eRNA co-expression network uncovers TF dependency and convergent cooperativity.

Enhancer RNAs (eRNAs) are non-coding RNAs produced by transcriptional enhancers that are highly correlated with their activity. Using a capped nascent RNA sequencing (PRO-cap) dataset in human lymphoblastoid cell lines across 67 individuals, we identified inter-individual variation in the expression of over 80 thousand transcribed transcriptional regulatory elements (tTREs), in both enhancers and promoters. Co-expression analysis of eRNAs from tTREs across individuals revealed how enhancers are associated with each other and with promoters. Mid- to long-range co-expression showed a distance-dependent decay that was modified by TF occupancy. In particular, we found a class of "bivalent" TFs, including Cohesin, that both facilitate and isolate the interaction between enhancers and/or promoters, depending on their topology. At short distances, we observed strand-specific correlations between nearby eRNAs in both convergent and divergent orientations. Our results support a cooperative model of convergent eRNAs, consistent with eRNAs facilitating adjacent enhancers rather than interfering with each other. Therefore, our approach to infer functional interactions from co-expression analyses provided novel insights into the principles of enhancer interactions as a function of distance, orientation, and binding landscapes of TFs.

Revealing eRNA interactions: TF dependency and convergent cooperativity.

Enhancer RNAs (eRNAs) are non-coding RNAs produced from transcriptional enhancers that are highly correlated with their activities. Using capped nascent RNA sequencing (PRO-cap) dataset in human lymphoblastoid cell lines across individuals, we identified inter-individual variation of expression in over 80 thousand transcribed transcriptional regulatory elements (tTREs), in both enhancers and promoters. Co-expression analysis of eRNAs from tTREs across individuals revealed how enhancers interact with each other and with promoters. Mid-to-long range interactions showed distance-dependent decay, which was modified by TF occupancy. In particular, we found a class of 'bivalent' TFs, including Cohesin, which both facilitates and insulates the interaction between enhancers and/or promoters depending on the topology. In short ranges, we observed strand specific interactions between nearby eRNAs in both convergent or divergent orientations. Our finding supports a cooperative convergent eRNA model, which is compatible with eRNA remodeling neighboring enhancers rather than interfering with each other. Therefore, our approach to infer functional interactions from co-expression analyses provided novel insights into the principles of enhancer interactions depending on the distance, orientation, and the binding landscapes of TFs.

RANKL-responsive epigenetic mechanism reprograms macrophages into bone-resorbing osteoclasts.

Monocyte/macrophage lineage cells are highly plastic and can differentiate into various cells under different environmental stimuli. Bone-resorbing osteoclasts are derived from the monocyte/macrophage lineage in response to receptor activator of NF-κB ligand (RANKL). However, the epigenetic signature contributing to the fate commitment of monocyte/macrophage lineage differentiation into human osteoclasts is largely unknown. In this study, we identified RANKL-responsive human osteoclast-specific superenhancers (SEs) and SE-associated enhancer RNAs (SE-eRNAs) by integrating data obtained from ChIP-seq, ATAC-seq, nuclear RNA-seq and PRO-seq analyses. RANKL induced the formation of 200 SEs, which are large clusters of enhancers, while suppressing 148 SEs in macrophages. RANKL-responsive SEs were strongly correlated with genes in the osteoclastogenic program and were selectively increased in human osteoclasts but marginally presented in osteoblasts, CD4+ T cells, and CD34+ cells. In addition to the major transcription factors identified in osteoclasts, we found that BATF binding motifs were highly enriched in RANKL-responsive SEs. The depletion of BATF1/3 inhibited RANKL-induced osteoclast differentiation. Furthermore, we found increased chromatin accessibility in SE regions, where RNA polymerase II was significantly recruited to induce the extragenic transcription of SE-eRNAs, in human osteoclasts. Knocking down SE-eRNAs in the vicinity of the NFATc1 gene diminished the expression of NFATc1, a major regulator of osteoclasts, and osteoclast differentiation. Inhibiting BET proteins suppressed the formation of some RANKL-responsive SEs and NFATc1-associated SEs, and the expression of SE-eRNA:NFATc1. Moreover, SE-eRNA:NFATc1 was highly expressed in the synovial macrophages of rheumatoid arthritis patients exhibiting high-osteoclastogenic potential. Our genome-wide analysis revealed RANKL-inducible SEs and SE-eRNAs as osteoclast-specific signatures, which may contribute to the development of osteoclast-specific therapeutic interventions.

