Fail Early, Fail Fast – A phenotypic rescue approach
New disease targets frequently emerge in literature, but the thorough target validation required to consider these targets for a drug discovery programme is often lacking.
In pharmacological or genetic perturbation studies using complex biological assays, undesired off-target effects cannot be easily distinguished from the intended mode of action at the desired target. This is especially evident in cancer drug development where it is important to discriminate on-target effects on cell viability from off-target effects resulting in non-specific loss of cellular fitness.
Neglecting the possibility of being deceived by off-target effects can have tremendous scientific and financial impact on a drug discovery programme. Ideally confidence in a preclinical drug target and a modulating compound is boosted in an early stage by more extensive analysis and validation of the actual target-disease or drug-target relationship. Rescue of a relevant phenotype by genetic restoration of a target mutation is a gold standard approach in drug discovery by which target validation can be achieved.
The cost of failure in drug development
Drug development is a cost and time-consuming process that can take more than 12 years (1) and cost about US$2.6 billion per approved compound (2). Such big figures are associated with high risks for the party investing in a drug development programme for which it is crucial to mitigate the chance of failure and thereby reduce costs involved as much as possible. However, in practice only about 14% of Phase I drugs reach approval. In oncology specifically, drug attrition rates are even higher, with only a meager 3.4% of drugs being licensed (3), which represents a huge financial burden on a development programme.
Discontinuation of drug development projects during clinical trial is often due to poor efficacy or safety, aspects that may be tackled by more thorough characterisation earlier in the pipeline utilising appropriate biological models and target validation approaches. CRISPR-Cas9 technology can be used in a simplified phenotype rescue approach, decreasing the significant pressure on profitability in drug discovery by limiting cost and failure as early as the target validation step.
Common target identification, hit finding and validation approaches
Currently the identification of a druggable target (gene, mRNA or protein) associated with a disease phenotype relies mainly on functional genomics and phenotypic screenings. For example, data generated by genome sequencing of individuals, tissues and cancers are great resources to identify genomic characteristics, gene mutations or genetic factors associated with disease.
Identification of a biological target has also benefited from the advances in functional genomic screening, such as RNA interference and CRISPR-Cas9 systems, high-throughput phenotypic screens using libraries of small molecules, biologics (antibodies and enzymes) and TIDES (peptides and oligonucleotides) are key to the identification of moieties that induce a desired therapeutic effect in a specific disease model. Also in silico analysis of drugtarget interaction is valuable for tool compound prediction and lead optimisation. Altogether these target identification and hit finding approaches are the very first steps in a drug development pipeline.
With the estimated probability for a novel target to reach preclinical stage at only 3%, compared to 17% for a known target (4), and more than half of 2018 FDA approvals targeting orphan diseases that are less well characterised (5), the subsequent validation of newly-identified targets is fundamental. The main objective in target validation is to confirm the involvement of the target in disease phenotype(s) and to select biologically plausible and relevant molecular targets for interference. Target validation can be broken down in two key steps:
1. Repetition of the target identification experiment to demonstrate reproducibility.
2. Introduction of changes in the experiment, such as a different cellular context or the use of different approaches to validate the functional role of the target in the disease phenotype.
A variety of cell-based assays have been developed in target validation using different strategies. The most popular approach modulates the cellular amount of the target by decreasing or disrupting the gene expression of a target using RNA interference or more recently by CRISPR-Cas9. The induced loss or gain-of-function allows for the determination of the physiological role of the target in disease phenotype without the presence of a drug. Such methods for accurate target validation are critical to improve the downstream probability of success, but need to be executed carefully as outlined below.
Pitfalls of poor target and hit validation
Although early sifting by proper target and hit validation can have a tremendous positive impact on downstream success rates, candidates are sometimes fed into downstream development programmes without complete knowledge of the underlying physiology and even improper assumptions on complex biology (6). Common pitfalls in interpreting the relevance of a (novel) target for the disease are:
– Failing to account for multiple gene targets. Often it is hypothesised that a single target or driver mutation is associated with disease prognosis, whereas in practice it is much more common that multiple genes are contributing to a diseased state. These factors may even counteract each other or redundant genes may exist to safeguard a homeostatic and resistant environment. Also if a single target has multiple roles in various pathways (7) it may not be favourable to challenge them all or it may be considered for drug repurposing (8).
– Cloudy forecasts. Despite great advancements in genome-wide association and expression profiling, it is still hard to predict whether altered gene or protein expression is a consequence or a cause of the disease.
– Misinterpreting target expression and knockdown levels. Assessment of target expression levels by RNA interference also poses confounding factors that need to be considered. Besides practical concerns such as residual expression by incomplete mRNA knockdown, high variation of knockdown efficiency between cell lines and a high incidence of confounding off-target effects (9), data could be misinterpreted by biological issues. For example, a gene may be processed at transcription or translation level by which genetic disruption would not decrease expression of all isoforms. Also, a gene could be expressed in excess, by which incomplete knock-down is not sufficient to establish the function of the target.
