Fifteen years ago, the pharmaceutical industry was abuzz: Computational modeling was about to transform the drug-discovery paradigm. Compound modeling and screening in silico would save lives, years of research, millions of dollars in costs and largely put an end to animal testing.1 Yet, when in silico success stories didn’t start pouring in, a kind of general malaise toward this approach fell over the pharmaceutical industry.
"Most of the revolutions that happened in silico,” says Douglas Krafte, Chief Scientific Officer of Icagen, Inc., “happened in the subfields such as biology and chemistry, not in the fully integrated design and generation of new drugs.”2 A 2009 article published on the topic of in silico models in Drug Discovery Today confirms this phenomenon: “Although the development of computational models to aid drug discovery has become an integral part of pharmaceutical research, the application of these models often fails to produce the expected impact on productivity.”3 The inability of these in silico methods to shorten the timetable from R&D to Investigative New Drug (IND) status — as the industry anticipated — made many of the larger pharmaceutical companies hold tight to more classical drug-discovery approaches, implementing computational models only as a type of follow-on when the traditional approaches failed.
Another major shortcoming of the early in silico approaches was the method’s inability to detect off-target pitfalls. Compounds detected as having a high affinity for a specific target, which in silico models had predicted, produced detrimental downstream effects caused either by off-target binding or toxic metabolites.1 The early computer simulations simply weren’t robust enough to test many of the possible downstream outcomes, and so, evaluating the cost-benefit ratio of in silico approaches, many larger pharmaceutical companies decided to continue on the well-trodden path.
Then, also in 2009, the venture capital firm Atlas Venture teamed up with Schrödinger LLC, a New York–based company that primarily developed chemical simulation software and computational models focused on physics simulations, and with additional seed money from Bill Gates created the in silico–based pharmaceutical company Nimbus Discovery. With its expertise in multifaceted simulations for complex chemical structures, Schrödinger — along with top-in-the-field in silico scientists — hoped to customize and expand its software suite in order to make headway in what was then a sluggish in silico drug-discovery domain.
And it did. In April of this year, Gilead Sciences Inc. (best known for its effective hepatitis C treatments) purchased Nimbus Apollo, one of a half-dozen drug-discovery subsidiaries under the Nimbus Therapeutics umbrella, for a deal valued at $1.2b ($400M up front with up to $800M in development-related milestone payments).4 The deal was based on Nimbus Apollo’s successful phase I testing of an acetyl-CoA carboxylase inhibitor, “a master regulator of fatty acid synthesis and oxidation, [that] has been a sought-after yet intractable target over the past two decades.”
Experience Over Teraflops
The success of Nimbus did not go unnoticed by companies like Icagen Inc., a drug-discovery pharmaceutical company known for bringing many successful compounds to IND. In July of this year, Icagen made a transformative move that changed the face of the company by acquiring Sanofi’s facility in Tucson. With the purchase of the Tucson facility, Icagen — known in the industry as the preeminent researcher and developer of compounds that target ion channels — acquired the second-best (only behind Schrödinger) high-performance computational suite for in silico drug discovery.
Although the computational power of the Tucson facility is praiseworthy, “in-house computational power doesn’t mean much these days,” says Anil Nair, director of In Silico Drug Discovery at Icagen. “Why build one house when you can rent houses all over the world? A pharmaceutical company could call up Amazon tomorrow and buy whatever cloud computing power they need for the day, a week, or a year: 200 teraflops, 400 teraflops — computing power greater than anything Icagen or even Schrödinger has in-house.”
What Dr. Nair is alluding to is that advanced software suites and massive computing capabilities are available everywhere today — what makes a difference is experience. Companies like Icagen, with almost a quarter of a century’s experience in drug discovery, are looking at the successes of Nimbus and realizing that by bringing in silico approaches to their existing platform, they can greatly reduce the time it takes to bring a candidate to IND. Some numbers being thrown around at Icagen these days are 24, or even 18 months to IND — a remarkable number considering the standard industry estimation is over five years.6
Besides being a transformative move, Icagen’s acquisition of the Sanofi facility brought “incredible talent, experience and capabilities,” says CEO Richie Cunningham, referring to the top scientists that will remain on-site at the Sanofi facility as part of the agreement. This includes a group of scientists with a very high level of expertise with in silico drug discovery “that positions Icagen as a Target to Lead Generation company at a time when there is a significant gap and need in the industry for quality leads. We at Icagen believe that selecting the right targets of interest based on market needs, combined with targets that fit our areas of expertise, is the key to our success.” What the move also brought about was the Sanofi Tucson facility’s reach in terms of drug discovery. Although Icagen had predominantly focused on ion channels and transporters, the facility at Tucson has a much wider range of targets that Icagen now adds to its repertoire.
