Metabolic Engineering 15 (2013) 167–173

Contents lists available at SciVerse ScienceDirect

Metabolic Engineering

1096-71

http://d

n Corr

E-m1 In

Biologia

Cardena2 Sc3 Ch

NJ 08904 Ca5 M

Chaoyan

journal homepage: www.elsevier.com/locate/ymben

Recombinant strains for the enhanced production of bioengineered rapalogs

Steven G. Kendrew a, Hrvoje Petkovic a,1, Sabine Gaisser a, Sarah J. Ready a, Matthew A. Gregory a,Nigel J. Coates a, Mohammad Nur-e-Alam a,2, Tony Warneck a, Dipen Suthar a, Teresa A. Foster a,Leonard McDonald b,3, Gerhard Schlingman b, Frank E. Koehn b, Jerauld S. Skotnicki b, Guy T. Carter b,4,Steven J. Moss a, Ming-Qiang Zhang a,5, Christine J. Martin a, Rose M. Sheridan a, Barrie Wilkinson a,n

a Biotica Technology Ltd, 3 Riverside Suite 5, Granta Park, Cambridge CB21 6AD, UKb Pfizer, Natural Products, Eastern Point Road, Groton, CT 06340, USA

a r t i c l e i n f o

Article history:

Received 9 August 2012

Received in revised form

16 October 2012

Accepted 3 November 2012Available online 17 November 2012

Keywords:

Rapamycin

Polyketide

Bioengineering

76/$ - see front matter & 2012 Elsevier Inc. A

x.doi.org/10.1016/j.ymben.2012.11.001

esponding author. Fax: þ44 1223 895888.

ail address: barrie.wilkinson@biotica.com (B.

stituto de Biomedicina y Biotecnologıa de

Molecular, Facultad de Medicina, Universida

l Herrera Oria, s/n., Santander 39011, Cantab

hool of Chemistry, Bangor University, Bangor

romocell Corporation, 685 U.S. Highway One

2, USA.

rter-Bernan Consulting, LLC, New City, NY 10

erck Sharp & Dohme R&D (China) Company

g District, Beijing 100015, PR China.

a b s t r a c t

The rapK gene required for biosynthesis of the DHCHC starter acid that initiates rapamycin biosynthesis

was deleted from strain BIOT-3410, a derivative of Streptomyces rapamycinicus which had been subjected to

classical strain and process development and capable of robust rapamycin production at titres up to

250 mg/L. The resulting strain BIOT-4010 could no longer produce rapamycin, but when supplied

exogenously with DHCHC produced rapamycin at titres equivalent to its parent strain. This strain enabled

mutasynthetic access to new rapalogs that could not readily be isolated from lower titre strains when fed

DHCHC analogs. Mutasynthesis of some rapalogs resulted predominantly in compounds lacking late post

polyketide synthase biosynthetic modifications. To enhance the relative production of fully elaborated

rapalogs, genes encoding late-acting biosynthetic pathway enzymes which failed to act efficiently on the

novel compounds were expressed ectopically to give strain BIOT-4110. Strains BIOT-4010 and BIOT-4110

represent valuable tools for natural product lead optimization using biosynthetic medicinal chemistry and

for the production of rapalogs for pre-clinical and early stage clinical trials.

& 2012 Elsevier Inc. All rights reserved.

1. Introduction

Biosynthetic engineering is established as a complimentaryapproach to semi-synthesis for the structural diversification and leadoptimization of pharmacologically active natural products. The avail-ability of DNA sequence for numerous biosynthetic gene clusters,methods for their manipulation and advances in our understanding ofbiosynthetic enzymology has greatly enhanced our ability to modifythe products of these pathways using genetic engineering (Walsh andFischbach, 2010; Wilkinson and Micklefield, 2007; Wong and Khosla,2012; Wu et al., 2012). Indeed, many clinically and commerciallyrelevant molecules have now been modified using these techniques,providing analogs that are not generally accessible by conventionalchemical techniques and which display usefully altered or improved

ll rights reserved.

Wilkinson).

Cantabria, Departamento de

d de Cantabria, Avda.

ria, Spain.

, UK.

, North Brunswick,

956, USA.

Ltd., 10 Jiu-Xian-Qiao Road,

pharmacological properties (Alexander et al., 2010; Marsden et al.,1998; Menzella et al., 2009; Sheehan et al., 2006; Zhang et al., 2008).Importantly, for biosynthetic engineering to play a role in any drugdiscovery program there must be a well-defined and reliable processfor the production of novel compounds at levels that can be readilyscaled to provide the multi-gram quantities of compound required forpre-clinical testing and beyond (Baltz, 2011).

A key example is found in the amino acid-linked polyketidemacrocycle rapamycin (1) (Fig. 1), a fermentation product ofStreptomyces rapamycinicus sp. nov. (previously Streptomyces hygro-

scopicus (Kumar and Goodfellow, 2008)) NRRL5491. The potentbiological activity of 1 arises from binding to the immunophilinFKBP12, and this complex subsequently binding to and inhibitingspecific functions of mTOR (mammalian/mechanistic target ofrapamycin) (Laplante and Sabatini, 2012; Wang and Proud, 2011).1 is used clinically as an immunosuppressive agent to inhibit organrejection after renal transplant surgery and as the pharmacologicalcomponent of drug-eluting stents in order to prevent restenosis.Semi-synthetic analogs of 1 have recently been approved for anumber of anticancer indications including advanced renal cellcarcinoma. Preclinical semi-synthetic analogs lacking the ability toinhibit mTOR function, but which still bind FK506-binding proteins(FKBPs), have shown potential for the treatment of stroke (Ruanet al., 2008). Despite these successes, the poor physicochemicalproperties displayed by 1 have promoted considerable interest in

Fig. 1. Structure and mutasynthesis of rapalogs using BIOT-4010 and BIOT-4110.

