Editor’s Note: At the author’s request on August 20th, 2015, revisions to this article were made following careful consideration. All changes are denoted in red.
Since its introduction in the 1980s, chromatin immunoprecipitation (ChIP) has become one of the most important and powerful techniques in the field of genetics, allowing researchers to characterize a given protein’s binding sites across the genome. With the advent of next-generation sequencing and whole genome analysis, a single ChIP can deliver gigabytes of information, including not just genomic binding sites but also sequence preference, characterization of chromatin structure, mutation data, and epigenetic regulation.
But ChIP is also notoriously technically challenging, highly sensitive to errors and artifacts, and often requires a great deal of upfront optimization before reliable results are obtained. Moreover, it is tremendously expensive, time-consuming, and if done improperly, can cause you to base an entire graduate project on shoddy data.
This technical review is meant to share my experiences of optimizing ChIP in human cell culture systems and provide readers with an accessible walkthrough of navigating their own ChIP experiments, including common pitfalls and troubleshooting procedures. It is aimed at researchers with at least an intermediate level of experience with genetic techniques. Generally, if you are just beginning to work at the bench and your boss wants you to figure out ChIP on your own, this means that either s/he considers you extremely talented, or that s/he has little appreciation for how difficult this technique can be. I would urge you to get familiar with more foundational genetics techniques before setting out on a ChIP experiment.
My ChIP protocol utilizes formaldehyde-based crosslinking and the Covaris S2 sonicator. There are other forms of ChIP that do not rely on crosslinking, and there are certainly other ways to shear DNA than using the S2 (probe sonicator and Bioruptor among them). The readout we use for our data is real time PCR or ChIP-seq, although, again, these are not the only options out there. For the sake of simplicity, and to keep within the area of my expertise, I will focus on our in-house methods and protocols with which we have had success.
Lastly, ChIP is a difficult procedure, with many different paths to success. Readers are encouraged to treat this review not as a gospel but as a signpost, a primer to get them started rather than a rigid protocol that must be followed to the letter. If anyone has any insights or critiques, by all means feel free to comment or e-mail me ([email protected]).
ChIP Overview
The premise of ChIP sounds quite simple. There are countless proteins that interact directly with the genome, and ChIP provides a means of figuring out which proteins interact with which regions of DNA. It does so by crosslinking protein complexes to the DNA with a fixative like formaldehyde- this creates covalent bonds that tether the proteins to associated DNA. The cells are then lysed, and the DNA is fragmented into small chunks of about 200-500 bp. The sonicated DNA is incubated with an antibody that binds to the protein of interest. The lysate is next incubated with beads coated with molecules (usually Protein A or Protein G) that bind the antibodies, such that the protein of interest- and its associated DNA- is now bound to the beads. The beads then go through a series of washes to get rid of any DNA fragments that are not tightly bound. Finally, the remaining DNA-protein complexes are eluted from the beads and decrosslinked at high temperatures, such that the covalent bonds that connected the proteins and DNA are broken. The researcher now has, in theory, enriched only the fraction of DNA that is bound by the protein of interest, and massively reduced the representation of all other regions of the genome. Seems reasonably straightforward, right?
There are indeed some ChIP experiments that are not especially difficult for an experienced biologist. Some targets, like well-studied transcription factors and histone marks, have a bevy of high-quality antibodies, protocols, and kits that are available online. If you have good lab hands, you should be able to get these working with relative ease.
But for less-studied systems and targets, ChIP can be very challenging. For instance, one reason that histone marks are so easy to ChIP for is that they are all over the genome; there are many, many regions that are associated with, say, H3K9 acetylation. But there are some target proteins that only interact with a tiny fraction of the genome, and so at the end of a successful ChIP, you are at best pulling down a few picograms of DNA, some of which is inevitably background. Another common issue for studying new or poorly characterized DNA binding proteins is the lack of quality antibodies: if you are trying to perform ChIP for a target protein using a lousy antibody, you will virtually never get really robust results.
This guide is meant to help readers navigate a standard ChIP protocol and avoid some of the common issues in optimizing the technique. Again, it is advice from someone who has learned the hard way what works and what doesn’t. The usual caveats- nothing here is written in stone, stay critical, be ready to figure things out yourself, etc- apply.
So, let’s get started.
