What is phenotypic screening? Phenotypic assays measure a phenotype in a physiological system. The term ‘phenotypic assay’ includes all preclinical assay formats that use physiological systems, eg animals, cells and biochemical pathways.

Phenotypic assays make few assumptions as to the molecular details of how the system works and provide an empirical method to probe effects in physiological systems. The phenotype most relevant to the practice of drug discovery is a phenotype that directly translates to the clinical disease (translational biomarker).

I have used the term ‘phenotypic screening’ in previous reports to include any screening where the molecular mechanism of action (MMOA) is not assumed1,4,6,7. In this context the phenotypic screening is a synonym for empirical screening. Using this definition phenotypic screening includes all screening that does not involve screening against an isolated target.

It is important to recognise that phenotypic assays have always played an important role in drug discovery. Much of early pharmacology and drug discovery was based on phenotypic assays. Phenotypic assays were used to identify leads that provided the desired efficacy. In his nobel lecture entitled ‘Selective inhibitors of dihydrofolate reductase’ George H. Hitchings Jr stated: “Those early, untargeted studies led to the development of useful drugs for a wide variety of diseases and has justified our belief that this approach to drug discovery is more fruitful than narrow targeting.”8

Drug Discovery
For the purpose of this work, drug discovery is defined as the identification of a new medicine candidate for testing in humans for efficacy and safety. The clinical testing is the development process. A drug is discovered when nominated for clinical trials; theoretically from this point it should fail or succeed based on its own merit. Its merit will be determined in the clinical development programme and biopharmaceutics evaluations. One key to increasing overall drug discovery productivity is to increase the success rate of clinical candidates. The practice of drug discovery and development is an iterative cycle of learning and confirming (Figure 1). The process to discover a new medicine is driven by an unmet medical need. It requires some level of understanding of the disease and the corresponding physiology that can be used to design and implement a drug discovery strategy. Ideally the understanding provides a translational biomarker that can be used to align preclinical drug discovery with clinical development. Examples of biomarkers include viral load for a virus infection or blood glucose levels for diabetes. Knowledge of the pathophysiology of the disease, genomic links and the identity of physiological regulators contribute to understanding ways to correct a disease phenotype. In practice this knowledge is used to establish assays that will identify drug leads. The leads are generally small molecules or biologics including recombinant proteins and monoclonal antibodies. The leads are then optimised to drug candidates that have sufficient efficacy, safety and biopharmaceutics properties for testing in human clinical studies. The clinical candidates then undergo clinical development and, when successful, will be approved as a new molecular entity (NME) by regulatory agencies such as the US FDA. Iterative learning and confirming happens at every stage of the process (Figure 1).

The fewer the iterations required to identify a compound suitable for registration as a new medicine, the more efficient the process. Rarely does the process occur in one iteration. For first-in-class medicines, where the available knowledge is incomplete there will be more failures, requiring more iterations. This is less of an issue with followers. Drug discovery has evolved different strategies based on the available knowledge. One approach starts with a hypothesis related to the disease. Historically this hypothesis was related to a phenotype such as clearing an infection by killing the infectious organism or blocking experimentally induced seizures in animals as a phenotype for epilepsy. More recently an increased understanding of the molecular basis of disease has led to a molecular approach to drug discovery, in which genetics and other molecular sciences identify a molecular hypothesis.

Consequently, current small molecule drug discovery strategies are polarised into two major types; target-based drug discovery (TDD) and phenotypic drug discovery (PDD). The focus of TDD is a gene-product, known as a target. Molecular and chemical knowledge are used to investigate specific molecular hypotheses. The focus of PDD is assays that measure a clinically meaningful phenotype in a physiological relevant system such as an animal or cell; the assays do not require prior understanding of the molecular mechanism of action (MMOA). Drug discovery with biologics is primarily a target informed approach. Successful drug discovery uses all strategies but there is debate and preferences on which is used first and under what circumstances.

What is the value of phenotypic screening to drug discovery?
Identification of MMOA: A major challenge in the identification of safe medicines is to identify molecular mechanisms of action (MMOAs) that provide both sufficient efficacy and safety1,4,5. Effective medicines work through MMOAs that provide a tolerable and useful therapeutic index. MMOAs can be thought of as ‘pharmacological hot spots’. MMOAs include the target and the biochemical mechanisms through which the target:drug complex interact with physiology to provide a safe and efficacious response. Due to the dynamic complexity of physiology both at the molecular and systems level it is difficult to a priori predict the exact interactions and molecules that will elicit a safe, therapeutically useful response. Knowing the parts of an efficient machine – a watch, an automobile or a computer – is not enough to describe how it works. The parts must collaborate in precise ways to provide accurate time, reliable transportation or processed information. MMOA is used to describe the complexities of how medicines interact with biology to provide an effective and safe response. A value of phenotypic assays is to identify first-inclass medicines and their respective MMOAs, since phenotypic assays do not require a priori identification of the target and MMOA. The empirical, phenotypic approach enables drug discovery to proceed with incomplete information.

