A. MAUREEN ROUHI, C&EN WASHINGTON

Many programs in place in industry and academia make compounds based on natural products or designed to be like natural products. They utilize three ways to go beyond natural products: building around natural product scaffolds with combinatorial chemistry techniques, assembling natural-product-like compounds through diversity-oriented synthetic routes, and creating new natural product derivatives based on paths made accessible by target-oriented synthesis. All can advance drug discovery.

Combinatorial libraries based on natural product scaffolds--core structures around which analogs and derivatives are produced--are not new. For example, combinatorial chemistry around scaffolds of tubulin inhibitors is paving the way for new anticancer drugs. Libraries of paclitaxel analogs have yielded four second-generation drug candidates in clinical trials. Those of curacin A have brought out viable candidates for drug development. And in the anti-infectives area, compounds active against vancomycin-resistant bacteria have emerged from combinatorial libraries of vancomycin dimers [Curr. Opinion Drug Discov. Dev., 5, 304 (2002)].

Scaffolds from natural products already known to have biological activity are a good place to start, says Larry Hardy, director of biology at Aurigene Discovery Technologies, a Boston-based drug discovery company. A scaffold that's amenable to convergent total synthesis, is itself small, has multiple diversity points (points around which the structure can be varied) that are distant from each other, and has good physical properties is an ideal candidate for a combinatorial library, he says.

ANOTHER THING to keep in mind is whether the scaffold is already highly represented in proprietary molecules. Although Aurigene uses scaffolds in the public domain, it is interested only in libraries that are patentable. "We wouldn't build around the core of penicillin, for example, because of intellectual property issues," Hardy says. Many of the company's scaffolds come from plants native to southern India.

A good example of a scaffold, Hardy says, is galanthamine, a natural product isolated from various plants, including daffodil bulbs, and now used to treat Alzheimer's disease. The group of Harvard University chemistry professor Matthew D. Shair prepared a 2,527-member library directed at discovering new compounds with biological activities different from that of galanthamine. One member, the new compound named secramine, inhibits protein trafficking by the Golgi apparatus [J. Am. Chem. Soc., 123, 6740 (2001)].

"We wouldn't necessarily make all the compounds" prepared by the Harvard group, Hardy says. Instead, at Aurigene, crystal structures of targets help direct what molecules will be made, he explains. "Combinatorial chemistry has great utility, but we need to seed it with information from natural products and refine it through rational design."

Libraries developed with guidance from natural products make a lot of sense, because they match the elements of conservatism and diversity simultaneously expressed by biological targets, says Herbert Waldmann, a professor of chemical biology at Max Planck Institute of Molecular Physiology, Dortmund, and the University of Dortmund, in Germany. The same can be said for privileged structures, which are structural motifs that bind strongly to various proteins.

K. C. Nicolaou, a chemistry professor at Scripps Research Institute and at the University of California, San Diego, is a leading advocate of developing libraries based on privileged structures derived from natural products. "Scaffolds of natural origin, which presumably have undergone evolutionary selection over time, might confer favorable bioactivities and bioavailabilities to library members," he and coworkers wrote in the first reports of their efforts in this direction [J. Am. Chem. Soc., 122, 9939, 9954, and 9968 (2000)].

Nicolaou's team prepared a 10,000-member combinatorial library based on 2,2-dimethylbenzopyran, a motif found in more than 4,000 compounds, including natural products, that exhibit a wide range of biological activities. "Nature obviously loves this skeleton for some reason," Nicolaou tells C&EN. Subsets of the libraries were distributed to various groups for biological investigations. The harvest has been rich.

The library has provided leads for, among others, agents against methicillin-resistant Staphylococcus aureus and inhibitors of NADH:ubiquinone oxidoreductase, an enzyme involved in oxidative phosphorylation. Most recently, in collaboration with Ronald M. Evans and Joseph P. Noel at the Salk Institute for Biological Studies, La Jolla, Calif., and others, the Nicolaou team optimized a lead from the library to a potent ligand of the farnesoid X receptor (FXR), a key receptor in bile acid and cholesterol metabolism [Org. Biomol. Chem., 1, 908 (2003)]. The developed compound, called fexaramine, has now made possible the structure elucidation of FXR [Mol. Cell, 11, 1079 (2003)].

"The three-dimensional structure of FXR was unknown until we found a lead from the benzopyran library," Nicolaou says. "Now we have a crystal structure of the ligand binding in the pocket of the receptor, which could be the beginning of a drug discovery program."

Meanwhile, there is a school of thought that believes the universe of natural products does not have enough members to probe all biological processes and/or treat human diseases. Therefore, what is required is a collection of small molecules that have the complexity of, but are not necessarily based on, natural products and that will populate regions of chemical space broadly.

FIVE YEARS AGO, Harvard University chemistry professor Stuart L. Schreiber and coworkers undertook a combinatorial chemistry effort toward preparing such a collection. They made 2 million compounds that can be used for chemical genetics--the use of small molecules to interrogate biological processes [J. Am. Chem. Soc., 120, 8565 (1998)]. Science named the feat one of the top 10 scientific achievements of 1998.

