RASSF1A promoter region CpG island hypermethylation in phaeochromocytomas and neuroblastoma tumours

Abstract

Deletions of chromosome 3p are frequent in many types of neoplasia including neural crest tumours such as neuroblastoma (NB) and phaeochromocytoma. Recently we isolated several candidate tumour suppressor genes (TSGs) from a 120 kb critical interval at 3p21.3 defined by overlapping homozygous deletions in lung and breast tumour lines. Although mutation analysis of candidate TSGs in lung and breast cancers revealed only rare mutations, expression of one of the genes (RASSF1A) was absent in the majority of lung tumour cell lines analysed. Subsequently methylation of a CpG island in the promoter region of RASSF1A was demonstrated in a majority of small cell lung carcinomas and to a lesser extent in non-small cell lung carcinomas. To investigate the role of 3p TSGs in neural crest tumours, we (a) analysed phaeochromocytomas for 3p allele loss (n=41) and RASSF1A methylation (n=23) and (b) investigated 67 neuroblastomas for RASSF1A inactivation. 46% of phaeochromocytomas showed 3p allele loss (38.5% at 3p21.3). RASSF1A promoter region hypermethylation was found in 22% (5/23) of sporadic phaeochromocytomas and in 55% (37/67) of neuroblastomas analysed but RASSF1A mutations were not identified. In two neuroblastoma cell lines, methylation of RASSF1A correlated with loss of RASSF1A expression and RASSF1A expression was restored after treatment with the demethylating agent 5-azacytidine. As frequent methylation of the CASP8 gene has also been reported in neuroblastoma, we investigated whether RASSF1A and CASP8 methylation were independent or related events. CASP8 methylation was detected in 56% of neuroblastomas with RASSF1A methylation and 17% without RASSF1A methylation (P=0.0031). These results indicate that (a) RASSF1A inactivation by hypermethylation is a frequent event in neural crest tumorigenesis, particularly neuroblastoma, and that RASSF1A is a candidate 3p21.3 neuroblastoma TSG and (b) a subset of neuroblastomas may be characterized by a CpG island methylator phenotype.

Introduction

Neuroblastoma and phaeochromocytoma are the commonest neural crest-derived tumours in children and adults respectively. Neuroblastoma is clinically variable with some tumours demonstrating spontaneous regression after little or no therapy, while other tumours have metastasized at presentation and have a poor prognosis. Familial neuroblastoma is rare and major susceptibility genes have not yet been isolated. The molecular pathology of sporadic neuroblastoma has been investigated extensively. Frequent alterations include N-myc amplification and gain of genetic material at 17q23-qter. Neuroblastoma tumour suppressor genes (TSGs) have been mapped by loss of heterozygosity (LOH) studies to 1p36, 11q23 and 14q23-qter (Maris and Matthay, 1999). In addition, about 15% of neuroblastomas demonstrate 3p allele loss (Ejeskar et al., 1998).

Phaeochromocytoma usually present with hypertension and 90% are benign. About 10% occur in susceptible individuals and germline mutations in the RET, VHL, NF1 and SDHD genes may predispose to phaeochromocytoma (Wallace et al., 1990; Eng et al., 1996; Woodward et al., 1997; Astuti et al., 2001). However somatic VHL and RET mutations are rare in sporadic phaeochromocytomas. Investigations of sporadic phaeochromocytomas for LOH have demonstrated frequent allele losses at 1p, 3p, 11 and 22. The incidence of 3p LOH in sporadic phaeochromocytoma is reported to be up to 50% (Zeiger et al., 1995; Vargas et al., 1997), which is much higher than the frequency of somatic VHL inactivation, consistent with a further phaeochromocytoma suppressor gene(s) (in addition to VHL) mapping to 3p.

