Researchers are increasingly relying upon fragment-based lead discovery in their search for novel and effective drug candidates. The approach utilizes a number of complementary technologies, with crystallography and structure-based drug design being two of these key techniques.
Last month, “Fragment-Based Lead Discovery 2008” was held in San Diego. “It was a great opportunity for all the pioneers in the field to get together and talk about where the research is going,” said Vicki Nienaber, Ph.D., CSO of ActiveSight and one of the conference organizers.
While fragment-based screening has been around for sometime, the approach taken by Plexxikon only proved feasible when it became possible to merge high-throughput screening with high-throughput crystallography, noted Rick Artis, Ph.D., vp, lead generation.
“It’s not just about functional biology approaches since the search for low-affinity hits using assay technology is fraught with disaster,” continued Dr. Artis. “By coupling that technology with crystallography, you get a relatively few high-quality true positives—and even then, to show that you have something useful, you have to validate them via chemistry.”
Plexxikon’s solution was to build a scaffold-focused screening library at the core of its lead discovery platform. “The key element was that we didn’t focus on small fragments,” said Dr. Artis. “We focused on somewhat larger molecules as potential scaffolds and that middle space had a profound impact on our program.”
Plexxikon’s scaffold-based drug discovery approach has yielded two clinical programs to date, in diabetes and oncology. In each case, the time from initiation to first-in-human study was less than two years. “In part, the success of these efforts is derived from the ability to pick starting points with tractable enabling chemistry in the structural context of a given binding site,” said Dr. Artis. “With this structural analysis in place, the subsequent design strategy focused on generating compact molecules with high atomic economy.”
Dr. Artis noted that the quality of compounds generated from this process has allowed for early pharmacokinetic screening and resulted in generally favorable properties when compounds are introduced to in vivo profiling.
“We really hit the sweet spot of this implementation, but we were fortunate in that we chose correctly in the first days of the company,” he said. “You can judge the platform by the compounds that come out of the program and into the clinic—that’s a key measure of how productive your platform is.”
Another key feature is that this platform is therapeutically agnostic. Plexxikon looks at the discovery opportunities not based on a particular disease, but on the protein targets that provide the best prospects for the approach in a variety of different areas.
“From a therapeutic perspective, we’d be walking away from too many potential drugs,” said Dr. Artis. “We decided to focus on bringing the best drug candidates forward and building relationships and partnering with companies with the clinical expertise to develop them.”
There are a number of approaches to screening and identifying good lead matter, but, according to Siegfried Reich, Ph.D., vp of SGX Pharmaceuticals, “none is in its own right a panacea. Fragment-based approaches offer a complementary and powerful way of identifying quality, ligand-efficient leads. Integrated into an overall drug discovery process that includes iterative structure-based design and a focus on retaining intrinsic favorable drug-like properties of the original hit, viable development candidates can be discovered.”
“Essentially, every company that presented at the meeting had a different spin on screening,” Dr. Reich reported. “Our approach involves rapid screening of a 1,400-member fragment library at the advanced photon source in Chicago where we have employees on the ground providing high-throughput and high-resolution crystallographic data for our projects.”
SGX’ system, termed FAST (fragments of active structures), allows rapid crystallographic screening of its fragment library against oncology targets of high therapeutic value as well as iterative structure-based design. Dr. Reich described several examples of the application of FAST toward the identification of advanced leads targeting BCR-Abl, HCV Pol, and MET.
“Based on information from the crystallographic screen, we can assess which are the most interesting leads with the highest potential for optimization,” Dr. Reich said. “But, of course, that leads to another aspect of fragment-based lead discovery; the fragments are small, consequently the initial hits have weak affinity. This is where the crystallographic information comes in. Knowing specifically how these compounds bind allows for their rapid optimization in most cases. And by carefully selecting the molecules for your library to have excellent lead-like properties and diversity in the first place, considerably smaller libraries can be employed.”
Roderick Hubbard, Ph.D., senior fellow at Vernalis and professor at the University of York, has been developing methods to analyze protein structure and designing ligands since the early 1980s.
Over the past six years, the team at Vernalis has developed, refined, and applied fragment-based methods in a number of structure-based discovery projects, said Dr. Hubbard. “This has required bringing together methods in high-throughput crystallography, computational chemistry, and NMR spectroscopy and integrating the approach with our medicinal chemistry program. This combination has helped us to discover and design compounds that open up new areas of drug discovery.”
