Johanna Samuelsson Ph.D., Senior Research Scientist at Active Motif covers some techniques to get killer data, even when your samples don’t make it easy.
ChIP Overview
I’m going to talk about chromatin immunoprecipitation, not necessarily focusing on all the specific details, but more the challenges with the technique, and those tricky samples that many of us are struggling with to get reliable ChIP data, but will start with a short introduction to epigenetics, which refers to functionally relevant and heritable modifications to the genome that do not influence a change to the nucleotide sequence.
These modifications, which includes DNA methylation and histone modifications, are regulators of the chromatin structure. And they ultimately influence gene expression. In addition to all these actual epigenetic modifications, there are also three main classes of chromatin regulators that influence these modifications. Those are the writers that actually deposits the modifications. Then you have the erasers that remove the histone modifications or the DNA methylation. Then you also have the readers, which are the interacting proteins that lack catalytic activity and simply recruit other proteins.Together, they regulate the state of the chromatin and then also the accessibility of transcription factors.
Chromatin Immunoprecipitation, or ChIP, is a powerful tool for studying interactions between specific proteins or modified forms of proteins in a genomic DNA region. And the first ChIP assay was developed by Gilmour and Lis in the mid-80s for monitoring the association of RNA Pol II and genes in vivo.
ChIP was later on adapted to mammalian cells as well in the late-90s by Boyd and Peggy Farnham. And just a few years later, Brian Straw and David Allis published a review in which they proposed that distinct histone modifications, that is, one or more tails, act sequentially or in a combination to form a histone code. That is then read by other proteins to bring about downstream events.
Now a decade later, we know that this is indeed the case. ChIP has generated us with a wealth of information since its invention about the histone modifications and also the proteins that regulate them. And this is relevant not only for insights into normal processes such as development, but also in diseases such as cancer where epigenetic changes such as DNA methylation and histone modifications have been shown to contribute to tumorigenesis.
The chromatin structure at active promoters. They show no or low levels of DNA methylation as well as enrichment of HJK4 tri-methylation. Temporarily repressed genes, on the other hand, they still show no or low levels of DNA methylation, but now an enrichment of H3K27 tri-methylation. Then permanently repressed genes and heterochromatic regions, which contains a lot of repetitive elements, they usually contain high levels of DNA methylation and also H3K9 tri-methylation.
Breakdown of the ChIP Procedure
So quickly, I want to start to give an overview of the different phases of the ChIP procedure. Living cells or frozen tissue are first fixed with formaldehyde, which results in protein protein and protein DNA crosslinks. And at the moment of fixation, these interactions become covalently attached. The cells are then processed. And the chromatin is isolated and sonicated or digested into smaller pieces.
Each fragment will then have the native population of proteins that is still bound to it. Then an antibody against a factor of interest is added, and immunoprecipitation’s performed to isolate the factor along with the DNA it’s associated with. Then the crosslinks are reversed, and DNA is isolated. And at this point, you will have a population of DNA fragments of all the genomic binding sites on the targeted factor.
This DNA can be used in qPCR to verify binding sites on a gene to gene basis. Or the DNA can be used for genomic analysis by re-hybridization or next generation sequencing.
I have to mention here that although ChIP has come a long way, there are still limitations to this technique. And within the protocol, there are multiple steps that have proven challenging for many people. So the chromatin preparation step is usually pretty straightforward when you’re working with cell culture.
For certain primary cells, such as T cells or also tissues like breast or fat, getting sufficient levels of good chromatin can be difficult. And optimization of the standard protocol will be needed to extract quality chromatin for these types of cells.
Then you also have the antibody selection. Most antibodies out there, they’re not sensitive or specific enough for ChIP purposes. Only about 25% of transcription factor antibodies and around 80% of histone modification antibodies actually work in ChIP. So it’s recommended to use a robust and ChIP-valid antibody from a reliable source. And at Active Motif, we currently have a list of over 200 ChIP validated antibodies.
Further downstream, the libraries can be problematic. A poor ChIP reaction, and also poor libraries, they can result in high duplication rates. Another issue for many people is the normalization and analysis of degenerated data sets.
Here at Active Motif, we have performed ChIP on over 200 different targets at this point. We have processed well over 10,000 samples for ChIP in over 9 species and 25 different tissue types– some I will show you here.
Many of these common challenges with ChIP is just less of an issue for based on experience. And we have also been able to incorporate a number of quality control steps to monitor our process as we go through the ChIP protocol.