Distinct regions on 3p are lost frequently in lung and other sporadic cancers suggesting that multiple TSGs map to 3p. Recently, we and others identified a novel 3p21.3 TSG, RASSF1, from a 120 kb interval identified by four overlapping homozygous deletions in lung and breast cancer cell lines (Dammann et al., 2000; Lerman and Minna for the International Lung Cancer Chromosome 3p21.3 Tumor Suppressor Gene Consortium 2000). A CpG island in the promoter region of isoform A of RASSF1 (RASSF1A) is methylated in the majority of lung tumours and in a variable proportion of breast tumours. (Dammann et al., 2000; Agathanggelou et al., 2001; Burbee et al., 2001). RASSF1A promoter region methylation is associated with loss of expression in tumour cell lines and expression is restored after treatment with 5′-aza-2′ deoxycytidine. In addition, in vivo and in vitro studies have demonstrated that RASSF1A suppresses the growth of tumour cell lines (Dammann et al., 2000; Burbee et al., 2001). These findings establish RASSF1A as a 3p21.3 TSG.

To investigate the role of 3p TSGs in the pathogenesis of neural crest tumours we analysed phaeochromocytomas for 3p allele loss and investigated phaeochromocytoma and neuroblastoma tumors for RASSF1A inactivation. We also investigated the relationship between RASSF1A and CASP8 methylation in neuroblastoma (Teitz et al., 2000).

3p LOH analysis and RASSF1A inactivation in sporadic phaeochromocytomas

We determined the frequency, extent and patterns of 3p loss in 41 sporadic phaeochromocytomas using 17 microsatellite markers as described in Fullwood et al. (1999) (D3S: 1304, 1537, 1038, 1259, 2431, 2337, 2432, 2407, 4597, 4604, 1621, 1289, 2408, 1481, 1284, 3507, 1274), targeted to regions containing known or putative 3p TSGs. 46% (19/41) of tumours demonstrated LOH at one or more loci, and four tumours showed LOH for all informative markers. 3p21.3 allele loss was found in 38.5% (15/39) informative tumours but two tumours with 3p21 allele loss did not show loss in other candidate TSG regions such as 3p12, 3p14 and 3p25-p26.

In view of the high frequency of 3p21.3 allele loss in phaeochromocytoma, we proceeded to investigate a subset of tumours (for which sufficient DNA was available) for RASSF1A CpG island promoter methylation status as described previously (Agathanggelou et al., 2001, Morrissey et al., 2001). 22% (5/23) of phaeochromocytomas demonstrated RASSF1A hypermethylation (all matching blood DNA samples were unmethylated) (Figure 1a). Complete digestion of the PCR products with TaqI and BstUI should have produced fragments of 92 bp, 81 bp, 31 bp and 89 bp, 83 bp, 32 bp respectively (Figure 1c). However, in most of the cases which were methylated the full spectrum of possible restriction enzyme digestion products were observed suggesting that there is a heterogeneity in the methylation pattern within the tumour cell population. We then sequenced five individual clones from each of two methylated phaeochromocytomas to determine the precise pattern of CpG methylation within the RASSF1A CpG island. In addition to the promoter fragments demonstrating either complete CpG methylation or methylation of majority of the CpGs, there were also clones of unmethylated DNA which may be attributed to the presence of contaminating normal tissue (tumour samples were not microdissected).

Figure 1
figure1

Representative results of bisulphite-PCR methylation analysis in (a) phaeochromocytomas and (b) neuroblastomas of the 5′ region of RASSF1A. Tumour and normal DNA was treated with bisulphite, amplified using specific primers and conditions reported previously (Agathanggelou et al., 2001) and digested using restriction enzymes that recognize methylated alleles only, bands separated by 3% agarose gel stained with ethidium bromide. Methylated alleles cleaved by restriction enzymes TaqI or BstUI. The sizes of the digested products (and 204 bp undigested product) are against the brackets. The 204 bp undigested product is seen in each lane. (c) Representation of the reverse sequence of the sodium bisulfite-treated RASSF1A promoter region CpG fragment amplified as described in Agathanggelou et al., 2001. The 16 CpG dinucleotides (shown as methylated) are underlined and the restriction sites for TaqI and BstUI are shown in capital letters

Although RASSF1A mutations appear to be rare in other tumour types in which epigenetic inactivation is common (e.g., lung and breast), we proceeded to analyse six phaeochromocytomas with 3p21.3 allele loss and no RASSF1A methylation and four tumours with RASSF1A methylation and no 3p21.3 LOH (but LOH on other chromosomes) for somatic RASSF1A mutations as described in Agathanggelou et al. (2001). No inactivating mutations were found.

