Correlation between dinucleotide periodicities and nucleosome positioning on mouse satellite DNA

The implications of satellite DNA instability on cellular function and evolution

Abundant tandemly repeated satellite DNA is present in most eukaryotic genomes. Previous limitations including a pervasive view that it was uninteresting junk DNA, combined with challenges in studying it, are starting to dissolve - and recent studies have found important functions for satellite DNAs. The observed rapid evolution and implied instability of satellite DNA now has important significance for their functions and maintenance within the genome. In this review, we discuss the processes that lead to satellite DNA copy number instability, and the importance of mechanisms to manage the potential negative effects of instability. Satellite DNA is vulnerable to challenges during replication and repair, since it forms difficult-to-process secondary structures and its homology within tandem arrays can result in various types of recombination. Satellite DNA instability may be managed by DNA or chromatin-binding proteins ensuring proper nuclear localization and repair, or by proteins that process aberrant structures that satellite DNAs tend to form. We also discuss the pattern of satellite DNA mutations from recent mutation accumulation (MA) studies that have tracked changes in satellite DNA for up to 1000 generations with minimal selection. Finally, we highlight examples of satellite evolution from studies that have characterized satellites across millions of years of Drosophila fruit fly evolution, and discuss possible ways that selection might act on the satellite DNA composition.

Evolutionary Dynamics of Abundant 7-bp Satellites in the Genome of Drosophila virilis

The factors that drive the rapid changes in abundance of tandem arrays of highly repetitive sequences, known as satellite DNA, are not well understood. Drosophila virilis has one of the highest relative amounts of simple satellites of any organism that has been studied, with an estimated >40% of its genome composed of a few related 7-bp satellites. Here, we use D. virilis as a model to understand technical biases affecting satellite sequencing and the evolutionary processes that drive satellite composition. By analyzing sequencing data from Illumina, PacBio, and Nanopore platforms, we identify platform-specific biases and suggest best practices for accurate characterization of satellites by sequencing. We use comparative genomics and cytogenetics to demonstrate that the highly abundant AAACTAC satellite family arose from a related satellite in the branch leading to the virilis phylad 4.5-11 Ma before exploding in abundance in some species of the clade. The most abundant satellite is conserved in sequence and location in the pericentromeric region but has diverged widely in abundance among species, whereas the satellites nearest the centromere are rapidly turning over in sequence composition. By analyzing multiple strains of D. virilis, we saw that the abundances of two centromere-proximal satellites are anticorrelated along a geographical gradient, which we suggest could be caused by ongoing conflicts at the centromere. In conclusion, we illuminate several key attributes of satellite evolutionary dynamics that we hypothesize to be driven by processes including selection, meiotic drive, and constraints on satellite sequence and abundance.

CORENup: a combination of convolutional and recurrent deep neural networks for nucleosome positioning identification

BACKGROUND

Nucleosomes wrap the DNA into the nucleus of the Eukaryote cell and regulate its transcription phase. Several studies indicate that nucleosomes are determined by the combined effects of several factors, including DNA sequence organization. Interestingly, the identification of nucleosomes on a genomic scale has been successfully performed by computational methods using DNA sequence as input data.

RESULTS

In this work, we propose CORENup, a deep learning model for nucleosome identification. CORENup processes a DNA sequence as input using one-hot representation and combines in a parallel fashion a fully convolutional neural network and a recurrent layer. These two parallel levels are devoted to catching both non periodic and periodic DNA string features. A dense layer is devoted to their combination to give a final classification.

CONCLUSIONS

Results computed on public data sets of different organisms show that CORENup is a state of the art methodology for nucleosome positioning identification based on a Deep Neural Network architecture. The comparisons have been carried out using two groups of datasets, currently adopted by the best performing methods, and CORENup has shown top performance both in terms of classification metrics and elapsed computation time.

