Chromatin profiling using ChIC/CUT&RUN

ChIC/CUT&RUN has been used to profile DNA-protein interactions in a variety of cell lines and tissue samples and a multitude of model organisms, including yeast, plants, and animal cells. So far, all these studies focused on mapping histone modifications and transcription factor binding sites.

The first report of CUT&RUN profiled the histone modification H3K27me3 in a human cell line and H2A in yeast cells (Skene and Henikoff, 2017). Comparison of the equivalent data sets generated by ChIP-Seq and ChIC/CUT&RUN demonstrated close similarities between the three different chromatin-profiling techniques in discovering ‘peak’ regions.

Over the last decade, ChIP-seq has been the predominant method to identify the genome-wide location of transcription factor binding sites. Compared to ChIP, the new ChIC/CUT&RUN  method is more cost-effective, quicker to use, requires only a fraction of starting material including single cells, has a more favorable signal-to-noise ratio, detects more defined ‘peaks’, and is compatible with automation.

This method is compatible with unfixed, native chromatin, even for transcription factor profiling, and therefore overcomes some of the difficulties associated with cross-linked ChIP protocols. ChIC/CUT&RUN has been used to profile multiple transcription factors, such as CTCF and pluripotency factors, and large chromatin-associated complexes, such as Polycomb Repressive Complexes and chromatin remodellers.

Table 6: ChIC/CUT&RUN with ChIP seq

During ChIC/CUT&RUN, an antibody against the protein of interest (such as a histone with a specific modification or a transcription factor) is used to guide Protein A/G-tagged MNase to the region in the genome where the protein is located. Activating the MNase then excises only those DNA sequences that are close to the protein of interest, and the small fragments are released from the nucleus. These DNA fragments are collected and used to generate NGS libraries with a low-input DNA library prep kit.

The protocol can be separated into five parts:

1. Nuclei extraction and binding to beads

​ChIC/CUT&RUN typically uses fresh, unfixed samples as starting material, but protocol adjustments can allow the use of samples cryopreserved in 10% DMSO (Janssens et al ., 2018; Skene et al., 2018).

​Very little material is necessary to perform the protocol, and the general recommendation is to start with less than 500,000 cells (mammalian).

Cells or nuclei released with the Henikoff recommended nuclear extraction buffer are bound to Concanavalin A magnetic beads, which have unique saccharide-binding properties. Other nuclear extraction protocols can be used providing their compatibility (see suggestions below about the use of Triton X-100 containing buffers). Using magnetic stands allows for easy washing of the samples during the rest of the protocol.

2. Permeabilization and antibody binding

The nuclei bound to Concanavalin A beads are permeabilized and simultaneously incubated with an antibody against the protein of interest (primary antibody) in a buffer containing digitonin and EDTA. EDTA allows the rapid cessation of cell metabolism and thereby inhibits endogenous DNAse activity, preserving the chromatin and reducing the overall background signal. The duration of this step can be adapted by the user; generally, ranging from 2 h to overnight. Antibody dilution of 1:100 or 0 .5 1 .0 µg is recommended as a starting point, but the amount of antibody used should be optimized. It is advised to always include positive (α-H3K27me3) and negative control (isotype control IgG) samples. Of note, Protein A and Protein G have different binding efficiencies to different antibody species (see here). Even though antibody compatibility is increased by using Protein A/G over Protein A alone, it might be necessary to use a secondary antibody in some cases. A secondary antibody guides MNAse to the target area by increasing the number of Protein A/G binding areas and thereby helps to recover low abundant target sequences.

3. MNase binding and cleavage of target sequences

To allow Protein A/G-MNase to be directed to the antibody-bound genomic target regions, the fusion protein is diluted in a wash buffer containing digitonin and incubated with the nuclei. Unbound enzyme fusion is then washed away, and MNase activity is activated at 0°C by adding Ca2+. Even though the cleavage itself is not particularly temperature-sensitive, the subsequent diffusion of the cleaved DNA fragments is sensitive, and a temperature rise would result in higher background. The digestion time can be adjusted if the final material contains a disproportional amount of high molecular weight fragments.

4. DNA recovery

The MNase activity is stopped by the addition of a STOP buffer containing EGTA. This buffer can optionally contain heterologous spike-in DNA to help calibrate the CUT&RUN profiles during data processing. The MNase generated fragments are released from the nuclei by increasing the incubation temperature and cleaned up using phenol/ chloroform/isoamyl alcohol extraction followed by ethanol precipitation.

A variation of the protocol can further restrict the premature release of the MNase-fusion protein from its binding site, its diffusion, and potential for non-specific cleavage. This version of the protocol uses a combination of low salt buffers and a high Ca2+ concentration to activate MNase activity and is ideal for targets mainly found in active open chromatin, but it can also be used when an antibody shows high levels of background signal.

