The Streptomyces glaucescens tcmKL polyketide synthase and tcmN

Effect of the Daunorubicin dpsH Gene on the Choice of Starter Unit and Cyclization Pattern Reveals That Type II Polyketide Synthases Can Be Unfaithful...
0 downloads 0 Views 2MB Size
Download PDF
J. Am. Chem. SOC. 1995,117, 6811-6821

6811

The Streptomyces glaucescens tcmKL Polyketide Synthase and tcmN Polyketide Cyclase Genes Govern the Size and Shape of Aromatic Polyketides B. Shen: R. G. SummersJvo E. Wendt-PienkowskiJ and C. R. Hutchinson*++,* Contribution from the School of Pharmacy and Department of Bacteriology, University of Wisconsin, Madison, Wisconsin 53706 Received February 7, 1995@

Abstract: The mechanism of a type I1 polyketide synthase was analyzed by combinatorially expressing components of the tetracenomycin (tcm) and the actinorhodin (act) polyketide synthase genes in various mutants or heterologous hosts. Structural analysis of metabolites produced by the recombinant organisms provided evidence to dissect the function of individual components of a type II PKS. Complementation studies with the S. glaucescens TcmK- and TcmL- mutants and constructs harboring the tcmK and actl-ORF2 or actl-ORF1 and tcmL genes demonstrated that a heterologous pair of genes encoding a j3-ketoacy1:acyl carrier protein synthase and chain length factor is often (but not always) nonfunctional. Isolations of the octaketides 6 and 7 from strains bearing actl-ORFl, actl-ORF2, and tcmM (pWHM766) or t c d , actl-ORF1, actl-ORF2, and tcmM (pWHM768), and the decaketides 8 and 9 from strains bearing tcmK, tcmL, and tcmM (pELE37) or t c d , tcmK, tcmL, and tcmM (pWHM731) established that the j3-ketoacy1:acyl carrier protein synthase and chain length factor proteins determine the chain length of the polyketide. While the addition of the t c d gene to the polyketide synthase core proteins consisting of the j3-ketoacy1:acyl carrier protein synthase, chain length factor and acyl carrier protein had no effect on the structures of the resulting metabolites, adding the tcmN gene to either pWHM766 and pWHM768 or pELE37 and pWHM731 resulted in the synthesis of octaketide 5 or decaketide 2, respectively. TcmN thus can alter the regiospecificity for the f i s t aldol cyclization from C-7/C-12 to C-9/C-14, suggesting that cyclases like TcmN determine the folding pattem of the linear polyketide intermediate. These activities, along with the choice of the starter unit, the loading of the extender unit to the PKS complex, and the function of T c d are discussed in an attempt to provide a rationale for the engineered biosynthesis of novel polyketides.

Polyketide metabolites are a large and diverse family of secondary metabolites found in bacteria, fungi, and plants.’ Many of them are clinically valuable antibiotics or chemotherapeutic agents or have other useful pharmacological activities (immunosuppressive,antiparasitic,insecticidal, etc).2 Ever since Birch3 outlined his “acetate hypothesis” in the early 1950s that polyketides could be biosynthetically derived from short fatty acids such as acetate, propionate, or butyrate, activated as acylthioesters, chemists and biochemists have been searching for a unified biosynthetic mechanism to account for the huge structural diversity found in polyketides. From the results of isotope labeling experiments, it has been evident for some time that the mechanism of polyketide biosynthesis is analogous to that of long-chain fatty acid biosynthesis catalyzed by the fatty acid synthases and that both polyketide and fatty acid biosynthesis utilize largely the same precursors. However, direct evidence supporting the analogy

* Corresponding author: C. R. Hutchinson, School of Pharmacy, University of Wisconsin, 425 North Charter St., Madison, WI 53706. Phone: (608)262-7582. Fax: (608)262-3134. email: crhutchi@facstaff. wisc.edu. School of Pharmacy. Department of Bacteriology. 5 Current address: Deuartment 47N. Abbott Laboratories AP9A. Abbott Park, IL 60064. Abstract published in Advance ACS Abstracts, June 15, 1995. (1) O’Hagen, D. The Polyketide Metabolites; Ellis Horwood: Chichester, U.K., 1991. (2) Monaghan, R. L.; Tkacz, J. S . Annu. Rev. Microbiol. 1990,44,271301. (3) Birch, A. J.; Donovan, F. W. Aus. J. Chem. 1953,6,360-368,373378; Birch, A. J.; Elliot, P. Aus. J. Chem. 1953, 6, 369-372; Birch, A. J. Science 1967, 156, 202-206.

