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.

Transcription Factors Drive Opposite Relationships between Gene Age and Tissue Specificity in Male and Female Drosophila Gonads.

Evolutionarily young genes are usually preferentially expressed in the testis across species. Although it is known that older genes are generally more broadly expressed than younger genes, the properties that shaped this pattern are unknown. Older genes may gain expression across other tissues uniformly, or faster in certain tissues than others. Using Drosophila gene expression data, we confirmed previous findings that younger genes are disproportionately testis biased and older genes are disproportionately ovary biased. We found that the relationship between gene age and expression is stronger in the ovary than any other tissue and weakest in testis. We performed ATAC-seq on Drosophila testis and found that although genes of all ages are more likely to have open promoter chromatin in testis than in ovary, promoter chromatin alone does not explain the ovary bias of older genes. Instead, we found that upstream transcription factor (TF) expression is highly predictive of gene expression in ovary but not in testis. In the ovary, TF expression is more predictive of gene expression than open promoter chromatin, whereas testis gene expression is similarly influenced by both TF expression and open promoter chromatin. We propose that the testis is uniquely able to express younger genes controlled by relatively few TFs, whereas older genes with more TF partners are broadly expressed with peak expression most likely in the ovary. The testis allows widespread baseline expression that is relatively unresponsive to regulatory changes, whereas the ovary transcriptome is more responsive to trans-regulation and has a higher ceiling for gene expression.

Testis single-cell RNA-seq reveals the dynamics of de novo gene transcription and germline mutational bias in Drosophila.

The testis is a peculiar tissue in many respects. It shows patterns of rapid gene evolution and provides a hotspot for the origination of genetic novelties such as de novo genes, duplications and mutations. To investigate the expression patterns of genetic novelties across cell types, we performed single-cell RNA-sequencing of adult Drosophila testis. We found that new genes were expressed in various cell types, the patterns of which may be influenced by their mode of origination. In particular, lineage-specific de novo genes are commonly expressed in early spermatocytes, while young duplicated genes are often bimodally expressed. Analysis of germline substitutions suggests that spermatogenesis is a highly reparative process, with the mutational load of germ cells decreasing as spermatogenesis progresses. By elucidating the distribution of genetic novelties across spermatogenesis, this study provides a deeper understanding of how the testis maintains its core reproductive function while being a hotbed of evolutionary innovation.

EHD2 shuttles to the nucleus and represses transcription.

EHD {EH [Eps15 (epidermal growth factor receptor substrate 15) homology]-domain-containing} proteins participate in several endocytic events, such as the internalization and the recycling processes. There are four EHD proteins in mammalian cells, EHD1-EHD4, each with diverse roles in the recycling pathway of endocytosis. EHD2 is a plasma-membrane-associated member of the EHD family that regulates internalization. Since several endocytic proteins have been shown to undergo nucleocytoplasmic shuttling and have been assigned roles in regulation of gene expression, we tested the possibility that EHD proteins also shuttle to the nucleus. Our results showed that, among the three EHD proteins (EHD1-EHD3) that were tested, only EHD2 accumulates in the nucleus under nuclear export inhibition treatment. Moreover, the presence of a NLS (nuclear localization signal) was essential for its entry into the nucleus. Nuclear exit of EHD2 depended partially on its NES (nuclear export signal). Elimination of a potential SUMOylation site in EHD2 resulted in a major accumulation of the protein in the nucleus, indicating the involvement of SUMOylation in the nuclear exit of EHD2. We confirmed the SUMOylation of EHD2 by employing co-immunoprecipitation and the yeast two-hybrid system. Using GAL4-based transactivation assay as well as a KLF7 (Krüppel-like factor 7)-dependent transcription assay of the p21WAF1/Cip1 [CDKN1A (cyclin-dependent kinase inhibitor 1A)] gene, we showed that EHD2 represses transcription. qRT-PCR (quantitative real-time PCR) of RNA from cells overexpressing EHD2 or of RNA from cells knocked down for EHD2 confirmed that EHD2 represses transcription of the p21WAF1/Cip1 gene.

