Structure-primed embedding on the transcription factor manifold enables transparent model architectures for gene regulatory network and latent activity inference

BACKGROUND

Modeling of gene regulatory networks (GRNs) is limited due to a lack of direct measurements of genome-wide transcription factor activity (TFA) making it difficult to separate covariance and regulatory interactions. Inference of regulatory interactions and TFA requires aggregation of complementary evidence. Estimating TFA explicitly is problematic as it disconnects GRN inference and TFA estimation and is unable to account for, for example, contextual transcription factor-transcription factor interactions, and other higher order features. Deep-learning offers a potential solution, as it can model complex interactions and higher-order latent features, although does not provide interpretable models and latent features.

RESULTS

We propose a novel autoencoder-based framework, StrUcture Primed Inference of Regulation using latent Factor ACTivity (SupirFactor) for modeling, and a metric, explained relative variance (ERV), for interpretation of GRNs. We evaluate SupirFactor with ERV in a wide set of contexts. Compared to current state-of-the-art GRN inference methods, SupirFactor performs favorably. We evaluate latent feature activity as an estimate of TFA and biological function in S. cerevisiae as well as in peripheral blood mononuclear cells (PBMC).

CONCLUSION

Here we present a framework for structure-primed inference and interpretation of GRNs, SupirFactor, demonstrating interpretability using ERV in multiple biological and experimental settings. SupirFactor enables TFA estimation and pathway analysis using latent factor activity, demonstrated here on two large-scale single-cell datasets, modeling S. cerevisiae and PBMC. We find that the SupirFactor model facilitates biological analysis acquiring novel functional and regulatory insight.

Tissue-specific RNA-seq defines genes governing male tail tip morphogenesis in C. elegans

Caenorhabditis elegans males undergo sex-specific tail tip morphogenesis (TTM) under the control of the transcription factor DMD-3. To find genes regulated by DMD-3, We performed RNA-seq of laser-dissected tail tips. We identified 564 genes differentially expressed (DE) in wild-type males vs. dmd-3(-) males and hermaphrodites. The transcription profile of dmd-3(-) tail tips is similar to that in hermaphrodites. For validation, we analyzed transcriptional reporters for 49 genes and found male-specific or male-biased expression for 26 genes. Only 11 DE genes overlapped with genes found in a previous RNAi screen for defective TTM. GO enrichment analysis of DE genes finds upregulation of genes within the UPR (unfolded protein response) pathway and downregulation of genes involved in cuticle maintenance. Of the DE genes, 40 are transcription factors, indicating that the gene network downstream of DMD-3 is complex and potentially modular. We propose modules of genes that act together in TTM and are coregulated by DMD-3, among them the chondroitin synthesis pathway and the hypertonic stress response.

Lhx3/4 initiates a cardiopharyngeal-specific transcriptional program in response to widespread FGF signaling

Individual signaling pathways, such as fibroblast growth factors (FGFs), can regulate a plethora of inductive events. According to current paradigms, signal-dependent transcription factors (TFs), such as FGF/MapK-activated Ets family factors, partner with lineage-determining factors to achieve regulatory specificity. However, many aspects of this model have not been rigorously investigated. One key question relates to whether lineage-determining factors dictate lineage-specific responses to inductive signals or facilitate these responses in collaboration with other inputs. We utilize the chordate model Ciona robusta to investigate mechanisms generating lineage-specific induction. Previous studies in C. robusta have shown that cardiopharyngeal progenitor cells are specified through the combined activity of FGF-activated Ets1/2.b and an inferred ATTA-binding transcriptional cofactor. Here, we show that the homeobox TF Lhx3/4 serves as the lineage-determining TF that dictates cardiopharyngeal-specific transcription in response to pleiotropic FGF signaling. Targeted knockdown of Lhx3/4 leads to loss of cardiopharyngeal gene expression. Strikingly, ectopic expression of Lhx3/4 in a neuroectodermal lineage subject to FGF-dependent specification leads to ectopic cardiopharyngeal gene expression in this lineage. Furthermore, ectopic Lhx3/4 expression disrupts neural plate morphogenesis, generating aberrant cell behaviors associated with execution of incompatible morphogenetic programs. Based on these findings, we propose that combinatorial regulation by signal-dependent and lineage-determinant factors represents a generalizable, previously uncategorized regulatory subcircuit we term "cofactor-dependent induction." Integration of this subcircuit into theoretical models will facilitate accurate predictions regarding the impact of gene regulatory network rewiring on evolutionary diversification and disease ontogeny.