*

@

0002-7863/95/1517-6811$09.00/0

between polyketide and fatty acid biosynthesis has come only recently from the molecular genetic and biochemical studies of antibiotic biosynthesis in Streptomyces species, principally. Sequence analysis of several sets of bacterial genes (plus a few from fungi) encoding polyketide synthases has revealed a highly conserved gene organization and a high degree of amino acid sequence similarity among the enzymes. Comparison of the polyketide synthases and fatty acid synthases has led to the conclusion that these systems must share a common mechanism of carbon chain assembly from similar if not identical precursor^.^-^ Yet, the details of how a polyketide synthase assembles the polyketide carbon skeleton remain obscure. Most of the mechanistic insights have been deduced from the effects of introducing mutations into the chromosomal copy of the polyketide synthase genes or by overexpressing the native or mutant polyketide synthase genes in various background^.^ Consequently, our understanding of the enzymology and biochemistry of polyketide synthases is rudimentary, and the purification or reconstitution of a complete polyketide synthase complex from a bacterium has so far not been reported, although significant progress has recently been made in this directi~n.’~-’~ (4) Hopwood, D. A.; Sherman, D. H. Annu. Rev. Genet. 1990,24, 3766. ( 5 ) Katz, L.; Donadio, S. Annu. Rev. Microbiol. 1993, 47, 875-912. (6) Hopwood, D. A.; Khosla, C. Ciba Found. Symp. 1992, 171, 88112. (7) Hopwood, D. A. Curr. Opin. Biotechnoi. 1994, 4, 531-537. (8J Vining, L. C.; Stuttard, C. Genetics and Biochemistry ofAntibioric Production; Butterworth-Heinemann: Boston, 1995; pp 323-498. (9) Hutchinson, C. R.; Fujii, I. Annu. Rev. Microbiol. 1995, 47, 201238.

0 1995 American Chemical Society

6812 J. Am. Chem. Soc., Vol. 117, No. 26, 1995

Shen et al.

