Dear Matt,

That is really a great question.

But I would say that you would not get a new protein:DNA interaction because after the formaldehyde fixation everything should be stopped. If you would get a new interaction, this would be too weak for the next steps.

all the best,

Sebastian

There is indirect evidence that such binding events do not occur: in native conditions most transcription factors get stripped off by extraction and washing. They cannot be ChIPped any longer. Accordingly it is very unlikely that a transcription factor that has not been crosslinked will bind to chromatin after extraction and will be pulled down efficiently.

Simple control: try to see if you can ChIP your factor under native conditions. If you cannot get enrichment, you should be on the safe side.

I also think that this plays no role. To form new DNA-protein interactions you need an intact and functional protein (after crosslinking, which alters proteins and after the treatment with the lysis buffer which contains detergents which denaturate protein complexes) and the right DNA to bind.

Try to do a ChIP without cross-linking, this should tell you if you need to worry. But I doubt it.

Interesting question - I hadn't considered that before. As many people stated earlier, one would assume that only proteins bound during the fixation step would remain bound to the DNA during lysis. Are you fixing with formaldehyde, UV cross-linker, or both? Also, depending on the type of lysis buffer, it is likely that the salts and detergents would also prevent any additional binding of protein to DNA as their native structures will likely be affected by the lysis buffer. If you do perform a control for this, I would be curious to know the results, but I am optimistic that only the proteins bound to the DNA during fixation would still be intact after lysis and washing.

Sounds like we are all on the same page - it shouldn't bind IF the lysis conditions are stringent enough. However, in light of the fact that I'm studying a strong DNA binding proteins and the lysis isn't fully denaturing (otherwise the antibody wouldn't work), I already setup the control experiment. Basically I'll be doing the ChIP protocol without the fixation step, and compare the results to my normal protocol. I'll let you know how it turns out. Thanks

I have an experience in chromatin fixation and immunoprecipitation in yeasts ( about 9 years). The cells are lysed by vortexing with glass beads in a buffer that contains 1% of SDS. Then the buffer is 10 fold diluted and antibodies against the DNA binding protein is added. Usually rabbit polyclonal antibodies is used they work good in these stringent conditions. When I set up this method in laboratory I used a negative control when there is no fixation of cells. I have reproducibly seen that the signal in this negative control is the same as from other negative controls that obtained with samples where no antibody are used, or the yeast strain has no protein. So, from these experiments I can conclude that in stringent conditions of cell lysis all interactions in lysates are inhibited.

From literature I know that the early methods used precipitation of chromatin from unfixed cell lysates prepared in mild conditions. These methods had difficulties with reproducibility, may be because of unwanted interactions in lysates. X-ChIP is more reliable successor of this technology.

For qPCR I suggest to transfer the protein-A-agarose beads with bound complexes into new tube for elution the bound protein-DNA complexes. This step reduce the DNA background that are non specifically bound to tube walls after incubation with cell lysates.

Dmitry, I appreciate you sharing your experience. I'm glad to hear you don't have any post-lysis binding. The mammalian ChIP protocols I've seen and used don't have that high of SDS % in the lysis buffer though - I'm only using 0.1%. If I seen any evidence that binding can occur in my buffer I will increase the % SDS in my next effort.

@ Tobias Straub - before there was ChIP, there were methods that allowed you to assay the binding of a txn factor to a specific DNA sequence without fixation, using milder lysis Buffer condition - so you are wrong. However, back to the original question - it depends how well you fix - if fixation is more or less complete ( that is efficient) - no new interactions should occur during lysis. Furthermore, just like in a regular IP protocol, depending on your salt concentration, washing should take care of most non-specific interactions. As to the lysis buffer, even mammalian ChIp protocols include a 1% SDS lysis buffer, which is the diluted to 0.1% SDS (1:10) during the actual IP. If you use the original ChIP protocol, you also wash with buffers containing DOC, so I don't see what your concern is. The only issue I can see is if you under-fix. I have had to fix cells at lower concentrations of formaldehyde because of a masked epitope, so I have had situations where I had to under-fix - still my interactions were specific. Hope this helps.

Matt, I can refer you to Abcam. On the website of this company there is a lot of information about quality of antibodies. For each antibody the application where it can be used is indicated. You need X-ChIP-grade antibodies. And also they provide experimental data and references about each antibody, so you can choose the most cited antibodies and go to the according references to find out the specific protocol the authors used. We used a lot of Abcam antibodies against histones and histone modifications in X-ChIP experiments. They work very good.

Hi Matt,

I too think that this is unlikely to happen and for the reasons that have been cited above. Normally factors that bind the DNA have their binding motif already engaged and locked by fixation.

I don’t think that your ChIP without fixation is worth to be performed for the simple reason that it is unreliable. This approach is only possible with histones. Transcription factors binding to the DNA obey equilibrium laws. In solution, they will be exchanging continuously their binding sites. This translates that they may occupy sites that are free of them in vivo. Because the number of molecules of any given transcription factor is lower than the number of their corresponding cognate binding sites and the process is random, each time you will be getting different sequence distribution. I thought that the whole Cross-linking process was meant to overcome that

The thing about crosslinking with formaldehyde is that it crosslinks proteins together - crosslinking must be performed for a few hours for proteins to be efficiently crosslinked to DNA; so, when crosslinking is performed for ChIP, generally what happens is that your protein of interest will be crosslinked to a nearby nucleosome. Formaldehyde fixation also causes epitope masking for this reason - since the nucleus is essentially a bag of proteins in close proximity, a lot of nonspecific crosslinks will occur that may obscure the epitope or epitopes recognized by your antibody.

