC, Canada).
Chromosome 3 transfer experiments
OV90 neor hybrids containing normal chromosome 3 fragments were generated by fusing OV90 neor cells with irradiated B78MC166 cells to fragment the entire genome. The B78MC166 cells were irradiated with 1000, 1500 or 2000 rads, using a Climax 6EX-SN643 apparatus (generating X-Rays). These irradiated cells were then fused to OV90 neor cells in the presence of either 48% or 50% polyethylene glycol 1600 (Fisher Scientific, Nepean, ON, Canada) for 60 s. The OV90 neor-derived hybrids were selected and maintained in the presence of 400 μg/ml of Geneticin® and 400 μg/ml of hygromycin B in OSE complete medium as described above (Lounis et al., 1994).
The hybrids were assessed for human chromosome 3, and mouse and human chromosome content by FISH analysis. A whole human chromosome-painting probe was generated upon Alu-PCR amplification (Dorin et al., 1992) and fluorescein isothiocyanate (FITC) labeling of DNA from PA-1, a diploid ovarian teratocarcinoma cell line harboring a balanced translocation between chromosomes 15 and 20 (Zeuthen et al., 1980). A whole mouse-painting probe was obtained upon direct FITC labeling of B78 DNA digested with EcoRI, BglII and BamHI. Chromosome 3 content was analysed using a human chromosome 3 painting probe (Cambio Biotin Labelled Chromosome 3 Paint; Cedarlane Laboratories Limited, Hornby, ON, Canada). FISH analyses were performed according to the manufacturer's instructions using Detection Reagents B1 (Cambio FITC Avidin; Cedarlane Laboratories Limited) and B4 (Cambio Biotinylated goat anti-avidin; Cedarlane Laboratories Ltd, ON, Canada).
Soft agarose, spheroid growth and tumorigenicity assays
The hybrids were assayed for their ability to grow in anchorage independent conditions by culturing 1 × 104 cells in 0.33 g per 100 ml agarose in OSE complete medium (Lounis et al., 1994) on a layer of 0.66 g per 100 ml agarose prepared with OSE complete medium. Three replicates of OV90 neor and each hybrid were performed. After 3 weeks, for each replicate, colonies were visualized and scored using Image Pro Plus® Software.
The hybrids were tested for their capacity to form three-dimensional aggregates or spheroids as described (Kelm et al., 2003). Briefly, 64 droplets each containing about 4000 cells in OSE complete medium were plated separately on non-coated plastic tissue culture plates. After a 4-day incubation period cell dispersion of the hybrids were compared with spheroid-forming OV90 neor.
The tumorigenic potential of the hybrids was assessed based on their ability to form tumors in 45-day-old female athymic nude mice (Swiss Nu/Nu CD1) at s.c. left gluteal (n=6 mice) and i.p. injection sites (n=6 mice). Each mouse was injected at either site with 2 × 106 cells suspended in phosphate-buffered saline. Animals were housed under clean conditions in a laminar flow environment with ad-lib access to food and water. Tumor formation was measured over 180 days. Animals were killed before neoplastic masses reached greater than 1.5 cm or after formation of ascites fluid, according to the guidelines of the Canadian Council on Animal Care.
DNA was extracted as described previously (Lounis et al., 1994). To extract DNA from a small number of cells (100–1000) generated from the earliest passages of MMCT derived hybrids, cells were trypsinized and incubated overnight in 50 μl of proteinase K digestion buffer consisting of 1 mg/ml proteinase K and 1 gram per100 ml of Tween-20 in TE buffer (pH 8.0), prior to PCR amplification.
Chromosome 3 content was assessed by PCR amplification of chromosome 3 microsatellite repeats as described previously (Lounis et al., 1998). Hybrids were scored for the presence or absence of alleles and intensity of signal in autoradiograms by comparison with OV90 neor and B78MC166. All primer sequences for polymorphic markers are available at The Genome Database (www.gdb.org).
