JOURNAL OF BACTERIOLOGY,0021-9193/97/$04.0010

Oct. 1997, p. 6416–6425 Vol. 179, No. 20

Copyright © 1997, American Society for Microbiology

Acyltransferase Domain Substitutions in ErythromycinPolyketide Synthase Yield Novel Erythromycin Derivatives

XIAOAN RUAN, ANA PEREDA, DIANE L. STASSI, DAVID ZEIDNER, RICHARD G. SUMMERS,MARIANNA JACKSON, ANNAPUR SHIVAKUMAR, STEPHAN KAKAVAS, MICHAEL J. STAVER,

STEFANO DONADIO,† AND LEONARD KATZ*

Antibacterial Discovery Research, Abbott Laboratories, Abbott Park, Illinois 60064

Received 17 June 1997/Accepted 29 July 1997

The methylmalonyl coenzyme A (methylmalonyl-CoA)-specific acyltransferase (AT) domains of modules 1and 2 of the 6-deoxyerythronolide B synthase (DEBS1) of Saccharopolyspora erythraea ER720 were replacedwith three heterologous AT domains that are believed, based on sequence comparisons, to be specific formalonyl-CoA. The three substituted AT domains were “Hyg” AT2 from module 2 of a type I polyketide synthase(PKS)-like gene cluster isolated from the rapamycin producer Streptomyces hygroscopicus ATCC 29253, “Ven”AT isolated from a PKS-like gene cluster of the pikromycin producer Streptomyces venezuelae ATCC 15439, andRAPS AT14 from module 14 of the rapamycin PKS gene cluster of S. hygroscopicus ATCC 29253. These changesled to the production of novel erythromycin derivatives by the engineered strains of S. erythraea ER720.Specifically, 12-desmethyl-12-deoxyerythromycin A, which lacks the methyl group at C-12 of the macrolactonering, was produced by the strains in which the resident AT1 domain was replaced, and 10-desmethylerythro-mycin A and 10-desmethyl-12-deoxyerythromycin A, both of which lack the methyl group at C-10 of themacrolactone ring, were produced by the recombinant strains in which the resident AT2 domain was replaced.All of the novel erythromycin derivatives exhibited antibiotic activity against Staphylococcus aureus. Theproduction of the erythromycin derivatives through AT replacements confirms the computer predicted sub-strate specificities of “Hyg” AT2 and “Ven” AT and the substrate specificity of RAPS AT14 deduced from thestructure of rapamycin. Moreover, these experiments demonstrate that at least some AT domains of thecomplete 6-deoxyerythronolide B synthase of S. erythraea can be replaced by functionally related domains fromdifferent organisms to make novel, bioactive compounds.

Erythromycin A is a clinically useful, broad-spectrum mac-rolide antibiotic produced by the gram-positive bacterium Sac-charopolyspora erythraea. Erythromycin belongs to a class ofnatural products called complex polyketides which includesmany important compounds, such as the antihelminthic agentavermectin and the immunosuppressants FK506 and rapamy-cin. The central feature of these compounds is a polyketide-derived macrocyclic lactone ring that is synthesized by an or-dered condensation of acyl thioesters. These rings are typicallybuilt from carbon chains of specific lengths, and they exhibitdifferent degrees of reduction and substitution around the ring(8, 29, 32). The production of polyketides has long been rec-ognized to resemble fatty acid biosynthesis (25, 43), but withimportant differences (15). Unlike a fatty acid synthase, whichusually starts fatty acid synthesis with acetyl coenzyme A(acetyl-CoA) and extends the chain by two carbons withoutside chains, using malonyl-CoA as the extender unit, a complexpolyketide synthase (PKS) is programmed to make distinctchoices at each step of carbon chain assembly. These includechoice of the starter from a number of possibilities, such asacetyl-CoA (oleandomycin), propionyl-CoA (erythromycin),and isobutyryl-CoA (avermectin), choice of either unbranchedor branched chain extenders (e.g., malonyl-CoA [rapamycin],methylmalonyl-CoA [erythromycin], and ethylmalonyl CoA[tylosin]), and determination of the degree of reduction of the

b-keto group that is generated after each condensation (rang-ing from no reduction through ketoreduction only, ketoreduc-tion and dehydration, or ketoreduction, dehydration, andenoyl reduction).

Polyketide programming is built into the structure of thePKS, which is determined at the genetic level. The genes (andenzymatic domains) of a complex PKS are organized in mod-ules (8, 29, 32, 37). Each module encodes all of the functionsnecessary for a single, specific condensation event and theattendant processing of the newly formed b-keto group. Eachmodule carries a ketosynthase (KS), an acyltransferase (AT),and an acyl carrier protein (ACP) domain. In addition, b-ketoreductase (KR), dehydratase, and enoyl reductase (ER)domains may also be present. The KS, AT, and ACP domainsare responsible for chain extension. The AT domain is believedto determine which extender is incorporated at each step ofpolyketide chain growth (10, 12, 18), while the KS domain isresponsible for catalyzing the actual condensation reaction.The ACP domains tether the growing polyketide chain to thePKS between condensations and also accept extender unitsfrom the AT domains in preparation for a condensation.

The macrolactone ring of erythromycin is synthesized by theerythromycin PKS, 6-deoxyerythronolide B synthase (DEBS)(Fig. 1). DEBS contains (i) six modules to program the sixsuccessive condensations between the starter propionyl-CoAand (ii) the six methylmalonyl-CoA extender units to build the14-membered, 6-deoxyerythronolide B (6-dEB) ring (Fig. 1).The precise order of the elongation steps is colinear with thegenetic order of the six modules: module 1 determines the firstcondensation, module 2 determines the second, module 3 de-termines the third, and so on until the sixth condensation stephas occurred (8). In the case of 6-dEB, the extender unit for

* Corresponding author. Mailing address: Abbott Laboratories, D-47P AP9A, 100 Abbott Park Road, Abbott Park, IL 60064. Phone:(847) 937-4132. Fax: (847) 938-3403. E-mail: leonard.katz@abbott.com.

† Present address: Biosearch Italia s.p.a., 21040 Gerenzano (VA),Italy.

6416

on January 7, 2021 by guesthttp://jb.asm

.org/D

ownloaded from

each condensation is methylmalonyl-CoA, which gives rise tothe characteristic pattern of alternating methyl groups aroundthe 6-dEB ring. Although only the (2S)-stereoisomer of meth-ylmalonyl-CoA is employed in the synthesis of erythromycin(23), the opposite (2R) stereochemistry is found in three of thesix methyl side chains. Whether the epimerizations take placeduring or following the condensation reactions remains to bedetermined.

Novel derivatives of erythromycin have been most fre-quently generated by chemical modifications to the antibiotic.However, in recent years derivatives have been made by ge-netically altered erythromycin-producing strains. 2-Noreryth-romycin A and its congeners 2-norerythromycin B, C, and Dwere the first examples of use of recombinant DNA techniquesto produce novel structures of erythromycin (24). Character-ization of the PKS genes responsible for the biosynthesis oferythromycin in S. erythraea led to the generation of derivativesthrough directed manipulation of the eryA genes. The novelcompounds 5,6-dideoxy-3a-mycarosyl-5-oxoerythronolide Band D6,7-anhydroerythromycin C were produced through tar-geted mutation of the KR domain in module 5 and the ERdomain in module 4, respectively (9, 10). The production ofthese novel compounds illustrated that recombinant DNA

technology can be effectively used to generate erythromycinderivatives that are difficult or impossible to make chemically.