Specification and survival of post-metamorphic branchiomeric neurons in the hindbrain of a non-vertebrate chordate

Tunicates are the sister group to the vertebrates, yet most species have a life cycle split between swimming larva and sedentary adult phases. During metamorphosis, larval neurons are largely replaced by adult-specific ones. Yet the regulatory mechanisms underlying this neural replacement remain largely unknown. Using tissue-specific CRISPR/Cas9-mediated mutagenesis in the tunicate Ciona, we show that orthologs of conserved hindbrain and branchiomeric neuron regulatory factors Pax2/5/8 and Phox2 are required to specify the "Neck", a compartment of cells set aside in the larva to give rise to cranial motor neuron-like neurons in the adult. Using bulk and single-cell RNAseq analyses, we also characterize the transcriptome of the Neck downstream of Pax2/5/8. Surprisingly, we find that Neck-derived adult ciliomotor neurons begin to differentiate in the larva, contrary to the long-held assumption that the adult nervous system is formed only after settlement and the death of larval neurons during metamorphosis. Finally, we show that manipulating FGF signaling during the larval phase alters the patterning of the Neck and its derivatives. Suppression of FGF converts Neck cells into larval neurons that fail to survive metamorphosis, while prolonged FGF signaling promotes an adult neural stem cell-like fate instead.

Structure primed embedding on the transcription factor manifold enables transparent model architectures for gene regulatory network and latent activity inference

The modeling of gene regulatory networks (GRNs) is limited due to a lack of direct measurements of regulatory features in genome-wide screens. Most GRN inference methods are therefore forced to model relationships between regulatory genes and their targets with expression as a proxy for the upstream independent features, complicating validation and predictions produced by modeling frameworks. Separating covariance and regulatory influence requires aggregation of independent and complementary sets of evidence, such as transcription factor (TF) binding and target gene expression. However, the complete regulatory state of the system, e.g. TF activity (TFA) is unknown due to a lack of experimental feasibility, making regulatory relations difficult to infer. Some methods attempt to account for this by modeling TFA as a latent feature, but these models often use linear frameworks that are unable to account for non-linearities such as saturation, TF-TF interactions, and other higher order features. Deep learning frameworks may offer a solution, as they are capable of modeling complex interactions and capturing higher-order latent features. However, these methods often discard central concepts in biological systems modeling, such as sparsity and latent feature interpretability, in favor of increased model complexity. We propose a novel deep learning autoencoder-based framework, StrUcture Primed Inference of Regulation using latent Factor ACTivity (SupirFactor), that scales to single cell genomic data and maintains interpretability to perform GRN inference and estimate TFA as a latent feature. We demonstrate that SupirFactor outperforms current leading GRN inference methods, predicts biologically relevant TFA and elucidates functional regulatory pathways through aggregation of TFs.

Cell Architecture-Dependent Constraints: Critical Safeguards to Carcinogenesis

Animal cells display great diversity in their shape. These morphological characteristics result from crosstalk between the plasma membrane and the force-generating capacities of the cytoskeleton macromolecules. Changes in cell shape are not merely byproducts of cell fate determinants, they also actively drive cell fate decisions, including proliferation and differentiation. Global and local changes in cell shape alter the transcriptional program by a multitude of mechanisms, including the regulation of physical links between the plasma membrane and the nuclear envelope and the mechanical modulation of cation channels and signalling molecules. It is therefore not surprising that anomalies in cell shape contribute to several diseases, including cancer. In this review, we discuss the possibility that the constraints imposed by cell shape determine the behaviour of normal and pro-tumour cells by organizing the whole interconnected regulatory network. In turn, cell behaviour might stabilize cells into discrete shapes. However, to progress towards a fully transformed phenotype and to acquire plasticity properties, pro-tumour cells might need to escape these cell shape restrictions. Thus, robust controls of the cell shape machinery may represent a critical safeguard against carcinogenesis.

