Label-free imaging of 3D pluripotent stem cell differentiation dynamics on chip.

Massive, parallelized 3D stem cell cultures for engineering in vitro human cell types require imaging methods with high time and spatial resolution to fully exploit technological advances in cell culture technologies. Here, we introduce a large-scale integrated microfluidic chip platform for automated 3D stem cell differentiation. To fully enable dynamic high-content imaging on the chip platform, we developed a label-free deep learning method called Bright2Nuc to predict in silico nuclear staining in 3D from confocal microscopy bright-field images. Bright2Nuc was trained and applied to hundreds of 3D human induced pluripotent stem cell cultures differentiating toward definitive endoderm on a microfluidic platform. Combined with existing image analysis tools, Bright2Nuc segmented individual nuclei from bright-field images, quantified their morphological properties, predicted stem cell differentiation state, and tracked the cells over time. Our methods are available in an open-source pipeline, enabling researchers to upscale image acquisition and phenotyping of 3D cell culture.

Classification of red cell dynamics with convolutional and recurrent neural networks: a sickle cell disease case study.

The fraction of red blood cells adopting a specific motion under low shear flow is a promising inexpensive marker for monitoring the clinical status of patients with sickle cell disease. Its high-throughput measurement relies on the video analysis of thousands of cell motions for each blood sample to eliminate a large majority of unreliable samples (out of focus or overlapping cells) and discriminate between tank-treading and flipping motion, characterizing highly and poorly deformable cells respectively. Moreover, these videos are of different durations (from 6 to more than 100 frames). We present a two-stage end-to-end machine learning pipeline able to automatically classify cell motions in videos with a high class imbalance. By extending, comparing, and combining two state-of-the-art methods, a convolutional neural network (CNN) model and a recurrent CNN, we are able to automatically discard 97% of the unreliable cell sequences (first stage) and classify highly and poorly deformable red cell sequences with 97% accuracy and an F1-score of 0.94 (second stage). Dataset and codes are publicly released for the community.

SHAPR predicts 3D cell shapes from 2D microscopic images.

Reconstruction of shapes and sizes of three-dimensional (3D) objects from two- dimensional (2D) information is an intensely studied subject in computer vision. We here consider the level of single cells and nuclei and present a neural network-based SHApe PRediction autoencoder. For proof-of-concept, SHAPR reconstructs 3D shapes of red blood cells from single view 2D confocal microscopy images more accurately than naïve stereological models and significantly increases the feature-based prediction of red blood cell types from F1 = 79% to F1 = 87.4%. Applied to 2D images containing spheroidal aggregates of densely grown human induced pluripotent stem cells, we find that SHAPR learns fundamental shape properties of cell nuclei and allows for prediction-based morphometry. Reducing imaging time and data storage, SHAPR will help to optimize and up-scale image-based high-throughput applications for biomedicine.

Adipose microtissue-on-chip: a 3D cell culture platform for differentiation, stimulation, and proteomic analysis of human adipocytes.

Human fat tissue has evolved to serve as a major energy reserve. An imbalance between energy intake and expenditure leads to an expansion of adipose tissue. Maintenance of this energy imbalance over long periods leads to obesity and metabolic disorders such as type 2 diabetes, for which a clinical cure is not yet available. In this study, we developed a microfluidic large-scale integration chip platform to automate the formation, long-term culture, and retrieval of 3D adipose microtissues to enable longitudinal studies of adipose tissue in vitro. The chip was produced from soft-lithography molds generated by 3D-printing, which allowed scaling of pneumatic membrane valves for parallel fluid routing and thus incorporated microchannels with variable dimensions to handle 3D cell cultures with diameters of several hundred micrometers. In 32 individual fluidically accessible cell culture chambers, designed to enable the self-aggregation process of three microtissues, human adipose stem cells differentiated into mature adipocytes over a period of two weeks. Coupling mass spectrometry to the cell culture platform, we determined the minimum cell numbers required to obtain robust and complex proteomes with over 1800 identified proteins. The adipose microtissues on the chip platform were then used to periodically simulate food intake by alternating the glucose level in the cell-feeding media every 6 h over the course of one week. The proteomes of adipocytes under low/high glucose conditions exhibited unique protein profiles, confirming the technical functionality and applicability of the chip platform. Thus, our adipose tissue-on-chip in vitro model may prove useful for elucidating the molecular and functional mechanisms of adipose tissue in normal and pathological conditions, such as obesity.

Automated optimization of endoderm differentiation on chip.

