Using Chemobiosynthesis and Synthetic Mini-Polyketide ... · polyketide, hence the need for imaginative genetic engineering approaches to overcome these challenges. The epothilones,

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Aug. 2010, p. 5221–5227 Vol. 76, No. 150099-2240/10/$12.00 doi:10.1128/AEM.02961-09Copyright © 2010, American Society for Microbiology. All Rights Reserved.

Using Chemobiosynthesis and Synthetic Mini-Polyketide Synthases ToProduce Pharmaceutical Intermediates in Escherichia coli�†

Hugo G. Menzella,* John R. Carney,‡ Yong Li,§ and Daniel V. Santi¶Kosan Biosciences Inc., Hayward, California 94545

Received 8 December 2009/Accepted 29 May 2010

Recombinant microbial whole-cell biocatalysis is a valuable approach for producing enantiomerically pureintermediates for the synthesis of complex molecules. Here, we describe a method to produce polyketideintermediates possessing multiple stereogenic centers by combining chemobiosynthesis and engineered mini-polyketide synthases (PKSs). Chemobiosynthesis allows the introduction of unnatural moieties, while a libraryof synthetic bimodular PKSs expressed from codon-optimized genes permits the introduction of a variety ofketide units. To validate the approach, intermediates for the synthesis of trans-9,10-dehydroepothilone D weregenerated. The designer molecules obtained have the potential to greatly reduce the manufacturing cost ofepothilone analogues, thus facilitating their commercial development as therapeutic agents.

Whole-cell biocatalysis is a rapidly developing technologyused to assist in developing synthetic routes to complex mole-cules in the pharmaceutical industry. The exquisite regio- andstereoselectivity of enzymes allows the facile introduction ofstereogenic centers with complete enantiomeric control, whichmay result in a significant reduction in the number of synthesissteps and therefore the final production cost (2, 14, 18). Inaddition, biocatalysis is one of the greenest technologies cur-rently available; since the protection and deprotection of func-tional groups are not required, high- and low-temperaturereactions can be circumvented, and organic solvents are notused (17).

Type I modular polyketide synthase (PKS) genes determinethe biosynthesis of valuable polyketide natural products, suchas erythromycin, epothilone, and many others. These genesencode enzymes consisting of modules of active sites (do-mains) that build the carbon chain of the final product in astepwise fashion using acyl-coenzyme A (CoA) starter andextender units (15).

Most of the known PKSs are microbial enzymes and possessa variable number of modules (Mod) preceded by a loadingdidomain (LM). The LM is composed of an acyl transferase(AT) domain that selects the starter acyl-CoA unit and an acylcarrier protein (ACP) domain that receives the acyl groupfrom the loading AT. The acyl group then is transferred to thefirst extender module and successively to downstream mod-ules. All extender modules contain an essential set of three

domains: ketosynthase (KS), AT, and ACP. The KS receivesthe acyl unit from the preceding module, while the AT trans-fers an appropriate acyl extender unit from its CoA ester to theACP. The KS then catalyzes a condensation between theacyl-KS and the �-carbon of the extender acyl-ACP to give anacyl-ACP. Additional domains may be present in some mod-ules and are responsible for the reduction of the keto groups ofthe growing polyketide chain. For example, modules may con-tain a ketoreductase (KR) that reduces the �-keto group ste-reospecifically to an alcohol. At the end of the assembly line, athioesterase (TE) domain on the C terminus of the last ex-tender module cleaves the polyketide chain from the PKS andconverts it to a lactone.

Thus, the structure of the two-carbon unit dictated by amodule is determined by the specificity of its AT domain, itscomplement of reductive domains, and carbon branch stereo-chemistry; the order of modules determines the sequence oftwo-carbon units in the polyketide product, and the number ofmodules determines carbon chain length.

Since the early 1990s, many research groups have been in-terested in understanding the rules of module-module interac-tions so as to genetically engineer microorganisms to createnovel polyketides (7). An ultimate goal is to produce complexmolecules by creating synthetic PKSs to be used directly asdrugs or as lead compounds for chemical optimization. Mean-while, even the combination of a few PKS modules can pro-duce molecules with multiple chiral centers (up to two permodule) that are difficult to obtain by chemical synthesis (5,21), thus assisting in the production of complex moleculescurrently made by total chemical synthesis.

