bpRNA-align: improved RNA secondary structure global alignment for comparing and clustering RNA structures.

Ribonucleic acid (RNA) is a polymeric molecule that is fundamental to biological processes, with structure being more highly conserved than primary sequence and often key to its function. Advances in RNA structure characterization have resulted in an increase in the number of accurate secondary structures. The task of uncovering common RNA structural motifs with a collective function through structural comparison, providing a level of similarity, remains challenging and could be used to improve RNA secondary structure databases and discover new RNA families. In this work, we present a novel secondary structure alignment method, bpRNA-align. bpRNA-align is a customized global structural alignment method, utilizing an inverted (gap extend costs more than gap open) and context-specific affine gap penalty along with a structural, feature-specific substitution matrix to provide similarity scores. We evaluate our similarity scores in comparison to other methods, using affinity propagation clustering, applied to a benchmarking data set of known structure types. bpRNA-align shows improvement in clustering performance over a broad range of structure types.

Improving deep models of protein-coding potential with a Fourier-transform architecture and machine translation task.

Ribosomes are information-processing macromolecular machines that integrate complex sequence patterns in messenger RNA (mRNA) transcripts to synthesize proteins. Studies of the sequence features that distinguish mRNAs from long noncoding RNAs (lncRNAs) may yield insight into the information that directs and regulates translation. Computational methods for calculating protein-coding potential are important for distinguishing mRNAs from lncRNAs during genome annotation, but most machine learning methods for this task rely on previously known rules to define features. Sequence-to-sequence (seq2seq) models, particularly ones using transformer networks, have proven capable of learning complex grammatical relationships between words to perform natural language translation. Seeking to leverage these advancements in the biological domain, we present a seq2seq formulation for predicting protein-coding potential with deep neural networks and demonstrate that simultaneously learning translation from RNA to protein improves classification performance relative to a classification-only training objective. Inspired by classical signal processing methods for gene discovery and Fourier-based image-processing neural networks, we introduce LocalFilterNet (LFNet). LFNet is a network architecture with an inductive bias for modeling the three-nucleotide periodicity apparent in coding sequences. We incorporate LFNet within an encoder-decoder framework to test whether the translation task improves the classification of transcripts and the interpretation of their sequence features. We use the resulting model to compute nucleotide-resolution importance scores, revealing sequence patterns that could assist the cellular machinery in distinguishing mRNAs and lncRNAs. Finally, we develop a novel approach for estimating mutation effects from Integrated Gradients, a backpropagation-based feature attribution, and characterize the difficulty of efficient approximations in this setting.

Discovery and Visualization of Age-Dependent Patterns in the Diurnal Transcriptome of Drosophila.

Many critical life processes are regulated by input from 24-hour external light/dark cycles, such as metabolism, cellular homeostasis, and detoxification. The circadian clock, which helps coordinate the response to these diurnal light/dark cycles, remains rhythmic across lifespan; however, rhythmic transcript expression is altered during normal aging. To better understand how aging impacts diurnal expression, we present an improved Fourier-based method for detecting and visualizing rhythmicity that is based on the relative power of the 24-hour period compared to other periods (RP24). We apply RP24 to transcript-level expression profiles from the heads of young (5-day) and old (55-day) Drosophila melanogaster, and reveal novel age-dependent rhythmicity changes that may be masked at the gene level. We show that core clock transcripts phase advance during aging, while most rhythmic transcripts phase delay. Transcripts rhythmic only in young flies tend to peak before lights on, while transcripts only rhythmic in old peak after lights on. We show that several pathways, including glutathione metabolism, gain or lose coordinated rhythmic expression with age, providing insight into possible mechanisms of age-onset neurodegeneration. Remarkably, we find that many pathways show very robust coordinated rhythms across lifespan, highlighting their putative roles in promoting neural health. We investigate statistically enriched transcription factor binding site motifs that may be involved in these rhythmicity changes.

Pan-tissue transcriptome analysis of long noncoding RNAs in the American beaver Castor canadensis.

