3D cell culture models are important tools in translational research but have been out of reach for high-throughput screening due to complexity, requirement of large cell numbers and inadequate standardization. Microfluidics and culture model miniaturization technologies could overcome these challenges. Here, we present a high throughput workflow to produce and characterize the formation of miniaturized spheroids using deep learning. We train a convolutional neural network (CNN) for cell ensemble morphology classification for droplet microfluidic minispheroid production, benchmark it against more conventional image analysis, and characterize minispheroid assembly determining optimal surfactant concentrations and incubation times for minispheroid production for three cell lines with different spheroid formation properties. Notably, this format is compatible with large-scale spheroid production and screening. The presented workflow and CNN offer a template for large scale minispheroid production and analysis and can be extended and re-trained to characterize morphological responses in spheroids to additives, culture conditions and large drug libraries.

Droplet microfluidics utilize a monodisperse water-in-oil emulsion, with an expanding toolbox offering a wide variety of operations on a range of droplet sizes at high throughput. However, translation of these capabilities into applications for non-expert laboratories to fully harness the inherent potential of microscale manipulations is woefully trailing behind. One major obstacle is that droplet microfluidic setups often rely on custom fabricated devices, costly liquid actuators, and are not easily set up and operated by non-specialists. This impedes wider adoption of droplet technologies in, e.g., the life sciences. Here, we demonstrate an easy-to-use minimal droplet production setup with a small footprint, built exclusively from inexpensive commercially sourced parts, powered and controlled by a laptop. We characterize the components of the system and demonstrate production of droplets ranging in volume from 3 to 21 nL in a single microfluidic device. Furthermore, we describe the dynamic tuning of droplet composition. Finally, we demonstrate the production of droplet-templated cell spheroids from primary cells, where the mobility and simplicity of the setup enables its use within a biosafety cabinet. Taken together, we believe this minimal droplet setup is ideal to drive broad adoption of droplet microfluidics technology.

Here, we build, train and apply an automated imaging and deep learning image analysis pipeline for optimization of assembly and culture conditions for miniaturized 3D cell spheroids production in microfluidic droplets. Miniaturization of spheroids, rapid assembly optimization and automated spheroid analysis would amount to a paradigm shift in early drug development. We expand an automated ultra-high-throughput workflow for minispheroid production in microfluidic droplets by training a convolutional neural network (CNN) model for automated minispheroid morphology assessment and classification. The CNN classifier was used to characterize minispheroid assembly of three different cell lines for a range of incubation times and surfactant concentrations.

Here we present a droplet microfluidics workflow for cell factory screening by yield of an intracellular product from isogenic microcolonies, i.e. minimal cell populations, encapsulated in picoliter droplets. This allows us to utilize all the benefits of droplet microfluidic screening in terms of throughput, but based on the signal from a population average, rather than the noisy single cell signal. We demonstrate microcolony sorting by integrated droplet fluorescence area of encapsulated E. coli, optimize triglyceride (TG) microcolony assay in droplets and apply the microcolony screening concept to analyze triglyceride (TG) production in S. cerevisiae.

3D cell culture models are an important tool in translational research but have been out of reach for high-throughput screening due to complexity, requirement of large cell numbers and inadequate standardization. Here, we present a high-throughput workflow to produce and characterize the formation of miniaturized spheroids using deep learning. We train a convolutional neural network (CNN) for cell ensemble morphology classification, benchmark it against more conventional image analysis, and characterize spheroid assembly determining optimal surfactant concentrations and incubation times for spheroid production for three cell lines with different spheroid formation properties. Notably, this format is compatible with large-scale spheroid production and screening. The presented workflow and CNN offer a template for large scale minispheroid production and analysis and can be extended and re-trained to characterize morphological responses in spheroids to additives, culture conditions and large drug libraries.

Cellular heterogeneity in isogenic cell populations is a major obstacle for single-cell screening campaigns, as the phenotype of individual cells might differ drastically from the mean, leading to large overlaps between productivity assessments of populations. At the other end of the spectrum, isogenic bulk assays provide a more accurate picture of a strain’s capacity at production scale, but suffers from low throughput and high reagent consumption.

Here, we present a screening format for cell factory variant libraries, aiming at combining the advantages of single-cell screening and bulk assay formats. Using microfluidic droplets, we compartmentalize yeast cell producer candidates, culture them to form isogenic microcolonies and sort colonies at higher throughput than bulk experiments to assess the genetic potential more accurately than in a single-cell screening format. To this end, we developed a fluorescence area-based sorting method that integrates the fluorescence signal from the entire fluorescence profile of a droplet and bases the sorting decision on that integrated fluorescence area. We validate the concept by sorting droplet microcolonies of fluorescent protein expressing Escherichia coli. Finally, we successfully sorted encapsulated iso-genic microcolonies of a low-producing and a high-producing strain of Saccharomyces cerevisiae by Triacylglycerol (TAG) production at 220 Hz, enriching the high-producing strain 4.45-fold.