Although the biosynthesis of polyketides found in nature isconfined to those that can be assembled with natural acyl-CoAprecursors, this limitation often can be overcome using che-mobiosynthesis (3, 9, 12, 20). Here, unusual chemical moietiesmay be introduced as the first unit of a polyketide chain byfeeding a PKS that has been disabled or deleted in an earlyextension module with a chemically synthesized carboxylic acidN-acetyl-cysteamine thioester (SNAC). In successful cases, thesynthetic thioester acylates the KS of the module immediately

* Corresponding author. Present address: Facultad de Ciencias Bio-quimicas y Farmaceuticas, Universidad Nacional de Rosario, Suipacha531, Rosario 2000, Argentina. Phone and fax: 54-341-4350661. E-mail:[email protected].

‡ Present address: Solazyme, Inc., 561 Eccles Avenue, South SanFrancisco, CA 94080.

§ Present address: TerraBay Pharmaceuticals Inc., 2 HuaTian Rd.,Suite 7012, Tianjin, P. R. China, 300384.

¶ Present address: Department of Pharmaceutical Chemistry, Box2240, 600 16th Street N457B, University of California, San Fran-cisco, CA.

† Supplemental material for this article may be found at http://aem.asm.org/.

� Published ahead of print on 11 June 2010.

5221

on January 18, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

downstream of the disabled/deleted one and is faithfullylengthened by subsequent extender modules.

There are several challenges in using chemobiosynthesis as ageneral approach to making a desired polyketide. First, it is notpossible to rationally predict whether a particular SNAC willbe accepted and processed by a given PKS module and, if it is,whether the extension of the unnatural starter unit will providean acceptable yield of product. Second, if the SNAC is ex-tended by the first module, it cannot be predicted whether theforeign polyketide chain will be extended by subsequent mod-ules. Finally, it is improbable that a naturally occurring PKSwill possess the appropriate sequence of modules necessaryto extend the SNAC and produce the desired unnaturalpolyketide, hence the need for imaginative genetic engineeringapproaches to overcome these challenges.

The epothilones, a family of polyketide compounds naturallyproduced by the myxobacterium Sorangium cellulosum, haveemerged as promising anti-cancer agents (10, 13, 19). Thepotent synthetic analogue trans-9,10-dehydroepothilone D(Fig. 1A), which is in human clinical trials, currently is pre-pared by complete chemical synthesis requiring 23 operationsto provide an overall yield of only about 1% (4). The analogue26-trifluoro trans-9,10-dehydroepothilone D, another promis-ing clinical candidate (16), likewise is difficult to prepare. Bio-synthetic approaches to making such analogues are especiallychallenging, because the C-4-gem-dimethyl group and the dou-ble bond at C-9–C-10 are rarely found in natural polyketides,and the routes for their incorporation into engineered biosyn-thetic pathways are largely unknown.

In this work, we describe an approach to create novelpolyketides with an unusual starter unit and an engineeredpattern of extension. Our strategy was validated by the biosyn-thesis of intermediates that facilitate the synthesis of epothi-lone analogues.

MATERIALS AND METHODS

Host and vectors. The Escherichia coli polyketide producer strain K207-3(BL21�prpBCD::T7prom prpE, T7prom accA1-pccB, T7prom sfp) and the pAngand pBru series of vectors for the expression of PKS constructs have beendescribed previously (8).

E. coli strain DH5� was used for plasmid preparation. The pAngII series ofvectors for the expression of PKS modules flanked by the N-terminal linker ofery5 (LNery5) and the C-terminal linker of ery2 (LCery2), designated LNery5-Mod-LCery2, were created by removing the NdeI-MfeI fragment of the pAngplasmids and inserting a DNA fragment containing a codon-optimized version ofLNery5 flanked by identical restriction sites. All of the vectors used in this workare listed in Table S1 in the supplemental material.

SNAC feeding to bimodules. K207-3 bacteria harboring pAngII donor plas-mids and pBru acceptor plasmids were grown in 2.5 ml LB with carbenicillin (50�g/ml) at 37°C to an optical density at 600 nm of 0.5. Cultures were induced withisopropyl-�-D-thiogalactopyranoside (IPTG; 0.5 mM) and arabinose (2 mg/ml),and 0.5 ml of a mixture of sodium glutamate (50 mM), sodium succinate (50mM), sodium propionate (5 mM), and SNAC 1 (1 mM) was added. Afterincubation at 22°C for 24 h with agitation, bacteria were removed by centrifu-gation, and supernatants were acidified with phosphoric acid to pH 2.5 andanalyzed after at least 30 min for polyketide production by liquid chromatogra-phy-mass spectrometry-mass spectrometry (LC-MS-MS).

