Phenotypic Versus Target-Based Screening for Drug Discovery

Target-based screening has been the method of choice in drug discovery for the past two decades, but phenotypic screening is having something of a renaissance. We look at the advantages and applications of both.

What does each screening approach involve?

Many of today’s drugs were discovered through phenotypic screening – testing a molecule in cells, isolated tissues or organs, or animals, to see whether it exerts the desired effects. The actual mechanism by which the drug exerts its effect – its target – was often determined later, and sometimes not for many years.

Historically, phenotypic screening has a slight advantage when it comes to identifying first-in-class drugs, while target-based screening has yielded more best-in-class drugs.¹ This has been attributed to the lack of bias when it comes to identifying a drug’s mechanism of action. But phenotypic assays are also more favorable with respect to identifying cell active compounds, says Professor Elizabeth Sharlow from the University of Virginia School of Medicine, Virginia, USA.

“Phenotypic assays are challenging because of the need for, often, complicated downstream target deconvolution methodologies,” she says. “And they are also, in some instances, more time consuming to implement which in the long term may impact throughput.  All of this needs to be factored into the overall screening paradigm and cost of the screening strategy.”

This lack of throughput, accompanied by the revolution in genomics, has led to drug companies and academia adopting a ‘target-first’ approach for many years, where a molecule known to be important in a disease process is used to screen vast compound libraries in the search for a ‘hit’ – a candidate drug. One advantage of such an approach, is that you can screen many millions of drug-like molecules, knowing that if you get a hit, you already have a candidate that is close to becoming a potential drug.

“Target-based assays, in general, are less time consuming to implement but sometimes are challenged by standard readouts such as enzymatic activity,” points out Professor Sharlow. “Then there’s more sophisticated target-based assays such as those based on protein-protein interactions, which, while more physiologically interesting, are much more complicated to implement.

Ultimately, both screening strategies are valuable and necessary for chemical probe and/or drug discovery, she says, and usually, a particular screening strategy is dependent upon the resources you have available.

However, from a drug discovery point of view, it seems there’s a slight resurgence in phenotypic screening.

The sweet spot – a combined targeted phenotypic approach

“Doing drug discovery screens within the cellular context already answers a few of the questions that bedevil the drug discovery process later on, such as ‘how do I make this more soluble?’, or ‘how do I get this into cells?’” says Dr Mike Howell, Head of Screening at the Francis Crick Institute, London, UK. “If you’re a medicinal chemist, you’re probably correctly of the opinion that you can use your magic to make things that are not very soluble or are not very good at getting into cells, better at getting into cells. But there’s a feeling that if you start off with a good cell-based assay then you’ve leapt some of those hurdles inherently by design.”

The sweet spot, says Howell, is the way in which the two approaches are increasingly being combined. “Many people will want a screen to look at the activity or localization of a protein within the cellular context – so it is both target-based and phenotypic to an extent.”

At the Francis Crick Institute, Howell explains that they err towards phenotypic approaches, in that they use predominantly cell-based assays, which have all the cellular influences on a particular process, but they will be primarily studying just one aspect of that process, such as the phosphorylation of a downstream target. “We use high-content image analysis of cell-based assays. It means you can measure multiple cellular read-outs both to quantify the primary target response and classify these responses using additional phenotype data on the off chance that some of those additional features tell you more about the process you were intending to study”

Phenotypic screening – the next level

This move towards combining the two approaches has been enabled by advances in technology that can more subtly manipulate molecular processes in cells. “We’re living in a time where we’ve gone through one way of knocking down genes using siRNA and now everyone’s doing it with CRISPR. An interesting point to note is the different results people are seeing. It just highlights that there isn’t a perfect technique, because they each ask different questions in different ways,” says Howell.

CRISPR has opened up a whole range of questions that weren’t possible to explore before, says Howell, such as following chromosome loss or mis-segregation in real-time. Simple over-expression screens that were once difficult because they required transfecting cDNAs and overexpressing them at levels much higher than necessary, are now much easier to control. The availability and use of a range of novel CRISPR techniques could now enable the generation of cellular models amenable to traditional high-throughput screening but that much more closely model the diseased cell state. I’m not saying it’s easy, but these things are now possible.”

Are biological assays always the answer?

So, what will take these screening approaches to the next level?  Sharlow cites the use of more physiologically relevant assay systems such as induced pluripotent stem cells and organoids as exciting, as well as a willingness to screen at a smaller scale (versus screening millions of chemotypes). “This allows for the use of sophisticated assay readouts including those in 3D configurations and could contribute to a higher translation of technologies to the clinic.”

In disease processes like cancer, incorporating the influence of the immune system in to the screen will be incredibly important, says Howell. “The next generation of phenotypic screening really has to include some other ways of having multicellular environments, 3D environments, the influence of other tissues, the influence of other systems.”

But there is also a risk of making screens too complex when it’s not necessary, Howell warns: “There’s this mantra of people wanting to do the most authentic assays, or as close to real life as possible. It’s always worth questioning whether there is the evidence to support the need to do that, or whether we are just doing things that feel virtuous by being difficult. To some extent it depends on what the information is that you’re trying to get at. If you’re trying to make a drug for a human being I think a more complex experiment may well be more informative, ultimately. If you’re trying to discover something about your favourite sub-cellular system that you use to do your research, then maybe it isn’t.”

The answer is that a comprehensive screening strategy will incorporate both targeted and phenotypic assays, with one format designated as the primary screen and the other as a secondary or follow up assay.

“You can always argue that the information will be richer from a phenotypic assay,” concludes Howell, “but the world still needs to make specific molecules that bind to specific proteins. The answer isn’t to use one or the other – you need both.”

Reference
1. Swinney, D. C. (2013). Phenotypic vs. Target-Based Drug Discovery for First-in-Class Medicines. Clinical Pharmacology & Therapeutics, 93(4), 299-301. doi:10.1038/clpt.2012.236