With the recent evidence that AT domains determine theextender used for each condensation step during polyketidechain growth (26), a series of novel erythromycins carrying sidechain alterations were generated by genetically engineeringDEBS with heterologous malonyl-CoA-incorporating AT do-mains. Here, we report that biosynthesis of the carbon back-bone of the erythromycin macrolactone ring can be repro-grammed through AT replacements in DEBS1, giving rise tonew erythromycin derivatives that lack methyl groups at eitherC-10 or C-12 of the macrolactone ring.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth media. Actinomycete strains andplasmids used in this study are listed in Table 1. Plasmid-containing Escherichiacoli strains were grown in Luria broth (30) supplemented with ampicillin (150mg/ml). For growth of S. erythraea strains in liquid cultures, either SGGP (44) orSCM (20 g of soytone, 15 g of soluble starch, 10.5 g of morpholinepropanesul-fonic acid [MOPS], 1.5 g of yeast extract, and 0.1 g of CaCl2 per liter of distilledH2O) medium was used. For the growth of S. erythraea strains on plates, R3Mmedium was used. One liter of R3M medium contains 4 g of yeast extract, 4 g ofCasamino Acids, 4 g of Bacto Tryptone, 0.25 g of K2SO4, 0.6 g of Tris base, 3.2 gof Tris-HCl, 2.5 ml of 1 M NaOH, 103 g of sucrose, 22 g of agar, and 940 ml of

FIG. 1. Biosynthetic pathway of erythromycin A. The erythromycin biosynthetic genes are clustered in a ca. 65-kb chromosomal segment of S. erythraea. The DEBSgenes are designated eryA (10), and a few of the flanking genes, eryF, eryK, and eryG, involved in the postpolyketide modifications, are shown. The eryB and eryC genesinvolved in the synthesis of the dideoxysugars L-mycarose and D-desosamine also are located in the cluster but are not shown. The hydroxyl groups introduced by EryFor EryK are highlighted.

VOL. 179, 1997 ERYTHROMYCIN DERIVATIVES VIA AT SUBSTITUTIONS 6417

on January 7, 2021 by guesthttp://jb.asm

.org/D

ownloaded from

H2O. After autoclaving at 121°C for 20 min, the following solutions were added:5,0003-concentrated trace elements solution, 0.2 ml (14); 50% glucose solution,20 ml; 2.5 M MgCl2 solution, 20 ml; 0.5% KH2PO4 solution, 5 ml; and 2.5 MCaCl2 solution, 20 ml. For selection of thiostrepton-resistant (Thior) S. erythraeastrains, 25 mg of thiostrepton/ml was used for growth on plates and 10 mg/ml wasused in liquid culture.

DNA manipulations. Standard molecular biology techniques were performedas described previously (30). For DNA sequencing, denatured plasmid DNA orsingle-stranded DNAs prepared from subclones of pUC18/19 or M13mp18/19(Bethesda Research Laboratories) were used as the templates. Nucleotide se-quences were determined by the dideoxynucleoside chain termination method(31), using either the fmol DNA cycle sequencing system (Promega Corp.,Madison, Wis.) or Sequenase version 2.0 with 7-deaza-dGTP (U.S. Biochemical,Cleveland, Ohio) according to the manufacturer’s instructions. The sequenceswere analyzed by using the Genetics Computer Group (Madison, Wis.) sequenceanalysis programs (6). Database searches for sequence homology to the deducedamino acid sequences were performed by using BLAST (1). Amino acid se-quence alignments were performed by using PILEUP and DENDROGRAM.

Cloning the DNA encoding “Ven” AT. A genomic library of Streptomycesvenezuelae ATCC 15439 was constructed in plasmid pNJ1 (38) and screened byusing a probe made from a 1.37-kb SmaI fragment containing part of the KSdomain of module 2 of eryAI (10). Cosmid pVen17 (Table 1) was isolated fromthe genomic library and characterized by Southern analysis. Three SstI fragmentswere found to hybridize to the probe. One of the fragments, 4.0 kb in size, wasthen subcloned into pUC19 to give pVen4.0. Sequence analysis revealed that theinsert in pVen4.0 contained a partial AT domain, and consequently the nucle-otide sequence was extended by using pVen17 DNA as the template to obtain afull-length sequence of the AT domain (“Ven” AT).

PCR primers and PCR subclonings. Seven pairs of oligonucleotide primerswere designed for PCR subcloning of the three heterologous AT domains andthe four regions of DNA that flank the DEBS AT1 and DEBS AT2 domains usedin this study (Table 2). Unique restriction sites introduced for convenient sub-cloning are underlined in each oligonucleotide. The template DNAs used forPCR cloning of the heterologous ATs were pAL58 for “Hyg” AT2 (28), pVen17for “Ven” AT (Table 1), and chromosomal DNA isolated from Streptomyceshygroscopicus ATCC 29253 for RAPS AT14. The template DNA used to sub-clone the regions flanking the DEBS AT1 and DEBS AT2 domains waspAIEN22, a pUC19 derivative containing a 22-kb DNA segment of the ery PKSgene cluster of S. erythraea.

Two PCR conditions were used with similar results to subclone the charac-teristically high-G1C-content actinomycete DNA fragments. (i) The 100-ml re-action mixture contained 10 ml of 103 PCR buffer (Bethesda Research Labo-ratories), 2 ml of 10 mM deoxynucleoside triphosphate mixture, 2 to 4 ml of 50mM MgCl2, 100 pmol of each oligonucleotide, and 10 to 50 ng of template DNA.Cycling was as follows: one cycle at 96°C for 6 min and 80°C for 1 min (Taq DNApolymerase added during this period); 30 cycles at 95°C for 1 min, 65°C for 1 min,and 72°C for 2 min, with a 5-min extension at 72°C for the last cycle. (ii) The

100-ml reaction mixture contained 10 ml of 103 Thermopol buffer (New EnglandBiolabs), 2% glycerol, 10% formamide, 100 pmol of each oligonucleotide, and100 to 200 ng of template DNA. The sample was heated to 99°C for 2 min andthen cooled to 80°C, at which time 16 ml of a deoxynucleoside triphosphatesolution (1.25 mM each dATP and dTTP, 1.5 mM each dCTP and dGTP) and 1ml of VentR DNA polymerase (New England Biolabs) were added. Cycling wasas follows: 30 cycles at 96.5°C for 35 s, 65°C for 1 min, and 72°C for 1.5 min,followed by one cycle at 72°C for 3 min.

The PCR-amplified DNA fragments carrying the heterologous ATs were in-dividually subcloned into the EcoRI/BamHI sites of pUC18 to give pUC18/“Hyg” AT2 and pUC18/“Ven” AT or into the EcoRI/HindIII sites of pUC19 togive pUC19/RapAT14. For the PCR-amplified flanking regions of DEBS AT1,the upstream fragment [corresponding to the segment nucleotides (nt) 2781 to3825; GenBank accession no. M63676] was subcloned into the EcoRI/BamHIsites of pUC19 to give pUC19/EryAT1/upstream; the downstream fragment(corresponding to the segment nt 4866 to 5912) was subcloned into theBamHI/HindIII sites of pUC19 to give pUC19/EryAT1/downstream (see Fig.3B). For the PCR-amplified flanking regions of DEBS AT2, the upstream frag-ment (corresponding to the segment nt 7090 to 8255) was subcloned into theHindIII/PstI sites of pUC19 to give pUC19/EryAT2/upstream; the downstreamfragment (corresponding to the segment nt 9282 to 10368) was subcloned intothe PstI/EcoRI sites of pUC19 to give pUC19/EryAT2/downstream (see Fig. 3B).The fidelity of these PCR-amplified fragments was confirmed by sequencing bothDNA strands of the subclones.

Genetic manipulation of S. erythraea ER720. Preparation of protoplasts of S.erythraea ER720, transformation of the protoplasts with integrative pWHM3derivatives (Table 1), and resolution of the integrants were carried out accordingto previously described procedures (41, 44). Protoplasts were independentlytransformed with the five pWHM3-derived AT replacement plasmids. To selectstable transformants (integrants) via homologous recombination, colonies arisingon the transformation plates were restreaked onto R3M plates containing thio-strepton (25 mg/ml). The resistant colonies were then inoculated into SGGPcontaining thiostrepton (10 mg/ml) to ensure Thior and to isolate chromosomalDNA for confirmation of the integration event by Southern analysis. The con-firmed integrants were next grown in SGGP without thiostrepton for 4 days andthen plated onto nonselective R3M plates for sporulation. Spore suspensionswere plated on R3M plates to obtain individual colonies, which were thenscreened for sensitivity to thiostrepton, indicating the loss of the plasmid se-quence from the chromosome through a second homologous recombinant event.Southern analysis of the thiostrepton-sensitive colonies confirmed the desiredgenotypes of the resolvants.