RASSF1A methylation status in neuroblastoma

In view of the common neural crest lineage of phaeochromocytoma and neuroblastoma, we proceeded to determine RASSF1A promoter methylation status in 67 primary neuroblastomas analysed previously for 3p LOH (Ejeskar et al., 1998). 55% (37/67) neuroblastomas demonstrated de novo RASSF1A CpG island promoter methylation (Figure 1b). To determine the pattern of CpG methylation within the fragment analysed, five individual clones from two methylated tumours were sequenced. As for phaeochromocytoma the majority of the 16 CpG dinucleotides in the fragment were methylated in each tumour, but there was clearly heterogeneity in many tumours. For example for tumour NB157, clone 2 had all 16 CpGs methylated, whilst in clone 1 CpGs 10, 12 and 14 were not methylated and in clone 3 CpG 13 was not methylated. We also analysed two neuroblastoma tumour lines (SK-N-SH and SK-N-FI) for RASSF1A methylation, both tumour lines were found to be completely methylated.

To investigate the relationship between RASSF1A methylation status and 3p21.3 allele loss, we typed each tumour for LOH at D3S4604 which maps close to RASSF1A. 16% (6/37) of informative tumours demonstrated 3p21.3 allele loss. All six tumours with 3p21.3 allele loss had RASSF1A methylation. In addition, 16 tumours with no LOH at 3p21.3 were methylated at RASSF1A (five of these 16 tumours had loss on other chromosomes (Martinsson et al., 1997a). The frequency of RASSF1A methylation in neuroblastomas with 3p21.3 allele loss was higher (6/6) than in tumours without 3p21.3 LOH (16/31) although it did not reach statistical significance (P=0.06).

To investigate whether somatic RASSF1A mutations might provide a ‘second-hit’ in neuroblastomas with RASSF1A methylation and no 3p21.3 allele loss we analysed five neuroblastomas with LOH on other chromosomes (Martinsson et al., 1997a) (we also analysed five further tumours with RASSF1A methylation and no 3p21.3 loss and no loss at other chromosomes) by SSCP. However, no inactivating mutations were found.

To determine the relationship of RASSF1A promoter methylation in the neuroblastoma cell lines SK-N-SH and SK-N-FI and RASSF1A expression, we treated the cells with the demethylating agent 5-azacytidine for 7 days. 5-azacytidine treatment restored RASSF1A expression in SK-N-SH and SK-N-FI cells, but had little or no change in the expression of RASSF1C after the 5-azacytidine treatment (Figure 2).

Figure 2
figure2

Expression of RASSF1A and RASSF1C before and after treatment of neuroblastoma cell lines SK-N-SH and SK-N-FI with 5-azacytidine. Both cell lines expressing RASSF1C but not RASSF1A by RT–PCR were grown in Eagle's Minimal Essential medium in the presence(+) and absence(−) of 5 μM 5-azacytidine for 7 days. RT–PCR primers and conditions described in Burbee et al., 2001

CASP8 methylation status in neuroblastoma

To investigate the relationship between RASSF1A and CASP8 methylation in neuroblastoma, methylation-specific PCR was carried out as described by Teitz et al. (2000). We found that 24 of 60 (40%) neuroblastomas were methylated for a CpG island located in the 5′ region of CASP8 (Figure 3). Since none of the neuroblastomas in this study were microdissected we always saw the PCR product from the unmethylated primer sets in every tumour analysed. Nineteen of 34 neuroblastomas with RASSF1A methylation also demonstrated CASP8 methylation but only four of 24 neuroblastomas without RASSF1A methylation demonstrated CASP8 methylation (P=0.0031).