BACKGROUND: Long noncoding RNAs (lncRNAs) have roles in gene regulation, epigenetics, and molecular scaffolding and it is hypothesized that they underlie some mammalian evolutionary adaptations. However, for many mammalian species, the absence of a genome assembly precludes the comprehensive identification of lncRNAs. The genome of the American beaver (Castor canadensis) has recently been sequenced, setting the stage for the systematic identification of beaver lncRNAs and the characterization of their expression in various tissues. The objective of this study was to discover and profile polyadenylated lncRNAs in the beaver using high-throughput short-read sequencing of RNA from sixteen beaver tissues and to annotate the resulting lncRNAs based on their potential for orthology with known lncRNAs in other species. RESULTS: Using de novo transcriptome assembly, we found 9528 potential lncRNA contigs and 187 high-confidence lncRNA contigs. Of the high-confidence lncRNA contigs, 147 have no known orthologs (and thus are putative novel lncRNAs) and 40 have mammalian orthologs. The novel lncRNAs mapped to the Oregon State University (OSU) reference beaver genome with greater than 90% sequence identity. While the novel lncRNAs were on average shorter than their annotated counterparts, they were similar to the annotated lncRNAs in terms of the relationships between contig length and minimum free energy (MFE) and between coverage and contig length. We identified beaver orthologs of known lncRNAs such as XIST, MEG3, TINCR, and NIPBL-DT. We profiled the expression of the 187 high-confidence lncRNAs across 16 beaver tissues (whole blood, brain, lung, liver, heart, stomach, intestine, skeletal muscle, kidney, spleen, ovary, placenta, castor gland, tail, toe-webbing, and tongue) and identified both tissue-specific and ubiquitous lncRNAs. CONCLUSIONS: To our knowledge this is the first report of systematic identification of lncRNAs and their expression atlas in beaver. LncRNAs-both novel and those with known orthologs-are expressed in each of the beaver tissues that we analyzed. For some beaver lncRNAs with known orthologs, the tissue-specific expression patterns were phylogenetically conserved. The lncRNA sequence data files and raw sequence files are available via the web supplement and the NCBI Sequence Read Archive, respectively.

LinearFold: linear-time approximate RNA folding by 5'-to-3' dynamic programming and beam search.

MOTIVATION: Predicting the secondary structure of an ribonucleic acid (RNA) sequence is useful in many applications. Existing algorithms [based on dynamic programming] suffer from a major limitation: their runtimes scale cubically with the RNA length, and this slowness limits their use in genome-wide applications. RESULTS: We present a novel alternative O(n3)-time dynamic programming algorithm for RNA folding that is amenable to heuristics that make it run in O(n) time and O(n) space, while producing a high-quality approximation to the optimal solution. Inspired by incremental parsing for context-free grammars in computational linguistics, our alternative dynamic programming algorithm scans the sequence in a left-to-right (5'-to-3') direction rather than in a bottom-up fashion, which allows us to employ the effective beam pruning heuristic. Our work, though inexact, is the first RNA folding algorithm to achieve linear runtime (and linear space) without imposing constraints on the output structure. Surprisingly, our approximate search results in even higher overall accuracy on a diverse database of sequences with known structures. More interestingly, it leads to significantly more accurate predictions on the longest sequence families in that database (16S and 23S Ribosomal RNAs), as well as improved accuracies for long-range base pairs (500+ nucleotides apart), both of which are well known to be challenging for the current models. AVAILABILITY AND IMPLEMENTATION: Our source code is available at https://github.com/LinearFold/LinearFold, and our webserver is at http://linearfold.org (sequence limit: 100 000nt). SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Sulforaphane Bioavailability and Chemopreventive Activity in Men Presenting for Biopsy of the Prostate Gland: A Randomized Controlled Trial.

Previous studies suggest compounds such as sulforaphane (SFN) derived from cruciferous vegetables may prevent prostate cancer development and progression. This study evaluated the effect of broccoli sprout extract (BSE) supplementation on blood histone deacetylase (HDAC) activity, prostate RNA gene expression, and tissue biomarkers (histone H3 lysine 18 acetylation (H3K18ac), HDAC3, HDAC6, Ki67, and p21). A total of 98 men scheduled for prostate biopsy were allocated into either BSE (200 µmol daily) or a placebo in our double-blind, randomized controlled trial. We used nonparametric tests to evaluate the differences of blood HDAC activity and prostate tissue immunohistochemistry biomarkers between treatment groups. Further, we performed RNA-Seq analysis on the prostate biopsies and identified 40 differentially expressed genes correlated with BSE treatment, including downregulation of two genes previously implicated in prostate cancer development, AMACR and ARLNC1. Although urine and plasma SFN isothiocyanates and individual SFN metabolites were statistically higher in the treatment group, our results did not show a significant difference in HDAC activity or prostate tissue biomarkers. This study indicates BSE supplementation correlates with changes in gene expression but not with several other prostate cancer biomarkers. More research is required to fully understand the chemopreventive effects of BSE supplementation on prostate cancer.