Characterization of erythromycin derivatives produced by recombinant S.erythraea strains. For thin-layer chromatography (TLC) analysis, cells weregrown in shake flasks in either SGGP or SCM for 4 to 5 days at 30°C and werethen removed by centrifugation. The resulting supernatant was adjusted to pH9.0 by the addition of NH4OH and extracted twice with an equal volume of ethylacetate. The organic phase, which contained the desired macrolides, was dried in

TABLE 1. Actinomycete strains and plasmids used

Strain or plasmid Description Reference or source

Actinomycete strainsS. erythraea ER720 Wild type, erythromycin producer 7S. venezuelae ATCC 15439 Wild type, pikromycin producer ATCCa

S. hygroscopicus ATCC 29253 Wild type, rapamycin producer ATCCS. erythraea

EryAT1/“Hyg” AT2 S. erythraea ER720 with DEBS AT1 replaced by “Hyg” AT2 This studyEryAT2/“Hyg” AT2 S. erythraea ER720 with DEBS AT2 replaced by “Hyg” AT2 This studyEryAT1/“Ven” AT S. erythraea ER720 with DEBS AT1 replaced by “Ven” AT This studyEryAT1/RapAT14 S. erythraea ER720 with DEBS AT1 replaced by RAPS AT14 This studyEryAT2/RapAT14 S. erythraea ER720 with DEBS AT2 replaced by RAPS AT14 This study

PlasmidspNJ1 Actinomycete-E. coli cosmid containing rep-pJVI 38pAL58 pNJ1 cosmid containing part of a “Hyg” PKS gene cluster isolated from

S. hygroscopicus ATCC 2925328

pVen17 pNJ1 cosmid containing a segment of “Ven” PKS genes of S. venezuelae This studypWHM3 E. coli-Streptomyces shuttle vector containing rep-pIJ702; replicates

poorly in S. erythraea39

pWHM3/EryAT1-flank pWHM3 containing EryAT1 flanking regions This studypWHM3/EryAT2-flank pWHM3 containing EryAT2 flanking regions This studypWHM3/EryAT1/“Hyg” AT2 pWHM3/EryAT1-flank containing “Hyg” AT2 This studypWHM3/EryAT2/“Hyg” AT2 pWHM3/EryAT2-flank containing “Hyg” AT2 This studypWHM3/EryAT1/“Ven” AT pWHM3/EryAT1-flank containing “Ven” AT This studypWHM3/EryAT1/RapAT14 pWHM3/EryAT1-flank containing RAPS AT14 This studypWHM3/EryAT2/RapAT14 pWHM3/EryAT2-flank containing RAPS AT14 This study

a ATCC, American Type Culture Collection, Rockville, Md.

6418 RUAN ET AL. J. BACTERIOL.

on January 7, 2021 by guesthttp://jb.asm

.org/D

ownloaded from

a SpeedVac and redissolved in ethyl acetate to give an appropriate concentrationfor spotting onto a TLC plate (Merck 60F-254 Silica Gel). Migration was allowedfor approximately 1.5 h, using isopropyl ether-methanol-NH4OH (75:35:2), andthe erythromycin derivatives were visualized with an anisaldehyde-sulfuric acid-ethanol (1:1:9) spray.

For TLC-bioautography, the TLC plate was developed as described above, airdried, and placed in a sterile bioassay dish (245 by 245 by 25 mm). The plate wasthen covered with 150 ml of antibiotic medium 11 (Difco) containing 0.1 ml of anovernight culture of Staphylococcus aureus as an indicator strain and incubatedovernight at 37°C to visualize zones of inhibition.

For mass spectrometry (MS), ethyl acetate extracts were separated by eitherhigh-pressure liquid chromatography (HPLC) or TLC. After TLC separation,the region of interest was scraped from the plate and extracted once with ethylacetate-methanol (1:1), followed by two extractions with ethyl acetate. Thecombined organic phases were then dried in a SpeedVac and submitted for MSanalysis. For HPLC separation, the sample was run through a Prodigy ODS (2)column (5 mm, 50 by 2 mm) on a Hewlett-Packard 1050 liquid chromatograph,using a gradient elution of 5 mM ammonium acetate and methanol at the flowrate of 0.3 ml/min. MS analysis was routinely performed with a Finnigan-MAT7000 mass spectrometer equipped with an atmospheric pressure chemical ion-ization source. Electrospray MS was performed with a Finnigan-MAT 752-7000mass spectrometer equipped with a Finnigan atmospheric pressure ionizationsource.

To isolate 10- and 12-desmethylerythromycins for nuclear magnetic resonance(NMR) structural studies, 30 liters of SCM was prepared in 42-liter LH Fer-mentation Series 2000 stainless steel stirred vessels and sterilized at 121°C and 15lb/m2 for 1 h. Antifoam (XFO-371; Ivanhoe Chemical Co., Mundelein, Ill.) wasadded initially at 0.01% and then was available on demand. The fermentors wereinoculated with 1.5 liters of second-passage seed growth of either the EryAT2/“Hyg” AT2 or EryAT1/“Hyg” AT2 strain of S. erythraea. The temperature wascontrolled at 32°C. The agitation rate was 260 rpm, and the air flow was 1.3vol/vol/min. The head pressure was maintained at 6 lb/m2, and the pH was

controlled at 7.3 with 5 M propionic acid. The fermentations were terminatedbetween 100 and 120 h. The resulting fermentation broths were then adjusted topH 9 with NH4OH and extracted twice with equal volumes of CH2Cl2. Theextracts were pooled and concentrated under vacuum to an oil.

For the 10-desmethylerythromycin sample, the concentrated crude materialwas extracted twice with an equal volume of 0.05 M aqueous potassium phos-phate (pH 4.25). The pooled aqueous phase was then adjusted to pH 9 withNH4OH and reextracted twice with an equal volume of ethyl acetate. Afterconcentration of the organic phase, the aqueous back extraction-organic reex-traction sequence was repeated. The oily residue remaining after concentrationof the final pooled ethyl acetate extracts was then digested in 5 ml each of theupper and lower phases of a solvent system consisting of n-hexane, ethyl acetate,and 0.02 M aqueous potassium phosphate (pH 6.6), 6:4:5 (vol/vol/vol), and waschromatographed on a custom droplet countercurrent chromatography instru-ment (100 vertical columns; 0.4-cm diameter by 24-cm length) (16), using thelower phase as the mobile phase. Fractions were analyzed according to bioac-tivity against S. aureus and by 1H NMR. Bioactive fractions were pooled andsubjected to a series of five countercurrent chromatographic separations in an Itomultilayer Planet Coil centrifuge. The solvent systems used at each step were asfollows: (i) trichloroethylene–chloroform–methanol–0.02 M aqueous potassiumphosphate (pH 6.4), 3:1:3:3 (vol/vol/vol/vol), with the lower phase mobile; (ii)n-hexane–ethyl acetate–0.02 M aqueous potassium phosphate (pH 7), 1:1:1 (vol/vol/vol), with the upper phase mobile; (iii) n-hexane–ethyl acetate–0.02 M aque-ous potassium phosphate (pH 8.1), 1:1:1 (vol/vol/vol), with the upper phasemobile; (iv) n-butanol–acetone–0.02 M aqueous potassium phosphate (pH 6.5),8:1:10 (vol/vol/vol), with the upper phase mobile; and (v) n-hexane–ethyl ace-tate–0.02 M aqueous potassium phosphate (pH 7), 1:1:1 (vol/vol/vol), with theupper phase mobile.