In our earlier work we noted that in the targetbased approach, drug discovery is generally hypothesis-driven, and in this case there are at least three hypotheses that must be correct to result in a new drug1. The first hypothesis applies to all discovery approaches: the hypothesis that activity in the preclinical screens used to select a drug candidate will translate effectively into clinically meaningful activity in patients. The other two related hypotheses are that the target selected is important in human disease and that the MMOA of drug candidates at the target in question is one that is capable of achieving the desired biological response. Successful first-in-class, target-based drug discovery requires the time and resources to investigate all three hypotheses and, in particular, the importance of hypothesis-testing to identify an appropriate MMOA may be an underappreciated challenge, that, if neglected, could contribute to increased attrition for such approaches. In other words, it is clearly difficult to rationally identify the specific molecular interactions from all the potential dynamic molecular interactions that will contribute to an optimal MMOA. Thus the key biochemical nuances important for translation of the molecular interaction between a drug and the target to an optimal pharmacological response could be missed with target-based approaches.

Translational biomarkers for phenotypic screening
Phenotypic assays have different types of endpoints depending on the goals; 1) empirical endpoints for basic research to understand the underlying biology that will lead to identification of translation biomarkers, 2) empirical endpoints to identify undesired effects related to toxicity of drug candidates, and 3) knowledge-based endpoints (biomarkers) for drug discovery which ideally are translational biomarkers that will be used to identify new drug candidates and their corresponding molecular mechanisms of action. The value of phenotypic assays is increased through effective alignment of phenotypic assay endpoints with the objectives of the relevant stage in the drug discovery and development cycle9.

Phenotypic assays for drug discovery
There is a need for more reliable predictable phenotypic assays. Arguably the success of phenotypic screening in areas such as infectious and metabolic diseases is due to the accessibility of accurate predictive phenotypic assays. In other disease areas such as CNS and cancer there is a need for more predictive models. In a recent report by Moffat and co-workers analysing the discovery of cancer drugs they found that recent phenotypic screening in cancer drug discovery has been hampered by a reliance on ‘classical’ nonspecific drug phenotypes such as cytotoxicity and mitotic arrest5. The point has been made by these authors and others that simple cellular models of cancer do not accurately translate to the complexity of three-dimensional tumours5,10. More disease relevant models including mechanistically informed phenotypic drug discovery (MIPDD) are needed to identify new anti-cancer medicines with new MMOAs.

Phenotypic assays used to discover first-in-class medicines
Retrospective analysis of the types of phenotypic assays used for the 28 NMEs categorised as firstin- class phenotypic in the analysis of how medicines were discovered by Swinney and Anthony are shown in Table 1. Ten of the medicines were identified using animals in which the phenotypic endpoints for the studies were well correlated with clinical indications. Levetiracetam, rufinamide and zonisamide were identified in well-establish models for anti-convulsant activity and aripiprazole in dopamine dependent activity known to be associated with anti-psychotic behaviour. Ziconotide was discovered in a model for pain and ranolazine in animal model measuring anti-anginal and antiischaemic effects. The endpoints for ezetimibe, nateglinide and pemirolast were blood cholesterol, blood glucose and cutaneous anaphylaxis, respectively. Nitisinone is used to treat tyrosineamia type 1 and was originally developed as a herbicide and repurposed for the rare disease when safety studies demonstrated an effect on tyrosine metabolism1,6. The phenotypic end points for those discovered using cell-based assays provide examples where cell death was used as the phenotypic marker. These included azacitidine, daptomycin, linezolid, nelarabine, retapamulin and sirulimus, all were approved for use as either anti-infective or anticancer therapies. Vorinostat was discovered by its ability to induce cytodifferentiation and growth arrest. The phenotypic markers for doconasol and cilostazol were viral replication and platelet aggregation, respectively. For the discovery of varenicline mesolimbic dopamine levels were measured and for fulvestrant the estrogenic effects.

The phenotypic marker for cinacalcet was an increase in calcium (Ca+2) in bovine parathyroid cells. The investigators were looking to agonise a calcium receptor. Miglustat was prepared and tested to interfere with glycoprotein synthesis and was repurposed for the Gaucher’s disease, a glycososphinoglipid storage disorder.

A current challenge for phenotypic drug discovery is that for many important diseases including diseases associated with ageing such as Alzheimer’s, the predictabilities of biomarkers and screening models are limited. Evolving technologies, including the use of stem cells, increase access and robustness of imaging technologies and systems biology provide opportunities to develop more predictive models.

