Research article

Position specific variation in the rate of evolution in transcription factor binding sites

Alan M Moses¹ , Derek Y Chiang² , Manolis Kellis^3,5 , Eric S Lander^4,5 and Michael B Eisen^1,2,6

¹  Graduate Group in Biophysics, University of California, Berkeley, CA 94720, USA

²  Department of Molecular and Cell Biology, University of California, Berkeley, CA 94720, USA

³  Department of Computer Science, Massachusetts Institute of Technology M.I.T., Cambridge, MA 02139, USA

⁴  Department of Biology, M.I.T., Cambridge, MA 02139, USA

⁵  Whitehead/MIT Center for Genome Research, Cambridge, MA 02139, USA

⁶  Department of Genome Sciences, Life Sciences Division, Ernest Orlando Lawrence Berkeley National Lab Berkeley, CA 94720, USA

author email corresponding author email

BMC Evolutionary Biology 2003, 3:19doi:10.1186/1471-2148-3-19

Published:	28 August 2003

Abstract

Background

The binding sites of sequence specific transcription factors are an important and relatively well-understood class of functional non-coding DNAs. Although a wide variety of experimental and computational methods have been developed to characterize transcription factor binding sites, they remain difficult to identify. Comparison of non-coding DNA from related species has shown considerable promise in identifying these functional non-coding sequences, even though relatively little is known about their evolution.

Results

Here we analyse the genome sequences of the budding yeasts Saccharomyces cerevisiae, S. bayanus, S. paradoxus and S. mikatae to study the evolution of transcription factor binding sites. As expected, we find that both experimentally characterized and computationally predicted binding sites evolve slower than surrounding sequence, consistent with the hypothesis that they are under purifying selection. We also observe position-specific variation in the rate of evolution within binding sites. We find that the position-specific rate of evolution is positively correlated with degeneracy among binding sites within S. cerevisiae. We test theoretical predictions for the rate of evolution at positions where the base frequencies deviate from background due to purifying selection and find reasonable agreement with the observed rates of evolution. Finally, we show how the evolutionary characteristics of real binding motifs can be used to distinguish them from artefacts of computational motif finding algorithms.

Conclusion

As has been observed for protein sequences, the rate of evolution in transcription factor binding sites varies with position, suggesting that some regions are under stronger functional constraint than others. This variation likely reflects the varying importance of different positions in the formation of the protein-DNA complex. The characterization of the pattern of evolution in known binding sites will likely contribute to the effective use of comparative sequence data in the identification of transcription factor binding sites and is an important step toward understanding the evolution of functional non-coding DNA.