Figure 3
figure3

Methylation- specific PCR for the detection of CASP8 methylated and unmethylated sequences in primary neuroblastoma samples. The primers specific for methylated and unmethylated sequences and PCR conditions are from Teitz et al., 2000. The neuroblastoma sample number is on top, both methylated and unmethylated CASP8 alleles are seen in all cases. Size of the PCR product corresponding to methylated CASP8 alleles (321 bp) and unmethylated CASP8 alleles (322 bp) are on the left margin of each panel

Neuroblastoma stage, prognosis and RASSF1A methylation status

We compared the RASSF1A methylation status in our tumour series to the results of investigations to determine N-myc amplification, 1p allele loss and 17q gain reported previously (Martinsson et al., 1997a). No correlation was detected between RASSF1A methylation and the presence or absence of any these parameters and no clear association between RASSF1A methylation and tumour stage or survival was detected.

The RASSF1A TSG is epigenetically silenced by promoter methylation in many tumour types (Dammann et al., 2000; Agathanggelou et al., 2001; Burbee et al., 2001; Morrissey et al., 2001). We found that RASSF1A was methylated in the majority of primary neuroblastomas analysed and demonstrated that RASSF1A promoter methylation is associated with transcriptional silencing (of RASSF1A but not RASSF1C) in SK-N-SH and SK-N-FI neuroblastoma cell lines. As in most other tumour types in which RASSF1A methylation has been reported, there was no evidence that somatic RASSF1A mutations are a frequent event in neuroblastoma or phaeochromocytoma. All 6 neuroblastomas with 3p21.3 allele loss also demonstrated RASSF1A methylation of the remaining allele consistent with Knudson's two hit model of tumorigenesis and with RASSF1A being the major 3p21.3 neuroblastoma TSG. However, many neuroblastomas with RASSF1A methylation did not show 3p21.3 LOH. This differs from lung cancer in which most small cell lung carcinomas (SCLC) with RASSF1A methylation also have 3p21.3 LOH. The presence of contaminating normal tissue in neuroblastoma tumors makes it difficult to determine whether RASSF1A methylated tumour cells have biallelic or monoallelic methylation. However, two neuroblastoma cell lines showed evidence of complete RASSF1A methylation consistent with homozygous inactivation by biallelic methylation or monoallelic methylation and allele loss. There is increasing evidence that TSG haploinsufficiency may promote tumourigenesis per se without the need for a second hit. Unfortunately neuroblastomas and phaeochromocytomas (n=20) with RASSF1A methylation and no 3p21.3 LOH were not informative at RASSF1A promoter SNPs (Agathanggelou et al., 2001) so it was not possible to determine directly whether tumour methylation was bi- or monoallelic.

Recently CASP8 was shown to be frequently inactivated in neuroblastomas by DNA methylation and gene deletion (Teitz et al., 2000). In our series of neuroblastomas we detected CASP8 methylation in 40% of neuroblastomas and demonstrated a significant correlation between RASSF1A and CASP8 methylation (P=0.0031), suggesting that a subset of neuroblastoma might have a CpG island methylator phenotype (CIMP) as described in colorectal cancer (Toyota et al., 1999). Further studies are required to determine if CIMP is one of the major pathways that contribute to neuroblastoma tumorigenesis. Such a finding would be significant because of the potential to develop novel forms of therapy with demethylating agents.

Phaeochromocytoma and neuroblastoma are both derived from the neural crest. Phaeochromocytomas arise from chromaffin cells derived from primitive cells of the neural crest that migrate into the paravertebral sympathetic ganglia and from there into the adrenal primordium, whilst neuroblastomas are composed of histologically primitive neuronal cells. Gene expression patterns in neuroblastoma and phaeochromocytoma reflect the differing origins of the tumours (particularly for undifferentiated neuroblastomas) (Hoehner et al., 1998). We note that although 3p allele loss is reported more frequently in phaeochromocytoma than in neuroblastoma, RASSF1A methylation was more common in the latter. The patterns of 3p allele loss in phaeochromocytoma suggested that multiple 3p TSGs may contribute to tumour development, whereas RASSF1A appears to have a major role in neuroblastoma. RASSF1A methylation is frequent in SCLC, but the existence of further 3p SCLC TSG(s) is indicated by the finding of homozygous deletions at 3p12 in SCLC cell lines (Latif et al., 1992; Sundaresan et al., 1998). We note that one phaeochromocytoma without 3p21.3 allele loss did demonstrate 3p12 LOH within the 3p12 critical interval. Alternatively RASSF1A haploinsufficiency (by 3p21 allele loss) might promote phaeochromocytoma tumourigenesis without the need for a ‘second hit’.

