In:
Proceedings of the National Academy of Sciences, Proceedings of the National Academy of Sciences, Vol. 109, No. 38 ( 2012-09-18)
Abstract:
We conclude that the periodic distribution of dinucleotides defines nucleosome positioning to a large extent. With the simple model presented here, both the position of nucleosomes and their affinity for certain DNA sequences can be predicted within several bps and several k B T . Three parameters suffice to predict nucleosome occupancy, in vitro and in vivo, with correlation coefficients of respectively 0.74 and 0.66. Comparison with experimental data confirms that positioning after salt reconstitution is dominated by the histone tetramer, whereas positioning in vivo depends on the fully wrapped histone octamer. The energy landscape for nucleosome positioning presented here provides a quantitative framework to distinguish the influence of other DNA-binding factors on nucleosome dynamics, such as that of transcription factors and chromatin remodelers. By allowing variation in the chemical potential, we may identify other effects that regulate nucleosome positioning, such as histone variants, post-translational modifications of histones, or the presence of DNA-binding factors that compete for the same DNA. In yeast and other eukaryotic cells, for example, the +1 nucleosome, which represents the first nucleosome downstream of the transcription start site (TSS) of a gene, is well positioned, and the degree of nucleosome positioning decreases progressively downstream in the coding region. Upstream of the TSS, a region that is depleted of nucleosomes is observed ( 5 ). If DNA sequence is the main determinant for the nucleosome distribution around the TSS, then the apparent chemical potential should be constant around the TSS. We clearly observed a decrease in chemical potential in in vivo chromatin at the +1 nucleosome and a slight increase around the -1 nucleosome, suggesting that factors other than DNA-histone interactions are responsible for the increased occupancy of histones that is observed around the TSS. To fit in vivo nucleosome occupancy maps, it was necessary to expand the positioning window to 147 bp; this suggests that in vivo, it is the full octamer that defines nucleosome positions, contrary to nucleosome positioning after salt-dialysis reconstitution, which is dominated by the 74 bp occupied by (H3/H4) 2 histone tetramer. Such behavior is similar to the shifted positions that are preferred after thermal or enzymatic remodeling of reconstituted chromatin fragments in vitro ( 4 ). To predict genome-wide nucleosome positions obtained in vivo, we expanded our model for binding multiple histone octamers on the same DNA substrate. Binding of a histone octamer will block potential binding sites for other histone octamers on the same DNA molecule and will position histone octamers around it, which is known as statistical positioning. A fourth parameter, the chemical potential, accounts for the average sequence-independent nucleosome affinity. A two-parameter fit of the dinucleotide periodicity and the chemical potential to 20-kbp sections of nucleosome occupancy maps from salt dialysis-reconstituted chromatin ( 3 ) accurately described experimentally determined genome-wide nucleosome distributions. Our model is based on a periodic probability function for the distribution of the dinucleotides TA, TT, AA, and GC. The choice of these dinucleotides is based on published studies ( 3 ). The three parameters that describe this function are the dinucleotide periodicity, probability amplitude, and nucleosome-binding window. We showed that a 10.1-bp periodicity together with a probability amplitude of 0.2 and a binding window of 74 bp, representing DNA binding by the (H3/H4) 2 histone tetramer, accurately reproduces the free energy of nucleosome formation and the preferred nucleosome position with nucleotide base-pair resolution for all documented nucleosome positioning sequences studied in vitro. Nucleosomes assembled in vitro by salt dialysis have a high preference for certain DNA sequences and tend to avoid other sequences. Genome-scale nucleosome mapping suggests that such sequence preferences might dominate nucleosome organization in vivo over enzymatic processes that (re-)position nucleosomes in the eukaryotic nucleus. To describe the influence of the DNA sequence on nucleosome organization, various models have been proposed to predict DNA sequence-dependent nucleosome affinity. These models are based on either the structural parameters obtained from crystal structures of DNA in the nucleosome or on genome-wide nucleosome occupancy distributions obtained from chromatin assembled in vitro or in vivo ( 2 , 3 ). However, none of these approaches can predict correctly both strong solitary nucleosome affinities and genome-wide nucleosome distributions. Here, we present a simple three-parameter model that captures DNA sequence preferences for nucleosome positioning for both in vitro and in vivo assembly and gives insight into the mechanisms that control nucleosome formation. Eukaryotic DNA is organized in nucleosomes consisting of 147-bp DNA segments wrapped around octamers of histone proteins ( 1 ). The positions of nucleosomes with respect to the underlying sequence, as well as the differences in affinity of the histone octamers for specific DNA sequences, can influence DNA accessibility and nucleosomal fiber folding and hence affect enzymatic processes such as transcription and nucleosome repositioning. For a structural understanding of such processes, understanding the effect of DNA sequence on nucleosome positioning is therefore imperative. Here, we introduce a model that predicts nucleosome positions and the free energy of their formation (see Fig. P1 ). Fig. P1. The binding preference of histone octamers and DNA is regulated by the underlying DNA ( A ). To capture this binding preference, we introduce a model based on periodic probability functions P for the distribution of the dinucleotides TA, TT, AA, and GC, which is defined by a sinus with an amplitude B , periodicity p , and width N . ( B ). Taking into account the 146 dinucleotides around a histone octamer, we obtain a free-energy landscape for nucleosome binding in which a histone octamer binds to the DNA sequence ( C ). Finally, depending on the presence of other histone octamers during nucleosome formation, as indicated by the chemical potential μ, the nucleosome density n is obtained along the DNA molecule by solving Percus’ equation resulting in a close agreement between experimental and modeled data. As shown for example by gene MSN4 on yeast chromosome 11, the presence of the -1 and +1 nucleosome around the TSS are respectively over- and underestimated by the model indicative of a role for other factors than DNA sequence ( D ).
Type of Medium:
Online Resource
ISSN:
0027-8424
,
1091-6490
DOI:
10.1073/pnas.1205659109
Language:
English
Publisher:
Proceedings of the National Academy of Sciences
Publication Date:
2012
detail.hit.zdb_id:
209104-5
detail.hit.zdb_id:
1461794-8
SSG:
11
SSG:
12
Bookmarklink
