The novel RASSF6 and RASSF10 candidate tumour suppressor genes are frequently epigenetically inactivated in childhood leukaemias

We investigated the CpG island methylation status of the 'classical' RASSF members, RASSF1–6, in a large series of childhood ALL to ascertain whether epigenetic inactivation of these genes occurs in childhood leukaemia. Surprisingly, we found that RASSF1–5 were infrequently methylated whilst RASSF6 was methylated in B-ALL (94%; 48/51), T-ALL (41%; 12/29) and 5/5 unclassified childhood leukaemias but not in normal blood or bone marrow control samples (0/8 and 0/1 respectively; figure 1A, B, table 1 and Additional file 1; figure S1–2). Given that RASSF6 epigenetic inactivation has not previously been observed in any cancer we next sought to determine the extent of methylation across the RASSF6 CpG island and determine whether methylation correlated with loss or downregulation of RASSF6 expression. Bisulphite sequencing of up to 12 individual alleles from 9 leukaemia cell lines, 6 B-ALL, 6 T-ALL, 1 pre-B-ALL, 6 normal blood and 1 normal bone marrow sample confirmed leukaemia-specific methylation across the entire RASSF6 CpG island in cell lines, B-ALL and T-ALL (figure 1C, and Additional file 1; figure S3 and Additional file 2; table S1). Whilst RT-PCR showed RASSF6 was expressed in all normal tissues analysed including bone marrow (see Additional file 1; figure S4A), and in the unmethylated leukaemia cell line U937 (figure 1D and Additional file 1; figure S4B), we observed absent or downregulated expression of RASSF6 in methylated leukaemia cell lines. Treatment of hypermethylated leukaemia cell lines with 5azaDC and TSA (Trichostatin A) reactivated RASSF6 expression (figure 1D and Additional file 1; figure S4B). Using a commercially available antibody that recognises human RASSF6 protein we also observed restoration of RASSF6 protein expression following 5azaDC and TSA treatment in protein lysates from leukaemia lines (figure 1E). Thus methylation of the RASSF6 CpG island results in loss of RASSF6 expression. As mentioned previously, several RASSF members have been shown to bind Ras proteins, including RASSF6 [11]. Previously we had also demonstrated that inactivation of RASSF2 by promoter hypermethylation inversely correlates with K-Ras mutation status in colorectal carcinomas [12]. We investigated our series of leukaemias for mutations to codons 12, 13 and 61 of both K- and N-Ras. K-Ras mutations were found in 2.2% (2/89) whilst N-Ras mutations were found in 11.2% (10/89) leukaemias. However, no association was found between Ras mutation status and RASSF6 methylation status.

The human RASSF10 gene has been predicted based on homology with RASSF9/P-CIP1 but has yet to be supported by experimental evidence. UCSC http://​www.​genome.​ucsc.​edu and Ensembl http://​www.​ensembl.​org human genome browsers predict a 615 amino acid protein encoded by a two exon gene separated by a short intron of 104 bp. Interestingly, no 5' or 3'UTRs were predicted, hence we reasoned RASSF10 had not been completely characterised. Sequencing of the only IMAGE clone available [IMAGE:4413366] [GenBank:BG034782] obtained from a human liver cDNA library revealed a 3'UTR of 478 bp beyond the predicted stop codon followed by a poly A-tail. Contrary to the UCSC and Ensembl genome browsers however, the IMAGE clone sequence suggested the RASSF10 gene did not contain introns and predicted alternative transcription and translation initiation sites located Chr11:12,987,466 and Chr11:12,987,700 respectively (UCSC genome browser, March 2006 freeze). Analysis of the sequence flanking this alternative ATG (G CCATG G) support its candidacy as a genuine translation initiation site since it matches the consensus 'Kozak' sequence of A/G CCATG G in which the important -3 and +4 A/G and G residues respectively (underlined) were conserved [13, 14]. Furthermore, translation initiation site prediction software at http://​dnafsminer.​bic.​nus.​edu.​sg also predicted the ATG at Chr11:12,987,700 is the translation initiation site within the RASSF10 gene. This suggested the RASSF10 gene contains a shorter open reading frame encoding a 507 amino acid protein. However, since the IMAGE clone may be incomplete we confirmed the transcription start site of the RASSF10 gene using 5'RACE. We first demonstrated expression of the RASSF10 transcript in DNaseI-treated normal human bone marrow RNA using RT-PCR (figure 2A). This RNA was then subjected to 5'RACE using RASSF10-specific antisense primers, which generated a major product of ~330 bp and several larger and smaller minor products (figure 2B). These products were gel purified, cloned and sequenced to reveal a transcription start site located at Chr11:12,987,444 (UCSC genome browser, March 2006 freeze) with two alternative transcription start sites at +119 bp and +169 bp (Chr11:12,987,565 and Chr11:12,987,613, respectively). This revealed the presence of a 5'UTR of up to 431 bp and confirms the RASSF10 gene likely encodes a much shorter 507 amino acid protein. That the predicted N-terminal 108 amino acids may not actually be part of the RASSF10 protein is supported by ClustalW alignments of RASSF7–10. These alignments show the predicted N-terminal 108 amino acids are extremely divergent with no significant homology to RASSF7–9 or with any other human protein (figure 2C and 2D). Interestingly, the Rassf10 murine protein [Genbank:NP_780488] also lacks this divergent N-terminus and is 508 amino acids in length. Therefore, we have characterised the human RASSF10 gene and verified expression in normal bone marrow. We provide evidence that RASSF10 is a single exon gene with a 5'UTR of up to 431 bp, a 3'UTR of 478 bp and an open reading frame encoding a protein of 507 amino acids, not 615 amino acids as predicted by current versions of human genome browsers.