ChIP Optimization Steps
Today I want to focus on those ChIP assays that are optimized for those samples that happen to be especially challenging to work in ChIP. We do have an optimized and highly sensitive ChIP protocol that is for low-abundance transcription factors and low input amounts. That is called ChIP High Sensitivity. And this has been proven to work with all these different species and tissue types.
But for each individual sample type, the chromatin protocol preparation usually has to be optimized. For example, the PBMCs, the peripheral blood mononuclear cells, such as T cells are very difficult to analyze, and thus, it’s very difficult to generate sufficient levels of chromatin from these cells. So we have developed a specific protocol from chromatin extraction from these cells, which includes a hypotonic buffer, flash freeze steps, and also additional sonication steps.
Here is an example of an experiment where we perform ChIP-seq against HJK4 mono- and tri-methylation, as well as H3K27 acetylation and tri-methylation in mouse-derived T cells both from wild type mouse and also non-disclosed double knockout mice. By using the modified protocol that we have developed specifically for T cells, we were able to get reliable ChIP-seq data from a variety of different antibodies.
ChIP Challenges from FFPE Samples
Today for this talk, I actually wanted to focus on an even trickier sample type for ChIP purposes, which is FFPE samples. Formalin-fixed, paraffin-embedded tissues is the most common method for tissue preparation for diagnostics, pathology and long term storage of clinical samples.
These tissues, they’re usually fixed in 10% of neutral-buffered formalin for approximately 48 hours. The tissue is then dehydrated through a series of ethanol baths to displace all the water. And then they’re infiltrated in wax.
The advantage with FFPE material is that there is a huge collection of clinical material available worldwide, estimated at over one billion. And all of those have known treatments and outcomes. And many times, DNA from these tissues have already been used for a variety of studies such as genetic analysis looking in to SNP mutations, as well as DNA methylation studies. So if we would be able to perform ChIP on these samples, we would be able to correlate these ChIP profiles with DNA methylation and mutation data from the same sample, and generate a multilevel genetic and epigenetic profiling of primary tissue samples.
In theory, this material can be used to identify epigenetic bio-markers that may be predicative of clinical outcome, and really highlights the value of performing of ChIP in primary tissues. However, now the challenge is that FFPE sample preparation is not standardized. And many of these samples can be very tricky to work with. For example, the time between the resection of the tissue and the actual fixation is essential to generate high-quality FFPE samples. They should be fixed immediately to guarantee high quality.
In animal models, this can be achieved through in vivo perfusion. However, in humans, surgical removement is the only option. And the way of conservation of this tissue between the surgery and the actual fixation is not standardized, and the immediate fixation of the tissue is not guaranteed.
Then you also have the harsh conditions and the long-term storage, which can lead to degradation and alterations in these samples. So this means that there’s a high level of variability between the different human specimens. Not to mention that these samples are usually very small, so it’s difficult to generate soluble chromatin from these samples.
But given all the benefits of these samples, we still felt that it was important to develop and optimize protocols that would include all the necessary steps to try to generate sufficient levels of quality chromatin from these FFPE samples.
Optimization for FFPE ChIP
So these steps that we have included, they will allow you to efficiently remove the paraffin, to generate sufficient yield of quality chromatin without destroying protein integrity of excessive DNA fragmentation. But I also have to note here that no FFPE sample is equal. So many times, individual optimization of the chromatin prep will be needed for certain samples.
Now, there are several ways to assess the chromatin yield in quality from these chromatin preps from FFPE samples. Fluorometric quantifications such as the Qubit together with qPCR-based assays can give you an estimate of the yield of solubilized chromatin.
qPCR can also give you an estimate of the fragmentation levels of these samples. For example, if you do not have sufficient amount of chromatin so you can run it on an agarose gel or run it on a bio-analyzer. Then you also have genome-wide sequencing that could give you information of the chromosome coverage in these chromatin preps.
Here, I show you an example of a chromatin prep that I performed from rat whole brain FFPE tissue slides. I I used a qPCR-based assay to try to determine the levels of chromatin in a soluble fraction versus the insoluble fraction that is the pellet. Just by increasing the number of sonication cycles, I was able to increase the amount of soluble chromatin. But I have to note here that you have to be careful in your attempts to solubilize chromatin so that you do not destroy protein integrity or cause extensive fragmentation of these samples.
Now, since the fixation procedure in model animals can be performed through in vivo profusions, these samples are usually of high quality, and the chromatin prep from these FFPE samples are usually pretty straightforward. However, as I mentioned earlier, human samples, they’re definitely trickier. And the quality of these samples can be highly variable. Here, I show you an example of chromatin preparations from several colon and kidney FFPE samples. You can see the normal in copper, the tumors in purple, and also metastasis here in red.
