Intrinsic structural variability of DNA allows multiple genomic encoding for nucleosomes: Comment on “Cracking the chromatin code: Precise rule of nucleosome positioning” by E.N. Trifonov

Abstract

The review by Trifonov [1] gives a fascinating account of the discovery of immense compaction of DNA in eukaryotic chromosomes and the consequent three-decade-long search for a chromatin code, to which the author himself contributed in no small measure. It culminates in the derivation of a sequence motif by the author that possibly represents the code. However, the story is rather one-sided, based as it is primarily on the author’s own results, with no mention of the large number of experimental studies ([2], and several references therein) which indicate that sequence plays little or no role in the formation of nucleosome structure. It leaves no room for the possibility that maybe, just maybe, there is no chromatin code. In this commentary, we would like to play the devil’s advocate and focus on the ambiguities surrounding the existence of a chromatin code. The theoretical postdiction by Trifonov implicitly assumes that dinucleotide steps take up unique structures. However, analysis of high-resolution DNA crystal structures [3] and molecular dynamics studies [4] have shown that several dinucleotide steps not only show large variation in their structures within the same distribution, but also display multimodal distributions. Further, the structure of RR/YY and YR/YR steps is dependent on their immediate neighbors [4]. Large variation in dinucleotide level structure is also seen in an analysis of nucleosome crystal structures [5]. In fact, it has been shown that “DNA overall flexibility increases considerably upon particle formation” [6]. Thus the absence of unique dinucleotide structures is a serious impediment to sequencebased structure prediction algorithms. An added complication is the presence of a large number of kinks in the nucleosome crystal structures [5]. Molecular dynamics studies have shown that kinks, similar to the ones observed in nucleosome structure, can spontaneously form in long, ∼ 90-mer oligonucleotides, and these kinks are stiff [7]. A linear elastic model does not apply to such stiff regions. Can one then really argue about which steps would (energetically) favor kinking into the major or minor grooves? For example, in free and other protein-bound DNA, the GG/CC step bends into the major groove, and almost never into the minor groove. However, in the nucleosome structure, several kinks into minor groove are observed at