WO2001004274A2

WO2001004274A2 - Ketoacyl synthase domains useful for priming of polyketide synthases

Info

Publication number: WO2001004274A2
Application number: PCT/US2000/018494
Authority: WO
Inventors: Stuart Smith; Anil Joshi; Vangipuram Rangan; Andrzej Witkowski
Original assignee: Children's Hospital Oakland Research Institute
Priority date: 1999-07-07
Filing date: 2000-07-06
Publication date: 2001-01-18
Also published as: AU6341500A

Abstract

The present invention provides methods to improve the efficiency of priming of a modular polyketide synthase, or to alter the specificity of priming of a polyketide synthase. β-ketoacyl synthase domains with enhanced decarboxylative activity and impaired coupling activity useful for improved or altered loading of polyketide synthases are utilized. Such β-ketoacyl synthases possess a non-nucleophilic amino acid residue at the position corresponding to residue 161 in the rat FAS. The modified β-ketoacylsynthase domain, together with an ACP and acyl transferase domain, constitute a loading module that can be used to prime polyketide synthesis. The loading module is expressed, according to the invention, either in the same polypeptide as the first functional module of the polyketide synthase or as a separate polypeptide with a suitable intermodular linker region to facilitate interaction with the first functional module.

Description

KETOACYL SYNTHASE DOMAINS USEFUL FOR PRIMING

OF POLYKETIDE SYNTHASES

GOVERNMENT SUPPORT

The U.S. Government has certain rights in this invention pursuant to Grant No. DK 16073 awarded by the National Institutes of Health.

FIELD OF THE INVENTION The present invention relates generally to the field of polyketides and polyketide synthases. In particular, the invention describes ketoacyl synthases with enhanced decarboxylase activity. The invention also describes methods for generating such ketoacyl synthases and methods for the use of such synthases for the improved priming of polyketide synthases.

BACKGROUND OF THE INVENTION Polyketides are a broad class of chemicals known to possess a wide array biological activities. Polyketides and derivatives thereof are currently in medicinal use as antibiotic agents, anticancer agents, immunosuppresive agents, cholesterol lowering agents and as veterinary agents. This remarkably broad structural class of compounds is in part classified by the common mode of biosynthesis of the diverse members of the class.

Many polyketides are synthesized by multienzyme assemblies of multidomain proteins comprising various enzymatic activities. The modules of such modular polyketide synthases (PKS) are evolutionarily related and highly similar to the animal fatty acid synthases (FAS). The multifunctional FAS has proven a valuable paradigm for elucidation of the functional organization and programming rules that enable the modular polyketide synthases to synthesize a wide variety of macrolide products (Donadio, S. et al. (1991) Science 252:675-679; Donadio, S. and Katz, L. (1992) Gene 111 :51-60). The PKS modules are used in a genetically prescribed sequence to perform elongation reactions, so that the intermediate product formed by one module becomes the substrate for the next, adjacent module. In particular, β-ketoacyl synthase (KS) domains catalyze the formation of new carbon- carbon bonds by condensation of a variety of acyl-chain precursors with an elongating carbon source, usually malonyl or methylmalonyl moieties, that are covalently attached via thioester linkage to an acyl carrier protein (ACP). These intermediates may be variably processed by each of the modules to leave either a keto, hydroxy, enoyl or fully reduced β-carbon atom. Complete β-carbon processing requires three activities, the β-ketoacyl reductase, dehydrase and enoyl reductase. The presence or absence of functional β-carbon processing enzymes in particular modules determines the final structure of the polyketide product.

All of the various members of the β-ketoacyl synthase family of enzymes likely utilize the same basic reaction mechanism involving successively, transfer of the first substrate from the ACP phosphopantetheine to a cysteine nucleophile, decarboxylation of the second substrate, either malonyl- or me hylmalonyl-ACP, to yield the carbanion, and finally a nucleophilic attack of the carbanion on the carbonyl of the cysteine-bound acyl substrate resulting in the formation of a new carbon-carbon bond. In the absence of the first substrate, decarboxylation of malonyl moieties takes place at a rate far lower than when accompanied by a condensation reaction. Presumably this phenomenon reflects the importance to the enzyme of generating the reactive carbanion species only when a recipient acyl chain is positioned on the active-site cysteine, thus avoiding the loss of malonyl moieties and establishment of a futile cycle. For instance, in the case of a derivative of the 6-deoxyerythronolide B synthase (DEBS), it has been shown that the purified synthase does not catalyze the decarboxylation of methylmalonyl-CoA, whereas this reaction is carried out in the course of the coupling reaction. (KJ. Weissman et al. (1998) Biochemistry 37:11012-11017)

Most β-ketoacyl synthases involved in the biosynthesis of fatty acids, polyketides and mycolic acid precursors share appreciable sequence similarity and are clearly related both structurally and evolutionarily (Siggaard- Andersen, M. (1993)

Prot. Seq. Data. Anal. 5:325-335). Thus, from multiple sequence alignments, residues have been identified that may play critical roles in catalysis. In addition to the cysteine nucleophile (Cys-161 in the rat FAS), three basic residues are universally conserved (corresponding to His-293, Lys-326 and His-331 in the rat FAS) that likely play important roles either catalytically or structurally (Fig. 1). A conserved glycine- rich region near the C-terminus has also been implicated as facilitating the entry of substrates into the active-site pocket (Huang, W. et al. (1998) EMBO J. 17:1183- 1191).

The co-linear relationship between the organization of catalytic modules and the final structure of polyketides has been exploited for the production of novel polyketides. A number of different strategies have been pursued to achieve diversity. The total number of carbon atoms has been changed by altering the number of modules present. The presence of keto, hydroxy and enoyl groups has been changed by altering the complement of β-carbon processing enzymes present in a given module. Katz (U.S. Pat. No. 5,824,513) provides recombinant methods for producing polyketides related to erythromycin through alteration of the genes encoding the deoxyerythronolide B synthase. Katz ('513) provides a number of means for altering the gene including: inactivation, addition, or deletion of beta- carbonyl processing domains; inactivation, activation, or deletion of domains involved in condensation to control the length of the polyketide; and substitution of one acyltransferase domain for an isologous one. These methods do not recognize the importance of loading modules for the control of polyketide synthesis. U.S. Pat. No. 5,712,146 claims libraries of polyketide synthase producing cell lines wherein the genes directing polyketide synthase production are composed of open reading frames (ORFs) derived from at least three different polyketide synthases. The '146 Patent does not recognize the importance of loading modules for the control of polyketide synthesis. Neither the '146 or '513 Patent recognizes the utility of utilizing ketoacyl synthase domains with impaired coupling activity and improved decarboxylative activity for priming polyketide synthesis.

An alternative approach that has been undertaken for the production of novel polyketides involves supplying synthetic polyketide precursors that enter the polyketide synthase later in the sequence, effectively bypassing the loading module (Chuck, J. et al. (1997) Chem. Biol. 4:757-766; Dutton, C.J. (1994) Tet. Lett. 35:327- 330).

The choice of the PKS or FAS priming moiety, for example: acetate, propionate, isobutyrate, etc., is determined by the specificity of the enzymes associated with the loading module, located immediately to the amino terminus of the first fully- functional elongating module. The loading modules of modular polyketide synthases typically consist of an acyltransferase (AT) and acyl carrier protein (ACP) domain. The acyltransferase picks up the primer substrate from its coenzyme A form and transfers it to the acyl carrier protein domain. From there the primer is translo- cated to the active-site cysteine residue of the ketoacyl synthase domain associated with module 1 , the first fully competent module, where it begins the elongation process. The 6-deoxyerythronolide B polyketide synthase, for instance, contains only an acyltransferase and ACP domain (Donadio, S. et al. (1991) Science 252:675-679). Yet other polyketide synthases lack loading modules entirely, as in the pyoluteorin polyketide synthase (Nowak-Thompson, B. et al. (1997) Gene 204:17-24).

