Bisulfite sequencing reproducibility - overlapping PCR amp primers give varying results (Nov/12/2008 )
I have been looking at the DNA methylation status a 1kb region around the TSS of several tissue specific genes. In the beginning I used smaller PCR fragments due to the fragility of bisulfite converted DNA but due to time/money/frustration and long T stretches in converted DNA, some of the PCR amplified products are now up to 600 bases with some overlap of regions covered. We get pretty good sequencing results but when we have overlapping fragments we sometimes see different trends in the methylation status. ie at the tail end of one fragment the "methylation" is higher then the same region that is in the beginning of the other fragment. We started with direct sequencing and we were measuring the ratio of C to T at the CpG site of interest and I thought the ratio differences in the two primers was due to background issues by the end of a long fragment due to incorrect C insertion by taq during sequencing because so little of it gets used. We have recently switched to cloning and we again see the problem but we aren't looking at ratios anymore there is only always a C or a T which kind of puts to rest my previous argument of the problem being during the sequencing step. Does anybody have any answers as to why overlapping PCR fragments give differing methylation status trends? Is there anyway to get rid of the variation? I thought of purposfully creating more overlapping primers but am afraid to and also I can't undo the overlap because the region was T rich thus we had a hard time getting primers to work in that area in the first place.
I also was worried that by using such large PCR fragments I might be selecting for more methylated fragments or fragments that were likely to be less converted thus appearing more methylated but with cloning we have seen 95 - 100% conversion with both samples that are completely methylated and unmethylated so I don't think we are skewing our data but I might have missed something in my logic.
And finally, this is a dumb question but why is the DNA fragile after bisulfite conversion. I know the conversion it self is harsh but after clean up why is it still fragile?
If anybody has any insights I would appreciate it!
I think, this is a very common issue in bisulfite sequencing. For some unknown reasons signals for cytosine pop up especially at the end of longer sequences. Even though there are no cytosines in this particular part of your sequence or the DNA is completely unmethylated. It seems to be a general electronical problem of sequencing machines in combination with bisulfite converted DNA. Due to the unnatural composition of bisulfite converted DNA the standard base callers can not adjust the signal intensities and different retention times of the fragments. We got (nearly) rid of that problem by using a different base-caller from Applied Biosystems (KB Base-Caller).
It is described that PCR products generated from bisulfite converted DNA seems to have a different fragment-length when analyzed on a high-resolution agarose gel or by capillary electrophoresis. Sorry, I have currently no access on my literature database. But I think it was published in the Journal of the American Chemical Society in 1998. They explain that bisulfite converted DNA is not as flexible as genomic DNA due to the lower complexity of its base-composition. It seems to be more 'stiff'. This also explains problems of standard sequencing machines (and base callers) to read longer bisulfite sequences.
Hope that helps...
sounds like you have PCR bias in your results and they are also a function of the primer efficiency.
There is no real way around it, we see it in our experiments also.
As a measure, you could try performing multiple PCR's on the same sample with the same primers and then pooling (so do triplicate PCRs and pool and then do your cloning and sequencing or whatever) this seems to smooth out any possible variation induced by primer efficiency and/or PCR bias.
As for why the DNA is so fragile, post conversion it is essentially a DNA/RNA hybrid (a DNA molecule with uracil incorporated). I am hazarding a guess but I would suspect that certain RNAses could then attack this and it therefore degrades.