Total RNA was extracted with TRIzol reagent (Gibco/BRL, Life Technologies Inc., Grand Island, NY, USA) using the recommended protocol. RNA was extracted from OV90 neor, and hybrids were grown to 80% confluency in 100 mm plastic dishes.
Microarray analysis of expression profiles
The quality of RNA for microarray analysis was assessed by gel electrophoresis and 2100 Bioanalyzer analysis using the RNA 6000 Nano LabChip kit (Agilent Technologies, Germany). Biotinylated hybridization target was prepared from total RNA as described (Tamayo et al., 1999). Expression data were collected using the Human Genome U133 Plus 2.0 DNA oligonucleotide microarray (Affymetrix®, Santa Clara, CA, USA). Hybridization and scanning were performed once per sample at the McGill University and Genome Quebec Innovation Centre. Detailed protocols are available at www.genomequebec.mcgill.ca. Gene expression levels were calculated for each probe set from the scanned image by Affymetrix® GeneChip (MAS5) and prepared as previously described (Novak et al., 2002; Presneau et al., 2003). As reproducibility of expression values is highly variable at low values of expression (Arcand et al., 2004), all normalized values below five were reassigned a threshold value of five based on the mean expression value of the lowest reliability scores (A calls). Only data from 23 887 of the 54 613 probes sets that contained at least one high reliability score (P call) across all samples tested were analysed. Differentially expressed genes associated with tumorigenic potential were identified using two-way comparative analyses between OV90 neor and each hybrid.
Probe sets that represented genes or ESTs on chromosome 3 were extracted from the main expression data file with Extractor© 2001 Lypny and Tonin (Presneau et al., 2003), using UniGene Homo sapiens database, May 2005 based on UniGene Build 184 retrieved from the National Center for Biotechnology Information (www.ncbi.nlm.nih.gov/UniGene). Additional mapping information (physical distances and cytoband positions) of all chromosome 3 probe sets was retrieved from the University of California Santa Cruz (UCSC) Human Genome Browser database May 2004 (hg 17) assembly (genome.ucsc.edu). Based on these databases, there were 3202 probe sets representing genes that map to chromosome 3.
Global gene expression patterns from raw data sets were also examined using Bioconductor and statistical methods (Gentleman and Carey, 2003). Background subtraction, normalization and expression value calculations were performed using the justGCrma function, which converts raw data derived from CEL files into normalized (by quantile normalization) expression values using the robust multiarray average expression measure combined with probe sequence information (Irizarry et al., 2003). The genefilter package removed genes with insufficient expression variation across all samples tested. Expression values retained in the analyses required estimated intensities greater than 100 fluorescence units in at least one sample, and a log base 2 scale of at least 0.5 for the interquartile range across all tested samples. Clustering analysis was performed using the hopach package, which implements a hybrid clustering method of existing partitioning and agglomerative hierarchical clustering algorithms. Differentially expressed genes were obtained using the limma package, which estimates the fold change between two predefined sample groups (OV90 neor, RH-3 and RH-7; RH-5, RH-6 and RH-10) by fitting a linear model and uses an empirical Bayes method to moderate the standard errors of the estimated log-fold changes for each probe set. This approach takes into consideration the small sample size of the comparative groups used in the present study. The P-value from the resulting comparison was adjusted for multiple testing: the false discovery rate was set to 0.05 as described previously (Benjamini and Hochberg, 1995).
Reverse Transcriptase-PCR analysis was performed as previously described (Presneau et al., 2005) to assess expression of NEDD9, ROBO1, SPARC, VEGF and VGL-3 in OV90 neor and the hybrids. The primer sequences for NEDD9 [5′ IndexTermggtgcttttcaaaacaactgc/3′tcaatcaagtcagccatctcc], ROBO1 [5′IndexTermtcccttagtactgcacgcct/3′agtggaaaacaagccaaaca], SPARC [5′IndexTermccccaatcacactagcaaca/3′agtctcaaaaccccagctca] VEGF [IndexTermtgggcaacttgtatttgtgtg/3′tgcactagagacaaagacgtga] and VGL-3 [5′IndexTermctgcattgaaataccatgaactt/3′ccaaccccacaaaaacagaa] were designed based on the 3′ transcribed sequence of the test gene using the Primer3 design software (www.broad.mit.edu/cgi-bin/primer/primer3.cgi/primer3_www.cgi). GAPDH was used as a control and primer sequences have been reported elsewhere (Presneau et al., 2003).
