Biocombinatorial techniques aid the development of small molecules to control the disease.
If you can’t beat it, find a way to live with it. That’s how the search for new anticancer drugs is shifting focus from cures to controls, according to University of Pittsburgh pharmacologist John Lazo. Controlling cancer means “you have the disease, and live with it, as do Parkinson’s patients, but it doesn’t kill you,” says Lazo.
In looking for drugs to slow tumor growth, biologists are becoming more target-driven, seeking promising molecules to arrest specific stages in the cancer-cell cycle. And this is where combinatorial chemistry—a.k.a. combichem—is playing an increasingly important role. Doing target validation without combichem, says Lazo, “is like walking with one leg.” Finding new anticancer drugs is becoming an interdisciplinary endeavor in which, out of necessity, biology meets chemistry.
Combinatorial chemistry—reacting a set of starting chemicals in all possible combinations to yield a library of tens of thousands of compounds—is now often called biocombinatorial chemistry. Collaborating with Lazo is University of Pittsburgh chemist Peter Wipf, who explains the process in this way: Start with a chemical compound called a platform that locks onto a specific molecular target, then change the platform incrementally to quickly produce hundreds of compounds to screen against that given target, and readily learn what specific compound structure is best at disrupting a biological pathway. Just such an approach is what Lazo and collaborators have used to identify a vitamin K analogue that inhibits breast tumor cell growth. That compound, says Lazo, “needs to be optimized”, and a second-generation analogue, with help from combichem, is now in the works.
As Lazo points out, it was difficult to get funding for early development of combinatorial libraries, and the National Cancer Institute (NCI) “did a lot of the heavy lifting.” From 1990 through 1997, John Weinstein’s NCI research group had screened more than 60,000 compounds against a panel of 60 human cancer cell lines. Each compound’s pattern of toxicity against the cell line’s bank is unique, like a fingerprint, and can be used to rank that compound with respect to specific anticancer activity, such as those not dependent on an intact p53 tumor suppressor gene.
Finding mechanisms
As Wipf points out, the discovery of new mechanism-based anticancer drugs is “closely linked” with advances in the understanding of cell biology and signaling processes. As an example, Wipf uses the discovery of a small molecule inhibitor of mitotic spindle bipolarity identified from a structurally diverse discovery library by cell biologist Thomas Mayer and colleagues at Harvard University (Science 1999, 286, 971–974). Mayer’s group selected 139 compounds from a library of 16,320 small molecules by using a mammalian whole-cell immunodetection assay to look for increases in phosphonucleolin staining that would indicate mitotic activity. Further screens, including one based on posttranslational modification and another based on visualizing microtubules and chromatin, identified compounds that affect mitosis. One compound, dubbed monastrol, inhibited motility of a motor protein required for spindle bipolarity, a real breakthrough because all previously identified small molecules only targeted tubulin.
Mayer’s approach—screening for small molecules that affect a particular pathway or process, rather than a single protein activity—is referred to as chemical genetics because of its conceptual similarity to classical forward genetic screens, such as irradiating Drosophila and then looking for eye-color mutants, and to distinguish it from the now common reverse genetics, in which a gene is identified before the phenotype it produces is known. As Mayer points out, because compounds that cause mitotic arrests by other mechanisms have shown antitumor activity in humans, monastrol may serve as a lead for the development of anticancer drugs. In other words, antimitotic can be equated with anticancer when it comes to describing drugs.
Wipf points to another important use of combichem: improving the biological and physiochemical profile of complex natural product lead structures that are not directly suited for human therapy but show attractive anticancer activities in vitro. Wipf’s group has recently developed libraries of potent antimitotic products showing greatly improved therapeutic potential (J. Am. Chem. Soc. 2000, 122, 9391–9395).
Another good example of mechanism-based therapeutics is the pharmacological rescue of mutant p53 conformation and function (Science 1999, 286, 2507–2510) by geneticist Farzan Rastinejad and colleagues at Pfizer Central Research in Groton, CT. Rastinejad’s group found compounds that stabilize the DNA binding domain of p53, a tumor suppressor gene, in the active conformation. The small synthetic compounds not only promoted the stability of wild-type p53 but also allowed mutant p53 to maintain an active conformation. The authors reported that a prototype compound caused the accumulation of conformationally active p53 in cells with mutant p53, enabling it to activate transcription and to slow tumor growth in mice. After further work to improve potency, Rastinejad and co-workers believe this class of p53 rescuers “could be developed into anticancer drugs of broad utility.”
Jackson Gibbs of Merck Cancer Research Laboratories in West Point, PA, has recently summarized progress on mechanism-based target identification and drug discovery in cancer research (Science 2000, 5460, 1969–1973). Gibbs lists several examples of molecular targets in tumor cells for cancer drug development: cyclin-dependent kinases (cdks); epidermal growth factor receptor (EGF); estrogen receptor (ER); farnesyltransferase (FTase); platelet-derived growth factor receptor (PDGF); and protein kinase C (PKC). As Gibbs points out, “nothing provides more compelling validation for a target than knowledge of the human genetics of a specific disease.” In breast cancer, for example, not only are several mechanism-based therapies in development, such as receptor antagonists, protein kinase inhibitors, therapeutic antibodies, antisense oligonucleotides, and viruses, but an increased use of tumor genotyping to guide the choice of cancer therapy is already underway.
