Precursor-directed polyketide biosynthesis in Escherichia coli

Article abstract of DOI:10.1016/j.bmcl.2003.08.008

Precursor-directed polyketide biosynthesis was demonstrated in the heterologous host Escherichia coli. Several diketide and triketide substrates were fed to a recombinant E. coli strain containing a variant form of deoxyerythronolide B synthase (DEBS) from which the first elongation module was deleted resulting in successful macrolactone formation from the diketide, but not the triketide, substrates.

Full text of DOI:10.1016/j.bmcl.2003.08.008

Bioorganic & Medicinal Chemistry Letters 13 (2003) 3701–3704
Precursor-Directed Polyketide Biosynthesis in Escherichia coli
Kenji Kinoshita,a Blaine A. Pfeifer,b Chaitan Khoslab and David E. Canea,*
aDepartment of Chemistry, Box H, Brown University, Providence, RI 02912-9108, USA
bDepartments of Chemical Engineering, Chemistry, and Biochemistry, Stanford University, Stanford, CA 94305-5025, USA
Received 23 June 2003; revised 6 August 2003; accepted 7 August 2003
Abstract—Precursor-directed polyketide biosynthesis was demonstrated in the heterologous host Escherichia coli. Several diketide
and triketide substrates were fed to a recombinant E. coli strain containing a variant form of deoxyerythronolide B synthase
(DEBS) from which the first elongation module was deleted resulting in successful macrolactone formation from the diketide, but
not the triketide, substrates.
# 2003 Elsevier Ltd. All rights reserved.
Polyketides are a group of natural products that possess
a wide range of biological and medicinal properties.
Examples include the antibiotic erythromycin, the anti-
tumor agent epothilone, and the immunosuppressant
FK506. Large modular megasynthases called polyketide
synthases (PKSs) biosynthesize their polyketide products
using simple building blocks such as acetyl coenzyme A
(acetyl-CoA), propionyl-CoA, malonyl-CoA, and
methylmalonyl-CoA.1 The most extensively character-
ized modular polyketide synthase is the 6-deoxyery-
thronolide B synthase (DEBS) found within the soil
bacterium Saccharopolyspora erythraea.2,3 DEBS, an
enzymatic complex containing three individual proteins
(DEBS1, DEBS2, and DEBS3), catalyzes the formation
of the erythromycin macrolide aglycone, 6-deoxyery-
thronolide B (6-dEB, 1), from a propionyl-CoA primer
and six methylmalonyl-CoA extender units (Fig. 1A).
Each of the six rounds of polyketide chain condensation
and functional group modification is mediated by a
dedicated group of active sites organized into a func-
tional unit known as a module (with two modules loca-
ted on each of the DEBS proteins). Each
multifunctional protein module is in turn made up of a
set of catalytic domains of 100–400 amino acids each
that are similar in both function and sequence to indi-
vidual enzymes of fatty acid biosynthesis.
other polyketides.4,5 This technique is based on the use
of a mutant of a modular PKS in which one or more of
the early-stage enzyme activities has been eliminated by
inactivation or deletion. The resultant mutant is thus
blocked in the biosynthesis of the natural polyketide
product. For example, DEBS(KS10) is a mutant of
DEBS that has been inactivated by site-directed muta-
genesis of the b-ketoacyl-ACP synthase domain of mod-
ule 1 (KS1). Because the KS1 domain would normally
catalyze the first condensation step of 6-deoxyery-
thronolide B biosynthesis, the DEBS(KS10) mutant is
incapable of carrying out the first round of polyketide
chain elongation from the natural propionyl-CoA star-
ter and methylmalonyl-CoA chain-elongation substrate
and is therefore unable to catalyze the formation of the
macrolide 6-dEB (1). Formation of the natural poly-
ketide can be restored, however, by introduction of a
synthetic derivative of the natural diketide intermediate
as the N-acetyl cysteamine (SNAC) thioester 2 (Fig.
1B). We have described precursor-directed biosynthesis
experiments in which 2 and various diketide and trike-
tide analogues were tested as substrates for Strepto-
myces coelicolor CH999/pJRJ2, an engineered strain
harboring DEBS(KS10).4,6À8
Notwithstanding earlier success with precursor-directed
formation of macrolide analogues, the manipulation of
modular PKSs is often hampered by the biological con-
straints posed by organisms (principally actinomycetes)
in which complex polyketides have thus far been pro-
duced. To address this problem, an Escherichia coli host
was recently engineered to produce complex polyke-
tides.9 Multiple changes were made to the E. coli genome
Precursor-directed biosynthesis is a promising approach
to generating novel analogues of erythromycin and
*Corresponding author. Tel.: +1-401-863-3588; fax: +1-401-863-
3556; e-mail: david_cane@brown.edu
0960-894X/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2003.08.008

