A Deep-Learning Pipeline for TSS Coverage Imputation From Shallow Cell-Free DNA Sequencing

Cell-free DNA (cfDNA) serves as a footprint of the nucleosome occupancy status of transcription start sites (TSSs), and has been subject to wide development for use in noninvasive health monitoring and disease detection. However, the requirement for high sequencing depth limits its clinical use. Here, we introduce a deep-learning pipeline designed for TSS coverage profiles generated from shallow cfDNA sequencing called the Autoencoder of cfDNA TSS (AECT) coverage profile. AECT outperformed existing single-cell sequencing imputation algorithms in terms of improvements to TSS coverage accuracy and the capture of latent biological features that distinguish sex or tumor status. We built classifiers for the detection of breast and rectal cancer using AECT-imputed shallow sequencing data, and their performance was close to that achieved by high-depth sequencing, suggesting that AECT could provide a broadly applicable noninvasive screening approach with high accuracy and at a moderate cost.

NucPosDB: a database of nucleosome positioning in vivo and nucleosomics of cell-free DNA

Nucleosome positioning is involved in many gene regulatory processes happening in the cell and it may change as cells differentiate or respond to the changing microenvironment in a healthy or diseased organism. One important implication of nucleosome positioning in clinical epigenetics is its use in the "nucleosomics" analysis of cell-free DNA (cfDNA) for the purpose of patient diagnostics in liquid biopsies. The rationale for this is that the apoptotic nucleases that digest chromatin of the dying cells mostly cut DNA between nucleosomes. Thus, the short pieces of DNA in body fluids reflect the positions of nucleosomes in the cells of origin. Here we report a systematic nucleosomics database - NucPosDB, curating published nucleosome positioning datasets in vivo as well as datasets of sequenced cell-free DNA (cfDNA) that reflect nucleosome positioning in situ in the cells of origin. Users can select subsets of the database by a number of criteria and then obtain raw or processed data. NucPosDB also reports the originally determined regions with stable nucleosome occupancy across several individuals with a given condition. An additional section provides a catalogue of computational tools for the analysis of nucleosome positioning or cfDNA experiments and theoretical algorithms for the prediction of nucleosome positioning from DNA sequence. We provide an overview of the field, describe the structure of the database in this context and demonstrate data variability using examples of different medical conditions. NucPosDB is useful both for analysis of fundamental gene regulation processes and training computational models for patient diagnostics based on cfDNA. The database currently curates ~400 publications on nucleosome positioning in cell lines and in situ as well as cfDNA from >10,000 patients and healthy volunteers. For open-access cfDNA datasets as well as key MNase-seq datasets in human cells, NucPosDB allows downloading processed mapped data in addition to the stable-nucleosome regions. NucPosDB is available at https://generegulation.org/nucposdb/.

Haplotype-aware single-cell multiomics uncovers functional effects of somatic structural variation

Somatic structural variants (SVs) are widespread in cancer genomes, however, their impact on tumorigenesis and intra-tumour heterogeneity is incompletely understood, since methods to functionally characterize the broad spectrum of SVs arising in cancerous single-cells are lacking. We present a computational method, scNOVA, that couples SV discovery with nucleosome occupancy analysis by haplotype-resolved single-cell sequencing, to systematically uncover SV effects on cis-regulatory elements and gene activity. Application to leukemias and cell lines uncovered SV outcomes at several loci, including dysregulated cancer-related pathways and mono-allelic oncogene expression near SV breakpoints. At the intra-patient level, we identified different yet overlapping subclonal SVs that converge on aberrant Wnt signaling. We also deconvoluted the effects of catastrophic chromosomal rearrangements resulting in oncogenic transcription factor dysregulation. scNOVA directly links SVs to their functional consequences, opening the door for single-cell multiomics of SVs in heterogeneous cell populations.