Supracellular organization confers directionality and mechanical potency to migrating pairs of cardiopharyngeal progenitor cells

Physiological and pathological morphogenetic events involve a wide array of collective movements, suggesting that multicellular arrangements confer biochemical and biomechanical properties contributing to tissue-scale organization. The Ciona cardiopharyngeal progenitors provide the simplest model of collective cell migration, with cohesive bilateral cell pairs polarized along the leader-trailer migration path while moving between the ventral epidermis and trunk endoderm. We use the Cellular Potts Model to computationally probe the distributions of forces consistent with shapes and collective polarity of migrating cell pairs. Combining computational modeling, confocal microscopy, and molecular perturbations, we identify cardiopharyngeal progenitors as the simplest cell collective maintaining supracellular polarity with differential distributions of protrusive forces, cell-matrix adhesion, and myosin-based retraction forces along the leader-trailer axis. 4D simulations and experimental observations suggest that cell-cell communication helps establish a hierarchy to align collective polarity with the direction of migration, as observed with three or more cells in silico and in vivo. Our approach reveals emerging properties of the migrating collective: cell pairs are more persistent, migrating longer distances, and presumably with higher accuracy. Simulations suggest that cell pairs can overcome mechanical resistance of the trunk endoderm more effectively when they are polarized collectively. We propose that polarized supracellular organization of cardiopharyngeal progenitors confers emergent physical properties that determine mechanical interactions with their environment during morphogenesis.

Regulation of Neurogenesis by FGF Signaling and Neurogenin in the Invertebrate Chordate Ciona

Neurogenesis is a complex sequence of cellular processes and behaviors driven by the coordinated expression of conserved effectors. The bipolar tail neurons (BTNs) of Ciona develop according to a highly dynamic, yet highly stereotyped developmental program and thus could serve as an accessible model system for neurogenesis, including underlying cell behaviors like neuronal delamination, migration, and polarized axon outgrowth. Here we investigate both the upstream events that shape BTN neurogenesis through spatiotemporal regulation of the conserved proneural factor Neurog, spatiotemporal, and the gene expression profile of differentiating BTNs downstream of Neurog activity. We show that, although early FGF signaling is required for Neurog expression and BTN specification, Fgf8/17/18 is expressed in tail tip cells at later stages and suppresses sustained Neurog expression in the anterior BTN (aBTN) lineage, such that only one cell (the one furthest from the source of Fgf8/17/18) maintains Neurog expression and becomes a neuron. Curiously, Fgf8/17/18 might not affect neurogenesis of the posterior BTNs (pBTNs), which are in direct contact with the Fgf8/17/18-expressing cells. Finally, to profile gene expression associated with BTN neurogenesis we performed RNAseq of isolated BTN lineage cells in which BTN neurogenesis was enhanced or suppressed by perturbing Neurog function. This allowed us to identify several candidate genes that might play conserved roles in neurogenesis and neuronal migration in other animals, including mammals.

Proteomic and interactomic insights into the molecular basis of cell functional diversity

The ability of living systems to adapt to changing conditions originates from their capacity to change their molecular constitution. This is achieved by multiple mechanisms that modulate the quantitative composition and the diversity of the molecular inventory. Molecular diversification is particularly pronounced on the proteome level, at which multiple proteoforms derived from the same gene can in turn combinatorially form different protein complexes, thus expanding the repertoire of functional modules in the cell. The study of molecular and modular diversity and their involvement in responses to changing conditions has only recently become possible through the development of new 'omics'-based screening technologies. This Review explores our current knowledge of the mechanisms regulating functional diversification along the axis of gene expression, with a focus on the proteome and interactome. We explore the interdependence between different molecular levels and how this contributes to functional diversity. Finally, we highlight several recent techniques for studying molecular diversity, with specific focus on mass spectrometry-based analysis of the proteome and its organization into functional modules, and examine future directions for this rapidly growing field.

From cell shape to cell fate via the cytoskeleton - Insights from the epidermis

Animal cells exhibit a wide range of shapes that reflect their diverse functions. Cell shape is determined by a balance between internal and external forces and therefore involves the cytoskeleton and its associated adhesion structures. Cell shape dynamics during development and homeostasis are tightly regulated and closely coordinated with cell fate determination. Defects in cell shape are a hallmark of many pathological conditions including cancer and skin diseases. This review highlights the links between cell shape and cell fate in the epidermis, which have been studied for over 40 years both in vitro and in vivo. Briefly discussing seminal experiments showing the strong coupling between keratinocyte cell shape and their fate we primarily focus on recent studies uncovering novel cellular and molecular mechanisms linking epidermal cell shape with cell growth, differentiation, asymmetric division, and delamination.