Dynamic and widespread control of poly(A) tail length during macrophage activation.

The poly(A) tail enhances translation and transcript stability, and tail length is under dynamic control during cell state transitions. Tail regulation plays essential roles in translational timing and fertilization in early development, but poly(A) tail dynamics have not been fully explored in post-embryonic systems. Here, we examined the landscape and impact of tail length control during macrophage activation. Upon activation, more than 1500 mRNAs, including proinflammatory genes, underwent distinctive changes in tail lengths. Increases in tail length correlated with mRNA levels regardless of transcriptional activity, and many mRNAs that underwent tail extension encode proteins necessary for immune function and post-transcriptional regulation. Strikingly, we found that ZFP36, whose protein product destabilizes target transcripts, undergoes tail extension. Our analyses indicate that many mRNAs undergoing tail lengthening are, in turn, degraded by elevated levels of ZFP36, constituting a post-transcriptional feedback loop that ensures transient regulation of transcripts integral to macrophage activation. Taken together, this study establishes the complexity, relevance, and widespread nature of poly(A) tail dynamics, and the resulting post-transcriptional regulation during macrophage activation.

Genome-Wide Identification of Polyadenylation Dynamics with TED-Seq.

Polyadenylation and deadenylation of mRNA are major RNA modifications associated with nucleus-to-cytoplasm translocation, mRNA stability, translation efficiency, and mRNA decay pathways. Our current knowledge of polyadenylation and deadenylation has been expanded due to recent advances in transcriptome-wide poly(A) tail length assays. Whereas these methods measure poly(A) length by quantifying the adenine (A) base stretch at the 3' end of mRNA, we developed a more cost-efficient technique that does not rely on A-base counting, called tail-end-displacement sequencing (TED-seq). Through sequencing highly size-selected 3' RNA fragments including the poly(A) tail pieces, TED-seq provides accurate measure of transcriptome-wide poly(A)-tail lengths in high resolution, economically suitable for larger scale analysis under various biologically transitional contexts.

Population-scale study of eRNA transcription reveals bipartite functional enhancer architecture.

Enhancer RNAs (eRNA) are unstable non-coding RNAs, transcribed bidirectionally from active regulatory sequences, whose expression levels correlate with enhancer activity. We use capped-nascent-RNA sequencing to efficiently capture bidirectional transcription initiation across several human lymphoblastoid cell lines (Yoruba population) and detect ~75,000 eRNA transcription sites with high sensitivity and specificity. The use of nascent-RNA sequencing sidesteps the confounding effect of eRNA instability. We identify quantitative trait loci (QTLs) associated with the level and directionality of eRNA expression. High-resolution analyses of these two types of QTLs reveal distinct positions of enrichment at the central transcription factor (TF) binding regions and at the flanking eRNA initiation regions, both of which are associated with mRNA expression QTLs. These two regions-the central TF-binding footprint and the eRNA initiation cores-define a bipartite architecture of enhancers, inform enhancer function, and can be used as an indicator of the significance of non-coding regulatory variants.

Single-Cell Transcriptome Analysis of Colon Cancer Cell Response to 5-Fluorouracil-Induced DNA Damage.

DNA damage often induces heterogeneous cell-fate responses, such as cell-cycle arrest and apoptosis. Through single-cell RNA sequencing (scRNA-seq), we characterize the transcriptome response of cultured colon cancer cell lines to 5-fluorouracil (5FU)-induced DNA damage. After 5FU treatment, a single population of colon cancer cells adopts three distinct transcriptome phenotypes, which correspond to diversified cell-fate responses: apoptosis, cell-cycle checkpoint, and stress resistance. Although some genes are regulated uniformly across all groups of cells, many genes showed group-specific expression patterns mediating DNA damage responses specific to the corresponding cell fate. Some of these observations are reproduced at the protein level by flow cytometry and are replicated in cells treated with other 5FU-unrelated genotoxic drugs, camptothecin and etoposide. This work provides a resource for understanding heterogeneous DNA damage responses involving fractional killing and chemoresistance, which are among the major challenges in current cancer chemotherapy.

m6A enhances the phase separation potential of mRNA.

