Design principles of chromatin organization


The introduction of high-throughput methods for determining nucleosome organization across entire genomes has provided a new perspective on understanding and modeling the regulation of eukaryotic gene expression. These studies have shown, first of all, that promoters are often depleted of nucleosomes compared with coding regions. Second, functionally related genes share nucleosome occupancy patterns in their promoters. Third, it appears that at least for a fraction of genome, promoter regions of highly transcribed genes are more depleted of nucleosomes compared with promoter regions of repressed genes. The question of how nucleosome occupancy and regulation of gene expression are related, appears to be the most complex and challenging issue since first, seminal studies of this relationship. This is partly due to the fact that there are multiple additional factors, such as chromatin remodelers, and the competition with transcription factors (TFs), influencing gene expression.


Intrinsic DNA sequence preferences of nucleosomes, has been a long-standing question for more than three decades. Yet, a general answer to this question at the genome-wide level is still debated and a matter of active research. It appears that there are two dominant sequence features for nucleosome positioning. First, nucleosomes are depleted from sequences enriched in poly(dA:dT) both in vivo and in vitro. This depletion is significantly stronger in vivo than in vitro. Second, nucleosomes are preferably positioned in sequences with AA/TT/AT and GG/CC/CG dinucleotides repeated with a period of about 10 nucleotides. The second sequence feature is observed to be stronger in vitro than in vivo, and overall, this periodicity shows a statistically weak signal.Comparison of genome-wide measurements of nucleosome occupancy in vivo and in vitro suggests that in a large fraction of the yeast genome in vivo, predominantly outside promoter regions enriched in poly(dA:dT) tracts, nucleosome occupancy is not intrinsically determined by the sequence, but it can be rather interpreted using a statistical positioning model. This model assumes the existence of physical barriers at specific genomic locations, inducing nucleosome ordering in the vicinity of such barriers. It was shown recently that the in vivo nucleosome occupancy can be reconstructed in a cell extract, in vitro, in the presence of adenosine triphosphate (ATP). This discovery suggests that barriers for statistical positioning of nucleosomes operate in an ATP-facilitated manner. This raises a key question: what mechanism provides such physical barriers for statistical nucleosome positioning? The latter question has become even more mysterious after it was shown that the transcription initiation complex is not an obvious barrier against which nucleosomes are organized. It was shown that specifically bound TFs might provide a barrier for statistical nucleosome positioning only for a limited fraction of the yeast genome, thus leaving the latter question open.


We hypothesize that nonspecific TF-DNA binding might provide such barriers genome-wide. We show that the free energy of nonspecific TF-DNA binding regulates the nucleosome occupancy genome-wide in yeast, in vivo. In particular, genomic regions depleted in nucleosomes possess a significantly lower free energy of nonspecific TF-DNA binding than genomic regions enriched in nucleosomes. We are interested in understanding the effect of nonspecific TF-DNA binding on the nucleosome occupancy in different eukaryotic organisms, such as yeast, C. elegans, and Drosophila.