Major Determinants of Nucleosome Positioning

The compact structure of the nucleosome limits DNA accessibility and inhibits the binding of most sequence-specific proteins. Nucleosomes are not randomly located on the DNA but positioned with respect to the DNA sequence, suggesting models in which critical binding sites are either exposed in the linker, resulting in activation, or buried inside a nucleosome, resulting in repression. The mechanisms determining nucleosome positioning are therefore of paramount importance for understanding gene regulation and other events that occur in chromatin, such as transcription, replication, and repair. Here, we review our current understanding of the major determinants of nucleosome positioning: DNA sequence, nonhistone DNA-binding proteins, chromatin-remodeling enzymes, and transcription. We outline the major challenges for the future: elucidating the precise mechanisms of chromatin opening and promoter activation, identifying the complexes that occupy promoters, and understanding the multiscale problem of chromatin fiber organization.

Precise genome-wide mapping of single nucleosomes and linkers in vivo

We developed a chemical cleavage method that releases single nucleosome dyad-containing fragments, allowing us to precisely map both single nucleosomes and linkers with high accuracy genome-wide in yeast. Our single nucleosome positioning data reveal that nucleosomes occupy preferred positions that differ by integral multiples of the DNA helical repeat. By comparing nucleosome dyad positioning maps to existing genomic and transcriptomic data, we evaluated the contributions of sequence, transcription, and histones H1 and H2A.Z in defining the chromatin landscape. We present a biophysical model that neglects DNA sequence and shows that steric occlusion suffices to explain the salient features of nucleosome positioning.

Imprinted control regions include composite DNA elements consisting of the ZFP57 binding site overlapping MLL1 morphemes

Mammalian genomes include DNA segments that are imprinted (CpG-methylated) only on one of the two parental chromosomes, leading to parent-of-origin-specific gene expression. The process is regulated by Imprinting Control Regions (ICRs) and germline Differentially Methylated Regions (gDMRs). Previously, ZFP57 was shown to recognize a methylated hexanucleotide in ICRs to maintain allele-specific gene repression. In Bioinformatics analyses, I found that the hexamer occurred frequently in mouse chromosomal DNA, suggesting that beside the ZFP57 binding site (ZFBS), ICRs contained sequence features with unknown characteristics. To identify such features, I examined chromosomal abundance of motifs in which the length of the hexamer was extended by one or several nucleotides. Results led to the discovery of a group of functionally significant composite DNA elements (ZFBS-Morph overlaps) that may play dual roles in the regulation of allele-specific gene expression. Importantly, results of genome-wide evaluations revealed that nearly 90% of the gDMRs included closely-spaced ZFBS-Morph overlaps.

Oligonucleotide sequence motifs as nucleosome positioning signals

To gain a better understanding of the sequence patterns that characterize positioned nucleosomes, we first performed an analysis of the periodicities of the 256 tetranucleotides in a yeast genome-wide library of nucleosomal DNA sequences that was prepared by in vitro reconstitution. The approach entailed the identification and analysis of 24 unique tetranucleotides that were defined by 8 consensus sequences. These consensus sequences were shown to be responsible for most if not all of the tetranucleotide and dinucleotide periodicities displayed by the entire library, demonstrating that the periodicities of dinucleotides that characterize the yeast genome are, in actuality, due primarily to the 8 consensus sequences. A novel combination of experimental and bioinformatic approaches was then used to show that these tetranucleotides are important for preferred formation of nucleosomes at specific sites along DNA in vitro. These results were then compared to tetranucleotide patterns in genome-wide in vivo libraries from yeast and C. elegans in order to assess the contributions of DNA sequence in the control of nucleosome residency in the cell. These comparisons revealed striking similarities in the tetranucleotide occurrence profiles that are likely to be involved in nucleosome positioning in both in vitro and in vivo libraries, suggesting that DNA sequence is an important factor in the control of nucleosome placement in vivo. However, the strengths of the tetranucleotide periodicities were 3-4 fold higher in the in vitro as compared to the in vivo libraries, which implies that DNA sequence plays less of a role in dictating nucleosome positions in vivo. The results of this study have important implications for models of sequence-dependent positioning since they suggest that a defined subset of tetranucleotides is involved in preferred nucleosome occupancy and that these tetranucleotides are the major source of the dinucleotide periodicities that are characteristic of positioned nucleosomes.

