|Home | About | Journals | Submit | Contact Us | Français|
Promoter region hypermethylation and transcriptional silencing is a frequent cause of tumour suppressor gene (TSG) inactivation in many human cancers. Previously, to identify candidate epigenetically inactivated TSGs in renal cell carcinoma (RCC), we monitored changes in gene expression in four RCC cell lines after treatment with the demethylating agent 5-azacytidine. This enabled us to identify HAI-2/SPINT2 as a novel epigenetically inactivated candidate RCC TSG. To identify further candidate TSGs, we undertook bioinformatic and molecular genetic evaluation of a further 60 genes differentially expressed after demethylation. In addition to HAI-2/SPINT2, four genes (PLAU, CDH1, IGFB3 and MT1G) had previously been shown to undergo promoter methylation in RCC. After bioinformatic prioritisation, expression and/or methylation analysis of RCC cell lines±primary tumours was performed for 34 genes. KRT19 and CXCL16 were methylated in RCC cell lines and primary RCC; however, 22 genes were differentially expressed after demethylation but did not show primary tumour-specific methylation (methylated in normal tissue (n=1); methylated only in RCC cell lines (n=9) and not methylated in RCC cell lines (n=12)). Re-expression of CXCL16 reduced growth of an RCC cell line in vitro. In a summary, a functional epigenomic analysis of four RCC cell lines using microarrays representing 11000 human genes yielded both known and novel candidate TSGs epigenetically inactivated in RCC, suggesting that this is valid strategy for the identification of novel TSGs and biomarkers.
Renal cell carcinoma (RCC) is a heterogeneous disorder. Most (~75%) of the tumours are classified as clear cell (conventional) and the next most frequent subtype is papillary RCC (~15% of all cases) (Kovacs et al, 1997). The most frequent genetic event in the evolution of clear cell RCC is inactivation of the VHL tumour suppressor gene (TSG) (Latif et al, 1993; Foster et al, 1994; Herman et al, 1994; Clifford et al, 1998), but epigenetic inactivation of TSGs by methylation of CpG dinucleotides in the promoter region has also been implicated in the pathogenesis of RCC (Morrissey et al, 2001; Dallol et al, 2002; Morris et al, 2003). Tumour suppressor gene promoter methylation has been studied most extensively in colorectal cancer, but TSGs that are frequently methylated in human cancers (e.g. p16ink4a, DAPK, NORE1A, MGMT, SDHB, RARB2 and CDH13) are infrequently methylated in RCC (Morris et al, 2003). This observation prompted us to speculate that, compared to other tumour types, a different repertoire of TSGs might undergo epigenetic inactivation in RCC. Therefore, to identify candidate novel epigenetically inactivated RCC TSGs, we performed a gene expression profiling of four RCC cell lines treated with the demethylated agent 5-azacytidine (5-AZA) (Morris et al, 2005). Analysis of genes differentially expressed before and after demethylation led us to identify HAI-2/SPINT2 as a novel epigenetically inactivated RCC TSG (Morris et al, 2005). In this study, we report the further analysis of our functional epigenomic screen to identify genes relevant in development of RCC.
Full details of the gene expression microarray experiments on four RCC cell lines (KTCL26, SKRC39, SKRC45 and SKRC47) have been described previously (Morris et al, 2005). All RCC cell lines (KTCL26, RCC4, UMRC2, UMRC3, SKRC18, SKRC39, SKRC45, SKRC47, SKRC54, 786-0 and Caki-1) analysed in this study were routinely maintained in DMEM (Invitrogen, San Diego, CA, USA) supplemented with 10% FCS at 37°C, 5% CO2. The demethylating agent 5-AZA-dC (Sigma, Gillingham, UK) was freshly prepared in ddH2O and filter sterilised. Cell lines were plated in 75-cm2 flasks in DMEM supplemented with 10% FCS at differing densities, depending upon their replication factor, to ensure that both control and 5-AZA-dC-treated lines reached approximately 75% confluency at the point of RNA extraction. Twenty-four hours later, cells were treated with 5μM 5-AZA-dC. The medium was changed 24h after treatment and then changed again after 72h. RNA was prepared 5 days after treatment using RNABee (AMS Biotechnology, Oxon, UK).