Polyketide synthases found in bacteria and fungi have been the Tcm polyketide synthase has a high substrate specificityfor classified into two groups. Type I polyketide synthases, acetyl-coA as the starter unitst3 Moreover, we have sugcatalyzing the biosynthesis of highly reduced polyketides such g e ~ t e d ' ~a, ~linkage ~ , ~ ~ between polyketide and fatty acid as macrolide and polyether antibiotics, are large multifunctional biosynthesis in S. glaucescens and have proposed that the proteins that harbor a distinct enzyme activity for every step loading of the malonyl-CoA extender unit to the TcmM acyl catalyzed and function largely nonreiteratively. Type II polyketide carrier protein of the Tcm polyketide synthase is catalyzed by synthases, catalyzing the biosynthesis of aromatic polyketides, a malonyl CoA:acyl carrier protein acyltransferase believed to be part of the fatty acid synthase cluster in the same organin contrast, are multienzyme complexes that carry a single set ism.2526 of reiteratively used activities and consist of several, largely monofunctional It appears that the means to To extend our investigation of the mechanism of the Tcm determine the assembly and processing of the nascent polyketide polyketide synthase, we describe here an approach, complechain for a type I polyketide synthase was to evolve a series of mentary to the cell-free system we reported previou~ly,'~ which enzyme activities possibly arranged in the order of the steps in involves combinatorially expressing components of the Tcm and polyketide biosynthesis, as typified by the 6-deoxyerythronolide actinorhodin (Act) polyketide synthases in a suitable StreptoB synthase from Saccharopolyspora e r y t h r ~ e a . ' ~ -In ' ~ this myces host. This approach was first used by Khosla, Hopwood, case, a set of domains (known as a module) containing the and their collaborator^,^^ who showed that hybrid polyketide requisite number of active sites for substrate loading and the synthases, made by combining genes for components of two condensation between the starter unit or acyl-coenzyme A (CoA) different polyketide synthases, are often functional. The use intermediates and the chain-extender units [R(COOH)COSof S. glaucescens TcmK- and TcmL- mutants23blocked in the CoA] plus the subsequent processing events (carbonyl or double early steps of Tcm C biosynthesis provide an opportunity to bond reduction and alcohol dehydration) associated with each examine the interactions between polyketide synthase compostep of polyketide construction can be recognized in the protein nents from chromosomal and plasmid bome gene products, sequence. The mechanism of a type I1 polyketide synthase is while the heterologous host Streptomyces lividans28and the S. not as decipherable since the latter only has one active site for glaucescens TcmIc29s30mutant devoid of the normal Tcm each type of reaction, as exemplified by the tetracenomycin polyketide synthase proteins minimize the uncertainties associ(Tcm) polyketide synthase from Streptomyces g l ~ u c e s c e n s . ' ~ ~ ~ ~ated - ~ ~with the background activities. Structural analysis of Therefore, the principal challenges to understanding the catalytic metabolites produced by recombinant bacteria bearing novel mechanism of the type I1 polyketide synthases are to determine combinations of tcm and act polyketide synthase genes was how the starter unit is specified, how the length of the growing expected to provide clues from which to deduce the specific polyketide chain is controlled, how the nascent polyketide function of the individual components of the Tcm polyketide chain is folded into the correct conformation for cyclization, synthase. We chose the combination between the Tcm polyketide and how the carbonyl and methylene groups are directed to synthase and the Act polyketide synthase from Streptomyces react to form the six-membered rings of the final cyclic ~ o e l i c o l o r , ~which ~ - ~ ~forms the shunt products aloesaponarin compounds. 11, 3a, and its 2-carboxy derivative, 3b34 (derived from the pathway leading to the main S. coelicolor metabolite actinorWe have been studying the biosynthesis of Tcm C, 1, a hodin 4), because the Tcm polyketide synthase synthesizes 2, a polyketide antitumor antibiotic produced by S. glaucescens, as decaketide resulting from a linear polyketide folded at C-11, a model of the type 11polyketide synthase (Figure 1 ). On the basis of sequence analysis20-z2and in vitro s t u d i e ~ , ' ~ , ' we '-'~ (23) Glossary: tcm , act and dps designate genes encoding polyketide reported that the enzymes produced by the tcmK, tcmL, and synthase or cyclase enzymes; Tcm, Act and Dps designate the protein tcmM genesz3 are responsible for the synthesis of the nascent, products of the corresponding genes; and TCM or ACT indicate the intermediates or products of tetracenomycin and actinorhodin biosynthesis, enzyme bound linear decaketide from one acetyl-coA and nine respectively. For instance, the rcml, rcmK, tcmL, tcmM, and tcmN genes malonyl-CoA molecules, and that this decaketide intermediate (also designated as tcmlKLMN) encode the TcmJ, TcmK, TcmL, TcmM, is subsequently folded and cyclized by the T c n W protein alone and TcmN proteins; and rcmK or TcmK- and tcmL or TcmL- indicate the geneotypes and phenotypes, respectively, of mutants that lack either of these or in combination with the T c d protein to form Tcm F2, 2, enzymes. Plasmid vectors for expression of tcm or act genes are designated the earliest isolable product released from the Tcm polyketide by pWHM and bacterial strains by WMH prefixes, respectively. synthase complex (Figure l).21924 We also demonstrated that (24) Shen, B.; Nakayama, H.; Hutchinson, C. R. J. Nar. Prod. 1993,56, (10) Gramajo, H.; White, J.; Hutchinson, C. R.; Bibb, M. J. J. Bacteriol. 1991, 173, 6475-6483. (11) Shen, B.; Gramajo, H.; Bibb, M. J.; Hutchinson, C. R. J. Bacteriol. 1992, 174, 3818-3821. (12) Roberts, G. A.; Staunton, J.; Leadlay, P. F. Eur. J. Biochem. 1993, 214, 305-311. (13) Shen, B.; Hutchinson, C. R. Science 1993, 262, 1535-1540. (14) Marsden, A. F. A.; Caffrey, P.; Aparicio, J. F.; Loughran, M. S.; Staunton, J.; Leadlay, P. F. Science 1994, 263, 378-380. (15) Aparicio, J. F.; Caffrey, P.; Marsden, A. F. A,; Staunton ,J.; Leadlay, P. F. J. Biol. Chem. 1994, 269, 8524-8528. (16) Bevitt, D. J.; Cortes, J.; Haydock, S. F.; Leadlay, P. F. Eur. J. Biochem. 1992, 204, 39-49. (17) Cortes, J.; Haydock, S. F.; Roberts, G. A,; Bevitt, D. J.; Leadlay, P. F. Nature 1990. 346. 176-178. (18) Donadio, S . ; Staver, M. J.; McAlpine, J. B.; Swanson, S. J.; Katz, L. Science 1991, 252, 675-679. (19) Donadio, S.: Katz, L. Gene 1992, I l l , 51-60. (20) Bibb, M. J.; Biro, S . ; Motamedi, H.; Collins, J. F.; Hutchinson, C. R. EMBO J. 1989, 8, 2727-2736. (21) Summers, R. G.; Wendt-Pienkowski, E.; Motamedi, H.; Hutchinson, C. R. J. Bacteriol. 1992. 174. 1810-1820. (22) Summers, R. G.; Wendt-Pienkowski, E.; Motamedi, H.; Hutchinson, C. R. J. Bacteriol. 1993, 175, 7571-7580.