RASSF1 has a predicted RAS association domain and was recently shown to be an effector of ras both in vivo and in vitro studies (Vos et al., 2000). Evidence for other putative functions, e.g. a phosphorylation target for ATM (Kim et al., 1999), have been suggested and the RASSF1A isoform that is methylated in various tumour types, also contains a predicted diacylglycerol (DAG) binding domain homologous to that found in the related gene NORE1. Inactivation of RASSF1A in neural crest tumours is consistent with other evidence implicating aberrations of RAS signalling pathways in neuroblastoma and phaeochromocytoma. Thus germline mutations and somatic mutations in NF1 gene can be associated with phaeochromocytoma and neuroblastoma and the NF1 protein, neurofibromin, downregulates RAS signaling by GTPase activity (Martinsson et al., 1997b; Klose et al., 1998; Hiatt et al., 2001). In addition, although RAS gene mutations are uncommon in neuroblastoma (Moley et al., 1991), multiple lines of evidence suggest that the RAS signalling pathway is an important modulator of cell growth and differentiation and expression of RAS proteins has been associated with a good prognosis in neuroblastoma (Tanaka et al., 1998). We have demonstrated that RASSF1A is inactivated frequently in early and late stage neuroblastomas. This finding provides opportunities to develop novel forms of therapy by reversing RASSF1A methylation silencing or correcting the effects of RASSF1A downregulation on RAS signal transduction pathways. The high frequency of RASSF1A inactivation by promoter region hypermethylation in neuroblastoma suggests that it is the critical 3p TSG involved in neuroblastoma development.

References

  1. Agathanggelou A, Honorio S, Macartney DP, Martinez A, Dallol A, Rader J, Fullwood P, Chauhan A, Walker R, Shaw JA, Hosoe S, Lerman MI, Minna JD, Moher ER,, Latif F . 2001 Oncogene 20: 1509–1518

  2. Astuti D, Douglas F, Eng C, Lennard T, Aliaganis I, Woodward ER, Evans G, Latif F, Maher ER . 2001 Lancet 357: 1181–1182

  3. Burbee DG, Forgacs E, Zochbauer S, Shivakumar L, Gao B, Randle D, Virmani A, Bader S, Sekido Y, Latif F, Gong K. Gazdar AF, Lerman MI, White M, Minna JD . 2001 JNCI. 93: 691–699

  4. Dammann R, Li C, Yoon JH, Chin PL, Bates S, Pfeifer GP . 2000 Nat. Genet. 25: 315–319

  5. Ejeskar K, Aburatani H, Abrahamsson J, Kogner P, Martinsson T . 1998 Br. J. Cancer 77: 1787–1791

  6. Eng C, Clayton D, Schuffenecker I, Lenoir G, Cote G, Gagel RF, van Amstel HK, Lips CJ, Nishisho I, Takai SI . 1996 JAMA 276: 1575–1579

  7. Fullwood P, Marchini S, Rader JS, Martinez A, Macartney D, Broggini M, Morelli C, Barbanti-Brodano G, Maher ER, Latif F . 1999 Cancer Res. 59: 4662–4667

  8. Hiatt KK, Ingram DA, Zhang Y, Bollag G, Clapp DW . 2001 J. Biol. Chem. 276: 7240–7245

  9. Hoehner JC, Hedborg F, Eriksson L, Sandstedt B, Grimelius L, Olsen L, Pahlman S . 1998 Lab. Invest. 78: 29–45

  10. Kim ST, Lim DS, Canman CE, Kastan M.B . 1999 Biol. Chem. 274: 37538–37543

  11. Klose A, Ahmadian MR, Schuelke M, Scheffzek K, Hoffmeyer S, Gewies A, Schmitz F, Kaufmann D, Peters H, Wittinghoferm A, Nurnberg P . 1998 Hum. Mol. Genet. 7: 1261–1268

