High-Throughput Screening Techniques To Improve Cellular Drug Discovery

Chronic diseases, such as cancer and diabetes, are a leading cause of death and disability globally. To treat these diseases, efforts for drug discovery have expanded beyond the traditional approaches of small molecules and antibodies to include cells. Among the various cell-based drug products, stem cells and immune cells rank as the most popular cell types to be investigated in clinical trials and approved by regulators.

Nevertheless, with advances in sequencing technologies, it has been discovered that even in a population of cells that are of the same type, there is heterogeneity in functions including cell-mediated cytotoxicity and paracrine secretion profiles. These variations in functions are an opportunity to discover more effective cell types as drugs to treat diseases. For instance, the ability to identify immune cells that are able to secrete more cytotoxic granzymes could make cancer immunotherapy more effective. Similarly, if we can identify mesenchymal stem cells that can secrete more growth factors, it will be highly beneficial for regenerative medicine.

For the last decade, microfluidic technologies have been extremely useful for high-throughput screening, including the selection of cells with desirable properties. They have also been integrated with artificial intelligence (AI)-based technologies for more accurate cell sorting. Beyond droplet microfluidics, where cells are sorted based on their reactions in a droplet, there is also a new movement termed “lab on a particle”. Particles are even smaller than a droplet, meaning even higher throughput and lower processing volumes. Here, we will describe the use of microfluidic methods for high-throughput screening of cell-based drug products to enhance drug discovery efforts.

Finding immune cells that kill better

Immune cell therapy, such as chimeric antigen receptor (CAR) T-cell therapy has revolutionized cancer treatment. Currently, there are six CAR T-cell therapy products approved by the US Food and Drug Administration (FDA), and more are expected to be approved in the next few years.

Immune effector cells such as cytotoxic T cells and natural killer cells are another option for cell therapies. These can mediate the killing and lysis of target cancer cells through binding of receptors in immune cells to ligands on cancer cells. The ability to identify immune effector cells that have preferential cytotoxic capabilities will guide the selection and study of these cell subtypes and improve engineering efforts to develop more effective therapies.

However, no method has enabled the isolation of immune effector cells with the best killing profiles. While techniques such as flow cytometry and electrical impedance can measure the loss of cancer cell viability, they measure cytotoxic killing efficacy at the bulk level. To tackle this problem, Luah and colleagues developed a method called proximity affinity intracellular transfer identification of killer cells (or PAINTKiller) to identify and isolate effector immune cells capable of lysing target cells.

The concept of PAINTKiller is an elegant one. When a target cell is lysed, it will release proteins that “paint” the killer or effector immune cells in close proximity, hence enabling the killer cells to be identified. This method does not require genetic modification and can be universally applied to any type of effector and target cell types.

Luah et al. postulated that for their concept to work, they need a universal, high-abundance intracellular fluorescence molecule for direct visualization, and a way to bind effector cells with the fluorescence molecule for subsequent cell isolation. They chose carboxyfluorescein succinimidyl ester (CFSE) as the cell permeable fluorescent cell staining dye. The effector cell is labelled with a bispecific antibody containing anti-CD45 that binds to immune cells, and anti-FITC that binds to CSFE.

As a proof of concept, the team analyzed how NK-92MI effector cells kill K562, a human erythroleukemic cell line. They found an increasing number of CSFE+ natural killer cells over time, demonstrating that CSFE-labelled proteins are released and successfully bound to their effector killer cells. To understand whether bystander natural killer cells could also be labelled, they team optimized the cell concentration and duration of co-culture. Although bystander natural killer cells could also be labelled with CSFE, it was negligible, indicating the excellent specificity and low background of PAINTKiller assay.

With these promising results, the team went on to use primary human-derived natural killer cells (as effector cells), T cells (as bystander cells) and K562 cells. The team found that based on the intensity change in CSFE+ cells, primary cells killed K562 cells more effectively than NK-92MI. While 70–80% of natural killer cells became CSFE+, <5% of T cells were CSFE+, indicating high specificity of the assay.

Next, to characterize whether the cytotoxicity functions of natural killer cells are correlated to their cytokine secretions, the team also labelled primary natural killer cells with anti-interferon-gamma (IFN-γ) antibodies. While cell lysis was a rapid event that occurred within an hour, cytokine secretion became more significant only after the third hour. This is aligned with the availability of pre-stored lytic granules to mediate cell lysis and the need for transcription and translation before cytokine secretions. Only about 8.6% of the natural killer cells are multifunctional effectors capable to lyse target cells and secrete cytokines.

The authors acknowledge that the PAINTKiller assay has been validated for only one kind of target cell lysis now but, in principle, can be used to measure and distinguish the killer of multiple target cells, such as tagging target cells with distinct fluorescent proteins via genetic modification. The choice of labelling florescent molecule CSFE is also diluted when cells divide and may not be as suitable for long-term experiments. The killing specificities of multiple effector and target cells can then be determined simultaneously and may be of interest to many high-throughput screening applications. Overall, this study revealed significant heterogeneity in the cytotoxic capability of immune effector cells and that the integration of single-cell sorting and sequencing as well as PAINTKiller can reveal molecular patterns associated with cell cytotoxicity to improve cell manufacturing and therapy.