N6-methyladenosine (m6A) is the most prevalent modified nucleotide in mRNA1,2, with around 25% of mRNAs containing at least one m6A. Methylation of mRNA to form m6A is required for diverse cellular and physiological processes3. Although the presence of m6A in an mRNA can affect its fate in different ways, it is unclear how m6A directs this process and why the effects of m6A can vary in different cellular contexts. Here we show that the cytosolic m6A-binding proteins-YTHDF1, YTHDF2 and YTHDF3-undergo liquid-liquid phase separation in vitro and in cells. This phase separation is markedly enhanced by mRNAs that contain multiple, but not single, m6A residues. Polymethylated mRNAs act as a multivalent scaffold for the binding of YTHDF proteins, juxtaposing their low-complexity domains and thereby leading to phase separation. The resulting mRNA-YTHDF complexes then partition into different endogenous phase-separated compartments, such as P-bodies, stress granules or neuronal RNA granules. m6A-mRNA is subject to compartment-specific regulation, including a reduction in the stability and translation of mRNA. These studies reveal that the number and distribution of m6A sites in cellular mRNAs can regulate and influence the composition of the phase-separated transcriptome, and suggest that the cellular properties of m6A-modified mRNAs are governed by liquid-liquid phase separation principles.

Chromatin run-on and sequencing maps the transcriptional regulatory landscape of glioblastoma multiforme.

The human genome encodes a variety of poorly understood RNA species that remain challenging to identify using existing genomic tools. We developed chromatin run-on and sequencing (ChRO-seq) to map the location of RNA polymerase for almost any input sample, including samples with degraded RNA that are intractable to RNA sequencing. We used ChRO-seq to map nascent transcription in primary human glioblastoma (GBM) brain tumors. Enhancers identified in primary GBMs resemble open chromatin in the normal human brain. Rare enhancers that are activated in malignant tissue drive regulatory programs similar to the developing nervous system. We identified enhancers that regulate groups of genes that are characteristic of each known GBM subtype and transcription factors that drive them. Finally we discovered a core group of transcription factors that control the expression of genes associated with clinical outcomes. This study characterizes the transcriptional landscape of GBM and introduces ChRO-seq as a method to map regulatory programs that contribute to complex diseases.

TED-Seq Identifies the Dynamics of Poly(A) Length during ER Stress.

Post-transcriptional RNA processing is a core mechanism of gene expression control in cell stress response. The poly(A) tail influences mRNA translation and stability, but it is unclear whether there are global roles of poly(A)-tail lengths in cell stress. To address this, we developed tail-end displacement sequencing (TED-seq) for an efficient transcriptome-wide profiling of poly(A) lengths and applied it to endoplasmic reticulum (ER) stress in human cells. ER stress induced increases in the poly(A) lengths of certain mRNAs, including known ER stress regulators, XBP1, DDIT3, and HSPA5. Importantly, the mRNAs with increased poly(A) lengths are both translationally de-repressed and stabilized. Furthermore, mRNAs in stress-induced RNA granules have shorter poly(A) tails than in the cytoplasm, supporting the view that RNA processing is compartmentalized. In conclusion, TED-seq reveals that poly(A) length is dynamically regulated upon ER stress, with potential consequences for both translation and mRNA turnover.

Aspm knockout ferret reveals an evolutionary mechanism governing cerebral cortical size.