For the 12-desmethylerythromycin sample, the concentrated crude materialwas extracted twice with an equal volume of 0.05 M aqueous potassium phos-phate (pH 6). The pooled aqueous phase was then adjusted to pH 8 with NH4OHand reextracted twice as described above. This process was repeated, and the oily

TABLE 2. PCR primers designed for AT replacement constructions

VOL. 179, 1997 ERYTHROMYCIN DERIVATIVES VIA AT SUBSTITUTIONS 6419

on January 7, 2021 by guesthttp://jb.asm

.org/D

ownloaded from

residue that remained after concentration of the pooled ethyl acetate extractswas then digested in 2.5 ml each of the upper and lower phases of a solventsystem consisting of n-hexane, ethyl acetate, and 0.02 M aqueous potassiumphosphate (pH 8), 1:1:1 (vol/vol/vol), and subjected to countercurrent chroma-tography in an Ito multilayer Planet Coil centrifuge, with the upper phasemobile. A second separation was then performed with a solvent system compris-ing n-hexane, ethyl acetate, and 0.02 M aqueous potassium phosphate (pH 6.4),6:4:5 (vol/vol/vol), with the upper phase mobile.

Complete assignment of the 1H and 13C NMR spectra of both 10-desmethyl-erythromycin A 6,9:9,12-spiroketal (10-desmethylanhydroerythromycin A) and12-desmethyl-12-deoxyerythromycin A was possible with the aid of correlatedspectroscopy, heteronuclear multiple quantum correlation, heteronuclear multi-ple bond correlation, and distortionless enhancement by polarization transferexperiments.

Nucleotide sequence accession number. The sequence of the “Ven” AT do-main reported in this study has GenBank accession no. AF015823.

RESULTS

AT domains used to replace DEBS AT1 and DEBS AT2.Three heterologous AT domains, “Hyg” AT2, “Ven” AT, andRAPS AT14, each thought to specify the use of malonyl-CoAfor polyketide chain extension, in contrast to methylmalonyl-CoA, which is used for condensations in the biosynthesis of6-dEB, were chosen to replace the AT domains of modules 1(DEBS AT1) and 2 (DEBS AT2) of DEBS1 (10). The “Hyg”AT2 domain was obtained from module 2 of the “Hyg” PKSgene cluster of the rapamycin producer S. hygroscopicus ATCC29253 (28). RAPS AT14 was obtained from module 14 of therapamycin PKS gene cluster (32) of S. hygroscopicus ATCC29253 and was proposed previously to encode a malonyl-CoA-specific AT domain based on correlation of the structure ofrapamycin with the organization of its PKS (32). The “Ven”AT domain was obtained from an unknown PKS-like genecluster in the pikromycin producer S. venezuelae ATCC 15439as described in Materials and Methods. Sequence comparisonsusing the deduced amino acid sequence of the “Ven” ATdomain indicated that it is most closely related to “Hyg” AT2and to the rapamycin AT domains that are thought to specifymalonyl-CoA as extenders (Fig. 2): 42% identical to “Hyg”AT2 and 40 to 44% identical to the RAPS ATs that are pre-

dicted to use malonyl-CoA as an extender unit (32). Thus, both“Hyg” AT2 and “Ven” AT also seemed likely to be specific formalonyl-CoA, based on their similarity to RAPS AT14 andother RAPS AT domains that were thought to specify the useof malonyl-CoA (Fig. 2).

Construction of the AT replacement plasmids. A cassettestrategy was developed to facilitate construction of severalplasmids, each harboring a different heterologous AT, to re-place DEBS AT1 or DEBS AT2. Two unique restriction sites,AvrII and NsiI, were introduced into the cassette by using PCRprimers (Table 2 and Fig. 3). The boundaries of the erythro-mycin and heterologous AT domains were defined by aminoacid sequence alignment of all known complex PKS AT do-mains, and these boundaries encompass the entire AT domainregions that were previously considered to be important for ATfunction (8). The boundaries were also chosen to tolerate theintroduction of new restriction sites with a minimum of dis-ruption to the original amino acid sequence of each AT do-main. Within these constraints, the entire DEBS AT1 andDEBS AT2 domains could be replaced when AvrII and NsiIsites were introduced at the N- and C-terminal boundaries,respectively. Likewise, the heterologous AT domains were alsoflanked by these restriction sites. As shown in Fig. 3A tworesidues near the N terminus (boxed) were changed to Pro (P)and Arg (R) by introducing the AvrII site. Notably, for the tworesidues changed at the N terminus, the P is conserved insimilar positions in all of the RAPS AT domains (alignmentnot shown), and the R is already present in three of the fiveATs shown in Fig. 3A. At the C terminus, P (boxed) is changedto Ala (A) upon introduction of the NsiI site (Fig. 3A). Alanineis naturally found at this position in the original sequence ofthe “Ven” AT domain.

Cassettes carrying DNA flanking the resident DEBS AT1(EryAT1) and DEBS AT2 (EryAT2) domains were con-structed in several steps. For construction of the EryAT1 flank-ing cassette (Fig. 3B), the upstream flank was isolated frompUC19/EryAT1/upstream (Materials and Methods) and in-serted into the EcoRI/BamHI sites of pUC19/EryAT1/down-stream to generate construct pUC19/EryAT1-flanks (map notshown). The EcoRI/HindIII fragment that contained bothflanking regions in their appropriate orientations was thenisolated from pUC19/EryAT1-flanks and subcloned into thesame sites of pWHM3 to generate the integrative cassettepWHM3/EryAT1-flank used for module 1 replacement (Fig.3B). Similar steps were taken to prepare a second integrativecassette, pWHM3/EryAT2-flank, for module 2 (Fig. 3B).

Individual heterologous AT domains were inserted betweenthe two flanking DNA regions of the DEBS AT1 or AT2domains as shown in Fig. 3B. The PCR-generated DNA frag-ments carrying “Hyg” AT2, “Ven” AT, and RAPS AT14 werefirst isolated from the corresponding subclones (pUC18/“Hyg”AT2, pUC18/“Ven” AT, and pUC19/RapAT14, respectively)by digestion with AvrII and NsiI and then subcloned in theintegrative cassette pWHM3/EryAT1-flank to generate threeplasmids, pWHM3/EryAT1/“Hyg” AT2, pWHM3/EryAT1/“Ven” AT, and pWHM3/EryAT1/RapAT14, for AT replace-ments at module 1 (Fig. 3B). Similar steps were taken togenerate pWHM3/EryAT2/“Hyg” AT2 and pWHM3/EryAT2/RapAT14 for AT replacements at module 2 (Fig. 3B).

Construction of the AT replacement strains of S. erythraeaER720. The AT domains of modules 1 and 2 in DEBS1 of S.erythraea ER720 were replaced by using the above-describedconstructs containing the different heterologous ATs in a two-step procedure that involves homology-based integration ofthe recombinant plasmids into the chromosome of S. erythraeaER720, followed by homology-driven resolution of the inte-

FIG. 2. Dendrogram generated from PILEUP analysis of the AT domains.Amino acid sequences of the “Hyg” AT2 (28) and “Ven” AT domains werealigned with those of the 14 RAPS AT and 6 DEBS AT domains.