EHD2 mediates trafficking from the plasma membrane by modulating Rac1 activity.

EHDs [EH (Eps15 homology)-domain-containing proteins] participate in different stages of endocytosis. EHD2 is a plasma-membrane-associated EHD which regulates trafficking from the plasma membrane and recycling. EHD2 has a role in nucleotide-dependent membrane remodelling and its ATP-binding domain is involved in dimerization, which creates a membrane-binding region. Nucleotide binding is important for association of EHD2 with the plasma membrane, since a nucleotide-free mutant (EHD2 T72A) failed to associate. To elucidate the possible function of EHD2 during endocytic trafficking, we attempted to unravel proteins that interact with EHD2, using the yeast two-hybrid system. A novel interaction was found between EHD2 and Nek3 [NIMA (never in mitosis in Aspergillus nidulans)-related kinase 3], a serine/threonine kinase. EHD2 was also found in association with Vav1, a Nek3-regulated GEF (guanine-nucleotide-exchange factor) for Rho GTPases. Since Vav1 regulates Rac1 activity and promotes actin polymerization, the impact of overexpression of EHD2 on Rac1 activity was tested. The results indicated that wt (wild-type) EHD2, but not its P-loop mutants, reduced Rac1 activity. The inhibitory effect of EHD2 overexpression was partially rescued by co-expression of Rac1 as measured using a cholera toxin trafficking assay. The results of the present study strongly indicate that EHD2 regulates trafficking from the plasma membrane by controlling Rac1 activity.

A conserved F box regulatory complex controls proteasome activity in Drosophila.

The ubiquitin-proteasome system catalyzes the degradation of intracellular proteins. Although ubiquitination of proteins determines their stabilities, there is growing evidence that proteasome function is also regulated. We report the functional characterization of a conserved proteasomal regulatory complex. We identified DmPI31 as a binding partner of the F box protein Nutcracker, a component of an SCF ubiquitin ligase (E3) required for caspase activation during sperm differentiation in Drosophila. DmPI31 binds Nutcracker via a conserved mechanism that is also used by mammalian FBXO7 and PI31. Nutcracker promotes DmPI31 stability, which is necessary for caspase activation, proteasome function, and sperm differentiation. DmPI31 can activate 26S proteasomes in vitro, and increasing DmPI31 levels suppresses defects caused by diminished proteasome activity in vivo. Furthermore, loss of DmPI31 function causes lethality, cell-cycle abnormalities, and defects in protein degradation, demonstrating that DmPI31 is physiologically required for normal proteasome activity.

Another tier for caspase regulation: IAPs as NEDD8 E3 ligases.

Many inhibitor of apoptosis proteins (IAPs) function as E3 ligases to ubiquitinate important cell death proteins, including caspases. Broemer et al. (2010) report recently in Molecular Cell that IAPs can also inhibit caspases by promoting conjugation of the UBL NEDD8.

Drosophila Past1 is involved in endocytosis and is required for germline development and survival of the adult fly.

Endocytosis, which is a key process in eukaryotic cells, has a central role in maintaining cellular homeostasis, nutrient uptake, development and downregulation of signal transduction. This complex process depends on several protein-protein interactions mediated by specific modules. One such module is the EH domain. The EH-domain-containing proteins comprise a family that includes four vertebrate members (EHD1-EHD4) and one Drosophila ortholog, Past1. We used Drosophila as a model to understand the physiological role of this family of proteins. We observed that the two predicted Past1 transcripts are differentially expressed both temporally and spatially during the life cycle of the fly. Endogenous Past1 as well as Past1A and Past1B, expressed from plasmids, were localized mainly to the membrane of Drosophila-derived cells. We generated mutants in the Past1 gene by excising a P-element inserted in it. The Past1 mutants reached adulthood but died precociously. They were temperature sensitive and infertile because of lesions in the reproductive system. Garland cells that originated from Past1 mutants exhibited a marked decrease in their ability to endocytose fluorescently labeled avidin. Genetic interaction was found between Past1 and members of the Notch signaling pathway, suggesting a role for Past1 in this developmentally crucial signaling pathway.