DNA from a total of 127 primary RCCs and 6 non-cancer-related kidneys was analysed. Local ethics committees approved the collection of samples and informed consent was obtained from each patient.
PCR cycling conditions consisted of an initial denaturing step of 95°C for 5min, followed by 30 cycles of 95°C denaturing at 45s, primer annealing at 55–60°C (gene-dependent) and product extension at 72°C for 45s. Semiquantitative analysis of expression was performed using LabWorks software (Ultraviolet products, Upland, CA, USA). GAPDH primers were 5′TGAAGGTCGGAGTCAACGGATTTGGT3′ and 5′CATGTGGGCCATGAGGTCCACCAC3′ (other RT–PCR primers and conditions upon request).
Bisulphite DNA sequencing was performed as described previously (Morris et al, 2003). Briefly, 0.5–1.0μg of genomic DNA was denatured in 0.3M NaOH for 15min at 37°C, and then unmethylated cytosine residues were sulphonated by incubation in 3.12M sodium bisulphite (pH 5.0; Sigma)/5mM hydroquinone (Sigma) in a thermocycler (Thermo Fisher Scientific, Waltham, MA, USA) for 20 cycles of 30s at 99°C and 15min at 50°C. The sulphonated DNA was recovered using the Wizard DNA cleanup system (Promega, Southampton, UK) in accordance with the manufacturer's instructions. The conversion reaction was completed by desulphonating in 0.3M NaOH for 10min at room temperature. The DNA was ethanol-precipitated and resuspended in water.
CpG islands were identified on the human genome browser and putative promoter regions were predicted by Promoter Inspector software (Genomatix, www.genomatix.de). Primer details are shown in Supplementary Table 1.
The CXCL16 expression construct was made by cloning the full-length human CXCL16 coding region from the SKRC 18 kidney cell line, into the EcoR1–BamHII sites of pCDNA3.1 vector (Invitrogen). Plasmid constructs were verified by sequencing. Six micrograms of empty vector or 6.8μg (equal molar amounts) of expression vector was transfected, by calcium phosphate method, into 5 × 105 SKRC39 cells. Forty-eight hours after transfection, cells were seeded in a serial dilution and maintained in DMEM and 10% foetal bovine serum supplemented with 1mgml−1 G418 (Gibco, Invitrogen, Paisley, UK). Surviving colonies were stained with 0.4% crystal violet (Sigma) in 50% methanol, 21 days after initial seeding, and counted. Each transfection was carried out in triplicate. Additionally, replicate experiments were carried out to obtain further clones for expression analysis.
A total of 56 genes (of 11000 transcripts analysed), each with a 5′ CpG island, that demonstrated >5-fold increased expression in one cell line or >2-fold in multiple cell lines were identified as candidate epigenetically inactivated RCC TSGs. In addition to SPINT2, four further genes (CDH1, PLAU, IGFB3 and MT1G) had previously been reported to undergo promoter region hypermethylation in RCC tumours (Yoshiura et al, 1995; Morris et al, 2003; Ibanez de Caceres et al, 2006). As we wished to identify novel genes, these were not analysed further. After bioinformatic evaluation (e.g. renal tissue specific expression patterns (array express (www.ebi.ac.uk/arrayexpress/), human genome browser (www.genome.ucsc.edu)), analysis of CpG islands for the presence of predicted promoter regions (www.genomatix.de) and consideration of likely role in tumorigenesis), 34 genes were selected for analysis of expression and/or CpG island methylation status in RCC cell lines±primary RCC tumours.