S.G. Kendrew et al. / Metabolic Engineering 15 (2013) 167–173168

the development of improved analogs. Total (‘diverted’) syntheticapproaches to 1 and analogs are impractical and semi-syntheticmethods offer limited scope (Graziani, 2009). Biosynthetic engineer-ing approaches to 1 are therefore extremely appealing and efficienttools to enable this have significant value.

The biosynthetic gene cluster for 1 has been sequenced (Schweckeet al., 1995) and its biosynthesis is well understood (Fig. 2A). Amodular polyketide synthase (PKS) and nonribosomal peptidesynthetase-like enzyme act together to generate prerapamycin (2),the first enzyme free macrocyclic intermediate (Gregory et al., 2004a).A series of monooxygenase and O-methyltransferase catalysed stepsthen complete the biosynthesis and their gene products are encodedby a region of DNA comprising the genes rapIJKLMNOQ (Gregory et al.,2006). Deletion of these genes from S. rapamycinicus NRRL5491 gavestrain S. rapamycinicus MG2-10 (BIOT-1712) which produced norapalogs. Ectopic expression of rapK in BIOT-1712 led to productionof prerapamycin at titres equivalent to those for 1 in the parent strain(Gregory et al., 2004a). We have since shown that RapK is achorismatase responsible for biosynthesis of (4R,5R)-4,5-dihydroxy-cyclohexa-1,5-dienecarboxylic acid (DCDC), the first committed inter-mediate towards biosynthesis of (4R,5R)-4,5-dihydroxycyclohex-1-enecarboxylic acid (DHCHC) which acts as the starter unit substratefor the rapamycin PKS (Andexer et al, 2011) (Fig. 2B).

Addition of exogenous racemic synthetic pseudo-starter unit(1Rn,3Rn,4Rn)-3,4-dihydroxycyclohexanecarboxylic acid (3) to grow-ing cultures of BIOT-1712 led to the efficient production of

prerapamycin (Gregory et al., 2004b, 2005). The alternative additionof DHCHC analogs was subsequently shown to enable the produc-tion of several further rapalogs through mutasynthesis (Gregoryet al., 2004b, 2005; Goss et al., 2006, 2010; Lowden et al., 2004), amethod which represents a versatile approach for the rapid struc-tural diversification of natural products. Several of these compoundswere produced as effectively as prerapamycin, although structurallymore diverse acids such as tetrahydro-2H-pyran-4-carboxylic acid(4) led to rapalogs at trace levels (Gregory et al., 2004b, 2005). Wenow report methods for the straightforward production of gram-plus quantities of novel rapalogs through the construction of strainsusing a combination of traditional and rational metabolic engineer-ing methods. These improvements enabled a thorough investigationof rapalog mutasynthesis to give pharmacologically relevant rapa-logs and enabled the production of material sufficient for pre-clinical evaluation.

2. Materials and methods

2.1. General methods and supporting information

Details of general methods, bioreactor procedures, media,analytical methods and specific compound isolation and charac-terisation are given in the supporting information.

Fig. 2. (A). Biosynthesis of prerapamycin (2) from DHCHC, seven molecules each of malonyl-CoA and (2S)-2-methylmalonyl-CoA and L-pipecolic acid (SAM,

S-adenosylmethionine)—the function of the gene products RapI, J, M, N, O and Q are described in Gregory et al., 2006; (B). Function of the chorismatase RapK to

generate DCDC from chorismic acid, and biosynthesis of the PKS starter acid DHCHC.

S.G. Kendrew et al. / Metabolic Engineering 15 (2013) 167–173 169

2.2. Fermentation and feeding parameters (Falcon tubes)

A single agar plug was used to inoculate RapV7 seed media(7 mL) in a disposable Falcon tube (50 mL) plugged with a foambung and cultured at 28 1C and 300 rpm (2.5 cm throw) for 48 h.MD6 production media (7 mL) was inoculated with this seedculture (0.5 mL) using a wide-bore tip and fermented for 6 days at26 1C and 300 rpm (2.5 cm throw). Where required, exogenouscarboxylic acid feeds were added after 24 h growth in productionmedia. Feeds were typically prepared as a 0.32 M stock solution inmethanol and 50 mL was added to each tube to give a finalconcentration of 2 mM.