Protein expression analysis. Samples (1 ml) of each culture were centrifugedat 14,000 � g for 3 min, resuspended in 1 ml 20 mM Tris, 150 mM NaCl, pH 7.5,and lysed by sonication. After 10 min of centrifugation at 14,000 � g, solublefractions equivalent to 10-�l cell suspensions were separated on NuPAGE Novex3 to 8% Tris-acetate gels (Invitrogen), stained by Sypro-red staining (MolecularProbes), and quantified with a Typhoon scanner using bovine serum albumin(BSA) standards.

Polyketide detection. Samples were analyzed by using a system consisting of aLeap Technologies HTC PAL sample handler, an Agilent 1100 high-perfor-mance liquid chromatography (HPLC) pump, and an Applied Biosystems API-3000 triple quadrupole mass spectrometer equipped with a Turbo ion spraysource. For the identification/characterization of triketides, samples (10 �l) werechromatographed on an Agilent Zorbax Eclipse XDB-C8 column (3.5-mm di-ameter, 2.1 by 150 mm) at 250 �l/min by holding a mobile phase of 10%acetonitrile (MeCN) (0.1% acetic acid [HOAc]) in H2O (0.1% HOAc) for 3 min,followed by a linear gradient to MeCN (0.1% HOAc) over 9 min. Additionalconditions were the following: source temperature, 375°C; declustering and fo-cusing potentials, 51 and 180 V, respectively; spray tip potential, 5,000 V; andcollision energy, 15 eV. Triketide 2 was identified by comparing characteristicmass spectra to that of an authentic synthetic standard.

The same LC-MS system was used for the detection of the tetraketides withdifferent HPLC and mass spectrometry conditions. Samples (10 �l) were injectedinto the same column, and its temperature was maintained at 45°C; the mobilephase used was a linear gradient from 10% MeCN (0.1% HOAc) in H2O (0.1%HOAc) to MeCN (0.1% HOAc) over 10 min. Multiple reaction monitoring(MRM) in positive-ion mode was used for detection.

The parent/daughter pairs of m/z 229/211, 229/193, 229/165, and 229/127 eachwere acquired with a dwell time of 200 ms and at unit resolution in the first andthird quadrupoles.

Additional conditions were the following: source temperature, 375°C; declus-tering and focusing potentials, 26 and 200 V, respectively; spray tip potential,4,600 V; and collision energy, 19 eV. The tetraketides were identified by com-paring their characteristic mass spectra to that of an authentic synthetic standard.Concentrations were estimated from the MRM data by comparing the arearesponse of samples to that of the standard at a known concentration.

RESULTS AND DISCUSSION

As an alternative to the complete biosynthesis of epothiloneanalogues, we considered using genetically engineered mi-crobes to prepare polyketide fragments that would serve asadvanced intermediates for chemical synthesis. A key interme-diate in the synthesis of compounds in this series is the C-3–C-9 aldehyde (Fig. 1A), which is made from a chiral materialand requires 12 chemical operations to produce. The diastereo-selectivity of the synthetic steps to create the two new stereo-genic centers is low (�4:1), and the purification of the desireddiastereomers is problematic, making the large-scale manufac-turing of these epothilone analogues extremely costly and,therefore, precluding their clinical development. We sought todevelop a biosynthetic method to produce tetraketide 3 or 4,which can be converted to the C-3–C-9 aldehyde by only fivestraightforward steps (4). Clearly, the economics of the semi-synthetic approach are attractive, but since the key intermedi-ates, tetraketides 3 and 4, have a double-bond moiety at aposition rarely found in polyketides, their biosynthesis was achallenging prospect.

