Accelerating Drug Discovery with High-Throughput Technologies

It’s pretty much a given that most (if not all) of the low-hanging fruit in pharma has been picked—the next blockbuster drug isn’t just waiting there to be discovered. Much of what remains seems to be up against traditionally “undruggable” targets—without obvious enzymology to go after, for example—or facing enormous financial, ethical, or other hurdles.

Yet, there is still considerable unmet need—for example, some of the most successful drugs on the market still have gaps in safety, efficacy, or just in terms of areas that they’ve not been meeting, says Rachel Grimley, Senior Vice President of Drug Discovery at Cancer Research Horizons. This affords “huge opportunities to find novel drugs with novel mechanisms of action, that are either addressing areas where there aren’t any therapies or are actually enabling the next-generation therapies.”

Several often-interrelated trends—some of which are hardly new—emerge as we survey the quest for novel drugs: small volumes (miniaturization), fast (high-throughput), increased information from a given assay (high content), smart (relevant), and able to learn on the fly (using artificial intelligence, AI). Here we examine some of the ways these play out in drug discovery labs today.

Binding is not always enough

Historically, high-throughput (HT) technologies in drug development emerged to accelerate discoveries of small molecule drug candidates by enabling companies to screen libraries with millions of compounds to test their activity against selected targets. Yet because binding did not always predict success, other parameters needed to be considered, explains Jacob Tesdorpf, Life Sciences Market Segment Leader at Revvity. “The question to be answered has moved from ‘which compound binds to the target?’ to ‘which compound can enter the relevant cell type, bind to the target and revert the cell to a healthy phenotype?’.”

The field of drug discovery is moving to high information, where smaller compound libraries are screened in more diverse and more complex assays to create activity profiles able to predict a broader set of drug effects, he continues.

There is a movement for models to be more physiologically relevant. These may include samples from a wider population diversity, sometimes taking advantage of patient-derived tissue, for example. Screening using spheroids, organoid cultures, and co-cultures are gaining in usage. “That’s less high-throughput, but it means we can assess the relevance of what we’re looking at upfront to understand whether it is likely to translate or not,” Grimley says. But keeping in mind that “less high-throughput” may still mean using 384- or 1536-well plates on highly- or fully-automated platforms.

An assortment of different biochemical assays is being tailored to capture the entire mechanism of the enzyme, she adds. For example, structural kinetic relationship (SKR) assays may be embedded in place of structure-affinity relationship (SAR) assays to differentiate compounds not just on their affinity or activity, but on a range of different kinetic parameters as well.

Functional genomics

Modern drug discovery relies heavily on correlating genotype to phenotype. One way to do this is with CRISPR-based screens, or by using custom pathway-focused screens allowing elucidation of cellular targets, Tesdorpf explains.

With HT functional genomics, we’re essentially modulating every individual gene, or group of genes, on a high-throughput basis, says Grimley. “We can do that basically all in one go as opposed to in a sequential manner. It really gives you kind of a holistic picture. And potentially, if you do them in gene families, you can also see the interplay of the different knock-outs or knock-ins, to actually understand that genetic level in much faster throughput than before.”

DNA-encoded libraries

“I’m really excited about the general field of targeted degradation as a new modality of drug discovery that without question is taking off,” effuses Matt Robers, Senior Research Scientist and Group Leader at Promega. DNA-encoded libraries (DELs) are used to identify ligands that can serve as scaffolds for targeted protein degradation. “It allows you to have an insane amount of chemical diversity in your HTS compound sets, in a very scalable way, without many of the challenges of creating and purifying target inhibitor libraries.

“I’m really compelled by this new workflow of using DELs to identify binders, converting those binders into degraders. And using those degraders to target undruggable proteins within the undruggable proteomes.”

“Researchers are feverishly looking to understand the possibilities and opportunities of the protein degradation field as quickly as possible,” concurs Benedict Cross, CTO of PhoreMost.

Cross has also noticed the recent introduction of in silico protein structural prediction into drug discovery workflows, enabled by machine learning approaches and the dozens of more finessed tools associated with them, “at levels which have been quite astonishing.”

High(er) content

Many labs are embracing imaging as a part of their high-throughput screening arsenal, taking advantage of systems that allow for integrated liquid handling and high-speed kinetics, for example, for fast-response assays such as Ca²⁺-flux and mitochondrial dynamics. This allows for more live-cell applications and analyses of biological processes at unseen levels.

Such platforms also enable assays such as cell painting, “which can trace even subtle changes of the cellular phenotype in response to drug treatment using a panel of organelle stains, high content screening, a high parameter image analysis followed by machine learning-based analysis,” explains Tesdorpf.

The classics

Many different parameters can be measured in the cellular environment, where target binding and function meet phenotype, enabled by stalwart technologies such as bioluminescence resonance energy transfer (BRET). Using directed evolution to engineer luciferases to be highly thermally stable and extremely bright has allowed miniaturized—and therefore high-throughput—assays with “sensitivity that can go down to really low levels of protein,” Robers points out.

Although mass spectrometry avoids some of the limitations of label-based assays, it has had limited application in high-throughput screening due to its cost per sample, throughput, and specificity, recounts Arndt Asperger, Senior Applications Scientist at Bruker Daltonics. Such issues can be overcome by combining the “speed and robustness of matrix-assisted laser desorption ionization (MALDI) sampling with trapped ion mobility spectrometry (TIMS) for rapid separation of isobars and isomers, and accurate mass detection using high-resolution quadrupole time-of-flight mass spectrometry (QTOF MS).” In comparison to chromatographic separation, TIMS is rapid enough to be incorporated into HTS workflows, he says, enabling the rapid and low-cost identification of high-quality leads.

With the emergence of biologics as a pharmaceutical tour-de-force, this article would be remiss not to mention a novel picodroplet-based microfluidic system for screening up to 40 million cells per day. Cyto-Mine “enables antibody discovery via the screening of entire genetic repertoires from cells collected from an immunized animal, and can screen engineered cells for the highest expressors in cell line development,” explains Frank Craig, CEO of Cyto-Mine’s maker Sphere Fluidics.

High-throughput screening has become much more than just massive-scale binding assays, embracing technologies from fields as diverse as high-content imaging and analysis, microfluidics, functional genomics, and mass spectrometry. All in the name of finding that small molecule or biologic that meets that unsolvable unmet healthcare need.