Although some of the older papers had suggested that hypermethylation and (global) hypomethylation are concurrent the more recent publications suggest that they can be quite independent of each other, depending on the cancer type (as reviewed by Peter Molloy e.g. 2007). Most recent papers use bead arrays that detect hypermethylation and hypomethylation at individual CpG sites in many genes (CpG islands only or more general, depending on the array version). These are usually analyzed by clustering methods that reveal correlations between some events - hyper or hypo. I am not sure to which extent the "genome-wide" data published so far (using arrays or WGS) allows an estimate of the overall relation between hypermethylation and hypomethylation, though, because only a fraction of the repeat sequences can be analyzed. You might like to look up some of the papers by the Laird or Feinberg groups, for instance, in this regard.

Dear Wenbing,

Cancer=!Cancer

Most cancer cells have specific methylation pattern, and this pattern can be used to classify them. But if you look on the precancerous cells, your question is very important. Which onkogene promoter are demethylated, which tumorsuppressor gene is methylated, which DNA repair gene is downregulated by dna methylation, and so on.... ; which is first, second,..

A current hypothesis suggests that oxidative stress might be a major factor in the development of a precancerous state. DNA methylation depletion is a consequence and therefore global DNA hypomethylation. It is known that tumor cells have global DNA hypomethylation that can be as high as 60% less than their normal counterparts.

IMHO a bunch of labs are currently trying to answer this question, which genes are hypo/hyper methylated in the stage of tumor initiation.

regards, Olivier

Patterns of cellular and HPV 16 methylation as biomarkers for cervical neoplasia

(2012) Journal of Virological Methods, 184 (1-2), pp. 84-92.

Epigenetics of cervical cancer. An overview and therapeutic perspectives

(2005) Molecular Cancer, 4, art. no. 38, 70 p.

Dear Olivier,

Thank you for you detailed answer. My question is addressing the initial stage of tumorigenesis. I think it's a very fascinating question.

Best, Wenbing

Dear Wenbing,

Maybe this paper might help you

http://www.ncbi.nlm.nih.gov/pubmed/22690664

Dear Elizabeth,

Thank you so much.

I doubt anybody can confidently say a pattern of DNA hypermetylation and hypomethylation is tumorigenic unless it leads to a clonal change. You can only compare the pattern of your cancer cells to their nearby normal cells to say a pattern is associated with cancer transformation.

To my knowledge, the methylation state is crucial for the normal functionality of the cell. For example a very tight control of SAMe (principal methyl donor molecule in cells) is needed for normal state. Chronic defect in SAMe follows hepatic injury for example and leads to hepatocellular carcinoma, and the same happens with a chronic excess of this molecule, hepatic injury and spontaneous development of HCC. So yes, methylation is critical in cancer, no doubt. There is some mouse models (MAT1A-KO and GNMT-KO) revealing that, most of these works done in liver, the metabolic organ of excellence.

So, to determine different methylation patterns in DNA of cancer cells is ok, but why not check the ethiology first?

Cheers!

Although some of the older papers had suggested that hypermethylation and (global) hypomethylation are concurrent the more recent publications suggest that they can be quite independent of each other, depending on the cancer type (as reviewed by Peter Molloy e.g. 2007). Most recent papers use bead arrays that detect hypermethylation and hypomethylation at individual CpG sites in many genes (CpG islands only or more general, depending on the array version). These are usually analyzed by clustering methods that reveal correlations between some events - hyper or hypo. I am not sure to which extent the "genome-wide" data published so far (using arrays or WGS) allows an estimate of the overall relation between hypermethylation and hypomethylation, though, because only a fraction of the repeat sequences can be analyzed. You might like to look up some of the papers by the Laird or Feinberg groups, for instance, in this regard.

Dear Wenbing

i think your question is indeed fascinating. My basic understanding is that hypomethylated genes are overexpressed and hypermethylated genes silenced, in cancer. This is not always the case and that is why in order to study in details the correlation of methylation status and expression of a specific gene/genes, one should always combine methylation and expression analysis.

However, one could "predict" that genes which help tumor growth will be hypomethylated in most cancers and vice versa.

i am sending you a link for a very interesting paper on methylation status and expression analysis in pancreatic cancer:

http://clincancerres.aacrjournals.org/content/17/13/4341.long

All the papers from Michael Goggins group are very informative about methylation and cancer.

Regards

Yiannis

The concepts of Hypermeth and Hypo meth depend on the geographical landscape of DNA you are looking at. In NORMAL cells - promoters are unmeth and ORFs are methylated HOWEVER, In Cancers - THIS is reversed. Promoter proximal CpG islands get hypermethylated while ORFs get HYPOmethylated. So yes - this is not a LAW!! there will be exception. But in cancer. majority of tumor suppressors - Hypermeth of CpG ilsands at promoters is seen - while  at Shores/ORFs hypomethylation is observed. The hypomethylation has more variability though which is called by Feinberg group as DIFFERENTIALLY methylated Regions. SO they BOTH happen in Cancers, but has geographical context.