The tetraketide target molecules 3 and 4 (Fig. 1B) wereshown previously to serve as starting material for the completesynthesis of trans-9,10-dehydroepothilone D (4). While thereare two possible stereomers of the �-methyl group of keto-lactone 4, in practice the rapid keto-enol equilibration of thefinal �-keto-ester results in the loss of stereochemical infor-mation at this position. Fortunately, either of the diaste-reomers 4a or 4b is a suitable biosynthetic target that wouldprovide equivalent access to the C-3–C-9 aldehyde (4).

We previously reported an approach to rapidly discoverpairs of heterologous PKS modules that can interact efficientlyto produce a polyketide (8). In this approach, members of alibrary of modules encoded by redesigned genes in a donorplasmid (pAng in Fig. 2) are coexpressed with members of alibrary of modules in an acceptor plasmid (pBru), and triketide

5222 MENZELLA ET AL. APPL. ENVIRON. MICROBIOL.

on January 18, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

FIG. 1. (A) Scheme for the synthesis of trans-9,10-dehydroepothilone D. (B) Production of tetraketide intermediate 3 by the double extensionof SNAC 1 with two D-type modules, 4a by extension with a D-G bimodular PKS and 4b by extension with a D-H bimodular PKS. (C) Structuresof two-carbon units added by the extension modules used in this work. D-type modules: eryM2, eryM5, eryM6, gldM3, sorM6, lepM10, rifM5, andepoM7; G-type modules: eryM3, rapM3, lepM4, and rapM6; H-type module: pikM6. ery, erythromycin; sor, soraphen; rap, rapamycin; gdm,geldanamycin; rif, rifamycin; lep, leptomycin.

5223

on January 18, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

production is analyzed by LC-MS in culture supernatants.More than 150 module-module interactions tested using thisassay yielded a library of �70 novel active bimodular PKSs;subsequently, the library has been expanded to �200 mini-PKSs that provide a library of some 100 active bimodularcombinations (unpublished results).

We decided to explore a systematic, stepwise method towardcreating novel polyketides with unusual starter units using ourpreexisting bimodular library. First, the SNAC containing thestarter unit is fed to an engineered E. coli strain expressingmembers of the Mod-TE (pBru) library that catalyze the firstdesired ketide extension, and the production of the expectedmolecule in the cultures is determined. Second, modules thatbest catalyze this first extension are cloned into donor vectorsand coexpressed with Mod-TEs (represented by further pBruplasmids) in the library that are known to extend the ketideproduct offered by the first module, and cultures are analyzedfor the presence of the final product (Fig. 3). To validate ourapproach, we sought to create designer PKSs to produce in-termediates that would facilitate the chemical synthesis ofepothilone analogues.

We have reported previously the abbreviated codes D, H,and G (8) for the two-carbon ketide units and PKS modulesthat encode them (Fig. 1C). The successful extension of SNAC1 by a bimodular PKS comprising two D-type modules is ex-pected to yield intermediate 3, while D-G and D-H combina-tions are expected to produce 4a and 4b, respectively. Of theeight D-type extension Mod-TEs in our library, only sorM6naturally receives and processes a substrate that resemblesSNAC 1. Nevertheless, LC-MS analysis revealed that six of theeight D-type extension Mod-TEs available in our library pro-duced detectable levels of triketide 2 when expressed in the E.coli polyketide producer strain K207-3 (11) in the presence ofSNAC 1. As shown in Table 1, four of the eight modules extendedSNAC 1 to provide the expected triketide 2 in yields of 1 to 20mg/liter, significantly more than those produced by sorM6, whichnaturally extends a substrate similar to that presented by SNAC 1.In keeping with earlier observations (1), all of the synthetic con-structs gave similar levels of soluble protein in the range of 70 to100 mg/liter. The high levels of protein expression likely are dueto the codon-optimized PKS genes, which typically produce 5- to10-fold more PKS than wild-type sequences.

FIG. 2. Two classes of expression plasmid used to test bimodular interactions in E. coli. pAng vectors contain a CloDF13 replication origin, astreptomycin resistance selection marker, and a PBAD promoter to drive the expression of LM-Module-LCeryM2 ORFs. pBru vectors contain aColE1 replication origin, a carbenicillin resistance selection marker, and a PBAD promoter to drive the expression of LNeryM3-Module-TE ORFs.Productive combinations of modules are revealed by the formation of triketide lactones (TKL).