Modular co-option of cardiopharyngeal genes during non-embryonic myogenesis

Background

In chordates, cardiac and body muscles arise from different embryonic origins. In addition, myogenesis can be triggered in adult organisms, during asexual development or regeneration. In non-vertebrate chordates like ascidians, muscles originate from embryonic precursors regulated by a conserved set of genes that orchestrate cell behavior and dynamics during development. In colonial ascidians, besides embryogenesis and metamorphosis, an adult can propagate asexually via blastogenesis, skipping embryo and larval stages, and form anew the adult body, including the complete body musculature.

Results

To investigate the cellular origin and mechanisms that trigger non-embryonic myogenesis, we followed the expression of ascidian myogenic genes during Botryllus schlosseri blastogenesis and reconstructed the dynamics of muscle precursors. Based on the expression dynamics of Tbx1/10, Ebf, Mrf, Myh3 for body wall and of FoxF, Tbx1/10, Nk4, Myh2 for heart development, we show that the embryonic factors regulating myogenesis are only partially co-opted in blastogenesis, and that markers for muscle precursors are expressed in two separate domains: the dorsal tube and the ventral mesenchyma.

Conclusions

Regardless of the developmental pathway, non-embryonic myogenesis shares a similar molecular and anatomical setup as embryonic myogenesis, but implements a co-option and loss of molecular modules. We then propose that the cellular precursors contributing to heart and body muscles may have different origins and may be coordinated by different developmental pathways.

Electronic supplementary material

The online version of this article (10.1186/s13227-019-0116-7) contains supplementary material, which is available to authorized users.

Discoidin-domain receptor coordinates cell-matrix adhesion and collective polarity in migratory cardiopharyngeal progenitors

Integrated analyses of regulated effector genes, cellular processes, and extrinsic signals are required to understand how transcriptional networks coordinate fate specification and cell behavior during embryogenesis. Ciona cardiopharyngeal progenitors, the trunk ventral cells (TVCs), polarize as leader and trailer cells that migrate between the ventral epidermis and trunk endoderm. We show that the TVC-specific collagen-binding Discoidin-domain receptor (Ddr) cooperates with Integrin-β1 to promote cell-matrix adhesion. We find that endodermal cells secrete a collagen, Col9-a1, that is deposited in the basal epidermal matrix and promotes Ddr activation at the ventral membrane of migrating TVCs. A functional antagonism between Ddr/Intβ1-mediated cell-matrix adhesion and Vegfr signaling appears to modulate the position of cardiopharyngeal progenitors between the endoderm and epidermis. We show that Ddr promotes leader-trailer-polarized BMP-Smad signaling independently of its role in cell-matrix adhesion. We propose that dual functions of Ddr integrate transcriptional inputs to coordinate subcellular processes underlying collective polarity and migration.

Evidence of Extracellular Vesicles Biogenesis and Release in Mouse Embryonic Stem Cells

Extracellular vesicles (EVs) released by mouse embryonic stem cells (mESCs) are considered a source of bioactive molecules that modulate their microenvironment by acting on intercellular communication. Either intracellular endosomal machinery or their derived EVs have been considered a relevant system of signal circuits processing. Herein, we show that these features are found in mESCs. Ultrastructural analysis revealed structures and organelles of the endosomal system such as coated pits and endocytosis-related vesicles, prominent rough endoplasmic reticulum and Golgi apparatus, and multivesicular bodies (MVBs) containing either few or many intraluminal vesicles (ILVs) that could be released as exosomes to extracellular milieu. Besides, budding vesicles shed from the plasma membrane to the extracellular space is suggestive of microvesicle biogenesis in mESCs. mESCs and mouse blastocyst express specific markers of the Endosomal Sorting Complex Required for Transport (ESCRT) system. Ultrastructural analysis and Nanoparticle Tracking Analysis (NTA) of isolated EVs revealed a heterogeneous population of exosomes and microvesicles released by mESCs. These vesicles contain Wnt10b and the Notch ligand Delta-like 4 (DLL4) and also the co-chaperone stress inducible protein 1 (STI1) and its partner Hsp90. Wnt10b and Dll4 colocalize with EVs biogenesis markers in mESCs. Overall, the present study supports the function of the mESCs endocytic network and their EVs as players in stem cell biology.

