Structure-based in silico modeling has quietly become the leading approach used in compound screening and lead optimization by pharmaceutical companies and contract research organizations.
In a horse race, the premise is simple: the best runner usually takes the lead and wins the race. When it comes to drug leads, the same premise applies, especially when the develop-ability profile is enhanced by the lead optimization process. Develop-ability of a compound can mean several things: whether the compound can be developed into an oral dosage form, whether or not it has good bioavailability, whether or not it has a good toxicity profile, etc.
One company that has focused much of its services and technology products around characterization of the ‘develop-ability’ of the compound is Symyx Technologies (Sunnyvale, Calif.). “The reason that Symyx is interested in this area is that we have our expertise in microscale experimentation, which allows you to get much more data even with limited amounts of compound,” says Stephen Cypes, director of life science research at Symyx. He adds that this exact scenario frequently occurs at the lead optimization stage, and, although most Symyx customers might perform the chemical modifications to optimize their leads themselves, “Symyx could assist in the process of making sure those modifications have enhanced some properties that make it more beneficial from a develop-ability standpoint to get it into a marketable dosage form.” The services Symyx performs to optimize limited quantities of active pharmaceutical product (API) include tests of solubility versus pH and excipients, its susceptibility to degradation under different stress conditions (including thermal, oxidative, and photolytic), ability to be formulated with excipients, ability to crystallize and characterization of polymorphs, ability to be formulated in an oral dosage form, salt selection studies, and more.
click to enlarge
EPIX computational model of transmembrane portion of mGluR5 negative allosteric modulator (NAM) with known ligand, Fenobam. The model was used during lead optimization to identify preclinical development candidate EPX-105287, synthesizing only 36 compounds. (Source: EPIX Pharmaceuticals) |
|
“These services can be performed in late discovery phase, so the customer can feel confident that the compound that they are taking into the clinic is something that can be delivered into humans,” says Cypes. “Without the microscale approach, there is the risk and what you’ve developed in solutions may not be able to be formulated into a commercial form.” Symyx encourages their clients to conduct lead optimization earlier in the process because its microscale technologies can work on very little active pharmaceutical ingredient (API).
Byeong Chang, PhD, the chief scientific officer of Symyx Technologies, focuses on the formulation of lead biologicals. “Typically, lead optimization is done after an engineering batch is produced, primarily because formulation studies require a substantial amount of drug substance, and because impurities may interfere with the accurate assessment of intrinsic stability of the product,” says Chang. “In this scenario, a formulation will be ready for preclinical studies.”
As a contract research organization (CRO), Charles River (Wilmington, Mass.) performs lead optimization services and proof-of-concept studies for pharmaceutical and biotech companies. Focusing mostly on the drug metabolism, efficacy, and toxicity aspects of the lead optimization process, Charles River typically optimizes a client’s lead using relevant animal models of major human diseases, such as mouse models of human cancer for testing novel oncology compounds, says Alain Stricker-Krongrad, PhD, chief scientific officer of Charles River. Endpoint analysis in the cancer models typically involves the use of an imaging technology such as MRI, bioluminescence, or PET scan to measure tumor growth, angiogenesis, or biochemical activity. It can be combined with quantitative analysis of biomarkers. For each therapeutic indication, these technologies are used to provide a seamless transition to development, expediting the process of bringing better drugs to the clinic.
A high-throughput surface plasma resonance (SPR) system from FujiFilm Life Science (New Haven, Conn.)—the AP-3000—can be used for label-free analysis of protein-small-molecule interactions. Designed to be utilized at the front end of lead optimization in a secondary screening role, the AP3000 is used “to select a subset of the chemical library hits that are most amenable to further lead optimization processes,” says Don Janezic, business development manager at FujiFilm Life Science. “So what we are able to do in lead optimization is to help [our pharmaceutical clients] focus the large number of hits that they’ve discovered in functional assays, and bring them down to a more manageable number based on their performance in a high-throughput, label-free direct binding assay.” The technology is intended to be used in conjunction with a traditional high-throughput screening approach.
click to enlarge
Sapient Discovery’s Drug Discovery process: The target’s three-dimensional structure is from X-ray crystallography or homology modeling, followed by binding/active site analysis. Virtual screening or fragment-based screening is utilized to find initial hits. These are assayed and then co-crystal structures determined to confirm the mechanism of binding. (Source: Sapient Discovery, LLC) |
|
According to Janezic, the AP-3000 is capable of analyzing ten 384-well microtiter plates loaded with compounds within a 24-hour period. Also, the system is sensitive enough to detect protein-small-molecule interactions with compounds at the low-hundred Dalton range of molecular weight, a feature that is advantageous for fragment-based drug discovery approaches, a frequent application of the system. Analysis is performed without the need to label the compound, thus limiting interference with protein-small-molecule interactions that often leads to false-positive and false-negative results. The types of analysis performed by the AP-3000 include structural affinity analysis (i.e., the affinity, or KD, of the compound for the target), on-off rates and steady-state affinity analysis. Compounds are ranked in order of increasing affinity for the target. “While some lead optimization is focused on a few compounds, when you start looking at a lead optimization process in medicinal chemistry or structure-based design groups, they care about structure-activity relationships of a group of compounds,” says Janezic. To pick winning lead compounds, much of the industry has implemented structure-based in silico approaches in its drug screening and optimization processes.