5224 MENZELLA ET AL. APPL. ENVIRON. MICROBIOL.

on January 18, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

The individual Mod-TEs containing a His6 tag also werepurified and tested for the production of triketide 2 in vitro;they showed a similar ranking of production, as was the case inexperiments with whole bacteria (data not shown).

The four modules that most successfully catalyzed the firstextension of SNAC 1 were reformatted into pAngII donorvectors in which the loading module was replaced by LNeryM5

(Fig. 3). Although the modules initially were evaluated in thecontext of the N-terminal linker of ery3 (LNery3), we haveshown in our previous work that the yields of Mod-TEs pre-ceded by LNery3 or LNery5 are essentially identical (6). In theresultant open reading frames (ORFs) encoding LNery5-Mod-

LCery2, the LM-to-LNery5 alteration precludes the priming ofthe module by intracellular acyl-CoAs and facilitates PKS pro-tein expression (6). As acceptors, we selected from 11 LNery3-Mod-TEs fulfilling two criteria: (i) they were, in theory, capa-ble of adding the desired second extensions (D type) to providethe 3-hydroxy-tetraketide 3 or the equilibrated mixture of keto-lactones 4a and 4b (G or H type); and (ii) each was shownpreviously to accept a substrate from one or more of the fourdonor modules that could catalyze the first extension of SNAC1 (8). Therefore, 6 D-type, 4 G-type, and 1 H-type moduleswere used as acceptors, each of which was shown previously toaccept substrate from one or more of the four selected donormodules. From the 44 possible combinations, 30 were createdand evaluated, 15 to provide tetraketide 3, 12 to give 4a, and 3to yield 4b. Bacteria were grown and PKS expression inducedin the presence of SNAC 1, and culture supernatants wereanalyzed for the expected tetraketide products. As shown inTable 2 and Fig. S1 in the supplemental material, all of thePKSs made products that showed LC retention times andMS-MS spectra identical to those of the synthetic standards.As observed with a chemically synthesized standard, the protonnuclear magnetic resonance (NMR) of purified tetraketide 4 ob-tained from fermentation showed about 15% of the enol form,confirming chemical the equilibration of 4a and 4b, and that thesame product(s) could be obtained from either type of bimodule.

FIG. 3. Method used to produce polyketide intermediates. PKS modules capable of catalyzing the extension of the unnatural starter unit areidentified by feeding the SNAC to bacteria expressing appropriate LNery3-Mod-TEs and analyzing for product. Modules active in the first SNACextension then are reformatted as LNery5-Mod-LCery2 donor modules in pAngII vectors, and these are coexpressed with appropriate LNery3-Mod-TEs from pBru vectors to determine which bimodular combinations can perform two extensions of the SNAC.

TABLE 1. In vivo extension of SNAC 1 by modules expressed asMod-TEs in E. coli strain K207-3

PKS Triketide 2 production(mg/liter)

LNeryM3eryM6-TE........................................................... 20LNeryM3eryM2-TE........................................................... 12LNeryM3eryM5-TE........................................................... 2LNeryM3gldM3-TE........................................................... 1LNeryM3sorM6-TE........................................................... 0.1LNeryM3lepM10-TE......................................................... TraceLNeryM3rifM5-TE ............................................................Not detectedLNeryM3epoM7-TE..........................................................Not detected

VOL. 76, 2010 POLYKETIDE PHARMACEUTICAL INTERMEDIATES IN E. COLI 5225

on January 18, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

The bimodule eryM6 � eryM5-TE made the largest amountof tetraketide 3. For tetraketide 4, similar yields were obtainedfrom the combinations eryM2 � eryM3-TE and eryM2 �pikM6-TE. Such results could not have been predicted. Forexample, one natural sequence of two modules (eryM2 �eryM3), presumably adapted by evolution, produced the highestlevel of anticipated product, whereas another (eryM5 � eryM6)did not. The latter observation is in agreement with earlier resultsshowing that eryM5 is considerably less efficient at processingSNACs than eryM6 (Table 1) (22). In all cases, the expectedcompound was the major product, and no other tetraketide prod-uct was detected. However, in most of the cases, traces of triketide2 were found. This compound presumably originated from theearly release of the polyketide chain after the extension of theSNAC by the first module. We conclude that the guided combi-natorial approach presented here is currently the most effectiveroute to success.