3702
K. Kinoshita et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3701–3704
Figure 1. Modular organization and function of the DEBS and its intermediates. (A) Each module contains three core catalytic domains, a keto-
synthase (KS), acyltransferase (AT), and acyl carrier protein (ACP), and a variable number of auxiliary domains, ketoreductase (KR), dehydratase
(DH), and enoylreductase (ER). In addition to the six modules, there is a loading didomain to prime module 1 and a thioesterase (TE) domain at the
C-terminus of module 6 that cyclizes the heptaketide and generates 6-dEB (1). (B) Formation of 6-dEB by DEBS(KS1O) is blocked but can be restored
by addition of the diketide-SNAC 2. (C) Formation of 6-dEB by DEBS(module1Á) is blocked but can be restored by addition of the diketide-SNAC 2.
to support 6dEB production, including introduction of
the three DEBS genes from Sac. erythraea, the sfp
phosphopantetheinyl transferase gene from Bacillus
subtilis,10 and the genes encoding a heterodimeric pro-
pionyl-CoA carboxylase (pcc) from S. coelicolor.11
When gene expression was coordinately induced, pro-
pionate could be converted into 6dEB by this metaboli-
cally engineered cellular catalyst. Indeed, when
cultivated at high cell densities, this derivative of E. coli
BAP1 harboring DEBS1, À2, and –3 could achieve
titers of 6-dEB in excess of 200 mg/L.12
T7 RNA polymerase promoter.13 To investigate pre-
cursor-directed biosynthesis using E. coli BAP1/
pBP175/pBP130, LB medium cultures were grown at
37 ꢁC to mid-log phase, then induced with 1 mM IPTG
and supplemented with 1 mM synthetic natural dike-
tide-SNAC thioester analogue 2 and 1 mM [1-14C]pro-
pionate at 20 ꢁC (Fig. 1C, Scheme 1). After an
additional 24 h incubation, the culture supernatants
were extracted with ethyl acetate and analyzed for the
formation of the expected macrolactone 6-dEB (1) by
analytical TLC and phosphoimaging (Fig. 2).14
To evaluate further the potential for manipulating
polyketide biosynthesis in E. coli, we have engineered an
E. coli BAP1 strain in which DEBS module 2 is used in
place of DEBS1(KS10), taking advantage of the
enhanced expression of the smaller protein (subunit
The identity of the 6-dEB sample was confirmed by
carrying out a preparative-scale incubation in the pre-
sence of 2 and [1-13C]propionate. Analysis of the resul-
tant labeled 6-dEB (1) by 13C NMR revealed the
expected pattern of 5 enhanced peaks, corresponding to
enrichment of C1, C3, C5, C7, and C9.15,16 The 1H
NMR spectrum of [1,3,5,7,9-13C5]-1 was essentially
identical to that of unlabeled 1 except for the expected
MW ꢀ150kDa) in
E. coli. E. coli BAP1
DEBS(module1Á) harbors DEBS module 2 plus pcc on
plasmid pBP175 (carrying kanamycin resistance) and
DEBS2 and DEBS3 on plasmid pBP130(carbenicillin
resistance), with gene expression controlled by the
IPTG- (isopropyl-b-d-thiogalactopyranoside) inducible
1
additional H–13C couplings for the H-3, H-5, and H-7
signals. The time-course of 6-dEB formation was also
monitored by TLC/phosphoimaging, with cultures

