Direct bisulfite sequencing of PCR-products: degree of CpG methylation - (Mar/05/2006 )
Could you, please, answer my question about the possibility to estimate CpG-methylation degree by means of direct sequencing of PCR-product.
I began to study CpG islands methylation in chronic leukemia at the German Cancer Research Center in Heidelberg not long ago. For this purpose we develope an approach based on oligonucleotide microarrays. Having analysed first templates, I am interested to test the results with bisulfite sequencing. Somewhere earlier in this forum, it was recommended to perform direct sequencing of PCR-product with a PCR-primer tagged with T7-sequence. As a start, I have made this for one PCR-product. So, two questions appeared.
1. Is it possible to estimate (at least semiquantitatively) methylation status of "C" in a given CG-dinucleotide from sequencing chromatogram?
2. "A"-peaks in DNA strand, which is extended while sequencing (corresponding to "T"s in original strand, not those in CG-dinucleotides of interest) are pretty often accompanied with lower peaks for "G"s. My first thought was that bisulfite conversion was not complete. But fortunately it seems not to be the case because "T"s in original sequence (which are not produced by deaminating of "C"s) also show major "A"-peaks as well as minor "G"-peaks (please, see attached picture). Could it be the general property of DNA-sequencing due to non-stringent pairing of T-nucleotide?
Thank you very much in advance!
indeed you can estimate the proportion of methylation at one particular methylation site from just the sequence chromatogram.
I would suggest you run a "calibrator" or standard. That is, the same sequence from which you know is a certain proportion is methylated or not, this can be done artificially by M.SssI methylase of your product and mixing in known proportions with the same product unmethylated and then performing bisulfite sequencing on this. I don't think many people do this as this is not reported in papers using direct sequencing, but it's a good thing to do to be sure for yourself.
As for the G peaks, I am interested to know what your fluorescence signals are. Usually such peaks are found where the overall flourescence intensity of your product is low or very close to background levels and this is why you are seeing the G's. if your product signals are high (above 500 or more) such peaks are not seen. if you are diluting your sequencing mix, maybe use it neat, or similarly, load more of your sample on for sequencing.
many thanks for your very helpful answer! I will definetely try calibrator. In fact this is what we are doing while selecting good-working oligonucleotides on the chip. Could you, please, tell me one detail of such a chromatogram analysis (I am making it for the first time). If I perform direct sequencing of a given PCR-product having different proportion of methylation (say, 25, 50, 75% methylation of each individual CG-dinucleotide), there will obviously be two peaks corresponding to each interrogating C (peaks for G and A in complementary strand). If I am write, methylation proportion should me calculated as the ratio of areas under corresponding peaks, namely Area(G)/(Area(G)+Area(A)). The question is whether fluorescence signals on chromatogram are already normalised or not.
Another question is what software can be used to calculate the areas under the peaks.
As for your note about G peaks, the low amount of PCR-product in a sequencing reaction can be the case, indeed, because a couple of other PCR-products were not able to be sequenced. And as I have been told due to their low amount. I simply did not know, that low fluorescence can produce unspecific Gs.
Thank you very much!
Indeed the peaks you are seeing with the G signal is actually background and is due to the low amounts of PCR template. Yes, the peaks are normalised in the basecaller software to give you the chromatogram readout hence the G background as this is amped up.
As for measuring the peak heights to determine methylation ratios, I know mutation surveyor from GE has the ability to do this quite well although I have not tried it myself as I would need to part with a huge sum of money to obtain a license for this. Indeed you are right, the way to do it is to measure the area under the peaks for T and C and express this as a ratio, there is a shortcut to this 4peaks is a freeware program where you can measure the peak height by placing the mouse over the peak, in fact measuring the peak will also suffice if you don't have the software to integrate the areas under the peaks for a more exact measure.
Indeed calibrate this with known ratios of methylation and see what you get, it should be very consistent with peak height alone.
thank you very much for a very helpful answer! I will try everything you have suggested.