In some modular polyketide synthases, loading modules are either specialized to facilitate the transfer of unusual primers, for example a dihydrocyclohexylcarbonyl moiety in the FK506 (Motamedi, H. and Shafiee, A. (1998) Eur. J. Biochem. 256:528-534) and rapamycin (Schwecke, T. et al. (1995) Proc. Natl. Acad. Sci USA 92:7839-7843) polyketide synthases. In some other modular polyketide synthases, specificity is quite broad. For example, the Streptomyces avermitilis avermectin-producing polyketide synthase can accept at least 44 different branched carboxylic acids. (C J. Dutton et al. (1991) J. Antibiot. 44:357- 365). The polyketide synthase responsible for the synthesis of erythromycin (DEBS) contains a loading module specific for propionyl-CoA (Brown, M.J.B. et al. (1995) J. Chem. Soc. Chem. Comm. 1995:1517; Kao, CM. et al. (1994) Science 265:509). Replacement of the DEBS loading module with that from the avermectin PKS significantly broadens the specificity of the DEBS (Marsden, A.F.A. et al. (1998) Science 279:199-202). A second strategy that has been pursued for altering the specificity of polyketide synthesis is to remove the priming module altogether, therefore removing whatever specificity it imparts to the polyketide synthase. Removal of the priming module of 6-deoxyerythronolide B synthase (DEBS) showed that it is not essential for the formation of erythromycin. (Pereda, A. et al. (1998) Microbiology 144:543-553). Removal of the priming module resulted in significant reduction of erythromycin production. Addition of the priming module, through coexpression as separate protein, increased the expression significantly, but not to the levels of the wild-type system.

The modular polyketide synthase loading modules of some other microorganisms contain an additional N-terminal ketoacyl synthase domain that is characterized by substitution of the usual active-site cysteine with a glutamine residue (referred to as the KS domain: KS=ketoacyl synthase, Q, the single letter code for glutamine). With the exception of the cysteine nucleophile, all of the known active- site residues of ketoacyl synthase domains are also well conserved in the KS domains that are associated with the loading modules of the modular polyketide synthases responsible for the synthesis of the macrolides 10-deoxymethynolide and narbonolide (S. venezuelae), platenolide (S. ambiofasciens) and the macrolide portion of niddamycin (S. caelestis). These KS domains are incapable of catalyzing the chain extension reaction that is the usual role for a ketoacyl synthase. The role of this unusual KS domain was heretofore unknown, nor was the utility of loading modules containing the KS module recognized.

Polyketides are an important class of pharmaceutical agents. Improving the biosynthetic yields of known or novel polyketides would be advantageous. Methods that can be used to alter the specificity of polyketide synthases to produce novel or unnatural polyketides are also highly desirable. While other methods have been identified for the production of polyketides, there still remains a need in the art for improved methods of polyketide synthesis. The present invention achieves this and other objectives by providing improved polyketide synthesis methods using ketoacyl synthase domains for the improved priming of polyketide synthases as well as improved methods using so-called chain extension moieties for priming polyketide synthesis. The present invention provides β-ketoacyl synthases and synthase domains with increased decarboxylative activity and impaired coupling activity useful for the improved priming or loading of polyketide synthases. BRIEF DESCRIPTION OF THE FIGURES

Figure 1 illustrates the alignment of the amino acid sequences from four regions of β-ketoacyl synthases. Figure 1 A shows the region corresponding to residues 157-167 of the rat FAS; Figure IB shows the region corresponding to residues 290-305 of the rat FAS; Figure 1C shows the region corresponding to residues 324-338 of the rat FAS; Figure ID shows the region corresponding to residues 391-402 of the rat FAS. Those β-ketoacyl synthases associated with multifunctional (type I) are distinguished by asterisks. Abbreviations used are: MAS, mycolic acid synthesis; PKS, polyketide synthase; FAS, fatty acid synthase; C, Caenorhaebditis; M., Mycobacterium; Sa., Saccharopolyspora; St. Streptomyces; E., Escherichia. The numbering system is for the rat FAS. Consensus residues, defined by 75% compliance, are denoted by black fill and conservative replacements by grey fill. The GenBank sequence identification numbers (Sequence ID) for the sequences are, in the order listed: 1, 66561; 2, 2117715; 3, 3876624; 4, 547900; 5, 416965; 6 & 7, 3800834; 8, 2558838; 9, 2317860; 10, 729876; 11, 119784; 12, 294666; 13, 3800749; 14, 1261947; 15, 1261948; 16, 119783; 17, 729460.

Figure 2 shows the elimination of the requirement for acetyl-CoA by the wild- type FAS in the presence of the CyslόlGln rat FAS mutant. Spectrophoto- metric assays were performed at 37 degrees C. All reaction mixtures contained in 0.2 mL: 0.1 M potassium phosphate, pH 6.6; 0.25 mM NADPH; 135 μM malonyl-CoA; and 50 μM CoASH. In addition, individual assays contained: (a) 10 μg CyslόlGln FAS and 65 μM acetyl CoA; (b) 1.5 μg wild-type FAS (no acetyl-CoA added); (c) wild-type and CyslόlGln FASs, 1.5 μg each (no acetyl-CoA added); (d) 1.5 μg wild- type and 4.5 μg CyslόlGln FASs (no acetyl-CoA added); (e) 1.5 μg wild-type and 10 μg CyslόlGln FASs (no acetyl-CoA added), additional malonyl-CoA was added at the time marked by the arrow; (f) 1.5 μg wild-type FAS and 65 μM acetyl-CoA.

Figure 3 shows general strategies for engineering of novel modular PKSs with covalently-attached loading modules containing KS^Q domains. Examples are shown of strategies that can be employed to introduce KS -containing loading modules (consisting of KS - AT- ACP domains) by covalent attachment to modules of other PKSs. In these examples, the hypothetical modular PKSs consist of three polypeptides, each containing two modules. Polypeptides and modules are identified by a superscript letter indicating the parental PKS. Loading modules are denoted by the letter 'L', those that contain a KS domain are distinguished by the suffix KS ; the KS loading module can be of the type that specifically decarboxylates either malonyl or methylmalonyl moieties. The linker regions that separate loading modules from the adjacent functional module are denoted by cross-hatching. Diagonal cross- hatching indicates the intermodular region associated with the KS -containing loading module from modular PKS A. Horizontal cross-hatching indicates the intermodular region associated with the first module of modular PKS B. Figure 4 illustrates engineering of novel modular PKSs with transacting loading modules containing a KS domain. Two examples are shown in which a KS -containing loading module is engineered so as to replace its carboxy-terminal linker with the carboxy-terminal linker of another PKS module that will facilitate interaction of the loading module with the amino-terminal linker of the module immediately downstream. The KS -containing loading module can be of the type that specifically decarboxylates either malonyl or methylmalonyl moieties. Complementary carboxy-terminal and amino-terminal interpoiypeptide 'linker regions' are shown in matching cross-hatching patterns (e.g., diagonal cross-hatching indicates peptide regions that facilitate interaction between the C-terminus of module 2 or PKS B and the N-terminus of module 3 of PKS B).

SUMMARY OF THE INVENTION It is one object of the current invention to provide β-ketoacyl synthases with increased decarboxylase activity by mutation of a conserved active site cysteine residue corresponding to residue 161 in the rat FAS. In a preferred embodiment of the invention, this active-site cysteine is replaced with glutamine. In another preferred embodiment, the active site cysteine is replaced with alanine.

A further object of the invention is to provide for the incorporation of such β-ketoacyl synthase domains with increased decarboxylative activity and impaired coupling activity into the loading modules of polyketide synthases to provide polyketide synthases with enhanced or improved priming. A loading module of the present invention contains a β-ketoacyl synthase domain and preferably also contains an acyl transferase domain and an acyl carrier protein domain. In one aspect of the invention, the acyl transferase domain is specific for an extender or elongating moiety, preferably malonyl-, methylmalonyl-, or ethylmalonyl-CoA. The loading module is expressed operably linked to the desired first functional module of the polyketide synthase. The incorporated loading module may replace all or part of the precursor polyketide synthase, and preferably replaces the precursor polyketide synthase loading module, if one were originally present.

Yet another object of the present invention is to provide β-ketoacyl synthase domains with increased decarboxylative activity and impaired coupling activity that are useful for priming polyketide synthases in trans (i.e., expressed in a loading module which is a distinct polypeptide from the polyketide synthase). In a preferred embodiment, a loading module containing such a synthase domain is coexpressed with a polyketide synthase in an appropriate organism. In other preferred embodiments, the loading modules also contain acyl transferase domains specific for either malonyl or methylmalonyl moieties. In a more preferred embodiment, the loading module contains peptide linker regions that facilitate physical interaction with the polyketide synthase. In a second preferred embodiment, such loading modules are used in vitro to enhance the loading of a polyketide synthase. The β-ketoacyl synthase protein also preferably incorporates a linker peptide region that facilitates its interaction with the first functional module of the polyketide synthase to be primed.