5. Library preparation and sequencing

Following the clean-up of the cleaved DNA fragments, standard end repair and adapter ligation methods are used to generate low-input DNA libraries according to the manufacturer’s guidelines. The Henikoff lab originally recommended a TRUseq library preparation approach, and besides, there have now been multiple reports using the NEBNext® Ultra™ II DNA Library Kit. Alternatively, similar to the automated CUT&RUN protocol (Jenssens et al . 2018), the phenol/chloroform/isoamyl alcohol extraction steps can be omitted, and the released DNA fragments can be directly used in the end repair and adapter ligation-based protocol.

The size distribution and concentration of libraries can be determined by running capillary electrophoresis (eg Bioanalyzer or TapeStation). Multiple libraries can be pooled together to obtain about 8 million paired-end sequencing reads per library. Due to the low background of ChIC/CUT&RUN libraries, 8 million paired-end reads are enough to profile histone modifications and even transcription factors.

Most conventional ChIP-seq data analysis tools can analyze ChIC/CUT&RUN data. There are a few analysis tools, such as the SEACR peak caller (Meers et al., 2019b), which are specifically designed by the Henikoff lab for ChIC/CUT&RUN data. For calibrated ChIC/CUT&RUN, the heterologous spike-in DNA added with the STOP buffer can be used to normalize signals across samples. Alternatively, E. coli DNA carried over from the recombinantly-produced Protein A/G-MNase can be used as a spike-in.​

In general, sample preparation is straightforward for ChIC/CUT&RUN as it can use fresh, unfixed samples. A single-cell suspension needs to be achieved in a way that is suitable for the cell type used. This can include using dissociation reagents like Accutase™, scraping cells off cell culture dishes, or mechanically dissociating tissues.

For ease of collection, samples can be cryopreserved in 10% DMSO and frozen down in a Mr.Frosty isopropyl alcohol chamber. Sample fixation is not required for either of the methods, however, if the beads clump during washes, then a light fixation of the sample using 0 .1% formaldehyde for only 2 min at room temperature before antibody incubation is beneficial.

As for ChIP, not all antibodies will work for ChIC/CUT&RUN. Most antibodies are not yet tested for compatibility with these newer methods, therefore testing and optimization will be required by the end user. ChIP-grade antibodies seem to largely work for ChIC/CUT&RUN, especially when shown to work for native ChIP.

When trialing antibodies that are not labeled as ChIP grade, assuming specificity is fully characterized, then it’s good to start with choosing antibodies that recognize the native form of the protein of interest; for example, antibodies that have been shown to work in immunoprecipitation or immunocytochemistry. Equally, when considering the concentration of antibody to use in a ChIC/CUT&RUN experiment, the concentration recommended for immunofluorescence-based methods seems to be a good starting point. Compatibility to Protein A and Protein G needs to be checked, and then appropriate secondary antibodies need to be used if necessary.

As with any other experiment type, including the correct controls is critical to ensure the experiment has worked as expected as well as to easily pinpoint areas to troubleshoot should the experiment fail.

Unlike ChIP, no input sample needs to be included in the protocols as a non-antibody guided MNase treatment or tagmentation would simply result in the identification of accessible chromatin. An IgG control should be included to get the background from the sample and set the baseline for the experiment. In terms of antibody controls, the Henikoff lab recommends using H3K27me3 as a positive control for ChIC/CUT&RUN experiments. The use of total unmodified histone controls, such as total H3, can be considered to allow the proportional representation of histone modifications.

The optimization of the ChIC/CUT&RUN protocol is required at several stages when using this technique.

In the initial versions of the protocols, the digitonin concentration used in the wash buffers needed to be tested to ensure efficient permeabilization of nuclei. This test might not be necessary if you follow improvements to the protocols, specifically by adding NP40 to the wash buffers. However, the efficiency of nuclei permeabilization should be kept in mind when using either technique on new material.

It is important to titrate the amount of antibody used per reaction, starting with recommended dilutions for ChIP or immunofluorescence assays.

Primary antibody incubation for 1 h at room temperature is sufficient, but this can be extended to 1 to 5 days in the cold room. This step can be optimized for the antibodies used in each experiment, keeping in mind that prolonged incubation might lead to an increase in background signal and, therefore, a less favorable signal-to-noise ratio overall.

The use of a secondary antibody can be considered for ChIC/CUT&RUN if experiencing low recovery with primary antibodies alone. Another reason to include a secondary antibody is to avoid an unfavorable pairing of Protein A/G with the primary antibody, which can be achieved by using a more favorable secondary antibody class.

MNase digestion and Tn5 tagmentation time can be adjusted depending on the protein of interest. Less abundant proteins may need more time to allow for the full recovery of all sites. It is important to keep in mind that prolonged digestion or tagmentation might result in untargeted cleavage and, therefore, higher background signals.

How many cells should I use as a starting material?

The genome of my model organism is smaller than human/mouse, should I increase the cell number to start with? Exceeding 500,000 cells per sample is not recommended. Generally, a good starting point is 50,000 cells. Unlike ChIP, ChIC/CUT&RUN is very sensitive and does not need millions of cells as starting material. Using too many cells will reduce the yield and, more importantly, the complexity of the library.

