The evolution and mutational robustness of chromatin accessibility in Drosophila.

BACKGROUND: The evolution of genomic regulatory regions plays a critical role in shaping the diversity of life. While this process is primarily sequence-dependent, the enormous complexity of biological systems complicates the understanding of the factors underlying regulation and its evolution. Here, we apply deep neural networks as a tool to investigate the sequence determinants underlying chromatin accessibility in different species and tissues of Drosophila. RESULTS: We train hybrid convolution-attention neural networks to accurately predict ATAC-seq peaks using only local DNA sequences as input. We show that our models generalize well across substantially evolutionarily diverged species of insects, implying that the sequence determinants of accessibility are highly conserved. Using our model to examine species-specific gains in accessibility, we find evidence suggesting that these regions may be ancestrally poised for evolution. Using in silico mutagenesis, we show that accessibility can be accurately predicted from short subsequences in each example. However, in silico knock-out of these sequences does not qualitatively impair classification, implying that accessibility is mutationally robust. Subsequently, we show that accessibility is predicted to be robust to large-scale random mutation even in the absence of selection. Conversely, simulations under strong selection demonstrate that accessibility can be extremely malleable despite its robustness. Finally, we identify motifs predictive of accessibility, recovering both novel and previously known motifs. CONCLUSIONS: These results demonstrate the conservation of the sequence determinants of accessibility and the general robustness of chromatin accessibility, as well as the power of deep neural networks to explore fundamental questions in regulatory genomics and evolution.

The evolution and mutational robustness of chromatin accessibility in Drosophila.

The evolution of regulatory regions in the genome plays a critical role in shaping the diversity of life. While this process is primarily sequence-dependent, the enormous complexity of biological systems has made it difficult to understand the factors underlying regulation and its evolution. Here, we apply deep neural networks as a tool to investigate the sequence determinants underlying chromatin accessibility in different tissues of Drosophila. We train hybrid convolution-attention neural networks to accurately predict ATAC-seq peaks using only local DNA sequences as input. We show that a model trained in one species has nearly identical performance when tested in another species, implying that the sequence determinants of accessibility are highly conserved. Indeed, model performance remains excellent even in distantly-related species. By using our model to examine species-specific gains in chromatin accessibility, we find that their orthologous inaccessible regions in other species have surprisingly similar model outputs, suggesting that these regions may be ancestrally poised for evolution. We then use in silico saturation mutagenesis to reveal evidence of selective constraint acting specifically on inaccessible chromatin regions. We further show that chromatin accessibility can be accurately predicted from short subsequences in each example. However, in silico knock-out of these sequences does not qualitatively impair classification, implying that chromatin accessibility is mutationally robust. Subsequently, we demonstrate that chromatin accessibility is predicted to be robust to large-scale random mutation even in the absence of selection. We also perform in silico evolution experiments under the regime of strong selection and weak mutation (SSWM) and show that chromatin accessibility can be extremely malleable despite its mutational robustness. However, selection acting in different directions in a tissue-specific manner can substantially slow adaptation. Finally, we identify motifs predictive of chromatin accessibility and recover motifs corresponding to known chromatin accessibility activators and repressors. These results demonstrate the conservation of the sequence determinants of accessibility and the general robustness of chromatin accessibility, as well as the power of deep neural networks as tools to answer fundamental questions in regulatory genomics and evolution.

Sex differences in interindividual gene expression variability across human tissues.

Understanding phenotypic sex differences has long been a goal of biology from both a medical and evolutionary perspective. Although much attention has been paid to mean differences in phenotype between the sexes, little is known about sex differences in phenotypic variability. To gain insight into sex differences in interindividual variability at the molecular level, we analyzed RNA-seq data from 43 tissues from the Genotype-Tissue Expression project (GTEx). Within each tissue, we identified genes that show sex differences in gene expression variability. We found that these sex-differentially variable (SDV) genes are associated with various important biological functions, including sex hormone response, immune response, and other signaling pathways. By analyzing single-cell RNA sequencing data collected from breast epithelial cells, we found that genes with sex differences in gene expression variability in breast tissue tend to be expressed in a cell-type-specific manner. We looked for an association between SDV expression and Graves' disease, a well-known heavily female-biased disease, and found a significant enrichment of Graves' associated genes among genes with higher variability in females in thyroid tissue. This suggests a possible role for SDV expression in sex-biased disease. We then examined the evolutionary constraints acting on genes with sex differences in variability and found that they exhibit evidence of increased selective constraint. Through analysis of sex-biased eQTL data, we found evidence that SDV expression may have a genetic basis. Finally, we propose a simple evolutionary model for the emergence of SDV expression from sex-specific constraints.

Differential compartmentalization of BMP4/NOGGIN requires NOGGIN trans-epithelial transport.

Using self-organizing human models of gastrulation, we previously showed that (1) BMP4 initiates the cascade of events leading to gastrulation, (2) BMP4 signal reception is restricted to the basolateral domain, and (3) in a human-specific manner, BMP4 directly induces the expression of NOGGIN. Here, we report the surprising discovery that in human epiblasts, NOGGIN and BMP4 were secreted into opposite extracellular spaces. Interestingly, apically presented NOGGIN could inhibit basally delivered BMP4. Apically imposed microfluidic flow demonstrated that NOGGIN traveled in the apical extracellular space. Our co-localization analysis detailed the endocytotic route that trafficked NOGGIN from the apical space to the basolateral intercellular space where BMP4 receptors were located. This apical-basal transcytosis was indispensable for NOGGIN inhibition. Taken together, the segregation of activator/inhibitor into distinct extracellular spaces challenges classical views of morphogen movement. We propose that the transport of morphogen inhibitors regulates the spatial availability of morphogens during embryogenesis.

The evolution of sex-biased gene expression in the Drosophila brain.

Genes with sex-biased expression in Drosophila are thought to underlie sexually dimorphic phenotypes and have been shown to possess unique evolutionary properties. However, the forces and constraints governing the evolution of sex-biased genes in the somatic tissues of Drosophila are largely unknown. By using population-scale RNA sequencing data, we show that sex-biased genes in the Drosophila brain are highly enriched on the X Chromosome and that most are biased in a species-specific manner. We show that X-linked male-biased genes, and to a lesser extent female-biased genes, are enriched for signatures of directional selection at the gene expression level. By examining the evolutionary properties of gene-flanking regions on the X Chromosome, we find evidence that adaptive cis-regulatory changes are more likely to drive the expression evolution of X-linked male-biased genes than other X-linked genes. Finally, we examine whether constraint owing to broad expression across multiple tissues and genetic constraint owing to the largely shared male and female genomes could be responsible for the observed patterns of gene expression evolution. We find that expression breadth does not constrain the directional evolution of gene expression in the brain. Additionally, we find that the shared genome between males and females imposes a substantial constraint on the expression evolution of sex-biased genes. Overall, these results significantly advance our understanding of the patterns and forces shaping the evolution of sexual dimorphism in the Drosophila brain.