Nanodesign Optimizes Antiestrogen Leads from Computer-Aided de novo Design

With the emphasis in drug discovery shifting from rational design to high-throughput synthesis and screening, it's refreshing to hear of a molecular modeling success. Score one for Nanodesign Inc. (Guelph, ON, Canada), which has optimized antiestrogen leads (for breast cancer therapeutics) uncovered through its molecular modeling discovery program.

Recent assays suggests that antiestrogens belonging to two chemical classes developed from scratch by Nanodesign are active at pharmacological dosages, validating the ability of Nanodesign's computer-aided design technology to develop novel, active compounds rapidly and efficiently.

In competitive binding assays, all the compounds tested showed significant activity compared to known compounds with high affinity for the estrogen receptor. This new information supports positive results obtained earlier from MCF-7, gene transfection, and cytotoxicity assays. The two novel chemical classes were discovered using Nanodesign's proprietary "evolutionary molecular design" (EMD).

"Our data demonstrates the ability of EMD to generate novel chemical classes with predicted synthesizability and biological activity," says Ian Anderson, Nanodesign president and CEO. "This success confirms the ability of the Nanodesign team to provide optimized drug leads to the pharmaceutical industry."

Evolutionary Molecular Design
Molecular modeling did not realize its potential early on—say in the late 1980s—because computer technology of that era was primitive compared with today's. As computational tools got better, however, combinatorial chemistry and high-throughput screening emerged, replacing modeling as the number-one buzzword in discovery technology.

Of course, modeling and high-throughput methods are not mutually exclusive by any stretch—in fact they're complimentary. Molecular modeling allows scientists to do everything in silico, with all the benefits we normally attribute to computers: multitasking, time-saving, multiple "discovery" projects occurring simultaneously, and virtual "testing" without committing a lot of resources. On the downside modeling is just that—a model, not an actual experiment.

EMD constructs computer models that mimic the mechanism by which drugs interact in the body. Models are then used to generate novel chemical structures. Selected structures are then evaluated for suitability for further development with pharmaceutical industry partners. According to Nanodesign, EMD's benefit is it addresses computationally complex problems of drug design and optimization, such as toxicity and bioavailability, early in the discovery process. EMD is covered by one issued patent (US Patent 5,699,268), plus additional patent filings and trade secrets.

Using structural information and biological activity data from a small set of active compounds (<10) and a limited number of inactive molecules, EMD creates a mathematical construct called a Virtual Receptor (VR). Structural information about the actual biological receptor is not required. The virtual receptor is functionally equivalent to its biological counterpart, and is used as a basis for computational generation of entirely novel structures as new drug leads.

Virtual receptors differentiate biologically active molecules from inactive molecules and are used to transmit relevant structure-activity data to the second component of EMD, the Molecular Assembler, which then generates chemical structures satisfying the design instructions specified by Virtual Receptors. The MA builds structures that are evaluated computationally by Virtual Receptors and screens for bioavailability, chemical synthesis accessibility, toxicity, and others. Candidates that consistently satisfy the design criteria are then moved forward into chemical synthesis and biological profiling programs.

Competitive Advantages
According to Nanodesign, EMD offers several advantages over traditional molecular modeling packages:

• Uses a small number of high- or moderate-affinity compounds and inactive structures as a training set for pattern recognition
  
• Does not require structural characterization of a biological target (e.g. receptor or enzyme)
  
• It is a design method, not a screening tool
  
• Based on identification of localised critical features of ligand binding surface and ability to account for more of the essential features in drug/target interactions
  
• Identifies structures predicted to be biologically active, bioavailable, chemically synthesizable, and novel (and therefore can be patented).

For more information: Barbara Fanning, Director of Business Development, Nanodesign Inc., Suite 300, Research Park Centre, 150 Research Lane, Guelph, ON Canada N1G 4T2. Tel: 519-823-9088. Fax: 519-823-9401.

By Angelo DePalma