Human induced pluripotent stem cells (hiPSCs) can serve as an unlimited source to rebuild organotypic tissues in vitro. Successful engineering of functional cell types and complex organ structures outside the human body requires knowledge of the chemical, temporal, and spatial microenvironment of their in vivo counterparts. Despite an increased understanding of mouse and human embryonic development, screening approaches are still required for the optimization of stem cell differentiation protocols to gain more functional mature cell types. The liver, lung, pancreas, and digestive tract originate from the endoderm germ layer. Optimization and specification of the earliest differentiation step, which is the definitive endoderm (DE), is of central importance for generating cell types of these organs because off-target cell types will propagate during month-long cultivation steps and reduce yields. Here, we developed a microfluidic large-scale integration (mLSI) chip platform for combined automated three-dimensional (3D) cell culturing and high-throughput imaging to investigate anterior/posterior patterns occurring during hiPSC differentiation into DE cells. Integration of 3D cell cultures with a diameter of 150 µm was achieved using a U-shaped pneumatic membrane valve, which was geometrically optimized and fluidically characterized. Upon parallelization of 32 fluidically individually addressable cell culture unit cells with a total of 128 3D cell cultures, complex and long-term DE differentiation protocols could be automated. Real-time bright-field imaging was used to analyze cell growth during DE differentiation, and immunofluorescence imaging on optically cleared 3D cell cultures was used to determine the DE differentiation yield. By systematically alternating transforming growth factor ß (TGF-ß) and WNT signaling agonist concentrations and temporal stimulation, we showed that even under similar DE differentiation yields, there were patterning differences in the 3D cell cultures, indicating possible differentiation differences between established DE protocols. The automated mLSI chip platform with the general analytical workflow for 3D stem cell cultures offers the optimization of in vitro generation of various cell types for cell replacement therapies.

Upscaling of pneumatic membrane valves for the integration of 3D cell cultures on chip.

Microfluidic large-scale integration (mLSI) technology enables the automation of two-dimensional (2D) cell culture processes in a highly parallel manner. Despite the wide range of biological applications of mLSI chips, manufacturing limitations of the central functional element, the pneumatic membrane valve (PMV), make the technology inaccessible for integrating tissue cultures and organoids with dimensions larger than tens of microns. In this study, we developed microtechnology processes to upscale PMVs for mLSI chips by combining 3D printing and soft lithography. Therefore, we developed a robust soft lithography protocol for the production of polydimethylsiloxane chips with PMVs from 3D-printed acrylate and wax molds. While scaled-up PMVs manufactured from acrylate-printed molds exhibited channel profiles with staircases, owing to the inherent 3D stereolithography printing process, PMVs manufactured from reflowed wax molds exhibited a semi-half-rounded channel profile. PMVs with different channel profiles showed closing pressures between 130 and 22.5 kPa, respectively. We demonstrated the functionality of the scaled-up PMVs by forming and maintaining 3D cell cultures from mouse fibroblasts (NIH3T3), human induced pluripotent stem cells (hiPSCs), and human adipose-derived adult stem cells (hASCs), with a narrow size distribution between 124 and 136 µm. Further, parallel and serial design of PMVs on an mLSI chip is used to first form and culture 3D cell cultures before fusing them within a defined flow process. Unit cell designs with upscaled PMVs enabled parallel formation, culturing, trapping, retrieval, and fusion of 3D cell cultures. Thus, the presented additive manufacturing strategy for mLSI chips will foster new developments for highly parallel 3D cell culture screening applications.

Dynamics of Individual Red Blood Cells Under Shear Flow: A Way to Discriminate Deformability Alterations.

In this work, we compared the dynamics of motion in a linear shear flow of individual red blood cells (RBCs) from healthy and pathological donors (Sickle Cell Disease (SCD) or Sickle Cell-ß-thalassemia) and of low and high densities, in a suspending medium of higher viscosity. In these conditions, at lower shear rates, biconcave discocyte-shaped RBCs present an unsteady flip-flopping motion, where the cell axis of symmetry rotates in the shear plane, rocking to and fro between an orbital angle ±Ï• observed when the cell is on its edge. We show that the evolution of ϕ depends solely on RBC density for healthy RBCs, with denser RBCs displaying lower ϕ values than the lighter ones. Typically, at a shear stress of 0.08 Pa, ϕ has values of 82 and 72° for RBCs with average densities of 1.097 and 1.115, respectively. Surprisingly, we show that SCD RBCs display the same ϕ-evolution as healthy RBCs of same density, showing that the flip-flopping behavior is unaffected by the SCD pathology. When the shear stress is increased further (above 0.1 Pa), healthy RBCs start going through a transition to a fluid-like motion, called tank-treading, where the RBC has a quasi-constant orientation relatively to the flow and the membrane rotates around the center of mass of the cell. This transition occurs at higher shear stresses (above 0.2 Pa) for denser cells. This shift toward higher stresses is even more remarkable in the case of SCD RBCs, showing that the transition to the tank-treading regime is highly dependent on the SCD pathology. Indeed, at a shear stress of 0.2 Pa, for RBCs with a density of 1.097, 100% of healthy RBCs have transited to the tank-treading regime vs. less than 50% SCD RBCs. We correlate the observed differences in dynamics to the alterations of RBC mechanical properties with regard to density and SCD pathology reported in the literature. Our results suggest that it might be possible to develop simple non-invasive assays for diagnosis purpose based on the RBC motion in shear flow and relying on this millifluidic approach.

Structural transitions and mechanochemical coupling in the nucleoprotein filament explain homology selectivity and Rad51 protein cooperativity in cellular DNA repair.