Sequence patterns defining the 5' boundary of human genes

In studies of both short and relatively long human genomic DNA, we found a clustering of the consensus site for the transcription factor GCF at the 5' boundary of a subset of human genes. In studies of promoter regions with known transcription initiation site, the cluster of consensus GCF site appeared near the transcription initiation site and in some sequences it extended into the transcribed region defining the leader mRNA. We also found a detectable correlation between the 5' boundary of human genes and recognition motifs for other transcription factors that bind to GC-rich sequences. But in these cases, the correlation was not as general as the correlation observed for the consensus GCF site.

A signal encoded in vertebrate DNA that influences nucleosome positioning and alignment

Evidence is provided that the nucleotide triplet con-sensus non-T(A/T)G (abbreviated to VWG) influences nucleosome positioning and nucleosome alignment into regular arrays. This triplet consensus has been recently found to exhibit a fairly strong 10 bp periodicity in human DNA, implicating it in anisotropic DNA bendability. It is demonstrated that the experimentally determined preferences for nucleosome positioning in native SV40 chromatin can, to a large extent, be pre-dicted simply by counting the occurrences of the period-10 VWG consensus. Nucleosomes tend to form in regions of the SV40 genome that contain high counts of period-10 VWG and/or avoid regions with low counts. In contrast, periodic occurrences of the dinucleotides AA/TT, implicated in the rotational positioning of DNA in nucleosomes, did not correlate with the preferred nucleosome locations in SV40 chromatin. Periodic occurrences of AA did correlate with preferred nucleosome locations in a region of SV40 DNA where VWG occurrences are low. Regular oscillations in period-10 VWG counts with a dinucleosome period were found in vertebrate DNA regions that aligned nucleosomes into regular arrays in vitro in the presence of linker histone. Escherichia coli and plasmid DNA, which fail to align nucleosomes in vitro, lacked these regular VWG oscillations.

A topological approach to nucleosome structure and dynamics: the linking number paradox and other issues

The linking number paradox of DNA in chromatin (two negative crossings around the octamer, associated with a unit linking number reduction), which is 21 years old this year, has come of age. After stirring much debate in the past, the initially hypothetical explanation of the paradox by DNA overtwisting on the nucleosome surface is now presented as a hard fact in recent textbooks. The first part of this article presents a historical perspective of the problem and details the numerous attempts to measure DNA local periodicity, which in one remarkable example sowed the seeds for the discovery of DNA bending. The second part is devoted to the DNA minicircle system, which has been developed in the author's laboratory as an alternative to the local-periodicity-measurement approach. It offers a simple proposal: a unit linking number reduction associated with a single crossing. This conclusion is contrasted with the latest high-resolution crystallographic data of the nucleosome in the third part of the article, and the fourth part examines the available evidence supporting an extension of these results to nucleosomes in chromatin. The last part addresses another basic question pertaining to nucleosome dynamics, the conformational flexibility of the histone tetramer.

Identification and characterization of genomic nucleosome-positioning sequences

Positioned nucleosomes are believed to play important roles in transcriptional regulation and for the organization of chromatin in cell nuclei. Here, we have isolated the DNA segments in the mouse genome that form the most stable nucleosomes yet characterized. In separate molecules we find phased runs of three to four adenine nucleotides, extensive CA repeats, and in a few cases phased TATA tetranucleotides. The latter forms the most stable nucleosome yet characterized. One sequence with CAG repeats was also found. By fluorescence in situ hydridization the selected sequences are shown to be localized at the centromeric regions of mouse metaphase chromosomes.