Fragment-based drug discovery, Dr. Hubbard noted, is an ideal space for smaller pharma. “High-throughput screening requires a huge investment in facilities and compounds. With fragment screening, you can get by with smaller libraries. Another benefit is that its very nature allows you to sample a large chemical space quickly. And a third benefit is that it gives you an opportunity to find hit compounds for new classes of targets.
“HTS relies on your compound collection containing exactly the right molecule that binds with high enough affinity to your target. The smaller size of fragments increases the chance you will find a suitable chemical start point to develop a new hit compound.”
According to Dr. Hubbard, Vernalis combined its medicinal chemistry experience with structural science to develop a fragment approach. “We put a lot of thought into the design and curation of our fragment library,” he said.
His group was also pragmatic about finding which fragments bound to which targets. “We have a process in place in which we use NMR to see which compounds bind competitively to a binding site. This binding is then cross-validated using other biophysical techniques such as surface plasmon resonance and HSQC NMR measurements.
“This confirms which fragments are binding to the protein so that we can concentrate on trying to determine the crystal structure of the fragment which provides the all-important structural information that can guide the evolution of the fragments into novel hit compounds.”
“It’s time to start looking at the next steps,” said Dr. Hubbard. “These include improved characterization of library content, design of novel fragments, decision support for fragment evolution, and targeting protein-protein interactions.”
Fragment-based drug discovery is not limited to protein-protein interactions. Christophe Verlinde, Ph.D., associate professor at the University of Washington (faculty.washington.edu/verlinde), has been working with a team of crystallographers in conjunction with the Medical Structural Genomics of Pathogenic Protozoa (MSGPP) consortium to solve the crystal structures for various parasites. “Our ultimate goal is not just to solve the crystal structure—we are also looking for the next generation of drugs,” he said.
To that end, Dr. Verlinde has carefully selected a collection of nearly 700 fragments from the MDL ACD database. From that, his team created 68 cocktails of 10 compounds that are shape-wise diverse. “We have explored the utility of these cocktails for initiating lead discovery in structure-based drug design by soaking numerous protein crystals obtained by the MSGPP consortium.”
Dr. Verlinde also discussed the history of his methodology and reported on fragment selection and cocktail design procedures and gave examples of the successes his group has obtained. “One basic problem is that you can see a blob of electron density, but the trick is to figure out which molecule from the cocktail fits.”
One of the findings from Dr. Verlinde’s lab is a parallel ligand discovery method that works by differential thermal unfolding of the target protein. “If you have a protein at a certain temperature it will melt or unfold. If you do it in the presence of a binding compound, the melting temperature will be higher. We confirm our binders with these methods.”
Putting It All Together
ActiveSight’s Dr. Nienaber, one of the pioneers in the fragment-based lead discovery (FBLD) field, noted that over the past decade technologies have progressed in both the detection of fragment binding and in their optimization to lead compounds, and that biophysical techniques such as SPR and calorimetry are becoming central assets to an FBLD program.
“Furthermore, the line between a traditional FBLD process and optimization of these early leads into clinical candidates is becoming blurred,” Dr. Nienaber said. “Technologies such as parallel synthesis, SPR, and calorimetry may be used hand-in-hand with high-resolution crystal structures to rapidly advance leads through optimization into clinical candidates.”
Dr. Nienaber said that her emphasis will be on emerging technologies and methodologies in the field. “One of the problems with looking at the HTS paradigm is that, by its nature, you need a lot of resources to run the program, so you have to go after targets with large markets.
“With fragment screening, it is possible first of all, for smaller companies to get involved in the drug discovery process. But the other and perhaps more important piece is that it gives everybody an opportunity to address some of the low-incidence populations—such as muscular dystrophy and Parkinson’s and Huntington diseases.
“We are working with the Cure Huntington’s Disease Initiative and have also initiated academic collaborations around other low-population diseases. In addition, our first program with HSP90 is moving forward.”
According to Dr. Nienaber, fragment co-crystal structures may be used to scaffold hop toward the goal of expanding or generating IP or improving the drug-like properties of a late-stage lead series.
“Our LENS technology pulls together multiple biophysical techniques with proprietary software tools to focus synthetic efforts in the transformation of fragment hits into drugs,” she said. “The real advantage to having been there at the inception of the field is that we can take everything we’ve learned and bring it to the next level. It’s been a lot of fun, and we’ve gotten a great team together to move these programs forward.”