ChIP Bioanalyzer pattern mystery... \o/

[Postdoc_Rach]

Hello ChIP-Seq community, I have pored over all the ChIP Bioanalyzer posts without finding an answer, and now I ask any advice from those of you with experience. Attached is a PDF of the Bioanalyzer results of my latest ChIP-Seq libraries, including Input, 9C/9D (tech replicates) and 27C/27D (tech replicates).

For these, one chromatin sample was crosslinked, sheared in a Bioruptor, and aliquoted into the appropriate antibody tubes.

After de-crosslinking/Prot. K, there were two methods of purification (to test the respective yields). Input, 9D and 27D were purified with 1.8 volumes of Ampure XP. Meanwhile, 9C and 27C were purified with the spin column that came with our ChIP kit. Libraries were produced in an identical manner for all DNAs using a kit, with similar amounts of starting DNA for each.

It seems to me that the "artifact" seen in 9C and 27C must arise from the column purification, but how this would translate to differences in the final library considering the adapter-ligated fragments are bead-purified before PCR? If it were a cross-linking effect, I would expect to see it in both replicates, so I assume it is a bubble product. (And therefore not terribly important, but this is driving me nuts.)

Any ideas greatly appreciated, thank you.

[nucacidhunter]

Not a straight answer but questions:
1- Extension time during PCR
2- Number of PCR cycles
3- Are these purified PCR’s profiles
4- Has 9D and 27D been amplified? They look exactly as the input with higher concentration which is not expected.

[Postdoc_Rach]

Thanks very much for the reply, Nucacidhunter.

Yes, you are looking at around 0.3 to 0.5 ng of purified (amplified) libraries here. The input sample was very concentrated and so was diluted 100-fold (perhaps too much) before loading on the Bioanalyzer chip, which is why the signal is so low.

For these libraries we use the Nugen Ultralow kit, and I used the amplification program in the protocol, which was 15 cycles and with an extension step of 30 seconds. Libraries are purified again with AmpureXP beads, and they were placed on a magnet while pipetting out for the BA chip.

We've sequenced libraries that look like the C samples (with the large peak at 3K) and they seem to have worked well enough in our limited experience. But, I have no good explanation for how the Ampure bead ChIP purification would change the library appearance so much.

If any obvious sequencing differences arise, I will post them here!

[nucacidhunter]

Your PCRs cut off with bead is similar at around 200bp in all libraries as expected. I am not sure that bubble is the cause of 9C and 27C large amplicons because there should be another smaller peak. Could it be that smaller fragments have been lost during column purification? I assume the input shown in the PDF is from bead purified input not the column ones.

[Postdoc_Rach]

Yes, that was also my (only) idea, that maybe the column somehow is losing the small fragments, leaving only larger ones to amplify.

Then the question becomes, what kind of defective spin column (included with a ChIP kit, no less) can't retain 200-1000 bp fragments, and where are the larger fragments in the AmpureXP samples? If I'm not mistaken, a 1.8x volume should retain nearly everything.

You are correct that the input was bead purified, although other times I have phenol/chloroform extracted the input and obtained the same size distribution. One time the input was column-purified, and the yield was appallingly poor (but normal-sized); I credited it to operator error, but maybe the columns are suspect after all.

In any case, this may be one of those things for which we just have to shrug and go forward. I sure will let you know how the sequencing goes though.

Thanks for your time and thoughts, Nucacidhunter! I am the only molecular biology researcher in our group, so it's a rare treat to bounce around ideas about the nuts and bolts of bench science.

[Postdoc_Rach]

Results are in

Hello all, the sequencing results are in. The difference between the C and D samples is that the C samples contain a sizeable bacterial contamination while the D samples do not.

That said, although they do correlate, I do not claim for sure that the bacterial DNA directly resulted in the Caliper graph differences.

As far as IP purification protocol differences go, my guess is that the AmpureXP ethanol washes in the D samples killed the bacteria off before going into library prep.

We will definitely use new IP reagents for the next round. Hope this helps someone down the line.