6420 RUAN ET AL. J. BACTERIOL.

on January 7, 2021 by guesthttp://jb.asm

.org/D

ownloaded from

grants to obtain the desired replacements. After protoplasttransformations with the AT replacement plasmids (;3 mg ofplasmid DNA typically used for each transformation), one tofive Thior integrative transformants were typically obtained.Integration of the plasmid into the chromosome was verified

by Southern hybridization using either pWHM3 vector se-quence or the heterologous AT sequence as a probe (data notshown). One integrant from each transformation was thenused for resolution of the integration event. Two to six thio-strepton-sensitive resolvants were commonly obtained from

FIG. 3. Scheme for the construction of the AT replacement plasmids. (A) Alignment of the amino acid sequences of the three heterologous ATs and two residentAT domains to be replaced. The boxed residues shown at the N and C termini of the AT domains were defined as the boundaries for the AT replacements. Two uniquerestriction sites, AvrII (CCTAGG) and NsiI (ATGCAT), were introduced in the boundary regions. The introduction of the AvrII site causes residue changes to P (Pro)and R (Arg) in the N terminus (as arrows indicate), and the NsiI site causes a change to A (Ala) in the C terminus (as the arrow indicates). (B) The flanking regioncassettes for the AT replacements were constructed as pWHM3/EryAT1-flank for module 1 and pWHM3/EryAT2-flank for module 2. Numbers above each flankingregion refer to the relevant sequence positions in the eryA sequence deposited in GenBank (accession no. M63676). Three PCR-generated fragments carrying theheterologous ATs were isolated from their corresponding subclones of pUC18/“Hyg” AT2, pUC18/“Ven” AT, and pUC19/RapAT14 (see Materials and Methods) bydigestion with AvrII and NsiI and subcloned in pWHM3/EryAT1-flank or pWHM3/EryAT2-flank, respectively, to generate the AT replacement plasmids in module 1(pWHM3/EryAT1/“Hyg” AT2, pWHM3/EryAT1/“Ven” AT, and pWHM3/EryAT1/RapAT14) or module 2 (pWHM3/EryAT2/“Hyg” AT2 and pWHM3/EryAT2/RapAT14).

VOL. 179, 1997 ERYTHROMYCIN DERIVATIVES VIA AT SUBSTITUTIONS 6421

on January 7, 2021 by guesthttp://jb.asm

.org/D

ownloaded from

the resolutions. These were then subjected to Southern anal-ysis (data not shown) to determine whether resolution resultedin the desired genotype wherein the resident AT domain hadbeen replaced by the heterologous AT. The five AT-replacedstrains of S. erythraea ER720 obtained were designated S. eryth-raea EryAT1/“Hyg” AT2, S. erythraea EryAT2/“Hyg” AT2, S.erythraea EryAT1/“Ven” AT, S. erythraea EryAT1/RapAT14,and S. erythraea EryAT2/RapAT14 (Table 1).

Novel erythromycin derivatives produced by the AT replace-ment strains. Macrolide compounds produced by the AT re-placement strains of S. erythraea ER720 were isolated as de-scribed in Materials and Methods and characterized by TLC,TLC-bioautography, MS, and (for S. erythraea EryAT1/“Hyg”AT2 and S. erythraea EryAT2/“Hyg” AT2) NMR analysis.

The AT replacements in module 1 (S. erythraea EryAT1/“Hyg” AT2, S. erythraea EryAT1/“Ven” AT, and S. erythraeaEryAT1/RapAT14) all appear to produce the same erythro-mycin derivative, based on both TLC and MS analysis of cul-ture broth extracts. As shown in Fig. 5, this new compound isvisible as a blue-staining spot that runs slightly faster thanerythromycin A. MS analysis of these samples after isolationfrom a TLC plate revealed a dominant M 1 H1 peak at m/z704, which is 30 units lower than for erythromycin A (M 1 H1

5 m/z 734). This result suggests that these compounds areerythromycin A derivatives lacking both a methyl group and ahydroxyl group. In agreement with the MS results, the majormacrolide species produced by S. erythraea EryAT1/“Hyg”AT2 was characterized by its 1H and 13C NMR spectra as12-desmethyl-12-deoxyerythromycin A (Table 3 and Fig. 4B).

To determine whether 12-desmethyl-12-deoxyerythromycinA is biologically active, the culture broth extracts preparedfrom the three AT replacement strains in module 1 were sub-jected to TLC-bioautography. A unique inhibition zone corre-sponding to the position of the novel TLC spot (Fig. 5) wasobserved in all samples assayed (data not shown). This resultindicates that the novel compound has significant bioactivityagainst S. aureus and may be the only biologically active mac-rolide produced by these strains, since no other inhibition zonewas observed even when a 10-fold-higher concentration of thecrude samples was applied (data not shown). To compare theamounts of compound produced by these three strains, cellswere grown in triplicate, extracts were normalized for cellgrowth, and the bioactivity of serially diluted extracts spottedon bioassay discs was measured by the size of the zone ofinhibition against S. aureus. S. erythraea EryAT1/“Hyg” AT2was found to produce approximately two- to fourfold morecompound than either S. erythraea EryAT1/RapAT14 or S.erythraea EryAT1/“Ven” AT.

Similar characterizations were carried out for the AT re-placement strains in module 2. S. erythraea EryAT2/“Hyg”AT2 and S. erythraea EryAT2/RapAT14 appear to produce thesame erythromycin derivatives, based on TLC and MS analysisof culture broth extracts. As shown in Fig. 5, two novel spots,one running slightly faster than erythromycin A and anotherrunning slightly slower than erythromycin A, were identified byTLC in extracts of culture supernatants. MS analysis of thesamples revealed two predominant species with the M 1 H1

molecular masses of m/z 720, corresponding to the slow-run-ning TLC spot, and m/z 704, corresponding to the fast-runningTLC spot. This is in agreement with erythromycin A deriva-tives lacking a methyl group and lacking both a methyl and ahydroxyl group, respectively. The predicted structures of thesetwo derivatives are shown in Fig. 4C. In addition, it was ob-served that the ratio of the two compounds (determined as theratio of the peak height at m/z 720 to that at m/z 704) variedslightly with culture conditions and/or the extraction process.

TLC-bioautography of the crude extract clearly indicatesthat both compounds are biologically active against S. aureus(data not shown). In contrast to the production of 12-des-methyl-12-deoxyerythromycin A, the yield of the 10-desmeth-ylerythromycin derivatives by S. erythraea EryAT2/“Hyg” AT2did not differ significantly from that produced by S. erythraeaEryAT2/RapAT14. In addition, as shown in Fig. 5, the produc-tion of 10-desmethylerythromycin derivatives by S. erythraeaEryAT2/“Hyg” AT2 and S. erythraea EryAT2/RapAT14seemed to be significantly lower than that of the 12-desmeth-ylerythromycin derivative, based on the intensities of the TLCstaining spots (Fig. 5).

For NMR structural studies, the erythromycin derivativesproduced by S. erythraea EryAT2/“Hyg” AT2 were isolatedfrom a 27-liter fermentation as described in Materials andMethods. Unfortunately, the expected compound, 10-des-methylerythromycin A, was not recovered in sufficient yield forstructure determination, but several milligrams of a presumedbreakdown product, the 6,9:9,12-spiroketal derivative of 10-

TABLE 3. NMR data for 12-desmethyl-12-deoxyerythromycin Aand 10-desmethylerythromycin A 6,9:9,12-spiroketal in CDCl3

Position

NMR chemical shifta

13C 1H

1 2 1 2

1 176.8 179.02 45.1 45.6 2.74 3.363 79.4 76.1 4.15 4.364 41.7 42.1 2.01 2.095 83.5 85.4 3.58 3.426 75.1 81.67 38.7 41.2 1.91/1.66 2.13/1.468 43.7 39.8 2.86 2.149 219.5 115.510 46.1 42.5 2.70 2.56/2.3811 66.9 80.1 4.05 3.8612 40.3 83.9 1.71/1.4613 73.7 81.6 5.06 5.1814 28.0 24.2 1.61/1.56 2.00/1.6615-H3 9.6 10.9 0.89 0.842-CH3 14.4 13.6 1.19 1.094-CH3 9.6 17.1 1.13 1.136-CH3 26.3 27.7 1.38 1.418-CH3 18.5 12.0 1.19 0.9010-CH3 10.6 1.1112-CH3 24.5 1.2719 103.2 102.4 4.47 4.2329 70.8 69.7 3.25 3.1839 65.4 65.8 2.49 2.5249 28.9 28.7 1.69/1.26 1.68/1.2559 69.2 69.4 3.54 3.4969-H3 21.3 21.2 1.23 1.22N(CH3)2 40.3 40.4 2.30 2.3010 96.6 94.7 4.85 5.2720 34.9 34.6 2.40/1.59 2.27/1.5330 72.7 72.840 77.9 78.3 3.03 2.9850 66.0 65.0 4.04 3.9860-H3 18.4 17.8 1.29 1.1830-CH3 21.5 21.6 1.25 1.20OCH3 49.5 49.3 3.33 3.26

a NMR chemical shift assignments for 12-desmethyl-12-deoxyerythromycin A(1) and 10-desmethylerythromycin A 6,9:9,12-spiroketal (2). The absence of 13Csignals for the 12-CH3 and 10-CH3 in 1 and 2, respectively, coupled with the newmethylene signals on C-12 (40.3 13C; 1.71/1.46 1H) in 1 and C-10 (42.5; 2.56/2.38)in 2 indicate that these compounds are the expected desmethylerythromycinderivatives.