DETAILED DESCRIPTION Definitions

In this document, except when noted, use of the term polyketide synthase should be read to include fatty acid synthases and mycolic acid synthases. Loading and priming are used synonymously throughout the specification in reference to the process of providing the primary fragment to the polyketide synthase.

Improved or enhanced loading should be read to include alteration of either the yield or specificity of polyketide synthase loading. "Extender moiety," "extender" and "elongation moiety" refer to activated chemical groups used in the stepwise chain extension of a polyketide. Typically, the "elongation moieties" are malonyl, methylmalonyl, and less typically, ethylmalonyl. Extender moieties may be carboxylic acids or be found in the activated thioester form.

"Priming moieties" refer to activated chemical groups used in the loading or priming of polyketide synthases. Most commonly, priming moieties are acetyl, propionyl or butyryl.

"Operably linked" when describing the relationship between two polynucleotide regions simply means that they are functionally related to each other. For example, the polynucleotides encoding two domains are operably linked if they are capable of being expressed so as to produce a product containing both domains in the desired reading frame. "Operably linked" also may describe the protein products encoded by a polynucleotide or separate polynucleotides. For instance, modules in a modular polyketide synthase are operably linked if the product of one enzymatic domain can serve as the substrate for the operably linked domain or module. "Complementary intermodular linker regions" or "complementary linker regions" can facilitate the operable linkage of proteins to form multi -polypeptide complexes.

"Complementary linker regions" are pairs of protein sequences that possess sufficient affinity for one another to promote the formation of intermolecular complexes in solution. Such linker regions or sequences particularly are those derived from polyketide synthase that facilitate assembly of the polyketide synthase.

"Purification" encompasses any procedure which results in the separation of the purified composition from at least one other composition with which it is found prior to purification.

Precursor Polyketide Synthases

DNA sequences of polyketide synthases are readily identified by homology to known polyketide synthases. Polyketide synthases are known in a large number of organisms especially in the Actinomycetes. Examples of Actinomycetes that produce polyketides include but are not limited to Micromonospora rosaria, Micromonospora megalomicea, Sacharapolyspora erythraea, Streptomyces antibioticus, Streptomyces albireticuli, Streptomyces ambofasciens, Streptomyces avermitilis, S. vene∑uelae, Streptomyces caelestis, Streptomyces fradiae, Streptomyces hygroscopicus, Streptomyces tsukubaensis, Streptomyces griseus, Streptomyces mycarofasciens, Streptomyces platensis, Streptomyces venezuelae, Streptomyces violaceoniger, and various Actinomadura, Dactylosporangium and Nocardia strains that produce polyether type of polyketides. This invention can be applied to the sequences of known polyketide synthase genes or to polyketide synthases yet to be discovered that possess a significant degree of homology to a known polyketide synthase.

Nucleic acids encoding polyketide synthases suitable for the present invention also may be produced by genetic engineering. Such a polyketide synthase has at least one module, but more typically has several modules either derived from different organisms or engineered from distinct domains derived from a single organism. The modules can be arranged in a single polypeptide chain, as in the FAS or in multiple polypeptide chains that assemble to form a multi-protein complex. Nucleic acids encoding polyketide synthases or polyketide synthase loading modules also can be constructed synthetically using oligonucleotide synthesis. One advantage of creating synthetic genes is to optimize the codon usage within the genes being used for optimal expression in a host organism of choice. Such host codon optimizations are routine in the art.

Ketoacyl Synthase Domain Variants

From the sequence of an isolated polyketide synthase gene, the module and domain structure readily can be identified by homology to known domains. Homology among this class of genes is relatively high. For the purposes of this invention, genes are aligned with suitable gapping to maximize homology and similarity. Domains are identified as ketoacyl synthase domains if they are substantially similar to known domains. Substantial sequence similarity is defined as amino acid residue identity of at least 20% between two or more sequences or between a sequence and a consensus sequence, and preferably an identity of at least 50%, more preferably 70% and most preferably 90% with suitable gapping to maximize homology. Such sequence alignments and comparisons can be made by a number of computational methods known to one skilled in the art (see, e.g., Siggaard- Andersen, M. (1993) Prot. Seq. Data. Anal. 5:325-335). In particular, ketoacyl synthase domains share a consensus active-site protein sequence motif of T(or A,G or S)A(or G)CS(or ATG)S(or ATG).

From the sequence alignment, the active site residue corresponding to cysteine residue 161 in the rat FAS can be identified. If the identified ketoacyl synthase domain DNA codes for a cysteine residue at the position corresponding to 161 in the rat FAS, it can be mutated to another residue by techniques well known in the art (see e.g., Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) "Molecular Cloning: A Laboratory Manual," Second Edition, Cold Spring Harbor, N.Y.; Ausubel, F.M., Brent, R., Kingston, R.E., Moore, D.D., Seidman, J.G. and Struhl, K. (Eds.) (1988) "Current Protocols in Molecular Biology," John Wiley, New York). The amino acid chosen is preferably a non-nucleophilic amino acid. "Non-nucleophilic amino acids" are alanine, asparagine, glutamine, glycine, isoleucine, leucine, and phenylalanine and valine. In a preferred embodiment, the chosen amino acid is glutamine. Non-natural or non-standard amino acids also may be used in the present invention provided that they do not contain nucleophihc side-chain functional groups such as alcohols, thiols or amines.

Acyl Transferase Domains

It is preferable to use the ketoacyl synthase variants of the present invention in conjunction with a specific acyl transferase ("AT") domain. Acyltransferase domains showing specificity for malonyl or methylmalonyl moieties have distinct consensus sequences (Haydock, S. et al. (1995) FEBS Lett. 374:246- 248.) Table 1 compares example AT domain sequences to the consensus sequences derived by Haydock et al. The acyltransferase domains that lie adjacent to the KS^Q domains of the modular polyketide synthases responsible for the synthesis of tylactone, in S. fradiae, and 10-deoxymethynolide and narbonolide, in S. venezuelae, contain the conserved sequence elements typical of the methylmalonyl-specific acyltransferases, whereas that associated with the niddamycin polyketide synthase, in S. caelestis contains the sequence elements characteristic of the malonyl-specific class of acyltransferases. Previously, it was not recognized that these domains could supply the substrate for decarboxylation by the adjacent KS domain. Thus, based on their chemical structures, it can be deduced that tylactone, 10-deoxymethynolide and narbonolide must be synthesized from a propionyl primer (derived from a methylmalonyl moiety) whereas niddamycin is synthesized from an acetyl primer (derived from a malonyl moiety) (Kuhstoss, S. et al. (1996) Gene 183:231-236; Kakavas, S.J. (1997) J. Bacteriol. 179:7515-7522; Xue, Y. et al. (1998) Proc. Natl. Acad. Sci. USA 95:12111-12116).

Table 1. Acyl Transferase Domain Consensus Sequences

Malonyl SEQ. ID NO. 8 ETGYA QxAxFGLL GHSxG Consensus¹

Niddamycin SEQ. ID NO. 9 ^"33RTEYT QTALYRTL GHSVG

Methylmalonyl SEQ. ID NO. 10 RVDVV MxSxAAxW GHSQG Consensus¹

Tylactone SEQ. ID NO. 11 ^"33RVDVV MVSLARY GHSQG

10-Deoxy- SEQ. ID NO. 12 ^"33RVDVV MVSLARVW GHSQG methynolide/ narbonolide

"x" indicates a position that is not conserved. ¹ Haydock, S. et al. (1995) FEBS Lett. 374:246-248

The nucleic acids encoding such specific AT domains can be obtained readily for known AT domains, such as those discussed above, by amplification by

PCR from genomic material derived from an appropriate organism. For example, malonyl-specific AT domains are known in modules 2, 5, 8, 9, 11, 12 and 14 of the rapamycin polyketide synthase from Streptomyces hygroscopicus (Schwecke, T. et al.

(1995), Proc. Natl. Acad. Sci. USA 92, 7839-7843); in modules 1, 2, 3 and 7 of the niddamycin polyketide synthase from Streptomyces caelestis (Kakavas, S.J. et al.