1288-1293. (25) Summers, R. G.; Ali. A.; Shen, B.; Wessel, W. A.; Hutchinson, C. R. Biochemistry, 1995, in press. (26) Hutchinson, C. R.; Decker, H.; Motamedi, H.; Shen, B.; Summers, R. G.; Wendt-Pienkowski, E.; Wessel, W. L. in Industrial Microorganisms: Basic and Applied Molecular Genetics; Baltz, R. H., Hegenman, G. D., Skatrud, P. L., Eds.; American Society for Microbiology: Washington, D.C., 1993; pp 203-216. (27) McDaniel, R.; Ebert-Khosla, S.; Hopwood, D. A,; Khosla, C. Science 1993, 262, 1546-1550. (28) Hopwood, D. A,; Kieser, T.; Wright, H. M.; Bibb, M. J. J. Gen. Microbiol. 1983, 129, 2257-2269. (29) Motamedi, H.; Wendt-Pienkowski, E.; Hutchinson, C. R. J. Bacteriol. 1986, 167, 575-580. (30) Decker, H.; Hutchinson, C. R. J. Bacteriol. 1993, 175, 3887-3892. (31) Femandez-Moreno, M. A.; Martinez, E.; Boto, L.; Hopwood, D. A.; Malpartida, F. J. Biol. Chem. 1992, 267, 19278-19290. (32)Sherman, D. H.; Bibb, M. J.; Simpson, T. J.; Johnson, D.; Malpartida, F.; Femandez-Moreno, M.; Maetinez, E.; Hutchinson, C. R.; Hopwood, D. A. Tetrahedron 1991, 47, 6029-6043. (33) Hallam, S. E.; Malpartida, F.; Hopwood, D. A. Gene 1988, 74,305320. (34) Bartel, P. L.; Zhu, J. S . ; Lampel, J. S . ; Dosch, D. C.; Connors, N. C.; Strohl, W. R.; Beale, J. M.; Floss, H. G. J. Bacteriol. 1990, 172,48164826.

Tefrucenomycin Polyketide Synthase from Streptomyces glaucescens

J. Am. Chem. Soc., Vol. 117, No. 26, 1995 6813

A AcW COA

mKLM

[

O

m

9xMalonflCoA

0 COSEnz 0

0

0

1-

iawI:2H tnnjHlNop

OH OH OH CHI

CH,

Tcm C (I)

Tcm F2 (2)

aclinmhdin (4)

i

OH 0

CH,

aloeaaponsrin Ii (R=H. 3a) 3.8-dihydroxy-1 -msthyianlhraquinone -2iarbor&

mid (R=C02H,3b)

Telracenomycln PKS Actinorhcdln PKS DaunONblCln PKS

d / / m KR

AT

U

O

X

H M T

Figure 1. 1. Biosynthetic pathways (A) and the gene organizations (B)for the production of Tcm C, 1. in S. glaucescens. actinorhodin, 4, in S. coelicolor, and daunorubicin in S. yeucerius. The abbreviations used are KS, 8-ketoacyl:acyl carrier protein synthase: CLF. chain length factor: ACP. acyl carrier protein: CYC, polyketide cyclase: KR. P-ketoacyl reductase: AT, acyltransferase. OX, oxygenase: MT, methyl transferase.

while the Act polyketide synthase produces 3a and 3h, octaketides derived by folding the linear polyketide at C-9 prior to the first cyclization (Figure 1A). On the basis of the presumption that functional hybrid polyketide synthases would result from expression of the fcm and act genes in the same ba~terium?~ our studies were aimed at determining the factors that control the chain length of the polyketide intermediate and the regiospecificity of the subsequent cyclizations to form the aromatic products. Five compounds, UWM1, 5, SEK4, 6, SEK4b, 7, SEKlS, 8. and SEK15h. 9, in addition to 2, were identified from the recombinant organisms (Figure 3 ). From the relationship between the composition of the hybrid polyketide synthase and the compounds produced, we conclude that the tcmKL genes (or genes whose products are quite similar to TcmK and TcmL) determine the length of the nascent polyketide chain made from acetate and malonate and that cyclases like that encoded by the tcmN gene dictate the regiospecificity of the first of a set of cyclization steps of the linear polyketide intermediate. Thus, the fcmKLMN genes govern the size and shape of the aromatic polyketides normally produced by S. glaucescens. Our results not only corroborate the findings of Khosla, Hopwood, and co-workers reported over the past year (vide infra) hut also reveal significant differences among the behavior of type Il polyketide synthases in different experimental systems.

