Regular article
Sequence and position-dependence of the equilibrium accessibility of nucleosomal DNA target sites1

https://doi.org/10.1006/jmbi.2000.3531Get rights and content

Abstract

We have previously shown that nucleosomes are conformationally dynamic: DNA sequences that in the time-average are buried inside nucleosomes are nevertheless transiently accessible, even to large proteins (or any other macromolecule). We refer to this dynamic behavior as “site exposure”. Here we show that: (i) the equilibrium constants describing this dynamic site exposure decrease progressively from either end of the nucleosomal DNA in toward the middle; and (ii) these position-dependent equilibrium constants are strongly dependent on the nucleosomal DNA sequence. The progressive decrease in equilibrium constant with distance inside the nucleosome supports the hypothesis that access to sites internal to a nucleosome is provided by progressive (transient) release of DNA from the octamer surface, starting from one end of the nucleosomal DNA. The dependence on genomic DNA sequence implies that a specific genomic DNA sequence could be a major determinant of target site occupancies achieved by regulatory proteins in vivo, by either governing the time-averaged accessibility for a given nucleosome position, or biasing the time-averaged positioning (of mobile nucleosomes), which in turn is a major determinant of site accessibility.

Introduction

The mechanisms whereby proteins gain access to DNA target sites in chromatin are not known. A problem that arises even at the lowest level of chromatin organization, the nucleosome, is that most of the DNA is inaccessible, yet essential processes such as transcription, replication, recombination, and repair must occur. Studies in vivo point to a need for ATP-powered molecular machines to “open chromatin”. Strikingly, however, studies from many laboratories reveal that nucleosomal DNA is accessible to apparently all DNA-binding proteins, even in the absence of exogenous factors, despite the fact that the crystallographic structure of the nucleosome shows that the binding sites would be occluded (Luger et al., 1997).

In earlier work Polach and Widom 1995, Polach and Widom 1996 we proposed a “site exposure model” that explains this behavior of nucleosomes in vitro and successfully predicts the results of widely disparate experiments, such as the equilibrium binding of a protein to a nucleosomal DNA target site, the kinetics of digestion of nucleosomal DNA at internal sites by restriction enzymes or at the ends of nucleosomal DNA by exonucleases, and the surprising cooperative binding of pairs of proteins to target sites contained within the same nucleosome. Let N represent a nucleosome in its native (most prevalent) conformation. The site exposure model supposes the existence of a transient state, S, of the nucleosome, at equilibrium with N, in which the nucleosome has transiently uncoiled some of its DNA, such that a DNA target site that was previously inaccessible is now freely accessible to a binding protein, R, or nuclease, E. These can bind to make a complex, RS or ES, the latter of which can go on to catalyze to yield products E and P. We suppose that sufficient DNA is uncoiled so as to make the subsequent steps on the site-exposed nucleosome equivalent to those on naked DNA. Thus:Nk21k12S+Rk32k23RSS+Rk32k23RS for equilibrium binding to a nucleosomal target or naked DNA, respectively, and:Nk21k12S+Ek32k23ESk34E+P and:S+Ek32k23ESk34E+P for nuclease digestion of nucleosomal or naked DNA, respectively.

Real nucleosomes in vitro behave as though such uncoiling processes are occurring constantly yet transiently, in a rapid pre-equilibrium. As a consequence of this behavior, even sites which are buried within the middle of the nucleosome are constantly but transiently accessible, with equilibrium constant for accessibility (site exposure) Keqconf=k12/k21. The apparent equilibrium binding affinity for proteins to nucleosomal target sites, and the observed rate constants for digestion of nucleosomal DNA, are reduced from their values on naked DNA by a factor equal to Keqconf. For further discussion of the site exposure model, see Polach and Widom 1998, Widom 1998.

In the site exposure model, equilibrium constants for site exposure are related to sums of free energies for all of the histone-DNA contacts that need to be broken in order to expose the site. This implies that the DNA sequence could be a major determinant of site exposure equilibrium constants. Sequences having especially high affinity for the histone octamer (or equivalently, strong nucleosome positioning sequences (Lowary & Widom, 1997)) should exhibit significantly reduced equilibrium site accessibility. DNA sequence motifs implicated in nucleosome positioning are found with statistically significant enhanced probability in eukaryotic genomes Widom 1996, Lowary and Widom 1998 and are known to actually confer increased nucleosome-positioning power on the DNA segments that contain these motifs Lowary and Widom 1998, Thastrom et al 1999. If such high-affinity sequences could strongly suppress site exposure equilibria, this could contribute importantly to gene regulation in two novel ways (see Discussion). Thus, one aim here is to test whether the choice of specific DNA sequence actually can have any significant effect on site exposure equilibrium constants.

A second aim of the study is to examine the position-dependence of equilibrium constants for site exposure. The site exposure model supposes that access to sites internal to a nucleosome is provided by progressive (transient) release of DNA from the octamer surface starting from an end of the nucleosomal DNA. Our earlier study analyzed sites within one side of the nucleosomal DNA, and obtained results consistent with this hypothesis: Keqconf decreased progressively with distance from the end of the nucleosomal DNA in toward the middle. However, the DNA constructs analyzed in that study had several limitations. They lacked appropriate sites in the other half, hence they did not allow measurements to be made along the full nucleosomal DNA length. Moreover, each DNA construct used contained only a small number of sites; results from several related, but different, DNA constructs were combined to yield the final picture. Hence, a second goal here is to allow measurement of Keqconf over the full length of the nucleosomal DNA using a single nucleosome population, to facilitate a clear analysis of the nature of position-dependences to Keqconf.

Section snippets

Purification and characterization of reconstituted nucleosomes

A new DNA sequence (601.2, Figure 1(a)) was developed for these studies, based on the following considerations. (i) It is derived from a non-natural sequence that was selected for especially high-affinity binding to histone octamer in nucleosome reconstitution (Lowary & Widom, 1998). Such a sequence is also expected to be a strong nucleosome-positioning sequence (useful for creating homogeneous samples) and to exhibit dramatically any DNA-sequence-dependent effects on equilibrium constants for

Discussion

Two aspects of the results are striking. First, the high affinity sequence exhibits greatly reduced Keqconf at all sites throughout the nucleosome. Compared to the sequences studied earlier, the high affinity sequence suppresses Keqconf by ten- to 100-fold throughout the nucleosome. Second, the overall pattern of Keqconfversus nucleosomal location is roughly symmetric about the mapped location of the nucleosomal center (dyad axis). Values of Keqconf decrease progressively from either end in

Preparation of DNA and histories

DNA sequence 601.2 was derived from sequence 601 (Lowary & Widom, 1998) using PCR to introduce specific base changes. Three oligonucleotides were used, in two successive steps of PCR synthesis (nucleotide changes are capitalized): R601, tacatgcacaggatgtatatatct; L601.1, gccctgCagaaGcTTggtCccgGggccgctcaattggtcgtagcaagctctGg-ATccgcttGaTcgAacgtacgcgctgtccccc; and L601.2, ggaccctatacgcggGcgccctgCagaaGcTTggtC. A first step of PCR used the primer pair L601.1 and R601 with the original 282 bp-long

Acknowledgements

We thank P.T. Lowary and K.J. Polach for valuable assistance and discussions. This work was supported by a grant from the NIH (to J.W.) and by an NIH Cell and Molecular Basis of Disease Traineeship (to J.A.). We acknowledge with gratitude the use of instruments in the Keck Biophysics Facility, which was established with a grant from the W.M. Keck Foundation.

Cited by (0)

1

Edited by T. Richmond

View full text