Hi Sebastien,

Only to make sure I understand your question. You have one single SNP that you think determines the methylation state of your promoter and its transcriptional status. Is it gene-rich or gene-poor CGI promoter? To my knowledge none of the DNA methylases have any known DNA binding specific sequence. That leaves you with two alternatives; either you have a change in the recognition site of a specific transcription factor that you can identify using bioinformatic tools such TRANSFACT (or others that some contributors will suggest to you) or the SNP induces a structural change in your promoter that makes it refractory to methylation, which is unlikely. 1 nucleotide change wouldn’t specify whether or not a nucleosome could be accommodated but I am not sure of what may define the deposition of H2AZ on proximal promoters. Also, there are conflicting reports for example on whether or not methylases have preferences of CpG located on internucleosomal regions over the ones located on nucleosomes. I guess in vivo, the association of DNMTs with remodeling factors could easily overcome this barer. I think you have a nice question to which you have to experimentally answer to.

Hi Sébastien, nice userpic ;)

If your SNP has an rs number, you can look up at chromatin immunopreciptation (ChIP) ENCODE data at UCSC and maybe thereby narrow your search for potential factors which can bind there..

Hi Maria and Halim,

Many thanks for your comments.

No, there is no rs number for this variant, because it's very rare. Yet, there are several other variants very close; one even concerns the nucleotide upstream. I will have look at ENCODE data.

And, yes, we have no proof that the two observations are linked, but the hypermethylated allele is associated with this rare SNV. This one has already been described by another team, which noted this relationship between the presence of the variant and the dominant mode of inheritance of the promoter hypermethylation. Yet, this team didn't or couldn't explain the link between the two observations. We suspect therefore that the primary cause is a genetic anomaly transmitted according to a dominant mode of inheritance, which has an impact on the hypermethylation of the promoter.

This promoter presents a CpG island divided into four parts. It's quite well-characterized.

According to bioinformatics, the variant is predicted to affect a transcription factor -it is not predicted by all bioinformatic programmes though-.

You're right, Halim, that only experimental studies could help to solve this question. This very little change in the DNA sequence might induce a change in conformation that would alter the methylation process. I find it very interesting, but very challenging in the same time.

Dear Sebastien,

The simplest explanation is that this variant disrupts binding/recognition of some factor that blocks methylation of this region - as you have already somewhat predicted based on bioinformatics. If your TF is blocking CpG methylation in its absence (caused by the variant) that region could be aberrantly methylated. The obvious experiment would be to inhibit access of that TF to the site...either by making mutations at the site, knocking down levels of the TF or in some other way blocking it from binding.

Please see Klengel, et al. Nature Neuroscience, 2013, "Allele-specific FKBP5 DNA demethylation mediates gene–childhood trauma interactions" for a specific example.

Pubmed link: http://www.ncbi.nlm.nih.gov/pubmed/23201972

This isn't a promotor region SNP but an intron based transcription factor binding element SNP. Basically it changes 3-D conformation of the POLII binding of the transcription start site. This leads to demethylation uder certain environmental conditions. Too involved to describe here. The paper is quite nice.

Methylation can be neotype specific. THere are a number of studies that have now combined genomewide methylation an SNP data to identify "methQTLs". For example see:

Bell, J.T., et al. DNA methylation patterns associate with genetic and gene expression variation in HapMap cell lines. Genome Biol 12, R10 (2011).

Gibbs, J.R., et al. Abundant quantitative trait loci exist for DNA methylation and gene expression in human brain. PLoS Genet 6, e1000952 (2010).

Kerkel, K., et al. Genomic surveys by methylation-sensitive SNP analysis identify sequence-dependent allele-specific DNA methylation. Nat Genet 40, 904-908 (2008).

We have observed this phenomenon as well in several genes where methylation at specific CpGs is related to genotype and an adjacent SNP.

is the phenomenon of hypermthylation only seen in this gene or is it widespread in the genome?

Dear Folami, Jeanette and John,

Thnak you very much for these very interesting comments and for the references. I have just started to read the articles. I wonder how frequent this phenomenon can be, as it enables to understand cases which could not be explained by clearly deleterious mutations, regarding for instance familial cancers. John, the references you gave suggest that the phenomenon is probably not so rare. With exome sequencing, we find a increasing number of rare variants, notably in promoting regions. With exome sequencing, teams find an increasing number of rare variants, notably in promoting and non-coding regions. There is no doubt that numerous observations of this type will reported soon for various pathologies. Conceptually, I find very exciting this bridge between genetics and epigenetics .

Hi Sebastien,

That is a good point. But I think association is only that until we find a mechanistic answer for the phenomenon or pathology we observe. It was funny for me to hear on BBC1 the analogy sometimes we make from statistical inferences. We may end up saying that sharks love ice-cream and hence their number and attacks increase in summer.

Is the SNP on the promoter present on both parental alleles? Maybe it is worth to browse on genome 1000 data base where there are over 38 millions computed SNPs between the recently sequenced 1000 genomes and see what you can find.

I agreed with all of your opinion. Our research group is also investigating the association between genetics and epigenetics. If the SNP has a miRNA, thereby it will have a different expression of the mature miRNA. Thereafter, the miRNA may regulate expression of DNMTs and cause aberrant methylation pattern. But I think this aspect will cause not only gene specific methyltion changes, but global methylation changes.

See Lienert et al Nat Genet. 43(11):1091-7. A brief model is displayed in Fig 5d.

Dear Wei, this complex relationship between SNV, miRNA regulation and methylation is also quite interesting. It should indeed concern a more comprehensive methylation change, as you suggested.

Thanks Jon for this additional reference. I will read it rightaway.

You’re true Halim. It’s easy to take a shortcut. And when we’re focused on a topic, we have a tendancy to hear and read about it everywhere. It’s also true that it’s often when you really spend time and energy to search something that you find it; otherwise, when you don’t care much about it, you miss it (thanks Lapalice).

Anyhow, I’ve seen that several teams focused on the association between genotype and methylation. For further reading, I found a few other relevant references:

Quon G et al. Nucleic Acids Res. 2013 Feb 1;41(4):2095-104.

Heijmans BT et al. Hum Mol Genet. 2007 Mar 1;16(5):547-54.

Drong AW et al. PLoS One. 2013;8(2):e55923.