(This article is sponsored by The Boston Group)

Given the tremendous increase in the number of potential molecular targets for therapeutic intervention, significant advances need to be made in technologies that uncover the roles that these proteins play in disease and then lead to the rapid discovery and development of drugs to treat these diseases at a scale and scope that complement the genome. Chemical Genomics offers a solution to this problem through the rapid discovery of small molecule ligands against thousands of targets without the need for functional or structural information. These small molecule ligands can then be used directly to validate the targets in cell-based assays and serve as lead compounds for drug discovery and development.

Drug Discovery
The target validation process often requires several years of focused research to understand the function of a target and build hypotheses around its role in disease pathology. It is at this point that the traditional drug discovery process can begin (Figure 1). The goal of this process is to develop molecules with low molecular weight, that allows them to penetrate membranes and enter diseased tissues, and with pharmacokinetic properties that allow them to persist long enough to be effective and yet be excreted such that they do not accumulate. If productive, these efforts result in lead compounds that enter pre-clinical studies where extensive rounds of synthesis and testing are done to optimize potency, stability and bioavailability, and to limit toxicity. Compounds emerging from this process then proceed into formal toxicology studies and if deemed safe, into clinical trials in humans.

Information derived from target validation studies can often be used to develop biochemical and cell-based screening assays around the target. Cell-based assay generation requires the creation of engineered cell lines, and is typically formatted to allow robotic manipulation of compounds that are then tested individually. These compounds can be derived from natural sources such as fermentation broths, traditional synthetic organic chemistry efforts, and combinatorial chemistry technologies. Compounds often score positively in these assays due to non-specific or toxic effects, thus requiring the creation of multiple counter screens to avoid false positives. Compounds discovered in screening assays are then put through an optimization process that requires iterative rounds of organic synthesis and testing in cell-based assays and later in animal model systems.

Another approach to discover new drug leads is through a process known as rational drug design. Here, detailed structural information derived from protein target-ligand co-crystals is used to guide the process of drug discovery. This strategy has been widely adopted and has proven to be very useful in the optimization of existing drug leads, but has had limited success as a drug discovery approach de novo. Rational drug design requires extensive physical biochemical characterization of the target and the identification or knowledge of target substrates or ligands, and thus is not applicable in the near-term to the thousands of new targets emerging from the genome.

In addition to small organic molecules, a number of successful therapeutics have resulted from the development of recombinant proteins and antibodies. Recombinant Erythropoeitin and Interferon-beta are effective treatments for patients suffering from anemia and multiple sclerosis, respectively. These protein drugs are encoded in the human genome and are administered by injection to patients suffering from life-threatening diseases. Sophisticated strategies to isolate all of the secreted and membrane-bound proteins from the genome have been devised and deployed. However, the potential of protein therapeutic drugs is limited by the number of human diseases where supplemental protein treatment is beneficial, by the cost of production of recombinant proteins, and by the difficulty and inconvenience of administration by injection. Several important diseases such as cancer and inflammatory bowel disease have been treated successfully in recent years with antibodies recognizing receptors and secreted proteins involved in these disease processes. The creation of mice engineered to produce human antibodies when immunized with human protein antigens has accelerated the process of therapeutic antibody discovery and development. However, like secreted proteins, the use of antibody drugs is limited by the expense involved in producing them and the difficulty and inconvenience of administration. In addition, antibodies often induce the host immune system to produce neutralizing humoral responses, limiting their application in chronic diseases. Finally, given that antibodies are soluble proteins, their use is limited to disease targets present in the circulation or on the cell surface.

Current drug discovery and development methodologies are inadequate to process the number and diversity of potential targets presented by the genome. Target validation strategies remain in the domain of basic research and are too time consuming and empirical for application to thousands of protein targets. Traditional small molecule screening approaches involve extensive screens and assays that require a detailed understanding of the biological role of the target before they can be undertaken and are limited to the diversity of compounds existing in current collections. Recombinant protein and antibody therapeutics are difficult and expensive to produce and are limited in their application. What is needed are technologies that couple the target validation and discovery processes, that can interrogate large numbers of potential drug leads quickly, and that can be applied universally to all classes of targets.

Chemical Genomics
Chemical genomics promises to discover thousands of ligands against protein targets derived from the genome that simultaneously validate the functions that these proteins play in human disease pathways, and provide small molecule ligands as potential drug leads. The foundation upon which chemical genomics is based is the selection of small molecule ligands by affinity. While this approach might appear to be simple conceptually, it is very powerful in that it facilitates the identification of potential drug leads quickly and circumvents the need for the most time consuming steps involved in traditional drug discovery such as functional validation, assay development and function-based screening. Like classical genetic approaches, affinity selection in chemical genomics allows for the unbiased interrogation of potentially millions of small molecule compounds with no knowledge of the target protein's structure or function. Selected small molecules can then be tested immediately in cell-based and animal models of human disease. Activity in these models provides validation for the target protein and indicates that the test compound has therapeutic potential. Additional compounds based upon this structure can then be synthesized and re-selected for increased affinity. This affinity maturation process through iterative, unbiased cycles of selection is a powerful means for producing potent binding molecules and conceptually resembles the evolution of antibody specificity and affinity generated in the mammalian immune system.

NeoGenesis Pharmaceuticals (Cambridge, MA, USA) has developed a ligand selection technology that leverages the power of mixture-based small molecule synthesis and LC/MS automated separation and detection methods. This technology, known as ALIS (automated ligand identification system), couples an automated affinity-based selection and detection process with rapid synthesis of large pools of diverse, drug-like compounds. Here, an initial library of 10 million small molecules is divided into approximately 5,000 pools of 2,000 compounds such that each molecule in a given pool can be identified uniquely by its mass. Protein targets are incubated with each compound pool in solution and then protein-ligand complexes are separated from unbound small molecules using size exclusion chromatography. Following denaturation of the protein-ligand complexes, protein-bound small molecules are identified instantly with LC/MS. Compound identification then leads to iterative rounds of synthesis of mixtures of thousands of compounds and further selection and identification. In this way, ligands can be identified against protein targets of unknown structure or function and then rapidly evolved to high affinity, drug-like compounds. NeoGenesis has used this technology to identify potent ligands against proteins of unknown function, and against known proteins that had proven to be intractable to traditional discovery methods. In addition, many of these compounds have been active in mammalian cells, thus allowing the validation of proteomic targets of unknown function. Finally, given that this technology can be applied universally to all classes of proteins and is automated, it can be scaled to find potent ligands to hundreds or even thousands of targets in the next few years.

Conclusions
To achieve the goal novel therapeutics to treat serious human diseases, significant technological progress needs to be made to industrialize the process of drug discovery. Chemical genomics provides a solution that bridges the gap between the thousands of genes and proteins of unknown function emerging from the genome and the great need for new drugs leads against validated targets.

|