Adams J, Huang P, Patrick D . (2002). Curr Opin Chem Biol 6: 486–492.
Agathanggelou A, Honorio S, Macartney DP, Martinez A, Dallol A, Rader J et al. (2001). Oncogene 20: 1509–1518.
Akutagawa N, Nishikawa A, Iwasaki M, Fujimoto T, Teramoto M, Kitajima Y et al. (2002). Jpn J Cancer Res 93: 644–651.
Arcand SL, Mes-Masson AM, Provencher D, Hudson TJ, Tonin PN . (2004). Mol Carcinog 41: 17–38.
Bardelli A, Parsons DW, Silliman N, Ptak J, Szabo S, Saha S et al. (2003). Science 300: 949.
Bayani J, Brenton JD, Macgregor PF, Beheshti B, Albert M, Nallainathan D et al. (2002). Cancer Res 62: 3466–3476.
Benjamini Y, Hochberg Y . (1995). J Roy Stat Soc Ser B 57: 289–300.
Blume-Jensen P, Hunter T . (2001). Nature 411: 355–365.
Chen HH, Mullett SJ, Stewart AF . (2004). J Biol Chem 279: 30800–30806.
Cheng N, Brantley D, Fang WB, Liu H, Fanslow W, Cerretti DP et al. (2003). Neoplasia 5: 445–456.
Cheng N, Brantley DM, Liu H, Lin Q, Enriquez M, Gale N et al. (2002). Mol Cancer Res 1: 2–11.
Cuthbert AP, Bond J, Trott DA, Gill S, Broni J, Marriott A et al. (1999). J Natl Cancer Inst 91: 37–45.
Dallol A, Forgacs E, Martinez A, Sekido Y, Walker R, Kishida T et al. (2002). Oncogene 21: 3020–3028.
Davidson B, Goldberg I, Gotlieb WH, Kopolovic J, Risberg B, Ben-Baruch G et al. (2003). Cancer Metastasis Rev 22: 103–115.
Dodson MK, Hartmann LC, Cliby WA, DeLacey KA, Keeney GL, Ritland SR et al. (1993). Cancer Res 53: 4456–4460.
Doherty AM, Fisher EM . (2003). Mamm Genome 14: 583–592.
Dorin JR, Emslie E, Hanratty D, Farrall M, Gosden J, Porteous DJ . (1992). Hum Mol Genet 1: 53–59.
Ehlen T, Dubeau L . (1990). Oncogene 5: 219–223.
Fullwood P, Marchini S, Rader JS, Martinez A, Macartney D, Broggini M et al. (1999). Cancer Res 59: 4662–4667.
Gentleman R, Carey V . (2003). Visualization and Annotation of Genomic Experiments. Springer: New York.
Hicklin DJ, Ellis LM . (2005). J Clin Oncol 23: 1011–1027.
Hoglund M, Gisselsson D, Hansen GB, Sall T, Mitelman F . (2003). Cancer Res 63: 3378–3385.
Hoglund M, Gisselsson D, Mandahl N, Johansson B, Mertens F, Mitelman F et al. (2001). Genes Chromosomes Cancer 31: 156–171.
Huebner K . (2001). Proc Natl Acad Sci USA 98: 14763–14765.
Irizarry RA, Bolstad BM, Collin F, Cope LM, Hobbs B, Speed TP . (2003). Nucleic Acids Res 31: e15.
Kelm JM, Timmins NE, Brown CJ, Fussenegger M, Nielsen LK . (2003). Biotechnol Bioeng 83: 173–180.
Kruk PA, Maines-Bandiera SL, Auersperg N . (1990). Lab Invest 63: 132–136.
