DNA sequence organization in chromatosomes.

Abstract

The chromatosome is a structural unit of chromatin which contains a histone octamer and one linker histone molecule (H5, H1a or H1b in chicken erythrocytes) bound to 168 bp of DNA (= core particle DNA extended by 22 bp). We have cloned and sequenced 280 DNA fragments of 163 to 173 bp in length isolated from chicken erythrocyte chromatosomes. We have analysed both this set and a subset of 171 clones whose lengths varied between 166 and 170 bp. The periodic modulation of the frequency of occurrence of trinucleotide sequences is neither as regular nor as pronounced for core particle DNA, even for the trinucleotide ApApA/TpTpT. Nevertheless for this trinucleotide the congruence of the preferential and avoided locations between the two sets is remarkable. We conclude that the rotational positioning of the DNA bound to the histone octamer is essentially the same in chromatosomes and nucleosome core particles. This conclusion is confirmed by Fourier analysis, which shows that within the central 145 bp the average rotational periods of about 10-20 bp and phases are very similar to those in core DNA. However the amplitude of the modulations is less, indicating that the binding of the linker histone can overcome to a certain extent the sequence-dependent bending preferences of DNA. Although the rotational signals are largely conserved the sequence organization of the chromatosomal DNA differs in significant ways from that of core particle DNA. The major difference is the preferential occurrence of short G+C-rich instead of A+T-rich sequences at the midpoint of the binding site. The second apparent difference is the exclusion of short oligo(dA).(dT) tracts from positions about 40 to 43 bp on either side of the midpoint in chromatosomal, but not in core particle DNA. Finally, we show that particular base-steps are preferentially enriched close to the termini of chromatosomal DNA and propose that these sequences may constitute one of the two DNA binding sites for the bivalent globular domain of histones H5 or H1. The implications for the positioning of GH5 relative to the path of the DNA superhelix are discussed.