Amplification and overexpression of E2F3 in human bladder cancer

Abstract

We demonstrate that, in human bladder cancer, amplification of the E2F3 gene, located at 6p22, is associated with overexpression of its encoded mRNA transcripts and high levels of expression of E2F3 protein. Immunohistochemical analyses of E2F3 protein levels have established that around one-third (33/101) of primary transitional cell carcinomas of the bladder overexpress nuclear E2F3 protein, with the proportion of tumours containing overexpressed nuclear E2F3 increasing with tumour stage and grade. When considered together with the established role of E2F3 in cell cycle progression, these results suggest that the E2F3 gene represents a candidate bladder cancer oncogene that is activated by DNA amplification and overexpression.

Main

Urological carcinomas of the bladder, common in Western countries, can be classified into two categories based on their histopathology and clinical behaviour. Around 70–80% of transitional cell carcinomas of the bladder present as superficial non-muscle invasive papillary carcinomas (pTa or pT1) that are associated with a high risk of recurrence (70%) following treatment, but low risk (10–20%) of progression to muscle invasion. The remaining 20–30% of bladder cancers show muscle invasion at the time of diagnosis (>T2), have no association with papillary tumours, and are thought to arise from carcinomas in situ (CIS).

A variety of genetic changes have been identified in human bladder cancer. The alterations include inactivation of the RB1, TP53 and INK4A/ARF genes, amplification and overexpression of MDM2, CCND1/CyclinD1 and ERBB2, and mutational activation of H-RAS and FGFR3 (Cappellen et al., 1999; Knowles, 2001). Comparative genomic hybridization (CGH) and loss of heterozygosity studies have also identified many consistent regions of chromosomal gain and loss that may be sites of additional oncogenes and tumour-suppressor genes in bladder cancer (Knowles, 2001). In these studies, gains and amplification at 6p22 have been frequently found in human bladder cancer (Hovey et al., 1998; Koo et al., 1999; Terracciano et al., 1999; Simon et al., 2000) and the presence of this amplicon correlates with tumour grade (Richter et al., 1999; Prat et al., 2001) and high tumour cell proliferation (Tomovska et al., 2001). Previous mapping studies have demonstrated that the 6p22 amplicon spans the closely related SOX4, PRL and E2F3 genes (Bruch et al., 2000; Veltman et al., 2003). In the current study, we have investigated whether amplification of the E2F3 gene is accompanied by its overexpression in human bladder cancer cell lines, and whether overexpression of E2F3 is found in primary human bladder cancer.

We initially used CGH onto cDNA microarrays to screen human bladder cancer cell lines for regions of genomic gain and amplification. Microarrays were co-hybridized with bladder cancer cell line DNA labelled with Cy5 and normal muscle DNA labelled with Cy3. Amplification of the E2F3 gene at 6p22 was detected in three (TCCSUP, 5637 and HT1376) bladder cancer cell lines. The result obtained for the TCCSUP cell line is shown in Figure 1a.

Figure 1

Amplification and expression of E2F3. (a) CGH onto cDNA microarrays. Duplicate results are presented in blue and red. The results for chromosomes 5 and 6 for the TCCSUP bladder cancer cell line are shown. (b) E2F3 Southern blot. Muscle DNA was used as the normal control. (c) CGH studies were repeated using an expanded cDNA microarray that contained 12 genes within a 6.5 Mb region spanning the E2F3 locus. Mapping positions are taken from the UCSC Human Genome Bioinformatics project April 2002 freeze (http://www.genome.ucsc.edu). Results for the TCCSUP, 5637, HT1376 and HT1197 cell lines, and for two primary bladder tumours designated T63 and T110, are shown. (d) E2F3 Northern blot. RNA from normal bladder was used as a control. Northern blot analysis identified E2F3 transcripts of 4.8 and 4.4 kb that correspond to the known sizes (http://www.genome.ucsc.edu) of the E2F3a (4.74 kb) and E2F3b (4.33 kb) transcripts. (e) E2F3 Western blot. Mouse monoclonal antibodies for Western blot analysis of E2F3 protein were obtained from Upstate Ltd (Milton Keynes, UK), and used at a dilution of 1 : 200. The antibody identified a protein doublet at 58 and 52 kDa. Use of antipeptide antibodies (Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA) corresponding to the common C-terminus of E2F3a and E2F3b and to the N-terminus of E2F3a, together with their competing peptides, allowed these bands to be assigned as E2F3a and E2F3b, respectively. Microarray-CGH studies were carried out as described by Clark et al. (2002). Southern, Northern and Western analyses were performed as described by Sambrook et al. (1989). The probe for Southern and Northern blot analysis was a 0.89 kb clone (IMAGE: 2163947) from exon 7 of the E2F3 gene. All cell lines were obtained from the American Type Culture Collection. All tissue samples were retrieved from the archives of the Department of Pathology, Royal Liverpool UK and the Department of Histopathology, Royal Marsden Hospital NHS Trust, UK. Normal human bladder tissues were obtained from patients in whom there was no history or suggestion of malignancy, either within the bladder or elsewhere