Kupprion C, Motamed K, Sage EH . (1998). J Biol Chem 273: 29635–29640.
Lacerte A, Lee EH, Reynaud R, Canaff L, De Guise C, Devost D et al. (2004). Mol Endocrinol 18: 1558–1569.
Li S, Crenshaw III EB, Rawson EJ, Simmons DM, Swanson LW, Rosenfeld MG . (1990). Nature 347: 528–533.
Lott ST, Lovell M, Naylor SL, Killary AM . (1998). Cancer Res 58: 3533–3537.
Lounis H, Mes-Masson AM, Dion F, Bradley WE, Seymour RJ, Provencher D et al. (1998). Oncogene 17: 2359–2365.
Lounis H, Provencher D, Godbout C, Fink D, Milot MJ, Mes-Masson AM . (1994). Exp Cell Res 215: 303–309.
Lovell M, Lott ST, Wong P, El-Naggar A, Tucker S, Killary AM . (1999). Cancer Res 59: 2182–2189.
Maeda T, Chapman DL, Stewart AF . (2002). J Biol Chem 277: 48889–48898.
Manderson EN, Mes-Masson AM, Novak J, Lee PD, Provencher D, Hudson TJ et al. (2002a). Genome Res 12: 112–121.
Manderson EN, Presneau N, Provencher D, Mes-Masson AM, Tonin PN . (2002b). Mol Carcinog 34: 78–90.
Manning AP, Mes-Masson AM, Seymour RJ, Tetrault M, Provencher DM, Tonin PN . (1999). Mol Carcinog 24: 218–225.
McMahon G . (2000). Oncologist 5(Suppl 1): 3–10.
Mertens F, Johansson B, Hoglund M, Mitelman F . (1997). Cancer Res 57: 2765–2780.
Mok SC, Chan WY, Wong KK, Cheung KK, Lau CC, Ng SW et al. (1998). Oncogene 16: 2381–2387.
Mok SC, Chan WY, Wong KK, Muto MG, Berkowitz RS . (1996). Oncogene 12: 1895–1901.
Novak JP, Sladek R, Hudson TJ . (2002). Genomics 79: 104–113.
Paley PJ, Staskus KA, Gebhard K, Mohanraj D, Twiggs LB, Carson LF et al. (1997). Cancer 80: 98–106.
Pejovic T, Heim S, Mandahl N, Baldetorp B, Elmfors B, Floderus UM et al. (1992). Genes Chromosomes Cancer 4: 58–68.
Petursdottir TE, Thorsteinsdottir U, Jonasson JG, Moller PH, Huiping C, Bjornsson J et al. (2004). Genes Chromosomes Cancer 41: 232–242.
Presneau N, Dewar K, Forgetta V, Provencher D, Mes-Masson AM, Tonin PN . (2005). Mol Carcinog 43: 141–154.
Presneau N, Mes-Masson AM, Ge B, Provencher D, Hudson TJ, Tonin PN . (2003). Oncogene 22: 1568–1579.
Provencher DM, Lounis H, Champoux L, Tetrault M, Manderson EN, Wang JC et al. (2000). In vitro Cell Dev Biol Anim 36: 357–361.
Rimessi P, Gualandi F, Morelli C, Trabanelli C, Wu Q, Possati L et al. (1994). Oncogene 9: 3467–3474.
Sanchez Y, el-Naggar A, Pathak S, Killary AM . (1994). Proc Natl Acad Sci USA 91: 3383–3387.
Sheng Z, Sun W, Smith E, Cohen C, Xu XX . (2000). Oncogene 19: 4847–4854.
Simon R, Desper R, Papadimitriou CH, Peng A, Alberts DS, Taetle R et al. (2000). Genes Chromosomes Cancer 28: 106–120.
Speevak MD, Berube NG, McGowan-Jordan IJ, Bisson C, Lupton SD, Chevrette M . (1995). Cytogenet Cell Genet 69: 63–65.