6422 RUAN ET AL. J. BACTERIOL.

on January 7, 2021 by guesthttp://jb.asm

.org/D

ownloaded from

desmethylerythromycin A, was isolated. From the 1H and 13CNMR spectra of this compound, it was clear that the methylgroup normally attached to C-10 was absent and that the C-10methine had been replaced by a methylene, as expected for a10-desmethylerythromycin derivative (Table 3). The massspectral data were also in agreement with the structural assign-ment, giving an M 1 H1 ion at m/z 702. It is known thaterythromycin A can be converted to its 6,9:9,12-spiroketal de-rivative in the presence of aqueous acid (42). It appears that10-desmethylerythromycin A is even more labile to such rear-rangement than the parent compound.

DISCUSSION

Methylmalonyl-CoA is the only extender unit used by theerythromycin PKS (DEBS) to build the polyketide core oferythromycin. To alter the polyketide backbone structure,three heterologous AT domains, “Hyg” AT2, RAPS AT14,and “Ven” AT, which were predicted to use malonyl-CoA,based on sequence similarity to other putative malonyl-CoA-specific acyltransferases, were chosen to replace the AT1 and

AT2 domains of the DEBS 1 subunit of the erythromycin PKS.The heterologous ATs were isolated from three different PKSclusters from two non-Saccharopolyspora actinomycetes. Oneof the three, RapAT14, was selected from the 14th module ofthe rapamycin PKS and, as shown in Fig. 2, appears to be theleast conserved among the seven rapamycin malonylacyltrans-ferase domains. The choice of RapAT14 was purely arbitrary.S. erythraea strains carrying these heterologous AT replace-ments produced novel erythromycin derivatives which lack amethyl group at the predicted position on the macrolactonering. All three AT replacements in module 1 appeared toproduce the same erythromycin derivative, 12-desmethyl-12-deoxyerythromycin A (12-desmethylerythromycin B). S. eryth-raea EryAT1/“Hyg” AT2 produced the highest level of thenovel compound, comparable to the level of erythromycin Aproduced by the parental strain. The other two strains carryingthe replacements at AT1 produced less 12-desmethyl-12-de-oxyerythromycin A. Substantially lower yields, however, wereobserved when the AT in module 2 of DEBS1 was replaced:the level of the 10-desmethylerythromycins produced by both

FIG. 4. Scheme showing AT replacements in module 1 or 2 of DEBS and compounds produced by replacement strains. (A) Erythromycin A produced by wild-typeS. erythraea ER720. (B) Erythromycin derivative lacking the methyl group at C-12 of the macrolactone ring produced by recombinant strains of S. erythraea ER720wherein the resident AT domain of module 1 was replaced by three heterologous ATs specifying malonate in place of methylmalonate. (C) Erythromycin derivativeslacking the methyl group at C-10 of the macrolactone ring produced by recombinant strains of S. erythraea ER720 wherein the resident AT domain of module 2 wasreplaced by two heterologous ATs specifying malonate in place of methylmalonate. Abbreviations: ACP, acyl carrier protein; KR, b-ketoreductase; KS, b-ketoacyl ACPsynthase; mmAT, methylmalonyl AT; mAT, malonyl AT; pAT, propionyl AT.

VOL. 179, 1997 ERYTHROMYCIN DERIVATIVES VIA AT SUBSTITUTIONS 6423

on January 7, 2021 by guesthttp://jb.asm

.org/D

ownloaded from

S. erythraea EryAT2/“Hyg” AT2 and S. erythraea EryAT2/RapAT14 was at least fivefold less than the level of erythro-mycin produced by S. erythraea ER720. Since wild-type levelsof 12-desmethylerythromycin were made in at least one of theAT1 constructs, the basis for the large variations in rates ofsynthesis cannot be attributed to either the limitation of mal-onyl-CoA for incorporation or the efficiency of processing thenascent polyketide lacking a side chain in a single position.Rather, it is likely that the large variations in rates of synthesisare related to the effects of introducing heterologous domainson the structure and functions of the other domains in themultifunctional polypeptide.

The AT domains accept the extender unit and pass it to theACP domain within the module that it occupies. Though ATdomain swapping may be permitted, detrimental effects on therates of extender unit transfer may occur when noncognateAT-ACP or putative AT-KS interactions take place. Thus, thelarge variations in the levels of polyketide production observedwhen different heterologous AT domains were used to replacea given resident AT domain in module 1 or 2 in DEBS 1 arenot unexpected. Kuhstoss et al. (20) found that when the AT-ACP domains of the starter module (responsible for loadingthe starter unit in spiramycin synthesis) were replaced by acorresponding region of the starter module of the tylosin PKS,the rate of synthesis of the hybrid polyketide actually went upthree- to fourfold. Until the structures of PKSs are understood,therefore, the effect of AT swapping on the rate of polyketidesynthesis cannot be predicted.

These effects may also explain the results observed whenreplacements using the three heterologous malonyl AT do-mains were attempted in other modules of DEBS. In summary,strains which carried “Hyg” AT2 (the domain that yielded thehighest level of hybrid polyketides in modules 1 and 2) inmodule 3, 4, 5, or 6 did not produce detectable levels oferythromycin analogs, nor did other strains carrying replace-ments using “Ven” AT or RapAT14 in selected modules (datanot shown). It is likely, therefore, that more information on thestructure of the PKS will be required before it can be predicted

a priori whether a given modular replacement will yield pro-ductive levels of novel polyketides.

Productive heterologous AT-ACP interactions in aromaticPKSs have also been reported (19). Aromatic PKSs are com-posed of discrete monofunctional (or difunctional) polypep-tides that assemble to form a complex. In these and laterexperiments (24a), one ACP was successfully substituted foranother. Most aromatic PKS clusters do not contain a genespecifying an AT function, but since malonate is always used asthe extender in aromatic polyketide synthesis, it is believed thatthese PKSs employ the AT component of the host fatty acidsynthase complex (27a, 35). Thus, the noncognate interactionsapparently took place between a fatty acid AT and a PKS ACP.

It was not surprising that hydroxylation at C-12 of the ma-crolactone ring, mediated by EryK, a P450 enzyme (34), didnot take place on the 12-desmethylerythromycin derivative.P450-mediated hydroxylation involves a free radical mecha-nism that should be several hundred-fold less favored energet-ically at the secondary (-CH2-) carbon rather than the tertiary[-CH(CH3)-] carbon normally found at C-12 of erythromycin.In the 10-desmethyl derivatives, approximately one-half of thecompounds were C-12 hydroxylated. Since the overall level ofproduction of the 10-desmethylerythromycins was much lowerthan the level of erythromycin produced by the parental strain,the finding of only 50% hydroxylation suggests very weak sub-strate activity of 10-desmethylerythromycin D for EryK. (In thenormal pathway to erythromycin A, erythromycin D is con-verted to erythromycin C by EryK: purified EryK exhibits a1,900-fold preference for erythromycin D over erythromycin Bas a substrate [21].) Interestingly, no 12-desmethylerythromy-cin D or 10-desmethylerythromycin C or D was produced inany of the AT1 or AT2 domain replacement strains. Theseresults indicate that EryG, the enzyme that O-methylates theC0-3 hydroxyl on the L-mycarose moiety, can use either the 10-or 12-desmethyl derivative efficiently as a substrate. This couldnot have been predicted since it was previously discovered thatsome erythromycin derivatives that contained changes in thestructure of the lactone ring, e.g., 6-deoxyerythromycin, 2-des-methylerythromycin, and D6,7-anhydroerythromycin C, had re-duced activity as substrates for EryG (9, 24, 27, 40).

