Chemical genomics allows scientists to rapidly characterize a large pool of small molecules against target proteins to identify those with the most promise

Gina Shaw
Shaw is a contributing editor based in Marlboro, N.J.

Many in the pharmaceutical industry agree that the field of chemical genomics is a promising one. What they have historically been unable to agree on, however, is exactly what the term means. But that is changing . . . sort of. Kenneth Carter, PhD, chief executive officer at Avalon Pharmaceuticals Inc., Germantown, Md., says "We look at it from the broadest possible definition, which is this: How does a particular chemical entity or chemical substructure, set of structures, or family of compounds affect the genome? And how do we capture that information in the broadest possible way?"

A molecule (in red) identified using diversity oriented synthesis selectively docks into a protein-binding pocket, thereby disrupting the protein-protein interaction known to confer resistance to chemotherapy. (Source: Infinity Pharmaceuticals Inc.)

Industry analyst Philip Ma, a principal with McKinsey and Co., defines it as follows: "At the end of the day, it's the combination of a systematic approach to chemistry, using common libraries and/or common scaffolds, combined with genomic targets or genomic technology." He concedes it is not a great definition, but it tends to cover all the bases.

Others echo Ma's definition. "It's a way of reaching out with a small molecule and perturbing the function of a gene, manipulating a gene product, or reaching out and touching a protein," says Alan Annis, PhD, vice president of new technologies with NeoGenesis Pharmaceuticals, which was purchased by Schering-Plough Corp., Kenilworth, N.J., in early 2005. "By using a small molecule, you can interrogate how that protein does cool things biologically."

Some take a broader view, some a narrower view, but all seem to agree that chemical genomics is coming of age. The term became a hot industry buzzword over the last several years, as companies such as NeoGenesis Pharmaceuticals and Infinity Pharmaceuticals Inc., Cambridge, Mass., entered the market. They began churning out capital for their high-throughput approaches to screening small molecules against proteins in order to identify gene function and fine-tune drug candidates at the same time. The mere fact that these companies are still around in 2005 indicates that the talk about chemical genomics wasn't just hype. Industry seems to agree: in January alone, Infinity Pharmaceuticals signed small-molecule collaboration agreements with Novartis and Johnson & Johnson, while Schering-Plough announced it was acquiring NeoGenesis outright.

One of the biggest bottlenecks in the drug discovery pipeline arises during phase III trials. According to pharmaceutical industry tracker Pharmaprojects, part of T&F Informa plc, London, the numbers of drugs in preclinical development and phase I and II trials has grown substantially over the past 10 years, by about 50% for the preclinical phase, 85% for phase I, and 90% for phase II. But that's where the growth stops: Phase III trials have languished at about the same levels for the last 10 years. Getting from phase II to phase III continues to be the hang-up for most new drug candidates, because, it's been suggested, many companies are wasting too much time on non-viable drug candidates in the earlier stages of the process. By using chemical genomics early on during the target validation phase of drug discovery, says Annis, pharmaceutical companies can do something very important: fail as fast as they possibly can.

Exploring the Dark Matter of the Genome Jim Inglese, PhD, doesn’t believe in the “druggable genome.” Among biomolecular researchers, that’s a bit sacrilegious, but the director of the Biomolecular Screening and Profiling Division at the NIH Chemical Genomics Center, Bethesda, Md., is entirely comfortable with his heresy. “It’s probably true that once every gene product and molecular interaction in the genome is defined, there will be some for which we can efficiently and effectively and non-toxically intervene with small molecules, and others that we cannot,” Inglese says. “But it’s fair to say that it’s dogma within the pharmaceutical industry that the current group of well-known targets represents the only gene products that can be used for intervention with a small molecule, because they’ve been used as drug targets in the past.” It makes sense, Inglese says: Pharmaceutical companies under enormous pressure to produce commercially important drugs in a short time are inclined to mine areas of the genome that have already proven successful. But with no pressure to produce drugs itself, the NIH Chemical Genomics Center doesn’t have to do that, so it has no plans to take aim at the heavily staked-out “druggable genome.” “In our view, the entire genome has therapeutic potential that needs to be explored. A lot of effort in industry is being put toward a few targets, and we’re trying to stay away from those and look at ones for which there isn’t necessarily any known therapeutic value, the ‘dark matter’ of the genome.” When NIH announced it would open a Chemical Genomics Center, some wondered if they weren’t staking out territory that competed with industry, drug development territory, rather than basic science or clinical research. To the contrary, says Christopher Austin, MD, the center’s director and the head of a lab of about 50 scientists. “What we’re doing is complementary to industry, not competitive,” he says. “The NIH’s role is to fund research that goes down those pathways the end of which we don’t know, because we don’t have that short-term commercial imperative.”