RASSF10 contains a large CpG island that covers the majority of the gene. Given the expression of RASSF10 in bone marrow we questioned whether RASSF10 is also inactivated by promoter methylation in the same series of childhood leukaemias. We investigated the RASSF10 CpG island for DNA hypermethylation using COBRA and bisulphite sequencing. The RASSF10 CpG island was heavily methylated in 7/7 leukaemia cell lines (figure 3A). Next, we investigated childhood B- and T-ALL, normal blood and normal bone marrow for evidence of RASSF10 CpG island hypermethylation. We found RASSF10 was methylated in 23/26 (88%) T-ALL and 8/51 (16%) B-ALL but was unmethylated in normal bone marrow and 21 normal blood samples (figure 3A, table 1). Direct bisulphite sequencing of the RASSF10 CpG island confirmed the presence of dense methylation across the entire region analysed (figure 3B). To investigate the effects of methylation on RASSF10 expression we analysed the leukaemia cell lines THP-1 and SUP-T1 using RT-PCR. We found low level expression of RASSF10 in THP-1 (partially methylated) and loss of RASSF10 expression in SUP-T1 (completely methylated) and in both cases expression was upregulated following 5azaDC and TSA treatment (figure 3C). Thus, methylation of the RASSF10 CpG island inactivates expression of the gene very frequently in T-ALL and in a subset of B-ALL.

Next, we investigated the remaining 'N-terminal' RASSF members to determine whether they too become inactivated in childhood leukaemias. RASSF7 and RASSF8 both contain 5' CpG islands encompassing their first exon, whilst RASSF9 does not contain a CpG island. As determined by COBRA and bisulphite sequencing, RASSF7 was unmethylated in all leukaemia cell lines analysed (0/5). RASSF8 methylation was found to be infrequent, occurring in 2/6 leukaemia cell lines, 2/19 (10%) childhood T-ALLs, and 4/46 (9%) B-ALL with no methylation detectable in normal bone marrow or normal blood (0/7). Methylation of the RASSF8 CpG island correlated with loss of RASSF8 expression which was restored following treatment with 5azaDC (and Additional file 1; figure S5).

Twelve leukaemia cell lines (DND-41, CCRF-CEM (CEM), U937, Jurkat (JKT), TALL-1, NALM1, NALM6, NALM16, NALM17, THP-1, SUP-T1 and MOLT-4) and 89 primary childhood ALL comprising 51 B-cell ALL (B-ALL), 32 T-cell ALL (T-ALL), 1 pre-B-ALL and 5 unclassified leukaemias were analysed. In addition a total of 21 normal blood samples and 1 normal bone marrow (BM, AMS Biotechnology) sample were used as controls. All DNA samples were obtained with informed consent from patients and family members.

Bisulphite modification was performed as described previously [23]. Promoter methylation status of RASSF1A and RASSF2 were determined using MSP as described previously [12]. The methylation status of RASSF3, RASSF4 and RASSF5A were determined by combined bisulphite restriction analysis (COBRA) as described in Hesson et al., [12], Hesson et al., [23] and Eckfeld et al., [24]. For RASSF6, primers were designed to amplify the entire CpG island encompassing the first exons of RASSF6A (NM_177532) and RASSF6B (NM_201431) from bisulphite modified DNA using semi-nested bisulphite PCR (see additional file 2; table S2 for primer sequences). The methylation status of RASSF7, RASSF8 and RASSF10 were determined by COBRA and direct bisulphite sequencing. For RASSF7 and RASSF8 primers were designed to amplify a portion of the CpG island encompassing exon 1. The RASSF10 CpG island covers the majority of the gene. We designed primers to amplify a portion of the CpG island encompassing the 5' region of the gene. Five microlitres of PCR product was incubated with 2 U TaqI restriction enzyme (TCGA) for 2 h at 65°C or 2 U BstUI restriction enzyme (CGCG) for 2 h at 60°C before visualisation on a 2% agarose gel. Samples selected for bisulphite sequencing were cloned (RASSF6) into the CloneJET vector (Fermentas) according to manufacturers instructions or sequenced directly (RASSF8 and RASSF10). Up to 12 individual colonies were chosen for colony PCR using the RASSF6 COBRA primers. Products were then sequenced to ascertain the methylation status of individual alleles and to determine the methylation index (MI). The MI was calculated as a percentage using the equation; number of CpG dinucleotides methylated/total number of CpG dinucleotides sequenced × 100.

