Chemoproteomics in Phenotypic Drug Discovery

Phenotypic drug discovery (PDD) stands in contrast to target-based drug discovery (TDD) – the former seeks novel compounds through observing the alterations they impose on disease pathophysiology. The latter identifies new drugs based on known mechanisms of action and interactions with drug targets. PDD is often more useful in discovering first-in-class medications: penicillin, the famous antibacterial, and zidovudine, the first treatment for HIV, were both discovered phenotypically. But perfecting and refining such compounds will inevitably require a better understanding of how they actually work at the molecular level – and this is where chemoproteomics becomes invaluable. 

The Value of Phenotypic Screening

TDD has dominated drug discovery pipelines for decades, providing streamlined development cycles and more predictable milestones. Despite this, an influential review comparing both PDD and TDD found that phenotypic approaches yielded 28 first-in-class drugs, compared to 17 for TDD – at a time when interest in PDD was not particularly strong. This work by Swinney & Anthony revitalized interest in phenotypic drug discovery, which is now considered part of mainstream pharmacology. Phenotypic screening is typically done in vitro on cell-based assays, or in vivo on animal models. Advancements in the field of cell-based screening, such as induced Pluripotent Stem Cells (iPSC) as well as gene editing using CRISPR, have also provided traction to PDD – improving the toolkit it can use to evaluate disease modulation.

Chemoproteomic Screens

Chemoproteomics refers to a vast array of methods used to characterize the interactions between proteins – which are typically the drug targets, and small molecules: the drug candidates. Chemoproteomics techniques are natural companions for PDD, as they can illuminate the molecular mechanism of action of what is otherwise a relatively target-agnostic approach for drug discovery. Applications of chemoproteomics have recently highlighted the potential of the field to expand the druggable proportion of the proteome. This is particularly significant for non-traditional treatment modalities which are not restricted by known binding, active or allosteric sites on proteins – such as Proteolysis Targeting Chimeras (PROTAC).

The Intersection of PDD with Chemoproteomics

Chemoproteomics has led to intriguing discoveries from phenotypic screens throughout recent times. In a 2006 study, affinity chromatography revealed that results observed during phenotypic assays of stem cells can be explained through interactions with multiple target proteins rather than just one – which is typically what TDD aims for. There are a number of novel methods in chemoproteomics which can be used to deconvolute targets in native environments, and are therefore less likely to produce artifacts. These include Drug Affinity Responsive Target Target Stability (DARTS), Thermal Shift Profiling, Stability of Proteins from Rates of Oxidation (SPROX), and others.

Additionally, while phenotypic screens may often be employed to expand druggable space, chemoproteomics has often revealed the interactions between novel compounds used in PDD and druggable targets. Such insights expand our knowledge of the druggable proteome, and illustrate how we need to widen our efforts even in the better understood, druggable, space.. This is indicative of just how much we have yet to learn even about the parts of the proteome we consider to be tractable.

The Great Repurposing

Phenotypic Drug Discovery does not merely focus on novel drugs – instead, the repurposing of drugs for new indications is also one of its strengths. This has been acutely demonstrated throughout the SARS-CoV-2 pandemic, where the pharma industry scrambled to discover new therapies and test old treatments against a new foe. Many antivirals approved for use in patients with COVID-19, such as remdesivir, ritonavir, and others, were previously known and had shown activity against the novel coronavirus. While the molecular modes of action of these compounds were understood – at least in the context of how they act upon the virus, their full effects on the body were less characterized. 

A study aiming to uncover these was done using Cellular Thermal Shift Assay Mass Spectrometry (CETSA MS), the first of its kind to use this chemoproteomics method to compare multiple antivirals. The study uncovered multiple protein interactions, the most notable of which was remdesivir’s potential to destabilize the TRIP13 protein. TRIP13 overexpression is implicated in multiple cancer indications, such as squamous head and neck carcinomas. The combination of PDD assays with cutting-edge chemoproteomics methods holds great promise for expanding our horizons – both in novel drugs, and what we know of current pharmaceuticals.

The Technological Revolutions

Chemoproteomics assays are often time-consuming and cost-prohibitive – particularly in high-throughput scenarios. This has given rise to multiple computational approaches, particularly molecular docking – which aims to simulate how small molecules will interact with proteins. Molecular docking can produce powerful pharmacophore models, which can reduce the resource commitment for drug development. A recent study showed potential for repurposing existing drugs for mental health indications through the use of computational chemoproteomic models.

Similarly, PDD also faces resource issues – particularly over the front-loaded expense and complexity of PDD investigations. Considering that early drug discovery is accompanied by high failure rates, it is unsurprising that TDD has remained the dominant form of studying new drugs. Novel developments in the space of Artificial Intelligence hold immense promise for solving these issues across both fields. Exploratory studies showed the potential of machine learning chemoproteomics models to predict binding site concentrations and targets in whole-cell lysates. 

New methods promise to increase the sensitivity of discovering new drug targets within phenotypic screens, a limitation of current chemoproteomics methods. As we see AI penetrate further in pharma, particularly the drug discovery space, it is inevitable that it will galvanize further chemoproteomics efforts in accelerating PDD. PDD has shown unique advantages in innovating drugs for unmet needs, and further improvements in the method promise to revolutionize both the druggable and undruggable proteome.

Nick Zoukas, Former Editor, PharmaFEATURES

Join Proventa International’s San Francisco Medicinal Chemistry Strategy Meeting to hear the latest news on computational approaches for chemoproteomics in Phenotypic Drug Discovery. Participate in closed roundtable discussions on the leading topics from across the field, such as Protein Degradation, innovations in protein structure prediction, and others!