"You don't always want to succeed," he says. "If you have a target that you think might be interesting, you don't want to waste a lot of person-years finding out that you're wrong. If all you need is a purified protein, and you can come up with chemical matter that you can use rapidly to probe biology, then you know one of two things: either you're on the righteous path to a drug, or you need to stop. That's the strength of chemical genomics: you can rapidly triage a large pool of small molecules against target proteins to identify the ones with the tightest binding, the least toxicity, and the fewest off-target effects."

click the image to enlarge The Automated Ligand Identification System (ALIS) created by Neogenesis can determine the affinities of individual compounds from a chemical mixture. (Source: Neogenesis Pharmaceuticals Inc.)

NeoGenesis' approach, now Schering-Plough's, uses a proprietary, mass-spectrometry-based technology called ALIS (Automated Ligand Identification System) to simultaneously determine the affinities of individual compounds from a chemical mixture. When a purified protein is combined with a mixture of compounds then titrated by a competitor compound, the affinities of each mixture component can be read directly from the resulting ALIS mass spectrometer responses.

The problem that ALIS solves, Annis says, is pinpointing which element of a compound mixture made the hit happen. Without that information, screening combinatorial mixtures is pointless. "The reason it works is that the compound that binds on the protein of interest lands on a mass spec detector, so you're not waiting for a 96-well plate to turn blue, which, of course, can be due to contaminants and other things found in big mixtures. With the accuracy of high-throughput mass spec, we can identify the high-affinity compound uniquely in mixtures of thousands of compounds." The ALIS mixtures are "mass-encoded," meaning that each compound in the mixture, which can include as many as 1,000, has a distinct mass, calculated to a difference of two decimal points. "By exploiting the power of a mixture, you can screen hundreds or thousands of compounds, working with both validated targets and relatively new targets," says Satish Jindal, PhD, NeoGenesis' founder.

New Chemical Matter Needed
To fully use the power of chemical genomics, novel chemical matter is needed.

Gene Expression Screening Goes Generic Todd Golub, PhD, director of the cancer program at the MIT Broad Institute, Cambridge, Mass., wasn’t trying to identify new drugs when he and his colleagues screened some 1,700 chemicals against five microarray-identified genes that serve as a “differentiation signature” for changes in leukemia cells. Instead, they wanted to make sure that their screening approach could correctly pinpoint which of these already well-characterized chemicals would, in fact, trigger those changes that would help leukemia cells behave like normal white blood cells. It did. Their new screening method, dubbed gene expression-based high throughput screening (GE-HTS), accurately identified eight chemicals that reproducibly trigger the differentiation of precursor cells into white blood cells, a process that is blocked in acute myelogenous leukemia. The implications for GE-HTS, says Golub, who published his findings in the March 2004 issue of Nature Genetics, go far beyond leukemia research or even oncology drug development. “With GE-HTS, gene expression is used as a signature for cellular states. It’s really a twist on phenotypic cell-based screen. You can take any phenotype of interest, define it in terms of that gene expression signature, and use them as readouts of the biological state. So you can make the assay completely generic, because any biological state transition could be approached in this way.” The GE-HTS screen is now being used in studies aimed at other cancers, such as prostate and sarcoma, as well as non-cancer indications. Avalon Pharmaceuticals is already adopting the concept for the purposes of drug development, says Golub, who is also an investigator at the Dana-Farber Cancer Institute, Boston. “If you’re faced with several candidate compounds and debating which to bring forward, having some understanding of the ways in which they differ in respect to their molecular effects on cells would be useful information to have. My feeling is that particularly as the costs of gene expression profiling continue to fall, this kind of molecular profiling of compounds, both early and late in the drug-development process, will become a standard part of the dossier of information used to decide which compounds are advanced.”