The human cerebral cortex is distinguished by its large size and abundant gyrification, or folding. However, the evolutionary mechanisms that drive cortical size and structure are unknown. Although genes that are essential for cortical developmental expansion have been identified from the genetics of human primary microcephaly (a disorder associated with reduced brain size and intellectual disability) 1 , studies of these genes in mice, which have a smooth cortex that is one thousand times smaller than the cortex of humans, have provided limited insight. Mutations in abnormal spindle-like microcephaly-associated (ASPM), the most common recessive microcephaly gene, reduce cortical volume by at least 50% in humans2-4, but have little effect on the brains of mice5-9; this probably reflects evolutionarily divergent functions of ASPM10,11. Here we used genome editing to create a germline knockout of Aspm in the ferret (Mustela putorius furo), a species with a larger, gyrified cortex and greater neural progenitor cell diversity12-14 than mice, and closer protein sequence homology to the human ASPM protein. Aspm knockout ferrets exhibit severe microcephaly (25-40% decreases in brain weight), reflecting reduced cortical surface area without significant change in cortical thickness, as has been found in human patients3,4, suggesting that loss of 'cortical units' has occurred. The cortex of fetal Aspm knockout ferrets displays a very large premature displacement of ventricular radial glial cells to the outer subventricular zone, where many resemble outer radial glia, a subtype of neural progenitor cells that are essentially absent in mice and have been implicated in cerebral cortical expansion in primates12-16. These data suggest an evolutionary mechanism by which ASPM regulates cortical expansion by controlling the affinity of ventricular radial glial cells for the ventricular surface, thus modulating the ratio of ventricular radial glial cells, the most undifferentiated cell type, to outer radial glia, a more differentiated progenitor.

Systematic Characterization of Stress-Induced RNA Granulation.

Upon stress, cytoplasmic mRNA is sequestered to insoluble ribonucleoprotein (RNP) granules, such as the stress granule (SG). Partially due to the belief that translationally suppressed mRNAs are recruited to SGs in bulk, stress-induced dynamic redistribution of mRNA has not been thoroughly characterized. Here, we report that endoplasmic reticulum (ER) stress targets only a small subset of translationally suppressed mRNAs into the insoluble RNP granule fraction (RG). This subset, characterized by extended length and adenylate-uridylate (AU)-rich motifs, is highly enriched with genes critical for cell survival and proliferation. This pattern of RG targeting was conserved for two other stress types, heat shock and arsenite toxicity, which induce distinct responses in the total cytoplasmic transcriptome. Nevertheless, stress-specific RG-targeting motifs, such as guanylate-cytidylate (GC)-rich motifs in heat shock, were also identified. Previously underappreciated, transcriptome profiling in the RG may contribute to understanding human diseases associated with RNP dysfunction, such as cancer and neurodegeneration.

RNA-DNA sequence differences in Saccharomyces cerevisiae.

Alterations of RNA sequences and structures, such as those from editing and alternative splicing, result in two or more RNA transcripts from a DNA template. It was thought that in yeast, RNA editing only occurs in tRNAs. Here, we found that Saccharomyces cerevisiae have all 12 types of RNA-DNA sequence differences (RDDs) in the mRNA. We showed these sequence differences are propagated to proteins, as we identified peptides encoded by the RNA sequences in addition to those by the DNA sequences at RDD sites. RDDs are significantly enriched at regions with R-loops. A screen of yeast mutants showed that RDD formation is affected by mutations in genes regulating R-loops. Loss-of-function mutations in ribonuclease H, senataxin, and topoisomerase I that resolve RNA-DNA hybrids lead to increases in RDD frequency. Our results demonstrate that RDD is a conserved process that diversifies transcriptomes and proteomes and provide a mechanistic link between R-loops and RDDs.

High-Resolution Mapping of RNA Polymerases Identifies Mechanisms of Sensitivity and Resistance to BET Inhibitors in t(8;21) AML.

Bromodomain and extra-terminal domain (BET) family inhibitors offer an approach to treating hematological malignancies. We used precision nuclear run-on transcription sequencing (PRO-seq) to create high-resolution maps of active RNA polymerases across the genome in t(8;21) acute myeloid leukemia (AML), as these polymerases are exceptionally sensitive to BET inhibitors. PRO-seq identified over 1,400 genes showing impaired release of promoter-proximal paused RNA polymerases, including the stem cell factor receptor tyrosine kinase KIT that is mutated in t(8;21) AML. PRO-seq also identified an enhancer 3' to KIT. Chromosome conformation capture confirmed contacts between this enhancer and the KIT promoter, while CRISPRi-mediated repression of this enhancer impaired cell growth. PRO-seq also identified microRNAs, including MIR29C and MIR29B2, that target the anti-apoptotic factor MCL1 and were repressed by BET inhibitors. MCL1 protein was upregulated, and inhibition of BET proteins sensitized t(8:21)-containing cells to MCL1 inhibition, suggesting a potential mechanism of resistance to BET-inhibitor-induced cell death.

