By David N. Leff

Name the biotech-developed therapeutic drug that's busted the biggest blocks on the pharmaceutical market. No, it's not Genentech Inc.'s tissue plasminogen activator (Activase), nor Amgen's erythropoietin (Epogen). The top bio-blockbuster is Bristol-Myers Squibb Co.'s anticancer agent paclitaxel (Taxol), which owes its name to the bark of the Pacific yew tree (Taxus brevifolia).

"It's the largest-revenue anticancer agent in history," observed protein chemist Daniel Santi, co-founder and CEO of Kosan Biosciences Inc., in Hayward, Calif. "Annual sales of Taxol and the related Aventis product, Taxotere, were $1.6 billion in 1998, with total sales expected to top $2 billion in 2000."

Taxol, combined with cisplatin, is first-line therapy for advanced ovarian cancer and non-small-cell lung cancer and, in combination with Herceptin, for breast cancer. It's second-line treatment for AIDS-related Kaposi's sarcoma.

But there are two flies in Taxol's ointment - which financially smells like a rose. First, many cancers are resistant to the drug, due to the multi-drug resistance mechanism that pumps it right out of the tumor cell it enters. Second, because it's water-insoluble, Taxol must be administered together with the surfactant Chremophor, which can cause severely adverse hypersensitivity side effects.

Luckily, there's another natural product on the horizon. which short-circuits Taxol's shortcomings. It goes by the name of epothilone, and was discovered 15 years ago by German scientists. They extracted the compound from a weirdly oddball soil bacterium, which they found in Southern Africa, and named Sorangium cellulosum.

"Epothilones were isolated in 1993," recounted Kosan's Santi, "and shown in 1995 to have potential as anticancer agents. S. cellulosum," he went on, "produces four slightly different epothilone versions, called A through D. All of them stabilize the microtubules in cells by the same mechanism of action as Taxol. But, they are effective against Taxol-resistant tumors. Also, they are water-soluble, so require little or no surfactant.

"For these reasons," Santi observed, "epothilone is widely perceived as a potential successor to Taxol." But these perceptions of Taxol's early demise, and epothilone's imminent rise, seem somewhat premature. S. cellulosum doesn't give up its therapeutic bounty so easily."

Sorangium's Output Slow And Stingy

"The bacterium has a very slow doubling time - once every 16 hours - and grudgingly yields only about 20 milligrams per liter of the drug," Santi noted, "which makes its production in the bacterium economically impractical. A and B" he went on, "are produced most abundantly in fermentation extracts, but D, with minimal output, has the highest therapeutic index. I understand - though to my knowledge it has not been announced - that some of the larger firms have taken epothilone into human patients, with very good results."

In a supply bind like this, protein chemists turn from fermentation to synthesis. "In tour de force efforts," Santi noted, "chemists from Sloan-Kettering and the Scripps Research Institute have reported the complete synthesis of epothilone. However, the complexity of the 20-or-more-step synthetic processes indicate that fermentation-based methods are likely to reign as practical approaches for the product's large-scale production."

Kosan, its CEO indicated, has a better way. "Although we and another group at the same time recently sequenced the entire 56-kilobase cluster of genes in S. cellulosum, which code for the six enzymes that make the epothilones," he told BioWorld Today, "we are the first to express them and produce the drug. More important, we have succeeded in transferring this entire gene cluster to a more compliant production organism, Streptomyces coelicolor."

In this week's issue of Science, dated Jan. 28, 2000, Kosan scientists report this feat under the title, "Cloning and heterologous expression of the epothilone gene cluster."

Epothilone is in good chemical company: it's a polyketide.

"Polyketides are a subgroup of natural products," Santi explained. "If you were to look at a list of the best-selling pharmaceuticals, you would see a lot of polyketides. For example, erythromycin, tetracycline, lovastatin (Mevacor), and all the other statins are polyketides. Many of the really big-selling pharmaceuticals are in this class.

"What is interesting about polyketides - and discovered only 10 years ago - is that they are made in the cells of their microorganisms by modular chemistries. That is, there is a code in the very large genes that make the enzymes that make the polyketides.

"Like, say, a DNA codon encoding a particular amino acid." Santi continued, "it allows us to go into other polyketide clusters, change their genes around, permutate the module, and thereby alter the polyketide structures. So that's what Kosan does for a living.

"Streptomyces," he pointed out, "is a soil microorganism renowned for producing these polyketides. And the particular Strep species, S. coelicolor, into which we moved the epothilone gene cluster, is sort of the E. coli of Streptomyces bacteria. It has a fast, two-hour doubling time."

Game's Name: Strain Improvement

This doesn't mean that Kosan's production program for the drug in its eager new host organism is home free. "Output is still very low," Santi said. "Once we start with this producer organism, what we have to do is go into the very tedious process called strain improvement.

"There," he recounted, "we will perform random mutagenesis on our recombinant S. coelicolor, and screen an enormous number of the mutants to find ones that were producing more. In conventional practice, we would repeat this until we had brought it up to a practical production level. That process can take anywhere from five to 10 years.

"So we are trying to short-circuit the process, using molecular biology. Now the home-run method that we have initiated, and I think we'll be betting on, is that we start out with a high producer of another polyketide, and change the genes in an attempt to morph that polyketide gene cluster into that of epothilone. Except here we'll be starting with something that's already producing, say, 10 grams per liter.

"If we went with what we've got now in S. coelicolor," Santi observed, "reaching practical production levels would probably take us 1.5 to two years. By this other approach, if we put the right manpower or person power on it, and hit it hard, we would know probably within a year whether we've got a real super-high producer."