(1997) J. Bacteriol. 179:7515-7522) and in the 'pik' module 2 of the 10- deoxyerythronolide and narbonolide polyketide synthase of Streptomyces venezulae (Xne, Y. et al. (1998) Proc. Natl. Acad. Sci. USA 95, 12111-12116). Methylmalonyl- specific AT domains are known in, for example, modules 1, 3, 4, 6, 7, 10 and 13 of the rapamycin polyketide synthase from Streptomyces hygroscopicus (Schwecke, T. et al. (1995)), in modules 4, 5 and 6 in the niddamycin polyketide synthase from

Streptomyces caelestis (Kakavas, S.J. et al. (1997)), in modules 1, 2, 3, 4, 5 and 6 in the erythromycin polyketide synthase from Saccharopolyspora erythreae (Schwecke, T. et al. (1995)), and in modules 1, 3, 4, 5 and 6 of the 10-deoxymethynolide and narbonolide polyketide synthase from Streptomyces venezulae (Xue, Y. et al. (1998)). Other nucleic acids also may be used that share substantial sequence similarity with the aforementioned malonyl and methylmalonyl consensus sequences.

Loading Modules

Loading modules according to the present invention preferably contain a ketoacyl synthase domain with increased decarboxylative activity and no coupling activity, an acyl transferase domain and an acyl carrier protein ("ACP") domain.

A nucleic acid encoding a loading module of the invention can be derived from a suitable polyketide synthase gene. For instance, the loading modules of polyketide synthases responsible for the synthesis of tylactone, in . S. fradiae (GenBank accession No. U78289), of 10-deoxymethynolide and narbonolide, in S. venezuelae (GenBank accession No. AF079138), and of niddamycin, in S. caelestis (GenBank accession No. AF016585), contain KS ketoacyl synthase domains, a specific acyl transferase domain and an acyl carrier protein domain.

Alternatively, a nucleic acid encoding a loading module of the invention can be constructed by one skilled in the art, using the standard techniques of molecular biology, to contain these domains such that they are operably linked . The ketoacyl synthase domain is preferably a KS domain derived from a naturally occurring polyketide synthase or a KS domain constructed according to the present invention. The AT domain is chosen as discussed above. The ACP domain may be any ACP domain from a polyketide synthase, but is preferably an ACP domain from a naturally occurring KS domain. The nucleic acid encoding the loading module of the invention can be propagated in an appropriate vector, preferably in a plasmid. The nucleic acid also can be assembled by ligating the nucleic acids encoding the three domains and used directly for the construction of a polyketide synthase.

Production of Polyketide Synthase Variants In a method of the present invention, polyketide synthases with improved priming are produced by assembly of a gene encoding a KS -containing loading module with that of a polyketide synthase. In particular, the method is useful for generating polyketide synthases that utilize malonyl, methylmalonyl or ethylmalonyl as priming moieties. The method is generally applicable to any precursor polyketide synthase. The nucleic acid encoding the precursor polyketide synthase first is isolated. Suitable precursor polyketide synthases include all naturally occurring polyketide synthases and designed polyketide synthases that are of a modular nature as discussed above. The precursor PKS gene is isolated by PCR, restriction digestion or any other suitable technique known to one with skill in the art. The precursor PKS gene also may be assembled from distinct modules by genetic engineering.

The first step is analyzing the domain structure of a polyketide synthase gene to be modified. The domain structure is best determined by analyzing the amino acid sequence predicted from the polyketide synthase gene and comparing the sequence information to sequences of domains with known function, such as those disclosed above. The invention also may be practiced without knowledge of the entire primary sequence.

The loading module containing a KS domain can be recombinantly introduced into a polyketide synthase gene of interest. Introduction can be achieved by a variety of techniques known to one skilled in the art, including homologous recombination or through subcloning through use of appropriate restriction endonucleases (see e.g., Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) in "Molecular Cloning: A Laboratory Manual," Second Edition, Cold Spring Harbor, N.Y.; Ausubel, F.M., Brent, R., Kingston, R.E., Moore, D.D., Seidman, J.G. and Kevin Struhl (Eds.) (1988) "Cuπ-ent Protocols in Molecular Biology," John Wiley, New York; Hopwood, D.A. et al. (1985) "Genetic manipulation of Streptomyces. A laboratory manual. The John Innes Foundation: Norwich). If the entire sequence of the precursor polyketide synthase is known, the gene first is amplified from the desired first functional domain to at least the end of the open reading frame, thus creating a polyketide synthase lacking a loading module. The desired first functional domain may be the first functional domain present in the PKS or may be a later domain, chosen such that the PKS variant produced will synthesize a polyketide composed of fewer extender units. The nucleic acid encoding the loading domain of the invention can be introduced by ligation, recombination or any other suitable cloning technique.

If the PKS of interest does not contain a loading module or for synthetic PKS designed without loading modules, the DNA encoding the ketoacyl synthase domain variant of this invention along with DNA encoding a specific acyl transferase domain and acyl carrier protein can be introduced to the PKS gene such that they are operably linked to the precursor polyketide synthase gene and will be expressed at the amino-terminal side of the module to be used as the first functional module of said PKS. In particular, it is advantageous to include an acyl transferase domain specific for either malonyl or methylmalonyl moieties.

In one application, the KS^Q loading module from either a naturally occurring PKS that uses a propionyl starter (i.e. the KS^Q-AT-ACP module specifically decarboxylates methylmalonyl moieties) or one that uses an acetyl starter (i.e. the KS -AT-ACP module specifically decarboxylates malonyl moieties) is used either to replace an existing 'non-KS module in the PKS of interest, or is added to a PKS that lacks any loading module (Fig. 3). In the case of the loading module replacements, the intermodular linker between the loading module and the first functional module can originate either from the KS^Q 'donor module' (e.g. Fig. 3, diagonal hatches, in unnatural modular PKS D) or the first module of the 'acceptor PKS' (e.g. Fig. 3, horizontal hatches, in unnatural modular PKS E); in either case, the linker sequence is that which separates the ACP of the loading module from the ketoacyl synthase of the first functional module. In cases where a KS loading module is introduced to a PKS that previously lacked a loading module, then the linker used must be that which is associated with the KS^Q donor loading module (e.g. Fig. 3, diagonal hatches, in unnatural modular PKS F). PKS D is engineered from PKS B by replacing its loading module and linker with the KS - containing loading module and its linker from PKS A. PKS E is engineered similarly except that the linker associated with the original loading module of PKS B is retained.

In the case of a polyketide synthase lacking a loading module, a KS - containing loading module can be added to the amino-terminus of the precursor polyketide synthase. PKS F is engineered from PKS C, which lacks a loading module, by the introduction, at the amino-terminus of module 1, of the KS - containing loading module and its linker from PKS A.

The site of integration of a KS^Q-containing loading module need not be limited to the amino-terminus of the first module. A KS -containing module may be placed in front of any downstream module. In the examples shown in Fig. 3, the KS^Q- containing loading module from PKS A, together with its carboxy-terminal linker, is fused to the amino-terminus of module 2 and module 3 of PKS B, yielding PKSs G and H. PKS H is engineered from PKS B by replacement of the amino-terminal linker of module 3 with the KS^Q-containing loading module and its linker from PKS A and also by deletion of polypeptide I . Modifications in the loading module can be combined with any other modifications, mutations or replacements in the modules responsible for catalysis of the chain extension reactions.

Two examples are shown in Figure 4 in which a KS^Q-containing loading module is engineered so as to replace its carboxy-terminal linker with the carboxy-terminal linker of another PKS module that will facilitate interaction of the loading module with the amino-terminal linker of the module immediately downstream. The KS -containing loading module can be of the type that specifically decarboxylates either malonyl or methylmalonyl moieties. Complementary carboxy- terminal and amino-terminal linker regions are shown in matching cross-hatching patterns. Thus novel PKSs D and E are engineered by replacing the linker normally associated with the KS -containing loading module from PKS A, with the carboxy- terminal interpolypeptide linker derived from either module 2 of PKS B or from module 4 of PKS C. In PKS D, the KS loading module will provide the priming substrate for the module that previously constituted module 3 of PKS B, whereas in PKS E, the KS loading module will provide the priming substrate for the module that previously constituted module 5 of PKS C. The advantage of this approach is that once KS -containing loading modules have been identified, or constructed, only modification of the carboxy-terminal linker region is required to facilitate its functional interaction with any other amino-terminal PKS module.

