The search for low-dose safe medicines

Since the early 2000s, researchers at Novartis have been exploring the therapeutic utility of covalent drugs in target-directed drug discovery projects. These powerful and permanent protein modifiers can be highly effective. But the deliberate use of this class of molecules has long been limited due to safety concerns – concerns that were often interpreted as a reluctance of drug discovery teams to include them in their project strategy. Nonetheless, interest in the area was renewed when a publication from Ben Cravatt’s lab at The Scripps Research Institute in California highlighted the potential use of this long-contested class of molecules. Although the research initially focused on their use as screening tools, the results had far-reaching impacts on the field of medicinal chemistry. “The whole field got a boost from that publication,” says Jeffrey McKenna, a longtime medicinal chemist and molecular library specialist at Novartis Biomedical Research. “Interest at the company had been growing for about a decade and that article showed that this approach was a technological advance that would be a necessary strategic addition for Novartis.” As a result, McKenna and colleagues reacted quickly and developed a dedicated screening library consisting of only covalent molecules, which Novartis approved in 2017. Starting with the roughly 2,000 covalent molecules that were already part of the company’s total screening library, McKenna and his colleagues have since grown this specialized library to over 15,000 compounds. McKenna continues to grow it by about 1,000 molecules per year. A second component to this effort was the initiation of a collaboration with the University of California, Berkeley, spearheaded by McKenna. This partnership aimed to combine the resources of Novartis with the state-of-the-art and exploratory research from academia to understand how this technology directed at interrogating the undrugged target landscape could be applied to drug discovery in its widest fashion. “When we saw that covalent molecules could also be used to find unique pockets on targets that you otherwise couldn’t find druggable sites for, that really spurred the field and technology as a whole to expand,” explains McKenna. To date, this collaboration has yielded over two dozen scientific publications, supporting the work of over 30 students and post-doctoral researchers. These insights have also been instrumental in shaping the course of covalent drug discovery at Novartis. In recent years, the Novartis covalent library has been used to screen against dozens of potential targets. Here, McKenna shares his thoughts on the field of covalent drugs and what it took to create this screening resource.

Starting with the roughly 2,000 covalent molecules that were already part of the company’s total screening library, McKenna and his colleagues have since grown this specialized li­brary to over 15,000 compounds. McKenna contin­ues to grow it by about 1,000 molecules per year.

Could you tell us a little about your own background and how you came to be in this field? Although my initial career plan was to be an army officer, an injury and a mindset change prompted me to follow my passion for chemistry. After my first work experience in the pharmaceutical industry, where I was focused on the synthesis of new potential drug molecules, I went back to school to get my doctoral degree in chemistry from Oxford. Since then, I’ve worked for over 25 years as a medicinal chemist, including about 20 years at Novartis. I’ve had a variety of roles at Novartis and I’ve also worked at four different sites – New Jersey in the United States, Horsham in the United Kingdom, Cambridge, Massachusetts, in the USA, and now finally Basel, Switzerland. What happened about 20 years ago that shifted people’s views on covalent drugs? Covalency, per se, isn't new, but it's had a little bit of a resurgence in the last 20 years. Historically, there have been covalent medicines, like aspirin, penicillin, and indeed many more recent examples that are a key component to the drug pharmacopeia that we all use regularly. However, what is not widely appreciated is that these were often discovered fortuitously and certainly not developed by intentional design. There were a lot of technological innovations in molecular biology, protein science, mass spectroscopy, and screening technologies from the early 2000s forward that opened whole new avenues for drug discovery across the board. For the first time, the field had the resources to look at drug-protein interactions at an atomic level. This was a huge benefit for noncovalent drugs, which received most of the attention, but these tools were equally useful for designing covalent drugs and things began moving forward on that front as well. What developments led to the creation of a dedicated covalent molecule library at Novartis? For me, it actually started with the 2016 publication from Ben Cravatt’s lab at The Scripps Research Institute in California. Cravatt’s group essentially combined the logic from two areas of drug discovery into a single new approach. Combined, these approaches created a strategy that could find unique binding sites on proteins that were generally considered undruggable. Covalency was key to its success. This approach made it possible to map binding sites on proteins and to open the access to target classes that we had otherwise thought were out of reach. Cravatt’s lab was the first to show that covalent fragment-like compounds could offer a whole new set of insights for drug targets.

Jeffrey McKenna, Director at Novartis Biomedical Research.