K. Kinoshita et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3701–3704
3703
Figure 3. Time course of 6-dEB (1) production by administration of 2
to E. coli BAP1/pBP175/pBP130.
substitution pattern and stereochemistry at C-2 and C-3
as the preferred syn-diketide substrate 2, but a chain
length more like that of a triketide, with a (4R)-methyl
substituent but no C-5 hydroxyl group. Indeed, we have
previously found that 3 is an excellent surrogate sub-
strate for DEBS module 2, with a kcat/Km 5.5Â that of
diketide 2, and a Km that is 9-fold lower.17 (By contrast,
the (4S)-diastereomer of 3, has a kcat/Km that is 0.8Â
that of 2.)17 Administration of branched-chain diketide
3 to E. coli BAP1/pBP175/pBP130in the presence of
[1-14C]propionate led to formation of the 14C-labeled
6-dEB analogue 4, (Scheme 1), with a relative pro-
ductivity compared to diketide 2 of 32% (ꢀ1.5 mg/L),
as determined by TLC/phosphoimaging (Fig. 2). The
structure of macrolactone 4 was confirmed by a pre-
parative scale incubation that for experimental con-
Scheme 1. Precursor-directed biosynthesis of 6-dEB (1) and analogues
by incubation of diketide and triketide analogues with E. coli BAP1/
pBP175/pBP130[DEBS(module1 Á)].
venience utilized 5,
a
3:2 (4R/4S) mixture of
diastereomers of diketide 3, in the presence of
[1-13C]propionate, giving rise exclusively to labeled 4
which exhibited 5 enhanced 13C NMR peaks (C1, C3,
C5, C7, and C9). Labeled 4 was analyzed by 1D NMR
(1H and 13C NMR) and 2D NMR [13C–1H COSY and
Heteronuclear Single Quantum Correlation (HSQC)] as
well as HRFAB-MS and proved to have the same
properties as authentic 4 previously obtained by feeding
either diketide 3 or the diastereomeric diketide mixture
5 to S. coelicolor CH999/JRJ2.16,17 (The formation of a
single diastereomer of 4 is due to exclusive utilization by
DEBS module 2 of the preferred diketide diastereomer
3).17 By contrast with the efficient utilization of dike-
tides 2 and 3, neither unsaturated triketide 6 nor 7 was
converted by E. coli BAP1 DEBS(module1Á) to the
expected macrolactones 8 and 9, respectively (Fig. 2,
Scheme 1), the products that were formed when these
anomalous substrates had been administered to the cul-
tures of S. coelicolor CH999/pJRJ2 harboring
DEBS1(KS10).4,6,7 Addition of b-oxidation inhibitors
such as 4-pentynoic acid or 3-(tetradecylthio)propionic
acid or other supplements such as 2,6-O-dimethyl-b-
cyclodextrin known to improve incorporation of poly-
ketide-SNAC derivatives failed to stimulate macrolide
analogue production from either unsaturated triketide 6
or 7.
Figure 2. TLC/phosphoimaging to measure relative productivity of
macrolide formation by incubation of diketide and triketide analogues
plus [1-14C]propionate with E. coli BAP1/pBP175/pBP130. Lane A:
authentic samples of macrolides 1, 4, and 8, lane B: negative control,
lane C: 2, lane D: 3, lane E: 6, lane F: 7. Solvent system: 50% EtOAc/
hexanes.
achieving a titer of up to 11.9 nmol/mL (ꢀ5 mg/L) of 1
within 72 h of IPTG induction and administration of
diketide 2 (Fig. 3). These yields of macrolactone are
within the titer range previously attained under similar
culture conditions using the E. coli BAP1 strain har-
boring the complete set of DEBS genes.9,12
Having established that E. coli BAP1 DEBS(module1Á)
would support precursor-directed biosynthesis from the
natural diketide 2, we examined the ability of this sys-
tem to tolerate variations in precursor structure
(Scheme 1). The branched-chain diketide 3 has the same
In summary, we have demonstrated the feasibility of
precursor-directed biosynthesis of 6-dEB 1 and its

