G to A transition in bisulfite-treated DNA - (Sep/07/2010 )

Hello,
I would like to ask to DNA methylation community if someone encountered and tried to explain a “G to A” transition on bisulfite treated DNA? I noticed this mutation on treated DNA only and, in the reverse strand all “C” resulted unmethylated. This mutation is related to gene expression. Other details: i) I used TaKaRa ex Taq for amplification, ii) I cloned the amplification products and ii) The “G to A” transition is always accomplished by a shift of dissociation peaks in real-time PCR. Possible explanation?
a) Methylation of cytosine on the reverse strand ? (In this case I would expect to find some methylated cytosines in the reverse strand. This was not the case)
Spontaneous deamination of complementary methylated cytosine? (In this case I would expect to see this mutation in the untreated DNA. This was not the case.

I am really getting crazy to find explanations and tempted to trash all these data.

Please help
Fulvio

I´m not sure what you mean by reverse strand. Because if you bisulfite treat DNA you will no longer have complementary strands. When performing a pcr using treated DNA you will only amplify one of the two DNA strands, but of course end up with a double stranded PCR product. Dependent on that you will get C to T transitions in one strand of your PCR product and G to A transitions in the other strand for unmethylated Cs when sequencing.
The melting curve peaks in realtime pcr depend on the degree of methylation in your samples. Amplicons from completely methylated samples usually have a higher Tm than umethylated samples cause by the different GC-contents of the PCR products.

Thanks for reply
I should explain a G to A transition on one strand of my bisulfite treated DNA. I though this would be the consequence of variation of methylation at the related C of the complementary strand BEFORE bisulfite treatment. But in this case the methylation status of the cytosine does not affect the G addition by DNA polymerase in the related position on the complementary strand. Another hypothesis would be that as cytidine becames methylated on the complemetary strand this is rapidly deaminated in vivo generating a T instead of the methylated C. But in this case I would have observed the G to A transition in the untreated DNA also. This was not the case.
what else?