Results The Chain Length Factor Alone Is Not Sufficient To Control the Chain Length. A typical type I1 polyketide synthase consists of three core proteins, the P-ketoacy1:acyl carrier protein synthase and its homologous companion, recently named the ”chain length factor”?7 and the acyl carrier protein, plus unique proteins called a polyketide cyclase, a P-ketoacyl reductase, and an acyltransferase, some of which may be absent in an individual type I1 polyketide synthase (Figure l e ) . Since it had been demonstrated early that interchanges of the acyl carrier protein component have little influence on the structures of the polyketides p r o d ~ c e d , 2 ’ . ’ we ~ ~ ~used ~ the TcmM acyl carrier protein throughout this study. Hybrid TcdAct polyketide synthases provided by heterologous pairs of genes (Le., combinations of genes from different species of bacteria) encoding the B-ketoacyl:acyl carrier protein synthasekhain length factor proteins were constructed in a derivative of plasmid pIJ486 where the transcription of the polyketide synthase genes is under ~~

~~

~

(35) Khosia. C.: McDaniel. R.: Eben-Khosla.S.: Toms. R.: Sherman, D. H. J. Bactrriol. 1993, 175, 2197-2204. (36) Khosla. C.: Eben-Khosla, S.: Hopwood. D. A. Molec Microhiol. 1992. 6 . 3237-3249. (37) McDaniel. R.: Eben-Khosla, S.: Howood. D. A,: Khosla, C. 3. Am. Chrm. Soc. 1993, 115, 11671-11675. (38) McDaniel. R.: Eben-Khosla. S.: Fu. H.: Hapwood. D. A,: Khosla. C. Pmc. Narl. Acad. Sci. U.S.A. 1994. 91. 11542-11546.

6814 J. Am. Chem. Soc., Vol. 117, No. 26, 1995

A

Shen ef ai.

-+

fcml-+ IaJ AAGCTCTGGACCGCCCrrGAC~C~A~GCCffiCCA~C~~CCACffiCTCAGCACG~~CAGACC~ffiCA

-

BcllORFI -+ 6: TGAGGATCCTCTAGAAGGTGBCTGAG&hAA.G -

Figure 2. Genetic arrangement of pWHM76Y. The genes cloned in the EcoRl and BamHl sites of ~11486are depicted by boxes shaded according to Figure IB. and the junctions between the genes are lettered A-D. The e m E * promoter is indicated by B bent arrow pointing in the direction of gene transcription. Only the restriction sites of interest are shown and are not necessarily unique. Gene sequence data not shown are available under GenBank accession numbers M80674 and X63449 for the fcm genes and the ncf genes, respectively. The genes for the Actb-ketoacyl:acyl carrier protein synthase and chain length factor (acfl-ORFI and actl-ORFZ, respectively) were obtained by PCR amplification from S. coelicolor M I I and the sequence of the genes was confirmed by DNA sequencing.The tcm and act genes were assembled at junction sites A-D. To maintain the possibility for translational coupling between the acf b-ketoacyl:acyl carrier protein synthase and either the nci or fcm chain length factor genes (junction C), the codon for the last residue of octl-ORFI was changed from alanine to serine. This alteration appears to be phenotypically silent. To ensure efficientexpression in S. gluucescens, the initiating TTG codon of actl-ORFI was changed to an ATG codon (iunction B). The DNA sequence of each of the fusion junctions is shown beneath the physical map of the cloned genes. Putative ribosome binding sites and the relevant start and stop codons for the cloned genes are singly and doubly underlined, respectively. The restriction enzyme sites shown in the physical map are overscored. An alternative GTG Stan codon for t c m l (shown in boldface type) is suggested by the results of site-directed mutagenesis experiments (E. Wendt-Pienkowski and C. R. Hutchinson. unpublished data).