2.3. Construction of plasmid pSG3998 and generation

of strain BIOT-4010

Our strategy took advantage of a naturally occurring MfeI siteclose to the 50-end of rapK. To generate upstream and down-stream areas of homology for integration the 7.3 kbp NcoI frag-ment from pR19 (Schwecke et al., 1995) was cloned intopLitmus28 that had been digested with NcoI and dephosphory-lated, and the 4.2 kbp NheI/PstI fragment from cosmid-2(Schwecke et al., 1995) was cloned into pLitmus28 digested withPstI-SpeI. This gave intermediate plasmids pLitmus28–7.3 andpLitmus28–4.2, respectively. To introduce the desired deletionfrom the MfeI site to an internal site of rapK two oligonucleotideswere used to amplify the required region, BioSG159:50–CCCCAATTGGTGTCGCTCGAGAACATCGCCCGGGTGA-30 and BioSG158: 50–CGCCGCAAGTAGCACCGCTCGGCGAAGATCTCCTGG-30

using plasmid pR19 as template. The resulting 1.5 kbp PCRproduct was treated with T4 polynucleotide kinase and clonedinto pLitmus28 that had been digested with EcoRV and depho-sphorylated, and the cloned PCR product was sequenced. The1.5 kbp MfeI-BglII fragment from this plasmid was excised andused to replace the 2.3 kbp MfeI-BglII fragment of pLitmus28–4.2.To complete the construct the 3.3 kbp MfeI-HindIII fragment ofthis plasmid was ligated into similarly digested pLitmus28–7.3.Finally, the deletion construct was transferred into the conjuga-tive Streptomyces vector pKC1132 (Bierman et al., 1992) as aHindIII/XbaI fragment. The final construct was designatedpSG3998. Plasmid pSG3998 was introduced into BIOT-3410 byconjugal transfer using into E. coli ET12567:pUZ8002 (Flett et al.,

1997) and allelic deletion accomplished following publishedprocedures (Gregory et al., 2004a). Apramycin sensitive colonies,representing candidate secondary recombinants, were thengrown to assess rapamycin production. Non-producers weretested further by addition of exogenous trans-4-hydroxyCHCA tothe production media after 24 h to confirm rapalog productionand verify the desired disruption of rapK. The best strain wasdesignated BIOT-4010.

2.4. Construction of plasmid pLL208 and generation

of strain BIOT-4110

The conjugative expression plasmid pLL150 that placesinserted genes under control of the actI/actII-Orf4 promotersystem was constructed by replacing the fC31 phage attachmentsite in pSGset1 (Gregory et al., 2006) with the fBT1 phageattachment site and integrase derived from pRT801 (Gregoryet al., 2003). This was achieved by digesting pSGset1 with SpeIand BstEII and ligating the resulting 2.4 kbp fragment with the3.9 kbp fragment of pRT801 digested with SpeI and BstEII. TherapNOQL genes were inserted into the pLL150 plasmid by ligatingthe SpeI-HindIII fragment from pMG236 (Gregory et al., 2004b)with pLL150 that had been digested with the same enzymes.Plasmid pLL208 was introduced into BIOT-4010 by conjugaltransfer using the methodology described above. Colonies exhi-biting good growth and sporulation were then screened to assessrapalog production. The best strain was designated BIOT-4110.

3. Results and discussion

3.1. Fermentation media development

Unless otherwise stated all fermentation experiments were runin Falcon tubes. Initial experiments with S. rapamycinicus usingpublished media recipes (Sehgal et al., 1976; Shelley and Banks,1992) gave poor yields of 1 which were highly variable. These mediawere developed further by varying nitrogen and carbon sources.Highest productivity was observed when soy flour and starch wereused as the main nitrogen and carbon sources, respectively, andthrough the addition of CaCO3 and MES (2-(N-morpholino)ethane-sulfonic acid) to stabilise pH. Further optimisation involved the

S.G. Kendrew et al. / Metabolic Engineering 15 (2013) 167–173170

introduction of secondary complex nitrogen sources such as yeastand corn steep liquor (CSL), a mineral nitrogen source (NH4)2SO4,salts and mineral elements. The addition of L-lysine (a precursor of L-pipecolic acid) was also found to be advantageous. During theexperiments we observed a relatively slow consumption of starchand therefore introduced the addition of commercial a-amylaseprior to medium sterilisation. The final production medium fromthese efforts was MD5FL (experimental section). As noted below theaddition of fructose to MD5FL to enable higher production with de-repressed strains led to MD6 production medium. The compatibleseed medium RapV7 was developed in parallel.

3.2. Strain development for increased production of 1

A streptomycin resistant mutant of S. rapamycinicus wasisolated by selection on MAM media agar plates containingincreasing concentrations of the antibiotic streptomycin followedby single spore isolation to identify a lineage with elevated 1 titreplus improved sporulation in order to allow robust spore stockpreparation. This technique has been termed ‘ribosome engineer-ing’ and selects for spontaneous mutations in ribosomal andtranscriptional machinery that have pleiotropic metabolic effects(Ochi et al., 2004). For example, mutations in the rpsL geneencoding ribosomal subunit S12 have been found to increasepolyketide titre in several Streptomycetes (Shima et al., 1996;Wang et al., 2008). The best S. rapamycinicus isolate (HY01)robustly produced 1 at �40 mg/L when grown under standardconditions in MD5FL media. Isolate HY01 was then subjected toUV mutagenesis irradiating at 254 nm and resulting isolatesscreened for the selection of strains no longer producing thepolyketide elaiophilin (a co-produced metabolite which causedproblems during compound isolation). After screening severalthousand colonies, approx. 100 colonies potentially lacking elaio-philin production were identified. Following careful re-testing,two strains were found to be stable elaiophilin non-producers, thebest of which produced 1 in the range 95–136 mg/L. After singlecolony isolation the best strain was named HY03.

