© 2003 Nature Publishing Group

86 | FEBRUARY 2003 | VOLUME 3 www.nature.com/reviews/cancer

Epigenetic modifications, such as DNA hyper-methylation and histone deacetylation, arecommonly detected in human cancer cells,where they have been shown to downregulate

tumour-suppressor genes. Sidransky and col-leagues have been using a combination of tech-niques to identify additional genes that aresilenced in this way, and have subsequentlyuncovered three new tumour suppressors.

Sidransky’s group treated oesophageal squa-mous-cell carcinoma (ESCC) cells with phar-macological agents to prevent the epigeneticsilencing of genes — cells were treated with 5-aza-2′-deoxycytidine to block DNA methyla-tion, and then with trichostatin A to inhibit his-tone deacetylase. This treatment reverses theformation of transcriptionally repressive chro-matin structures at methylated promoters.Microarray analysis was then used to comparegene-expression patterns between transcrip-tionally silenced and unsilenced cancer cells. Atotal of 58 silenced genes were identified by thisapproach: 44 (76%) were found to containdense CpG islands — well-known methylationsites — in their promoter regions. Some 13 of22 tested gene promoters were methylated inthe ESCC cell lines, and 10 of these genes werefound to be downregulated at the mRNA level.

Three of the genes — CRIP1, APOD andNMU — had growth-suppressive activity whentransfected into ESCC cells, as determinedusing colony-formation assays. NMU is pro-posed to be involved in normal oesophagealmucosa integrity, and its receptor, FM3, is a G-protein-coupled receptor that can signalthrough phosphatidylinositol 3-kinase-γ —recently shown to block the growth of humancolon cancer cells. CRIP1 is believed to be atranscription factor that was shown to induceapoptosis in cancer cell lines, and APOD hasbeen associated with cell growth arrest.

The epigenetically silenced genes werefound to cluster in specific chromosomalregions. Many of these loci — such as 3q26,4q12 and 14q24 — have also been shown toharbour chromosome deletions of loss of het-erozygosity in cancer cells. This approachcould therefore be used to identify newtumour-suppressor genes and loci in othertypes of cancer cell.

Kristine Novak

References and linksORIGINAL RESEARCH PAPER Yamashita, K. et al.Pharmacological unmasking of epigenetically silenced tumorsuppressor genes in esophageal squamous cell carcinoma.Cancer Cell 2, 485–495 (2002)FURTHER READING Sidransky, D. Emerging molecularmarkers of cancer. Nature Rev. Cancer 2, 210–219 (2002)WEB SITEDavid Sidransky’s lab: http://urology.jhu.edu/faculty/sidransky/

As its name suggests, the HIC1(hypermethylated in cancer 1) gene ishypermethylated and hence transcriptionallysilent in several types of cancer, indicatingthat it might act as a tumour suppressor.Now, Stephen Baylin and colleagues havedeveloped a mouse model to furtherinvestigate this possibility, and show thatcomplete loss of Hic1 predisposes mice to agender-dependent array of tumour types.

The authors generated Hic1+/– mice andinvestigated the incidence of tumourformation. The heterozygous mice began todevelop tumours at 70 weeks, and by 100weeks 34.2% had developed malignanttumours, compared with 14.2% of the wild-type mice. Surprisingly, the tumour typesthat were identified differed according to thesex of the mice — male heterozygotesdeveloped carcinomas at an increasedfrequency, whereas female heterozygotesmostly developed lymphomas and sarcomas.The tumours that developed in Hic1heterozygotes were also more aggressive thanthose that developed in wild-type mice.

So, was the remaining wild-type allele lostin these cancers? Gross chromosomaldeletions were not observed, so the authorsinvestigated the methylation status of theHic1 promoter. As in humans, Hic1 has twoalternative promoters, 1a and 1b; it is 1bthat seems to be hypermethylated in mosthuman cancers. In mice, however,methylation-specific polymerase chainreaction revealed that the 1a promoter ispredominantly methylated in mostlymphomas and sarcomas. All carcinomasfrom heterozygous mice that do not have amethylated 1a promoter were found tocontain dense methylation of the 1bpromoter, and, in total, 87% of malignanttumours from the heterozygous mice haddense methylation of one of the alternativepromoters. This promoter methylationcorresponded with a lack of Hic1 protein, asit does in human cells.

HIC1 is therefore the first candidatetumour suppressor that is transcriptionallysilenced, rather than mutated, and that acts as a tumour suppressor in mice. The

identification of other such genes, and theirconfirmation using mouse models, will nodoubt follow.

Emma GreenwoodReferences and links

ORIGINAL RESEARCH PAPER Chen, W. Y. et al.Heterozygous disruption of Hic1 predisposes mice to agender-dependent distribution of malignant tumors. NatureGenet. 33, 197–202 (2003)WEB SITEStephen Baylin’s lab:http://www.hopkinsmedicine.org/graduateprograms/cmm/baylin.html

Gender-dependent tumour suppression

E P I G E N E T I C S

Breaking the silence

E P I G E N E T I C S

H I G H L I G H T S