Cover Story - Drug platforms

Haas, M. SciBX 2(2); doi:10.1038/scibx.2009.37
Published online Jan. 15 2009

(Back)boning up on polyketides

by Michael J. Haas, Senior Writer

Polyketides are the last major class of natural products that have not been successfully synthesized in Escherichia coli, the microorganism of choice for biosynthesis and biomanufacturing. Researchers at the University of California, Los Angeles, think they have overcome a major barrier to this biosynthesis by engineering E. coli to express a modified fungal enzyme that controls the synthetic process.¹ A first-line application could be the production of doxorubicin analogs that are less toxic than the parent drug but more difficult and expensive to synthesize by other means.

"Many existing aromatic polyketides are synthesized by organisms that cannot be manipulated easily—or at all—genetically," said team leader Yi Tang, an associate professor of chemical and biomolecular engineering at UCLA. With this platform, "we open up the biosynthesis of this family of compounds to the powerful E. coli genetic tools."

Polyketides are metabolites of bacteria, fungi and other organisms and have pharmaceutical applications that include antibiotics and cancer therapies. A key obstacle to their biosynthesis has been engineering E. coli to express the polyketide synthases (PKSs) that build up the polyketide backbone.

Engineering efforts over the past 15 years have focused on bacterial PKSs because their molecular mechanisms are better understood than those of fungal PKSs. But attempts to express bacterial PKSs in E. coli yielded insoluble and inactive enzymes.

These failures led Tang and colleagues to speculate that engineering E. coli to express fungal PKS genes, rather than bacterial ones, might produce a functional enzyme.

In 2007, a team led by Tang reported in the Journal of the American Chemical Society that PKS4 derived from the fungus Gibberella fujikuroi could be expressed and remain functional in E. coli.² However, PKS4 closed the polyketide backbone into a ring structure and thereby precluded the potential for subsequent structural changes by downstream enzymes.

As reported in the PNAS study, Tang's team subsequently set out to remove PKS4's ability to form the ring structure while retaining its ability to synthesize polyketide backbones. Building on a study by a group at Johns Hopkins University that showed which PKS4 domain mediated ring formation,³ Tang's team replaced that domain with an inactive linker to produce a smaller, modified version of PKS4 (PKS_WJ).

In vitro studies showed that PKS_WJ built polyketide backbones but did not cyclize them. Mass spectrometry and NMR experiments showed that the modified enzyme produced polyketide structures that the original PKS4 did not.

Finally, the team used PKS_WJ to carry out the biosynthesis of a known polyketide of the anthracycline family, SEK26, in E. coli.

New and improved

The platform could make a wide array of new polyketides accessible for research purposes and could allow modification of existing polyketides to improve their pharmacological properties.

Chaitan Khosla, professor of chemical engineering, chemistry and biochemistry at Stanford University, said the platform overcomes obstacles to polyketide biosynthesis that have limited the number and yield of pharmacologically interesting structures that can be explored.

The UCLA findings should aid the drug discovery process by making polyketides "accessible to a vast number of scientists and engineers who might otherwise lack the expertise to access them," he said.

Gerald Walsh, CEO of Gem Pharmaceuticals LLC, thought that one potential application of the platform might be the biosynthesis of doxorubicin analogs that are less toxic than the parent drug.

Doxorubicin is a polyketide of the anthracycline family that is widely used to treat cancer, but its use is limited by serious cardiotoxicity. Walsh noted that the drug's cardiotoxic effects are associated with its C13 carbonyl group and that the compound lacking this group—13-deoxydoxorubicin—is difficult to synthesize.

"The method could be very valuable if it could be used to manufacture 13-deoxy analogs of doxorubicin" inexpensively and at high yields, he said.

In 2005, Gem terminated a Phase II trial of GPX-100 (13-deoxydoxorubicin) to treat metastatic breast cancer. The compound was active but also had cardiotoxicity.

Gem has since found that the C5 quinone group of doxorubicin also contributes to cardiotoxicity, said Raymond Tesi, president and CEO of Coronado Biosciences Inc. His company in-licensed CNDO101, a fourth-generation doxorubicin analog that lacks both the C5 quinone and C13 carbonyl groups, from Gem in 2007.

Based on the GPX-100 trial data, "we concluded that both structural elements must be removed from the anthracycline structure to eliminate cardiotoxicity," Tesi said.

CNDO101 is in a Phase I trial to treat advanced solid tumors, and Coronado expects to begin a Phase II trial in relapsed small cell lung cancer (SCLC) this year.

Tesi said the current manufacturing process for CNDO101 is complex and expensive. "We will follow the development of this biosynthetic process with interest," he said.

Unnatural advance

Tang's team is now extending the platform to the biosynthesis of larger polyketides than those reported in PNAS. The group is also incorporating functional groups that are not found in naturally occurring polyketides.

Tang agreed with Walsh and Tesi that an important next step is synthesizing marketed cancer drugs and developing derivatives of those compounds.

He told SciBX the platform is not yet ready for synthesizing such drug compounds—or for biomanufacturing purposes—because first the system would have to be engineered to express the enzymes needed to incorporate the desired compound's specific functional groups. That process, although relatively straightforward, would involve "some laborious molecular biology," which is not his team's current focus, he said.

Nevertheless, Tang said, "our work showed that the PKSs from the two kingdoms of life indeed can be functionally combined in the E. coli platform." He said the platform will allow researchers to examine how these interactions occur and whether the interactions are different from naturally occurring protein-protein interactions.

Tang said the findings reported in PNAS are not patented.