the control of the strong, constitutive ermE* promoter (Table I).?’ Figure 2 shows details of the construction of ~WHM769,2~ a typical example of plasmids containing acf and fcmpolyketide synthase genes. The plasmids were introduced by transformation into the Tcm C non-producing strains S. glaucescens WMH1061,” -WMH1068, and -WMH1077 to examine the relationship between the gene combinations and the production of known or new metabolites. S. glaucescens WMH1061 has a point mutation in the fcmK gene,’2.” S.glaucescens WMH1068 has a deletion mutation in the fcmLgene,”.” and S.glaucescens WMH1077 has a point mutation in the promoter region of the rcmGHlJKLMNO 0peron,2~.’~-’‘’which prevents significant expression of these nine fcm genes. Table 2 summarizes the metabolites identified from the recombinant bacteria. Introduction of pWHM732 (fcmlKLMN) into either the WMH1061 or WMH1068 strains restored the production of 1 and served as a positive control, demonstrating that the plasmid home gene products are able to interact effectively with those from the chromosomal genes. As anticipated, the hybrid polyketide synthases resulting from introduction of pWHM750, pWHM751, and pWHM752, each of which carry the normal tcmK gene along with acfl-ORF2, restored the production of 1to WMH1061. However, despite the high sequence similarity between TcmK and ActI-Orfl, complementation was not observed when either pWHM762, pWHM763, or pWHM764, each carrying the act/-ORFI gene along with fcmL, was introduced into WHM1061; only a trace amount of 1 was observed with pWHM765 that has the productivity-enhancing fcml gene discussed below. The complementation results obtained with the WMH1068 strain that lacks TcmL were different. Not only the plasmids harboring the fcmL gene, such as pWHM762, pWHM763, pWHM764, and pWHM765, but also those in which the fcmL gene was replaced by acfl-ORF2,such as in pWHM750, pWHM751, and pWHM752. restored production of 1 to WMH1068. When these cassettes of hybrid polyketide synthase genes were introduced into WMH1077, no significant amount of any metabolite was observed except in the case of pWHM752. which contains fcml, fcmK, acfl-ORFZ,fcmM, and rcmN and caused the production of a mixture of 2 and 5 in small quantity, in sharp contrast to

the positive control pWHM732 hearing fcmlKLMN that produced 2 exclusively. Taken together, these results show that a heterologous pair of P-ketoacy1:acyl carrier protein synthase1 chain length factor proteins can he nonfunctional, as evident in the combinations of acfl-ORFl/fcmL that failed either to complement the mutation in WMH1061 &e., to restore Tcm C synthesis) or to synthesize any metabolite in WMH1077. However, the chain length factor seems to have a relaxed fidelity because the combination of fcmWact1-ORF2 was able to complement WMH1068, albeit less effectively than fcmWfcmL on the basis of the relative yields of 1. Since the Act polyketide synthase specifies an octaketide intermediate, the fact that fcmW acfl-ORF2 restored the production of 1 to WMH1068, via a decaketide intermediate, strongly suggests that the chain length factor alone is not sufficient to control the chain length. In fact, populations of both C I and ~ C ~metabolites O were identified as 2 and 5 in the case of the WMH1077 transfonnant carrying tcmliVacfl-ORF2/fcmMN. These results differ from the hehavior of the chain length factor encoding genes reported by Sherman, Hopwood, and co-worker~’~ where only the gralORF2 gene (encoding a protein functionally identical to ActOtfl)among the three tested was capable of restoring actinorhodin production to an acfl-ORFZ mutant hut are consistent with the report of McDonald et aI.*’ that the tcmL gene in combination with actl-ORFI did not produce a functional polyketide synthase. The Joint Activities of the /3-KetnacykAcyl Carrier Protein Synthase and Chain Length Factor Determine the Chain Length. Since the above results show that a heterologous pair of P-ketoacy1:acyl carrier protein synthasekhain length factor proteins can be nonfunctional in certain cases, we next constructed cassettes of hybrid polyketide synthase genes with the P-ketoacyl:acyl carrier protein synthasekhain length factorencoding pair from the same bacterium. We initially introduced these pIJ486-based cassettes into both S. glaucescens WMH1077 and S.lividans to examine their abilities to synthesize polyketide metabolites. Upon thin layer chromatography (TLC) andlor high performance liquid chromatography (HPLC) analysis of (39)Kim, €.A:Hopwood. D. A.: Sherman. D. H. J. Rrrcreriol. 1994. 176, IKJ-1804.

Tetracenomycin Polyketide Synthase from Streptomyces glaucescens

J. Am. Chem. SOC., Vol. 117, No. 26, 1995 6815

1-

decaketide linear intermedlat

1

tcmKLM

Acetyl CoA

+

OH SEKlS (8)

9 x Malonyl CoA

-no

OH 0 OH SEKl SI(0)

Acetyl CoA

+

spontaneous

7 x Malonyl CoA

SEK4 (6) 0

1

act/-ORFf,.%cmM

1

lntermediat

[ op 1CHI

I tc"ortcmJN

0

SEK44 (7)

..

LEN '

OH

0

cn3 o

0

ROWA~OH

OR UWMZ (IO) &CH3 UWM3 (11) R=CH&Hs

Figure 3. Proposed biosynthetic mechanisms and structures of identified polyketides produced by the act/tcm hybrid polyketide synthase plasmids in S. glaucescens or S. lividans.

the crude extracts from both hosts, we found that the product distributions in both systems were almost identical. Since the WMH1077 strain consistently gave better yields, it subsequently became the host of choice. Table 3 summarizes the major polyketide metabolites identified from the recombinant bacteria. Although the productivity varied between batches even for the same construct, the comparison of yields between different constructs are significant because they represent the average results of many fermentations. Mass spectral (MS) analysis of these metabolites led us to classify the recombinant bacteria into two groups: the

octaketide producers with pWHM766, pWHM767, pWHM768, and pWHM769 containing actl-ORF1 and -0RF2 combined with different tcm genes and the decaketide producers with pELE37, pWHM722, pWHM731, and pWHM732 containing only tcm genes. The two groups correlated with the origin of the B-ketoacy1:acyl carrier protein synthasekhain length factor pair in each construction (Table 1): the octaketide producers utilize the Act ,&ketoacyl:acyl carrier protein synthasekhain length factor proteins and the decaketide producers contain the Tcm ,&ketoacyl:acyl carrier protein synthasekhain length factor proteins.

