Trends in Target-Based Drug Discovery
Target discovery has advanced from phenotypic studies to sophisticated -omics and computational methods, transforming drug identification approaches.
Complete the form below to unlock access to ALL audio articles.
Prior to the sequencing of the human genome and the advent of recombinant biology in the early 2000s, drug discovery was predominantly undertaken through phenotypic or pharmacological studies. Using these methods, drugs were tested for therapeutic activity with the target being identified at a later stage – often decades later for some approved drugs.1
Today, a wealth of functional -omics and data analytical approaches is enabling a wide range of target discovery strategies. In this article, we highlight how target discovery has advanced and explore the growing role that computational approaches are playing in modern drug target identification.
Functional genomics
The sequencing of the human genome transformed target identification back at the dawn of the 21st century, heralding in an era of functional genomics, generally defined as “the broad, combined use of genetic information and genome-modifying technologies to connect genes and functional regulatory elements with their representative phenotypes”.2
Revolutionizing Cancer Therapy With Bispecific Antibodies
Discover the advantages of using bsAbs for immunotherapies.View Whitepaper
Genetic tools, such as RNA interference, allowed researchers to study whether gene expression differences caused a loss, gain or modification of function, and gave a direct route to validating the functional relevance of putative target genes.2 However, reproducibility was an issue, and the advent of CRISPR-Cas9 gene editing has largely superseded this approach, enabling a mix of gain-, loss- and modification-of-function screens while also making it possible to generate cell- or animal-based models for target validation. There are now many variations of CRISPR technology, and they are a key asset in the toolbox being exploited for target discovery.3
A disease mechanism through systems biology
Although functional genomics continues to play an important role in target discovery, the functional relevance of genes identified is not always clear. Increasingly, researchers are turning to a range of -omics technologies, made possible by advances in tools such as mass spectrometry, RNA sequencing and data analytics. This is heralding in a new era of systems biology, says Matteo Barberis, professor of systems biology at the University of Surrey, in which it is possible to overlay different maps derived from -omics data and use these to pinpoint the key players in health and disease.
“Traditional approaches to suggest or discover targets come from bioinformatics, which applies arbitrary thresholds to determine the significance of a molecule,” said Barberis. “Although useful, this tells us nothing about the biological context of these molecules, and how they interact with other molecules during normal physiological and pathological processes.”
An alternative is to integrate multiomics approaches with existing biochemical knowledge to build mechanistic maps. Barberis achieves this by layering multiomics data onto a biochemical map and then uses the resulting multi-scale map to inform mathematical models to predict disease development. In this way, it becomes possible to visualize how the different up- and down-regulated molecules link together in different contexts.
Accelerate Drug Discovery With Small Molecule Microarrays
Explore the benefits of small molecule microarrays for rapid hit identification across target types.View Poster
“We’re trying to build an integrated functional view of a healthy human so we can then highlight the key mechanisms – and targets – that are specific for disease,” said Barberis. “Many disease targets found by bioinformatics studies can also be found in healthy organisms, but although these entities might be linked in both health and disease, it is the way they communicate that can be very different.”
This approach was recently used in a collaborative study with physician Andrea Biondi from the University of Milano-Bicocca in Italy, from where Barberis also graduated, to pinpoint mechanisms underlying phosphatidylinositol 4-kinase alpha (PI4KA)-related disorder, a condition which causes wide-ranging neurological and gastrointestinal symptoms.4 The results indicated that PI4KA-related disorder is an immunological syndrome of several processes – mainly altered lipid metabolism and energy production that in this case cause B-cell deficiency.
“We believe other diseases could also be considered as syndromes, because there are so many processes that go wrong which are common to several inflammatory conditions,” said Barberis. “The power of our target discovery approach is that it’s disease independent. We step back from the disease definition and look only at the mechanism.”
Harnessing computational approaches in target discovery
While Barberis’ approach uses existing biochemistry knowledge to find insights about common targets, others are looking to machine learning to find disease-agnostic targets.
Ekta Khurana is an associate professor of computational genomics at Weill Cornell Medicine and develops computational approaches for analyzing gene function data derived from patient samples.
“Finding new targets in oncology is challenging, and although you can find some of them from genetic sequencing analysis, many tumors remain with no known targets,” said Khurana. “Only recently it has become possible to obtain chromatin accessibility data from patient samples, which can tell you when and how different genes are expressed by revealing the switches that control these genes.”
Khurana used ATAC-seq (assay for transposase-accessible chromatin with sequencing) and RNA sequencing data to construct gene regulatory networks for 371 patients with 22 cancer types. Using machine learning, they integrated the chromatin and RNAseq data (i.e., which genes are being transcribed in a cell at a given time), which then reveals where the switches are that turn these genes on and off at that time.5
Harnessing Structural Insights in Drug Discovery
Explore how structural insights transform target validation and drug design.View Webinar
“The underlying idea is that by integrating these datasets using machine learning, we tried to look at the patient tumor at the molecular level and not from the label of kidney or colon cancer,” said Khurana. “This is based on the idea that the main proteins driving tumor growth can be similar even across different organs.”
Once the switches and genes they produce are identified it becomes possible – with the power of machine learning – to identify the master control switches, i.e., the proteins that are regulating the whole machinery. “We have the 3D chromatin structure for some time points for some samples, but we certainly don't have it at the scale of 400 patients, and that's where machine learning comes in,” said Khurana. “The kind of data we are getting now was not feasible until recently, and there are so many data modalities that can be used right now and so many conceptually innovative ways to use them. We now have a different lens for identifying what these targets are.”
Summary
With the advent of large-scale multiomics and advanced data analytic technologies, strategies for target discovery are shifting from focusing on disease-specific molecules and phenotypes, to understanding the system-levels changes that occur during the development of different types and stages of disease.
Understanding how these networks of potential targets interact in the transition from health to disease should aid in identifying the key controller targets that play roles across complex conditions. When modulated therapeutically at an early stage of pathogenesis, these targets could be exploited to effectively prevent and treat people in a disease-agnostic way.