Prenatal stress and the developing brain: Risks for neurodevelopmental disorders

The prenatal period is increasingly considered as a crucial target for the primary prevention of neurodevelopmental and psychiatric disorders. Understanding their pathophysiological mechanisms remains a great challenge. Our review reveals new insights from prenatal brain development research, involving (epi)genetic research, neuroscience, recent imaging techniques, physical modeling, and computational simulation studies. Studies examining the effect of prenatal exposure to maternal distress on offspring brain development, using brain imaging techniques, reveal effects at birth and up into adulthood. Structural and functional changes are observed in several brain regions including the prefrontal, parietal, and temporal lobes, as well as the cerebellum, hippocampus, and amygdala. Furthermore, alterations are seen in functional connectivity of amygdalar-thalamus networks and in intrinsic brain networks, including default mode and attentional networks. The observed changes underlie offspring behavioral, cognitive, emotional development, and susceptibility to neurodevelopmental and psychiatric disorders. It is concluded that used brain measures have not yet been validated with regard to sensitivity, specificity, accuracy, or robustness in predicting neurodevelopmental and psychiatric disorders. Therefore, more prospective long-term longitudinal follow-up studies starting early in pregnancy should be carried out, in order to examine brain developmental measures as mediators in mediating the link between prenatal stress and offspring behavioral, cognitive, and emotional problems and susceptibility for disorders.

From pattern to process: studies at the interface of gene regulatory networks, morphogenesis, and evolution

The development of anatomical structures is complex, beginning with patterning of gene expression by multiple gene regulatory networks (GRNs). These networks ultimately regulate the activity of effector molecules, which in turn alter cellular behavior during development. Together these processes biomechanically produce the three-dimensional shape that the anatomical structure adopts over time. However, the interfaces between these processes are often overlooked and also include counter-intuitive feedback mechanisms. In this review, we examine each step in this extraordinarily complex process and explore how evolutionary developmental biology model systems, such as butterfly scales, vertebrate teeth, and the Drosophila dorsal appendage offer a complementary approach to expose the multifactorial integration of genetics and morphogenesis from an alternative perspective.

Lineage specification of early embryos and embryonic stem cells at the dawn of enabling technologies

Establishment of progenitor cell populations and lineage diversity during embryogenesis and the differentiation of pluripotent stem cells is a fascinating and intricate biological process. Conceptually, an understanding of this developmental process provides a framework to integrate stem-cell pluripotency, cell competence and differentiating potential with the activity of extrinsic and intrinsic molecular determinants. The recent advent of enabling technologies of high-resolution transcriptome analysis at the cellular, population and spatial levels proffers the capability of gaining deeper insights into the attributes of the gene regulatory network and molecular signaling in lineage specification and differentiation. In this review, we provide a snapshot of the emerging enabling genomic technologies that contribute to the study of development and stem-cell biology.

Concerted action of neuroepithelial basal shrinkage and active epithelial migration ensures efficient optic cup morphogenesis

Organ formation is a multi-scale event that involves changes at the intracellular, cellular and tissue level. Organogenesis often starts with the formation of characteristically shaped organ precursors. However, the cellular mechanisms driving organ precursor formation are often not clear. Here, using zebrafish, we investigate the epithelial rearrangements responsible for the development of the hemispherical retinal neuroepithelium (RNE), a part of the optic cup. We show that in addition to basal shrinkage of RNE cells, active migration of connected epithelial cells into the RNE is a crucial player in its formation. This cellular movement is driven by progressive cell-matrix contacts and actively translocates prospective RNE cells to their correct location before they adopt neuroepithelial fate. Failure of this migration during neuroepithelium formation leads to ectopic determination of RNE cells and consequently impairs optic cup formation. Overall, this study illustrates how spatiotemporal coordination between morphogenic movements and fate determination critically influences organogenesis.