Transcription factors regulate the expression of genes by binding to certain DNA sequences. But the outcome can be markedly different, depending on whether the binding is stable or short-lived. See Letter p.251
To reproduce, differentiate or even just respond to changes in their surroundings, cells need to control the expression of thousands of genes. One way of doing this is to use transcription factors — proteins that bind to regulatory regions on their target genes and either activate or repress the transcription of DNA into RNA. Over the past decade, researchers have analysed the binding sites of hundreds of these proteins on the genomes of many organisms and cell types, and measured the gene-expression patterns within the same cells. In such studies, the overall degree of occupancy by a transcription factor at a regulatory region is commonly interpreted as an indication of the protein's ability to control the expression of the gene. However, transcription factors also bind to thousands of genes in a weak, and probably non-functional, manner1. On page 251 of this issue, Lickwar et al.2 illuminate this matter by reporting the results of a systematic, genome-wide study of the binding dynamics of a particular transcription factor. The authors find that transcription levels have a stronger link to the kinetics of binding than to the total occupancy of the factor.
The DNA-binding sites of the transcription factor Rap1 along the genome of the yeast Saccharomyces cerevisiae were mapped more than a decade ago3. The mapping used a genome-wide protein–DNA binding assay, known as chromatin immunoprecipitation (ChIP)-on-chip, or ChIP-seq, which identifies the genomic locations of a transcription factor over a huge number of live cells and therefore averages the transcription-factor occupancy over a large population. This is still the method of choice in similar genome-wide studies. However, a high occupancy of a transcription factor at a specific site — as detected by this technique — can mean either that the factor is constantly bound to this DNA location in some of the cells, or that it is transiently bound in many cells.
To distinguish between the two possibilities, Lickwar et al.2 adapted a strategy, previously used to measure the turnover of DNA-bound proteins4,5,6, to address the question of transcription-factor binding stability. The authors created yeast cells that produced two Rap1 variants, Rap1–Flag and Rap1–Myc, each one with a 'tag' consisting of a specific peptide that could be recognized by antibodies. Furthermore, Rap1–Flag was produced constantly, whereas Rap1–Myc's expression was experimentally inducible. The authors then measured the binding of each Rap1 variant to the yeast genome in a dense time series after Rap1–Myc induction. Although the inducible protein quickly outcompeted Rap1–Flag at some genomic sites, it was slowly incorporated into other sites, which suggested that Rap1 binding to the first sites was less stable than to the other sites.
The researchers then applied a mathematical model5 to estimate the rate of Rap1 turnover at more than 400 target genes. They found that, among genes with high Rap1 occupancy, those with slower Rap1 turnover showed higher transcription levels than those with faster Rap1 turnover. That is, the transcription level depended on how long Rap1 remained bound. Of note, the genomic sites that exhibited fast Rap1 turnover in this analysis2 have previously been reported5,6 to have fast turnover rates of nucleosomes (protein complexes around which DNA is packaged) and of the general transcription factor TBP (which facilitates the binding of the core transcriptional machinery). Overall, the results are consistent with those of other studies7,8 that showed that transcription factors and nucleosomes compete for some genomic sites and that this competition leads to inefficient transcription.
Lickwar et al. suggest a model for the binding dynamics of transcription factors that activate transcription (Fig. 1). In this model, on binding to a target site, the factor has to recruit the core transcriptional machinery. This process takes some time. Therefore, if the factor's binding to the DNA site is unstable, it will not lead to productive transcription. Indeed, it has been shown9 that short, repeated pulses of Msn2 — another transcription factor — into the cell nucleus do not activate target genes, whereas longer pulses do. Therefore, for transcription factors to be effective activators, they require stable binding to their target DNA.
Moreover, the researchers speculate that a constant turnover or 'treadmilling' of nucleosomes and transcription factors acts as a distinct mechanism for transcriptional regulation. Unlike static gene repression10, in which transcription is prevented by the nucleosome's protection of DNA, a site that has a treadmilling transcription factor is poised for activation. When, somehow, the nucleosome is removed or its affinity for DNA is decreased, the factor can quickly achieve stable binding to its target sequence and so activate the gene's transcription. Several mechanisms would allow for the targeted eviction of nucleosomes, including chromatin-remodelling enzymes (which move nucleosomes on DNA), chemical modifications of histones (the protein components of nucleosomes) or the replacement of certain histones with specific variants.
Lickwar and colleagues' study2 explains how different regulatory regions can present similar levels of transcription-factor occupancy and different transcriptional levels. But it also raises further exciting questions. Do the different turnover rates of transcription factors play a key part in gene regulation? Or do they just reflect some other aspects of the transcriptional process, such as stabilization of protein–DNA binding by interactions with the transcriptional machinery? Are nucleosomes the only competition for factor binding to DNA, or is competition with other transcription factors important too?
To fully understand how transcription factors work, we should consider not only their overall binding occupancy, but also their binding dynamics. This line of research will form the basis for a much-needed quantitative understanding of transcription regulation kinetics.
Macarthur, S. et al. Genome Biol. 10, R80 (2009).
Lickwar, C. R., Mueller, F., Hanlon, S. E., McNally, J. G. & Lieb, J. D. Nature 484, 251–255 (2012).
Lieb, J. D., Liu, X., Botstein, D. & Brown, P. O. Nature Genet. 28, 327–334 (2001).
Schermer, U. J., Korber, P. & Hörz, W. Mol. Cell 19, 279–285 (2005).
Dion, M. F. et al. Science 315, 1405–1408 (2007).
van Werven, F., van Teeffelen, H., Holstege, F. & Timmers, H. Nature Struct. Mol. Biol. 16, 1043–1048 (2009).
Polach, K. J. & Widom, J. J. Mol. Biol. 254, 130–149 (1995).
Voss, T. C. et al. Cell 146, 544–554 (2011).
Hao, N. & O'Shea, E. K. Nature Struct. Mol. Biol. 19, 31–39 (2011).
Hager, G. L., McNally, J. G. & Misteli, T. Mol. Cell 35, 741–753 (2009).
About this article
Critical Reviews in Biochemistry and Molecular Biology (2013)