Part I: Preparation
Early Considerations
The first and most important consideration at the outset of any ChIP experiment should be- should I do this ChIP experiment? This sounds simple, but it really is important to consider this carefully. It is, again, oftentimes a long, challenging, and expensive procedure. If you have only a limited amount of time at your current lab, it may not be worthwhile to go through all of the optimization, only to leave once you finally have it working. Remember that if you are planning to do ChIP-seq, it may take a long while to get the data back, and then you will likely have to validate your ChIP-seq results on a new set of samples by real-time PCR. Be sure that you have a clear goal in doing this experiment, so that when the data comes back, you won’t be completely awash in several megabases of sequencing data. If your readout is PCR, find positive and negative control regions- regions in which your target does and doesn’t bind, respectively, and work out a PCR program that amplifies with these primers efficiently.
Read the literature. Chances are that someone somewhere has already done some of the legwork for your project- why reinvent the wheel? If all your ChIP does is reproduce what someone has already demonstrated, that is not very interesting or productive. Instead, try to find a unique angle that you or your lab is especially tailored to pursue. The primary literature may also give you a heads-up on potential problems for your particular target- for example, whether it is easily degraded, or when in the cell cycle it is expressed. This advice is especially critical for young scientists, who in youthful enthusiasm charge into the benchwork without doing any homework.
Last but not least, it is very possible that someone has already done ChIP for your target, and there are multiple online resources that can provide you with that data without requiring you to lift a pipette. Some critical resources here are the USCS Genome Browser, the ENCODE Project, and the NCBI. If you aren’t familiar with these resources, get familiar with them, now. Find someone who can show you the ropes. If these resources immediately provide you with the answers you hope to get from ChIP, then you’ve just saved yourself many hours of blood, sweat and tears. If your questions still aren’t answered, then you’ve armed yourself with at least a working knowledge of your protein of interest.
At the risk of being tiresome, I would like to underline this one last time: Do your homework. Read the literature. Study the ChIP-Seq tracks at the USCS genome browser. Do not assume your PI knows everything. Before even dusting off your bench, make sure you know what you are getting into.
That being said, let’s say that your question isn’t answered online, and you want to proceed with a ChIP experiment. Where to start?
Picking the Right Antibody
If you want to ChIP for something that is already heavily studied, you are likely in luck- there are many well-vetted antibodies for various histone marks and transcription factors. Many companies even offer kits for certain targets that are quite reliable. Antibodies to histone marks in particular are very robust, and in fact I would recommend including one IP reaction in each experiment with a histone mark- such as H3K27 monomethylation- as a positive control. If, at the end of your experiment, you see poor enrichment for histone marks, that means you screwed up bad.
Picking the antibody is often the most important step in ChIP, especially if you are studying an obscure target that lacks high-quality antibodies. Companies will indicate which of their antibodies are suitable for ChIP and which aren’t, although these recommendations are not always trustworthy. The only way to know if an antibody works well for ChIP is to try it. I recommend buying two or three antibodies and seeing which works best, but since they are very expensive, that may not be possible for some labs. Consult the literature and other people in the field and make an educated choice. Also, there are some antibody companies that are more reliable than others. Cell Signaling and Abcam are among the best, so buy from them if you can.
Be mindful of what region of the protein an antibody binds, and whether it is a monoclonal or polyclonal antibody. If you are studying a mutated or truncated protein, knowing the epitope sequence is critical. Often polyclonal antibodies are better suited for ChIP than monoclonals. Whereas polyclonal antibodies can bind multiple regions of your protein, monoclonals only bind one, and it could be that crosslinking hides or destroys that single epitope.
Picking the Right Cell Line and the Right Conditions
Also put some time into thinking which cell line would be a good model for your question. Obviously, the cell line should express your target protein. You can check online resources to confirm this- the CBioPortal has mRNA expression data for many cell lines, and you can check for protein expression at the Protein Atlas. It’s probably a good idea to double-check your cell line upfront, for mRNA and protein by real time PCR and Western blot, respectively, especially if you’ve bummed a non-STR validated cell line off another lab.