To improve biomass formation during the early stage offermentation and boost the productivity of 1, MD5FL productionmedium was supplemented with D-glucose. However, addition ofD-glucose at concentrations 420 g/L significantly reduced theyield of 1, indicating the presence of glucose repression (Hodgson,1982). To eliminate D-glucose repression we followed the proce-dure described by Hodgson. We supplied the S. rapamycinicus

cells with D-fructose, a carbon source whose utilization wasshown to be D-glucose repressed, together with the toxic D-glucose analog 2-deoxy-D-glucose (2DOG). 2DOG is not metabo-lized but is nonetheless capable of activating the D-glucoserepression system and hence prevent utilization of the alternativecarbon source. In parallel, we identified the minimal inhibitoryconcentration at which 2DOG completely inhibited the growth ofHY03 in minimal medium containing D-fructose at a concentra-tion of 10 mM and observed a complete inhibition of growth at aconcentration of 50 mM 2DOG. We then followed the two-stageapproach of Hodgson using a UV mutagenized spore preparation.Several resulting isolates displayed a normal growth pattern onminimal medium supplemented with D-fructose and were stillable to grow on glucose as a single carbon source. To confirm thisisolates were grown in minimal medium containing equalamounts of D-fructose and D-glucose. Several mutants consumedboth sugars simultaneously confirming diauxic growth and con-firming the de-repression phenotype. The isolate displaying thebest morphological traits and 1 titre was compared to that ofHY03 and showed an increase in the titre of 1 in response toincreasing concentrations of both D-fructose and D-glucose (up to60 g/L of D-glucose) whereas the titre of 1 was significantly

decreased for HY-03. This was subjected to further rounds ofsingle colony isolation which led to selection of isolate HY07capable of producing 1 in the range of 130–150 mg/L using MD6production medium (MD5FL plus 20 g/L D-fructose).

Isolate HY07 was finally subjected to mutagenesis using longwave UV radiation (365 nm) in the presence of 8-methoxypsoralen.The best isolate produced 1 at titres of 4200 mg/L under standardconditions and the best titre recorded was 278 mg/L. This isolatewas subjected to an additional round of single colony isolation toyield strain HY11 (BIOT-3410). BIOT-3410 was capable of producing1 at 4200 mg/L when grown in bioreactors.

3.3. Deletion of rapK yields BIOT-4010 for enhanced titre

mutasynthesis

In order to access biologically active rapalogs (i.e., with fullpost-PKS processing) we targeted the selective removal of rapK inBIOT-3410 through in-frame deletion. Initial efforts to remove theentire coding region of rapK produced strains that gave predomi-nantly incompletely processed rapalogs when fed exogenousstarter acid (data not shown). Detailed LCMS/MS analysis andcomparison to purified standards (Gregory et al., 2006) showedthese to lack the 9-keto group, indicating a pronounced polareffect upon expression of the rapJ gene which is located imme-diately downstream of rapK. The strategy employed in these firstexperiments either moved the start codon of rapK to the startcodon of rapJ, or alternatively deleted all but 15 bp of rapK

(encoding the first four and the last amino acids). Production offully post-PKS processed rapalogs was restored (to the start codonto start codon mutant) by ectopic expression of a gene cassettecontaining rapJ (data not shown). We believe a cryptic promoterfor the expression of downstream genes may be present in therapK coding region and must be retained for transcription of rapJ.To yield an effective mutasynthesis strain lacking any polar effectwe introduced an in frame deletion that removed the majority ofrapK but retained DNA coding for 4 amino acids at the N-terminusand 80 amino acids at the C-terminus of the RapK protein. Thisretains the part of the rapK coding sequence believed to encodethe promoter sequence for the downstream gene rapJ. To achievethis, the deletion construct pSG3998 was introduced into BIOT-3410 by conjugal transfer from E. coli, followed by sub-culture onMMAM media lacking apramycin selection in order to promote asecondary recombination event. Multiple apramycin sensitivecolonies were selected and incubated with and without trans-4-hydroxycyclohexanecarboxylic acid (5) which was added after24 h of growth. The majority of these strains (�66%) producedthe desired compound 39-desmethoxyrapamycin (BC210, 6)(Lowden et al., 2004) only when fed 5. After single colonyisolation, the best of these (BIOT-4010) produced 6 at approxi-mately 200 mg/L when fed 5 (2 mM), although titres of 6 in excessof 400 mg/L were occasionally observed. These titres representbioconversions of 11.3% (200 mg/L) and 22.6% (400 mg/L) respec-tively for the prochiral substrate 5. The genotype of BIOT-4010was confirmed by: (1) feeding the RapK product DCDC which fullyrecapitulated the production of 1 which includes compete post-PKS processing and verifies that no genomic disruption hasoccurred in the regions surround rapK; and, (2) ectopic expressionof rapK which again fully recapitulated the production of 1(Andexer et al., 2011).

3.4. Robust rapalog production in bioreactors using BIOT-4010

To demonstrate that BIOT-4010 was effective for the scale upof rapalog production it was examined in five separate bioreactorexperiments (15 L working volume in 22 L vessel). The mean titreof 6 observed on addition of 5 was 190718 mg/L (10.7%

S.G. Kendrew et al. / Metabolic Engineering 15 (2013) 167–173 171

bioconversion; n¼5) indicating a robust process with goodproductivity. The process and production medium used in bior-eactors was slightly modified from that used for Falcon tubes.Most notably the new medium MD6-5/1 did not contain a-amylase (see supporting information). Following extraction andpurification, these fermentations provided multi-gram quantitiesof pure 6 after isolation (spectroscopic analysis was consistentwith that published previously for this molecule (Lowden et al.,2004)). An example of a typical bioreactor run is given in thesupporting information.