Leukaemia cell lines were maintained in RPMI1640 (Sigma) supplemented with 10% FCS, 2 mM Glutamine, 20 mM HEPES, 1 mM Sodium Pyruvate and 12.6 mM Glucose Monohydrate at 37°C, 5% CO₂. Cells were treated with 5 μM of the DNA demethylating agent 5azaDC (Sigma) freshly prepared in ddH₂O and filter-sterilised. The medium (including 5 μM 5azaDC) was changed every day for 5 days. Cells were also treated on day 4 with 0.1 μM TSA for 24 hrs. RNA was prepared using RNA bee (AMS biotechnology) according to manufacturers' instructions. cDNA was generated from 1 μg total RNA using SuperScript III (Invitrogen) and polyN primers. We analysed expression of RASSF6 using primers specific for exons 2 and 4 that amplify both the A and B isoforms (F = 5'-ATGATGGCTCACCAGTACCC-3' and R = 5'-GGTCGTTTTACTCCCCAGAA-3'). We analysed expression of RASSF10 using the primers 5'-CCATGACCCAGGAGAAACAG-3' (F) and 5'-TGCTGGCGAATTGTGTGGTC-3' (R). Since RASSF10 is a single exon gene we first DNaseI (Fermentas) treated all RNA samples prior to cDNA synthesis and included control experiments in which reverse transcriptase was omitted (No RT controls) in all samples analysed. We analysed expression of RASSF8 using the primers F = 5'-TCCATTGAGAAACAGCTGGA-3' and R = 5'-TGGCACAAATCAAAAAGGAA-3'. In all cases a GAPDH control was included using conditions described previously [12]. RASSF6, RASSF8 and RASSF10 were amplified from 50 ng cDNA using 0.8 μM of each primer, 3 mM MgCl₂, 0.25 mM dNTPs and 1 U Fast start Taq (Roche).

Leukaemia lines THP-1 and CEM were treated with 5azaDC as above. Protein extracts were then prepared by lysis in 50 mM TrisHCl pH7.5, 1 mM EDTA, 1 mM EGTA, 50 mM Sodium Fluoride, 5 mM Sodium Pyrophosphate, 1 mM Sodium Orthovanadate, 0.27 M Sucrose, 1% Triton and protease inhibitor cocktail (Roche). Twenty micrograms of protein extracts were resolved by SDS-PAGE, transferred to PVDF membrane and probed for the presence of RASSF6 using 1 μg/mL anti-RASSF6 rabbit-raised polyclonal antibody (ProteinTech Group).

One microgram of total RNA from bone marrow (AMS Biotechnology) was DNaseI (Fermentas) treated according to manufacturers instructions. We synthesised cDNA using a 5'/3' RACE kit (2^nd generation, Roche) and the RASSF10 antisense primer 5'-TGCTGGCGAATTGTGTGGTC-3'. The 5' UTR of the RASSF10 transcript was obtained using nested PCR with the nested antisense RASSF10-specific primers 5'-GCTGTTTCTCCTGGGTCATG-3' and 5'-TATCTTCTTTTCCGAAGGATCCATG-3' used in combination with a polyT-anchor primer. Control experiments in which reverse transcriptase was omitted were also performed to eliminate DNA contamination. PCR products were gel extracted, sequenced and the sequences obtained aligned against the human genome browser.

PCR primers were used to amplify exons 2 and 3 of the K- and N-Ras genes from genomic DNA. The primers were as follows: K-Ras exon 2 F 5'-TTTGTATTAAAAGGTACTGGTGGAG-3'; K-Ras exon 2 R 5'-CCTTTATCTGTATCAAAGAATGGTC-3'; K-Ras exon 3 F 5'-CTGTGTTTCTCCCTTCTCAGG-3'; K-Ras exon 3 R 5'-AGAAAGCCCTCCCCAGTCCT-3'; N-Ras exon 2 F 5'-GATGTGGCTCGCCAATTAAC-3'; N-Ras exon 2 R 5'-GAATATGGGTAAAGATGATCCGAC-3'; N-Ras exon 3 F 5'-GTTAGATGCTTATTTAACCTTGGC-3'; N-Ras exon 3 R 5'-TGTGGTAACCTCATTTCCCC-3'. PCR products were directly sequenced using ABI BigDye cycle sequencing kit (Perkin-Elmer).