The soluble chromatin and the insoluble chromatin fractions. These samples, they have been stored under less than optimal conditions with no temperature or humidity control for over 10 years. And again, the yield of soluble chromatin and insoluble chromatin was estimated by qPCR.
There are high levels of variability both between the two tissue types, but also between samples within the same tissue types when it comes to yield of soluble chromatin. While most of these colon samples were able to generate sufficient amounts of soluble chromatin to be considered for ChIP, three out of the four kidney samples, as you can see here, gave way too low levels of chromatin to be able to be considered for ChIP.
I do have to mention, though, that we have included suggestion of additional steps that you can consider to try to increase the levels of solubilized chromatin. In these samples, they proved to be extra tricky when you want to get enough chromatin for ChIP.
Getting Good Yield from FFPE Samples
Considering the many challenges of FFPE samples such as the limited amount of starting material and also this over-fixed state of chromatin, it can be very tricky to generate chromatin sufficient yields and of good quality to be used for ChIP-seq analysis. You also have to have a ChIP protocol that is sensitive enough for these samples. We have developed the high-sensitivity ChIP protocols which is optimized to work on limited cell numbers. And we applied this protocol with the needed optimizations to chromatin generated from FFPE tissues.
As an initial step to determine that your chromatin is of sufficient quality for ChIP purposes, we always recommend that you first perform an immunoprecipitation with a controlled antibody such as the ChIP-validated H3K4 tri-methylation prior to immunoprecipitation with the antibody against your protein of interest. And also, due to the tricky nature of these samples, we always recommend that you use robust and ChIP-validated antibodies.
It’s also important to note that the cellular complexity and density can vary between tissues, and also between tissue sections in large tissues. So this can affect the quality and the amount of chromatin from different parts of the tissue.
In this experiment, I have performed ChIP on rat whole brain as well as micro-dissected hippocampus and prefrontal cortex FFPE tissue slides from rat. I used antibodies against two active histone marks, H3K4 tri-methylation and H3K9 acetylation. While the level of enrichment varied just a little bit between samples, which was likely due to the varied amounts of input chromatin for each reaction, the ChIP profiles were consistent for the active gene GAPDH for both antibodies between these different tissue sections.
We also validated this ChIP reaction on rat whole brain by next generation sequencing. And here, I’ll show you the enrichment at two actively expressed genes showing the expected ChIP profiles.
Now, I also have to mention that the main challenge in these experiments is that the amount of starting material is so low. For this ChIP-seq experiment, I used 200 nanograms of chromatin, which is equivalent to approximately 25,000 cells. This means that the sequencing library complexity will be low. This is the same problem that is encountered in any ChIP experiments where you use limited amount of starting material. Although the overall ChIP-seq data will give you a lower signal than a normal ChIP, the data is still interpretable.
ChIP on Human FFPE Samples
Now, the main challenge is getting the ChIP to work on human FFPE samples, as the quality of these samples can be highly variable, and the colon FFPE samples that I performed ChIP on here, they were over 10 years old and stored on the less than optimal conditions– just laying around the lab. And as we recommend, I first performed a ChIP against H3K4 tri-methylation.
The successful enrichment of GAPDH, both in the normal and tumor sample, over the negative control region, confirmed the integrity of antigens in these chromatin preps, as well as the ability of the antibody to recognize and bind the target even in these heavily crossing samples.
Then I went ahead with several other ChIP-validated antibodies against the two repressive histone marks– H3K27 tri-methylation and H3K9 tri-methylation, as well as RNA Pol II and the transcription YY1, all showing successful enrichment of their positive control regions over the negative control regions. And together, these results confirm the successful generation of quality chromatin from these colon samples, and also the successful immunoprecipitations with all these antibodies on these samples.
Then I went ahead and validated this ChIP by next generation sequencing, both against the histone mark H3K4 tri-methylation that I showed you some examples of here, but also against the transcription factor YY1.
In this first example, we can see peaks in both normal and tumor at the two promoters here. But we do see an additional peak at the PTK7 gene promoter only in the tumor and not in the normal colon. And this is actually a known cancer-associated gene. On the right-hand side, we do again see enrichment of H3K4 tri-methylation in tumor but are not in the normal tissue. (I’ve blinded the gene name here, as this is still unpublished data.)
We also then looked at the H3K4 tri-methyl peak overlap between the normal and tumor sample. And as seen here, there are a number of regions that are marked by H3K4 tri-methylation that are common in both normal and tumor. But there’s also a number of regions with enrichment in only one of the samples, such as the normal or the tumor. So now you can actually look at these peaks and try to correlate this with the clinical treatment and outcome information that you have from these samples.