In silico modelers
FORMA Therapeutics (Cambridge, Mass.) is one of many companies that focus on a structural biology approach to drug discovery. The structure-guided drug discovery process has two prongs—X-ray crystallography followed by computational analysis—based on FORMA’s proprietary platform called CS map, which is a fragment-based computational method. Steve Tregay, PhD, founder, president, and chief executive officer of FORMA describes the inner workings of this technology. A probe set of fragments is allowed to minimize across a protein structure surface, which is typically performed across 30 or 40 fragments. Then, they search for consensus binding of those fragments across that protein surface. “What this allows us to do is to very carefully map out hot spots on the protein surface and identify the active sites, the allosteric regulatory sites, and broad open surfaces like protein-protein interactions that will allow us to identify key hot spot interactions within a much broader protein-protein interaction surface,” says Tregay. “This structure-guided analysis of protein surfaces and associated hot spots provides a rational process for target selection, small-molecule discovery, and optimization.”
click to enlarge
Outsourcing tasks for lead optimization and drug formulation may help move promising drug candidates to clinical trials faster. (Source: Symyx) |
|
Another company that subscribes to the use of an in silico modeling platform technology for lead optimization is EPIX Pharmaceuticals (Lexington, Mass.). In terms of workflow, use of the platform has been applied to G-protein-coupled receptors or ion channels in advance of the lead generation phase. It is utilized during lead optimization to examine and optimize the virtual binding characteristics of lead compounds for both target and off-target properties (e.g. selectivity, ADME). The platform incorporates the combination of medicinal chemistry and computational approaches. “We use the computational models to test hypotheses during lead optimization in a very dynamic and iterative process,” says Sheila DeWitt, PhD, vice president of discovery and manufacturing at EPIX Pharmaceuticals. “Every compound we make is assessed both by medicinal chemistry and computational chemistry to try to limit the number of compounds we make to address specific questions while improving the overall therapeutic profile.”
“During lead optimization, we’re trying to optimize properties in a limited amount of compounds in the shortest amount of time, leveraging what we’ve learned during lead generation,” says DeWitt. “And that has allowed us to shorten the timelines in lead optimization from the industry-standard.” In a typical in silico drug screen, these computational receptor or channel-based models are used to filter and prioritize between 100,000 and 300,000 compounds from a library of four million compounds that are commercially available from a variety of vendors. Once prioritized, 100 to 300 compounds are purchased and tested, and, from that reduced number of compounds, hits are identified and then become the starting point for hit-to-lead activities including identification of a lead scaffold. Although docking of lead compounds occurs in silico, determination of hits occurs in in vitro, receptor, or channel-based assays, which are typically performed by CROs. During lead optimization of a lead scaffold, approximately 10 percent of virtually designed compounds are prioritized for synthesis and actually tested in vitro. Testing a significantly smaller number of compounds not only reduces the timeline, but there is also a large cost-savings in using an in silico modeling approach.
In a third example, drug discovery efforts at Sapient Discovery, LLC (San Diego, Calif.) are founded on its virtual-screening platform in which a series of X-ray crystallographic structures is analyzed for sequence homology for the novel target. The homology-based model for the novel target can be used to perform virtual screening. From a virtual compound library of greater than 20 million readily available compounds, Sapient Discovery selects between 300 and 500 compounds to purchase and then performs a biochemical assay to screen for compounds that bind in at least the micromolar range, thus completing the lead generation process. Sapient then partners with larger pharmaceutical companies, as well as biotechnology companies and virtual companies, to continue to optimize hits from their leads, and then to further develop those hits into clinical candidates.
click to enlarge
Scientists in Symyx’s Contract Development and Manufacturing Organization develop drug formulations using advanced scientific informatics and high-throughput experimentation technologies. (Source: Symyx) |
|
“We prefer to do co-crystal structure determination on the lead that we generated to find out exactly where the compound binds on the protein surface or in the protein active site,” says Kal Ramnarayan, PhD, president and chief scientific officer of Sapient. “Developing a co-crystal structure is the first thing that we do to help to speed up the optimization process because, by then, we are concerned with whether or not the compound binds to the protein of interest.”
Like many industry experts, Ramnarayan believes that lead optimization should begin as early as possible. “Lead optimization combined with co-crystallization studies should, in my opinion, begin immediately, once the biochemical assay and functional assay have validated the compound, so that you can reject compounds that are likely to be problematic before they go into animals,” he says. And, with structure-based in silico approaches performing virtual high-throughput screening, early elimination of poorly developable compounds has become the rule, not the exception.
Summary
Lead optimization is a necessary process that, like other processes in drug discovery, has followed two major trends—increased outsourcing and increased use of virtual methods, both of which promise to significantly reduce drug-development costs. Lead-optimization services cut costs by providing the necessary expertise and technology on a contractual basis, while the use of virtual screening methods and structural data on the target or compound has simultaneously reduced the number of compounds screened and costs.
About the Author
James Netterwald is president and CEO of BioPharmaComm LLC, a provider of writing, editing, and consulting services to the life science, pharma-biotech, and public relations industries.
