Image restoration of degraded time-lapse microscopy data mediated by near-infrared imaging.

Time-lapse fluorescence microscopy is key to unraveling biological development and function; however, living systems, by their nature, permit only limited interrogation and contain untapped information that can only be captured by more invasive methods. Deep-tissue live imaging presents a particular challenge owing to the spectral range of live-cell imaging probes/fluorescent proteins, which offer only modest optical penetration into scattering tissues. Herein, we employ convolutional neural networks to augment live-imaging data with deep-tissue images taken on fixed samples. We demonstrate that convolutional neural networks may be used to restore deep-tissue contrast in GFP-based time-lapse imaging using paired final-state datasets acquired using near-infrared dyes, an approach termed InfraRed-mediated Image Restoration (IR2). Notably, the networks are remarkably robust over a wide range of developmental times. We employ IR2 to enhance the information content of green fluorescent protein time-lapse images of zebrafish and Drosophila embryo/larval development and demonstrate its quantitative potential in increasing the fidelity of cell tracking/lineaging in developing pescoids. Thus, IR2 is poised to extend live imaging to depths otherwise inaccessible.

Arrested coalescence of multicellular aggregates.

Multicellular aggregates are known to exhibit liquid-like properties. The fusion process of two cell aggregates is commonly studied as the coalescence of two viscous drops. However, tissues are complex materials and can exhibit viscoelastic behaviour. It is known that elastic effects can prevent the complete fusion of two drops, a phenomenon known as arrested coalescence. Here we study this phenomenon in stem cell aggregates and provide a theoretical framework which agrees with the experiments. In addition, agent-based simulations show that active cell fluctuations can control a solid-to-fluid phase transition, revealing that arrested coalescence can be found in the vicinity of an unjamming transition. By analysing the dynamics of the fusion process and combining it with nanoindentation measurements, we obtain the effective viscosity, shear modulus and surface tension of the aggregates. More generally, our work provides a simple, fast and inexpensive method to characterize the mechanical properties of viscoelastic materials.

Periodic formation of epithelial somites from human pluripotent stem cells.

During embryonic development, epithelial cell blocks called somites are periodically formed according to the segmentation clock, becoming the foundation for the segmental pattern of the vertebral column. The process of somitogenesis has recently been recapitulated with murine and human pluripotent stem cells. However, an in vitro model for human somitogenesis coupled with the segmentation clock and epithelialization is still missing. Here, we report the generation of human somitoids, organoids that periodically form pairs of epithelial somite-like structures. Somitoids display clear oscillations of the segmentation clock that coincide with the segmentation of the presomitic mesoderm. The resulting somites show anterior-posterior and apical-basal polarities. Matrigel is essential for epithelialization but dispensable for the differentiation into somite cells. The size of somites is rather constant, irrespective of the initial cell number. The amount of WNT signaling instructs the proportion of mesodermal lineages in somitoids. Somitoids provide a novel platform to study human somitogenesis.

MOrgAna: accessible quantitative analysis of organoids with machine learning.

Recent years have seen a dramatic increase in the application of organoids to developmental biology, biomedical and translational studies. Organoids are large structures with high phenotypic complexity and are imaged on a wide range of platforms, from simple benchtop stereoscopes to high-content confocal-based imaging systems. The large volumes of images, resulting from hundreds of organoids cultured at once, are becoming increasingly difficult to inspect and interpret. Hence, there is a pressing demand for a coding-free, intuitive and scalable solution that analyses such image data in an automated yet rapid manner. Here, we present MOrgAna, a Python-based software that implements machine learning to segment images, quantify and visualize morphological and fluorescence information of organoids across hundreds of images, each with one object, within minutes. Although the MOrgAna interface is developed for users with little to no programming experience, its modular structure makes it a customizable package for advanced users. We showcase the versatility of MOrgAna on several in vitro systems, each imaged with a different microscope, thus demonstrating the wide applicability of the software to diverse organoid types and biomedical studies.

Maximizing content across scales: Moving multimodal microscopy and mesoscopy toward molecular imaging.

Molecular imaging aims to depict the molecules in living patients. However, because this aim is still far beyond reach, patchworks of different solutions need to be used to tackle this overarching goal. From the vast toolbox of imaging techniques, we focus on those recent advances in optical microscopy that image molecules and cells at the submicron to centimeter scale. Mesoscopic imaging covers the "imaging gap" between techniques such as confocal microscopy and magnetic resonance imagingthat image entire live samples but with limited resolution. Microscopy focuses on the cellular level; mesoscopy visualizes the organization of molecules and cells into tissues and organs. The correlation between these techniques allows us to combine disciplines ranging from whole body imaging to basic research of model systems. We review current developments focused on improving microscopic and mesoscopic imaging technologies and on hardware and software that push the current sensitivity and resolution boundaries.

A role for the fusogen eff-1 in epidermal stem cell number robustness in Caenorhabditis elegans.

Developmental patterning in Caenorhabditis elegans is known to proceed in a highly stereotypical manner, which raises the question of how developmental robustness is achieved despite the inevitable stochastic noise. We focus here on a population of epidermal cells, the seam cells, which show stem cell-like behaviour and divide symmetrically and asymmetrically over post-embryonic development to generate epidermal and neuronal tissues. We have conducted a mutagenesis screen to identify mutants that introduce phenotypic variability in the normally invariant seam cell population. We report here that a null mutation in the fusogen eff-1 increases seam cell number variability. Using time-lapse microscopy and single molecule fluorescence hybridisation, we find that seam cell division and differentiation patterns are mostly unperturbed in eff-1 mutants, indicating that cell fusion is uncoupled from the cell differentiation programme. Nevertheless, seam cell losses due to the inappropriate differentiation of both daughter cells following division, as well as seam cell gains through symmetric divisions towards the seam cell fate were observed at low frequency. We show that these stochastic errors likely arise through accumulation of defects interrupting the continuity of the seam and changing seam cell shape, highlighting the role of tissue homeostasis in suppressing phenotypic variability during development.

