Drug Discovery & Development

Drug Discovery & Development
Reed Business Information
Rockaway, NJ, 07866

Chemical Genomics Is the "Big Tent" of Drug Discovery

Despite confusion on a definition of chemogenomics, researchers in this field use the gamut of analytical tools to test molecules against genes and proteins

Angelo DePalma, PhD
DePalma is a freelance writer based in Newton, N.J.

click the image to enlarge
A sensitive capillary electrophoresis (CE) assay screens for protein-small molecule affinity. The technique first establishes baseline CE behavior for the native protein, including retention time and peak characteristics. Molecules in crude microbial or yeast extracts that bind to the target protein cause a shift in electrophoretic behavior corresponding to affinity or binding. Chemists zero in on active molecules through chemical and chromatographic fractionations. (Source: Cetek)

Chemical genomics, now firmly established in the latest wave of post-genome sequence era discovery, is experiencing an identity crisis. As more groups adopt chemical genomics, either as the centerpiece of their discovery efforts or as one tool among many, the field is coming to resemble the proverbial "big tent," where all discovery tools are welcome.

Classical chemical genomics, popularized by professor Stuart Schreiber at Harvard University, Cambridge, Mass., tests single compounds against a gene or protein system to elucidate biological mechanisms. This original idea soon spawned a plethora of methods, all of which claim chemical genomics, chemogenomics, or even "chemical genetics" as their scientific basis. As a result, chemical genomics has evolved into a kind of catch-all category for techniques that unravel biology by interrogating proteins, genes, tissues, or organisms with small organic molecules.

One approach mutates genes randomly with small molecules, looks for interesting phenotypes, identifies relevant genes, and reverse-engineers the pathway to arrive at targets (and ultimately candidate drugs). A more focused approach, says Kleanthis Xanthopoulos, PhD, CEO of Anadys Pharmaceuticals Inc., San Diego, uses a combination of drug-like chemicals or known drugs to probe for multiple, previously unknown genomic targets and pathways.

At a superficial level, chemical genomics is analogous to gene perturbation through antisense, RNA interference (RNAi), or chemical mutagens, except that small-molecule-induced interactions tend not to be as pronounced, and the targets are more likely proteins than DNA or RNA.

By one view, chemical genomics seeks to reconcile the 2,000 to 3,000 recognized disease-mediating genomic targets with organic chemical space. Anadys' strategy involves sampling as much chemical diversity as they can afford and to use screening as the interface for identifying where to look for compounds. "Once that happens, we take the traditional approach of medicinal chemistry with iterative rounds of interaction with our biology," says Xanthopoulos.

Anadys focuses about 80% of its effort on antivirals and the remainder on antibiotics. The company demonstrated proof-of-concept for two of its drugs, against hepatitis C and hepatitis B. Programs for other viral targets and antibacterials are at the early clinical or preclinical stages.

click the image to enlarge
Aided in part by its chemical genomics program, Exelixis identified more than 25 kinase targets and determined the crystal structure for them and about 250 kinase-inhibitor complexes. The company plans to file an investigational new drug (IND) application in 2005 for XL880, a Spectrum Selective Kinase Inhibitor that targets C-MET. Pictured here, IC50 = 0.4 nM, t1/2 > 15 hours. (Source: Exelixis)

What's in a name?
The terms chemical genomics and chemical genetics are used interchangeably, and probably inaccurately, says Kevin Fitzgerald, PhD, group leader for emerging technologies at Bristol-Myers Squibb, Hopewell, N.J. "Strictly speaking, chemical genomics should refer to how small molecules affect gene expression, whereas in chemical genetics, you either remove or overexpress a gene and investigate the consequences on a compound's activity." Furthermore, says Fitzgerald, probing protein function through small molecules might be better named chemical proteomics. "The global term 'chemical genomics' is almost meaningless in the context in which it is often used," he says.

Some leading researchers, such as Xanthopoulos, don't understand the term "chemical genomics" at all. "And I'm a genomicist by training. It's very loose terminology. I can give you five different approaches to drug discovery and they all fall under the heading of chemical genomics."

