The actual bisulfite treatment - protocol and troubleshooting (Nov/18/2004 )
With this new (agarose bead) protocol, I get great PCR and great sequence, so.... I think it's ok, but thanks for your concern Anyone have any input? I'm not seeing much methylation, but that wouldn't be a result of things not getting desulfonated properly, since in my understanding the 5mC doesn't react with bisulfite in the first place, right?
Yes you are absolutely correct, 5mC is unreactive to the bisulfite reaction.
Thanks for that protocol labtechie! It worked great - it even successfully converted some really dodgy samples that the chemicon kit had failed to work on. Just to add, I've cut out the number of washes at the end and it still seems to work fine.
Jon
Hi Jon! That makes me so happy. Questions: How many washes did you cut it down to? You still have to wash initially in TE and then in water before PCR, right? Are you seeing any CpG methylation? I see none, I assume because that's a negative result but I'm worried that perhaps my protocol works TOO well, I need a positive control 
Do most people reading this buy fully methylated DNA or is there a region I can amplify that is known to always be methylated? Also, the only artifacts I get are sometimes a non-converted nonCpG C a few bp away from a CpG. Do you get that? Interesting artifact, a result of steric interactions or secondary DNA structure maybe?
Just curious about all these things... so glad your protocol works!
Ellen
P.S. All: I do NOT pre-digest my DNA with a restriction enzyme and this protocol still works FINE.
Do most people reading this buy fully methylated DNA or is there a region I can amplify that is known to always be methylated? Also, the only artifacts I get are sometimes a non-converted nonCpG C a few bp away from a CpG. Do you get that? Interesting artifact, a result of steric interactions or secondary DNA structure maybe?Hi Ellen,
as a positive negative control, instead of buying methylated DNA it might be worth your while to methylate some DNA yourself. You can use SssI methylase (NEB #M0226) and this enzyme methylates every CG residue available. As a control I would methylate a plasmid such as pGEM and then "spike" my sample with the methylated plasmid prior to bisulfite treatment. You can assume that all CG's on the plasmid are methylated after SssI methylase and therefore should all come up as CG after bisulfite sequencing.
As for non-CG non conversion close to CG residues, this is due to incomplete denaturation of the strands and thus inefficent bisulfite conversion.
Hope this helps
Nick
Hi Ellen
I just did 2 TE washes and 1 water wash and it seemed to be fine.
I have some universally methylated control DNA and this showed methylation using the protocol, so it is fine (although like you I found no methylation in my actual test samples !)
I haven't done much sequencing on the samples yet, so can't comment on the non-converted C's, but will let you know as soon as I get round to it.
I also didn't enzyme digest and don't appear to have problems.
Cheers
Jon
Thanks everyone for all the help.
Nick: when you say "spike" do you mean that I should have the plasmid in the same bead as my sample so as to ensure that it is getting the EXACT same treatment? Or can I have it as a different bead and just do the same protocol at the same time?
Jon (and anyone else): When you get no methylation in your test samples, I assume these are CpG islands yes? I figure mine is a perfect negative result, but my boss is not happy. He thinks that there should be some methylation somewhere even though I show him the literature that most CpG islands are NOT methylated at all. Anyone have any really good literature on this they can point me to?
Ellen
He thinks that there should be some methylation somewhere even though I show him the literature that most CpG islands are NOT methylated at all. Anyone have any really good literature on this they can point me to?
Ellen,
yes, ideally you should have the plasmid in the same bead as your sample to ensure you are getting the exact treatment for both DNA's.
As for your island of interest, is it part of a promoter of a gene? if so, has the gene been switched off? There are cases in the literature where the CpG island promoter is unmethylated and yet switched off, there are also cases where the island is methylated and the gene is expressed. I have found such a case!!
in terms of literature, there are a number of good review papers published by Bird's Group and Feinberg's Group on the topic. Can't tell you off the top of my head now though sorry!
You are right to tell your supervisor that most CpG islands are not methylated. You may have to look at the chromatin environment surrounding your gene as chromatin proteins can modulate gene expression, that maybe a whole new kettle of fish!
Good luck with it all!
Nick
Ellen
I think your results are probably fine - if the CpG's you were looking at were methylated then they wouldn't be converted.
It is possible there could be very low-level methylation - perhaps test this using a quantitative method like MS-SNuPe - if the level was very low (e.g. 5%) you may not pick a methylated clone from the number you are sequencing.
On a technical note about the bead protocol - how long do you think the treated DNA is stable for after the protocol (if stored in a fridge in TE)??
cHEERS
jON
Jon-- I've used treated DNA beads up to about three weeks after and they were fine-- some protocols I've read say five days. I don't have the technical expertise to give you any better advice than that though.
Thanks, methylation buddies, for backing me up on the "CpG islands are not methylated" debate with my boss. I'll have to do a positive control. But if I do my own methylation of a plasmid, how do I know it worked? I mean, if I methylate the plasmid, spike my samples, then treat, then sequence the plasmid and the CpGs are not methylated, how do I know if the methylation step failed or if the treatment is too strong?
So complicated.... thanks for all the help,
Ellen