I disagree with the above assessment that native ChIP is unreliable and that TFs will continuously exchange binding sites after lysis; it's not known whether this occurs following cell lysis, particularly in the cold as most lysis is performed. We recently mapped the yeast epigenome after native chromatin extraction and showed that we could map the binding not only of nucleosomes but also particles that are likely TFs (http://www.ncbi.nlm.nih.gov/pubmed/22025700).

Gabriel,

I am sorry, you referred to a work which has nothing to do with this matter. One thing is to perform mapping by in vivo nuclease protection assay and another thing is to have in solution native chromatin fragments for IP. Just think of EMSA competition experiment to understand what I meant.

I can assure you that the SDS in the lysis buffer after crosslinking open most epitopes. I once raised this question regarding Restriction enzyme digestion after crosslinking chromatin and apart from being referred to a specific work, I done it myself. The action of formaldehyde is quite quick, it is very specific because it only crosslinks molecules that are at a valence or two distance from each others. That means there has to be an interface of interaction between the two proteins or proteins-nucleic acids.

I suggest you ask S. Henikoff , for whom I have a great respect, what he thinks

The work referenced does entail the generation of soluble, unfixed chromatin via MNase digestion - this is what was sequenced to generate the profiles shown in the paper. The fact that we are able to see regions of MNase protection at very small sizes (down to ~25 bp) suggests that, during the preparation of native chromatin, binding of non-nucleosomal factors could be quite stable. We have also successfully used MNase digestion of native chromatin for Pol II ChIP-seq (http://www.ncbi.nlm.nih.gov/pubmed/22085965).

My own work involves using MNase-digested native chromatin as input for ChIP-seq of chromatin remodeling enzymes in yeast (accepted at PLOS Genetics, I can send you a manuscript if you like). I found that I am able to get substantial enrichment of the Isw2 remodeler at sites previously determined both by X-ChIP-chip and ChIP-exo, as well as correlations between chromatin occupancy and nucleosome turnover within gene bodies for Isw1 and Chd1, which act to suppress excessive histone turnover in gene coding regions (http://www.ncbi.nlm.nih.gov/pubmed/22922743).

On the topic, here's a paper that shows that formaldehyde crosslinking may not always preserve protein-DNA interactions:

http://www.sciencedirect.com/science/article/pii/S1046202398907241#

Hi Gabriel,

I think there is a bit of misunderstanding. So, this is not about argumentation only to make things clear. I may have exaggerated a bit “continuously exchanging” but it was for the purpose of illustration. I am saying that you cannot apply native ChIP to transcription factors that bind the DNA through specific sequences and I stand by it. Usually they have high affinity and they can be extracted at relatively low salt concentration (300-400 mM NaCl or even less). They can exchange in solution if there is an unoccupied cognate site. It is pure thermodynamics. In vivo their binding is tightly regulated. Pol Ii could be regarded as an exception, as well as some other factors that bind DNA non-specifically and very strongly. We should not confuse high affinity binding with strong binding. They are salt-resistant to extractions...

The MNase digestion only shows site occupancy in vivo that correlate well with ChIP seq performed with some transcription factors.

I didn’t say that formaldehyde preserve every kind of protein interactions. It is a blind try even though we know all aspects of it chemistry. Contrarily to what one may expect (at least it was for me), I think for nucleosomes it only pick up very few despite all the interface of interactions of the octamer with the DNA. So, you can even open/displace the histone octamer without releasing it from the DNA and digest with restriction enzyme a site that it is inaccessible otherwise.

I will be willing to read your paper and congratulations by the way.

Abdelhalim,

I didn't mean to come across as argumentative; I tend to get worked up over this issue! The only point I wanted to make is that we can see what appears to be legitimate binding events of various proteins without crosslinking.

Anyway: I see your point; thermodynamically speaking exchange will occur at some rate, even at low temperature and ex vivo. As to the mapping of site-specific TFs via native ChIP, we are working on this issue, so we will see if it can be done!

Regarding the crosslinking of nucleosomes themselves, this paper may be of interest: http://www.cell.com/abstract/0092-8674(78)90278-7. It seems that crosslinking of histones within nucleosomes is quite inefficient and must be done for days in order to get octamers.

Cheers Gabriel! I will be looking for it. I will be looking for it. I am open-minded. I like new methods. I myself was engaged in developing a new epigenetic method (ChILL) Chromatin-immunolinked ligation within a European consortium but for a commercial use. The technique was patented by Rolf Ohlsson from Sweden. It combines 3C/4C with ChIP but there is no IP. Still, though, uses formaldehyde crosslinking followed by RE digestion, but everything is done within the cell. We had hard time to demonstrate the validity of the method, optimization, etc…due to technical inconsistencies in the original version (4 labs including one from the company). The project (3 years and almost 2 millions euros) expired before reaching all its targets. I am a bit disconnected about it final outcome, but I know Rolf’s lab is still engaged in it and there are no available kits, which was the real objective.

Thanks for the ref