A deep recurrent neural network discovers complex biological rules to decipher RNA protein-coding potential.

The current deluge of newly identified RNA transcripts presents a singular opportunity for improved assessment of coding potential, a cornerstone of genome annotation, and for machine-driven discovery of biological knowledge. While traditional, feature-based methods for RNA classification are limited by current scientific knowledge, deep learning methods can independently discover complex biological rules in the data de novo. We trained a gated recurrent neural network (RNN) on human messenger RNA (mRNA) and long noncoding RNA (lncRNA) sequences. Our model, mRNA RNN (mRNN), surpasses state-of-the-art methods at predicting protein-coding potential despite being trained with less data and with no prior concept of what features define mRNAs. To understand what mRNN learned, we probed the network and uncovered several context-sensitive codons highly predictive of coding potential. Our results suggest that gated RNNs can learn complex and long-range patterns in full-length human transcripts, making them ideal for performing a wide range of difficult classification tasks and, most importantly, for harvesting new biological insights from the rising flood of sequencing data.

The polycomb group mutant esc leads to augmented levels of paused Pol II in the Drosophila embryo.

Many developmental control genes contain paused RNA polymerase II (Pol II) and are thereby "poised" for rapid and synchronous activation in the early Drosophila embryo. Evidence is presented that Polycomb group (PcG) repressors can influence paused Pol II. ChIP-Seq and GRO-Seq assays were used to determine the genome-wide distributions of Pol II, H3K27me3, and H3K4me3 in extra sex combs (esc) mutant embryos. ESC is a key component of the Polycomb repressive complex 2 (PRC2), which mediates H3K27me3 modification. Enhanced Pol II occupancy is observed for thousands of genes in esc mutant embryos, including genes not directly regulated by PRC2. Thus, it would appear that silent genes lacking promoter-associated paused Pol II in wild-type embryos are converted into "poised" genes with paused Pol II in esc mutants. We suggest that this conversion of silent genes into poised genes might render differentiated cell types susceptible to switches in identity in PcG mutants.

In Vivo Characterization of an AHR-Dependent Long Noncoding RNA Required for Proper Sox9b Expression.

Xenobiotic activation of the aryl hydrocarbon receptor (AHR) by 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) prevents the proper formation of craniofacial cartilage and the heart in developing zebrafish. Downstream molecular targets responsible for AHR-dependent adverse effects remain largely unknown; however, in zebrafish sox9b has been identified as one of the most-reduced transcripts in several target organs and is hypothesized to have a causal role in TCDD-induced toxicity. The reduction of sox9b expression in TCDD-exposed zebrafish embryos has been shown to contribute to heart and jaw malformation phenotypes. The mechanisms by which AHR2 (functional ortholog of mammalian AHR) activation leads to reduced sox9b expression levels and subsequent target organ toxicity are unknown. We have identified a novel long noncoding RNA (slincR) that is upregulated by strong AHR ligands and is located adjacent to the sox9b gene. We hypothesize that slincR is regulated by AHR2 and transcriptionally represses sox9b. The slincR transcript functions as an RNA macromolecule, and slincR expression is AHR2 dependent. Antisense knockdown of slincR results in an increase in sox9b expression during both normal development and AHR2 activation, which suggests relief in repression. During development, slincR was expressed in tissues with sox9 essential functions, including the jaw/snout region, otic vesicle, eye, and brain. Reducing the levels of slincR resulted in altered neurologic and/or locomotor behavioral responses. Our results place slincR as an intermediate between AHR2 activation and the reduction of sox9b mRNA in the AHR2 signaling pathway.

Improving deep models of protein-coding potential with a Fourier-transform architecture and machine translation task.

