Centrosome clustering and cyclin D1 gene amplification in double minutes are common events in chromosomal unstable bladder tumors

Malignant tumors typically arise from multiple events within the developing cancer cells. Genetic damage is a hallmark of malignant cells and plays a key role in both the initiation and the progression of tumorigenesis [1].

Bladder cancer, along with most solid tumors, is characterized by multiple numerical and structural chromosome aberrations which in general associate with progression [2, 3]. Amplification of 11q13 involving cyclin D1 gene (CCND1) is among the most common sites of gene amplification in T1-T2 high grade tumors [4‐6]. Cyclin D1 plays an important role in cell cycle, binds to cyclin dependent kinases (CDK4/6), and promotes phosphorylation of RB1, orchestrating progression through the G1 restriction point.

Gene amplification involving oncogenes, a common mechanism to overexpress cancer-related genes, might be present in cancer cells as double-minute chromosomes (DMs) or homogeneously staining regions (HSRs). DMs are circular extrachromosomal autonomously-replicating DNA fragments lacking a centromere. HSRs are amplified intrachromosomal sequences that may be located in the same region of the amplified gene or in another chromosomal region [7]. The 11q13 amplicon is generally located at the same chromosome region of the single-copy genes involved (CCND1, etc.) [8]; other amplifications, such as those involving MYCN in neuroblastomas, are inserted in several places in the genome other than chromosome 2, where MYCN gene is mapped [9, 10].

Numerical chromosome instability (CIN), which occurs very frequently in cancer cells [11], contributes to aneuploidy and plays a critical role in tumorigenesis as a key element of genomic instability [11, 12]. Chromosome missegregation resulting from the deregulation of the spindle checkpoint is thought to be a potential cause of CIN. However, the molecular basis of this causative relation remains largely unknown [13]. The centrosome, a major microtubule-organizing center in animal cells, plays a vital role during mitosis as a spindle pole, and is crucial for accurate chromosome segregation to daughter cells [14]. In previous studies, centrosome amplification, defined as an increase in the centrosome number, has been identified in many different tumors, including bladder cancer [15, 16]. In addition, centrosome amplification has been recently shown to initiate tumorigenesis in flies [17]. Several studies have demonstrated that centrosomal abnormalities and chromosome copy-number heterogeneity frequently co-exist in bladder tumor cells [18‐20]. More recently, Jin et al. [21] found that multipolar mitosis and anaphase bridges are common, often concurrent, mitotic abnormalities in urothelial carcinomas, both in vivo and in vitro. The same authors identified several types of chromosome segregation abnormalities, including telomere dysfunction, sister-chromatid non-disjunction, and supernumerary centrosomes in urothelial cancer cell lines. These studies strongly support the hypothesis that CIN is present in bladder carcinomas.

The aim of this study was to describe how CCND1 amplicons and chromosome 11 copy number heterogeneity represent in vivo features of chromosomal instability in superficial bladder carcinomas. To that end, 21 paraffin-embedded cancer tissue samples were analyzed using comparative genomic hybridization (CGH) and fluorescence in situ hybridization (FISH). In seeking a basis for the chromosomal heterogeneity, we investigated centrosome and mitotic spindle integrity by immunofluorescent staining. Our results demonstrate, for the first time, that CCND1 amplification in DM and HSR could co-exist in the same bladder tumor. A correlation between HSR fragmentation and the appearance of DMs, which were subsequently eliminated by micronuclei extrusion, was also observed. Interestingly, we found that those patients whose tumors showed CCND1 amplification in DMs had a significantly shorter overall survival rate. Finally, the correlation between chromosome instability and centrosome abnormalities showed that the coalescence of centrosomes into two functional spindle poles was common in unstable bladder tumors.

In the present study, the gene copy number variation analysis of CCND1 in formalin fixed paraffin embedded tissue sections revealed a complex and unprecedented pattern of cellular behavior in non-muscle invasive bladder tumors. Our results suggested that copy number changes of CCND1 could be used as a biomarker to detect chromosome instability in bladder cancer. Bladder tumors were classified according to the CIN index, and we have shown a positive correlation between high heterogeneity, centrosome abnormalities and CCND1 gene amplification.

A positive correlation between the level of chromosomal instability and the tumor grade was identified; this phenomenon was previously described in bladder cancer [16]. Focusing exclusively on chromosome 11, the present study classifies the majority of Ta tumors as stable (CIN-negative group). These results are in agreement with the 2004 WHO classification that distinguishes two different entities in non-muscle invasive bladder tumors: one (Ta low-grade G1/G2) genetically stable in which gene amplifications are rare, and the other (T1 high-grade) with a high degree of genetic instability including high level amplifications [24]. Despite the low number of samples analyzed, it is remarkable that our classification of tumors according to their CIN index included all T1G3 samples in the CIN-positive group. CIN-negative group was just composed of grade 1 and grade 2 tumors, being all but one Ta. CCND1 amplification was exclusively observed in CIN-positive samples, suggesting that CCND1 might be involved in the generation of centrosomal abnormalities [25]. Moreover, we showed amplification of CCND1 as DMs in three CIN-positive samples. To our knowledge, this is the first report in the literature to show genomic amplification of CCND1 as DMs in bladder tumors.

The differences between FISH and metaphase CGH results for the CIN-positive samples highlight intratumoral heterogeneity. CGH detected the dominant genomic alterations present in at least some 60% of the tumor population [26]; however, it did not detect either the alterations that appear in a small number of cells or ongoing chromosomal instability. The association of DMs, centrosome aberrations and intercellular CIN observed in this study may indicate that the CIN phenotype does not become the major clonal population in bladder cancer.

