Many drug discovery researchers think smaller these days, specifically about fragments. “A big trend more specific to small molecule drug discovery is fragment-based drug design (FBDD),” said Molly He, Ph.D., senior scientist at Sunesis Pharmaceuticals (www.sunesis.com). “It has gained increasing popularity among the pharmaceutical and biotech companies because of its ability to access large chemical space using small amounts of compounds—it’s a new paradigm for drug discovery.”
Fragment screening and targeting protein-protein interactions generated considerable buzz at the CHI “Drug Discovery Chemistry 2007” conference. Harren Jhoti, Ph.D., founder and CSO of Astex Therapeutics (www.astex-therapeutics.com), noted that a key reason for the emergence of fragment-based drug discovery is its sheer practicality.
“We only need to have a modest collection of fragments—roughly 2,000; but because they are so small, between 150 and 200 daltons, they can sample so much chemical space. The larger the fragment size in your library, the larger the number of compounds that you are going to need to sample the same amount of chemical space.”
Astex uses a number of biophysical techniques as screening tools including high-throughput x-ray crystallography in its fragment-based discovery approach. “X-ray crystallography is an intelligent methodology and has the advantage of giving the best picture of how the fragment sits in the target, and when paired with NMR allows you to create a structure-based lead,” noted Dr. Jhoti.
Astex’ discovery program utilizes high-throughput x-ray crystallography and NMR to screen fragment libraries. According to Dr. Jhoti, the company has proven that this approach for lead generation has distinct advantages over conventional bioassay-based, high-throughput screening in that low-affinity fragments with novel structures can be identified as starting points for hit-to-lead chemistry.
“These fragment hits can then be rapidly optimized for potency and DMPK properties using iterative cycles of medicinal chemistry and structure-based drug design,” Dr. Jhoti continued. “Our fragment-based drug discovery approach, which we call Pyramid, enables us to handcraft molecules with regard to potency and selectivity, and consequently our molecules tend to be more compact. There is a clear correlation between the size of the molecule and the success of its development. Compounds that are too large tend to be too complex to succeed and suffer high rates of attrition in development.”
Astex has developed lead compounds using this approach for targets such as the cyclin-dependent kinases and aurora kinases, both of which are key proteins involved in cancer. AT7519 and AT9283 were identified using Pyramid and are now being tested in clinical trials as potential anticancer therapies.
“To the best of my knowledge, we are the only company to go from fragment to compound in the clinic twice,” said Dr. Jhoti. “In addition, we have a third compound that we expect to go into the clinic next year. Having identified five drug candidates in four years, it should be noted that all five compounds started as fragments. We see that as reasonable productivity.”
SGX Pharmaceuticals (www.sgxpharma.com) has also developed a fragment-based drug discovery platform that utilizes high-throughput x-ray crystallography for lead identification and optimization. Stephen K. Burley, M.D., D.Phil., CSO, talked about SGX523, a MET inhibitor that, he says, is highly selective, orally bioavailable, and intended to exploit a targeted approach to cancer treatment.
“This is a hot target. Activating MET mutations, MET gene amplification, and aberrant MET activation are implicated in a huge number of human cancers,” said Dr. Burley. The MET protein controls cell growth and motility and is thought to play an important role in metastases of many tumors.
Dr. Burley also presented SGX’ structure-guided, fragment-based approach known as FAST™ (fragments of active structures). Each member of the FAST fragment/scaffold library was selected to be amenable to rapid chemical elaboration at two or three points of chemical diversity using parallel organic synthesis.
Initial lead optimization involves using knowledge of the cocrystal structure of the target-fragment complex and advanced computational chemistry tools to guide synthesis of small-focused linear (1-D) libraries. These linearly elaborated fragments/scaffolds are then evaluated with in vitro biochemical and cellular assays and cocrystal structure determinations. Active compound series are prioritized for further medicinal chemistry and compound development efforts using the results of in vitro and in vivo ADME and toxicology studies.
“Our structure-based drug discovery approach to MET yielded SGX523, which has exquisite selectivity for MET over all other protein kinases tested. Not only is it orally bioavailable and of low molecular weight,” said Dr. Burley, “but it also has excellent ADME characteristics and in vitro safety properties, covers virtually all of the clinically characterized MET mutants, and was well-tolerated in animal studies.” SGX expects to file an IND for SGX523 by Q1 2008, and begin clinical studies shortly thereafter. “We believe this compound has the potential to be a best-in-class MET inhibitor.”
NMR Fragment Screening
As a discovery tool, Bill Metzler, Ph.D., associate director, macromolecular NMR, Bristol-Myers Squibb (www.bms.com), says that NMR-fragment screening is not only easy to implement, but it also can provide detailed information such as binding-site locations not readily obtainable by other biophysical methods.
“Applying fragment-based approaches to protein-protein interactions is an area that we’re just delving into,” said Dr. Metzler. “Protein-protein interactions is a huge area of potential targets, many of which have been validated already. While identifying inhibitors of protein-protein interactions has been refractory to standard screening methods, this area lends itself to a fragment-based discovery approach. One just needs to be judicious when selecting which targets to work on.”