“We are actively expanding the PAINTKiller assay in two key areas,” said Dr. Cheow Lih Feng, assistant professor at the National University of Singapore and corresponding author of the study. “First, we are adapting the assay to detect in vivo killing activities, enabling a deeper understanding of immune responses in physiological environments.”

“Second, we are enhancing its capability to simultaneously measure the killing of multiple target cell types, broadening its application for screening and functional analysis,” he continued. “PAINTKiller-sorted and culture-expanded cells have demonstrated sustained superior cytotoxicity, positioning the assay as a valuable tool for cell therapy workflows. Its potential for functional enrichment of immune effector cells could significantly enhance the quality and efficacy of cell manufacturing workflow.”

Identifying the best stem cells for regenerative medicine

Mesenchymal stem cells can be used in regenerative medicine to replace dead cells and promote tissue repair and it has been found that paracrine signaling by stem cells is the main mechanism in which they elicit regenerative properties. They secrete extracellular vesicles that play crucial roles in intercellular communication and possess anti-inflammatory properties to treat diseases such as osteoarthritis and myocardial damage. Despite this knowledge, there is no available method to isolate single mesenchymal stem cells based on their ability to produce and secrete extracellular vesicles, as current methods focus on bulk enrichment that mask the heterogeneity of single stem cells.

A study by Koo et al. aimed to design a method to better isolate mesenchymal stem cells. The researchers developed cavity-containing hydrogel particles called nanovials. This uses a microfluidic device that generates uniform water-in-oil emulsions to create millions of monodisperse nanovials with an inner cavity coated with biotinylated gelatin. Mesenchymal stem cells adhere onto the gelatin coating and secreted extracellular vesicles are captured on the nanovial surface with conjugated antibodies. The captured extracellular vesicles are then labelled with fluorescent antibodies against other surface markers of extracellular vesicles and the live secreting cells can be sorted using a fluorescence-activated cell sorter for downstream applications.

After validating that mesenchymal stem cell-derived extracellular vesicles have high surface expressions of CD63 AND CD9, the team used the corresponding antibodies to capture and label the extracellular vesicles. Significant heterogeneity was observed in the ability of mesenchymal stem cells to secrete extracellular vesicles. Using flow cytometry, the team sorted out cells into low, medium and high secretors to study how persistent is their ability to secrete extracellular vesicles. The high secretors consistently proliferated faster and their ability to differentiate into adipogenic, osteogenic and chondrogenic lineages were unaffected, even when they were cultured in nanovials.

Using the sorted samples, the team quantified their differential gene expressions according to extracellular vesicle biogenesis gene ontology signature. They found that the extracellular vesicle biogenesis signal was enriched among high-secretor–dominant clusters compared to low-secretor clusters. Genes that mediate vesicular trafficking and endosomal recycling such as TSG101 and HGS were enriched in high-secretor–dominant clusters.

Next, the team performed a head-to-head comparison of mesenchymal stem cells. They compared the effects of high- versus low-secretor clusters in cardiac repair using a mouse model of myocardial infarction, after verifying that the former cluster produced extracellular vesicles with superior paracrine functions. On day 28, they observed that high-secretor mesenchymal stem cells were able to restore left ventricular ejection fraction (LVEF) and left ventricular fractional shortening (LVFS) significantly better than wild-type heterogeneous and low-secretor mesenchymal stem cells. The authors concluded that this is likely attributed to the difference in extracellular vesicle-mediated paracrine activity.

Overall, this work suggests that by selecting stem cells with greater ability to produce more extracellular vesicles over multiple generations, it can improve therapeutic outcomes. In particular, when enriched populations of high-secretor stem cells were used, it led to greater recovery in functions and structural features after a myocardial injury. This nanovial approach enables high-throughput screening and sorting of mesenchymal stem cells to reduce heterogeneity in extracellular vesicle production and improve batch-to-batch consistency. The isolated stem cells may also be cryopreserved for master cell banks to establish cell clones based on increased extracellular vesicle productivity.

“Now that we can sort cells based on functional properties, such as the secretion of extracellular vesicles, the hope is we can define cell therapies more precisely, and not by surrogate markers, leading to better therapies,” said Dr. Dino Di Carlo, professor and department chair at the University of California, Los Angeles, and corresponding author of this paper.

“The FDA group that regulates cell and gene therapies asked me to present to them about nanovial technology and functional cell selection,” he continued. “Since we have more precise tools, future cell therapy regulations and final products may be more robust and repeatable. With our collaborator, Ke Cheng, we have continued work that is expected to be funded to advance the approach to large animal studies, and towards the clinic.”

High-throughput screening methods are a growing trend, as there is significant heterogeneity in biological processes from target binding, target killing and secretory profiles. The ability to isolate single cells in a microfluidic droplet or vial before studying their functions and connecting it to their transcriptomics and proteomics at the single-cell level will change how cell-based drugs can be further enhanced in therapeutic and safety outcomes.