Improvement of the chromatin immunoprecipitation (ChIP) assay by DNA fragment size fractionation.

The chromatin immunoprecipitation (ChIP) assay is a valuable technique that allows researchers to establish what proteins (e.g., transcription factors) bind in vivo to a specific region of DNA, usually a gene promoter (1). In vitro proteinDNA binding assays, such as the electrophoretic mobility shift assay, are useful in identifying potential transcription factors that bind to a specific DNA sequence and regulate transcription of a gene. However, it is not unusual to identify several different transcription factors, or isoforms of a transcription factor, that can bind, at least in vitro, to a particular sequence. Thus, determining which transcription factor is physiologically relevant, or determining whether the transcription factor bound to a promoter region changes under different physiological or pathophysiological conditions, can be difficult if not impossible without some in vivo analysis. The ChIP assay, which involves (i) crosslinking proteins to DNA with formaldehyde, (ii) shearing of DNA in small fragments, (iii) immunoprecipitation of fragments with an antibody directed against a known transcription factor, and (iv) PCR amplification of the DNA region containing the sequence of interest, offers the ability to obtain a “snapshot” of which proteins are bound to specific cis-elements in vivo. It is not unusual for a given transcription factor to bind to several sites in a promoter, particularly when considering the fact that eukaryotic promoter regions often span several kilobases. Thus, for this technique to provide useful information, the DNA must be efficiently sonicated so that the sizes of DNA subjected to analysis are no longer than 200–300 bp. If this is not achieved, one can neither accurately nor confidently assign the binding of the transcription factor to a defined region of the gene. Proposed solutions have included increasing the sonication time, introducing glass beads during sonication, and using a microtip on the sonicator to enhance shearing of the DNA (3,4). It is also possible to subject the DNA to appropriate restriction enzymes, although in many cases the restriction is not 100% complete, and/or useful restriction sites are not present or accessible. In this report, we describe a modified ChIP assay that includes a DNA fragment fractionation step that provides a solution to the problems described above. Our own experiments, which probed for the presence of the transcription factor CCAAT/enhancer binding protein α (C/EBPα) on the promoter of the gene coding for phosphoenolpyruvate carboxykinase (PEPCK), illustrated the problem associated with insufficient shearing of DNA. It had been previously shown by in vitro binding assays (5,6) that C/EBPα can bind to four different sites on the PEPCK promoter; three of these are clustered in a region spanning -320 to -230, while another is centered at -85 (Figure 1). We decided to investigate whether C/EBPα was bound to the -320/-230 region in isolated hepatocytes using the ChIP assay. The ChIP assay was carried out as described previously (1), incorporating the use of glass beads during the sonication step. Immunoprecipitation of DNA fragments was performed using an antibody for C/EBPα (Santa Cruz Biotechnology, Santa Cruz, CA, USA), and PCR analysis was performed to assess the binding of C/EBPα to the -310/-230 region by using primers -333 and -149, which amplify a fragment of 184 bp. As shown in Figure 2A, lane 1, a PCR product of this size was observed. However, when these same immunoprecipitated fragments were subjected to PCR using primers -1227 and -1008, a region that does not contain C/EBP binding sites, an appropriate size fragment of 220 bp was also observed (Figure 2A, lane 2). These data suggested that the sonication step was inefficient and that some longer DNA molecules were immunoprecipitated that contained both regions. This possibility was verified by using the primer pairs -1227 and -149, which amplified a fragment of approximately 1 kb (Figure 2A, lane 3). The presence of these longer, unsheared fragments prevented us from being able to delineate the site to which the C/EBPα was binding. Extending the time or intensity of sonication did not change the results obtained (data not shown), so we decided to modify the standard protocol such that the maximal length of DNA fragments analyzed would be limited to those less than 250 bp. Following the step in which the cross-links between the protein and DNA complexes were reversed, the immunoprecipitated DNA was fractionated on a 1% low melting temperature agarose gel, with 100-bp DNA markers run in an adjacent lane. The region of the gel containing immunoprecipitated DNA fragments of 100–250 bp was excised with a scalpel, and the DNA was purified from the agarose as described previously (2) and subjected to PCR analysis. Figure 2B shows the results of our modified ChIP assay and a number of control experiments to demonstrate (i) the effect that the fractionation step had Benchmarks