Factors to consider for phenotypic drug discovery (PPD) (Figure 2) 
Phenotypic drug discovery provides a new opportunity to complement target-based programmes. There is overlap between the approaches but also differences that provide opportunities for innovation. In addition to the opportunities to establish more predictive assay and identify translational biomarkers, as discussed above, there are opportunities to innovate in the optimisation and development phases to improve on the approaches to optimise a hit from a phenotypic screen to a clinical candidate and, subsequently, a registered medicine. In the section below are discussed some of the areas with opportunities for innovation.

The format of assays and importance of relating translational biomarkers were discussed in the previous section. Other aspects that are important are the robustness, reliability and reproducibility of the data. It is critical to have screening data that can drive chemical optimisation and differentiate activity of molecules. Another factor for consideration is the cycle time of the assays, obviously the longer the assay the fewer iterations that can take place within a given time frame.

Optimisation of hits –medicinal chemistry
Hits are optimised from phenotypic screens into viable clinical candidates based on classical medicinal chemistry optimisation. Optimisation will be based primarily on experimental pharmacology data as opposed to structural and computational information, reinforcing the need for phenotypic assays that are reproducible and robust with easily interpretable data. An additional challenge for the medicinal chemistry optimisation is the desire and ability to optimise a molecule without knowing the details of the structural interactions. An empirical approach to medicinal chemistry optimisation is required. Opportunities to integrate phenotypic screening with the newer understanding of the drug-like molecules of properties may provide some insights. An obvious opportunity exists for analysis of the chemical characteristics of molecules discovered with phenotypic screens as compared to those from target-biased approaches.

Safety evaluation
A very important feature of optimisation is the safety evaluation. Regardless of the approach, empirical phenotypic evaluation of safety is required to identify non-specific toxicities. This does not change with a phenotypic approach, other than some of this screening is incorporated in the early discovery assays as part of hit validation. One concern of some when using phenotypic screening is ‘how to derisk for potential mechanism-based toxicity (on-target) if you do not know how the drug works?’ One practical approach to address this question is to use a series of molecules with similar structure, but different pharmacology to see if an observed toxicity relates to the pharmacology. 

Clinical development
There is value to relate the molecular mechanism of action to a specific patient population and the drug dose required to provide the desired pharmacological response. This is the underlying principle for personalised medicines. Many have used this argument as a reason not to pursue PDD or at least not proceed into clinical development without knowing a target or endpoint that can be used to determine patient populations and dosing. A significant challenge is that rarely are genetics and pharmacology as well matched as they are for some of the cancer phenotypes, including c-braf11 and abl inhibitors12.

How do we proceed into clinical development to choose patient populations and effective doses with incomplete knowledge? Unfortunately, our knowledge of most diseases and the underlying molecular causes are incomplete. An additional challenge is that knowledge of the cause, for instance a genetic defect or multiple genetic defects, rarely provides a specific molecular solution, despite our hopes7. Plenge, Scholnick and Altshuler recently noted most preclinical discovery programmes have incomplete supporting material to inform the drug discovery strategy13. This challenge applies not only to phenotypic drug discovery, but to all types of drug discovery. The challenge to address incomplete knowledge with a combination of empirical and knowledgebased approaches provides one of the greatest opportunities to impact drug discovery success and productivity.

Regulatory approval
The US FDA does not requirement knowledge of mechanism for approval. The requirement is that the medicine provide efficacy at a safe dose. With that said, more mechanistic knowledge is helpful to define patient populations, drug class and potential toxicities. There are opportunities to define what type of mechanistic information is useful and valuable.

Identification of Mechanism/target
Questions worth considering are ‘how much mechanistic information is required to help derisk development? Do you need to know a target? Or can you proceed with less detailed mechanistic information? Is identification of the target on the critical path or a nice to have? If drug action is considered from the standpoint that ‘the target is not the exclusive definition for a mechanism’, then different types of mechanistic information should be considered to help direct the clinical programme. For example, mechanism could be defined by a specific pathway where the drug candidate acts and/or a specific translational biomarker that can be measured clinically, such as blood pressure, viral load or cholesterol levels. In my opinion including identification of the target on the critical path for a phenotypic screening programme is a recipe for failure. Despite all the new tools it is still a challenge to identify an exact molecular mechanism. And in some cases the desirable pharmacology may be due to polypharmacology, where the drug works at a number of different sites or the action may involve a complex multicomponent transient physiological system. Obviously, it is somewhat easier for infectious diseases in which resistance mutations can be found, but even this does not inform as to the exact mechanism. The target and mechanism of the cholesterol- lowering drug ezitimibe was not identified until after it reached the market, despite considerable effort14. However, it was known that it worked by blocking cholesterol absorption through a novel mechanism of action. I think it is more important to consider what is the minimal information needed to move the compound forward in clinical testing than to determine the target and exact mechanism of action; this can be done in parallel with a goal to inform development and follower activities.

