qPCR for ChIP quantification - cannot get nice standard curves.

[JWC]

Dear all,

I realise that standard curve questions are very common, but the majority of problems I have seen are to do with low efficiencies. I have the opposite problem - efficiencies of 200 or even 300% on standard curves.

Brief outline of my experiment:

in vivo ChIP using mouse embryos

1) Crosslink dissected embryos
2) Isolate and sonicate chromatin
3) Incubate with antibody + protein G magnetic beads
4) Wash beads and elute immunoprecipitated chromatin
5) Reverse crosslinks and purify chromatin
6) Quantify IP at sites of interest with SybrGreen qPCR

A) an "input" fraction of chromatin is taken from step 2, skips steps 3 and 4 and joins the other samples at step 5. This fraction is the reference for the qPCR and represents total genomic DNA.

I decided to run a standard curve for each set of primers using a serial dilution of the purified input fraction in order to estimate primer efficiencies for use in the Pffafl method. The standard curve started with 4ng of input DNA and consisted of 1:10 serial dilutions across a total of 7 orders of magnitude.

Unfortunately, the first four primers I tried on the standard curve produced slopes of around -1.5 to -1.8 which produces efficiences of around 250% to 350%. This is clearly very bad. This is the best of the standard curves so far (you'll have to remove the ***, I can't post full links as a new member): http:/***/postimage.org/image/csyl2mvs3/

I am using the standard cycling parameters on the ABI 7900HT (40 cycles of 15'' 95C, 1' 60C) and 250nM of each primer in a 10ul reaction mix with 2ul of DNA solution per reaction.

New primers are one suggestion I guess will come up, but the fact that 4 out of 4 primer pairs I have tried so far show this behaviour makes me suspicious. Does anyone have any advice for the painful process of extracting some meaningful data from these samples? Any help would be greatly appreciated.