Despite many decades of success, small molecule drug discovery appears to be increasingly challenged. The likelihood of approval of a small molecule at phase 1 is approximately half that of a biologic; pharma companies have shifted their pipelines dramatically towards large molecules over the past decade and have laid off thousands of medicinal chemists; and many of the advantages of small molecules – intracellular targeting and access to immune-privileged sites like the brain – are being eroded by new modalities and innovative delivery techniques.
This raises the inevitable question: is small molecule drug discovery a dying art, reminiscent of a golden age, but now destined for the antiquities section of the nearest molecular museum?
Old-school drug discovery
For decades, “traditional” small molecule drug discovery has been dominated by compounds that obey Lipinski’s rule of five (for those of you that believe in it). These rules guided chemists in their pursuit of classical chemical beauty: molecules that are small and efficient – less than 500 molecular weight – reflecting the types of disease targets that fueled the industry’s historic success, i.e. those with a well-defined binding site or small allosteric pocket on an enzyme, receptor or ion channel.
Playing in this space has become increasingly difficult. Much of the low hanging fruit has been picked over after many decades of success and we now find ourselves taking on substantially more risk:
- Chemical risk because we are left with targets that are seemingly intractable e.g. PTP1B, KRas, c-Myc and lack simple, obvious or precedented active sites; and
- Biological risk as we have turned to exciting but emerging biologies e.g. epigenetics and cancer metabolism.
It’s clear that business as usual will not be a winning strategy here. Our chemistry needs to innovate beyond the recycling of tired chemotypes to a point where we can explore diversity and functionality in new and powerful ways. I think that there are reasons to be optimistic that chemistry is entering a new period of enlightenment:
(a) More, different stuff. Warp Drive Bio’s turbocharged natural product platform is generating rich chemical diversity full of unusual linkages, three dimensionality and handedness (chirality), features that are less apparent in conventional compound libraries. DNA-encoded libraries (X-Chem, Hitgen) are now becoming productized, allowing staggering amounts of chemical space to be searched (100 billion compounds versus the ~5 million or so compounds that are commercially available for purchase). It remains to be seen whether these approaches will deliver hits that can be elaborated into viable drug candidates but the history of fragment-based drug discovery suggests that these technologies will find their killer app … in time.
(b) Small molecules with “superpowers”. Small molecules can do more than just bind reversibly to an enzyme’s active site. Avila Therapeutics, acquired by Celgene, and Principia BioPharma have demonstrated how proteins can be ‘silenced’ through tight (covalent) bonding while, newly-formed Arvinas, uses functionalized small molecules to recruit the cell’s protein degradation machinery, even in the absence of an active site e.g. for pseudokinases.
(c) Looking inside the black box. Structure-based approaches have reached a tipping point and can now be used to drive discovery prospectively, largely replacing brute force, trial-and-error chemistry. For example, Heptares Therapeutics has opened up the whole GPCR space to rationale design while, here at Nimbus Discovery, we are integrating new computational technologies developed with Schrödinger to predict allosteric sites, engineer selectivity against very similar off targets and drive potency using water energetics.
(d) Big data. Using massive amounts of structure-activity relationship data, Numerate is cracking targets where structural information is completely lacking e.g. GPCRs, ion channels while MedChemica is helping design safer molecules by compiling toxicology data across the industry.
We have observed many of these trends first hand with our acetyl CoA carboxylase (ACC) program at Nimbus. A natural product co-crystal structure provided the inspiration for a massive virtual screen that broadly surveyed chemical space. This led to identification of hits that, when bound to an allosteric pocket, could block protein dimerization. We then optimized our hits to Development Candidate in only 16 months and with 225 compounds synthesized. This was possible because our computational technology provided a roadmap for where and how to place chemical functionality, in the context of real-time in vivo efficacy, safety and ADME readouts.
Breaking the rules
As the Impressionists rebelled against classical Realism, the conventional definition of a small molecule drug is becoming blurred. The fact that orally-bioavailable drugs can now be designed with molecular weight of over eight hundred means that we are starting to discern the rules for these non-traditional small molecules. Recent examples include the HCV NS5A inhibitors (e.g. Gilead’s ledipasvir) and the BCL-2 inhibitors (e.g. Abbott’s ABT-199).
There’s undoubtedly a lot more to learn. This is illustrated nicely by the fact that cyclosporine achieves robust oral bioavailability despite having a molecular weight of 1,200. Exactly how this occurs is a bit of a mystery but a lot of smart minds at Peptidream, Bicycle Therapeutics and elsewhere are designing macrocycles in the 600-2000 molecular weight range that are cell permeable and, in select cases, orally bioavailable. Given their larger footprint, macrocycles allow us to imagine how to drug gnarly protein-protein interaction (PPI) targets which a colleague once described to me as like “trying to get a ball to stick to a table”.
Now that small molecules are being linked to proteins in the form of antibody-drug conjugates, the concept of a small molecule is being stretched to new limits. Adcentris®, for example, takes a highly toxic small molecule (monomethyl auristatin E), with a non-existent therapeutic index, and transforms it into an important new treatment for lymphoma.
To provide my perspective on the question raised above, I believe that small molecule chemistry, as traditionally defined and practiced, has limited utility in today’s world … but new chemistries that embrace the trend towards redefinition and convergence, offer real promise. This new world allows us to match our chemistry to the problem being solved, thinking that is clearly apparent in the recent Biogen Idec-Isis deal where the best of three modalities is selected on a target-by-target basis. In doing so, small molecules (however you define them) will continue to stake out competitive white space, especially in situations where patient convenience, cost of goods or tissue penetration e.g. the brain, are of the utmost importance.
We also need to be aware that good chemistry provides a robust starting point for drug discovery but ultimate success requires an integrated approach to managing biological, clinical and commercial risk. This starts with a clear understanding of what a drug does to the body (efficacy, safety) and what the body does to the drug (ADME). Clinical proof of concept should then be demonstrated in a clearly defined patient population that allows the new mechanism to shine, enabling registration studies to be designed that support a competitive product label.
Small molecule drug discovery is as challenging as it has ever been but the recent approvals of Imbruvica™, Stribild®, Sovaldi™, and Tecfidera®, show that the rewards are undoubtedly there for us as an industry. Most importantly, small molecules continue to deliver real value for patients and there is no reason why that should not continue as long as we embrace the opportunities at hand.