  12. Latif F, Tory K, Modi WS, Graziano SL, Gamble G, Douglas J, Heppell-Parton AC, Rabbitts PH, Zbar B, Lerman MI . 1992 Genes Chromosomes Cancer 5: 119–127

  13. Lerman MI, Minna JD . for The International Lung Cancer Chromosome 3p21.3 Tumor Suppressor Gene Consortium. 2000 Cancer Res. 60: 6116–6133

  14. Maris JM, Matthay KK . 1999 J. Clin. Oncol. 17: 2264–2279

  15. Martinsson T, Sjoberg RM, Hallstensson K, Nordling M, Hedborg F, Kogner P . 1997a Eur. J. Cancer 33: 1997–2001

  16. Martinsson T, Sjoberg RM, Hedborg F, Kogner P . 1997b Cancer Genet. Cytogenet. 95: 183–189

  17. Moley JF, Brother MB, Wells SA, Spengler BA, Biedler JL, Brodeur GM . 1991 Cancer Res. 51: 1596–1599

  18. Morrissey C, Martinez A, Zatyka M, Agathanggelou A, Honorio S, Astuti D, Morgan NV, Moch H, Richards FM, Kishida T, Yao M, Schraml P, Latif F, Maher ER . 2001 Cancer Res. In press

  19. Sundaresan V, Chung G, Heppell-Parton A, Xiong J, Grundy C, Roberts I, James L, Cahn A, Bench A, Douglas J, et al . 1998 Oncogene 17: 1723–1729

  20. Tanaka T, Sugimoto T, Sawada T . 1998 Cancer 83: 1626–1633

  21. Teitz T, Wei T, Valentine MB, Vanin EF, Grenet J, Valentine VA, Behmm FG, Look AT, Lahti JM, Kidd VJ . 2000 Nat. Med. 6: 529–535

  22. Toyota M, Ahuja N, Ohe-Toyota M, Herman JG, Baylin SB, Issa JP . 1999 Proc. Natl. Acad. Sci. USA 96: 8681–8686

  23. Vargas MP, Zhuang Z, Wang C, Vortmeyer A, Linehan WM, Merino MJ . 1997 Hum. Pathol. 28: 411–415

  24. Vos MD, Ellis CA, Bell A, Birrer MJ, Clark GJ . 2000 J. Biol. Chem. 275: 35669–35672

  25. Wallace MR, Marchukm DA, Andersen LB, Letcher R, Odeh HM, Saulino AM, Fountain JW, Brereton A, Nicholson J, Mitchell AL, et al . 1990 Science 249: 181–186

  26. Woodward ER, Eng C, McMahon R, Voutilainen R, Affara NA, Ponder BA, Maher ER . 1997 Hum. Mol. Genet. 6: 1051–1056

  27. Zeiger MA, Zbar B, Keiser H, Linehan WM, Gnarra JR . 1995 Genes Chromosomes Cancer 13: 151–156

Download references

Acknowledgements

This work is supported in part by British Heart Foundation, Association for International Cancer Research, and Cancer Research Campaign. S Honorio is supported by Fundacao para a Ciencia e a Technologia. Drs Martinsson and Kogner supported by funds from the Swedish Cancer Society and Swedish Children's Cancer Foundation.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Farida Latif.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Astuti, D., Agathanggelou, A., Honorio, S. et al. RASSF1A promoter region CpG island hypermethylation in phaeochromocytomas and neuroblastoma tumours. Oncogene 20, 7573–7577 (2001). https://doi.org/10.1038/sj.onc.1204968

Download citation

Keywords

  • methylation
  • 3p tumour suppressor gene
  • neuroblastoma
  • phaeochromocytoma

Further reading

Search