The ion-channel drug market is expected to grow to $21.4B by the end of 2018, up from $12B in 2012.
Narrowing the Pipeline: How In Silico Works
In many ways, scientists using in silico approaches for drug discovery have much in common with video game developers. When performing virtual screening, millions of compounds are docked into three-dimensional models of a specific target and scored based on the interactions. Molecular dynamics closely follow target conformation over time. Such simulations can be carried out for protein ligand complexes with solvent and membrane systems. The larger the system, the more the computing power needed and the longer the calculation (some can take weeks). Most steps require state-of-the art computer graphics hardware.
By supporting a drug discovery program in silico from the very beginning, a pharmaceutical company can weed out thousands of compounds that don’t have a strong affinity for the target and compounds with potential liabilities. These include off-target activities, metabolism issues and suboptimal compound physico-chemical properties.
Fewer, stronger candidates means fewer compounds that chemists have to produce for any one specific target. Fewer compounds for any specific target means that a company can partner with a provider like Icagen and, without changing its size or makeup, run more drug-screening programs. A company that had its chemists producing 2,000 compounds for one drug screening program can now (by utilizing the advantages of in silico approaches) have its chemists produce only ~200 well-designed compounds for the same protein target and thus significantly accelerate the discovery program and use the spared resources to take on new programs
Bringing In Silico Approaches To Ion Channels and Transporters
The ion-channel drug market is expected to grow to $21.4b by the end of 2018, up from $12b in 2012.7,8 Some of the most universally difficult targets for in silico approaches are ion channels and transporters, due to challenges in mapping their three-dimensional structures. Ion channels and transporters are integral to many cellular processes, and the too-many-to-list diseases and ailments that arise when they malfunction have caused Icagen to build a considerable suite of traditional drug-screening approaches targeting them. These include an extensive portfolio of cell lines expressing difficult targets such as all the major ion channels and transporters, a label-free X-ray fluorescence method for direct ion flux measurements and assays for drug candidate screening based on several different technologies combined with robotic ultra-high-throughput screening, available at the Tucson facility.
Icagen’s already-established ion channel discovery suite, paired with their now-considerable in silico capabilities, makes it an ideal partner for a pharmaceutical company looking to create compounds for an array of targets for which structural information is available. On the plate now are well-mapped-out structures for targets, whose mutations play a big role in human disease, such as inherited developmental and metabolic disorders and certain cancers. An ambitious client looking to target ion channels and transporters could find no better partner than Icagen.
- Ericson, John. “Breakthroughs Might Mean the End of Animal Testing.” Newsweek. 18 Mar. 2014. Web.
- Unless otherwise noted, all quotes come from phone interviews conducted by the author on August 19, 2016 and September 7, 2016.
- Brown, Scott P., Steven W. Muchmore, Philip J. Hajduk. “Healthy Skepticism: Assessing Realistic Model Performance.” Drug Discovery Today 14.7–8 (2009): 420-427. Web.
- Weisman, Robert. “Gilead to Buy Nimbus Unit for Up to $1.2b.” The Boston Globe. 4 Apr. 2016. Web.
- Nimbus Therapeutics Announces Initiation of Clinical Studies for ACC Inhibitor. Nimbus Therapeutics. 21 Apr. 2015. Web.
- Schuhmacher, Alexander, Markus Hinder, Oliver Gassmann. Value Creation in the Pharmaceutical Industry: The Critical Path to Innovation. New Jersey: Wiley, Apr. 2016. Print.
- Ion Channel Modulators: Market Research Report. Rep. Global Industry Analysts. Mar. 2013. Web.
- Wickenden, Alan D., Birgit Priest, Gul Erdemli. “Ion Channel Drug Discovery: Challenges and Future Directions.” Future Medicinal Chemistry 4:5 (2012). Web.