Ribosomes are information-processing macromolecular machines that integrate complex sequence patterns in messenger RNA (mRNA) transcripts to synthesize proteins. Studies of the sequence features that distinguish mRNAs from long noncoding RNAs (lncRNAs) may yield insight into the information that directs and regulates translation. Computational methods for calculating protein-coding potential are important for distinguishing mRNAs from lncRNAs during genome annotation, but most machine learning methods for this task rely on previously known rules to define features. Sequence-to-sequence (seq2seq) models, particularly ones using transformer networks, have proven capable of learning complex grammatical relationships between words to perform natural language translation. Seeking to leverage these advancements in the biological domain, we present a seq2seq formulation for predicting protein-coding potential with deep neural networks and demonstrate that simultaneously learning translation from RNA to protein improves classification performance relative to a classification-only training objective. Inspired by classical signal processing methods for gene discovery and Fourier-based image-processing neural networks, we introduce LocalFilterNet (LFNet). LFNet is a network architecture with an inductive bias for modeling the three-nucleotide periodicity apparent in coding sequences. We incorporate LFNet within an encoder-decoder framework to test whether the translation task improves the classification of transcripts and the interpretation of their sequence features. We use the resulting model to compute nucleotide-resolution importance scores, revealing sequence patterns that could assist the cellular machinery in distinguishing mRNAs and lncRNAs. Finally, we develop a novel approach for estimating mutation effects from Integrated Gradients, a backpropagation-based feature attribution, and characterize the difficulty of efficient approximations in this setting.

An improved assembly of the "Cascade" hop (Humulus lupulus) genome uncovers signatures of molecular evolution and refines time of divergence estimates for the Cannabaceae family.

We present a chromosome-level assembly of the Cascade hop (Humulus lupulus L. var. lupulus) genome. The hop genome is large (2.8 Gb) and complex, and early attempts at assembly were fragmented. Recent advances have made assembly of the hop genome more tractable, transforming the extent of investigation that can occur. The chromosome-level assembly of Cascade was developed by scaffolding the previously reported Cascade assembly generated with PacBio long-read sequencing and polishing with Illumina short-read DNA sequencing. We developed gene models and repeat annotations and used a controlled bi-parental mapping population to identify significant sex-associated markers. We assessed molecular evolution in gene sequences, gene family expansion and contraction, and time of divergence from Cannabis sativa and other closely related plant species using Bayesian inference. We identified the putative sex chromosome in the female genome based on significant sex-associated markers from the bi-parental mapping population. While the estimate of repeat content (~64%) is similar to the estimate for the hemp genome, syntenic blocks in hop contain a greater percentage of LTRs. Hop is enriched for disease resistance-associated genes in syntenic gene blocks and expanded gene families. The Cascade chromosome-level assembly will inform cultivation strategies and serve to deepen our understanding of the hop genomic landscape, benefiting hop researchers and the Cannabaceae genomics community.

RNA structure and multiple weak interactions balance the interplay between RNA binding and phase separation of SARS-CoV-2 nucleocapsid.

The nucleocapsid (N) protein of SARS-CoV-2 binds viral RNA, condensing it inside the virion, and phase separating with RNA to form liquid-liquid condensates. There is little consensus on what differentiates sequence-independent N-RNA interactions in the virion or in liquid droplets from those with specific genomic RNA (gRNA) motifs necessary for viral function inside infected cells. To identify the RNA structures and the N domains responsible for specific interactions and phase separation, we use the first 1,000 nt of viral RNA and short RNA segments designed as models for single-stranded and paired RNA. Binding affinities estimated from fluorescence anisotropy of these RNAs to the two-folded domains of N (the NTD and CTD) and comparison to full-length N demonstrate that the NTD binds preferentially to single-stranded RNA, and while it is the primary RNA-binding site, it is not essential to phase separation. Nuclear magnetic resonance spectroscopy identifies two RNA-binding sites on the NTD: a previously characterized site and an additional although weaker RNA-binding face that becomes prominent when binding to the primary site is weak, such as with dsRNA or a binding-impaired mutant. Phase separation assays of nucleocapsid domains with double-stranded and single-stranded RNA structures support a model where multiple weak interactions, such as with the CTD or the NTD's secondary face promote phase separation, while strong, specific interactions do not. These studies indicate that both strong and multivalent weak N-RNA interactions underlie the multifunctional abilities of N.

Age-dependent effects of blue light exposure on lifespan, neurodegeneration, and mitochondria physiology in Drosophila melanogaster.