There is also some confusion regarding the molecular targets of chemical genomics. Some authors speak of genes, others of proteins. "Regardless of what they call it, chemical genomics á la Schreiber almost always means the effect of chemicals on proteins," says Kurt Jarnagin, PhD, vice president of technology at Iconix Pharmaceuticals Inc., Mountain View, Calif.

Iconix concentrates on the effect of chemicals on protein transcription in rats and rat hepatocytes. To study and categorize such systems, Iconix created DrugMatrix, a database and related services that handicaps or prioritizes drug-like compounds according to pharmacology and toxicology, both on and off target. The DrugMatrix chemogenomic database uses 600 compounds (approximately 400 FDA-approved drugs, 60 drugs approved in Europe and Japan, 25 withdrawn drugs, and 100 toxicants). Molecules are profiled in up to seven rat tissues, using multiple doses, time points, and biological replicates. In all, the database represents more than 3,200 unique drug-dose-time-tissue combinations.

By comparing new structures against this small but representative and highly-characterized compound collection, DrugMatrix helps scientists predict a new structure's properties, or "benefits and liabilities" in Jarnagin's parlance, and assists in lead optimization based on genomic, biologic, and chemical profiles.

"We thought what was missing from gene expression analysis was contextwhat drugs and druglike molecules do," says Jarnagin. "So, we built this contextual database based on 15,000 microarray experiments." Users can take candidate molecules and compare their profiles to entries in this database to predict pharmacology and toxicology, allowing more direct, simpler use of gene expression profiles.

Iconix offers discovery organizations its contextual database through various fee-for-service arrangements. One way is by analyzing compounds or dosed tissues against its database and providing data on potential toxicology, pharmacologic side effects, and general pharmacology. "Because this report is based on our large contextual reference data set, it has much greater resolution and authority than a study done on a single compound," says Jarnagin. Iconix also builds proprietary, project-centered databases in specific organisms/tissues and target classes, or companies may purchase the complete DrugMatrix database and maintain it on site.

click the image to enlarge
A DrugMatrix screen shows the expression levels of the 500 most significantly modulated genes resulting from in vivo treatments rats at different timepoints and doses. The confidence interval is based on the variation in three biological replicates. (Source: Iconix Pharmaceuticals)

Probing target classes
Unlike Iconix, which deeply, exhaustively characterizes about 600 carefully selected compounds, Exelixis Inc., South San Francisco, Calif., has screened a three million-plus compound library against a relatively narrow range of kinases. This strategy looks deeply into a target class, generating large datasets while providing information that is unavailable from screening millions of compounds against disparate target types.

However, the data is never an end in itself. "Our goal is not to create a huge database and offer it for sale, but to develop the best compounds to take forward," says chief science officer Gregory Plowman, MD, PhD.

To Plowman, chemical genomics is especially powerful against target classes, for example GPCRs, NHRs, ABC transporters, ATPases, and of course kinases. "You can learn a lot from one assay to the next, which is something we've leveraged in our discovery and development programs."

In addition to one in-licensed drug in phase III for an orphan oncology indication, Exelixis's pipeline includes three kinase agents in or just out of phase I, and potentially five more that should enter the clinic "in a year or so." Plowman uses the term "spectrum-selective kinase inhibitors" since these molecules combine activity against both vascular- and epithelial-specific kinases. "You can think of these as doing what Herceptin or Erbitux do on the epithelial side and what Avastin does on the vasculature, but combined into one molecule."

In line with its target-class approach, Exelixis has plans to move into other target classes, particularly GPCRs and nuclear receptors for cancer and metabolic diseases.

Back to Nature
Although drug developers are at no loss for molecules, the temptation to go "back to nature" and mine chemical diversity in organisms remains strong. But as rich as natural sources may be in druglike or druggable molecules, that abundance constitutes an embarrassment of sorts. Isolating and testing individual molecules from tissues containing hundreds or thousands of compounds is impractical.