Different SP1 binding dynamics at individual genomic loci in human cells

Using a tamoxifen-inducible time-course ChIP-sequencing (ChIP-seq) approach, we show that the ubiquitous transcription factor SP1 has different binding dynamics at its target sites in the human genome. SP1 very rapidly reaches maximal binding levels at some sites, but binding kinetics at other sites is biphasic, with rapid half-maximal binding followed by a considerably slower increase to maximal binding. While ∼70% of SP1 binding sites are located at promoter regions, loci with slow SP1 binding kinetics are enriched in enhancer and Polycomb-repressed regions. Unexpectedly, SP1 sites with fast binding kinetics tend to have higher quality and more copies of the SP1 sequence motif. Different cobinding factors associate near SP1 binding sites depending on their binding kinetics and on their location at promoters or enhancers. For example, NFY and FOS are preferentially associated near promoter-bound SP1 sites with fast binding kinetics, whereas DNA motifs of ETS and homeodomain proteins are preferentially observed at sites with slow binding kinetics. At promoters but not enhancers, proteins involved in sumoylation and PML bodies associate more strongly with slow SP1 binding sites than with the fast binding sites. The speed of SP1 binding is not associated with nucleosome occupancy, and it is not necessarily coupled to higher transcriptional activity. These results with SP1 are in contrast to those of human TBP, indicating that there is no common mechanism affecting transcription factor binding kinetics. The biphasic kinetics at some SP1 target sites suggest the existence of distinct chromatin states at these loci in different cells within the overall population.

Proximal-end bias from in-vitro reconstituted nucleosomes and the result on downstream data analysis

The most basic level of eukaryotic gene regulation is the presence or absence of nucleosomes on DNA regulatory elements. In an effort to elucidate in vivo nucleosome patterns, in vitro studies are frequently used. In vitro, short DNA fragments are more favorable for nucleosome formation, increasing the likelihood of nucleosome occupancy. This may in part result from the fact that nucleosomes prefer to form on the terminal ends of linear DNA. This phenomenon has the potential to bias in vitro reconstituted nucleosomes and skew results. If the ends of DNA fragments are known, the reads falling close to the ends are typically discarded. In this study we confirm the phenomenon of end bias of in vitro nucleosomes. We describe a method in which nearly identical libraries, with different known ends, are used to recover nucleosomes which form towards the terminal ends of fragmented DNA. Finally, we illustrate that although nucleosomes prefer to form on DNA ends, it does not appear to skew results or the interpretation thereof.

High frequency of intron retention and clustered H3K4me3-marked nucleosomes in short first introns of human long non-coding RNAs

Background It is established that protein-coding exons are preferentially localized in nucleosomes. To examine whether the same is true for non-coding exons, we analysed nucleosome occupancy in and adjacent to internal exons in genes encoding long non-coding RNAs (lncRNAs) in human CD4+ T cells and K562 cells. Results We confirmed that internal exons in lncRNAs are preferentially associated with nucleosomes, but also observed an elevated signal from H3K4me3-marked nucleosomes in the sequences upstream of these exons. Examination of 200 genomic lncRNA loci chosen at random across all chromosomes showed that high-density regions of H3K4me3-marked nucleosomes, which we term ‘slabs’, are associated with genomic regions exhibiting intron retention. These retained introns occur in over 50% of lncRNAs examined and are mostly first introns with an average length of just 354 bp, compared to the average length of all human introns of 6355 and 7987 bp in mRNAs and lncRNAs, respectively. Removal of short introns from the dataset abrogated the high upstream H3K4me3 signal, confirming that the association of slabs and short lncRNA introns with intron retention holds genome-wide. The high upstream H3K4me3 signal is also associated with alternatively spliced exons, known to be prominent in lncRNAs. This phenomenon was not observed with mRNAs. Conclusions There is widespread intron retention and clustered H3K4me3-marked nucleosomes in short first introns of human long non-coding RNAs, which raises intriguing questions about the relationship of IR to lncRNA function and chromatin organization.

Nucleosome positioning on large tandem DNA repeats of the '601' sequence engineered in Saccharomyces cerevisiae

The so-called 601 DNA sequence is often used to constrain the position of nucleosomes on a DNA molecule in vitro. Although the ability of the 147 base pair sequence to precisely position a nucleosome in vitro is well documented, in vivo application of this property has been explored only in a few studies and yielded contradictory conclusions. Our goal in the present study was to test the ability of the 601 sequence to dictate nucleosome positioning in Saccharomyces cerevisiae in the context of a long tandem repeat array inserted in a yeast chromosome. We engineered such arrays with three different repeat size, namely 167, 197 and 237 base pairs. Although our arrays are able to position nucleosomes in vitro as expected, analysis of nucleosome occupancy on these arrays in vivo revealed that nucleosomes are not preferentially positioned as expected on the 601-core sequence along the repeats and that the measured nucleosome repeat length does not correspond to the one expected by design. Altogether our results demonstrate that the rules defining nucleosome positions on this DNA sequence in vitro are not valid in vivo, at least in this chromosomal context, questioning the relevance of using the 601 sequence in vivo to achieve precise nucleosome positioning on designer synthetic DNA sequences.