Speevak MD, Chevrette M . (1996). Mol Cell Biol 16: 2214–2225.
Taetle R, Aickin M, Panda L, Emerson J, Roe D, Thompson F et al. (1999a). Genes Chromosomes Cancer 25: 46–52.
Taetle R, Aickin M, Yang JM, Panda L, Emerson J, Roe D et al. (1999b). Genes Chromosomes Cancer 25: 290–300.
Tamayo P, Slonim D, Mesirov J, Zhu Q, Kitareewan S, Dmitrovsky E et al. (1999). Proc Natl Acad Sci USA 96: 2907–2912.
Teodoridis JM, Hall J, Marsh S, Kannall HD, Smyth C, Curto J et al. (2005). Cancer Res 65: 8961–8967.
Tse C, Xiang RH, Bracht T, Naylor SL . (2002). Cancer Res 62: 542–546.
Wang L, Darling J, Zhang JS, Liu W, Qian J, Bostwick D et al. (2000). Genes Chromosomes Cancer 27: 1–10.
Whang-Peng J, Knutsen T, Douglass EC, Chu E, Ozols RF, Hogan WM et al. (1984). Cancer Genet Cytogenet 11: 91–106.
Xian J, Clark KJ, Fordham R, Pannell R, Rabbitts TH, Rabbitts PH . (2001). Proc Natl Acad Sci USA 98: 15062–15066.
Xiang R, Davalos AR, Hensel CH, Zhou XJ, Tse C, Naylor SL . (2002). Cancer Res 62: 2637–2643.
Yiu GK, Chan WY, Ng SW, Chan PS, Cheung KK, Berkowitz RS et al. (2001). Am J Pathol 159: 609–622.
Yoneda J, Kuniyasu H, Crispens MA, Price JE, Bucana CD, Fidler IJ . (1998). J Natl Cancer Inst 90: 447–454.
Yonescu R, Currie JL, Hedrick L, Campbell J, Griffin CA . (1996). Cancer Genet Cytogenet 87: 167–171.
Zabarovsky ER, Lerman MI, Minna JD . (2002). Oncogene 21: 6915–6935.
Zeuthen J, Norgaard JO, Avner P, Fellous M, Wartiovaara J, Vaheri A et al. (1980). Int J Cancer 25: 19–32.
Zheng JP, Robinson WR, Ehlen T, Yu MC, Dubeau L . (1991). Cancer Res 51: 4045–4051.
We thank Suzanna Arcand, Henriette Gourdeau, Lise Portelance, Manon de Ladurantaye, Marise Roy, Philippe O. Gannon and Jean-Simon Diallo for technical assistance and helpful discussions. NALC is the recipient of a studentship from the Research Institute of the McGill University Health Centre. VO is a recipient of studentships from the Canadian Institutes of Health Research (CIHR) and Canderel fund of the Institut du Cancer de Montréal. ENM is a recipient of a studentship from the Natural Sciences and Engineering Research Council of Canada (NSERC), CIHR and the Fonds de Recherche en Santé du Québec (FRSQ). A-MM-M is a recipient of Chercheur National fellowship from the FRSQ. DMP is a Clinicien-Chercheur Senior from the FRSQ. This research was supported by a grant from the CIHR to PNT, A-M M-M, MC and DMP.
About this article
Cite this article
Cody, N., Ouellet, V., Manderson, E. et al. Transfer of chromosome 3 fragments suppresses tumorigenicity of an ovarian cancer cell line monoallelic for chromosome 3p. Oncogene 26, 618–632 (2007). https://doi.org/10.1038/sj.onc.1209821
- ovarian cancer
- chromosome 3p
- radiation hybrids
- tumor suppressor gene
Multicellular detachment generates metastatic spheroids during intra-abdominal dissemination in epithelial ovarian cancer
3 Biotech (2018)
Development Genes and Evolution (2016)
Low levels of IGFBP7 expression in high-grade serous ovarian carcinoma is associated with patient outcome
BMC Cancer (2015)
BMC Medical Genomics (2015)