Base-pair-resolution genome-wide mapping of active RNA polymerases using precision nuclear run-on (PRO-seq).

We provide a protocol for precision nuclear run-on sequencing (PRO-seq) and its variant, PRO-cap, which map the location of active RNA polymerases (PRO-seq) or transcription start sites (TSSs) (PRO-cap) genome-wide at high resolution. The density of RNA polymerases at a particular genomic locus directly reflects the level of nascent transcription at that region. Nuclei are isolated from cells and, under nuclear run-on conditions, transcriptionally engaged RNA polymerases incorporate one or, at most, a few biotin-labeled nucleotide triphosphates (biotin-NTPs) into the 3' end of nascent RNA. The biotin-labeled nascent RNA is used to prepare sequencing libraries, which are sequenced from the 3' end to provide high-resolution positional information for the RNA polymerases. PRO-seq provides much higher sensitivity than ChIP-seq, and it generates a much larger fraction of usable sequence reads than ChIP-seq or NET-seq (native elongating transcript sequencing). Similarly to NET-seq, PRO-seq maps the RNA polymerase at up to base-pair resolution with strand specificity, but unlike NET-seq it does not require immunoprecipitation. With the protocol provided here, PRO-seq (or PRO-cap) libraries for high-throughput sequencing can be generated in 4-5 working days. The method has been applied to human, mouse, Drosophila melanogaster and Caenorhabditis elegans cells and, with slight modifications, to yeast.

Chromatin Architecture of the Pitx2 Locus Requires CTCF- and Pitx2-Dependent Asymmetry that Mirrors Embryonic Gut Laterality.

Expression of Pitx2 on the left side of the embryo patterns left-right (LR) organs including the dorsal mesentery (DM), whose asymmetric cell behavior directs gut looping. Despite the importance of organ laterality, chromatin-level regulation of Pitx2 remains undefined. Here, we show that genes immediately neighboring Pitx2 in chicken and mouse, including a long noncoding RNA (Pitx2 locus-asymmetric regulated RNA or Playrr), are expressed on the right side and repressed by Pitx2. CRISPR/Cas9 genome editing of Playrr, 3D fluorescent in situ hybridization (FISH), and variations of chromatin conformation capture (3C) demonstrate that mutual antagonism between Pitx2 and Playrr is coordinated by asymmetric chromatin interactions dependent on Pitx2 and CTCF. We demonstrate that transcriptional and morphological asymmetries driving gut looping are mirrored by chromatin architectural asymmetries at the Pitx2 locus. We propose a model whereby Pitx2 auto-regulation directs chromatin topology to coordinate LR transcription of this locus essential for LR organogenesis.

Genome-wide dynamics of Pol II elongation and its interplay with promoter proximal pausing, chromatin, and exons.

Production of mRNA depends critically on the rate of RNA polymerase II (Pol II) elongation. To dissect Pol II dynamics in mouse ES cells, we inhibited Pol II transcription at either initiation or promoter-proximal pause escape with Triptolide or Flavopiridol, and tracked Pol II kinetically using GRO-seq. Both inhibitors block transcription of more than 95% of genes, showing that pause escape, like initiation, is a ubiquitous and crucial step within the transcription cycle. Moreover, paused Pol II is relatively stable, as evidenced from half-life measurements at ∼3200 genes. Finally, tracking the progression of Pol II after drug treatment establishes Pol II elongation rates at over 1000 genes. Notably, Pol II accelerates dramatically while transcribing through genes, but slows at exons. Furthermore, intergenic variance in elongation rates is substantial, and is influenced by a positive effect of H3K79me2 and negative effects of exon density and CG content within genes.DOI: http://dx.doi.org/10.7554/eLife.02407.001.

