ChIP-seq and beyond: new and improved methodologies to detect and characterize protein-DNA interactions

Abstract

Chromatin immunoprecipitation experiments followed by sequencing (ChIP-seq) detect protein-DNA binding events and chemical modifications of histone proteins. Challenges in the standard ChIP-seq protocol have motivated recent enhancements in this approach, such as reducing the number of cells that are required and increasing the resolution. Complementary experimental approaches - for example, DNaseI hypersensitive site mapping and analysis of chromatin interactions that are mediated by particular proteins - provide additional information about DNA-binding proteins and their function. These data are now being used to identify variability in the functions of DNA-binding proteins across genomes and individuals. In this Review, I describe the latest advances in methods to detect and functionally characterize DNA-bound proteins.

Figure 1. Comparison of experimental protocols

Experiments to detect different aspects of DNA binding proteins and chromatin conformation share many of the same steps. a | ChIP-seq for DNA binding proteins such as transcription factors. Recent variations on the standard protocol include using endonuclease digestion instead of sonication (ChIP-exo) to increase the resolution of binding site detection and eliminate contaminating DNA, and amplification after ChIP for samples with limited cells. b | ChIP-seq for histone modifications use MNase digestion to fragment DNA and can also now be run on small samples with the additional post-ChIP amplification. c | DNase-seq relies on digestion by the DNaseI nuclease to identify regions of nucleosome-depleted open chromatin where there are binding sites for all types of factors, but it cannot identify what specific factors are bound.

Figure 2. General analysis pipeline for sequence tag experiments

Experiments using short sequence reads that identify regions with a particular molecular characteristic share many of the same analysis steps. Poor quality reads can be filtered initially, but often the inability to align these reads to the genome sufficiently removes bad sequences. Alignment using one of many possible software programs (Table 1) is followed by filtering artifacts that arose during the PCR amplification step when sequencing, or that appear due to the underrepresentation of certain sequences in the reference genome, such as pericentromeric satellite sequences. Often, reads aligning to more than some number of genomic locations are removed. For experiments identifying independent locations, ‘peak’ calling tools (Table 1) identify genomic regions of signal enrichment indicating a bound protein, histone modification, or open chromatin. In contrast, chromatin interaction experiments use aligned paire-end reads to find evidence of interacting distal genomic regions. DNaseI footprints indicate local protection from DNaseI digestion within a larger DHS region due to a bound protein. The distribution of alleles in sequences spanning heterozygous variants can be analyzed to determine if a bias towards sequences with one of the two alleles exists. This may reflect a functional difference caused by the underlying genotype.

Figure 3. DNaseI footprints correspond to bound proteins

The distribution of DNaseI digestion sites with DNaseI hypersensitive regions is not uniform with peaks and troughs in the signal, where troughs are due to protection by bound proteins. Transcription factor binding motif databases such as JASPAR can be searched using the sequence from each footprint to predict what factor is bound. Shown here is data from the proximal promoter region of the FMR1 gene. DNaseI footprints had been identified previously at this locus in lymphoblastoid cells. More recent data from DNase-seq was used to recapitulate these results in a single experiment. Note: This figure is taken from . It is figure 1C in that manuscript.

Figure 4. Detecting chromatin interactions

In three-dimensional space, distal genomic regions on the same or different chromosomes interact mediated by one or more DNA binding proteins. a | Chromatin conformation capture experiments use a ligation step to join distant fragments that are interacting in three-dimensional chromatin space providing information on possible targets for DNA-bound proteins. b | ChIA-PET similarly detects chromatin interactions using a ligation step to pair non-adjacent interaction regions, but using a ChIP step to more specifically identify interactions with a particular bound protein, such as RNA PolII. It should be noted that the DNA that is actually sequenced as part of the paired-end sequencing do not necessarily correspond to the precise region of interaction but are dictated by the presence of restriction enzyme targets. c | The UCSC Genome Browser contains data from both ChIA-PET and 5C experiments. Shown here is a 275Kb region showing interactions near the transcription start site of the ST7 gene. For both experiments, solid blocks represent the sequenced paired ends with lines connecting them when they appear on the same chromosome. Darker shading indicates more confidence in this mapping based on multiple instances of complementary paired end sequences. The ChIA-PET experiment was performed using a RNA PolII antibody, and the corresponding signal from RNA PolII binding is displayed immediately below. Also included beneath these chromatin interaction annotations are signals from DNase-seq experiments in the same cell type. Regions of enriched DNase-seq signal indicate nucleosome-depleted DNA that represents putative regulatory elements. Together, chromatin interaction data provide clues as to the gene targets for these regulatory regions. Note: CTCF weblogo from, figure 2.

Figure 5. Allele-specific bias in a CTCF ChIP-seq experiment

Sequence-based experiments allow for the investigation of functional differences across individuals due to their underlying genotype. This schematic depicts a region with an enriched number of sequence reads from a ChIP-seq experiment, where each red and blue box indicates an aligned read with blue reads aligned to the forward strand, and red reads to the reverse strand. As is typical of a factor ChIP-seq experiment, forward strand reads accumulate 5′ of the site while reverse strand reads accumulate 3′ of the site. Contained within this locus is a heterozygous polymorphism, denoted by A and T bases. Only one-third of the spanning reads contain the T allele while two-thirds contain the A allele indicating an allelic-imbalance.