Besides validation of the relationship between the target and the phenotype, the drug-target interaction and its predicted biological effect should also be confirmed. Key points that should not be overlooked herein are:
– Off-target effects. In complex biological assays it is not always evident whether the resulting druginduced phenotype is (entirely) due to the anticipated response at the desired target or if off-target effects should be taken into account. For example, cellular fitness is governed by many different processes in parallel. Unintentionally disrupting any of these pathways may impact cell viability independent of the intended on-target effect of an anti-cancer drug candidate.
– Choose your model wisely. Another important consideration for accurate drug-target validation is the model system used. A simple cell line will likely react differently to a stressor than a 3D cell model or primary patient material that may be a better reflection of the in vivo situation. Even more so, individual background mutations differing between patients may interfere with the efficacy of a compound. For this reason, drug-target interaction and phenotype modulation need to be confirmed in relevant cellular and disease contexts.
All in all, this highlights the importance of not disregarding comprehensive target and hit validation and carefully confirming the hypothesis before proceeding to the next stages of development for a novel compound.
Phenotypic rescue approach: a gold standard in target validation
Many of the considerations described above can be tackled by a phenotypic rescue approach. In such experiments, the role of a target and modifying drug in the disease phenotype is investigated by manipulating the target in the cellular context, such as:
– The modification or deletion of the active site of the target
– The restoration from the disease-associated variant of the target to the wild type variant of the target
– The precise deletion of the region of interaction with the drug.
Such approaches can confirm that the target is causative for the disease and putative off-target effects can be distinguished from on-target effects without confounding factors such as different cellular backgrounds. Currently, phenotype rescue is often performed by decreasing or disrupting other pathways to compensate for the mutant phenotype, by transient expression or integration of the original or mutated form of the target elsewhere in the genome. Each of these strategies has drawbacks of its own, however.
Indirect rescue via another pathway is often hard to interpret given the complexity of interacting cellular systems. Re-expression systems may lead to overexpression artefacts by disrupting the natural balance of genes expressed in a homeostatic state. Also for transient expression systems the timing may not coincide with the disease-causing mechanisms and integration elsewhere in the genome may disrupt the function of the locus where it is integrated. It may thus be clear that improvement of the methods commonly used for phenotypic rescue is highly desired.
Added value of CRISPR-Cas9 technology in phenotypic rescue experiments
There is no doubt that the CRISPR-Cas9 technology has revolutionised many facets of science and target identification and validation approaches for drug discovery are no exception to that. It is now relatively easy to edit a specific target gene of interest and genome-wide knockout libraries are commercially available, making it a popular tool for phenotypic screens. Besides this well-known application in target discovery, the CRISPR-Cas9 toolbox also opens new possibilities for performing experiments aimed at target and hit validation.
Phenotypic rescue experiments can be improved by exploiting the CRISPR-Cas9 technology in various ways (Figure 1).
Complete abolishment of target gene expression using CRISPR-Cas9 provides the maximal window of a phenotypic effect for the particular target without the confounding residual expression from incomplete knockdown by RNA interference. Thereby interpretation of knockout and rescue experiments is more straightforward. Furthermore, the impact of a certain mutation on a disease phenotype can be assessed in a physiologically relevant system by precise correction of the mutation at the endogenous locus, thus not affecting expression levels.
Restoration of a disease-associated mutation is also a valuable approach for drugs that supposedly impact only the mutated and not the wild type form of a target: a resulting insensitivity of the cell line to the drug confirms specificity for the mutated form and rejects off-target effects. Similarly, the involvement of a target in a drug-sensitive phenotype can be studied in more detail by precise mutation of its putative active site.
Mutating the hypothesised drug interaction site can help to elucidate a drug mechanism, but also to determine specificity of a drug. Although in general these are effective approaches for sound target and hit validation, caution should be exerted for cancer lines specifically since correction of a driver mutation may be technically challenging as it could result in loss of cellular fitness in vitro.
All such target and hit validation strategies using CRISPR-Cas9 can be reproducibly performed in several independent cell lines containing their own unique background mutations to confirm robustness in various patient populations. In principle, the same toolbox can be repeatedly used for all cell lines, albeit some loci seem to be more resistant to editing in one cell line compared to the other. Also, delivery of the system is challenging in some cells and tissues. So even though the addition of CRISPR-Cas9 technology to the phenotype rescue toolbox is a great add-on, there are some limitations to take into account.
The way forward to success
In order to reduce the enormous costs of drug development there is an urgent need for more stringent filtering of putative drug targets and candidate compounds in early drug discovery. This can be achieved by investing more time and effort in characterisation and validation of the drug-target interaction on a disease phenotype. As outlined in this article, rescue of a disease-relevant phenotype by genetic restoration using CRISPR-Cas9 is a valuable and recommended tool to use next to the more established methods for target identification, hit finding and validation.
As all interventions and model systems have drawbacks and pitfalls of their own, it is strongly preferred to use a combination of approaches for detailed assessment of a putative clinical candidate prior to moving forward, with a strong focus on target and hit validation after the initial discovery.