6816 J. Am. Chem. Soc., Vol. 117, No. 26, 1995

Shen et al.

Table 1. Plasmids Used in This Study

they are encoded by heterologous pair of genes encoding these proteins if the latter results in a functional enzyme complex. Cyclases Like TcmN Appear To Determine the Folding pWHM762 actl-ORFl/tcmLM this work Pattern of the Linear Polyketide Intermediates. The isolation pWHM763 this work actl-ORFMcmLMN of 8, along with small amounts of 9, from strains bearing the pWHM764 this work tcmJ/actI- ORFUtcmLM pWHM765 this work tcmJ/actI-ORFl/tcmLMN tcmKLM genes encoding the polyketide synthase core proteins pWHM750 this work tcmK/actI-ORF2/tcmM is remarkable because 8 apparently results from folding the pWHM75 1 this work tcmK/actI-ORF2/tcmMN linear decaketide intermediate at C-9 instead of C-11, the pWHM752 this work tcmJK/actI-ORF2/tcmMN position intrinsic to the formation of 2 prior to the subsequent actI-ORFl/actI-ORF2/tcmM this work pWHM766 ring-forming cyclizations. This unexpected regiospecificity for pWHM767 actI-ORFl/actl-ORF2/tcmMN this work the first cyclization, although p r e ~ e d e n t e d , ~was ~ ? ~completely ' tcmJ/actI-ORFl/actI-ORF2/tcmM this work pWHM768 pWHM769 this work tcmJ/actI-ORFl/actI-ORF2/tcmMN diverted to the C-11 based folding pattern by the addition of pELE37 21,22 tcmKLM the tcmN or t c d N genes to the same construct, as evident in pWHM722 21,22 tcmKLMN the isolation of 2 from strains canying pWHM722 (tcmKLMiV) pWHM73 1 21,22 tcmJKLM or pWHM732 (tcmJKLMN)(Figure 3). Do these results imply pWHM732 this work tcmJKLMN that the P-ketoacy1:acyl carrier protein synthase, chain length factor, and acyl carrier protein proteins of a given type I1 The structures of the metabolites isolated were established polyketide synthase only specify the synthesis of the linear mainly by 'H and I3C NMR spectroscopic analyses and were polyketide intermediate, whose subsequent cyclizations are confirmed by isotope labeling experiments with [ 1-I3C]-, govemed completely by the polyketide cyclases such as T c d [2-I3C]-, and [ 1,2-I3C2]acetate. The metabolite, 2, synthesized and TcmN to give a particular aromatic carbon skeleton? If by strains with tcmKLMN or t c d K L M N , was identified as Tcm so, in the absence of such cyclases, the linear decaketide F2.24 The major metabolite, 8, isolated from strains bearing intermediate formed by the TcmK, -L, and -M proteins must tcmKLM or t c d K L M , was clearly a decaketide on the basis of undergo spontaneous cyclizations driven by either its kinetic the high resolution fast atom bombardment (FAB) MS analysis, reactivity or a thermodynamically-determinedproperty of the which yielded a [M HI+ ion at 385.0920 (C20H1608,calc. cyclized products to give a mixture of different carbon skeletons, 384.0845). 'H, 'W'H COSY, and I3C NMR analyses of 8 such as 8 and 9. This idea is consistent with the isolation of 6 revealed that it was identical to SEK15.@The minor metabolite, and 7 from strains bearing pWHM766 containing actl-ORFU 9, copurified along with 8, was produced in such a low quantity actI-ORF2/tcmM because each of the latter compounds arises that no significant NMR data were collected. However, it from a different folding pattem of the nascent octaketide. behaved identically to authentic SEK15b upon either TLC or To answer this question, we added tcmN or t c d plus tcmN HPLC analysis, leading us to assign 9 as SEK15b38(Figure 3). to the pWHM766 construct to test if the folding pattem of the Similar methods were applied to establish the structures of linear octaketide could be diverted from C-9 as in 6 or C-13 as the octaketide metabolites. The molecular formulas of the two in 7 to C-11. Indeed, both pWHM767 containing actl-ORFU major compounds, 6 and 7, isolated from strains bearing actlactI-ORF2/tcmMN and pWHM769 containing tcd/actI-ORFI/ ORFl/actI-ORF2/tcmM or tcd/actI-ORFl/actI-ORF2/tcmM, acti-ORF2/tcmMN synthesized a new metabolite, 