Tuning precision polypharmacology: our investment in Harmonic Discovery
By Joel Dudley and Nick Olsen
The hurdles to successful drug development are truly staggering. First, drug developers must identify a biological target for therapeutic development. Target identification is arduous — numerous biological and clinical factors factor into the target selection process. Once drug developers identify a therapeutic target (most often a protein target), they determine the appropriate mechanism to increase or decrease the activity of that target. Then, drug developers must precisely engineer a bioactive ligand (usually a small molecule chemical, antibody, or other protein) that reliably and specifically interacts with a particular target to drive a desired biological activity. In addition to target-ligand activity, a therapeutic drug must stably survive human digestion and metabolism and find its way to diseased cells and tissues.
Researchers have found a way to discover medicines that have significantly improved the quality of life for billions of people worldwide over the past century despite the complexities of successful drug development. And certainly, our methods have improved considerably since the discovery of morphine, often credited as the first modern medicine. With pioneering experiments and work on sickle-cell anemia in the 1940s, many historians credit Linus Pauling for starting the movement toward molecular medicine — the concept that a defective molecule could drive an illness. As science accelerated through the 1980s, advances in molecular biology provided additional tools for interrogating biology that gave researchers a fine-grained view into underlying biological signatures of disease. The FDA’s approval of Herceptin in 1998, a drug targeting the overexpressed estrogen receptors that the Her-2/Neu oncogene, marked a breakthrough for genetically-targeted drug development.
As we have moved into the 21st century, next-generation sequencing and other molecular profiling tools have continued to improve our precision of diagnosis. Consequently, therapeutic development is becoming more targeted and more precise. For example, in 2020, the FDA approved two drugs that specifically targeted RET-driven malignancies (selpercatinib and pralsetinib), which are caused by specific mutations or gene fusions in a family of kinases. Older kinase inhibitors showed mediocre efficacy, but this new class showed significant clinical benefit. Identifying these mechanisms was impressive enough, let alone that researchers could precisely target drugs to this particular kinase conformation when there is significant structure conservation across the kinase family.
The molecularly-targeted medicine concept has focused on rational drug discovery and over-indexing single molecular targets — driving a one-drug, one-target mindset. However, most common diseases involve complex mechanisms acting in a symphony of multiple genes and other molecular factors. Even when canonical knowledge indicates that a drug targets a single protein, it is often promiscuous and interacts with other closely related proteins, a phenomenon known as “polypharmacology.” Interactions with the wrong proteins can drive significant toxicity and safety issues. Interactions with the right combinations, however, can drive harmonious responses.
An over-focus on the one-drug, one-target paradigm remains a challenge in drug discovery up until this day. Over the past few years, covalent medicines have returned to vogue (named as such, given that they form covalent bonds with the protein they are targeting), given their ability to be highly specific. However, in diseases like cancer, tumors evolve and are quick to find mechanisms to resist and escape the therapy, as we’ve seen in the case of Ibrutinib. This resistance emerges because signaling pathways have evolved to be robust and redundant. From an evolutionary biology perspective, this makes sense. You don’t want the cell to die if you lose a protein at a step in a pathway.
Ideally, you want to design a solution that embraces this complexity of disease and redundancy of signaling pathways and takes advantage of the polypharmacological nature of small molecule chemistry. However, we’re lacking the toolkit to do this effectively today. Today, we’re excited to announce our investment in a team working on this problem — Harmonic Discovery.
Harmonic Discovery’s vision is to design therapeutics that unlock the full potential of kinase therapeutics. Molecules that are built atom-by-atom and finely tuned for each indication.
Kinases are a great target class to pursue this strategy. Once considered undruggable, kinases are critical nodes in oncology and inflammatory diseases. They are intimately involved in crucial functions like cell growth, proliferation, and survival. The structural similarity of the active site across the entire family has evolved to bind a common substrate, ATP; this is wholly unlike other target classes that interact with structurally distinct ligands.
Harmonic Discovery’s platform is an end-to-end, machine learning-first infrastructure that integrates all aspects of kinase drug discovery. Harmonic Discovery’s platform builds on advancements in computational power, machine learning, computational chemistry, and high-throughput biological perturbations. Their models integrate the vast quantity of kinome data, learn from the manifold of structural similarity and binding affinity, and generate synthesizable chemistry tuned for polypharmacological effects.
More importantly, the team approaches this problem with fresh eyes but grounded rationality. The founding team — Rayees, Jason, and Marcel — bring together a trio of skill sets fit for this company. Rayees comes from a background in machine learning and computational chemistry; Jason brings a clinical mindset, and Marcel brings a medicinal chemist’s intuition. They believe strongly in learning from historical successes and failures in the field while applying state-of-the-art techniques to areas of highest unmet need.
We are thrilled to support the team in this journey.
For more detail on the platform, check out brief descriptions of oral presentations given at the American Chemical Society Fall 2022 meeting that introduce key aspects of the platform. Additionally, if you’re passionate about making a difference in this field, join the team here.