That's the challenge, says Julian Adams, PhD, Infinity's chief scientific officer. "We've already taken the low-hanging fruit, and we need new chemical matter to chase down new biological targets." Historically, about 30% of drug compounds have come from natural products. These compounds—like the taxanes, chemotherapy drugs originally derived from the Pacific yew tree—differ in important ways from their chemically synthesized counterparts. They are more three-dimensional in structure, sometimes higher in molecular weight, and often contain a complex stereochemical array of substituents. "They're more voluptuous, you might say," says Adams. Because of this, they boast increased activity and improved selectivity against drug targets. "The pharmaceutical industry, on the other hand, tends to make flat molecules devoid of stereochemistry."

Infinity blends combinatorial chemistry with the world of natural products, deliberately introducing stereochemical complexity into the libraries it synthesizes in order to make compounds far more complex than those typically found in the pharmaceutical industry. "Diversity-oriented synthesis," as it is called, can cut six months to a year off the target-validation process, Adams says. In fact, Infinity expects to file its first investigational new drug (IND) application with the US Food and Drug Administration this quarter on a molecule called IPI-504 that it first targeted just one year ago. "In under a year, we were able to take this compound and develop it through preclinical research, scale it up, and make a sterile, high-purity, stable drug product suitable for administration to humans," Adams says. IPI-504 targets the heat shock protein 90 (Hsp90) system, which regulates protein balance and protein integrity in cells. It's thought to be an important target for many cancers, but Infinity's first trial will be in multiple myeloma.

Part of Infinity's advantage, Adams says, is the fact that all of its compounds are interrelated. "From a screening exercise, we can know within a matter of a few weeks if there's a pattern that's worth following up on, and we can do that easily. It's not a molecule in isolation. It has a lot of similar chemicals displayed with it in a 3D space that should start to form a pattern. Initially, we can start to work with the most potent of these agents, and then do secondary and tertiary rounds of synthesis, followed by more traditional medicinal chemistry to fine-tune the product."

Boosting the Odds
Unlike NeoGenesis and Infinity, Avalon Pharmaceuticals comes at chemical genomics from the other end of the equation: the biology side. It uses two high-throughput analysis systems to examine how very subtle differences in a compound's chemistry affect the genome at multiple levels. The first of these, the high-throughput integrated transcriptional screen (HITS) system, is based on a highly modified PCR platform. "It allows us to examine up to a few dozen genes across tens or even hundreds of thousands of compounds," says Kenneth Carter, PhD, Avalon's chief executive officer. After analyzing cell lines to identify the best model systems for specific diseases, the cell lines are treated with a range of candidate compounds and expressed transcripts are isolated and converted to cDNA. Once scientists have identified gene sets that correlate with the desired biological outcome, the stored cDNA samples can be queried to find out which compounds led to those changes.

More traditional screens might allow for just one data point, for example, associated with an enzymatic activity or a cell survival measurement. "On the other hand, by measuring one or two dozen carefully selected genetic markers, you get a much broader snapshot of what a particular compound is doing, even at the earliest stage, in a system that costs essentially about the same as a traditional enzymatic screen," Carter says.

In February, Avalon announced its first collaborative agreement to end-license a clinical compound, VX-944, in partnership with Vertex Pharmaceuticals Inc., Cambridge, Mass., which initially discovered the compound. It has shown promise in preclinical studies in treating hematologic malignancies and may have a much broader effect as well. "We think that by bringing it in-house and looking at it in a wide variety of settings, we'll be able to define the biomarkers that will be helpful in assessing efficacy and establishing patient stratification," Carter says. "We think our approach will also help to very quickly define whether VX-944 might also be useful in solid tumor indications like colon, prostate, and breast."

The real promise of chemical genomics, Carter predicts, will be to increase the probability of success of individual programs. "You will have analyzed the biological effects of compounds on a much broader and more fundamental level before they ever get into human clinical trials. And once you're in trials, you can use these advantages to define efficacy biomarkers and stratify your patient populations in a much more individualized way."