Periodicity of dinucleotides in nucleosomes derived from simian virus 40 chromatin

It is thought that statistical analysis of dinucleotide periodicities can provide insight into the general features of nucleosome forming sequences. The chromatin of simian virus 40 (SV40) provides a model for a unique DNA sequence that is found in association with histones in vivo. I have therefore analyzed the periodicity of dinucleotides in a collection of cloned nucleosomal DNA fragments prepared from SV40 chromatin isolated under relatively mild conditions, in order to learn about the generality of results obtained from the statistical approach and to examine the SV40 data set in the context of models that have been proposed to explain the molecular basis of nucleosome formation. In one study, I assumed a symmetry in the distribution of dinucleotides with respect to the nucleosome dyad position and considered complementary dinucleotides to be equivalent, i.e. AA = TT and GG = CC. The results showed a periodic signal for GG/CC but not for AA/TT, purine-purine, and pyrimidine-pyrimidine dinucleotides. In a second study, the SV40 nucleosomal DNA fragments were aligned and examined with respect to the late strand of the viral genome to determine the distribution of dinucleotides in one direction. Fourier analysis revealed periodic signals for AA/TT (10.26 bp) and GG/CC (10.0 bp) and indicated that AA dominates the occurrences of AA/TT and GG dominates the occurrences of GG/CC. The results of both studies implied that there might be an asymmetry and a directionality in the distribution of certain dinucleotides in nucleosomes.

Introns of the chicken ovalbumin gene promote nucleosome alignment in vitro

A defined in vitro chromatin assembly system was used to examine the nucleosome alignment induced by histone H5 throughout a 12 kilobase pair chicken genomic DNA fragment containing the ovalbumin gene. In contrast with total fragmented chicken DNA and several anonymous cloned fragments, much of the gene permitted histone H5 to space nucleosomes at physiological intervals in an extended array. Nucleosomes at the 3'-end of the gene and on approximately 4 kilobase pairs of 5'-flanking ovalbumin sequence did not become aligned to appreciable extents. Analysis of cloned 2-3 kilobase pair subfragments suggested that a strong nucleosome alignment signal, specifying a 196 +/- 5 base pair repeat exists in intron E. A second discrete region of the gene, which mapped approximately to intron A, exhibited nucleosome alignment with a spacing periodicity of about 200 base pairs. The ovalbumin cDNA did not permit nucleosome alignment. These findings suggest that some of the introns contain signals that direct nucleosome alignment over the ovalbumin gene in a way conducive to its regulation.

In vitro assembly of a positioned nucleosome near the hypersensitive region in simian virus 40 chromatin

Previous studies have identified a nucleosome near a potential late boundary for the nuclease-hypersensitive region in simian virus 40 chromatin. We have performed in vitro reconstitution analysis to determine whether the underlying DNA sequence encodes for the assembly of this nucleosome and applied hydroxyl radical and DNase I footprinting techniques to examine the structure of the reconstituted nucleosome. Both methods revealed the formation of a precisely positioned nucleosome in vitro, on a fragment spanning the strong in vivo nucleosome location site determined previously in the viral chromatin. The center of the positioned nucleosome maps between nucleotide 384 and 387 on simian virus 40 DNA. The corresponding nucleosome core includes the major-late transcription site (12 base-pairs within the core), the MspI site, and a segment shown previously to adopt a bent structure in the absence of proteins. The hydroxyl radical produces a strikingly well-defined cleavage pattern over the bent DNA incorporated in nucleosomes. The dominant periodicity of DNA in this nucleosome is 10.26 base-pairs per turn. The distribution of the .OH cut sites in the positioned nucleosome provides strong support for models in which the minor grooves of the A/T-rich tracts are oriented toward the histone core while the minor grooves of the G/C-rich sequences are facing outward.