Cetek Corp., Marlborough, Mass., has overcome this natural product roadblock with CE Assay Screening, a sensitive capillary electrophoresis (CE) assay for protein-small molecule affinity. The technique first establishes baseline CE behavior for the native protein, including retention time and peak characteristics. Molecules in crude microbial or yeast extracts that bind to the target protein cause a shift in electrophoretic behavior corresponding to affinity or binding. Chemists zero in on active molecules through chemical and chromatographic fractionations familiar to natural products chemists. According to company president James Little, PhD, CE Assay is between 10 and 50 times more sensitive than typical in vitro assays.

Unlike traditional chemical genomics, which starts with a known small molecule and presumes little about the target, CE Assay Screening works in precisely the opposite manner. The method nevertheless satisfies the most basic requirement of chemical genomics, namely to use small molecular weight entities to probe the behavior of genes and proteins. One could say that the drug target is being used to probe the genome of the source organism, as identified by the small molecules produced by that genome's protein products.

In fact CE Assay handles multiple protein targets provided they are related in structure and CE behavior. Of particular interest are proteins that differ only with respect to post-translational modifications. In one project for a large-pharma customer, Cetek determined the relative activities of natural product extracts against phosphorylated and unphosphorylated protein targets.

Cetek began as a service company and counts about 25 pharmaceutical firms as its present or former customers, including Cubist, Fujisawa, Genome Therapeutics, Johnson & Johnson, Millennium Pharmaceuticals, Pharmacia, and Schering-Plough.

Targets for drugs
Where traditional screening-based drug discovery relies on rigorous target validation, chemical genomics is concerned more with finding targets for drugs rather than drugs for targets. "It flips the process on its head," says David Szymkowski, PhD, director of biotherapeutics at Xencor Inc., Monrovia, Calif.

Through his previous stint at Roche and his Xencor experience, Szymkowski has become an expert on chemical genomics and has published extensively on the subject. He believes chemogenomic studies of existing drugs, especially the several hundred currently marketed for which mechanisms are unknown or poorly understood, could be a rich source of new compounds and might significantly shorten times between therapeutic generations. Such programs carry little risk, he says, because "you already know the drug works."

As an example he cites COX-2 inhibitors. It took thousands of years between the discovery of willow bark's analgesic properties and the synthesis of aspirin. "Another hundred years passed before we learned how methyl salicylic acid worked. Then in the 1980s, the discovery of COX-1 and COX-2 pathways led to the selective COX-2 inhibitors." With the proper chemical genomics tools, he contends, the entire process might have been compressed to a decade or so.

Other prime chemical genomic candidates, says Szymkowski, include "fallen angels," which are compounds that have failed in human, animal, or preclinical studies. "Chemogenomics can identify the toxicology target, which would help in producing safer analogs or even new chemical approaches."

Looks like traditional discovery
Philosophically, chemical genomics borrows heavily from the established ethos of drug discovery methods. Graffinity Pharmaceuticals AG, Heidelberg, Germany (which dubs itself "The Chemical Genomics Company") makes chemical genomics chemistry and biology work together at an early stage through the action of small molecules on protein targets. Like Cetek (see sidebar), Graffinity's approach resembles traditional multi-compound, single-target drug discovery.

Graffinity screens proteins against a library of 10,000 compound fragments with molecular weights of up to 250 daltons, and about 90,000 drug-like molecules (up to about 400 daltons). All of the approximately 100,000 chemical moieties are contained on 15 gold-coated plates. Researchers detect compound-protein interactions through surface plasmon resonance, a sensitive but "assay-less" technique which requires no prior knowledge of the protein or the compound-target interaction.

The fragment-based approach samples a huge number of organic chemical characteristics through a relatively small library. "We start with fragments containing enough information to give some idea of specificity and selectivity," says Kristina Schmidt, PhD, manager of technology. "Hits" are then synthesized by linking two or more active fragments, and optimizing them through medicinal chemistry or chemical genomics.