3.5. BIOT-4010 enables mutasynthesis of previously

inaccessible rapalogs

The availability of BIOT-4010 allowed us to explore theproduction of rapalogs from synthetic starter units previouslyidentified as poor substrates for the rapamycin PKS. Of interestwas tetrahydo-2H-pyran-4-carboxylic acid (4) which gave rapa-logs at trace levels using BIOT-3016 (Gregory et al., 2004b, 2005).When fed to BIOT-4010 growing in Falcon tubes 4 gave newrapalogs with the anticipated MS characteristics. The fully pro-cessed rapalog BC346 (7) was produced at reasonable titre, withsignificant accumulation of the equivalent rapalog lacking the27-methoxy group (BC347, 8) (1471.0 and 42.372.1 mg/L,respectively (3.3% bioconversion; n¼3)) (see Table 1). We thentransferred production into two bioreactors run in parallel, with 4being added after 24 h (2 mM). The harvest titres at day 6 were 22and 23 mg/L of 7 (1.3 and 1.3% bioconversion) and 34 and 36 mg/L

Table 1Comparative titres of mutasynthetic rapalogs produced using BIOT-4010 and BIOT-4

(1S*,3R*,4S*)-3-fluoro-4-hydroxy-cyclohexanecarboxylate (9). Analyses are from n¼3 e

FeedTetrahydo-2H-pyran-4-carboxylic acid (4) M

Product (mg/L) 7 8 7þ8 7/8 ratio 1

BIOT-4010 14.071.0 42.372.1 56.373.1 0.33 5

BIOT-4110 34.371.5 21.371.5 55.773.1 1.61 9

mAU

200

150BIOT-4010 fed 4

50

100

0 2 4 6

0

mAU

150

200

BIOT-4110 fed 4

50

100

0 2 4 6

0

Fig. 3. BIOT-4110 demonstrates enhanced post-PKS processing of mutasynthetic ra

27-desmethoxy rapalogs exist in two interconverting forms that elute as two separate

elute as single peaks.

of 8 (2.0 and 2.1% bioconversion) respectively (see Fig. 3A). Aftercombination and extraction we obtained a crude sample contain-ing 7 (426 mg) and 8 (635 mg) (1.25 mmol total rapalogs;bioconversion of 2.1% based on 60 mmol of 4 fed). A portion ofthis material was processed to purity (495%) and the structuresof both compounds were confirmed by standard spectroscopicmethods.

3.6. Ectopic expression of post-PKS biosynthetic enzymes enhances

rapalog processing

To address the issue of accumulating partially processedrapalogs additional copies of the post-PKS genes rapN, rapO andrapQ were introduced into BIOT-4010. We have previously shownthat rapN encodes for a cytochrome P450 monooxygenase respon-sible for 27-hydroxylation of rapamycin (rapO encodes its cognateferrodoxin) and that rapQ encodes the O-methyltransferase whichacts on the 27-hydroxyl group introduced by RapN (Gregory et al.,2006). An additional copy of rapL was also introduced whichincludes a strong transcriptional terminator and is typicallyplaced at the end of all our expression cassettes as we haveempirically observed a positive effect on the productivity ofresulting strains. The rapNOQL genes were cloned into an inte-grative vector to give pLL208, which was introduced into BIOT-4010 by conjugal transfer. Multiple isolates were selected andscreened for their ability to produce fully post-PKS processedrapalogs. The most effective isolate in this regard was namedBIOT-4110.

110 when fed starter acids tetrahydo-2H-pyran-4-carboxylic acid (4) or methyl

xperiments for each strain/ starter acid combination.

ethyl (1Sn,3Rn,4Sn)-3-fluoro-4-hydroxy-cyclohexanecarboxylate (9)

0 11 10þ11 10/11 ratio

9.071.7 84.071.7 143.073.5 0.70

9.774.6 41.372.1 141.376.8 2.40

8

7

min8 10 12

7

8

min8 10 12

palogs compared to BIOT-4010 (HPLC traces of day 6 fermentation extracts).

peaks on HPLC as observed for compound 8. Rapalogs with a 27-methoxy group

mAU

500

300

400BIOT-4010 fed 9 10 11

100

200

min0 2 4 6 8 10 12

0

mAU

400

500

BIOT-4110 fed 9

10

100

200

300 11

min0 2 4 6 8 10 12

0

Fig. 4. BIOT-4110 demonstrates enhanced post-PKS processing of mutasynthetic rapalogs compared to BIOT-4010 (HPLC traces of day 6 fermentation extracts).

27-desmethoxy rapalogs exist in two interconverting forms that elute as two separate peaks on HPLC as observed for compound 11. Rapalogs with a 27-methoxy group

elute as single peaks.