The nucleoprotein filament (NPF) is the fundamental element of homologous recombination (HR), a major mechanism for the repair of double-strand DNA breaks in the cell. The NPF is made of the damaged DNA strand surrounded by recombinase proteins, and its sensitivity to base-pairing mismatches is a crucial feature that guarantees the fidelity of the repair. The concurrent recombinases are also essential for several steps of HR. In this work, we used torque-sensitive magnetic tweezers to probe and apply mechanical constraints to single nucleoprotein filaments (NPFs). We demonstrated that the NPF undergoes structural transitions from a stretched to a compact state, and we measured the corresponding mechanochemical signatures. Using an active two-state model, we proposed a free-energy landscape for the NPF transition. Using this quantitative model, we explained both how the sensitivity of the NPF to the homology length is regulated by its structural transition and how the cooperativity of Rad51 favors selectivity to relatively long homologous sequences.

High Aspect Ratio Sub-Micrometer Channels Using Wet Etching: Application to the Dynamics of Red Blood Cell Transiting through Biomimetic Splenic Slits.

Nanoparticles delivering drugs, disseminating cancer cells, and red blood cells (RBCs) during splenic filtration must deform and pass through the sub-micrometer and high aspect ratio interstices between the endothelial cells lining blood vessels. The dynamics of passage of particles/cells through these slit-like interstices remain poorly understood because the in vitro reproduction of slits with physiological dimensions in devices compatible with optical microscopy observations requires expensive technologies. Here, novel microfluidic PDMS devices containing high aspect ratio slits with sub-micrometer width are molded on silicon masters using a simple, inexpensive, and highly flexible method combining standard UV lithography and anisotropic wet etching. These devices enabled revealing novel modes of deformations of healthy and diseased RBCs squeezing through splenic-like slits (0.6-2 × 5-10 × 1.6-11 µm3 ) under physiological interstitial pressures. At the slit exit, the cytoskeleton of spherocytic RBCs seemed to be detached from the lipid membrane whereas RBCs from healthy donors and patients with sickle cell disease exhibited peculiar tips at their front. These tips disappeared much slower in patients' cells, allowing estimating a threefold increase in RBC cytoplasmic viscosity in sickle cell disease. Measurements of time and rate of RBC sequestration in the slits allowed quantifying the massive trapping of spherocytic RBCs.

Serum pantetheinase/vanin levels regulate erythrocyte homeostasis and severity of malaria.

Tissue pantetheinase, encoded by the VNN1 gene, regulates response to stress, and previous studies have shown that VNN genes contribute to the susceptibility to malaria. Herein, we evaluated the role of pantetheinase on erythrocyte homeostasis and on the development of malaria in patients and in a new mouse model of pantetheinase insufficiency. Patients with cerebral malaria have significantly reduced levels of serum pantetheinase activity (PA). In mouse, we show that a reduction in serum PA predisposes to severe malaria, including cerebral malaria and severe anemia. Therefore, scoring pantetheinase in serum may serve as a severity marker in malaria infection. This disease triggers an acute stress in erythrocytes, which enhances cytoadherence and hemolysis. We speculated that serum pantetheinase might contribute to erythrocyte resistance to stress under homeostatic conditions. We show that mutant mice with a reduced serum PA are anemic and prone to phenylhydrazine-induced anemia. A cytofluorometric and spectroscopic analysis documented an increased frequency of erythrocytes with an autofluorescent aging phenotype. This is associated with an enhanced oxidative stress and shear stress-induced hemolysis. Red blood cell transfer and bone marrow chimera experiments show that the aging phenotype is not cell intrinsic but conferred by the environment, leading to a shortening of red blood cell half-life. Therefore, serum pantetheinase level regulates erythrocyte life span and modulates the risk of developing complicated malaria.

Probing Rad51-DNA interactions by changing DNA twist.

In eukaryotes, Rad51 protein is responsible for the recombinational repair of double-strand DNA breaks. Rad51 monomers cooperatively assemble on exonuclease-processed broken ends forming helical nucleo-protein filaments that can pair with homologous regions of sister chromatids. Homologous pairing allows the broken ends to be reunited in a complex but error-free repair process. Rad51 protein has ATPase activity but its role is poorly understood, as homologous pairing is independent of adenosine triphosphate (ATP) hydrolysis. Here we use magnetic tweezers and electron microscopy to investigate how changes of DNA twist affect the structure of Rad51-DNA complexes and how ATP hydrolysis participates in this process. We show that Rad51 protein can bind to double-stranded DNA in two different modes depending on the enforced DNA twist. The stretching mode is observed when DNA is unwound towards a helical repeat of 18.6 bp/turn, whereas a non-stretching mode is observed when DNA molecules are not permitted to change their native helical repeat. We also show that the two forms of complexes are interconvertible and that by enforcing changes of DNA twist one can induce transitions between the two forms. Our observations permit a better understanding of the role of ATP hydrolysis in Rad51-mediated homologous pairing and strand exchange.