One of the two programs he described involves inhibiting the active transport of proteins into the nucleus, which is facilitated by the interaction of the nuclear localization sequence of the imported protein with karyopherin. A hypothesis-driven fragment screening approach was used to identify submicromolar inhibitors of the binding of karyopherin to the transcription factor NF-kB.
In the second program, his group initiated studies to identify small molecules that could block the engagement of the T-cell receptor CD28 with its B-cell counter-receptors, CD80 and CD86, thereby preventing T-cell proliferation and inducing antigen-specific unresponsiveness.
“A fragment screen of CD28 identified numerous small molecules that bound CD28 with mM affinity,” said Dr. Metzler. “To improve affinity, we built a homology model of CD28 and performed a virtual screen biased by the fragment hits. After subsequent deck mining several small molecules were identified that have improved the albeit still-limited ability to block CD28 binding to CD80 and CD86.”
“The main lesson that we learned when applying fragment-screening methods,” said Dr. Metzler, “is that structural information is absolutely critical to fragment-to-lead progression. Second, we cannot be misled by a compound’s initial potency. We need to think smaller—small additions to a compound can grant big returns, whereas larger additions often lead to poor pharmacological properties.”
Targeting Protein-Protein Interactions
With a focus on discovering small molecule agents that inhibit Prolexys Pharmaceuticals (www.prolexys.com) has identified novel and druggable therapeutic targets for cancer and asthma. “Prolexys is focused on discovering protein-protein interactions,” said Sudhir R. Sahasrabudhe, Ph.D., CSO and scientific founder.
“We created two technology platforms—HyNet, an advanced yeast two-hybrid based approach to identify binary interactions between proteins, and HySpec, a protein science and mass spectrometry based workflow that permits identification of individual proteins within a protein complex. By using yeast-based and biochemical high-throughput screens, we have identified small molecules that have been tested in cell and animal models of disease.”
The company reported results of lead programs in two areas: small molecules that inhibit the protein interactions in the beta-catenin pathway and protein interaction modulators in the HSP20 pathway. Its primary screen was a high-throughput assay using 14-3-3 gamma protein and a pHSP20 peptide coupled to a fluorophore,” said Dr. Sahasrabudhe.
“We screened over 58,000 compounds, and 268 of those tested caused 20 percent or greater inhibition of pHSP20 peptide and 14-3-3 gamma interactions. We retested those for activity and further evaluated to see if they behaved in a dose-responsive manner. After that analysis, seven scaffolds were selected for secondary screening.”
Dr. Sahasrabudhe’s group found that a fluorescence polarization assay provided a high-throughput method for screening pHSP20 agonists. Cell-based magnetic-twisting cytometry and traction microscopy (performed in collaboration with Jeff Fredberg’s laboratory at the Harvard School of Public Health) provided sufficient throughput to be used as a secondary screen.
Compounds that qualified through this cell-based secondary screen were tested in ex vivo assays that measured bronchodilation in bronchial ring tissue samples.
Prolexys is working on colon cancer. The company generated drug discovery targets for its beta-catenin program by identifying a beta-catenin protein interaction network derived from the initial list of 59 Wnt pathway genes identified from Biocarta, KEGG, and information available in public domain.
“We validated our targets by over-expressing (using cDNA) or suppressing (using RNAi) the candidate genes in cells that were engineered to respond to beta-catenin activation,” said Dr. Sahasrabudhe.
“Our target product profile is one that exhibits tumor cell selective toxicity, specifically inhibits Wnt pathway activation, and causes growth inhibition of human tumor cells implanted in mice. We have a good proof of concept, and we anticipate having a clinical candidate by the fourth quarter of next year.”
Inhibition of TNF-alpha
Sunesis Pharmaceuticals has been studying small molecule antagonists of protein-protein interactions such as the interaction between TNF-alpha and TNF-alpha receptor that is involved in autoimmune diseases including rheumatoid arthritis and Crohn’s disease.
Dr. He presented a mechanism of inhibition where a small molecule inhibitor disrupted the TNF-alpha trimer and disabled the TNF-alpha molecules from binding to its receptor. “Up until now, viable leads for analogous small molecule inhibitors of TNF-alpha have not been reported. Such drugs, with attendant advantages in manufacturing, patient accessibility, administration, and compliance, would represent a major advance in the treatment of TNF-alpha mediated diseases.”
Dr. He’s group described a small molecule TNF-alpha inhibitor initially identified from an in vitro screen for compounds capable of disrupting TNF binding to TNF receptor R1. “By using x-ray diffraction data collected at beam line 7–1 at SSRL,” said Dr. He, “we were able to determine a cocrystal structure of TNF-alpha with the small molecule inhibitor.” She noted that the structure revealed a novel interaction in which one of the subunits of the TNF-alpha trimer is replaced by the small molecule.
“The resulting TNF-alpha dimer retained the same basic structural subunit fold as in the native trimer, but the angle between the subunits within the dimer was slightly widened,” she said. “Interestingly, a large fraction of the contact surface with the small molecule involves six tyrosine residues from the TNF-alpha dimer.”
According to Dr. He, the described results should enable the design of assays that may allow for the identification of potent small molecule inhibitors that inactivate multimeric proteins via a rapid predissociation-independent subunit dissociation process.