S.G. Kendrew et al. / Metabolic Engineering 15 (2013) 167–173172

The capability of BIOT-4110 was exemplified by feeding thestarter acids 4 and racemic methyl (1Sn,3Rn,4Sn)-3-fluoro-4-hydroxycyclohexanecarboxylate (9) to Falcon tubes (Table 1).The methyl ester 9 was utilised due to ease of synthesis; it isknown that simple carboxyl esters are effective as exogenousfeeds for mutasynthesis (Goss et al., 2010). We recently describedthe synthesis of the ethyl ester of 9 and generation of aprerapamycin analog after feeding to BIOT-1712 (Goss et al.,2010). For the current study 9 (�30 g) was generated throughan alternative synthetic route described previously in the patentliterature Moss et al. (2011). As observed for 4, when fed to BIOT-4010, 9 gave the expected fully processed equivalent (BC261, 10)as the minor product with the partially processed analog lackingthe 27-methoxy group (BC359, 11) as the major product (seeTable 1 and Figs. 3A and 4A, respectively). In contrast, uponaddition of both starter acid feeds to BIOT-4110 a significant shiftto the production of fully processed rapalogs was observed,making them the major product of fermentation with no sig-nificant change in the total rapalog titre (Table 1 and Figs. 3B and4B, respectively). We then showed that this effect was transferredto bioreactors by running two parallel fermentations, one inocu-lated with each strain and both then fed with 9 (2 mM after 24 h).The titres and product distribution for these experiments werevery similar to those for the experiments run in Falcon tubes.Following harvest the resulting broths were combined and afterwork up gave extracted yields of 10 (1478 mg) and 11 (1619 mg)(3.5 mmol total rapalogs; bioconversion of 11.7% based on30 mmol of the correct 9 diastereoisomer fed (60 mmol race-mate)). A portion of this material was processed to purity (495%)and the structures confirmed by standard spectroscopic methods.

The utility of BIOT-4110 was further exemplified by theisolation of rapalogs which derive from a 3-hydroxybenzoic acid(3HBA, 12) starter acid, our target being the fully processedrapalog BC348 (13). We previously reported that BC325 (14),the 27-desmethoxy analog derived from 10, is a natural minormetabolite produced by BIOT-4010 and that the fully processedequivalent 13 could not be observed even when exogenous 12was fed to boost productivity (Andexer et al., 2011). In contrast,addition of exogenous 12 to growing cultures of BIOT-4110 gave a

low but significant level of 13. To isolate 13 we grew BIOT-4110 intwo bioreactors and fed 12 (2 mM after 24 h). Upon harvest andwork up we observed final titres of 13 (2 mg/L) and 14 (30 mg/L).The resulting extract contained 47 mg of 13 and 755 mg of 14(0.94 mmol total rapalogs; 1.5% bioconversion based on 60 mmolof 12 fed). After purification, 13 (4.6 mg; 495% purity) wasobtained and the structure was confirmed by standard spectro-scopic methods.

4. Conclusions

Traditional techniques of strain mutagenesis coupled withmedia and fermentation optimisation are well established forimproving the productivity of microorganisms responsible for thebiosynthesis of natural products by fermentation. However,genome sequencing, systems biology and a growing body ofgenomics-based experimental data are now influencing currentand future approaches in this area (Baltz et al., 2011; Wu et al.,2012). For example, techniques such as genome shuffling mayhave benefits in terms of speed and strain stability (Chen et al.,2009; Xu et al., 2008; Zhang et al., 2002). Importantly, forbiosynthetic engineering it has been established that specificbiosynthetic pathway mutations made (and tested) in a researchstrain can be transferred successfully to a commercially relevantstrain optimised for productivity (Baltz, 2011; Li et al., 2009;Rodriguez et al., 2003). This allows biosynthetic alterations to betested in more tractable strains and, critically, in parallel to theprocess of strain and process optimisation required to establishcommercial levels of productivity.

We demonstrate here the effective combination of traditionaland rational genetic approaches in order to give recombinantstrains capable of producing multi-gram quantities of new rapa-logs using biosynthetic engineering. Our results clearly show thatbioengineered polyketides can be produced at excellent titres,often equivalent to that of the parent compound. Based on thesuccess of our experiments BIOT-4010 and BIOT-4110 are routi-nely used for all of our biosynthetic engineering work on the 1pathway and have potential as general tools to enable

S.G. Kendrew et al. / Metabolic Engineering 15 (2013) 167–173 173

biosynthetic medicinal chemistry approaches for natural productlead optimization.

Appendix A. Supporting information

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.ymben.2012.11.001.

References

Alexander, D.C., Rock, J., He, X., Brian, P., Miao, V., Baltz, R.H., 2010. Development ofgenetic system for lipopeptide combinatorial biosynthesis in Streptomycesfradiae and heterologous expression of the A54145 biosynthetic gene cluster.Appl. Environ. Microbiol. 76, 6877–6887.

Andexer, J.N., Kendrew, S.G., Nur-E-Alam, M., Lazos, O., Foster, T.A., Zimmermann,A.S., Warneck, T.D., Suthar, D., Coates, N.J., Koehn, F.E., Skotnicki, J.S., Carter,G.T., Gregory, M.A., Martin, C.J., Moss, S.J., Leadlay, P.F., Wilkinson, B., 2011.Biosynthesis of the immunosuppresants FK506, FK520, and rapamycininvolves a previously undescribed family of enzymes acting on chorismate.Proc. Natl. Acad. Sci. U.S.A. 108, 4776–4781.