What happened next? I had been involved in evaluating this technology that we foresaw to be of broad utility at Novartis and I needed a collection of covalent compounds as a source set to screen. At the same time, my colleague, Christian Markert, was interested in generating a covalent library for use in an internal program he was working on. The two projects dovetailed and so we put together a proposal in which the convergence of these interests made a compelling case for funding. For me, this resulted in a close and ongoing collaboration with the University of California, Berkeley, to further develop this chemical mapping strategy for new protein targets. We chose Berkeley because we already had had ongoing collaborations there, and also because an alumnus from Ben Cravatt’s lab, Dan Nomura, was leading a covalent chemo-proteomics group there. It also meant that we were able to start to build out an internal Novartis covalent library with the intent that it would be screened across numerous targets and disease areas. How do you pick which molecules to add to the library? To begin with, there was a small core of covalent molecules already in the corporate library, because people had purchased or made them for previous projects. When we started actively designing, purchasing, and synthesizing new additions for the library, we looked to include a sensible sampling of chemical diversity. This has meant triaging and choosing a wide variety of compounds to expand the library: The key considerations are the molecular shapes they describe, which are determined by the scaffolds, the properties of the molecules, which are dictated by the types of atoms included within them, and lastly, the intrinsic chemical reactivity of the molecules that overall drove a selection that made sense. Can you explain in more detail? Unlike reversible noncovalent inhibitors, covalent inhibitors form permanent chemical bonds to their target proteins through a chemical reaction. There are a variety of potential reactions and reaction partners that we can strategically include in the library. Coupled with this overall idea, we were also particularly, but not exclusively, interested in those molecules containing reactive groups that could be targeted by cysteine: This is a reactive amino acid in most of our proteins. To get just a little more scientific for a moment, why is cysteine so important? Cysteine is rare within our proteome – it makes up only two to three percent of all the amino acids in our proteins. It controls numerous cellular processes, and cysteine is one of the more reactive amino acids under natural conditions in our bodies. We designed the library to include a large number of molecules with “warheads” capable of potentially forming a bond with cysteine. There are a handful of other amino acids that are also reactive, which we may harness when needed. Most of the library is dedicated to potential cysteine binders, but we’ve included covalent molecules that are compatible with these other amino acids as well. How have safety considerations been addressed in the design of the covalent library? With time and experience, we developed a number of tests to help us navigate the threats of toxicity due to molecules that are just too reactive. A hyperreactive compound can mean that a molecule isn’t specific about which proteins it binds, or that a molecule is so reactive that it won’t even survive a biological system long enough to reach its intended protein target. When we consider new covalent molecules as potential additions to the library, we use this set of standard tests to profile the reactivity of the molecule. If the molecule turns out to be too reactive, then we steer away from it. The collaboration with the University of California, Berkeley, has been central to these efforts since the start. Could you say more about this ongoing collaboration? The potential of covalent molecules to find new druggable binding sites on proteins really spurred the whole field and technology forward. Our collaboration with Berkely allowed Novartis to support and adopt the most state-of-the-art exploratory developments in this technology. With their help, we uncovered several unique modalities that we wouldn’t have thought were compatible with a covalent approach. That has additionally spawned interesting insights and different ideas that we continually review and look to apply in unique ways to enhance projects and the Novartis portfolio. What do you think is on the horizon for covalent drugs in the future? I think one of the greatest potentials is the opportunity to target proteins and protein classes that we haven’t been able to target with conventional modalities. There’s also great promise in these new areas to achieve the perfect scenario for the use of covalent drugs, which is to separate dosing frequencies from the duration of the compounds’ effect and to achieve this at very low doses of these medicines. This is equally valid for more traditional targets too, where low doses and high specificities are required. In simple terms, covalent drugs have an advantage in that they only need to engage their target once – after the drug reacts with the protein, it’s not going anywhere, so to speak. After that, there’s no need to give another dose until the protein itself is cleared from the body and a new protein resynthesized in its place. The resynthesis rate of the target protein is a key determinant of when to use covalency in drug discovery. This minimal dosing strategy, combined with the high specificity that we can design into these molecules, really provides an opportunity for highly safe and effective drugs for therapeutic areas with high unmet need.

Research Pursuing potential. The search for low-dose safe medicines Interview K.E.D. Coan. Photos Annick Ramp. Interview with Jeffrey McKenna, Director at Novartis Biomedical Research.