Blue light is a predominant component of light emitting devices (LEDs), which are increasingly present in our environment. There is already accumulating evidence that blue light exposure causes damage to retinal cells in vitro and in vivo; however, much less is known about potential effects of blue light on non-retinal cells. That blue light may be detrimental at the organismal level independent from retinal effect was recently shown by findings that it reduces lifespan in worms and also in flies with genetically ablated retinas. Here, we investigated the effects of blue light exposure across the fly lifespan and found that susceptibility to blue light stress is strongly age-dependent. The blue light of the same intensity and duration reduced survival and increased neurodegeneration more significantly in old flies than in young flies. These differences appear to be caused, at least in part, by impairments of mitochondrial respiratory function. We report that blue light significantly reduces the activity of Complex II in the electron transport system and decrease the biochemical activity of succinate dehydrogenase in both young and old flies. In addition, complex I and complex IV activities are reduced by age, as are ATP levels. We therefore propose that older flies are more sensitive to blue light because the light-induced mitochondrial damage potentiates the age-related impairments in energy metabolism that occurs even in darkness. Taken together, our results show that damaging effects of blue light at the organismal level are strongly age dependent and are associated with reduced activity of specific components of energy producing pathways in mitochondria.

Chronic blue light leads to accelerated aging in Drosophila by impairing energy metabolism and neurotransmitter levels.

Blue light (BL) is becoming increasingly prevalent in artificial illumination, raising concerns about its potential health hazard to humans. In fact, there is evidence suggesting that acute BL exposure may lead to oxidative stress and death of retinal cells specialized for photoreception. On the other hand, recent studies in Drosophila melanogaster demonstrated that chronic BL exposure across lifespan leads to accelerated aging manifested in reduced lifespan and brain neurodegeneration even in flies with genetically ablated eyes, suggesting that BL can damage cells and tissues not specialized for light perception. At the physiological level, BL exposure impairs mitochondria function in flies, but the metabolic underpinnings of these effects have not been studied. Here, we investigated effects of chronic BL on metabolic pathways in heads of eyes absent (eya 2 ) mutant flies in order to focus on extra-retinal tissues. We compared metabolomic profiles in flies kept for 10 or 14 days in constant BL or constant darkness, using LC-MS and GC-MS. Data analysis revealed significant alterations in the levels of several metabolites suggesting that critical cellular pathways are impacted in BL-exposed flies. In particular, dramatic metabolic rearrangements are observed in heads of flies kept in BL for 14 days, including highly elevated levels of succinate but reduced levels of pyruvate and citrate, suggesting impairments in energy production. These flies also show onset of neurodegeneration and our analysis detected significantly reduced levels of several neurotransmitters including glutamate and Gamma-aminobutyric acid (GABA), suggesting that BL disrupts brain homeostasis. Taken together, these data provide novel insights into the mechanisms by which BL interferes with vital metabolic pathways that are conserved between fly and human cells.

How the Dorsal gradient works: insights from postgenome technologies.

Gradients of extracellular signaling molecules and transcription factors are used in a variety of developmental processes, including the patterning of the Drosophila embryo, the establishment of diverse neuronal cell types in the vertebrate neural tube, and the anterior-posterior patterning of vertebrate limbs. Here, we discuss how a gradient of the maternal transcription factor Dorsal produces complex patterns of gene expression across the dorsal-ventral (DV) axis of the early Drosophila embryo. The identification of 60-70 Dorsal target genes, along with the characterization of approximately 35 associated regulatory DNAs, suggests that there are at least six different regulatory codes driving diverse DV expression profiles.

Promoter elements associated with RNA Pol II stalling in the Drosophila embryo.

RNA Polymerase II (Pol II) is bound to the promoter regions of many or most developmental control genes before their activation during Drosophila embryogenesis. It has been suggested that Pol II stalling is used to produce dynamic and rapid responses of developmental patterning genes to transient cues such as extracellular signaling molecules. Here, we present a combined computational and experimental analysis of stalled promoters to determine how they come to bind Pol II in the early Drosophila embryo. At least one-fourth of the stalled promoters contain a shared sequence motif, the "pause button" (PB): KCGRWCG. The PB motif is sometimes located in the position of the DPE, and over one-fifth of the stalled promoters contain the following arrangement of core elements: GAGA, Inr, PB, and/or DPE. This arrangement was used to identify additional stalled promoters in the Drosophila genome, and permanganate footprint assays were used to confirm that the segmentation gene engrailed contains paused Pol II as seen for heat-shock genes. We discuss different models for Pol II binding and gene activation in the early embryo.