Amplification of the E2F3 gene in these three cell lines was confirmed by Southern analysis (Figure 1b). To map this amplicon, CGH studies were carried out onto an expanded cDNA microarray containing all the 12 genes within a 6.5 Mb region spanning the E2F3 locus. These experiments (Figure 1c) show that E2F3 maps within the peak region of amplification in the TCCSUP, 5637 and HT1376 bladder cancer cell lines, and in two primary transitional cell carcinomas of the bladder (T63 and T110). The PRL and SOX4 genes, which mapped outside the peak region of amplification, had also previously been excluded as candidate oncogenes by Bruch et al. (2000). They found that the PRL gene was not expressed at all in bladder cancer cell lines, and that SOX4 expression was only enhanced in one of four lines containing the 6p22 amplicon.

Northern and Western analyses were used to determine whether amplification of the E2F3 gene is accompanied by increased expression of E2F3 mRNA and protein. Although E2F3 transcripts were diffuse, when compared to normal bladder, overexpression of E2F3 mRNA could be observed in the three cell lines that exhibited amplification of E2F3 (Figure 1d). These three cell lines also contained the highest levels of E2F3 protein. Relatively high levels of E2F3 protein were also found in the bladder cancer cell lines J82 and T24 in the absence of DNA amplification when compared to the HT1197, SW780 and SCaBER cell lines (Figure 1e).

Immunohistochemical analysis of bladder cancer cell lines demonstrates that overexpressed E2F3 is present predominantly in cell nuclei (Figure 2a), while cell lines showing low levels of E2F3 (Figure 2b) exhibit faint cytoplasmic staining. Immunohistochemical screening of a series of normal bladders (n=20) failed to identify nuclear E2F3 staining in any but some terminally differentiated umbrella cells at the surface of the urothelium (Figure 2c): in no case was nuclear E2F3 staining found in the subsurface epithelium. Immunohistochemical analysis of primary transitional cell carcinoma of the bladder was carried out using tissue arrays containing triplicate cores of each cancer. In these studies, intense nuclear staining was observed in around one-third (33/101) of primary bladder transitional cell carcinomas (Figures 2d–f), with the proportion of cells containing intense staining varying from 5 to 60%. The proportion of tumours with nuclear E2F3 protein increased with tumour grade (G1, 2/16, 12.5%; G2, 11/44, 25%; G3, 20/38, 53%) (χ² for trend=10.24, DF=1, P=0.0014), in agreement with previous reports that the presence of the 6p22 amplicon in bladder cancer correlates with tumour grade (Richter et al., 1999; Prat et al., 2001). Our results also suggest a correlation with tumour stage (χ²=8.48, DF=1, P=0.0034), since 56% (18/32) of tumours with evidence of muscle invasion (pT2-4) exhibited nuclear staining compared to 18% (15/59) of superficial non-muscle invasive cancers (pTa and pT1).

Figure 2

E2F3 expression detected by immunohistochemistry. Immunohistochemistry on the bladder cancer cell lines (a) 5637 and (b) SCaBER was carried out on formalin-fixed cell pellets. (c) Normal urothelium. (d–f) Primary transitional cell bladder cancer. Examples of tumours scored as (d) negative and (e, f) positive are shown. Immunochemistry was carried out on formalin-fixed tissue or cell lines exactly as described (Cornford et al., 2000). The Upstate E2F3 antibody (Milton Keynes, UK) was used at a dilution of 1 : 200. Tissue arrays were constructed using a Beecher Manual tissue arrayer. Triplicate tissue cores of 0.6 mm were taken from each tumour. In selected cases, up to eight cores were taken to check the consistency of immunohistochemical staining. All images are × 40. Staining: brown, E2F3; blue, haematoxylin counter stain