Executing a phenotypic strategy versus target
The early phase of phenotypic screening requires biology intensive resources, whereas the target-based screening once a target is selected is a chemistry and DMPK intensive effort. It is important to realise this when embarking on a phenotypic programme. More time will be needed to develop and validate assays and leads. A clearly thought out series of counterscreens will be useful to ensure the specificity and lack of toxicity of hits. Note that these are usually conducted later in a targeted approach following some optimisation. Opportunities for innovation occur at all these stages.

Bringing it all together
Renewed awareness of the value of phenotypic screening to drug discovery creates many new opportunities to increase drug discovery success and productivity. For all drug discovery approaches to be successful there is a need for validated translational biomarkers and predictive models of disease to guide drug discovery. The more relevant the system is to physiology the better it will predict the clinical success. Unfortunately, predictive phenotypic assays and relevant biomarkers are not available for most human diseases. One of the hopes for the genetic revolution was to identify specific genotypes, genes and targets that could be used to guide preclinical drug discovery to identify new medicines.

This approach has not been as widely successful as hoped. Aligning these efforts to identify translational biomarkers for phenotypic assays should increase the successful discovery of new medicines. Many of the features important to successfully execute a phenotypic drug discovery strategy are also relevant to target-based drug discovery, including how to move forward with incomplete understanding of the disease pathobiology, chemistry and mechanisms of drug action. Rarely is our understanding of these complete. The significant challenge and opportunity is to use available knowledge in combination with empirical approaches such as phenotypic screening to identify, create and assess drug discovery opportunities. New insights for drug discovery success will come from combining empiricism with the understandings from basic research in the context of realisation that the pharmacological knowledge will almost always be incomplete.

I wish to acknowledge Dr Jonathan Lee of Lilly for his thoughtful discussions and informative discussions that helped with the preparation of this manuscript. 

Dave Swinney has more than 25 years of industrial drug discovery experience. He is currently CEO of the non-profit Institute for Rare and Neglected Diseases Drug Discovery in Mountain View, CA, aka iRND3, where he is using new insights to identify clinical candidates and effective mechanisms of drug action.

Swinney, DC and Anthony, J. How were new medicines discovered? Nat Rev Drug Discov, 2011. 10(7): p. 507-19.

2 Sams-Dodd, F. Is poor research the cause of the declining productivity of the pharmaceutical industry? An industry in need of a paradigm shift. Drug Discov Today, 2013. 18(5-6): p. 211-7.

3 Lee, JA et al. Modern phenotypic drug discovery is a viable, neoclassic pharma strategy. J Med Chem, 2012. 55(10): p. 4527-38.

4 Swinney, DC. Phenotypic vs target-based drug discovery for first-in-class medicines. Clin Pharmacol Ther, 2013. 93(4): p. 299-301.

5 Moffat, JG, Rudolph, J and Bailey, D. Phenotypic screening in cancer drug discovery – past, present and future. Nat Rev Drug Discov, 2014. 13(8): p. 588-602.

6 Swinney, DC. The contribution of mechanistic understanding to phenotypic screening for first-in-class medicines. J Biomol Screen, 2013. 18(10): p. 1186-92.

7 Swinney, DC and Xia, S. The discovery of medicines for rare diseases. Future Med Chem, 2014. 6(9): p. 987-1002.

8 Hitchings, GH Jr. Selective inhibitors of dihydrofolate reductase. 1988, http://www.nobelprize.org/nob el_prizes/medicine/laureates/19 88/hitchings-lecture.pdf,

9 Swinney, DC. The value of translational biomarkers to phenotypic assays. Front Pharmacol, 2014. 5: p. 171.

10 Li, Q et al. 3D models of epithelial-mesenchymal transition in breast cancer metastasis: high-throughput screening assay development, validation, and pilot screen. J Biomol Screen, 2011. 16(2): p. 141-54.

11 Flaherty, KT et al. Inhibition of mutated, activated BRAF in metastatic melanoma. N Engl J Med, 2010. 363(9): p. 809-19.

12 Capdeville, R et al. Glivec (STI571, imatinib), a rationally developed, targeted anticancer drug. Nat Rev Drug Discov, 2002. 1(7): p. 493-502.

13 Plenge, RM, Scolnick, EM and Altshuler, D. Validating therapeutic targets through human genetics. Nat Rev Drug Discov, 2013. 12(8): p. 581-94.

14 Garcia-Calvo, M et al. The target of ezetimibe is Niemann-Pick C1-Like 1 (NPC1L1). Proc Natl Acad Sci U S A, 2005. 102(23): p. 8132-7.