Predicting Drosha and Dicer Cleavage Sites with DeepMirCut.

MicroRNAs are a class of small RNAs involved in post-transcriptional gene silencing with roles in disease and development. Many computational tools have been developed to identify novel microRNAs. However, there have been no attempts to predict cleavage sites for Drosha from primary sequence, or to identify cleavage sites using deep neural networks. Here, we present DeepMirCut, a recurrent neural network-based software that predicts both Dicer and Drosha cleavage sites. We built a microRNA primary sequence database including flanking genomic sequences for 34,713 microRNA annotations. We compare models trained on sequence data, sequence and secondary structure data, as well as input data with annotated structures. Our best model is able to predict cuts within closer average proximity than results reported for other methods. We show that a guanine nucleotide before and a uracil nucleotide after Dicer cleavage sites on the 3' arm of the microRNA precursor had a positive effect on predictions while the opposite order (U before, G after) had a negative effect. Our analysis was also able to predict several positions where bulges had either positive or negative effects on the score. We expect that our approach and the data we have curated will enable several future studies.

A draft phased assembly of the diploid Cascade hop (Humulus lupulus) genome.

Hop (Humulus lupulus L. var Lupulus) is a diploid, dioecious plant with a history of cultivation spanning more than one thousand years. Hop cones are valued for their use in brewing and contain compounds of therapeutic interest including xanthohumol. Efforts to determine how biochemical pathways responsible for desirable traits are regulated have been challenged by the large (2.8 Gb), repetitive, and heterozygous genome of hop. We present a draft haplotype-phased assembly of the Cascade cultivar genome. Our draft assembly and annotation of the Cascade genome is the most extensive representation of the hop genome to date. PacBio long-read sequences from hop were assembled with FALCON and partially phased with FALCON-Unzip. Comparative analysis of haplotype sequences provides insight into selective pressures that have driven evolution in hop. We discovered genes with greater sequence divergence enriched for stress-response, growth, and flowering functions in the draft phased assembly. With improved resolution of long terminal retrotransposons (LTRs) due to long-read sequencing, we found that hop is over 70% repetitive. We identified a homolog of cannabidiolic acid synthase (CBDAS) that is expressed in multiple tissues. The approaches we developed to analyze the draft phased assembly serve to deepen our understanding of the genomic landscape of hop and may have broader applicability to the study of other large, complex genomes.

Lactate dehydrogenase expression modulates longevity and neurodegeneration in Drosophila melanogaster.

Lactate dehydrogenase (LDH) catalyzes the conversion of glycolysis-derived pyruvate to lactate. Lactate has been shown to play key roles in brain energetics and memory formation. However, lactate levels are elevated in aging and Alzheimer's disease patients, and it is not clear whether lactate plays protective or detrimental roles in these contexts. Here we show that Ldh transcript levels are elevated and cycle with diurnal rhythm in the heads of aged flies and this is associated with increased LDH protein, enzyme activity, and lactate concentrations. To understand the biological significance of increased Ldh gene expression, we genetically manipulated Ldh levels in adult neurons or glia. Overexpression of Ldh in both cell types caused a significant reduction in lifespan whereas Ldh down-regulation resulted in lifespan extension. Moreover, pan-neuronal overexpression of Ldh disrupted circadian locomotor activity rhythms and significantly increased brain neurodegeneration. In contrast, reduction of Ldh in neurons delayed age-dependent neurodegeneration. Thus, our unbiased genetic approach identified Ldh and lactate as potential modulators of aging and longevity in flies.

The lncRNA Locus Handsdown Regulates Cardiac Gene Programs and Is Essential for Early Mouse Development.

Precisely controlled gene regulatory networks are required during embryonic development to give rise to various structures, including those of the cardiovascular system. Long non-coding RNA (lncRNA) loci are known to be important regulators of these genetic programs. We have identified a novel and essential lncRNA locus Handsdown (Hdn), active in early heart cells, and show by genetic inactivation that it is essential for murine development. Hdn displays haploinsufficiency for cardiac development as Hdn-heterozygous adult mice exhibit hyperplasia in the right ventricular wall. Transcriptional activity of the Hdn locus, independent of its RNA, suppresses its neighboring gene Hand2. We reveal a switch in a topologically associated domain in differentiation of the cardiac lineage, allowing the Hdn locus to directly interact with regulatory elements of the Hand2 locus.