Baltz, R.H., 2011. Strain improvement in the postgenomic era. J. Ind. Microbiol.Biotechnol. 38, 657–666.

Bierman, M., Logan, R., O’Brien, K., Seno, E.T., Nagaraja Rao, R., Schoner, B.E., 1992.Plasmid cloning vectors for the conjugal transfer of DNA from Escherichia colito Streptomyces spp. Gene 116, 43–49.

Chen, X., Wei, P., Fan, L., Yang, D., Zhu, X., Shen, W., Wu, Z., Cen, P., 2009.Generation of high-yield rapamycin-producing strains through protoplasts-related techniques. Appl. Microbiol. Biotechnol. 83, 507–512.

Flett, F., Mersinias, V., Smith, C.P., 1997. High efficiency intergeneric conjugaltransfer of plasmid DNA from Escherichia coli to methyl DNA-restrictingstreptomycetes. FEMS Microbiol. Lett. 155, 223–229.

Graziani, E.I., 2009. Recent advances in the chemistry, biosynthesis and pharma-cology of rapamycin analogs. Nat. Prod. Rev. 26, 602–609.

Gregory, M.A., Till, R., Smith, M.C., 2003. Integration site for Streptomyces phagephiBT1 and development of site-specific integrating vectors. J. Bacteriol. 185,5320–5323.

Gregory, M.A., Gaisser, S., Lill, R.E., Hong, H., Sheridan, R.M., Wilkinson, B., Petkovic,H., Weston, A.J., Carletti, I., Lee, H.L., Staunton, J., Leadlay, P.F., 2004a. Isolationand characterization of prerapamycin, the first macrocyclic intermediate inthe biosynthesis of the immunosuppressant rapamycin by S. hygroscopicus.Angew. Chem. Int. Ed. Engl. 43, 2551–2553.

Gregory, M.A., Gaisser, S., Petkovic, H., Moss, S.J., 2004b. Genetic engineering ofStreptomyces hygroscopicus and other microbes for production of poly-ketides and other natural products (Biotica Technology Ltd). PCT Int. Appl.WO2004007709.

Gregory, M.A., Petkovic, H., Lill, R.E., Moss, S.J., Wilkinson, B., Gaisser, S., Leadlay,P.F., Sheridan, R.M., 2005. Mutasynthesis of rapamycin analogues through themanipulation of a gene governing starter unit biosynthesis. Angew. Chem. Int.Ed. Engl. 44, 4757–4760.

Gregory, M.A., Hong, H., Lill, R.E., Gaisser, S., Petkovic, H., Low, L., Sheehan, L.S.,Carletti, I., Ready, S.J., Ward, M.J., Kaja, A.L., Weston, A.J., Challis, I.R., Leadlay,P.F., Martin, C.J., Wilkinson, B., Sheridan, R.M., 2006. Rapamycin biosynthesis:elucidation of gene product function. Org. Biomol. Chem. 4, 3565–3568.

Goss, R.J., Lanceron, S.E., Wise, N.J., Moss, S.J., 2006. Generating rapamycinanalogues by directed biosynthesis: starter acid substrate specificity ofmono-substituted cyclohexanecarboxylic acids. Org. Biomol. Chem. 4,4071–4073.

Goss, R.J., Lanceron, S., Deb Roy, A., Sprague, S., Nur-E-Alam, M., Hughes, D.L.,Wilkinson, B., Moss, S.J., 2010. An expeditious route to fluorinated rapamycinanalogues by utilising mutasynthesis. ChemBioChem. 11, 698–702.

Hodgson, D.A., 1982. Glucose repression of carbon source uptake and metabolismin Streptomyces coelicolor A3(2) and its perturbation in mutants resistant to 2-deoxyglucose. J. Gen. Microbiol. 128, 2417–2430.

Kumar, Y., Goodfellow, M., 2008. Five new members of the Streptomyces viola-ceusniger 16S rRNA gene clade: Streptomyces castelarensis sp. nov., comb. nov.,Streptomyces himastatinicus sp. nov., Streptomyces mordarskii sp. nov., Strepto-myces rapamycinicus sp. nov. and Streptomyces ruanii sp. nov. Int. J. Syst. Evol.Microbiol. 58, 1369–1378.

Laplante, M., Sabatini, D.M., 2012. mTOR signalling in growth control and disease.Cell 149, 274–293.

Li, C., Hazzard, C., Florova, G., Reynolds, K.A., 2009. High titer expression oftetracenomycins by heterologous expression of the pathway in a Streptomyces

cinnamonensis industrial monensin producer strain. Metab. Eng. 11, 319–327.Lowden, P.A.S., Bohm, G.A., Metcalfe, S., Staunton, J., Leadlay, P.F., 2004. New

rapamycins by precursor-directed biosynthesis. ChemBioChem. 5, 535–538.Marsden, A.F.A., Wilkinson, B., Cortes, J., Dunster, N.J., Staunton, J., Leadlay, P.F.,

1998. Engineering broader specificity into an antibiotic-producing polyketidesynthase. Science 279, 199–202.

Menzella, H.G., Tran, T.T., Carney, J.R., Lau-Wee, J., Galazzo, J., Reeves, C.D.,Carreras, C., Mukadam, S., Eng, S., Zhong, Z., Timmermans, P.B., Murli, S.,Ashley, G.W., 2009. Potent non-benzoquinone ansamycin heat shock protein90 inhibitors from genetic engineering of Streptomyces hygroscopicus. J. Med.Chem. 52, 1518–1521.