Several lines of evidence highlight the importance of E2F3 in cell cycle progression and proliferation. E2F3^−/− mouse embryo fibroblasts have a proliferative and cell cycle defect when compared to their wild-type counterparts, and a critical threshold level of one or more E2F3-regulated genes appears to determine the timing of the G1/S transition and rate of DNA synthesis (Leone et al., 1998; Humbert et al., 2000; Wu et al., 2001). Inhibition of E2F3 activity by antibody microinjection impairs entry into the S phase (Leone et al., 1998) and, in transgenic mouse studies, E2F3 expression has been demonstrated to contribute to the ectopic proliferation of neuronal cells and lens fibre cells that occur in Rb^−/− null mice (Huang et al., 2003). Importantly, E2F3 is critical for the transcriptional activation of genes that control proliferation in both normal and transformed cells (Leone et al., 1998). Recently, an E2F3 metagene that could potentially define the expression phenotype of an E2F3 oncogenic pathway has been described (Huang et al., 2003). When considered with these biological properties of E2F3, our results suggest that the E2F3 gene represents a candidate bladder cancer oncogene that is activated by DNA amplification and/or overexpression. The idea that the E2F3 gene has a role in promoting progression of bladder cancer cells is consistent with the observation that the presence of the 6p22 amplicon in human bladder cancer has been shown to be associated with higher cancer cell proliferation rates (Tomovska et al., 2001).

Recently, E2F3 has been shown to bind to the E-box factor TFE3 (Giangrande et al., 2003), which has also been implicated in cancer development. We have previously shown that the entire TFE3 protein becomes fused to the N-terminal domains of the pre-mRNA splicing proteins PRCC, PSF or NonO in human papillary renal cancer (Sidhar et al., 1996; Clark et al., 1997; Skalsky et al., 2001). It is proposed that the E2F3–TFE3 interaction, which involves the marked box domain of E2F3, might contribute to the specificity of E2F3 function (Giangrande et al., 2003).

Many of the genetic alterations found in bladder cancer are believed to function through removal of pRb tumour-suppressor control at the G1/S transition in the cell cycle (Knowles, 2001) (Figure 3). Genetic changes may remove pRb itself, elevate cyclin D1, remove both p14ARF and the CDK inhibitor p16 (each encoded by the INK4A/ARF gene), or involve a variety of alterations that lead directly or indirectly to removal of the CDK inhibitor p21: namely amplification and overexpression of MDM2 and TP53 gene mutation (Knowles, 2001). Downregulation of expression of the p21 gene itself has also been reported in human bladder cancers (Stein et al., 1998). The demonstration that amplification and overexpression of E2F3 occurs in bladder cancer is consistent with the model presented in Figure 3, since removal of pRb is thought to exert its proliferative effect through releasing functional E2F3 transcription factor. The discovery of amplification and overexpression of E2F3 in bladder cancer further highlights the importance of the INK4A/ARF locus in the suppression of this disease (Figure 3). INK4A/ARF encodes both p14ARF (p19ARF in mouse), which interacts with E2F3, promoting its degradation (Martelli et al., 2001), and p16, which represses the activity of CDK4 (Serrano et al., 1993). p14ARF can also block CDK4 activity through the MDM2/p53/p21 pathway.

Figure 3

Control of the pRB pathway. INK4A/ARF encodes two proteins, p16 and p14ARF, that are negative regulators of the pRB pathway. Genes frequently mutated (blue), amplified (red) or downregulated (green) in human bladder cancer are indicated

Previous analyses of the three cell lines TCCSUP, 5637 and HT1376, which we have shown to contain amplification and high levels of expression of E2F3, have consistently demonstrated loss or aberrant expression of pRb, and the presence of wild-type INK4A/ARF (Rieger et al., 1995; Markl and Jones, 1998; Florl and Schulz, 2003). These observations intriguingly suggest that co-operation between pRb removal and overexpression of E2F3 may be required for bladder cancer carcinogenesis. Our results may also have relevance to the design of novel drugs targeting bladder cancer. Drugs directed against upstream targets of E2F3 such as pRB and CDKs (Ortega et al., 2002) might be expected to be less effective against cancers overexpressing E2F3. There would also be concern that amplification and upregulation of E2F3 could represent a mechanism for developing resistance to such drugs.