Despite the technological advances, drug discovery is still a lengthy, expensive, and difficult stage in the drug development process. Screening assays are a critical part of rapidly and thoroughly analysing vast numbers of compounds for desired biochemical activity. In a recent interview with Proventa International, Dr Prasun Mishra, Founder and CEO for Agility Pharmaceuticals, discussed the benefits and challenges of phenotypic screening in drug discovery. 

For daily articles on the latest pharma trends and innovations, as well as interviews with leading experts and in-depth industry White Papers, subscribe to PharmaFeatures.com.

Proventa International: You recently spoke about reinstating phenotypic screening (or PDD: phenotypic drug discovery) in drug discovery at our Medicinal Chemistry Strategy Meeting. Could you explain what phenotypic screening is?

Prasun Mishra: Phenotypic screening is a type of screening used in biological research and drug discovery to identify substances such as small molecules, peptides, or RNAi screens, that alter the phenotype of a cell or an organism in the desired manner. Briefly, phenotypic screens look at the effects of these substances on genes, pathways, and inhibition/induction, which are causing the desired change in the phenotype.

PDD is an alternative method of drug discovery that aims to address some of the challenges with target-based drug discovery (TDD). What advantages do you believe PDD offers over target-based methods?

PM: Traditional target-based methods are employed once a target is known: By inhibiting this target you get the therapeutic benefit/advantage. While this works for the disease cases where a mutation causes the disease, this is not the case for a majority of diseases. The biology is complex, and usually one or more pathways are involved in driving a complex disease phenotype. 

So then it becomes important to address that complexity of disease biology, which cannot be done by the classic approach of targeting a single gene. Hence, phenotypic screens can be used to understand both the biology of the disease and the master regulator genes/pathways associated with that disease phenotype. I believe that phenotypic screens are the right approach to address this challenge, which is why the drug discovery field is moving towards PDD right now.

Could you go into more detail about the relevance of high-content imaging in phenotypic screening?

PM: High-content imaging screening is a type of phenotypic screening where the whole cell phenotype can be imaged using various cellular markers. In this manner, high content screening utilises molecular imaging to assess the phenotypic changes at the cellular level – sometimes in real-time – upon treatment with a substance. It’s also called cellomics as we are trying to look at the cellular phenotype through the eye of a microscope or a profound assay or imaging system. 

How well has phenotypic screening been integrated into animal models?

PM: Phenotypic screening can be utilised to replace animal experiments. For example, in the case of cancer, one can use patient-derived tumour cells and grow them into spheroids. These assays are powerful not only in replacing animal-based screening assays but also can be used as a secondary screen to get an idea of hit confirmation. For example, if you have the top hundred compounds from a large screening, then you can take those molecules and run them through a secondary screen of spheroids to understand how they would be performing in three-dimensional spheroid assays. Essentially, you mimic environments in patient tumours. 

Patient-derived spheroids can give you an answer as to how candidate drugs will contribute to the shrinking of 3D tumours. It provides a means to avoid a large-scale secondary screen using animal models. Using complex co-culture screens, one can narrow down the list of hundreds to the top three to five candidate drugs. These few lead compounds can be further optimised utilising animal experiments which are necessary for IND-enabling studies towards FDA approval. 

So, in the above example, by utilising PDD, you have not only narrowed down the scope of active molecules but also have budgeted your project with a limited number of animal experiments. 

Are there any limitations of PDD that need to be addressed? The high rate of false positives, for example?

PM: We as a community have addressed some [major] challenges associated with PDD. The key is to run a very good positive, negative, and no treatment control (blank). This then allows us to run our data analytics against these controls to give us true hits. Analysing the spectrum of data through these provides very good insights into the candidate drugs. We have also developed a list of common false positives and false negatives associated with a screening assay. So, utilising the proper controls and getting rid of false hits, we have overcome some of the key limitations associated with phenotypic screens. 

There is a saying in our field: “Garbage in, garbage out.” If your screen is designed poorly, you will have a high rate of false positives. Designing a high-throughput screen is a robust process. One has to spend some time diligently designing and validating a phenotypic screen to gain confidence in a screen. Furthermore, gaining confidence in a readout helps precisely measure what you want to measure. Once the screen and readout are optimised, you can screen millions, if not billions, of compounds in a very short time using this powerful tool. 

If a phenotypic screen is designed well, it can lead to amazing findings. This can transform basic biology into what I call “high-throughput biology.” Many research projects and brilliant minds have been stuck on one protein, one pathway, one drug hypothesis. Since then, a lot of progress has been made in the field of PDD. Also, the whole genome is not druggable, PDD allows us to identify upstream and/or downstream effectors to an undruggable gene to be able to still inhibit a pathway.

Some of the new papers in the PDD field are very promising, reiterating the importance of phenotypic screens as well as the importance of high-throughput approaches to find ways to inhibit certain disease phenotypes. I am confident that if we utilise PDD and high-throughput approaches to understand disease biology, it will propel us faster in finding curative therapies for some of the incurable diseases.

Can phenotypic screens replace animal models? 

PM: Yes, I think that if we use at least some of these smart assays that we have developed, such as co-culture assays, as well as various complex phenotypes/ tissues, we can replace the need for animal screens with phenotypic screening. 

That said, I would acknowledge that this is an optimistic statement, as you still have to do IND-enabling studies utilising animal experiments to gain FDA approval. However, as I said before, you can narrow your leads, and then utilise very few animal experiments for confirmation studies. So, I hope I have convinced you that one can minimise costs as well as spare animal lives by smartly utilising phenotypic screening.

In general, how important is AI and ML in further improving PDD?

PM: AI and ML will help mainly with data analytics of high-throughput biology. Because of the complexity within the body, the cells, the pathways, and complex interactions, phenotypic screening is the way to go. However, one of the downsides is that you are bombarded with massive datasets, so you have to be prepared to analyse and have a way to make useful conclusions in the right way. So, this is where AI, ML, and smart algorithms are very helpful. 

You can automate not only the cell culture process and screening but also the data analysis process. So, once you’ve streamlined everything, you get the best possible insights into high-throughput biology using a phenotypic screen accompanied by the AI/ML tools.

Charlotte Di Salvo, Junior Medical Writer
Proventa International