A second approach places a functional KS module in trans with any amino-terminal module of any PKS. This approach allows initiation of polyketide synthesis, via a substrate-specific KS loading module, at any amino-terminal module in any PKS. The strategy for introducing a trans-zcung KS domain incorporates a design feature based on the recent paper by Gokhale, R.S., Tsuji, S.Y., Cane, D.E. & Khosla, C. [Dissecting and exploiting intermodular communication in polyketide synthases. Science 284, 482-485 (1999)]. In this case one optimizes the probability of the KS loading module interacting with the functional module of interest by replacing the intermodular linker that follows the ACP domain of the KS^Q module with the linker that is found naturally at the carboxy-terminus of the module that would normally catalyze the elongation step preceding that catalyzed by the functional module of interest. The strategy is illustrated in Fig. 4. Three hypothetical naturally occurring modular PKSs are shown, A, B and C. PKS A has a loading module containing a KS domain, whereas PKSs B and C have loading modules lacking a KS^Q domain. The unnatural PKSs D and E are constructed, respectively, by taking the KS -containing loading module from the naturally occurring PKS A and replacing the linker that connects it to module 1 with the linker sequence derived from the carboxy-terminus of module 2 associated PKS B or of module 4 of PKS C. These 'interpolypeptide linkers' facilitate interaction with the complementary linker at the amino-terminus of module 3 associated with PKSs B or C [Gokhale et al. (1999)]. Modifications in the loading module can be combined with any other modifications, mutations or replacements in the modules responsible for catalysis of the chain extension reactions. Host Cells and Vectors

The PKS gene constructed by the methods of the present invention, is preferably cloned in a vector containing suitable control sequences for regulating the transcription of the polyketide synthase gene. The choice of regulatory region chosen will be dependent on the host cell chosen for expression of the polyketide synthase. It is generally preferred that the regulatory sequence for polyketide synthases be from the host cell in which the precursor polyketide synthase occurs naturally. For instance, an engineered Erythromycin synthase was expressed using the wild-type regulatory sequence from S. erythraea (Marsden et al. (1998) Science 279:199-202). The preferred vector of the invention also serves as a shuttle for the introduction of the polyketide synthase into a suitable host organism capable of producing the polyketide synthase. The plasmid can be propagated extrachromasomally or can be designed to be stably integrated into the host cell's genome by recombination. Host cells suitable for use in the present invention depend on the desired method of polyketide production. Most commonly, host cells will be polyketide producing cell lines in which the wild-type polyketide synthase gene or genes have been deleted or disabled. Such host cells have been described derived from S. coelicolor and also from S. lividans (Ziermann, R. and Betlach, M.C., Biotechniques 1999 26:106-10). It is also possible to produce polyketides in cells that normally do not produce polyketides, such as E. coli and S. cerevisiae (Kealey J.T. et al. (1998) Proc. Natl. Acad. Sci. USA 95:505-509; Ghokale et al. (1999)).

Polyketide production also can occur in vitro using purified PKS proteins. For this method, the vector chosen preferably encodes a signal sequence for the secretion of the PKS product from a suitable host cell, such as E. coli. Other host cells also may be used including yeast cells, e.g. Pichia pastoris and S. cerevisiae, mammalian cells, plant cells, and insect cells. The polyketide synthase also may be purified from an Actinomycete or other polyketide producing host. Polyketide Production

Polyketide production can be carried out within the host organism. In the case of a suitably chosen polyketide-producing host cell, cells transformed with the nucleic acid encoding the polyketide synthase of the invention are cultured under conditions that favor polyketide production. The DNA encoding the novel modular PKSs is cloned into an appropriate expression plasmid (e.g. pRM5; Khosla et al. U.S. Pat. No. 5,672,491) and used to transform the host cells (e.g. Streptomyces coelicolor CH999; Marsden et al. 1998). Typically expression is continued for several days after which the fermentation broth is extracted with ethyl acetate and the polyketide products purified by thin layer and high-performance liquid chromatography and characterized by nuclear magnetic resonance spectrometry and electrospray mass spectrometry. Preferred host cells are derived from polyketide producing cells by deleting or inactivating the wild-type polyketide synthase. A particularly preferred host cell is S. coelicolor strain CH999, from which the entire gene cluster for actinorhodin production has been deleted. Host cells also may be used that are not derived from polyketide producing cells. E. coli BL21 (DE3) cells have been used to produce active polyketide synthases for purification and characterization in vitro. When using such cells, it is advisable to coexpress a phosphopantetheinyl gene (e.g. the sfp gene from Bacillus subtilis) along with the polyketide synthase genes. This is necessary since E. coli does not efficiently phosphopantetheinylate foreign ACPs (Gokhale, R.S. et al. (1999) Science 284, 482-485).

The PKS variant of the invention also can be purified and used to produce polyketides in vitro. In particular, it is advantageous to express the polyketide synthase polypeptides as fusion proteins with an affinity domain. An affinity domain useful for the invention has the property of specifically binding to a partner domain with an interaction strength sufficient to effect purification. A large number of affinity domains known in the art of protein purification are useful for the present invention. Particularly useful affinity domains useful for the invention include hexahistidine tags, that bind to chelated metal supports. Fusion with an affinity domain may be at the N-terminus or at the C-terminus. If N- or C-terminal affinity tags are found not to aid in purification (perhaps as a result of being inaccessible to the surface of the protein) the affinity domain instead can be fused at a position internal to the PKS module sequence, particularly at those positions homologous to those chosen for purification of FAS (Joshi, A.K. et al. (1998) J. Biol. Chem. 273, 4937-4943). Of course, the site of affinity tag fusion must not interfere with the proper function of the PKS, which may be assayed using methods described herein or otherwise known in the art.

Detecting improved priming by the polyketide synthase variant of the invention relative to the precursor polyketide synthase can be carried out in a number of ways known to one with skill in the art of polyketide production. For instance, it is common to carry out polyketide production in the presence of a labeled primer or extender substrate. Isotopically labeled malonyl and methylmalonyl moieties are particularly useful for monitoring the invention. Polyketide production is carried out

13 2 in the presence of the C and H labeled primer moiety for detection by nuclear magnetic resonance spectroscopy, which is particularly useful for identifying the polyketide product. Radioactive isotopes also can be incorporated. Use of malonyl and methylmalonyl moieties labeled with H or ¹⁴C results in the production of radioactive polyketides suitable for analysis and detection using thin layer chromatography (TLC), high performance liquid chromatography (HPLC), capillary electro- phoresis (CE), or any other applicable technique known to one with skill in the art.

Without intending to limit the invention in any way, particular embodiments of the present invention are illustrated in the following Examples.

EXAMPLE 1

Construction of a ketoacyl domain variant Construction ofcDNAs Encoding His(,. and FLAG-Tagged FASs and

Expression of the Proteins in Sf9 cells. The strategies for construction of cDNAs encoding the wild-type FAS, single domain-specific mutants and for introduction of His₆ or FLAG tags have been described in detail previously (Witkowski, A. et al. (1996) Biochemistry 35:10569-10575; Joshi, A.K. et al. (1998) Biochemistry 37:2515-2523; Joshi, A.K. et al. (1998) J. Biol. Chem. 273:4937-4943). The Cyslόl mutant FASs were constructed by first generating mutated partial cDNA fragments by polymerase chain reaction amplification, using pFAS 74.20 (partial FAS cDNA in pUCBM20) as a template together with the appropriate primers (Joshi, A.K. and Smith, S. (1993) Biochem. J. 296:143-149). Oligonucleotide FASl 152B (SEQ. ID NO. 1 : 5'- cac tag aat tcT TCA GGG TTG GGG TTG TGG AAA TGC; upper case characters correspond to bp 1152-1176 of the rat FAS cDNA, lower case characters represent nucleotides added for the introduction of restriction sites) was used as the antisense primer. The mutagenic sense primers were derived from bases 530-575 of the rat FAS cDNA (GenBank Accession No. M84761 ; GenBank peptide sequence: 66561). SEQ. ID NO. 2 contains the triplet GCA encoding the Cyslόl Ala mutant. SEQ. ID NO. 3 contains the triplet GGA encoding the CyslόlGly mutant. SEQ. ID NO. 4 contains the triplet AGC encoding the CyslόlSer mutant. SEQ. ID NO. 5 contains the triplet ACC encoding the CyslόlThr mutant. SEQ. ID NO. 6 contains the triplet AAT encoding the Cyslόl Asn mutant, and SEQ. ID NO. 7 contains the triplet CAA encoding the Cyslόl Gin mutant.

