Identification of DNA hypermethylation of SOX9 in association with bladder cancer progression using CpG microarrays

Abstract

CpG island arrays represent a high-throughput epigenomic discovery platform to identify global disease-specific promoter hypermethylation candidates along bladder cancer progression. DNA obtained from 10 pairs of invasive bladder tumours were profiled vs their respective normal urothelium using differential methylation hybridisation on custom-made CpG arrays (n=12 288 clones). Promoter hypermethylation of 84 clones was simultaneously shown in at least 70% of the tumours. SOX9 was selected for further validation by bisulphite genomic sequencing and methylation-specific polymerase chain reaction in bladder cancer cells (n=11) and primary bladder tumours (n=101). Hypermethylation was observed in bladder cancer cells and associated with lack of gene expression, being restored in vitro by a demethylating agent. In primary bladder tumours, SOX9 hypermethylation was present in 56.4% of the cases. Moreover, SOX9 hypermethylation was significantly associated with tumour grade and overall survival. Thus, this high-throughput epigenomic strategy has served to identify novel hypermethylated candidates in bladder cancer. In vitro analyses supported the role of methylation in silencing SOX9 gene. The association of SOX9 hypermethylation with tumour progression and clinical outcome suggests its relevant clinical implications at stratifying patients affected with bladder cancer.

Figures

Figure 1

CpG arrays. (A) Quality control: the relative percentage of methylated and unmethylated genes was similar among the pairs of bladder tumours and normal urothelium samples under study. (B) CpG island arrays identify hypermethylated candidates in patients with bladder cancer. (C) Summary of known genes among the 84 clones simultaneously hypermethylated in 7 out of 10 samples under analysis. The number of cases found differentially expressed with a Cy5/Cy3 ratio higher than 2 for these genes are also indicated.

Figure 2

Analysis of CpG island methylation status of the promoter of SOX9 by bisulphite genomic sequencing in bladder human cancer cell lines (n=11; including nonmuscle invasive, invasive, metastatic transitional and squamous cells). The upper part indicates the nucleotide sequences of the CpG island region analysed by bisulphite sequencing, sequencing primers highlighted in yellow and red and the area amplified in the chromatograms in blue. The mid-section shows a schematic depiction of the SOX9 CpG islands around the transcription start sites. CpG dinucleotides are represented in squares. The presence of ‘Cs’ in the dinucleotide CpG reflects methylated cytosines (black squares), while the presence of ‘Ts’ in the dinucleotide CpG reflects unmethylated cytosines (white squares). Cell lines with black squares indicate the presence of methylation confirmed in at least two of the clones that were sequenced for each of the cell lines under analyses. The bottom part displays representative examples of the chromatograms obtained by bisulphite genomic sequencing of human cancer cell lines (a magnified boxed fragment is displayed). Normal lymphocytes (NL) were used as a negative sequencing control.

Figure 3

CpG island methylation is associated with gene silencing of SOX9. The upper part shows methylation-specific PCRs for SOX9 in human bladder cancer cell lines. The presence of a PCR band under the lane M indicates a methylated gene, while the presence of a PCR band under the lane U indicates an unmethylated gene. Normal lymphocytes (NL) and in vitro-methylated DNA (IVD) were used as negative and positive controls for unmethylated and methylated PCRs, respectively. Sequencing information is included as well, highlighting methylated cell lines by genome sequencing in dark grey. Reverse transcription polymerase chain reaction analysis of SOX9 expression is displayed. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) transcript expression was used as a transcript loading control. Western blot analysis of protein expression is also shown. Tubulin expression was used as a protein loading control. The hypermethylated cell lines show relatively low transcript and protein expression of the coded protein as compared to unmethylated cell lines.

Figure 4

The treatment with the demethylating agent AZA reactivates gene expression of SOX9. The upper part displays the reverse transcription polymerase chain reaction analysis of SOX9 expression. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression was used as a transcript loading control. The hypermethylated J82 cell line did not express SOX9, and restored SOX9 transcript expression after AZA exposure. The mid-section shows western blot analysis of protein expression. Tubulin expression was used as a protein loading control. The hypermethylated cell line did not express the coded protein. The treatment with the demethylating agent reactivated SOX9 protein expression. The unmethylated RT4 cell line did not show changes in transcript or protein expression. The bottom part displays immunofluorescence analysis of SOX9 expression after AZA exposure. The methylated cell line did not show any staining for the protein, while the unmethylated ones showed its characteristic staining pattern.

Figure 5

Association between SOX9 hypermethylation with cancer progression and clinical outcome in bladder tumours. (A) Representative pairs of normal urothelium (NB) and primary bladder tumours (T) analysed by MS-PCR for SOX9. The presence of a PCR band under the lane M indicates a methylated gene, while the presence of a PCR band under the lane U indicates an unmethylated gene. Normal lymphocytes (NL) and in vitro-methylated DNA (IVD) are used as negative and positive controls for unmethylated and methylated PCRs, respectively. (B) Summary of the number of unmethylated (U) and methylated (M) cases for SOX9 regarding their tumour stage and tumour grade. (C) Kaplan–Meier curve describing the association of SOX9 hypermethylation with poor overall survival.