Protocol Online logo
Top : Forum Archives: : DNA Methylation, Histone and Chromatin Study

Preferential methylation of specific CpG sites? - (Apr/25/2007 )

Hi,

I just recently did some quantitative analysis of a portion of the RASSF1 promotor in normal luminal breast epithelial cells and the results I got are somewhat confusing. Looking at a segment of about 9 CpG sites, 7 of them seemed to have methylation in the 15-20% range and 2 sites had methylation at 50-60%. Has anyone noticed a trend like this in quantitative analysis before? Or would I be safe in assuming that some sort of technical error may have occurred? I'm planning on getting this particular sample re-sequenced, but talking to Dr. Jean-Pierre Issa at AACR he seemed to suggest that such a phenomenon is extremely rare and that usually the entire CpG island gets methylated to relatively the same degree.

Does anyone else have a take on this?

-cancergeek-

I have, but not in the case for cancer. Many cases in cancer, and I would say Issa is a champion for it, is the CIMP phenotype.

NOS gene is regulated by the methylation of one CpG dinucleotide and we are beginning to see that the methylation of particular sites within a CpG island are the main "switch" for gene expression and they tend to fall within a transcription factor binding site. This is probably what you are seeing here cancer geek.

are you looking at a homogeneous tumor sample/cell line? what you are seeing could be an affect of a mixed population of cells.

Nick

-methylnick-

QUOTE (cancergeek @ Apr 26 2007, 04:50 AM)
Hi,

I just recently did some quantitative analysis of a portion of the RASSF1 promotor in normal luminal breast epithelial cells and the results I got are somewhat confusing. Looking at a segment of about 9 CpG sites, 7 of them seemed to have methylation in the 15-20% range and 2 sites had methylation at 50-60%. Has anyone noticed a trend like this in quantitative analysis before? Or would I be safe in assuming that some sort of technical error may have occurred? I'm planning on getting this particular sample re-sequenced, but talking to Dr. Jean-Pierre Issa at AACR he seemed to suggest that such a phenomenon is extremely rare and that usually the entire CpG island gets methylated to relatively the same degree.

Does anyone else have a take on this?



Someone else in my lab has used BSP primers to amplify TIMP3 CpG island in esophagus caner epithelial cells. He designed 3 BSP primers to amplify 3 different regions in the island. The result showed that only one region was methylated while the other two were not. So it may be true that sometimes not entire CpG island gets methylated. Or maybe you can use a different methlytion analysis method to confirm it?

Just out of curiosity, how do you do a quantitative analysis of levels of methylation? I only know the PCR method, which tells methylation but not in a quantitative way. Can you explain in some details cause I’ve only been working on the methylation area for several months?

-Telmimore-

QUOTE (Telmimore @ Apr 25 2007, 09:25 PM)
QUOTE (cancergeek @ Apr 26 2007, 04:50 AM)
Hi,

I just recently did some quantitative analysis of a portion of the RASSF1 promotor in normal luminal breast epithelial cells and the results I got are somewhat confusing. Looking at a segment of about 9 CpG sites, 7 of them seemed to have methylation in the 15-20% range and 2 sites had methylation at 50-60%. Has anyone noticed a trend like this in quantitative analysis before? Or would I be safe in assuming that some sort of technical error may have occurred? I'm planning on getting this particular sample re-sequenced, but talking to Dr. Jean-Pierre Issa at AACR he seemed to suggest that such a phenomenon is extremely rare and that usually the entire CpG island gets methylated to relatively the same degree.

Does anyone else have a take on this?



Someone else in my lab has used BSP primers to amplify TIMP3 CpG island in esophagus caner epithelial cells. He designed 3 BSP primers to amplify 3 different regions in the island. The result showed that only one region was methylated while the other two were not. So it may be true that sometimes not entire CpG island gets methylated. Or maybe you can use a different methlytion analysis method to confirm it?

Just out of curiosity, how do you do a quantitative analysis of levels of methylation? I only know the PCR method, which tells methylation but not in a quantitative way. Can you explain in some details cause I’ve only been working on the methylation area for several months?


I have been utilizing pyrosequencing which is a luciferase based method for quantitative detection of the amount of dNTP incorporation. You can find more information about it at www.pyrosequencing.com. And to answer your question Methylnick I am examining an enriched population of normal luminal breast epithelial cells isolated from breast via magnetic antibody separation.

-cancergeek-

@cancergeek.

Hmmmm seems like an interesting result. as a general rule, CpG islands are completely methylated or are either completely unmethylated. But having said that, there are exceptions. RASSF1 is a well characterised loci for methylation I would say.

Assuming that what you are seeing is real, could you replicate the result with another method other than pyrosequencing? I am thinking a clonal analysis with bisulfite sequcning.

Nick

-methylnick-

The hippocampus specific exon1,7 of the GR receptor in rats is controlled by a difference in the methylation of 1 excl.gif CpG within the CpG island within this exon... TFs may by the answer to this problem.
But also a technical problem may oocur. I have seen something similar in the BDNF gene promoter in the rat, where a CpG island contains a large stretch of about 10 CpG's. In all samples, even the mixing experiments, I had an increased methylation reading in the middle of the stretch. I am pretty much sure that there is a problem within the PCR steps.

K.

-krümelmonster-

Where does this "rule of thumb" come from? Can anyone direct me to some references? In my experience of looking at CpG islands (7 genes 10 different tissues from the same animals) that sometimes the amplicons that I am looking at (containing up to 50 CpG sites) are methylated similarly along the length of the CpG island, but in other islands there are vast (very repeatable) differences with certain CpG sites always (regardless of tissue or developmental stage) being vastly different to others on the same amplicon. From the data that I have (most generated using Sequenom MassArray technology) I wouldn't say that rule of thumb holds.

Just my experience.

-ultragirl-

I also think this rule of thumb doesn't hold. As emntioned above, there are examples for one CpG within the whole island making the difference ... I could imagine, that this rule of thumb has something to do with the clonal approach and that the "newer" quantitative methods are more accurately answering this question.

-krümelmonster-

Well I don't believe in the "rule of thumb", I believe most people in the field are basing this rule on the early experiments in DNA methylation by Bird and collegues where they found CpG islands to be in the HpaII tiny fraction which would mean that they are unmethylated.

This is then extrapolated to include all CpG islands in general by most people in the field and I believe is exemplified by the MSP assay which I think many people wrongly use to measure methylation across the whole island.

It's just an observation/and feeling I have gotten from the reviewers of my paper that the rule of thumb holds. There are many "rule of thumbs" with regard to methylation which I believe are wrong also.

Nick

-methylnick-

Thanks MethylNick, the words you wrote are what I have been trying to convince people of (pretty much exactly your words) since I started generating this data! One day they will listen :-)

-ultragirl-