Bisulfite tx'd DNA sequence analysis - what to do with CG to TA conversion - bisulfite sequencing analysis (Nov/10/2007 )

Hello all. I've come across your Bioforum and have read through many of your topics - very useful.

I am having difficulties analyzing some sequences using MethTools for CpG methylation status and was hoping perhaps one or more of you could provide some guidance.

I treated genomic DNA with the Zymo EZ DNA methylation kit for a control sample and my test sample (mice).

The DNA was then eluted in water.

Primers were designed using MethPrimer for a sequence I call DP341. I had previously sequenced 12 clones for this DP341 on non-bisulfite tx'd DNA with primers (not MethPrimers) in the same region however, with 10 of the clones having 100% identity (2 clones each had 1 SNP).

I PCR'd the bisulfite tx'd DNA and obtained a specific DP341 product. The PCR product was cleaned and subcloned into pGEM. Insert positive clones were sequenced.

The bisulfite treatment was successful as all Cs not in CpGs were converted to Ts in both control and test samples.

My problem arises in that some of my clones have a TA "SNP" in place of either a CG or a TG. I do not see this variation in my non-bisulfite treated DNA.

I know that methylated CpGs can mutate to TAs over time. Why would I see this conversion from a CG to a TA only in the bisulfite treated DNA samples?

There are a few clones in which the normal nucleotide sequence should be a AG and yet some of my bisulfite treated clones have an AA.

I am willing to provide a text file of sequences or PDF file of the alignment if anyone thinks they have ideas to help me!

This SNP issue is posing a problem analyzing the methylation status of my clones using MethTools.

Thanks for reading this long topic and for your anticipated ideas.

hi there harley,

what polymerase are you using? we are using normal Taq from promega and I have seen this occur before, and we rule it down to polymerase error, if you see it in all your clones then there is a PCR bias occurring in concert.

it is something that can not be avoided as I have tried to use a proofreading enzyme which has very few errors and these invariably fail.

Nick

Are you cloning your results prior to sequencing, or are you sequencing the PCR product directly? If you are sequencing the PCR product, then this can be explained by lack of specificity for the modified strand in your amplification primers. If you are amplifying both converted strands, rather than selecting for one or the other, then you will see conversions of G's to A's at sites which are unmethylated (in the opposite strand). Since you see both G->A and C->T conversions, you must be looking at conversions on both strands, which probably means your amplifcation primers are not selecting for modified strands. If you are sequencing from PCR product, then you are likely sequencing a mixture, which may show up as having modifications at both sites, although I would guess that you would see ambiguous calls at those locations. Have you looked at the electropherogram?

MethylNick and Phage434. Thank you for your thoughts regarding my bisulfite sequencing problem.

In answer to your questions...

I am performing the PCR on bisulfite tx'd DNA with primers designed by MethPrimer. That PCR product is size checked on a gel, then cleaned, diluted, ligated to pGEM, transformed into Top10 cells, and cloned. Insert positive clones are selected for SP6/T7 sequencing. I do not see the conversion issues in the primer sequence.

I am using "regular" Taq polymerase for my PCRs.

I have reviewed the electropherograms quite closely, forward and reverse directions, and there is no ambiguity hence I don't think I have two sequences.

I would be interested in your further evaluation based on this information. If this is a Taq "error" ... do you suggest I edit my sequence to read CG so that MethTools does not provide false information regarding methylation at those sites?

If this is lack of specificity of my primers for the modified strand..yikes now what? Is this something I can edit in my sequence to account for the problem and then put that edited sequence into MethTools? I have trouble thinking backwards, in reverse complement etc so please advise with that in mind!

I've attached a PDF of an alignment of my clones against the "mother sequence" for easier viewing of my SNPs, demethylation etc.

Thanks for your help - last minute crunch to get this data to work! Regards - harley

OK, I'm mystified. Since you are sequencing individual clones, you should be seeing either one or the other of the modified strands, even if the primers are not selecting for one of the modified strands over the other. I cannot see how you get from a CG pair in the genomic sequence to a TA pair in the PCR product, except by error. And it is not happening once, but many times.

I have only one thought. It might be possible for two different PCR products to anneal together (one from each modified strand). They would anneal well, since most bases match. If then cloned into a vector, and the vector transformed, then that vector would undergo base mismatch repair in the cell, leading to the results that you see. I put this in the three-miracles-at-once category of explanation, however, and my Occam razor is buzzing at full speed.

another thing I see is some pcr/clonal bias harley,

how much gDNA do you start for bisfulfite conversion, the reason you are seeing many instances of a CG-TA conversion is that it probably arose from one Taq mistake that is amplified itself.

are you performing multiple PCRs on the one sample and then pooling prior to cloning? how is your cloning efficiency?

Nick