The Dawn of a New Era in Drug Discovery? Drug Screening and the Increasing Biological Complexity of Testing Models

Abstract

Whole-organism phenotype screening and complex in vitro model technology increase the likelihood in identifying successful lead compounds and lower drug attrition rates at later and more expensive stages of the drug discovery process.

Developments in in vitro high-throughput drug screening continues to increase the efficiency at which potential drug candidates are identified. The adoption of in silico technologies for drug binding have also enhanced the ability of wet-laboratory experiments to identify compounds that elicit desired biological responses. However, further downstream in the drug discovery process are clinical trials that pose challenges still difficult for researchers to overcome. Indeed, most lead compounds identified during the early stages of the drug discovery process fail to make it to market. This failure can usually be attributed to compounds’ lack of efficacy in humans, despite their promising performance in vitro and in traditional animal models such as rodents. Additionally, lead compounds often show unexpected toxicity in humans.¹

A player gaining traction in addressing these issues encountered during the drug discovery process is the zebrafish (Danio rerio), an organism featured at the recent ACS Fall 2021 Meeting session titled “Thinking Outside the Well: Novel Assays in Drug Discovery and Development” hosted by the Division of Chemical Toxicology. The use of zebrafish for whole-organism phenotypic screening allows for preliminary toxicity studies to occur simultaneously while lead compounds are identified. Furthermore, zebrafish and humans share a great degree of genomic similarity. Dr. Randall Peterson (University of Utah), the opener of the session, also highlighted the prolific reproduction behavior of zebrafish which allows for relatively high-throughput assays. Notably, phenotypic screens for behavior disorders that are traditionally performed through labor-intensive and time-consuming rodent model studies have the potential to be performed more efficiently with zebrafish. Combining the advantages of zebrafish model studies with recent advancements in generating disease models in whole organisms, the Peterson laboratory, among others, is working to develop high-throughput screening methods for drugs that affect social behavior. These works are motivated by the need to understand the underlying causes of neurological phenotypes such as autism and schizophrenia.²

Several compounds identified from zebrafish phenotype screenings have made it to preclinical or clinical trial phases. As for the compounds that are not identified as “hits” in the zebrafish assays, is it because the molecule indeed imparts no effect on the zebrafish? Or rather is it because the compound was never absorbed into the zebrafish’s system? This is an important question that Dr. Donna Huryn (University of Pittsburgh) brought to the session, and one in which her laboratory is interested. The consideration on absorption invokes the famous Lipinski’s rule of five which offers guidelines on what makes a compound “drug-like”. This evaluation is based on characteristics of molecular weight (MW < 500 Da), number of hydrogen-bond donors (HBD ≤ 5), number of hydrogen-bond acceptors (HBA ≤ 10), and lipophilicity (log P ≤ 5). However, the metrics of Lipinski’s rule of five are, naturally, based on human physiology. This begs the question of whether a new set of rules needs to be developed for zebrafish. As a start to an answer, the Huryn laboratory surveyed a library of compounds that have been identified as “hits” in zebrafish phenotype screening studies and noted the characteristics considered in Lipinski’s rule of five. The metrics of the “zebrafish Lipinski’s rule of five,” while showing some overlap to the original rule of five, showed some statistically significant deviations from its predecessor (MW ≤ 500, HBD ≤ 3, HBA ≤ 7, and log P ≤ 5.3).³

With animal-testing comes the consideration of ethical practices and concern on the reproducibility of experimental results. These ethical and experimental shortcomings of animal testing are a strong motivator for the development of technology coined as complex in vitro models (CIVMs), another point of focus of the session. In essence, CIVMs are artificially constructed cellular structures that are amenable for drug discovery experiments. Importantly, CIVMs are housed in vessels such as well-plates to allow for screening for lead compounds with relative efficiency. The four major classes of CIVMs, listed in increasing biological complexity, are spheroids, organoids, bioengineered tissues, and organ-on-a-chip systems. A key advantage of CIVMs is their three-dimensionality (as opposed to the “two-dimensionality” of traditional cell-culture models) that allows for increased biological mimicry and complexity. As noted by the session’s closing speaker, Dr. Marc Ferrer (National Center for Advancing Translational Sciences), CIVMs in theory are customizable to model any normal or diseased physiological state.⁴ One example of a CIVM presented during this session was a liver-on-a-chip developed by Dr. Thalita Zanoni and co-workers (Charles River), which importantly demonstrated that CIVMs have the potential to even mimic metabolic processes.⁵

Just as with whole-organism screening, CIVM technology serves to further close the gap between the results of early stage drug discovery processes and clinical studies. Having an increasingly robust system in place for the identification of lead compounds could reduce drug attrition rates at the later and more costly stages of drug discovery. Because these methodologies are arguably still in their infancy, the approaches to establish more physiologically complex testing platforms, and their metrics for success, are varied across different laboratories. In his writing and presentations, Dr. Ferrer echoes the scientific community’s hope that one day a standardized practice for these advanced drug screening technologies will be established, allowing for these robust methodologies to be adopted for general use.