Mechanisms of Binding Specificity among bHLH Transcription Factors

The transcriptome of every cell is orchestrated by the complex network of interaction between transcription factors (TFs) and their binding sites on DNA. Disruption of this network can result in many forms of organism malfunction but also can be the substrate of positive natural selection. However, understanding the specific determinants of each of these individual TF-DNA interactions is a challenging task as it requires integrating the multiple possible mechanisms by which a given TF ends up interacting with a specific genomic region. These mechanisms include DNA motif preferences, which can be determined by nucleotide sequence but also by DNA’s shape; post-translational modifications of the TF, such as phosphorylation; and dimerization partners and co-factors, which can mediate multiple forms of direct or indirect cooperative binding. Binding can also be affected by epigenetic modifications of putative target regions, including DNA methylation and nucleosome occupancy. In this review, we describe how all these mechanisms have a role and crosstalk in one specific family of TFs, the basic helix-loop-helix (bHLH), with a very conserved DNA binding domain and a similar DNA preferred motif, the E-box. Here, we compile and discuss a rich catalog of strategies used by bHLH to acquire TF-specific genome-wide landscapes of binding sites.

RSC-mediated nucleosome positioning and GRFs form barriers in promoters to limit divergent noncoding transcription

The directionality of gene promoters - the ratio of protein-coding over divergent noncoding transcription - is highly variable and regulated. How promoter directionality is controlled remains poorly understood. We show that the chromatin remodelling complex RSC and general regulatory factors (GRFs) dictate promoter directionality by attenuating divergent transcription. Depletion of RSC increased divergent noncoding transcription and decreased protein-coding transcription at promoters with strong directionality. Consistent with RSCs role in regulating chromatin, RSC depletion impacts nucleosome occupancy upstream of the nucleosome depleted region where divergent transcription initiates, suggesting that nucleosome positioning at the 5 prime border of gene promoters physically blocks the recruitment of the transcription machinery and inhibits initiation of divergent transcription. Highly directional promoters were also enriched for the binding of GRFs such as Reb1 and Abf1. Furthermore, ectopic targeting of divergent transcription initiation sites with GRFs or the dCas9 protein can suppress divergent transcription. Our data suggest that RSC-mediated nucleosome positioning and GRFs play a pervasive role in repressing divergent transcription. We propose that any DNA binding factor, when stably associated with cryptic transcription start sites, can form a barrier for repressing divergent transcription. Our study provides an explanation as to why certain promoters are more directional than others.

GC content but not nucleosome positioning directly contributes to intron-splicing efficiency in Paramecium

Eukaryotic genes are interrupted by introns that must be accurately spliced from mRNA precursors. With an average length of 25 nt, the >90,000 introns of Paramecium tetraurelia stand among the shortest introns reported in eukaryotes. The mechanisms specifying the correct recognition of these tiny introns remain poorly understood. Splicing can occur co-transcriptionally and it has been proposed that chromatin structure might influence splice site recognition. To investigate the roles of nucleosome positioning in intron recognition, we determined the nucleosome occupancy along the P. tetraurelia genome. We showed that P. tetraurelia displays a regular nucleosome array with a nucleosome repeat length of ~151 bp, amongst the smallest periodicities reported. Our analysis revealed that introns are frequently associated with inter-nucleosomal DNA, pointing to an evolutionary constraint to locate introns at the AT-rich nucleosome edge sequences. Using accurate splicing efficiency data from cells depleted for the nonsense-mediated decay effectors, we showed that introns located at the edge of nucleosomes display higher splicing efficiency than those at the centre. However, multiple regression analysis indicated that the GC content, rather than nucleosome positioning, directly contributes to intron splicing efficiency. Our data reveal a complex link between GC content, nucleosome positioning and intron evolution in Paramecium.