Moss, S.J., Gregory, M.A., Zhang, M.-Q., 2011. (Biotica Technology Ltd). PCT Int.Appl. WO2011045594.

Ochi, K., Okamoto, S., Tozawa, Y., Inaoka, T., Hosaka, T., Xu, J., Kurosawa, K., 2004.Ribosome engineering and secondary metabolite production. Adv. Appl.Microbiol. 56, 155–184.

Rodriguez, E., Hu, Z., Ou, S., Volchegursky, Y., Hutchinson, C.R., McDaniel, R., 2003.Rapid engineering of polyketide overproduction by gene transfer to indust-rially optimized strains. Microbiol. Biotechnol. 30, 480–488.

Ruan, B., Pong, K., Jow, F., Bowlby, M., Crozier, R.A., Liu, D., Liang, S., Chen, Y.,Mercado, M.L., Feng, X., Bennett, F., von Schack, D., McDonald, L., Zaleska, M.M.,Wood, A., Reinhart, P.H., Magolda, R.L., Skotnicki, J., Pangalos, M.N., Koehn, F.E.,Carter, G.C., Abou-Gharbia, M., Graziani, E.I., 2008. Binding of rapamycinanalogs to calcium channels and FKBP52 contributes to their neuroprotectiveactivities. Proc. Natl. Acad. Sci. U.S.A. 105, 33–38.

Schwecke, T., Aparicio, J.F., Molnar, I., Konig, A., Khaw, L.E., Haydock, S.F., Oliynyk,M., Caffrey, P., Cortes, J., Lester, J.B., Bohm, G.A., Staunton, J., Leadlay, P.F., 1995.The biosynthetic gene cluster for the polyketide immunosuppressant rapa-mycin. Proc. Nat. Acad. Sci. U.S.A. 92, 7839–7843.

Sehgal, S.N., Blazekovic, T.M., Vezina C., 1976. Rapamycin and Process of Prepara-tion (Ayerst). US Patent US3,993,749.

Sheehan, L.S., Lill, R.E., Wilkinson, B., Sheridan, R.M., Vousden, W.A., Kaja, A.L.,Crouse, G.D., Gifford, J., Graupner, P.R., Karr, L., Lewer, P., Sparks, T.C., Leadlay,P.F., Waldron, C., Martin, C.J., 2006. Engineering of the spinosyn PKS: directingstarter unit incorporation. J. Nat. Prod. 69, 1702–1710.

Shelley, P., Banks, R.R.M., 1992. 3-Demethylrapamycin or derivatives thereof,processes for their preparation and their use as antifungal agents andimmunosuppressants (SmithKline Beecham PLC). PCT Intl. Appl, WO9214737.

Shima, J., Hesketh, A., Okamoto, S., Kawamoto, S., Ochi, K., 1996. Induction ofactinorhodin production by rpsL (encoding ribosomal protein S12) mutationsthat confer resistance in Streptomyces lividans and Streptomyces coelicolor

A3(2). J. Bacteriol 178, 7276–7284.Walsh, C.T., Fischbach, M.A., 2010. Natural products version 2.0: connecting genes

to molecules. J. Am. Chem. Soc. 132, 2469–2493.Wang, X., Proud, C.G., 2011. mTORC1signalling: what we still don’t know. J. Mol.

Cell. Biol. 3, 206–220.Wang, G., Hosaka, T., Ochi, K., 2008. Dramatic activation of antibiotic production in

Streptomyces coelicolor by cumulative drug resistance mutations. Appl.Environ. Microbiol 74, 2834–2840.

Wu, M.-C., Law, B., Wilkinson, B., Micklefield, J., 2012. Bioengineering naturalproduct biosynthetic pathways for therapeutic applications. Curr. Opin.Biotechnol 23, 931–940.

Wilkinson, B., Micklefield, J., 2007. Mining and engineering natural-productbiosynthetic pathways. Nat. Chem. Biol. 3, 379–386.

Wong, F.T, Khosla, C., 2012. Combinatorial biosynthesis of polyketides – aperspective. Curr. Opin. Chem. Biol 16, 117–123.

Xu, B., Jin, Z., Wang, H., Jin, Q., Jin, X., Cen, P., 2008. Evolution of Streptomyces

pristinaespiralis for resistance and production of pristinamycin by genomeshuffling. Appl. Microbiol. Biotechnol. 80, 261–267.

Zhang, Y.-X., Perry, K., Vinci, V.A., Powell, K., Stemmer, W.P., del Cardayre, S.B.,2002. Genome shuffling leads to rapid phenotypic improvement in bacteria.Nature 415, 644–646.

Zhang, M.Q., Gaisser, S., Nur-E-Alam, M., Sheehan, L.S., Vousden, W.A., Gaitatzis, N.,Peck, G., Coates, N.J., Moss, S.J., Radzom, M., Foster, T.A., Sheridan, R.M.,Gregory, M.A., Roe, S.M., Prodromou, C., Pearl, L., Boyd, S.M., Wilkinson, B.,Martin, C.J., 2008. Optimizing natural products by biosynthetic engineering:discovery of nonquinone Hsp90 inhibitors. J. Med. Chem. 51, 5494–5497.