ChIP Applications in Clinical Samples
What are the importance of actually doing ChIP in these clinical samples? I think I’ve tried to highlight that a little bit through the talk., but I want to go through it in a little bit more detail. And as you know, epigenetic changes have been shown to contribute to wide variety of diseases such as cancer. In normal sense, active promoters are usually associated with low levels of DNA methylation and enrichment of H3K4 tri-methylation. Heterochromatic regions which contains a lot of repetitive elements, on the other hand, they are usually permanently silenced with high levels of DNA methylation and H3K9 tri-methylation.
In tumors, though, we often see the aberrant silencing of tumor suppressor genes by DNA hypermethylation and the associated enrichment of H3K9 tri-methylation. In the heterochromatic regions, on the other hand, we often see DNA hypomethylation, and if the chromatin is still in a repressive state, we will see enrichment of H3K27 tri-methylation. But if it’s now in an open and more active state, we will see enrichment of H3K4 tri-methylation.
I also have to mention here that most of the knowledge that we have generated so far about these system modifications and chromatin regulators come in vitro from experiments. They are performed in cell lines. But it’s also important to try to validate these results in primary tissues. So I will now show you two examples of data that we have generated in collaboration with Manuel Perucho, who is the director of the IMPPC Institute in Barcelona.
In this first example, the Perucho lab have previously found frequent demethylation of an SST1 promoter in human cancers. And here, I show you an example of two different colon cancers showing variable degree of demethylation in the tumors.
When they perform ChIP in cell lines, they found that this hypomethylation was associated with the decrease in H3K9 tri-methylation, but then it increased in H3K27 tri-methylation. So by performing ChIP on these colon FFPE samples, we were able to confirm these changes in the histone modifications in primary tissues in several different colon cancer patients.
And again, in colon cancer case 709 and 726, In the normal tissue, you have high levels of H3K9 tri-methylation and low levels of H3K27 tri-methylation. On the other hand, in the tumor, we now see a switch with reduced levels of H3K9 tri-methylation and an increase in H3K27 tri-methylation.
Further, when we perform ChIP-Seq, as I showed you earlier, on one of those colon samples, we encountered this gene that showed increased levels of H3K4 tri-methylation in the tumor, as I show you here, and also decreased levels of H3K27 tri-methylation compared to the normal tissue.
By performing ChIP on several other colon FFPE samples, we were able to verify that, as you can see here, by qPCR. In the normal tissue at this gene, we show low levels of H3K4 tri-methylation and high levels of H3K27 tri-methylation. In the tumor, on the other hand, we see a switch with reduced levels of H3K27 tri-methylation and increased levels of HJK4 tri-methylation.
In addition, when Sergio Alonso, who’s an associate investigator at the IMPPC, investigated the DNA methylation levels of the CpG Island associated with the promoter of this gene. He found frequent hypo-methylation in several of these colon tumors. He also found increased expression levels in those same tumors. Then all of this correlated well with the changes in histone modifications that we have observed by doing ChIP on these FFPE samples.
So all together, these ChIP reactions on these FFPE samples, they provided us with valuable insights into the alterations and histone modifications in these primary tumors.
Getting Reliable ChIP Data
So to summarize, epigenetic research and the ChIP technology are now expanding into diverse fields such as the clinic, and the ability to do ChIP on low-abundance transcription factors and these difficult samples that I’ve shown you today gives valuable insights into the epigenome, both at normal development and in disease. Performing ChIP on these FFPE samples, that enables a multi-level genetic and epigenetic profiling in primary tissues that comes with pathological, clinical, and outcome information.
I’ve shown you today that FFPE ChIP-seq is now possible on as low as 200 nanograms of chromatin. That is equivalent to approximately 25,000 cells. Also, through the collaboration with Manuel Perucho, we have been able to generate FFPE ChIP data in these colon samples, identifying multiple alteration and histone modifications that are associated with tumorigenesis. I think this really highlights the importance of performing ChIP in these primary tissues.
So what are the outputs of all the ChIP experiments that we have performed here at the Active Motif? Well, working together with all our academic collaborators has allowed us to develop a lot of expertise in ChIP, and also to optimize these specialized ChIP protocols that we are now making commercially available for everyone. Our program is to develop and commercialize these assays, and optimize of them, and try to develop good-quality ChIP-valid antibodies for those of you who want to perform these assays yourself.
We also provide an end-to-end service for those of you who prefer that we do the ChIP for you. We do everything from start to finish. You just send us the samples. We even do the analysis.