SEQ. SEQUENCE ID NO.

2 5' -TCAAAGGACC CAGCATTGCC CTGGACACAG CCGCATCCT CTAGCCT

3 5' -TCAAAGGACC CAGCATTGCC CTGGACACAG CCGGATCCT CTAGCCT

4 5' -TCAAAGGACC CAGCATTGCC CTGGACACAG CCAGCTCCT CTAGCCT

5 5' -TCAAAGGACC CAGCATTGCC CTGGACACAG CCACCTCCT CTAGCCT

6 5' -TCAAAGGACC CAGCATTGCC CTGGACACAG CCAATTCCT CTAGCCT

7 5' -TCAAAGGACC CAGCATTGCC CTGGACACAG CCCAATCCT CTAGCCT

Authenticity of the amplification products was confirmed by DNA sequencing and the appropriate fragments were moved stepwise into the full-length, wild-type construct. The final constructs for Cysl61Ser/Gln/Asn mutants also encoded a C-terminal Hisό affinity tag and those for the Cysl61Ala/Gly mutants encoded a C-terminal FLAG tag, although the presence of the tags was not exploited in these studies. These FAS cDNA constructs, in the context of the pFASTBAC 1 vector (Life Technologies, Inc., Rockville, MD), were used to generate recombinant baculovirus stocks by the transposition method employing a baculovirus expression system (Luckow, V.A. et al. (1993) J. Virol. 67, 4566-4579). Sβ cells were then infected with the purified recombinant viruses and cultured for 48 h at 27 °C. The tagged FAS proteins were partially purified from the cytosols as described earlier (Joshi, A.K. and Smith, S. (1993) Biochem. J. 296: 143-149) and then subjected to final purification by affinity chromatography (Joshi, A.K. et al. (1998) J. Biol. Chem. 273:4937-4943); glycerol (10%, v/v) was included in all buffers used for chromatography .

EXAMPLE 2 Assay for Decarboxylation of Malonyl and Methylmalonyl Moieties

Decarboxylase activity was assayed by quantification of acetyl-CoA, β-ketobutyryl-CoA and triacetic acid lactone ("TAL") formed from malonyl-CoA. Enzymes were incubated at 37 °C or 10 °C for 1 to 2 min with 110 μM [2-¹⁴C]malonyl-CoA and 50 μM CoASH in 0.2 M potassium phosphate buffer, pH 6.6; CoASH was omitted from the reaction when activity of the wild-type FAS was assayed. Reactions were quenched with perchloric acid (Witkowski, A. et al. (1996) Biochemistry 35:10569-10575) and the products were identified by reversed- phase HPLC (Joshi, A.K. and Smith, S. (1993) J. Biol. Chem. 268:22508-22513). Triacetic acid lactone was identified in the eluate by comparison with the elution position of a chemically synthesized standard. Overall decarboxylase activity was calculated based on the total amount of acetyl units formed from malonyl-Co A and corresponds to acetyl moieties that were released from the enzyme by transfer to CoASH, or utilized as primers for condensation and subsequently released as either β- ketobutyryl-CoA or triacetic acid lactone. Methylmalonyl decarboxylation was assayed under similar conditions except that [2- C]malonyl-CoA was replaced by non-radioactive methylmalonyl-CoA. In this case, the product, propionyl-CoA, was identified from its elution position on reversed-phase HPLC. A unit of activity is equivalent to 1 μmole of malonyl or methylmalonyl moieties decarboxylated per minute at the specified temperature. The Cys 161 Gin mutation increases malonyl decarboxylase activity

150-fold over that of the wild type ("wt") FAS. Replacement of Cys-161 with other residues either reduces, or only slightly increases, malonyl decarboxylase activity. In the case of the wild-type FAS, most of the acetyl moieties produced by decarboxylation of malonyl moieties are condensed with either one or two malonyl moieties, resulting ultimately in the release of either β-ketobutyryl-CoA or triacetic acid lactone from the enzyme. However, in the case of the Cyslόl Gin mutant, no condensation can take place and all of the acetyl moieties produced are released as acetyl-CoA (Table 2).

The malonyl decarboxylase activity associated with the Cyslόl Gin mutant is stimulated approximately two-fold by the addition of 20-50 μM CoASH, indicating that the malonyl moieties likely are decarboxylated following transfer to the FAS, so that release of the β-ketobutyryl product requires the addition of CoASH as an acceptor. In the case of the wild-type FAS, where the rate of decarboxylation is relatively low, and where much of the product is released as triacetic acid lactone, no free CoA needs to be added in order to achieve maximum activity; presumably the free CoASH released on transfer of the malonyl moieties from CoA ester to the FAS is sufficient to meet the acceptor requirement for release of β-ketobutyryl moieties. That the decarboxylation of malonyl moieties requires the initial loading of malonyl moieties onto the phosphopantetheine thiol attached to Ser-2151 of the acyl carrier protein (ACP), via the active-site serine residue of the malonyl/acetyltransferase ("MAT"), Ser581, is also supported by the observation that mutation of either of these serine residues to alanine eliminates decarboxylase activity (Table 2).

The ability of the modified FAS to catalyze the decarboxylation of malonyl moieties, in the absence of an accompanying condensation, results in the formation and release of acetyl-CoA. The rate of decarboxylation is at least as great at that which accompanies the condensation reaction and results in the formation and release of β-ketobutyryl-CoA by the wild-type FAS (Table 2). Conversion of the condensing enzyme into a potent malonyl decarboxylase can be attributed uniquely to the replacement of the cysteine nucleophile, preferably with glutamine. Table 2. Characterization of FASs Mutated or Chemically Modified at Cysteine-161.

Malonyl Decarboxylation β- Products ketobutyryl

FAS FAS CoA Activity β-ketobut acetyl synthesis synthesis TAL CoA CoA munit/mg munit/mg munit/mg % % % wt 2029±43 130 ± 10 3.3 ± 0.2 49 49 2

C161Q* 0 0 495 ± 16 0 0 100

Iodoacet- 53 ± 10 2.8 ± 0.1 6.4 ± 0.3 0 15 85 amide- treated wt C161S 13 ± 1 5.8 ± 0.7 1.3 ± 0.2 7 42 51

C161A* 0 0 7.0 ± 0.1 0 0 100

C161N 0 0 1.5 ± 0.1 0 0 100

C161G 0 0 0.2 ± 0.0 0 0 100

C161T 0 0 0

S2151A 0 0 0.0 ± 0.0 (ACP-)t S581A 0 0 0.0 ± 0.1 (MAT-)J

All reactions were carried out at 37 °C.

* Methylmalonyl decarboxylation rates assessed for the C161Q and C161 A mutants were 120±7 and 1.5+0.2 munits/mg, respectively; the only product detected was propionyl-CoA. Units of activity are μmole of NADPH oxidized per minute for the FAS assay; for the other assays, units are μmole of product formed per minute. ⁺ Mutation of these residues abolishes ACP or MAT activity.

EXAMPLE 3 Priming FAS with a mutant Ketoacyl Synthase Domain Assay for Fatty Acid Synthesis-Overall fatty acid synthesizing activity was measured spectrophotometrically at 37 °C or 10 °C (Smith, S. and Abraham, S. (1975) Meth. Enzymol. 35:65-74). A unit of activity is equivalent to 1 μmole of NADPH oxidized per min at the specified temperature.

The Cyslόl Gin mutant is inactive in the spectrophotometric assay for fatty acid synthesis (Fig 2a). In this assay, the activity of the wild-type FAS is dependent on the presence of the co-substrate, acetyl-CoA, as well as malonyl-CoA and NADPH (Fig. 2, compare b and f). However, when the CyslόlGln mutant FAS is included in the same assay together with the wild-type FAS, the dependency on added acetyl-CoA is eliminated (Fig. 2, compare b with c, d and e). In reaction mixtures containing the highest levels of the Cyslόl Gin mutant, the initial rate of reaction is similar to that observed in the presence of added acetyl-CoA but drops to zero after about 8 min (Fig. 2e); the addition of more malonyl-CoA at this time sparks the reaction back to the original rate (denoted by arrow). These results are consistent with the Cyslόl Gin mutant having catalyzed the decarboxylation of malonyl-CoA thus providing a supply of acetyl moieties required by the wild-type FAS. In the presence of high levels of the mutant, the availability of malonyl-CoA eventually becomes rate-limiting for the wild-type FAS (Fig. 2e).

