Modular Polyketide Synthases--Programming and Engineering Chemical Diversity

The extraordinary diversity of organic molecules in nature has long fascinated chemists, biologists, and the pharmaceutical industry. What are their structures, how are they synthesized, and what do they do? One remarkable class of molecules are the polyketides (Fig. 1) (1-4). Since the 1900s, researchers have isolated thousands of polyketides from plants, marine organisms, and soil microorganisms. Within this huge family of molecules, they discovered many with valuable medicinal, veterinary, and agrochemical properties. Today, polyketides comprise a large fraction of therapeutics and generate over $10 billion in combined annual sales; the antibiotic erythromycin and cholesterol-lowering drug mevacor are just two notable examples. Thus current polyketide research is driven by the following two questions: (1) How does nature generate this remarkable diversity of molecules? (2) Can we generate novel pharmaceuticals by genetically manipulating their biosynthetic enzymes?

Initial insights into the origins of these molecules arose by feeding labeled precursors to polyketide-producing organisms.(1) Between the 1950s and 1980s, chemists determined that complex polyketides, like fatty acids, are derived from simple carboxylic acid precursors assembled into a carbon chain, with chemical functionalities incorporated in a "processive" manner during chain formation. While these elaborate biosynthetic pathways could include 30 or more reaction steps, the underlying catalytic mechanisms remained a mystery because of the inability to directly study the enzymes involved.

A crucial breakthrough came in 1991, with the cloning and sequencing of the 6-deoxyerythronolide B synthase (DEBS), a modular polyketide synthase (PKS) from the erythromycin-producer Saccharopolyspora erythraea (Fig. 2A) (5, 6). DEBS catalyzes the formation of 1, and its DNA sequence elegantly revealed distinct active sites present in a one-to-one correspondence for each biosynthetic step. Each of the 6 DEBS "modules," located on three large subunits (each MW > 300,000), possess active sites [KS, AT, ACP(7)] to lengthen the polyketide intermediate by a carboxylic acid monomer, and a specific subset of reductive domains [KR, DH, ER(8)] to incorporate a particular functionality (carbonyl, hydroxyl, alkene, or methylene group) at the newest region of the chain. A terminal thioesterase (TE) cyclizes the full-length polyketide. The rational active site organization of DEBS immediately suggested how modular PKSs could control the overall selectivity of biosynthetic pathways, while simultaneously allowing for the generation of structural diversity.

With cloned genes and DNA sequences, modular PKSs were finally amenable to direct analysis. Again however, rapid progress was hampered by technology, in this case by the primitive genetic tools for most polyketide-producing organisms and by the difficulty manipulating large PKS genes (> 30 kb). In 1994, we attempted to bridge this gap by constructing an expression system for modular PKSs in Streptomyces coelicolor, a polyketide-producing soil bacterium with more established genetic tools. This plasmid-based system led to the first heterologous expression of a modular PKS, with the overexpression of the DEBS proteins and production of the 14-membered macrolactone 1 (Fig. 2A) (9).

The modular organization of DEBS generated many hypotheses regarding the structural and functional features of these enzymes, and our experimental system enabled us to test them by genetically manipulating PKS genes. For example, are modules structurally and functionally independent entities? What is the substrate tolerance of the TE? To address these questions, we created a series of DEBS module truncations fused to the terminal TE (Fig. 2B) (10-12). These mutants produced the polyketides 2, 3, and 4, respectively, demonstrating the dispensability of downstream modules for function (13). Moreover, the TE displayed relaxed specificity for alternate chain lengths by successfully cyclizing the 12-membered ring lactone 4. However, the TE was unable to cyclize the module 3 intermediate into an 8-membered ring and instead formed 3, possibly due to specificity limits or the presence of a C5-hydroxyl. To test these hypotheses, we turned to gain-of-function mutagenesis.

"Gain-of-function" mutagenesis (14) is the introduction of catalytic activities into modules that previously lacked them. Two examples are shown in Fig. 2C. In the first mutant, the reductive portion of module 2 (a KR that generates a 3S C3-hydroxyl; cf. Fig. 2A) is replaced by the reductive domains from module 1 of another modular PKS, that of the immunosuppressant rapamycin (a DH-ER-KR region that generates a methylene in rapamycin) (15). In the second mutant, the KR from module 4 of the rapamycin PKS is used (16). The first mutant produced the 8-membered ring lactone 5 with a C5 methylene, establishing the heterologous function of the rapamycin PKS domains (17). Moreover, this experiment indicated that the C5-hydroxyl prevents 8-membered ring formation in 3 and that the TE can be used to generate molecules that are difficult to synthesize chemically (18). The second mutant produced the triketide 6 with a 3R C3-hydroxyl, revealing that stereochemistry is "locally" controlled and can be transferred with KR domains (19). These and other gain-of-function and active site specificity changes (14, 20-25) have demonstrated our general ability to control polyketide structure by genetically manipulating PKSs.