The Hmr and Lhr hybrid incompatibility genes suppress a broad range of heterochromatic repeats.

Hybrid incompatibilities (HIs) cause reproductive isolation between species and thus contribute to speciation. Several HI genes encode adaptively evolving proteins that localize to or interact with heterochromatin, suggesting that HIs may result from co-evolution with rapidly evolving heterochromatic DNA. Little is known, however, about the intraspecific function of these HI genes, the specific sequences they interact with, or the evolutionary forces that drive their divergence. The genes Hmr and Lhr genetically interact to cause hybrid lethality between Drosophila melanogaster and D. simulans, yet mutations in both genes are viable. Here, we report that Hmr and Lhr encode proteins that form a heterochromatic complex with Heterochromatin Protein 1 (HP1a). Using RNA-Seq analyses we discovered that Hmr and Lhr are required to repress transcripts from satellite DNAs and many families of transposable elements (TEs). By comparing Hmr and Lhr function between D. melanogaster and D. simulans we identify several satellite DNAs and TEs that are differentially regulated between the species. Hmr and Lhr mutations also cause massive overexpression of telomeric TEs and significant telomere lengthening. Hmr and Lhr therefore regulate three types of heterochromatic sequences that are responsible for the significant differences in genome size and structure between D. melanogaster and D. simulans and have high potential to cause genetic conflicts with host fitness. We further find that many TEs are overexpressed in hybrids but that those specifically mis-expressed in lethal hybrids do not closely correlate with Hmr function. Our results therefore argue that adaptive divergence of heterochromatin proteins in response to repetitive DNAs is an important underlying force driving the evolution of hybrid incompatibility genes, but that hybrid lethality likely results from novel epistatic genetic interactions that are distinct to the hybrid background.

RNA-DNA differences are generated in human cells within seconds after RNA exits polymerase II.

RNA sequences are expected to be identical to their corresponding DNA sequences. Here, we found all 12 types of RNA-DNA sequence differences (RDDs) in nascent RNA. Our results show that RDDs begin to occur in RNA chains ~55 nt from the RNA polymerase II (Pol II) active site. These RDDs occur so soon after transcription that they are incompatible with known deaminase-mediated RNA-editing mechanisms. Moreover, the 55 nt delay in appearance indicates that they do not arise during RNA synthesis by Pol II or as a direct consequence of modified base incorporation. Preliminary data suggest that RDD and R-loop formations may be coupled. These findings identify sequence substitution as an early step in cotranscriptional RNA processing.

Defining NELF-E RNA binding in HIV-1 and promoter-proximal pause regions.

The four-subunit Negative Elongation Factor (NELF) is a major regulator of RNA Polymerase II (Pol II) pausing. The subunit NELF-E contains a conserved RNA Recognition Motif (RRM) and is proposed to facilitate Poll II pausing through its association with nascent transcribed RNA. However, conflicting ideas have emerged for the function of its RNA binding activity. Here, we use in vitro selection strategies and quantitative biochemistry to identify and characterize the consensus NELF-E binding element (NBE) that is required for sequence specific RNA recognition (NBE: CUGAGGA(U) for Drosophila). An NBE-like element is present within the loop region of the transactivation-response element (TAR) of HIV-1 RNA, a known regulatory target of human NELF-E. The NBE is required for high affinity binding, as opposed to the lower stem of TAR, as previously claimed. We also identify a non-conserved region within the RRM that contributes to the RNA recognition of Drosophila NELF-E. To understand the broader functional relevance of NBEs, we analyzed promoter-proximal regions genome-wide in Drosophila and show that the NBE is enriched +20 to +30 nucleotides downstream of the transcription start site. Consistent with the role of NELF in pausing, we observe a significant increase in NBEs among paused genes compared to non-paused genes. In addition to these observations, SELEX with nuclear run-on RNA enrich for NBE-like sequences. Together, these results describe the RNA binding behavior of NELF-E and supports a biological role for NELF-E in promoter-proximal pausing of both HIV-1 and cellular genes.