Replacement of the cysteine nucleophile with Gin, Ser, Ala, Asn, Gly or Thr completely eliminates the ability to catalyze the condensation reaction and consequently eliminates overall FAS activity (Table 2). Only replacement of Cys-161 with serine produces a β-ketoacyl synthase with residual catalytic activity, as described earlier (Joshi, A.K. et al. (1997) Biochemistry 36:2316-2322).

EXAMPLE 4

Modification of FAS by iodoacetamide The FAS storage buffer was replaced with 0.25 M potassium phosphate, pH 5.8, containing 1 mM EDTA and 0.5 mM tris(2- carboxyethyl)phosphine (Calbiochem-Novabiochem Corp., San Diego, CA) by centrifugation through a BioGel P-30 (Bio-Rad Labs, Hercules, CA) gel filtration column (Penefsky, H.S. (1977) J. Biol. Chem 252:2891-2899). The modification reaction was carried out at 20 °C using 0.73 mM iodoacetamide; to protect the phosphopantetheine thiol from modification, 0.96 mM malonyl-CoA was included in the reaction mixture. The reaction was quenched by addition of mercaptoethanol to 10 mM and the reaction buffer was replaced with storage buffer by repeated dilution and concentration in a Centricon-100 device (Amicon, Inc., Beverly, MA). No condensation products are produced by the Cyslόl Ala, Cyslόl Asn or CyslόlGly mutants, which also lack an appropriate nucleophile at position 161, although in these cases the rate of formation of acetyl-CoA is much lower than for the

CyslόlGln mutant (Table 2). For comparison the properties of the iodoacetamide- treated FAS were also studied. The iodoacetamide treated FAS possesses a side chain

[CH₂SCH₂CONH₂] that is very similar to that of the Gin side-chain

[-CH₂CH₂CONH₂]. As has been reported by others, the β-ketoacyl synthase and overall FAS activities are markedly lowered by the modification, whereas the malonyl decarboxylase activity is increased (Kresze, E.-B. et al. (1977) Eur. J. Biochem. 79:191-199; Tomoda, H. et al. (1984) J. Biochem. 95:1705-1712); most of the acetyl moieties produced are released directly by transfer to CoA. However, the increment in decarboxylase activity is modest when compared to that produced by the

Cys 161 Gin mutation.

EXAMPLE 5 Replacement of the DEBS loading module and its intermodular linker with the

Niddamycin loading module 1 and its intermodular linker.

This example illustrates that a naturally occurring KS containing loading module can be used to replace the loading module of a polyketide synthase to change the substrate specificity. An approximately 3 kb fragment of the niddamycin loading module is amplified using the polymerase chain reaction (PCR) from DNA isolated from S. caelestis using the primer set nid.Tl/nid.Bl (SEQ. ID NO. 13 and 14). The resulting fragment is purified using QIAquick PCR purification kit (QIAGEN, Valencia, CA). The eluted fragment is digested with Nhel and Hindlll restriction endonucleases, purified by agarose gel electrophoresis and the DNA fragment eluted from the gel using GENECLEAN kit (Bio 101, Vista, CA). A -460 bp fragment of the first module of DEBS ("DEBS 1 ") is amplified by PCR using the primer set debs.Tl/ debs.Bl (SEQ. ID NO. 16 and 17), digested with Hindlll and Sfil endonucleases and purified as above. The two fragments are ligated together through the Hindlll site and cloned into appropriately digested plasmid pCK12, which carries the entire DEBS 1+ TE domain coding sequence (Kao,C.M., Luo,G., Katz,L., Cane,D.E. and Khosla,C, J. Am. Chem. Soc. 117, 9105 (1995)), through Nhel & Sfil sites, resulting in an in-frame replacement of DEBSl loading domain by niddamycin loading domain.

The resulting plasmid is transformed into S. coelicolor CH999 host cells. The cells are plated on R2YE medium plates containing 50μg/liter thiostrepton and grown at 30°C (Hopwood, D. A. et.al, Genetic Manipulation of Streptomyces. A laboratory Manual (John Innes Foundation, Norwich, United Kingdom (1985)). Select colonies are grown in liquid R2YE culture media including 50μg/liter thiostrepton. The cells are cultured, then centrifuged, and the liquid media extracted with ethyl acetate. The procedure is repeated with C-labeled malonyl-CoA. The polyketide produced is purified by reversed phase HPLC using a gradient of acetonitrile in water. The mass of the purified polyketide, determined by electrospray ionization mass spectrometry, shows incorporation of an acetyl rather than a propionyl moiety. The identity of the polyketide produced is also confirmed using C-NMR. The C-NMR shows a signal expected for the incorporation of a labeled acetyl moiety.

Thus an acetyl moiety, produced by the niddamycin loading module through decarboxylation of a malonyl moiety, is now used to initiate polyketide synthesis in place of the propionyl moiety normally supplied by the DEBS loading module. EXAMPLE 6

Replacement of the DEBS loading module with the Niddamycin loading module. In this variation of the strategy of Example 5, the intermodular linker originates from DEBS rather than from niddamycin. The same switch from a propionyl starter to an acetyl starter is observed. All procedures are performed as described in Example 5. The -3 kb fragment containing the niddamycin loading module is isolated as described above. A -530 bp fragment of debs 1 is obtained by PCR using the primer set debsl T2/debsl Bl (SEQ. ID NO. 17 and 19). Both fragments are digested with Hindlll, purified and ligated to give an -3.5 kb fragment. The fragment, as above, is ligated to the DEBSl module of plasmid pCK12 through the Nhel and Sfil sites. The plasmid is transformed in to S. coelicolor CH999 host cells. Polyketide production and identification are carried out as described in Example 5. As in the preceding example, the modified PKS is primed with decarboxylated malonyl moieties, resulting in the incoporation of an acetyl rather than propionyl starter unit.

EXAMPLE 7 Introduction of the niddamycin loading module into the pyoluteorin PKS. In this example, the niddamycin loading module, together with its associated intermodular linker is placed in front of module 1 of the pyoluteorin PKS. The pyoluteorin PKS, which lacks a loading module of its own, normally uses a starter unit derived from proline by a mechanism yet unknown. The novel PKS generated by introduction of the niddamycin loading module would utilize an acetyl starter derived from decarboxylation of a malonyl moiety.

Plasmid pPYOl is first genarated from the parental plasmid pCK7 (Kao,C.M., Katz,L.and Khosla,C. Science 265, 509(1994)) by replacing the three eryA genes between restriction sites Pad and Nsil, with an ~12kb fragment generated by long range PCR (primer set pyo T2/pyo B2), encompassing the two pyoluteorin synthase genes plt &pltC (Thompson, B.N., Gould, S.T., Loper, J.E. (1997) Gene 204:17). As above, in Examples 5 and 6, the -3 kb fragment encoding the niddamycin loading module is amplified, digested with Hindlll and purified. A -3 kb fragment of the pyoluteorin module 1 is amplified using the primer set pyo.Tl/pyo.Bl (SEQ. ID NO. 21 and 22). The pyoluteorin fragment is digested with Hindlll and Kpnl and ligated through the Hindlll site to the niddamycin loading domain. The resulting ~6kb chimeric fragment is next used to replace the Pacl-Kpnl fragment of pltB gene in pPYOlto to generate plasmid pPYO2. Plasmid pPYO2 is used to transform CH999 host cells as above. The cells are cultured and the polyketide produced as desribed above. EXAMPLE 8

Engineering of a trans-acting PikA loading module interacting with module 3 of DEBS.

In this example the KSQ containing PikA loading module from S. venezuelae is expressed as an individual protein and functions in trans. The loading module is expressed fused to the -60 amino acid interpolypeptide linker derived from the carboxy-terminus of DEBSl. The linker region interacts with the complementary amino-terminal interpolypeptide linker of DEBS2 and facilitates direct functional communication of the loading module with module 3 of DEBS2. Cells are transformed with the plasmid encoding the KSQ containing loading module and with a plasmid encoding module 3 of DEBS2, such that both polypeptides are produced.