Although our strategy was successful in producing either thehydroxy- or keto-tetraketide, 3 or 4, respectively, the latter hasthe clear advantage of requiring one fewer chemical conver-sions to provide the C-3–C-9 aldehyde. Thus, future effortsdirected toward optimizing the biosynthetic production of tet-raketide 4 by further protein engineering and fermentationprocess development should yield titer improvements to pro-duce this starting material in a cost-effective manner.

In summary, a procedure has been established to utilize apreexisting library of bimodular PKSs to create polyketideswith unnatural starter units. First, PKS modules capable of

catalyzing the extension of the unnatural starter unit are iden-tified by feeding the SNAC to bacteria expressing appropriateMod-TEs and analyzing for product. Next, modules active inthe first SNAC extension are reformatted as donor modules,and these are coexpressed with appropriate Mod-TEs to de-termine which bimodular combinations can perform two ex-tensions of the SNAC. The process can, in principle, be con-tinued to provide polyketides of increasing size. The feasibilityof this approach has been validated by developing bimodularsystems producing tetraketide lactones with multiple chiralcenters and an unusual starter, which serve as valuable inter-mediates in the chemical synthesis of 9,10-dehydro-12,13-des-oxyepothilone and its analogues. We anticipate that the use ofthe semisynthetic C-3–C-9 aldehyde intermediate describedhere will eliminate seven steps in the synthesis, including thoseinvolved in the creation of stereogenic centers and the prob-lematic purification operations associated with them.

ACKNOWLEDGMENTS

We thank Janice Lau for helping with fermentation and the isolation ofpolyketides, Gary Ashley and Yue Chen for assisting with the preparationof SNACs, and David Hopwood for the critical review of the manuscript.

REFERENCES

1. Chandran, S. S., H. G. Menzella, J. R. Carney, and D. V. Santi. 2006.Activating hybrid modular interfaces in synthetic polyketide synthases bycassette replacement of ketosynthase domains. Chem. Biol. 13:469–474.

2. Chotani, G., T. Dodge, A. Hsu, M. Kumar, R. LaDuca, D. Trimbur, W.Weyler, and K. Sanford. 2000. The commercial production of chemicalsusing pathway engineering. Biochim. Biophys. Acta 1543:434–455.

3. Jacobsen, J. R., C. R. Hutchinson, D. E. Cane, and C. Khosla. 1997. Pre-cursor-directed biosynthesis of erythromycin analogues by an engineeredpolyketide synthase. Science 277:367–369.

4. Li, Y., and Y. Chen. 2008. Process for the preparation of epothilones. PatentWO 2008141130.

5. McDaniel, R., M. Welch, and C. R. Hutchinson. 2005. Genetic approaches topolyketide antibiotics. 1. Chem. Rev. 105:543–558.

6. Menzella, H. G., J. R. Carney, and D. V. Santi. 2007. Rational design andassembly of synthetic trimodular polyketide synthases. Chem. Biol. 14:143–151.

7. Menzella, H. G., and C. D. Reeves. 2007. Combinatorial biosynthesis for drugdevelopment. Curr. Opin. Microbiol. 10:238–245.

8. Menzella, H. G., R. Reid, J. R. Carney, S. S. Chandran, S. J. Reisinger, K. G.Patel, D. A. Hopwood, and D. V. Santi. 2005. Combinatorial polyketidebiosynthesis by de novo design and rearrangement of modular polyketidesynthase genes. Nat. Biotechnol. 23:1171–1176.

9. Menzella, H. G., T. T. Tran, J. R. Carney, J. Lau-Wee, J. Galazzo, C. D.Reeves, C. Carreras, S. Mukadam, S. Eng., Z. Zhong, P. B. Timmermans, S.Murli, and G. W. Ashley. 2009. Potent non-benzoquinone ansamycin heatshock protein 90 inhibitors from genetic engineering of Streptomyces hygro-scopicus. J. Med. Chem. 52:1518–1521.

10. Michaud, L. B. 2009. The epothilones: how pharmacology relates to clinicalutility. Ann. Pharmacother. 43:1294–1309.

11. Murli, S., J. Kennedy, L. C. Dayem, J. R. Carney, and J. T. Kealey. 2003.Metabolic engineering of Escherichia coli for improved 6-deoxyerythronolideB production. J. Ind. Microbiol. Biotechnol. 30:500–509.