3704
K. Kinoshita et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3701–3704
macrolactone analogue 4 by feeding diketides 2 and 3 to
an engineered E. coli strain harboring DEBS modules 2
through 6. Thus E. coli BAP1/pBP175/pBP130lacking
DEBS module 1 and its attached loading domain can be
used with the diketides examined to give the same mac-
rolactones as were produced by feedings with S. coeli-
color CH999/pJRJ2 harboring DEBS(KS10). The
unsaturated triketides 6 and 7, however, were not con-
verted to the expected macrolactone products by the
engineered E. coli strain, possibly reflecting uptake and/
or metabolic differences between E. coli and S. coelico-
lor. Precursor-directed biosynthesis using S. coelicolor
CH999/pJRJ2 has already proved to be powerful tool
for rational synthesis of structural analogues of ery-
thromycin and related macrolides. We believe that
E. coli may offer many biological advantages over
S. coelicolor, including metabolic rate, scale-up poten-
tial, and ease of metabolic engineering.
12. Pfeifer, B.; Hu, Z.; Licari, P.; Khosla, C. Appl. Environ.
Microbiol. 2002, 68, 3287.
13. The plasmid encoding the DEBS2 and DEBS3 genes
(pBP130) has been reported earlier.9 Plasmid BP175 (con-
taining module 2 and the pcc) is a derivative of pBP144
(containing DEBS1 and the pcc9) with module 2 replacing
DEBS1. The construction of module 2 has been described
previously: Tsuji, S.; Cane, D. E.; Khosla, C. Biochemistry
2001, 40, 2326.
14. Small scale feeding using [1-14C]propionate was performed
as follows: an E. coli BAP1/pBP175/pBP130transformant was
inoculated into 1 mL LB medium in the presence of carbeni-
cillin (100 mg/mL) and kanamycin (50 mg/mL) at 37 ꢁC and
250rpm. Cultures were grown to mid-log phase (OD 600=0.6),
cooled at 20 ꢁC for 15 min, and then centrifuged. The cell pel-
lets were resuspended in 1 mL of fresh LB media and induced
with 1 mM IPTG. In addition, [1-14C]propionate (56 mCi/
mmol) and either diketides 2 or 3 or triketide analogues 6 and
7 were each administered at final concentrations of 1 mM. The
culture was then shaken for an additional 24 h at 20 ꢁC. At
this point the culture was centrifuged and 1 mL of the super-
natant was extracted with 2 mL of ethyl acetate. The extract
was dried in vacuo and subjected to thin-layer chromato-
graphy (TLC) analysis. Negative controls included cultures
without substrate.
Acknowledgements
This work was supported by grants from the NIH to
D.E.C. (GM22172) and to C.K. (GM67937).
15. Large-scale feeding using [1-13C]propionate was per-
formed as follows: A transformant of E. coli BAP1 DEBS/
pBP175/pBP130was used to start a 3-mL LB culture with
carbenicillin (100 mg/mL) and kanamycin (50 mg/mL) at
37 ꢁC and 250rpm. Cultures were grown to mid-log phase
(OD600=0.7) and then centrifuged at 3000 rpm. The cell pel-
lets were resuspended in 3 mL of fresh LB media. This culture
was used to inoculate 200 mL LB media with the same anti-
biotic concentrations as above. These cultures were grown at
250rpm and 37 ꢁC to mid-log phase (OD600=0.5–0.7), cooled
for 15 min in a 20 ꢁC bath, and induced with 1 mM IPTG. 1
mM [1-13C]propionate and 0.5 mM substrate 2 or 5 were
added at the same time, and the cultures were incubated at
20 ꢁC for 48 h. The sample was then centrifuged and the super-
natant extracted twice with 100 mL ethyl acetate (2Â100 mL),
and the derived macrolide products, 1 and 4, respectively, were
purified by SiO2 gel column chromatography (ca. 1 g) from the
concentrated extract. The products obtained were then ana-
lyzed by 1D (1H and 13C) and 2D NMR and by high-resolution
mass spectrometry to confirm their individual structures.
16. 1: 13C NMR (CDCl3, 100 MHz) d 213.42 (C9), 178.36
(C1), 79.53 (C3), 76.52 (C5), 37.68 (C7) (enriched carbon
atoms only); HRFAB-MS ([M+Na]+, NBA/NaI): Calcd for
References and Notes
1. O’Hagan, D. O. The Polyketide Metabolites; Ellis Hor-
wood: Chichester, UK, 1991.
2. Donadio, S.; Staver, M. J.; Mcalpine, J. B.; Swanson, S. J.;
Katz, L. Science 1991, 252, 675.
3. Cortes, J.; Haydock, S. F.; Roberts, G. A.; Bevitt, D. J.;
Leadlay, P. F. Nature 1990, 348, 176.
4. Jacobsen, J. R.; Hutchinson, C. R.; Cane, D. E.; Khosla, C.
Science 1997, 277, 367.
5. Weissman, K. J.; Bycroft, M.; Cutter, A. L.; Hanefeld,
U.; Frost, E. J.; Timoney, M. C.; Harris, R.; Handa, S.;
Roddis, M.; Staunton, J.; Leadlay, P. F. Chem. Biol. 1998,
5, 743.
6. Jacobsen, J. R.; Cane, D. E.; Khosla, C. J. Am. Chem. Soc.
1998, 120, 9096.
7. Kinoshita, K.; Williard, P. G.; Khosla, C.; Cane, D. E. J.
Am. Chem. Soc. 2001, 123, 2495.
8. Cane, D. E.; Kudo, F.; Kinoshita, K.; Khosla, C. Chem.
Biol. 2002, 9, 131.
13
(12C16 C5H38O6)Na+: 414.2734, Found: 414.2754. 4: 13C
9. Pfeifer, B. A.; Admiraal, S. J.; Gramajo, H.; Cane, D. E.;
Khosla, C. Science 2001, 291, 1790.
10. Lambalot, R. H.; Gehring, A. M.; Flugel, R. S.; Zuber, P.;
LaCelle, M.; Marahiel, M. A.; Reid, R.; Khosla, C.; Walsh,
C. T. Chem. Biol. 1996, 3, 923.
NMR (CDCl3, 400 MHz) d 213.29 (C9), 178.28 (C1), 79.56
(C3), 76.46 (C5), 37.79 (C7) (enriched carbon atoms only);
HRFAB-MS
13
([M+Na]+,
NBA/NaI):
Calcd
for
12
(
18C5H42O6)Na+: 442.3095, Found: 442.3100.
17. Kinoshita, K.; Khosla, C.; Cane, D. E. Helv. Chim. Acta
Submitted for publication.
11. Rodriguez, E.; Gramajo, H. Microbiology 1999, 145, 3109.