A -3.2 kb nucleic acid fragment encoding the loading module from Streptomyces venezuelae is amplified by the PCR using primer set pik.Tl /pik.Bl (SEQ. ID NO. 23 and 24). The fragment is purified by agarose gel electrophoresis and digested with Hindlll. The DEBSl carboxy-terminal interpolypeptide linker (-156 bp) is amplified by PCR using the primer set debs.T3/debs.B2 (SEQ. ID NO. 18 and 20). The fragment encoding the interpolypeptide linker fragment is digested with Hindlll and ligated to the above -3.2 kb fragment derived from the pik loading module. The resulting -3.4 kb fragment is purified by agarose gel electrophoresis and used to replace the entire eryA gene in plasmid pCK7. The resulting plasmid pPikl is used to transform S. coelicolor CH999 cells. The cells are cultured as described above. Cells expressing the PikA loading module, DEBS2 & DEBS3 produce a multi- polypeptide assembly that functions as a polyketide synthase. The KSQ containing loading module associated with module 3 in trans primes module 3 with acetyl moieties derived from PikA loading domain.

The following primers are used for amplification of genetic sequences for examples 6-8 are shown in Table 3. Table 3. Primer sequences.

Primer SEQ. Sequence¹³ location³ ID No. nid.Tl SEQ. ID 5 ' -attaqctaqcATGGCAGGGCATGGTGACGCCA 604- No. 13 Nhe I 625 nid.Bl SEQ. ID 5 ' - acataaqcttGATCGGATCACCCGCCTTCGCCT 3608-

No. 14 Hind III 3630 md.B2 SEQ. ID 5 ' -atataaqcttCGCCAGGTGGTCGGCCACCGCCCGT 3516- No. 15 Hind III 3540 debs.Tl SEQ. ID 5 ' - atataaqcttGTCGTCGCGATGGCCTGCCGGCT 2268-

No. 16 Hind III 2290 debs.T2 SEQ. ID 5 ' - 2196-

No. 17 at a aaqc t tGCGGGAACCGAGGTCGCACAACGGGAA 2222 Hind I I I debs.3 SEQ. ID 5 ' -atataaqcttAACGCCTCCGCGGTCGCCGGTT 10926-

No. 18 Hind I I I 10947 debs. l SEQ. ID 5 ' -attaGGCCCTCCAGGCCGAGCGTGTA 2727- No. 19 Sfi I 2748 debs.2 SEQ. ID 5 ' - 11194-

No. 20 atatqqatccTCAATCGCCGTCGAGCTCCCGGCCGA 11219 BarriΑ I pyo.Tl SEQ. ID 5 ' - atataaqcttGCAATCATCGGGAGTGGATGCCGCT 4878-

No. 21 Hind III 4902 pyo.B l SEQ. ID 5 ' - acqqcqqtaccGGCCCCACGGTGT 8026-

No. 22 Kpn I 8049 pιk.Tl SEQ. ID 5 ' -attaqctaqcATGTCTTCAGCCGGAATTACCAGGA 2912-

No. 23 Nhe I 2947 pιk.Bl SEQ. ID 5 ' 6128-

No. 24 at ataaqcttGAGAGCCTCGGGGGTGGGGAAGTCGA 6153 Hind I I I pyoT2 SEQ. ID 5 ' - GACTGAATGGATGCTCGTGCGCCCAT 4836-

No. 25 4861 pyoB2 SEQ. ID 5 ' - CGTCAGGCCTCGGCCACGCAGCCCT 17569- No. 26 17593

"The nucleotide numbers for various genes are according to: [1] Niddamycin: Kakavas, S. J., Katz, L. & Stassi, D. Identification and characterization of the niddamycin polyketide synthase genes from Streptomyces caelestis. J. Bacteriol. 179:7515-7522 (1997)

[2] DEBS: Donadio, S., Staver, M. J., McAlpme, J. B., Swanson, S. J. & Katz, L. Modular organization of genes required for complex polyketide biosynthesis. Science 252, 675-679 (1991). [3] Pyoluteorin: Nowak-Thompson, B., Gould, S. J. & Loper, J. E. Identification and sequence analysis of the genes encoding a polyketide synthase required for pyoluteorin biosynthesis in Pseudomonas fluorescens Pf-5. Gene 204, 17-24 (1997). [4] Xue, Y., Zhao, L., Liu, H -W. & Sherman, D. H. A gene cluster for macrolide antibiotic biosynthesis m Streptomyces venezuelae: architecture of metabolic diversity. Proc. Natl. Acad. Sci. USA 95, 12111-12116 (1998)

Upper case letters in o go sequence mdicate the naturally occurring nucleotides, lower case letters indicate the nucleotides engineered mto the ohgomer The locations of the restriction sites are underlined

All publications, issued patents, and patent publications referenced in the instant application are hereby incorporated by reference in their entirety.

Claims

CLAIMS What is claimed is:

1. A polyketide synthase generated by a method comprising:

(a) isolating a first polynucleotide encoding a first polyketide synthase;

(b) identifying a first functional module in said first polyketide synthase;

(c) identifying a second polynucleotide encoding a loading module, wherein said loading module is not naturally associated with said first polyketide synthase, and wherein said loading module comprises:

(i) an acyl carrier protein; (ii) an acyl transferase domain; and

(iii) a ketoacyl synthase domain variant with a non- nucleophilic residue at the position corresponding to residue 161 in the rat fatty acid synthase (Seq. Id. No. 27); and

(d) co-expressing said first and said second polynucleotides to generate a second polyketide synthase wherein the priming substrate for the first functional module is provided by the loading module.

2. The polyketide synthase of Claim 1 wherein said second polynucleotide is operably linked to said first polynucleotide such that said loading module and said first functional module are expressed as a single polypeptide chain.

3. The polyketide synthase of Claim 2 wherein said acyl transferase domain is specific for priming with malonyl moieties.

4. The polyketide synthase of Claim 2 wherein said acyl transferase domain is specific for priming with methylmalonyl moieties.

5. The polyketide synthase of Claim 2 wherein said acyl transferase domain is specific for priming with ethylmalonyl moieties.

6. The polyketide synthase of Claim 2 wherein said non-nucleophilic residue is glutamine.

7. The polyketide synthase of Claim 6 wherein said acyl transferase domain is specific for priming with malonyl moieties.

8. The polyketide synthase of Claim 6 wherein said acyl transferase domain is specific for priming with methylmalonyl moieties.

9. The polyketide synthase of Claim 6 wherein said acyl transferase domain is specific for priming with ethylmalonyl moieties.

10. A polynucleotide encoding the polyketide synthase of Claim 2.

11. A host cell transformed with the polynucleotide of Claim 10.

12. A method for producing a polyketide comprising the steps of: a) transforming a host cell with the polynucleotide of Claim 10; b) growing said host cells under appropriate conditions for polyketide production; and c) purifying said polyketide.

13. The polyketide synthase of Claim 1 wherein said second polynucleotide is operably linked to said first polynucleotide such that said loading module and said first functional module are expressed as separate polypeptide chains with complementary intermodular linker regions.

14. The polyketide synthase of Claim 13 wherein said acyl transferase domain is specific for priming with malonyl moieties.

15. The polyketide synthase of Claim 13 wherein said acyl transferase domain is specific for priming with methylmalonyl moieties.

16. The polyketide synthase of Claim 13 wherein said acyl transferase domain is specific for priming with ethylmalonyl moieties.

17. The polyketide synthase of Claim 13 wherein said non-nucleophilic residue is glutamine.

18. The polyketide synthase of Claim 17 wherein said acyl transferase domain is specific for priming with malonyl moieties.

19. The polyketide synthase of Claim 17 wherein said acyl transferase domain is specific for priming with methylmalonyl moieties.

20. The polyketide synthase of Claim 17 wherein said acyl transferase domain is specific for priming with ethylmalonyl moieties.

21. The first and second polynucleotides encoding the polyketide synthase of Claim 13.

22. A host cell transformed with the polynucleotides of Claim 21.

23. A method for producing a polyketide comprising the steps of: a) transforming a host cell with the polynucleotide of Claim 21; b) growing said host cells under appropriate conditions for polyketide production; and c) purifying said polyketide.