If I use more cells, should I increase the amount of Concanavalin A beads I use?

The number of cells used per experiment should not exceed 500,000 cells. Since the number of beads in the protocol is optimized for the use of 50,000 to 500,000 cells, there should be no need to increase the number of beads recommended in the protocol.

How should I cryopreserve my cells to make them compatible with ChIC/CUT&RUN?

It is recommended to use 10% DMSO in an appropriate buffer or media and slowly freeze using a Mr. Frosty isopropyl alcohol chamber. Flash freezing is not recommended.

I have cells fixed and frozen for ChIP, can I use them for ChIC/CUT&RUN? 

Depending on the fixation and cryopreservation conditions used for the ChIP sample, it might be possible. However, it is unlikely to work with the standard protocols because cell numbers for ChIP samples are likely to exceed recommendations for ChIC/CUT&RUN samples and the strong fixation conditions (eg double fixation and quenching) used will impair MNase or Tn5 activity.

The problem of cell numbers in ChIP samples exceeding the recommended cell starting material for ChIC/CUT&RUN can be easily circumvented by splitting the sample into multiples. However, the other problems of fixation and freezing of samples remain. Fixation used in ChIP protocols is likely too harsh to be compatible with either ChIC/CUT&RUN, as it will most likely impair the ability of the MNase or Tn5 to fragment/tagment the target sequences, and multiple reports have hinted at this incompatibility. Additionally, there is a possibility of epitope masking due to fixation. Sample fixation is, therefore, not recommended. When cryopreserving cells for ChIC/CUT&RUN, the use of 10% DMSO and a Mr.Frosty isopropyl alcohol chamber to gently freeze the samples is required. Flash freezing included in most ChIP protocols is not recommended.

My beads clump during the washes, what am I doing wrong?

The problem of clumping can be caused by using too high a ratio of cells to beads or the lysis of nuclei/cells in the digitonin-containing buffer, which would release DNA and lead to clumping. The first thing would be to check that the recommended number of cells is not exceeded. Furthermore, a light fixation with formaldehyde (0.1% formaldehyde for 2 min at room temperature) before incubation with the antibody is recommended to reduce the clumping of beads in digitonin-containing wash buffers. However, keep in mind that fixation could affect epitope availability and, therefore, will need to be tested for each antibody.

Why is there EDTA in the buffer for the primary antibody incubation?

The addition of EDTA in the permeabilization and antibody incubation buffer is recommended as it chelates Mg2+ and thereby stops any ATP-dependent cellular processes, including replication and chromatin remodeling, and it will also stop endogenous DNases.

Where can I get the Protein A/G-MNase or Protein A-Tn5 from?

Both fusion proteins were initially provided by the Henikoff lab. There are protocols on how to make your own fusion proteins (Kaya-Okur et al., 2019; Meers et al., 2019a), whereas the plasmids, Protein A-MNase, Protein A/G-MNase, and Protein A-Tn5 are all commercially available from reagent suppliers.

Concerning E. coli DNA carryover, what is a normal percentage of E.coli reads to sample reads?

The carryover depends on the amount of starting cells as well as the abundance of the antibody epitope, with an inverse correlation between the number of starting cells and the number of sequencing reads that map to E. coli DNA. The carryover also depends on the amount of Concanavalin A beads used, whereby the more beads used, the more E.coli reads. Overall, calibration using E.coli carryover DNA should only be used for experiments that are comparable in the number of cells and beads used.

As a guide, the Henikoff lab has reported a range of E. coli read percentages from 0.01% to 11.5% of the total reads, with IgG samples usually having a higher amount of E. coli reads (2% to 11.5%). They furthermore reported that when using only a few hundred or thousand cells in an experiment, the percentage of E. coli reads can reach about 30% to 70%. Additionally, when testing a commercially available fusion protein they only got 0.01% of the total reads back as E.coli reads, which might be too little to use for data normalization.

Is there a way of quality controlling the experiment before sequencing?

Similar to ChIP, qPCR could be used to examine carefully designed regions of interest. Keep in mind that the reactions need to be performed on the amplified library and not on the generated DNA fragment pool. In the case of ChIC/CUT&RUN, the released fragments often contain large genomic regions that have not been directly targeted by the antibody-directed MNase. These fragments will not contribute to the libraries prepared from these samples, but they would be amplified by qPCR. To avoid false positives, the template for qPCR should be an amplified library, and appropriate controls should be carried out as well.

The Bioanalyzer/TapeStation trace for my transcription factor experiment looks just like the IgG control, what did I do wrong?

Good news, most likely you didn’t do anything wrong! Provided that your positive control sample (eg H3K27me3) returned a nice nucleosomal patterning, the only way to know whether your transcription factor sample has worked is to sequence the samples.

Unfortunately, it is quite common for transcription factor libraries to look very similar to the IgG libraries, and yet after sequencing, they still produce good-quality data sets. You could opt for a qPCR approach to compare signals in the IgG versus the transcription factor libraries for known transcription factor target sites.