Now, Schreiber says that the 2 million compounds are disappointing. When mapped on chemical space, the compounds form clumps, because most of the structural diversity is due to appendage modification. The focus has since shifted to diversity-oriented synthesis, which is combinatorial chemistry to produce diverse skeletons and stereochemistries with high appending potential, rather than appendage variations on a single skeleton (C&EN, March 3, page 51).

Part of the inspiration for diversity-oriented synthesis is natural products, says Peter Wipf, a chemistry professor at the University of Pittsburgh whose research includes diversity-oriented synthesis. "Natural products are diverse because of the presence of many rings--fused and bridged--with functionalizations at different positions, usually oxygenations and aminations. We apply those observations to the compounds we want to synthesize."

Diversity-oriented synthesis is not about, but does not exclude, use of natural product scaffolds, Wipf adds. "You want to generate structures that are quite different. If you simply play with a scaffold, you'll probably still be limiting yourself to less diversity than you can access otherwise."

Wipf says the Schreiber group's 2 million-compound effort bridges simple, appendage-modifying com- binatorial chemistry and diversity-oriented synthesis. "They used known reactions and readily available compounds," he explains. "But the reactions dramatically changed the structures of starting materials. In particular, a lot of rings were introduced. The next step now is to increase the number of reactions that can make such dramatic changes."

Methodologies are being worked out by many groups, including those of Schreiber, Shair, and Wipf. With diverse skeletons, far fewer than 2 million compounds might suffice.

Does diversity-oriented synthesis matter for drug discovery? Some people think so. For example, at VivoQuest, an early-stage pharmaceutical company based in Valley Cottage, N.Y., creating combinatorial libraries through diversity-oriented synthesis is a major focus, says Anthony Sandrasagra, director of biology. The firm's chemistry director, Zhen Yang, trained with Nicolaou and Schreiber and is steeped in both total synthesis of natural products and combinatorial synthesis of natural-product-like compounds.

Others believe that diversity-oriented synthesis will deliver tools to probe basic biology but not necessarily drug leads. "When you follow nature's leads, you already have biological activity somewhere there," Nicolaou says. "The chance of finding activity among compounds unrelated to nature, I believe, is less."

Waldmann concurs. "For basic biology, I fully subscribe to diversity-oriented synthesis," he says. "No doubt, that is a valid and proven way to find compounds for chemical genetics. It has the potential to discover new opportunities not provided by nature. But when it comes to drugs, compounds that are meaningful to nature, I believe, will have a higher success rate."

Diversity-oriented synthesis makes a lot of sense for the high-throughput-screening paradigm of current drug discovery, says Samuel J. Danishefsky, a chemistry professor at Columbia University and director of the Laboratory for Bioorganic Chemistry at Sloan-Kettering Institute for Cancer Research, New York City. "But I think the dividend in terms of drugs will take longer to be realized."

Danishefsky is a prime proponent of yet another way to go beyond natural products for drug discovery. And that is by preparing hypothesis-driven natural product analogs through paths made accessible by the mastery of molecular architecture that total synthesis enables. Medicinal chemistry starting with a natural product can only go so far. Structural changes that can lead to better properties may not be possible with the natural product as starting material. But they might be implemented if the compound is built from scratch. Structural modifications of epothilones implemented in this manner by the Nicolaou and Danishefsky groups in the late 1990s have led to designed epothilones that are in clinical trials for cancer chemotherapy.

A more recent example is the case of radicicol, a compound with novel antitumor properties in vitro that is not active in vivo. Danishefsky and coworkers guessed that converting the epoxide in radicicol to a cyclopropane ring would lead to a compound that's active in vivo.

It would have been prohibitive to make that change starting with radicicol, he says. "The rest of the molecule would have crumbled under the force of the reactions you would have had to use. So we built in the cyclopropane from the beginning." Danishefsky's lab has developed two routes to cycloproparadicicol [Angew. Chem. Int. Ed., 42, 1280 (2003); J. Am. Chem. Soc., 125, 9602 (2003)]. As predicted, the compound is active in vivo.

Another recent example is the designed epothilone called 26-trifluoromethyl-(E)-9,10-dehydrodesoxyepothilone B. On the surface, it is only slightly different from the original natural product epothilone B. It has a double bond where the original has an epoxide and a trifluoromethyl group where the original has a methyl group. The changes, the Danishefsky team thought, would yield an analog with much better pharmacokinetic properties and bioavailability.

The compound now has been prepared and tested against tumors in mice. At a dose of 30 mg per kg, the compound obliterates tumors with no relapse for more than two months [Angew. Chem. Int. Ed., 42, 4761 (2003)]. Danishefsky says the structural changes, especially the introduction of the trifluoromethyl group, could never have been implemented without mastery of epothilone architecture learned through total synthesis.

This strategy will not turn out enough compounds to satiate the pharmaceutical industry's capacity for screening, but it is highly effective, Danishefsky says. His lab's pipeline of vaccines and drugs proves it, he adds.

"The pharmaceutical industry has a conception of the format through which future discoveries will be made, and natural products were not on their radar screen," Danishefsky observes. "The mavens of the pharmaceutical industry seem to think that a discovery made outside that format can't be worth much. Some of these guys would have turned down the Gettysburg Address because it was handwritten by an aging single author rather than turned out by some pricey word-manufacturing institute that hit upon it by chance."

---

Chemical & Engineering News
Copyright © 2003 American Chemical Society