After analysis of expression pre- and post-treatment with 5-AZA in up to 11 RCC cell lines, nine genes (ATF5, SLC1A4, ID3, STC2, DUSP6, SEMA3C, CD44, HMGA1 and IRF7) were excluded from further investigation as they were not commonly silenced in the pretreatment cell lines (data not shown).
All genes analysed had a 5′ CpG island (as identified by the human genome browser (http://genome.ucsc.edu/)) and promoter regions were predicted by Genomatix promoter inspector software (http://www.genomatix.de/). The methylation status of 5′ CpG island promoter regions were analysed by direct sequencing of bisulphate-modified DNA of RCC cell lines and primary tumours for 25 genes. Methylation of 5′ CpG dinucleotides was rare (ICAM1, IGSF4, EHM2, CLDN1, MUC1, SEMA5A) or absent (CTGF, PHD3, RRM2, PMAIP1, GPR39, MYL2, BAP1, SLC25A21, FBLN1, H2B, ECE1, FZD8) for 18 genes. For ICAM1, IGSF4, EHM2, CLDN1 and MUC1, no methylation was detected in 20 sporadic RCC tumours. These findings suggest that the alterations in expression after demethylation observed for these 18 genes were caused either by (a) methylation of other in cis regulatory regions or (b) that expression of these genes might be regulated by genes that were epigenetically inactivated.
Seven genes demonstrated promoter region hypermethylation concordant with gene expression in RCC cell lines (e.g. KTN19; Figures 1A–C). Four of the seven genes (SST, PTGS1, ISG15, THY1) were frequently methylated in RCC cell lines (Table 1) but were not methylated in primary RCC (n=20) and although ENG was frequently methylated in RCC cell lines and primary tumours (Table 1), methylation was also detected in adjacent normal tissues and in normal renal tissue from patients without cancer (n=6).
Two of the seven genes, CXCL16 and KTN19, demonstrated frequent promoter region hypermethylation in RCC cell lines (40 and 62%, respectively) and in primary RCCs (42% (n=62) and 38% (n=66), respectively, but not in normal renal tissue from patients without cancer (n=6). For primary RCC with KTN19 promoter methylation, methylation was rarely (14%, 3/22) detected in the adjacent normal tissue. In contrast, for tumours with CXCL16 methylation (but not for unmethylated tumours), methylation was also detected in the adjacent normal tissue.
We have expanded our previous analysis of HAI-2/SPINT2 promoter methylation in RCC (38%) to incorporate all tumours analysed for KTN19 and CXCL16 promoter methylation. However, no significant correlation was detected between methylation at SPINT2, KTN19 and CXCL16 (P>0.1) (data not shown).
The effect of re-expression of CXCL16 on cell growth was assessed by in vitro colony formation assays. Following transfection of a wild-type CXCL16 expression plasmid (or empty vector) into SKRC39 (an RCC cell line, which is heavily methylated at the CXCL16 promoter), there was a significantly reduced number of G418 resistant colonies (mean 40.1%, t=4.16 P=0.014) compared to SKRC39 cells transfected with an empty vector control in three independent experiments (Figure 2).
Functional epigenomic screens have proven to be a successful strategy for identifying epigenetically inactivated TSGs in a number of different tumour types (Yamashita et al, 2002; Sato et al, 2003; Lodygin et al, 2005). However, to our knowledge, only one other functional epigenomic study of RCC has been reported. Thus, Ibanez de Caceres et al (2006) performed gene expression microarrays in four RCC cell lines (all different to those analysed in our study) after treatment with a demethylating agent (5-AZA-2 deoxycytidine) and a histone deacetylation inhibitor (trichostatin A) and found that between 111 and 170 genes demonstrated a 3-fold upregulation of expression after treatment in each cell line. Then they proceeded to analyse 12 genes that were upregulated 3-fold in at least three of the four cell lines and were expressed in renal tubular cells (BIRC3, NP, GADD45A, NFKB1A, CYCS, TGM2, IGFBP1, COL1A1, CTGF, IGFBP1, GDF15 and PLAU) and found that three genes (IGFBP1, IGFBP3 and COL1A1) demonstrated tumour-specific methylation and seven genes did not show promoter methylation. Despite differences in experimental details and microarrays employed, their findings are similar to those from our more extensive analysis. Thus, it is clear that although upregulation of gene expression after demethylation treatment will often not indicate promoter region hypermethylation, and even if methylation is present, it may not be tumour specific, this experimental approach does provide a strategy for identifying novel TSGs.