Rethinking embryology in vitro: A synergy between engineering, data science and theory.

Pluripotent stem cells, in the recent years, have been demonstrated to mimic different aspects of metazoan embryonic development in vitro. This has led to the establishment of synthetic embryology: a field that makes use of in vitro stem cell models to investigate developmental processes that would be otherwise inaccessible in vivo. Currently, a plethora of engineering-inspired techniques, including microfluidic devices and bioreactors, exist to generate and culture organoids at high throughput. Similarly, data analysis and deep learning-based techniques, that were established in in vivo models, are now being used to extract quantitative information from synthetic systems. Finally, theory and data-driven in silico modeling are starting to provide a system-level understanding of organoids and make predictions to be tested with further experiments. Here, we discuss our vision of how engineering, data science and theoretical modeling will synergize to offer an unprecedented view of embryonic development. For every one of these three scientific domains, we discuss examples from in vivo and in vitro systems that we think will pave the way to future developments of synthetic embryology.

Stochastic loss and gain of symmetric divisions in the C. elegans epidermis perturbs robustness of stem cell number.

Biological systems are subject to inherent stochasticity. Nevertheless, development is remarkably robust, ensuring the consistency of key phenotypic traits such as correct cell numbers in a certain tissue. It is currently unclear which genes modulate phenotypic variability, what their relationship is to core components of developmental gene networks, and what is the developmental basis of variable phenotypes. Here, we start addressing these questions using the robust number of Caenorhabditis elegans epidermal stem cells, known as seam cells, as a readout. We employ genetics, cell lineage tracing, and single molecule imaging to show that mutations in lin-22, a Hes-related basic helix-loop-helix (bHLH) transcription factor, increase seam cell number variability. We show that the increase in phenotypic variability is due to stochastic conversion of normally symmetric cell divisions to asymmetric and vice versa during development, which affect the terminal seam cell number in opposing directions. We demonstrate that LIN-22 acts within the epidermal gene network to antagonise the Wnt signalling pathway. However, lin-22 mutants exhibit cell-to-cell variability in Wnt pathway activation, which correlates with and may drive phenotypic variability. Our study demonstrates the feasibility to study phenotypic trait variance in tractable model organisms using unbiased mutagenesis screens.

Long-term time-lapse microscopy of C. elegans post-embryonic development.

We present a microscopy technique that enables long-term time-lapse microscopy at single-cell resolution in moving and feeding Caenorhabditis elegans larvae. Time-lapse microscopy of C. elegans post-embryonic development is challenging, as larvae are highly motile. Moreover, immobilization generally leads to rapid developmental arrest. Instead, we confine larval movement to microchambers that contain bacteria as food, and use fast image acquisition and image analysis to follow the dynamics of cells inside individual larvae, as they move within each microchamber. This allows us to perform fluorescence microscopy of 10-20 animals in parallel with 20 min time resolution. We demonstrate the power of our approach by analysing the dynamics of cell division, cell migration and gene expression over the full ∼48 h of development from larva to adult. Our approach now makes it possible to study the behaviour of individual cells inside the body of a feeding and growing animal.

Using noise to control heterogeneity of isogenic populations in homogenous environments.

We explore the extent to which the phenotypes of individual, genetically identical cells can be controlled independently from each other using only a single homogeneous environmental input. We show that such control is theoretically impossible if restricted to a deterministic setting, but it can be achieved readily if one exploits heterogeneities introduced at the single-cell level due to stochastic fluctuations in gene regulation. Using stochastic analyses of a bistable genetic toggle switch, we develop a control strategy that maximizes the chances that a chosen cell will express one phenotype, while the rest express another. The control mechanism uses UV radiation to enhance identically protein degradation in all cells. Control of individual cells is made possible only by monitoring stochastic protein fluctuations and applying UV control at favorable times and levels. For two identical cells, our stochastic control law can drive protein expression of a chosen cell above its neighbor with a better than 99% success rate. In a population of 30 identical cells, we can drive a given cell to remain consistently within the top 20%. Although cellular noise typically impairs predictability for biological responses, our results show that it can also simultaneously improve controllability for those same responses.

Modeling leukocyte-leukocyte non-contact interactions in a lymph node.

The interaction among leukocytes is at the basis of the innate and adaptive immune-response and it is largely ascribed to direct cell-cell contacts. However, the exchange of a number of chemical stimuli (chemokines) allows also non-contact interaction during the immunological response. We want here to evaluate the extent of the effect of the non-contact interactions on the observed leukocyte-leukocyte kinematics and their interaction duration. To this aim we adopt a simplified mean field description inspired by the Keller-Segel chemotaxis model, of which we report an analytical solution suited for slowly varying sources of chemokines. Since our focus is on the non-contact interactions, leukocyte-leukocyte contact interactions are simulated only by means of a space dependent friction coefficient of the cells. The analytical solution of the Keller-Segel model is then taken as the basis of numerical simulations of interactions between leukocytes and their duration. The mean field interaction force that we derive has a time-space separable form and depends on the chemotaxis sensitivity parameter as well as on the chemokines diffusion coefficient and their degradation rate. All these parameters affect the distribution of the interaction durations. We draw a successful qualitative comparison between simulated data and sets of experimental data for DC-NK cells interaction duration and other kinematic parameters. Remarkably, the predicted percentage of the leukocyte-leukocyte interactions falls in the experimental range and depends (~25% increase) upon the chemotactic parameter indicating a non-negligible direct effect of the non-contact interaction on the leukocyte interactions.