AtEHDs, novel Arabidopsis EH-domain-containing proteins involved in endocytosis.

SUMMARY: Endocytosis is an essential process by which the eukaryotic cell internalizes exogenous material. Studies in yeast and mammalian cells have revealed that endocytosis is a complex molecular process depending on regulated interactions between a variety of proteins and lipids through specific modules. One such module is the Eps15 homology (EH) domain, a conserved modular protein-interaction domain found in several endocytic proteins. The EH-domain-containing proteins function as regulators of endocytosis through their ability to interact with other proteins involved in this process. Here we describe the isolation and characterization of two Arabidopsis EH-domain-containing proteins (AtEHD1 and AtEHD2). We show that the two proteins are involved in endocytosis in plant systems and demonstrate that the Arabidopsis EHD proteins function similarly to mammalian EHDs. Similarly to hEHD2, over-expression of AtEHD2 has an inhibitory effect on endocytosis. While transgenic plants over-expressing AtEHD1 had no detectable phenotype, downregulation of AtEHD1 caused retardation of entry of endocytosed material into plant cells.

AtEHDs in endocytosis.

Endocytosis regulates many important and diverse processes in eukaryotic life. EH domain containing proteins function as regulators of endocytosis through protein-protein interactions. Several interactors of mammalian EHDs were identified, including clathrin machinery components. The four human EHD proteins share high homology at the protein level and possess similar domains, but appear to be involved in different stages of intracellular trafficking. EHD1 regulates recycling through the endocytic recycling compartment (ERC). EHD2 has been found to inhibit internalization in mammalians when overexpressed.We have recently investigated the importance of EH domain containing proteins in plant endocytosis. We were able to show that both of the Arabidopsis EHD homologs, termed AtEHD1 and AtEHD2, play important roles in plant endocytosis. Knockdown of AtEHD1 delayed internalization, and overexpression of AtEHD2 inhibited endocytosis. Thus, the function of plant EHDs is highly homologous to that of mammalian EHDs.

EHD3: a protein that resides in recycling tubular and vesicular membrane structures and interacts with EHD1.

Here we report the characterization of an eps15 homology (EH) domain containing protein designated EHD3. EHD3 was mapped to human chromosome 2p22-23, while the murine Ehd3 homolog was mapped to chromosome 17p21. Both the human and the mouse genes contain a polymorphic (CA) repeat in their 3'UTR. One 3.6-kb Ehd3 transcript was mainly detected in adult mouse brain and kidney and at day 7 of mouse development. On the other hand, human tissues exhibited two, 4.2- and 3.6-kb, EHD3 RNA species. They were predominantly expressed in heart, brain, placenta, liver, kidney and ovary. EHD3, expressed as a green fluorescent fusion protein was localized to endocytic vesicles and to microtubule-dependent, membrane tubules. There was a clear colocalization of EHD3-positive structures and transferrin-containing recycling vesicles, implying that EHD3 resides within the endocytic recycling compartment. Shuffling the N-terminal domain of EHD1 (previously shown to reside in the transferrin-containing, endocytic recycling compartment) with that of EHD3 resulted in a chimeric EHD protein that was localized mainly to tubules instead of the endocytic vesicles, implicating the N-terminal domain as responsible for the tubular localization of EHD3. Mutant EHD3 forms, missing the N-terminal or the C-terminal domains, lost their tubular localization. Results of two-hybrid analyses indicated that EHD1 and EHD3 interact with each other. In addition, EHD1 and EHD3 could be coimmunoprecipitated from cellular extracts, confirming the interaction implied by two-hybrid analysis. Moreover, coexpression of EHD1 and EHD3 resulted in their colocalization in microtubule-dependent tubules as well as in punctate forms. Based on its specific intracellular localization and its interaction with EHD1, we postulate that EHD3 localizes on endocytic tubular and vesicular structures and regulates their microtubule-dependent movement.