Graffinity's fragment library has led to two preclinical candidates: DPP-4 for type 2 diabetes and an oral thrombin inhibitor. The company also has ongoing discovery relationships with Serono, Eli Lilly and Co., Genentech, and Novartis Pharmaceuticals, and has worked with Pfizer.

In the haste to categorize and over-differentiate chemical genomics, it is easy to forget how much it shares with other discovery methods. "The analogies between HTS and chemical genomics is of great interest to us," says John Anson, PhD, vice president of product development at GE Healthcare, Cardiff, UK.

GE Healthcare's predecessor company, Amersham Biosciences, developed many of the homogeneous assays which enable today's automated, high-throughput discovery tools. GE Healthcare continues to offer assays and instrumentation that serve chemical genomics as well as more established markets. For example, the company's IN Cell analyzer for high-throughput subcellular microscopy works as well for traditional cell work as for chemical genomics studies. And its LEADseeker instrument, which uses scintillation proximity, fluorescence, and luminescence, screens more than 1.4 million compounds per day in experiments where subcellular resolution is not required.

Anson also says that chemical genomics is much bigger than drug discovery. "Pharmaceutical researchers are only interested in a small subset of protein targets and how they interact with millions of organic molecules," he observes. "Academics have a much broader perspective, because their goal is fundamental understanding of biological processes." For now, he sees chemical genomics more as a research tool than drug discovery platform, a view that coincides with that of Christopher Austin, MD, director of the NIH Chemical Genomics Center (see story at left).

Are we there yet?
However one names the interplay between chemistry and biology, it is obvious the two disciplines have much to offer drug discovery when used together, especially within the context of modern molecular biology tools such as RNAi and microarray technology.

"You can ask a lot more sophisticated questions about compounds and activity today than just five years ago," says Fitzgerald of BMS, who believes that interrogating genes and proteins with small molecules early in discovery will eventually pay off with safer, more effective medicines. But when?

"Drug discovery occurs on a 12- to 15-year time scale," he says, "so the newer high-throughput techniques are only beginning to fill pipelines now. The impact of chemical genomics has been even more rapid and is already affecting the way we develop new compounds. In fact, one of our marketed drugs benefited from this approach."

NIH Funds Chemical Genomics Initiative
On June 9, 2004, the National Institutes of Health (NIH) established the NIH Chemical Genomics Center (NCGC), the first nationwide effort to apply chemical genomics to biology and drug development. NCGC aims to create a consortium of up to ten chemical genomics screening centers at academic institutions and other locations across the United States during 2005. Meanwhile on the NIH campus, a staff of about 50 scientists will begin screening small molecules by the end of 2004 to support NCGC.

As part of its plan NIH will establish a repository of up to 1 million chemical compounds, data from which will reside in a central database freely accessible to all scientists. This repository, PubChem, will be managed by the National Center for Biotechnology Information at the National Library of Medicine and should begin operation by early autumn 2004.

NCGC will focus on those 24,000-plus protein/gene targets (out of 25,000 to 30,000) for which no small-molecule probe is known. The center plans to screen more than 100,000 small molecules in multiple high-throughput assays within its first year, an effort described as "unprecedented" by the center's director, Christopher Austin, MD.

The center has selected several ultra-high throughput target and pathway screening technologies from privately-owned Kalypsys Inc., San Diego, a drug discovery and platform technology company. Kalypsys will provide materials, services, and a robotic screening system with a potential throughput of one million assay wells per day and online storage capacity for 2.2 million compounds. Kalypsys claims its second-generation UHTS system runs more than 200 individual cell-based and biochemical screens10 times the number of commercially available HTS systems.

The center is not specifically out to discover new drugs, says director Austin, although he hopes that will be one result of the effort. NIH's goal is to understand the genome, basic biology, and disease mechanisms. "The science and history of drug development tells us unequivocally that the tools we produce will only be useful as probes. There is often confusion about this point, I think because no one has ever made such compounds on this scale for this purpose. People think we are trying to 'make drugs' since that is the only context in which they are used to hearing about small molecules. Thdis novelty is one of the reasons we, and the community, are so excited about this initiative."