As expected, centrosome amplification is correlated with CIN. Approximately, 75% (9/12) of CIN-positive samples and none of the CIN-negative samples showed centrosomal abnormalities. During the analysis of metaphase figures, multipolar and pseudo-bipolar spindles were identified in some CIN-positive tumors with supernumerary centrosomes. The presence of extra centrosomes within tumor cells might be deleterious as multipolar mitosis may generate sufficient high levels of aneuploidy to compromise cell viability. Several cancer cell lines overcome this problem by clustering their extra centrosomes at the two poles of the spindle, thus ensuring bipolar chromosome segregation [27‐29]. This phenomenon was observed in some samples in the present study showing abnormal large centrosomes and bipolar spindles, what confirms that centrosome clustering occurs in bladder cancer. It is interesting to note that bipolar spindles were also observed in three CIN-positive samples displaying a string-like centrosome similar to what was observed by Pihan et al. [30] in malignant tumors and tumor-derived cell lines.

In addition, the high chromosomal instability observed in our samples with centrosome clustering suggests that other factors might cause chromosomal instability. In fact, chromosome lagging, defined as a delayed movement of one chromatide in anaphase, was observed when anaphasic cells were studied. This is consistent with studies carried out by Thompson and Compton [31] on human cell lines, where the authors identified defective kinetochore-spindle attachments leading to anaphase lagging as a cause of chromosome missegregation. Recently, Ganem et al. [32] demonstrated that extra centrosomes alone are sufficient to promote chromosome missegregation during bipolar cell division. According to these authors, cells passing a transient multipolar spindle intermediate accumulate merotelic kinetochore-spindle attachment errors before centrosome clustering and anaphase.

DNA sequence amplification is one of the hallmarks of genomic instability in cancer. The target genes driving the 11q13 amplicon have been extensively reported, and at least four cores of amplification have been established in breast cancer [33, 34]. However, the evolution of this amplicon in tumor cells remains unclear. The 11q13 amplicon is usually located in the same chromosomal region as the amplified target gene [7, 8, 35, 36]. In our study, HSRs were usually located in the same chromosomal region as the amplified target genes, thus strongly supporting the hypothesis that the 11q13 amplicon is of intrachromosomal origin [8]. The presence of dicentric chromosomes and anaphase bridges in cell populations undergoing amplification is consistent with the role of the breakage-fusion-bridge (BFB) cycle in explaining intrachromosomal amplifications [37]. In the present study, the HSR-bearing chromosome 11 was often observed to be involved in nucleoplasmatic bridges and dicentric chromosomes.

Nevertheless, several mechanisms for the genesis of extrachromosomal amplifications (i.e., DMs) have been proposed [38]. A yeast model system was used to demonstrate that hairpin-capped double-strand breaks occurring at the location of human Alu-quasipalindromes trigger both DM and HSR gene amplification. According to this model, the nature of the amplicons depends on the chromosomal location of the amplified gene relative to double-strand break formation [39]. Within our sample set, the co-existence of CCND1 amplification in DMs and HSR is noteworthy. In tumor samples with both types of amplification, metaphasic cells with both HSR and DMs were observed in the transition zone between the HSR and DM carrier cell sub-populations. These data demonstrate a striking correlation between the presence of DMs and the observed fragmentation of HSR, thus suggesting a possible mechanism for excising amplified sequences in HSRs, giving rise to DMs. A similar phenomenon has been described in human cell lines with dihydrofolate-resistance gene amplification [40, 41]. Our findings strongly suggest that the same mechanism operates in tumor cells in vivo. Moreover, as seen in patient U-443, fragmentation of the HSR might occur in a metastatic form, after remaining stable for a long period of time in the primary tumor.

All samples with DMs also exhibited micronuclei; approximately 50% of them were CCND1-positive. The fact that the CCND1 signal was not present in the remaining 50% of the micronuclei indicates that other genomic regions were being actively eliminated from these cells. These findings suggest that micronuclei extrusion could induce rapid and dramatic changes, not only in the CCND1 gene with DMs, but also in other acentric fragments or even affecting whole chromosome copy numbers, therefore exacerbating genomic instability. The removal of amplified CCND1 sequences by micronuclei extrusion in bladder tumor cells was consistent with results reported by Valent et al. [42] regarding DMs containing MYCN neuroblastoma. Furthermore, in the present study, some cells showed one to three copies of the CCND1 on chromosome 11, but with an adjacent CCND1-positive micronucleus, suggesting that in some cells the normal copy number for this gene is restored by DMs extrusion.

It is known that centrosome amplification is a source of CIN, as are chromosomal lagging and micronuclei formation. Centrosome clustering partially reduces chromosomal instability [28], and increases cell viability by avoiding multipolar mitosis. Our observations that centrosome clustering is a common feature of chromosomally unstable bladder tumors, and the appearance of new drugs that specifically target centrosome clustering, such as griseofulvin [43], highlights the importance of further studying the role of centrosome abnormalities in bladder cancer.

The present study describes the in vivo behavior of CCND1 amplification in chromosome unstable T1 bladder tumors. We also demonstrate that the coalescence of centrosomes into two functional spindle poles is a common feature of these tumors.

Our study is the first report in the literature regarding the simultaneous CCND1 amplification in DM and HSR in bladder cancer cells. Our findings suggest a striking correlation between HSR fragmentation and the appearance of DMs which subsequently are removed by micronuclei extrusion. Of interest, we found that only those patients whose tumors showed CCND1 amplification in DMs had a significantly shorter overall survival rate. Further studies in a larger sample size should be necessary in order to confirm our results.

Coalescence of supernumerary centrosomes was observed in 80% of the most unstable tumors, highlighting the importance of this phenomenon in bladder cancer.