Lately it seems that allostery is constantly in the news, especially in stories covering a series of high-profile financings by companies touting a range of approaches and a steady stream of news from top tier academic institutions. The interest in allostery is driven by a desire to drug targets that remain inaccessible to the current pharmacopeia. Allostery allows us to reach beyond the limits of traditional drug discovery. In this post, I want to make the case for why the conventional ‘active site’ approach to drug discovery is challenging and why Nature might give us clues to a solution.
Conventional approaches to drug discovery generally focus on the active site of the protein, the ‘engine’ where fuel molecules such as ATP are consumed or where chemistry is performed on the enzyme’s substrate. Unfortunately, the active site presents significant challenges for a number of reasons:
• Drug molecules must compete with very high concentrations of substrate (the engine’s fuel,) and this can lead to many orders of magnitude reduction in potency
• Nature performs chemistry in a select number of ways, so even if a molecule can successfully target an active site, it is likely that the molecule will cross react with a series of other proteins, leading to side effects and poor tolerability in humans
• Active sites are often either very lipophilic (i.e. greasy) or charged, leading to medicines with similarly poor physical properties. Such molecules can be very difficult to formulate into low dose, once-a-day pills that lead to good patient compliance
In addition, many proteins lack active sites entirely, meaning that a large class of proteins are out of reach with this approach.
At HotSpot, we are pursuing a new approach to drug discovery that uses Nature as its blueprint. Nature has long been the inspiration for innovation, and in many cases this has led to the creation of entirely new industries. Watching birds fly inspired the first aviators to build wings and then airplanes; the sun prompted physicists to ask how stars can create seemingly limitless energy, which in turn led to scientists to harnessing fusion here on earth; and the human brain provided the blueprint for neural networks and the field of machine learning.
A similar mindset can be applied in the field of drug discovery.
Over many millions of years, evolution has developed precise ways of controlling protein function. These usually involve chemical modifications of the protein that lead to protein rearrangements or interactions with new partner proteins, resulting in the inhibition or activation of a cellular function.
Cells take advantage of a diverse menu of chemical modifications, including adding or removing negatively charged phosphates (phosphorylation), carbon-containing groups (methylation, acetylation) and longer fatty acid chains (myristolation).
It is remarkable to think that adding just four atoms of a phosphate group can dictate whether a protein a thousand times larger is active, but such is the precision with which Nature orchestrates proteins within the body. Understanding exactly how Nature’s regulatory chemistry works is at the heart of HotSpot’s approach to drug discovery. If we can uncover the mechanisms that Nature itself uses to control proteins, we can design small-molecule medicines that replicate the same mechanisms. In many cases we are able to generates medicines with similar precision and potency.
These is good reason to believe that such an approach is feasible and could lead to important new medicines. An important example exists in the metabolic enzyme, Acetyl CoA carboxylase (ACC), an enzyme responsible for the rate-limiting step of fat synthesis. Addressing fat accumulation in the liver is an important step in treating diseases such as non-alcoholic steatohepatitis (NASH), which led to significant industry interest in ACC. Animal models that lacked the gene encoding ACC showed broad improvements in metabolic and liver health, leading to a gold rush to develop an effective ACC-targeting therapeutic. The pharmaceutical industry spent about 20 years focused on the active site of ACC without success, invariably leading to non-selective molecules with poor drug properties.
In her previous life at Nimbus Therapeutics, HotSpot Co-Founder Gerry Harriman pioneered an approach to drug a regulatory hotspot on ACC. Nature regulates ACC via the phosphorylation of a flexible “tail” at one end of the protein. Once phosphorylated, the tail binds tightly into the interface between two of the protein’s subunits. The binding of the tail prevents the two protein subunits from coming together, thereby inhibiting the protein.
The Nimbus team, collaborating with computational chemistry leader Schrödinger, recognized that small drugs could be created that replicate the effect of the phosphorylated tail. Using knowledge of the protein’s molecular structure and computer-based drug discovery methods, Gerry’s team developed the first hotspot-targeted molecules, differentiated from the active-site forerunners. These molecules were exquisitely selective since the natural regulatory site was not present elsewhere in the cell. The inhibitors showed dramatically-improved pharmacology, in part because potency was maintained in vivo but also because the molecules disrupted how the ACC protein came together in a complex. The idea that a tiny small molecule can prevent two sizeable proteins coming together may seem remarkable, but the natural regulatory mechanism works in exactly the same way.
In addition, the pocket on the ACC protein has never been targeted before with small molecules. It represents completely novel biological real estate that allows chemists to move away from the overused active-site chemical libraries to fresh new chemotypes. In the hands of the Nimbus team, this led to molecules with pristine chemical properties enabling the drug substance to be formulated as a low dose, once-a-day oral pill.
Armed with these desirable properties, the ACC hotspot inhibitors were able to progress rapidly to clinical studies and were shortly thereafter acquired by Gilead. The ACC hotspot inhibitor is currently in Phase 2b studies for NASH having shown striking effects in smaller studies.null
The success of the ACC program led our founders to consider how regulatory hotspots could be uncovered systematically. Our team has developed a platform leveraging computer-based and experimental tools to identify new protein sites and design molecules that serve as the basis for new drug discovery projects.
Machine learning approaches are used to uncover proteins that are regulated through specific control mechanisms and chemical modifications. The substrate for machine learning is the entire corpus of biomedical literature comprising 30 million publications, which are evaluated and retrained by a panel of experts. Protein structures derived from x-ray structures are then used to pinpoint the regions of proteins involved in regulatory function. HotSpot has uncovered approximately 200,000 pockets each of which has been scored as a regulatory binding site. Each pocket is then evaluated for known links to protein function using a broad range of mutational databases, and its suitability for small molecule drug targeting is determined.
The list of in silico-ranked pockets has led to important insights for our team. We observe groupings of pockets by structure function, insights within and across target families and stand-alone pockets with previously unknown function.
HotSpot is focusing on a number of high-value target classes that have been under-exploited by current chemical matter and drug discovery approaches. For example, within the kinase class, HotSpot is focusing on the highly validated but unsuccessfully-drugged targets S6 Kinase (S6K) and PKC-theta.
Unfortunately, it is clear that conventional chemistry libraries focused on active sites are not going to be successful for regulatory hotspots. A recent analysis performed by our team indicates that only 3% of commercially available compounds are predicted to interact with regulatory regions. Given this, we are developing our own library of chemical scaffolds biased towards the unique properties of hotspots along with specialized assays to assess hotspot binding in vitro and in cells.
Our strategy is succeeding. Our scientists have delivered the first and only allosteric inhibitors for a series of important proteins, including S6K and PKC-theta. Initial in vitro and in vivo pharmacology experiments indicate that hotspot inhibitors are indeed structurally differentiated compared to traditional active site inhibitors and can be exploited using specially designed chemical libraries to identify molecules that specifically bind the hotspot.
Taken together, our approach uncovers unexploited parts of proteins and then generates starting points for chemistry. Our platform, called SpotFinder™, delivers target-molecule pairs for biological pathways that have remain inaccessible to date. We are now positioning our programs to move rapidly into clinical trials.