According to Wipf, the most aggressive cancers, such as prostate, liver, breast, brain, and some types of skin cancers, are the primary targets for small-molecule therapeutics that interfere “directly and decisively” with cellular proliferation. It is difficult to imagine that there will be “one drug or even one family of drugs” to fight cancer, says Wipf. Instead, Wipf expects the chemotherapeutic arsenal against cancer to grow quickly, almost to the point where genetic fingerprinting will be used to identify the most effective drug regime on a case-by-case basis. “Obviously,” believes Wipf, “combinatorial chemistry is well-positioned to provide a plethora of lead structures and fine-tune the multiple mechanism-based therapeutics that will reign in the bewildering array of cellular mishaps that cause cancer.”
Changing cultures
The collaboration between biologists and chemists, while on the rise, requires a culture change, according to Lazo. Speaking as a biologist, he says it is “rare to find chemists who understand biologists.” As a result, Lazo holds biweekly meetings with his chemist collaborators and believes that there should be as much interaction as possible between the two camps. Lazo bemoans the physical barriers to interaction, pointing out that at many U.S. medical schools, which typically house the pharmacologists, the chemistry department is not even within walking distance. To Lazo, the interaction between biologists and chemists is necessary to avoid making combichem libraries just for the sake of making libraries. A far more efficient approach ensues, he believes, from “interrogating the target.”
Lazo sees pharmacologists as long-enamored of the power of molecular biology but only recently recognizing the power of chemistry. Pharmacologists who do not use combichem will end up plagued by the same fate as those “who didn’t embrace molecular biology,” says Lazo, “it will be like driving your car without one cylinder. Building bridges between medicine and pharmacology, between biologists and chemists—combichem gives you those tools.” As for biologists, they are “used to libraries of DNA, now it’s libraries of compounds. You have to think about automation, it changes your mind set.” The danger for the future, he believes, is that researchers will throw up their hands at having “too many targets” but perhaps people who complain about too many targets are analogous to Mozart’s detractors who denigrated his music for having too many notes.
Jürgen Drews of International Biomedicine Management Partners in Basel, Switzerland, sees the history of drug discovery as driven by chemistry but increasingly guided by pharmacology and the clinical sciences (Science 2000, 287, 1960–1964). New drugs, believes Drews, “are no longer generated solely by the imagination of chemists but result from a dialogue between biologists and chemists.” Drews questions whether recent attempts to design combinatorial libraries with a high degree of structural diversity make biological sense. Is the molecular diversity seen by chemists and calculated by structural descriptors actually “seen” by a biological target molecule?
Drews calls for design and sampling of compound libraries to be guided “not only by structural descriptors, but also by descriptors of biological activity.” The best way to do this, according to Drews, is to screen all compounds in a library against a panel of functionally dissimilar proteins to determine the binding affinity of each compound for each protein. The result is a set of binding affinities—an affinity fingerprint—for each compound. Affinity fingerprints can be compared to bring to light similarities or differences in biological activity. As Drews points out, drug discovery “has become so complex that it cannot be contained within the confines of the pharmaceutical industry.”
Going forward
Combinatorial chemistry started off in the early 1990s with the promise of revolutionizing drug discovery. The hope was that large collections of new compounds would accelerate the identification of new drug candidates. But, as University of Tübingen chemists Jörg Rademann and Günther Jung intone (Science 2000, 287, 1947–1948), “the route from design and synthesis of compound libraries to identification of a biologically valid lead structure is still long and tedious.” To really gain from the process of automated synthesis, analysis and screening must also be automated, and the enormous amounts of data generated must be handled efficiently. As more genomic information becomes available and more protein targets emerge, “more diverse compound libraries will have to be made and ways to efficiently extract information from such libraries will have to be developed,” state Rademann and Jung. And, they believe, “the bottleneck in drug discovery has shifted from generation of lead structures to their transformation into orally active drugs with the desired physiological properties and performance results in clinical trials.” They add that “it will no longer be enough to synthesize large and diverse libraries,” but it will become more important to integrate demands such as bioavailability at an early stage of drug research.
Rademann and Jung call for creating efficient interfaces between combinatorial syntheses and bioassays, and they have several suggestions for achieving this goal. Synthesis and screening for ligands of immunological receptors can be linked directly by using equimolar mixtures of compounds in solution and can be linked even more closely by doing both on the same support, on pins and resin beads. Or, instead of on beads, screening can be done on surfaces, such as those used for DNA microarrays, to produce screening chips that could be reused in several assays. For screening on surfaces, label-free detection, such as reflective interference spectroscopy based on white light interference, could be used to eliminate the extra synthesis and isolation steps required when classical fluorescent or radioactive labels are used.
What does the future hold? Wipf weighs in: “As our first-line therapeutic defenses against infectious diseases and cancer fall victim to biological resistance, there is a rapidly increasing need for fundamentally new agents with new modes of action. Combinatorial chemistry in combination with high-throughput, high-content, biological screening is an ideal driving force for innovation in the drug discovery process.”
Biochemist Said Sebti, head of the drug discovery program at the University of South Florida’s Moffitt Cancer Center in Tampa, agrees. Combichem speeds up the identification of lead compounds that can be further optimized via medicinal chemistry, says Sebti. And, when coupled with high-throughput screening, combichem not only speeds the process of drug discovery but also greatly increases the hit rate for identification of lead compounds, adds Sebti. In his own work, Sebti has used combichem in the discovery of GFB-111, a platelet-derived growth factor that has antitumor and anti-angiogenic activities (Nat. Biotechnol. 2000, 18, 1065–1070). The role of combichem in cancer research can only increase as data from the human genome project become available and “result in more molecular targets that will need combichem to identify leads,” says Sebti.
Mona Mort is a freelance writer based in Tucson, AZ. Send your comments or questions regarding this article to mdd@acs.org or the Editorial Office by fax at 202-776-8166 or by post at 1155 16th Street, NW; Washington, DC 20036.