12. Murli, S., K. S. MacMillan, Z. Hu, G. W. Ashley, S. D. Dong, J. T. Kealey,C. D. Reeves, and J. Kennedy. 2005. Chemobiosynthesis of novel 6-deoxy-erythronolide B analogues by mutation of the loading module of 6-deoxy-erythronolide B synthase 1. Appl. Environ. Microbiol. 71:4503–4509.

13. Perez, E. A. 2009. Microtubule inhibitors: differentiating tubulin-inhibitingagents based on mechanisms of action, clinical activity, and resistance. Mol.Cancer Ther. 8:2086–2095.

14. Pollard, D. J., and J. M. Woodley. 2007. Biocatalysis for pharmaceuticalintermediates: the future is now. Trends Biotechnol. 25:66–73.

15. Reeves, C. D. 2003. The enzymology of combinatorial biosynthesis. Crit. Rev.Biotechnol. 23:95–147.

16. Rivkin, A., F. Yoshimura, A. E. Gabarda, Y. S. Cho, T. C. Chou, H. Dong,and S. J. Danishefsky. 2004. Discovery of (E)-9,10-dehydroepothilonesthrough chemical synthesis: on the emergence of 26-trifluoro-(E)-9,10-dehy-dro-12,13-desoxyepothilone B as a promising anticancer drug candidate.J. Am. Chem. Soc. 126:10913–10922.

17. Tao, J., and J. H. Xu. 2009. Biocatalysis in development of green pharma-ceutical processes. Curr. Opin. Chem. Biol. 13:43–50.

TABLE 2. In vivo production of tetraketide 3 or 4a, 4b, and 4c inE. coli strain K207-3 by extension of SNAC 1 by bimodular PKSs

PKSProduction (mg/liter) of tetraketide:

3 4a, 4b, and 4c

eryM6 � eryM5-TE 5.3eryM6 � eryM2-TE 4.9eryM6 � eryM6-TE 3.6eryM6 � sorM6-TE 3.5eryM6 � gldM3-TE 2.5eryM2 � eryM2-TE 2.2eryM2 � eryM5-TE 2.8eryM2 � eryM6-TE 2.0eryM2 � sorM6-TE 2.1eryM5 � eryM6-TE 1.8eryM5 � eryM2-TE 1.2eryM5 � eryM5-TE 0.8eryM5 � rifM5-TE 1.4gldM3 � eryM6-TE 0.8gldM3 � eryM2-TE 0.7eryM2 � eryM3-TE 93.2eryM2 � pikM6-TE 81.6eryM2 � rapM3-TE 51.6eryM2 � lepM4-TE 0.7eryM2 � rapM6-TE 74eryM6 � eryM3-TE 7.4eryM6 � pikM6-TE 34.4eryM6 � rapM3-TE 0.6eryM6 � lepM4-TE 0.02eryM6 � rapM6-TE 0.2eryM5 � eryM3-TE 0.1eryM5 � pikM6-TE 0.7eryM5 � rapM3-TE 0.5eryM5 � lepM4-TE 0.7eryM5 � rapM6-TE 0.5

5226 MENZELLA ET AL. APPL. ENVIRON. MICROBIOL.

on January 18, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

18. Walsh, C. 2001. Enabling the chemistry of life. Nature 409:226–231.19. Wang, J., H. Zhang, L. Ying, C. Wang, N. Jiang, Y. Zhou, H. Wang, and H.

Bai. 2009. Five new epothilone metabolites from Sorangium cellulosum strainSo0157-2. J. Antibiot. (Tokyo) 62:483–487.

20. Weissman, K. J. 2007. Mutasynthesis–uniting chemistry and genetics fordrug discovery. Trends Biotechnol. 25:139–142.

21. Weissman, K. J., and P. F. Leadlay. 2005. Combinatorial biosynthesis ofreduced polyketides. Nat. Rev. Microbiol. 3:925–936.

22. Wu, N., S. Y. Tsuji, D. E. Cane, and C. Khosla. 2001. Assessing the balancebetween protein-protein interactions and enzyme-substrate interactions inthe channeling of intermediates between polyketide synthase modules.J. Am. Chem. Soc. 123:6465–6474.

VOL. 76, 2010 POLYKETIDE PHARMACEUTICAL INTERMEDIATES IN E. COLI 5227

on January 18, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from