Finally, the development of an in vitro system for the biosynthesis of 1 in our laboratory (26) allowed us to explore modular PKS structure. While the dimeric nature of these enzymes had been demonstrated, (26, 27) little else was known about their three-dimensional structure. By constructing several inactive DEBS mutants with specific KS or ACP mutations and restoring polyketide biosynthesis by heterodimer complementation in specific mutant pairs, we provided evidence for the organizational model in Fig. 3 (28). Here, in a manner completely analogous to the fatty acid synthases of higher eukaryotes, each module forms a head-to-tail homodimer, which in aggregate create two equivalent clusters of active sites for polyketide biosynthesis. This model presented important constraints on the higher order structure of modular PKSs and outlined a strategy to map the relationships of other active sites within the catalytic clusters (29).

The future prospects for PKS research are extremely promising, as new experimental approaches harness the power of genetics, chemistry, protein biochemistry, and engineering technologies (9, 26, 30, 31). In addition to expanding our understanding of these fascinating enzymes, the engineering of PKSs has provided a new route to drug discovery. Indeed, the construction of novel polyketide libraries by combinatorially manipulating PKS genes is already underway. Ultimately, these strategies could potentially generate polyketide diversity greater than that currently encountered in nature.

References

1. D. O'Hagan, The polyketide metabolites, 176 (E. Horwood, New York, 1991).
2. L. Katz, Chem. Rev. 97, 2557 (1997).
3. C. Khosla, Chem. Rev. 97, 2577 (1997).
4. J. Staunton and B. Wilkinson, Chem. Rev. 97, 2611 (1997).
5. S. Donadio, M. J. Staver, J. B. McAlpine, S. J. Swanson, L. Katz, Science 252, 675 (1991).
6. J. Cortes, S. F. Haydock, G. A. Roberts, D. J. Bevitt, P. F. Leadlay, Nature 348, 176 (1990).
7. KS: ß-ketoacyl-acyl carrier protein synthase, AT: acyltransferase, ACP: acyl carrier domain.
8. KR: ß-ketoreductase, DH: dehydratase, ER: enoyl reductase.
9. C. M. Kao, L. Katz, C. Khosla, Science 265, 509 (1994).
10. C. M. Kao, G. Luo, L. Katz, D. E. Cane, C. Khosla, J. Am. Chem. Soc. 117, 9105 (1995).
11. C. M. Kao, G. Luo, L. Katz, D. E. Cane, C. Khosla J. Am. Chem. Soc. 118, 9184 (1996).
12. J. Cortes et al. Science 268, 1487 (1995).
13. Recently, it has been shown that modules do not require upstream modules for function (J. R. Jacobsen, D. E. Cane, C. Khosla, Biochem. 1998, 37, 4928).
14. R. McDaniel et al. J. Am. Chem. Soc. 119, 4309 (1997).
15. C.M. Kao et al. J. Am. Chem. Soc. 119, 11339 (1997).
16. C. M. Kao et al. J. Am. Chem. Soc. 120, 2478 (1998).
17. Preliminary data suggest that the observed C3 reduction in 5 may result from the action of an endogenous S. coelicolor dehydrogenase. (H. Fu, and R. Gokhale, unpublished results.).
18. The ability of the TE to catalyze 16-membered ring lactone formation has also been demonstrated J.R. Jacobsen, C.R. Hutchinson, D. E. Cane, C. Khosla, Science 277, 367 (1997).
19. The triketide fallout product 6 from a three-module PKS suggests that the downstream module 3 strongly discriminates against 3R hydroxy intermediates in vivo.
20. R. McDaniel, C. M. Kao, S. J. Hwang, C. Khosla, Chem. & Biol. 4, 667 (1997).
21. M. Oliynyk, M. J. B. Brown, J. Cortes, J. Staunton, P. F. Leadlay, Chem. Biol. 3, 833 (1996).
22. S. Kuhstoss, M. Huber, J. R. Turner, R. N. Rao, Gene 183, 231 (1996).
23. L. Liu, A. Thamchaipenet, H. Fu, M. Betlach, G. Ashley, J. Am. Chem. Soc. 119, 10553 (1997).
24. X. Ruan et al. J. Bact. 179, 6416 (1997).
25. A. F. Marsden et al. Science 279, 199 (1998).
26. R. Pieper, G. Luo, D. E. Cane, C. Khosla, Nature 378, 263 (1995).
27. J. Staunton et al. Nature Struct. Biol. 3, 188 (1996).
28. C. M. Kao, R. Pieper, D. E. Cane, C. Khosla, Biochemistry 35, 12363 (1996).
29. R. S. Gokhale, J. Lau, D. E. Cane, C. Khosla, Biochemistry 37, 2524 (1998).
30. J. R. Jacobsen, C.R. Hutchinson, D.E. Cane, C. Khosla, Science 277, 367 (1997).
31. R. Pieper, R. S. Gokhale, G. Luo, D. E. Cane, C. Khosla, Biochemistry. 36, 1846 (1997).

To Advertise   |   Find Products