Replacement of a methylmalonyl with a malonyl extender ina polyketide through swapping of the AT1 domain of DEBS 1with the AT domain from the second module of the rapamycinPKS was first reported by Oliynyk et al. (26). Although thatwork used a DEBS 1 model system that produced a triketidelactone, it showed unambiguously that the AT domains alonein DEBS 1 (and likely all complex PKSs) specify the structureof the extender unit incorporated during polyketide synthesis.In addition, a comparison of all AT domains in the rapamycinPKS with those that were previously described for DEBS al-lowed Schwecke et al. (32) to demonstrate that ATs could begrouped into two clusters on the basis of their sequences: thosespecifying malonate and those specifying methylmalonate. Thefindings reported here confirm and expand the previous workby demonstrating that two AT domains in DEBS 1 can bereplaced by three different heterologous, malonyl-specifyingATs. However, the signature sequences proposed by Haydocket al. (12) for malonyl-incorporating ATs (ETGYAxxxxxxxQx-AxFGLL; 11 conserved amino acids) may apply only to therapamycin PKS. The corresponding domain in “Hyg” AT2 hasonly four conserved residues (rTdhAxxxxxxxexAxyrLw), andthe “Ven” AT domain has only three conserved residues(rTvftxxxxxxxexsxFrLf). Nonetheless, the AT groups proposedby Schwecke et al. (32) are very important. In two of the threecases, the AT used for replacement in either module 1 ormodule 2 of DEBS 1 was from a PKS-like gene cluster whose

FIG. 5. TLC analysis of the erythromycin derivatives produced by AT re-placement strains of S. erythraea ER720. Cell culture broth of the AT replace-ment strains in modules 1 and 2 were extracted as described in Materials andMethods. Lane 1, wild-type strain S. erythraea ER720; lanes 2 to 4, 12-desmethyl-12-deoxyerythromycin A producers S. erythraea EryAT1/“Hyg” AT2, S. erythraeaEryAT1/RapAT14, and S. erythraea EryAT1/“Ven” AT, respectively; lane 5,erythromycin A standard (5 mg); lanes 6 and 7, 10-desmethylerythromycin A and10-desmethyl-12-deoxyerythromycin A producers S. erythraea EryAT2/“Hyg”AT2 and S. erythraea EryAT2/RapAT14, respectively. Positions corresponding toerythromycin A are indicated by arrows.

6424 RUAN ET AL. J. BACTERIOL.

on January 7, 2021 by guesthttp://jb.asm

.org/D

ownloaded from

corresponding polyketide was not yet identified. Thus, ratio-nally designed changes in DEBS can be made with uncharac-terized or partially characterized PKS components. It appearspromising, therefore, that novel, rationally designed complexpolyketides can be produced through the assembly of PKScomponents from a variety of organisms with only sequenceinformation to guide the choice of AT domains.

ACKNOWLEDGMENTS

We thank T. Kavanaugh, L. Zulawinski, and S. Nannapeneni forsynthesis of oligonucleotides, J. Wang and K. Matuszak for MS anal-ysis, and D. Whittern and S. Malvar for NMR experiments.

REFERENCES1. Altschul, S. F., W. Gish, W. Miller, E. W. Myers, and D. J. Lipman. 1990.

Basic local alignment search tool. J. Mol. Biol. 215:403–410.2. Amy, C. M., A. Witkowski, J. Naggert, B. Williams, Z. Randhawa, and S.

Smith. 1989. Molecular cloning and sequencing of cDNAs encoding theentire rat fatty acid synthase. Proc. Natl. Acad. Sci. USA 86:3114–3118.

3. Aparicio, J. F., I. Molnar, T. Schwecke, A. Konig, S. F. Haydock, L. E. Khaw,J. Staunton, and P. F. Leadlay. 1996. Organization of the biosynthetic genecluster for rapamycin in Streptomyces hygroscopicus: analysis of the enzymaticdomains in the modular polyketide synthase. Gene 169:9–16.

4. Caffrey, P., D. J. Bevitt, J. Staunton, and P. F. Leadlay. 1992. Identificationof DEBS1, DEBS2, and DEBS3, the multienzyme polypeptides of the eryth-romycin-producing polyketide synthase from Saccharopolyspora erythraea.FEBS Lett. 304:225–228.

5. Cortes, J., S. F. Haydock, G. A. Roberts, D. J. Bevitt, and P. F. Leadlay. 1990.An unusually large multifunctional polypeptide in the erythromycin-produc-ing polyketide synthase of Saccharopolyspora erythraea. Nature 348:176–178.

6. Devereux, J., P. Haeberli, and O. Smithies. 1984. A comprehensive set ofsequence analysis programs for the VAX. Nucleic Acids Res. 12:387–395.

7. Dewitt, J. P. 1985. Evidence for a sex factor in Streptomyces erythreus. J.Bacteriol. 164:969–971.

8. Donadio, S., and L. Katz. 1992. Organization of the enzymatic domains inthe multifunctional polyketide synthase involved in erythromycin formationin Saccharopolyspora erythraea. Gene 111:51–60.

9. Donadio, S., J. B. McAlpine, P. J. Sheldon, M. Jackson, and L. Katz. 1993.An erythromycin analog produced by reprogramming polyketide synthesis.Proc. Natl. Acad. Sci. USA 90:7119–7123.

10. Donadio, S., M. J. Staver, J. B. McAlpine, S. J. Swanson, and L. Katz. 1991.Modular organization of genes required for complex polyketide biosynthesis.Science 252:675–679.

11. Fu, H., R. McDaniel, D. A. Hopwood, and C. Khosla. 1994. Engineeredbiosynthesis of novel polyketides: stereochemical course of two reactionscatalyzed by a polyketide synthase. Biochemistry 33:9321–9326.

12. Haydock, S., J. F. Aparicio, I. Molnar, T. Schwecke, A. Konig, A. F. A.Marsden, I. S. Galloway, J. Staunton, and P. F. Leadlay. 1995. Divergentstructural motifs correlated with the substrate specificity of (methyl)malonyl-CoA:acylcarrier protein transacylase domains in modular polyketide syn-thases. FEMS Lett. 374:246–248.

13. Haydock, S. F., J. A. Dowson, N. Dhillon, G. A. Roberts, J. Cortes, and P. F.Leadlay. 1991. Cloning and sequence analysis of genes involved in erythro-mycin biosynthesis in Saccharopolyspora erythraea: sequence similarities be-tween EryG and a family of S-adenosylmethionine-dependent methyltrans-ferases. Mol. Gen. Genet. 230:120–128.

14. Hopwood, D. A., M. J. Bibb, K. F. Chater, T. Kieser, C. J. Bruton, H. M.Kieser, D. J. Lydiate, C. P. Smith, J. M. Ward, and H. Schempf. 1985.Genetic manipulation of Streptomyces: a laboratory manual. The John InnesFoundation, Norwich, England.

15. Hopwood, D. A., and D. H. Sherman. 1990. Molecular genetics of polyketidesand its comparison to fatty acid biosynthesis. Annu. Rev. Genet. 24:37–66.