It may be the case that your ChIP antibody doesn’t work well for Western blot, and vice versa. This is because you are interrogating the same protein under two very different conditions. On one hand, a good Western antibody may not bind well to the crosslinked protein, as crosslinking often destroys or obscures epitopes. Conversely, an antibody may bind well to the crosslinked protein in a mostly-native state, but may not bind at all when it is under denatured conditions, as it often is in Western blots. It’s usually not necessary to have an antibody that works well in both ChIP and Western assays; it’s OK to have one antibody for Westerns and one for ChIP.
In theory, when you check for your protein by Western blot, you want to see a single crisp band at the expected molecular weight. If you see multiple bands, those can be versions of your protein that have post-translational modifications (like phosphorylation), degradation products of your protein, or proteins that your antibody binds to non-specifically. While having multiple bands by Western is not ideal for a ChIP antibody, it sometimes can’t be avoided. Besides, the chances that those non-specific bands belong to a nuclear protein that interacts with DNA are probably quite low, so it shouldn’t make much of a difference for your ChIP.
Depending on your protein of interest, it may be helpful at this point to recognize the conditions under which your protein is best expressed. For instance, many proteins are expressed only during a specific stage of the cell cycle. SUMOylation is a post-translational modification that often requires the cells to be incubated at 42 degrees to stabilize the SUMOylation mark. It may be that only the phosphorylated version of your protein successfully binds DNA. Again, these are all general considerations to keep in mind.
Lastly, remember that cell line models are just that- models. If you want to ask a biological question, make sure that the cell line you pick is a good model for answering that question. For instance, in prostate cancer, two of the most widely used cell line models are DU-145 and PC-3 because they were established long ago and are well-studied. But recent analyses suggest that they are in fact not very representative of prostate cancer in vivo, and that there are other prostate lines that more closely resemble the properties of primary prostate cancer that we study.
Preparing Your Reagents
So now you have your antibody and your cell line, and you’re ready for a trial ChIP. Here are the reagents you should prepare beforehand:
LB1 Buffer | LB2 Buffer | Shearing Buffer | ||
50 mM HEPES, pH 8 | 10 mM Tris-HCl, pH 8 | 50 mM Tris-HCl, pH 8 | ||
140 mM NaCl | 200 mM NaCl | 5 mM EDTA | ||
1 mM EDTA | 1 mM EDTA | 0.2% SDS | ||
10% glycerol | 0.5 mM EGTA | |||
0.5% Igepal | ||||
0.25% Triton X-100 | ||||
Dilution Buffer | Elution Buffer | |||
2 mM EDTA | 1% SDS | |||
150 mM NaCl | 0.75% sodium bicarbonate | |||
20 mM Tris-Hcl, pH 8.0 | ||||
TSE I | TSE II | TSE III | ||
0.1% SDS | 0.1% SDS | 0.25 M LiCl | ||
1% Triton | 1% Triton | 1% Igepal C-630 | ||
2 mM EDTA | 2 mM EDTA | 1 mM EDTA | ||
20 mM Tris-Hcl, pH 8.0 | 20 mM Tris-Hcl, pH 8.0 | 10 mM Tris-Hcl, pH 8.0 | ||
150 mM NaCl | 500 mM NaCl | 1% deoxycholate |
These buffers vary a little bit between labs and applications, but generally they are all pretty much the same. This is likely stating the obvious, but you want to use high-quality, molecular-grade reagents for ChIP. It’s a long and expensive procedure, and so it’s not worth trying to be cheap and using ancient buffers and powders that have been sitting around the lab for a decade. In particular, sodium bicarbonate in solution goes bad fast when exposed to the open air, so you want to prepare the elution buffer fresh each time immediately before elution.
You also want to prepare the 1% formaldehyde solution fresh each time, immediately before you begin harvesting your cells for ChIP. We use methanol-free ampules of 16% formaldehyde and dilute it down to 1% with PBS, such that we have 20 mL per T175 of cells. Using a fresh ampule each time ensures that nothing funny happens to your formaldehyde stock from one run to the next- formaldehyde solutions also have a tendency to change in composition and crosslinking efficiency over longer periods of time.
This concludes my primer on all the upfront preparation for ChIP. In the next installment, we’ll get into the guts of the actual procedure, highlight the steps that may require some optimization, and alert the reader to some common pitfalls that beset chromatin immunoprecipitation.
Continue to the next article in this ChIP series, Part II: A Starter Chromatin Immunoprecipitation (ChIP) Protocol.