Keratin 19, an intermediate filament protein responsible for the structural integrity of epithelial cells as a gene, frequently demonstrated promoter methylation in sporadic RCC but not in normal kidney tissue from RCC patients and non-cancer patients. Previously, keratin 19 was reported to be downregulated in head and neck cancers. Furthermore, although re-expression of keratin 19 did not affect the growth rate of transfected cell lines, in vitro invasiveness after treatment with HGF was reduced after re-expression (Crowe et al, 1999).
We found that the transmembrane chemokine CXCL16 (Matloubian et al, 2000; Shimaoka et al, 2000; Wilbanks et al, 2001; Shimaoka et al, 2003) was also frequently methylated in RCC cell lines and in primary tumours but not in normal kidney from non-cancer controls. Although CXCL16 methylation was detected in adjacent normal renal tissue from RCC patients with tumour methylation, this might indicate a premalignant field defect (as described in bronchial epithelium); however, contamination by tumour cells cannot be excluded completely. Interestingly, re-expression of CXCL16 significantly reduced the formation of kidney cell line colonies.
A significant rationale for identifying RCC-associated hypermethylated TSGs is their potential role as biomarkers to identify high-risk individuals and presymptomatic tumours by analysis of urine samples (Battagli et al, 2003; Hoque et al, 2004). In this context, it is valuable to identify both genes that are methylated early in tumorigenesis (but which may not show tumour-specific methylation) and also genes that are specific for tumours.
The functional epigenomic approach we pursued, in addition to identifying genes associated with tumour-specific methylation, also identified significant numbers of genes that (a) appeared to demonstrate cell line-specific methylation (SST, PTGS1, ISG15, THY1) or (b) were upregulated by demethylation but had no apparent promoter region methylation. The former group, although apparently not frequently methylated in primary tumours, might be considered as candidate genes implicated in progression (assuming that acquisition of methylation in cell culture might have a growth advantage). In this context, it is interesting to note that PTGS1 methylation is frequent in prostate cancer (Bastian et al, 2005), THY1 has been implicated as a candidate TSG in nasopharyngeal cancer (Lung et al, 2005) and SST promoter methylation has been described in ~90% of colorectal cancers (Mori et al, 2006).
Genes that were upregulated by demethylation but not methylated included genes that have been described previously as TSGs in other cancer types (e.g. BAP1 (Jensen and Rauscher, 1999), IGSF4 (Kikuchi et al, 2006), RRM2 (Gautam and Bepler (2006)), PMAIP1 (Shibue et al, 2003), claudin-1; (Higashi et al, 2007) and ICAM1; (Georgolios et al, 2006)). These observations would be consistent with the hypothesis that expression of these genes is reactivated secondary to changes in promoter methylation at upstream regulators. Hence, further analysis of this class of genes may provide insights into key pathways that are dysregulated in RCC and identify candidate upstream regulators for epigenetic analysis.
We have reported the identification and systematic analysis of candidate RCC TSGs identified from a functional epigenomics screen. To date, the only two functional epigenomic studies performed in RCC have each analysed four RCC cell lines on arrays containing <15000 genes, but each have identified novel genes with tumour-specific methylation. Compared to other genes reported to be methylated in RCC, SPINT2, CXCL16 and KTN19 are relatively frequently methylated (Figure 3). Therefore, we suggest that further, more extensive, studies are warranted to identify additional potential RCC TSGs and methylated biomarkers for early cancer detection.
We thank Cancer Research UK for financial support.