16. Hostettmann, K., and A. Marston. 1990. Liquid-liquid partition chromatog-raphy in natural product isolation. Anal. Chim. Acta 236:63–76.

17. Kao, C. M., G. L. Luo, L. Katz, D. E. Cane, and C. Khosla. 1995. Manipu-lation of macrolide ring size by directed mutagenesis of a modular polyketidesynthase. J. Am. Chem. Soc. 117:9105–9106.

18. Katz, L., and S. Donadio. 1993. Polyketide synthesis: prospect for hybridantibiotics. Annu. Rev. Microbiol. 47:875–912.

19. Khosla, C., R. McDaniel, S. Ebert-Khosla, R. Torres, D. H. Sherman, M. J.Bibb, and D. A. Hopwood. 1993. Genetic construction and functional analysisof hybrid polyketide synthases containing heterologous acyl carrier proteins.J. Bacteriol. 175:2197–2204.

20. Kuhstoss, S., M. Huber, J. R. Turner, J. W. Paschal, and N. Rao. 1996.Production of a novel polyketide through the construction of a hybridpolyketide synthase. Gene 183:231–236.

21. Lambalot, R. H., D. E. Cane, J. J. Aparicio, and L. Katz. 1995. Overproduc-tion and characterization of the erythromycin C-12 hydroxylase, EryK. Bio-chem. 34:1858–1866.

22. MacNeil, D. J., J. L. Occi, K. M. Gewain, T. MacNeil, P. H. Gibbons, R.McDaniel, S. Ebert-Khosla, D. A. Hopwood, and C. Khosla. 1993. Engi-neered biosynthesis of novel polyketides. Science 262:1546–1550.

23. Marsden, A. F. A., P. Caffrey, J. F. Aparicio, M. S. Loughran, J. Staunton,and P. F. Leadlay. 1994. Stereospecific acyl transfers on the erythromycin-producing polyketide synthase. Science 263:378–379.

24. McAlpine, J. B., J. S. Tuan, D. P. Brown, K. D. Grebner, D. N. Whittern, A.Buko, and L. Katz. 1987. New antibiotics from genetically engineered acti-nomycetes. J. Antibiot. 40:1115–1122.

24a.McDaniel, R., S. Ebert-Khosala, D. A. Hopwood, and C. Khosla. 1995.Rational design of aromatic polyketide natural products by recombinantassembly of enzymatic subunits. Nature 375:549–554.

25. Molnar, I., J. F. Aparicio, S. F. Haydock, L. E. Khaw, T. Schwecke, and D.O’Hagan. 1993. Biosynthesis of fatty acid and polyketide metabolites. Nat.Prod. Rep. 10:593–624.

26. Oliynyk, M., M. J. B. Brown, J. Cortes, J. Staunton, and P. F. Leadlay. 1996.A hybrid modular polyketide synthase obtained by domain swapping. Chem.Biol. 3:833–839.

27. Paulus, T. J., J. S. Tuan, V. I. Luebke, G. T. Maine, J. P. DeWitt, and L. Katz.1990. Mutation and cloning of eryG, the structural gene for erythromycinO-methyltransferase from Saccharopolyspora erythraea, and expression oferyG in Escherichia coli. J. Bacteriol. 172:2541–2546.

27a.Revill, W. P., M. J. Bibb, and D. A. Hopwood. 1995. Purification of amalonyltransferase from Streptomyces coelicolor A3(2) and analysis of itsgenetic determinant. J. Bacteriol. 177:3946–3952.

28. Ruan, X., D. Stassi, S. Lax, and L. Katz. A second type-I PKS gene clusterisolated from Streptomyces hygroscopicus ATCC29253, a rapamycin-produc-ing strain. Gene, in press.

29. Ruby, C. L., and S. J. Danis. 1992. Complex organization of the Streptomycesavermitilis genes encoding the avermectin polyketide synthase. Gene 115:119–125.

30. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: alaboratory manual, 2nd ed. Cold Spring Harbor Laboratory, Cold SpringHarbor, N.Y.

31. Sanger, F., S. Nicklen, and A. R. Coulson. 1977. DNA sequencing withchain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 74:5463–5467.

32. Schwecke, T., J. F. Aparicio, I. Molnar, A. Konig, L. E. Khaw, S. F. Haydock,M. Oliynyk, P. Caffrey, J. Cortes, J. B. Lester, G. A. Bohm, J. Staunton, andP. F. Leadlay. 1995. The biosynthetic gene cluster for the polyketide immu-nosuppressant rapamycin. Proc. Natl. Acad. Sci. USA 92:7839–7843.

33. Serre, L., E. C. Verbree, Z. Dauter, A. R. Stuitje, and Z. S. Derewenda. 1995.The Escherichia coli malonyl-CoA:acyl carrier protein transacylase at 1.5 Aresolution. J. Biol. Chem. 270:12961–12964.

34. Stassi, D., S. Donadio, M. J. Staver, and L. Katz. 1993. Identification of aSaccharopolyspora erythraea gene required for the final hydroxylation step inerythromycin biosynthesis. J. Bacteriol. 175:182–189.

35. Summers, R. G., A. Ali, B. Shen, W. A. Wessel, and C. R. Hutchinson. 1995.Malonyl-coenzyme A:acyl carrier protein acyltransferase of Streptomycesglaucescens: a possible link between fatty acid and polyketide biosynthesis.Biochemistry 34:9389–9402.

36. Summers, R. G., S. Donadio, M. J. Staver, E. Wendt-Pienkowski, C. R.Hutchinson, and L. Katz. Characterization of ten genes from the erythro-mycin biosynthetic gene cluster of Saccharopolyspora erythraea that are in-volved in L-mycarose and D-desosamine production. Microbiology, in press.

37. Swan, D. G., A. M. Rodriguez, C. Vilches, C. Mendez, and J. A. Salas. 1994.Characterization of a Streptomyces antibioticus gene encoding a type Ipolyketide synthase which has an unusual coding sequence. Mol. Gen.Genet. 242:358–362.

38. Tuan, J. S., J. M. Weber, M. J. Staver, J. O. Leung, S. Donadio, and L. Katz.1990. Cloning of genes involved in erythromycin biosynthesis from Saccha-ropolyspora erythraea using a novel actinomycete-Escherichia coli cosmid.Gene 90:21–29.

39. Vara, J., M. Lewandowska-Skarbek, Y.-G. Wang, S. Donadio, and C. R.Hutchinson. 1989. Cloning of genes governing the deoxysugar portion of theerythromycin biosynthesis pathway in Saccharopolyspora erythraea (Strepto-myces erythreus). J. Bacteriol. 171:5872–5881.

40. Weber, J. M., J. O. Leung, S. J. Swanson, K. B. Idler, and J. B. McAlpine.1991. An erythromycin derivative produced by targeted gene disruption inSaccharopolyspora erythraea. Science 252:114–117.

41. Weber, J. M., and R. Losick. 1988. The use of a chromosome integrationvector to map erythromycin resistance and production genes in Sacchar-opolyspora erythraea (Streptomyces erythreus). Gene 68:173–180.

42. Wiley, P. F., K. Gerzon, E. H. Flynn, M. V. Sigal, O. Weaver, U. C. Quarck,R. R. Chauvette, and R. Monahan. 1957. Erythromycin. X. Structure oferythromycin. J. Am. Chem. Soc. 79:6062–6070.

43. Witkowski, A., V. S. Rangan, Z. I. Randhawa, C. M. Amy, and S. Smith. 1991.Structural organization of the multifunctional animal fatty-acid synthase.Eur. J. Biochem. 198:571–579.

44. Yamamoto, H., K. H. Maurer, and C. R. Hutchinson. 1986. Transformationof Streptomyces erythraeus. J. Antibiot. 39:1304–1313.

VOL. 179, 1997 ERYTHROMYCIN DERIVATIVES VIA AT SUBSTITUTIONS 6425

on January 7, 2021 by guesthttp://jb.asm

.org/D

ownloaded from