5 , which were established to be Cl6H1407 (calc. 318.0740) by high reacted readily with CH30H or CH3CH20H to form adducts resolution FAB MS analysis, which gave almost identical [M 10 and 11, respectively. The mass spectra of 5 , 10, and 11 have been discussed above and Table 4 summarizes the 'H and H]+ peaks at 319.0822 and 319.0823, respectively, although I3C NMR data of 10 and 11. The [1-I3C]- and [2-I3C]acetate the fragmentation patterns of the two compounds were different. labeling experiments specifically enriched 11 at eight alternate Upon NMR analysis, 6 and 7 displayed identical 'H, 'W'H carbons with the two carbons at 55.05 and 18.51ppm unlabeled, COSY, I3C, and lH/l3C HETCOR data to those of SEK440and confirming that 5 is derived from eight acetate equivalents with SEK4b:' respectively. The structure of 5 was more elusive. the two unlabeled carbons in 11 derived from CH3CH20H. The Electron spray MS analysis of 5 revealed a molecular weight one bond I3C-l3C coupling constants for 11 were determined of 300, indicative of C16H1206.Moreover, when treated with through a [ 1,2-I3C2]acetate feeding experiment that resulted CH30H or CH3CH20H in the presence of a trace of AcOH, 5 in the assignment of the eight pairs of acetate units (Figure 4A). formed two adducts, UWM2, 10, and UWM3, 11. High The 'W'H and 'WI3C HETCOR spectra of 11 enabled the resolution electron ionization (EI) MS analyses of 10 and 11 assignment of the H-C connectivities. These assignments were yielded molecular weights of 3 14.0818 and 328.0970, suggesting unequivocally confirmed by heteronuclear multiple bond cormolecular formulas of C17H1406(calc. 3 14.0790) and C18H1606 relation (HMBC) studies (Figure 4B), resulting in the structure (calc. 328.0947), respectively. Thus, it was clear that 5 is a of 11 as shown. It is apparent that 5 will readily react with octaketide metabolite. Therefore, these results show that alcohols at its pyrone moiety to form the corresponding ketals plasmids harboring actI-ORFl/actI-ORF2 synthesized exclusuch as 10 or 11, particularly in the presence of acids like AcOH, sively octaketides and that plasmids harboring tcmUtcmL which was used during the isolation. The structure of 5, synthesized exclusively decaketides, suggesting that it is the therefore, supports the hypothesis that a type I1 polyketide joint activities of the P-ketoacy1:acyl carrier protein synthase/ cyclase can determine the folding pattern of the linear polyketide chain length factor proteins that determine the chain length of intermediate, as observed with the TcmN cyclase that initiated a polyketide made by a type I1 polyketide synthase (Figure 3). the C- 11 folding pattern for both a decaketide and an octaketide Consequently, there is a much stronger correlation between the intermediate. size of the polyketide produced when the P-ketoacy1:acyl carrier protein synthase and chain length factor proteins of the Discussion polyketide synthase are from the same bacterium than when Studies of the genetics and biochemistry of the biosynthesis (40) Fu, H.; Ebert-Khosla, S.; Hopwood, D. A,; Khosla, C. J. Am. Chem. of Tcm C (1) in S. glaucescens, along with parquel studies of SOC.1994, 116, 4166-4170. actinorhodin biosynthesis in S. coelicolor, have provided a (41) Fu, H . ; Hopwood, D. A,; Khosla, C. Chem Eiol. 1994, 1, 205210. paradigm for the biosynthesis of aromatic polyketide metabolites p 1asm id

genotype

+

+

ref

Tetracenomycin Polyketide Synthase from Streptomyces glaucescens

J. Am. Chem. SOC., Vol. 117, No. 26, 1995 6817

Table 2. Complementation of Tcm PKS Mutants by Cassettes of Hybrid act/tcm PKS Genes

WMH1061 WMH1068 WMH1077

pWHM762

pWHM763

pWHM764

-

-

-

1 -

1

1

-

-

pWHM765 1 (trace) 1 -

Table 3. Engineered Biosynthesis of Polyketides by S. Strains Bearing Hybrid acthcm PKS Genes plasmid polyketide identified ( m g k ) pELE37 8 (1): 9 (trace) 6 , 7 (trace) pWHM766 pWHM731 8 (5),b 9 (