A model approach to accelerated drug discovery

Abstract

Structural Bioinformatics, Inc. (SBI; San Diego, CA, USA) have announced an alliance with DuPont Pharmaceuticals, Co. (Wilmington, DE, USA) that gives DuPont access to SBI's proprietary protein-modelling technology – a technology that facilitates the rapid identification of small-molecule drug leads from novel gene sequences. Under the terms of the US$100 million, multiyear agreement, DuPont will pay to use SBI's technology to explore numerous gene targets.SBI's technology uses the three-dimensional protein-structure information that is encoded in expressed gene sequences to rapidly identify small-molecule inhibitors or antagonists that have a desired therapeutic effect. SBI believes that the technology will enable pharmaceutical scientists to access the potential of the human genome. ‘Sequences per se are not useful in drug discovery as they are linear pieces of information. By way of analogy, our technology is similar to the difference between the digital code for Microsoft Word™ and the text that shows up on the screen after it has gone through the operating system’, says Dr Edward Maggio, president and chief executive officer of SBI.With high-throughput screening, pharmaceutical companies typically test 100 000 molecules against a new target and expect to get a 0.1–0.01% hit rate. It is a slow and expensive process – not only do they incur testing costs but also the much more significant acquisition and synthesis costs for compounds. ‘By using computers, we are able to computationally pre-screen millions of virtual molecules and select just 50–100 compounds for in vitro testing. Our hit rate with SBI's technology is 5–20%’, claims Maggio. ‘It means that we are able to cut out 99.9 % of the testing and compound acquisition costs.’As no single, universally applicable algorithmic approach exists for predicting all protein structures, SBI has taken useful elements from ∼8–12 separate predictive techniques and has combined them. Although current homology-modelling techniques are good at predicting the core structures of proteins that share high sequence homology, they are less successful at predicting the surface structures.Much of SBI's success is due to the inclusion of the model of loop-structure prediction, which requires no prior structural information, and certain energetic computational analyses that have been developed by Dr Kal Ramnarayan and his team at SBI. ‘This approach allows us to accurately predict the loop structures with no assumptions about the shape of the loops’, says Maggio.A limitation of homology modelling is that one X-ray crystallographic or nuclear magnetic resonance (NMR) structure within a protein family needs to be known to identify the other family members, although scientists at SBI-Advanced Technologies (Hoersholm, Denmark) and at the Danish Technical University (Copenhagen, Denmark) are working to reduce or eliminate this limitation by developing novel, advanced concepts in protein-structure prediction.A key feature of SBI's drug discovery process is the DynaPharm™ template technology (see Fig. 1Fig. 1), with which virtual constructs of dynamic protein surfaces are generated to screen virtual libraries of small molecules. The libraries, known as CombiLib™ modules, are built around core scaffolds (the initial chemical structure on which the final molecule is based) or chemical classes that possess attractive drug properties, are easily modifiable and that meet customers' criteria of drug-like characteristics.Fig. 1The DynaPharm™ process. (a) A protein structure is used to generate (b) DynaPharms, which model protein-surface structures that are used to generate or select (c) active small-molecule drug leads.View Large Image | Download PowerPoint Slide‘Once we have produced a DynaPharm from the protein surface we are literally days away from having active molecules. The limiting step is the speed with which we can synthesize the 50 or so molecules to test’, says Maggio.Over the last year, SBI has successfully applied its technology to six clinically important protein targets on behalf of three other corporate pharmaceutical collaborations. In each case, multiple chemical series of antagonists were generated for each target in the viral protease, protein kinase, hormone and apoptosis protein classes. An endothelin receptor antagonist, developed by their predecessor company, Immuno Pharmaceutics, and currently owned by Texas Biotechnology (Houston, TX, USA), is currently undergoing Phase II clinical trials for congestive heart failure.In addition to drug discovery programmes with corporate partners, SBI is conducting its own drug discovery efforts into Bcl-2, to treat cancer, and into erythropoietin (EPO) agonists to treat anaemia, amongst others. This year they received two Small Business Innovation Research (SBIR) grants from the National Cancer Institute (Bethesda, MD, USA). The first is to develop non-peptide, small molecule antagonists that target the HER2/NEU gene product; the HER2 gene is amplified and overexpressed in ∼30 % of all human cancers and is associated with poor prognosis, chemoresistance and aggressive, metastatic growth. The second grant is to develop inhibitors of a bacterial metalloprotease; bacterial metalloproteases might be promising targets for developing new classes of antibiotics.