A robust validation procedure could entail:
1. Validation in different model systems, eg cell lines from various tissue types, several individual patients, 2D versus 3D cultures.
2. Confirmation of results generated using RNA interference methods by CRISPR-Cas9 gene editing and vice versa.
3. Reproduction of the genetic ablation-induced phenotype by intervention with small molecules.
4. Rescue experiments by genetic restoration of the disease-associated mutation, abolishment of a targets’ active site and/or putative drug-target interaction.
Increased awareness of such gold standard approach in drug discovery will help to boost profitability of a drug development programme by early failure of poor candidates. DDW
This article originally featured in the DDW Summer 2019 Issue
Dr Anne-Marie Zuurmond is a Director at Charles River. She has been active as project leader in a CRO environment for more than 14 years, managing projects in various therapeutic areas, including neurology, fibrosis, arthritis and metabolic syndrome. She was responsible for the implementation of the CRISPR/Cas9 technology at Charles River and is now leading genome engineering projects using this technology to support in vitro drug discovery. She earned her doctoral degree in molecular biology and MSc in biology from University of Leiden.
Dr Geraldine Servant is Senior Scientist at Charles River. She received her PhD from Pierre and Marie Curie University in the field of Molecular Biology. As a molecular biologist with 12 years of experience in France and the United States, she has been a key player in innovative research. Currently, her dream would be to work on research and development of bio-medicines and targeted therapy for the treatment of cancer or rare diseases. She is also striving to work as a project manager to pilot innovative scientific projects, develop systems that analyse living organisms and original and efficient therapeutic treatments for a more adapted medicine for patient benefits.
Dr Lieke Geerts is Senior Scientist at Charles River. She received her PhD degree in neuroscience from the VU University in Amsterdam and is currently a project leader at Charles River. Since joining the company she has gained experience in assay development, drug target validation and small molecule testing using a wide range of technologies and in various disease areas. A main focus has been the integration of CRISPR/Cas9 into the Charles River toolbox for supporting drug development programmes.
Dr Laure Grand Moursel is Senior Scientist at Charles River. She received her PhD in neuropathology and cerebrovascular disorders from the Leiden University Medical Center. Before joining Charles River Laboratories she worked at Biomethodes on the directed evolution of proteins of therapeutic interest and at the Leiden University Medical Center where she specialised in the development of Heavy Chain antibodies. She currently works on drug target identification and validation using CRISPR/Cas9 gene editing technology in Charles River Early Discovery.
Dr Jeroen DeGroot is Senior Director at Charles River. He received his PhD in medical biology from the University of Utrecht. Before being acquired by Charles River Laboratories, Dr DeGroot was the Director of Cell Biology at BioFocus and before that was a Research Manager at TNO for more than 13 years. At Charles River, he supports clients with innovative contract research services, from initial target identification, through hit identification, H2L & LO programmes, via in vivo pharmacology and regulatory safety studies to IND.
Dr Ian Waddell is Executive Director, Biology at Charles River. He received his PhD in molecular medicine from the University of Dundee (UK) and secured a lectureship in Dundee before working at AstraZeneca, followed by CRUK at the Manchester Institute. Dr Waddell is the Executive Director of Biology in CRL, has a very strong interest in efficient drug discovery and is a lean six sigma black belt.
1 Dimasi, JA, Feldman, L, Seckler, A, Wilson, A. Trends in risks associated with new drug development: Success rates for investigational drugs. Clin Pharmacol Ther. 2010 Mar;87(3):272-7.
2 DiMasi, JA, Grabowski, HG, Hansen, RW. Innovation in the pharmaceutical industry: New estimates of R&D costs. J Health Econ. 2016 May;47:20-33.
3 Wong, CH, Siah, KW, Lo, AW. Estimation of clinical trial success rates and related parameters. Biostatistics. 2019 Apr;20(2):273-86.
4 Van den Haak, MA. Industry Success Rates 2004. CMR International; Surrey, UK: 2004. 2004.
5 Kinch, MS, Griesenauer, RH. 2017 in review: FDA approvals of new molecular entities. Drug Discov Today. 2018;23(8):1469-73.
6 Vasaikar, S, Bhatia, P, Bhati,a P, Chu Yaiw, K. Complementary Approaches to Existing Target Based Drug Discovery for Identifying Novel Drug Targets. Biomedicines. 2016 Nov 21;4(4):27.
7 Jeffery, CJ. Multifunctional proteins: examples of gene sharing. Ann Med. 2003;35(1):28-35.
8 GNS, HS, GR, S, Murahari, M, Krishnamurthy, M. An update on Drug Repurposing: Rewritten saga of the drug’s fate. Biomed Pharmacother. 2019 Feb;110:700-16.
9 Echeverri, CJ, Beachy, PA, Baum, B, Boutros, M, Buchholz, F, Chanda, SK et al. Minimizing the risk of reporting false positives in large-scale RNAi screens. Nat Methods. 2006 Oct;3(10):777-9.