Employing modular polyketide synthase ketoreductases as biocatalysts in the preparative chemoenzymatic syntheses of diketide chiral building blocks

Article abstract of DOI:10.1016/j.chembiol.2011.07.021

Chiral building blocks are valuable intermediates in the syntheses of natural products and pharmaceuticals. A scalable chemoenzymatic route to chiral diketides has been developed that includes the general synthesis of α-substituted, β-ketoacyl N-acetylcysteamine thioesters followed by a biocatalytic cycle in which a glucose-fueled NADPH-regeneration system drives reductions catalyzed by isolated modular polyketide synthase (PKS) ketoreductases (KRs). To identify KRs that operate as active, stereospecific biocatalysts, 11 isolated KRs were incubated with 5 diketides and their products were analyzed by chiral chromatography. KRs that naturally reduce small polyketide intermediates were the most active and stereospecific toward the panel of diketides. Several biocatalytic reactions were scaled up to yield more than 100 mg of product. These syntheses demonstrate the ability of PKS enzymes to economically and greenly generate diverse chiral building blocks on a preparative scale.

Full text of DOI:10.1016/j.chembiol.2011.07.021

Chemistry & Biology
Article
EmployingModular Polyketide Synthase Ketoreductases
as Biocatalysts in the Preparative Chemoenzymatic
Syntheses of Diketide Chiral Building Blocks
Shawn K. Piasecki,1 Clint A. Taylor,2 Joshua F. Detelich,2 June Liu,2 Jianting Zheng,2 Arkady Komsoukaniants,2
Dionicio R. Siegel,1,2 and Adrian T. Keatinge-Clay1,2,
1Institute for Cellular and Molecular Biology
*
2Department of Chemistry and Biochemistry
The University of Texas at Austin, Austin, TX 78712, USA
*Correspondence: adriankc@mail.utexas.edu
DOI 10.1016/j.chembiol.2011.07.021
SUMMARY
the syntheses of complex polyketides because the required
synthetic fragments often harbor multiple stereocenters and
Chiral building blocks are valuable intermediates in are challenging to prepare. The complexity of synthesizing
reduced polyketides can be appreciated by examining the
synthetic routes established for 6-deoxyerythronolide B (6-dEB),
the syntheses of natural products and pharmaceuti-
cals. A scalable chemoenzymatic route to chiral dike-
the precursor of erythromycin antibiotics (Masamune et al.,
1981; Myles et al., 1990; Evans et al., 1998; Crimmins and Slade,
2006; Stang and White, 2009). Although several groups over the
tides has been developed that includes the general
synthesis of a-substituted, b-ketoacyl N-acetylcys-
teamine thioesters followed by a biocatalytic cycle
in which a glucose-fueled NADPH-regeneration sys-
tem drives reductions catalyzed by isolated modular
polyketide synthase (PKS) ketoreductases (KRs). To
last 30 years have worked on the synthesis of this archetypal
macrolide core, the shortest is 22 steps with an overall yield of
7.8% (Stang and White, 2009).
The soil bacterium Saccharopolyspora erythraea synthesizes
identify KRs that operate as active, stereospecific
biocatalysts, 11 isolated KRs were incubated with 5
6-dEB with DEBS (6-dEB synthase), a PKS that contains diverse
catalytic activities: acyltransferases (ATs) catalyze the trans-
diketides and their products were analyzed by chiral esterification of small organic extender units from acyl-CoAs
onto the PKS, ketosynthases (KSs) catalyze decarboxylative
Claisen condensations that create carbon-carbon bonds
between these extender units and the growing polyketide chain,
chromatography. KRs that naturally reduce small
polyketide intermediates were the most active and
stereospecific toward the panel of diketides. Several
biocatalytic reactions were scaled up to yield more
than 100 mg of product. These syntheses demon-
strate the ability of PKS enzymes to economically
and greenly generate diverse chiral building blocks
on a preparative scale.
ketoreductases (KRs) catalyze stereospecific, NADPH-coupled
reductions of b-keto groups formed through condensation, a
dehydratase (DH) catalyzes an elimination reaction to yield an
olefin, and an enoylreductase (ER) catalyzes the stereospecific,
NADPH-coupled reduction of that olefin (Figure 1A) (Khosla
et al., 2007). Growing polyketides are passed along the PKS
assembly line through trans-thioesterification reactions between
acyl carrier proteins (ACPs) and KSs to a thioesterase (TE) that
catalyzes the cyclization of a heptaketide to yield 6-dEB. Thus,
enzymes in the PKS cooperate to transform seven molecules
INTRODUCTION
Modular polyketide synthases (PKSs) are enzymatic assembly of propionate into one molecule of 6-dEB under ambient,
lines that control the substitution and stereochemistry of aqueous conditions. As the current syntheses of 6-dEB and
complex polyketides in the biosyntheses of such compounds related natural products rely on organic solvents, chiral auxilia-
as the antibacterial erythromycin, the antifungal amphotericin, ries, and metals that can be both expensive and environmentally
and the immunosuppressant rapamycin (Staunton and Weiss- harmful, employing PKS enzymes in such syntheses would be
man, 2001; Khosla et al., 2007; Smith and Tsai, 2007; Sherman advantageous.
and Smith, 2006). Many synthetic organic chemists seek to
Aldol reactions employed in organic syntheses to form
emulate the reactions they catalyze and obtain the products carbon-carbon bonds and set stereocenters in the same reac-
they produce; however, if PKS enzymes, themselves, were har- tion sometimes lack the desired level of stereocontrol. The
nessed as biocatalysts, coveted chiral building blocks and bio- equivalent process in PKSs is carefully regulated by two
logically active molecules would be more readily accessed.
enzymes: KSs that mediate carbon-carbon bond formation
Libraries of chiral building blocks are highly desired by and KRs that guide the setting of stereocenters. The combina-
synthetic organic chemists to accelerate the synthesis of natural tion of stereocenters produced is dependent on the KR type.
products and pharmaceuticals (Hilterhaus and Liese, 2007; KRs that perform reductions yielding L-b-hydroxyl groups are
Patel, 2008). Such compounds would be particularly helpful in referred to as ‘‘A-type,’’ and KRs that perform reductions
Chemistry & Biology 18, 1331–1340, October 28, 2011 ª2011 Elsevier Ltd All rights reserved 1331

Chemistry & Biology
Biocatalysis with Modular PKS Ketoreductases
Figure 1. Employing PKS KRs to Chemoenzymatically Synthesize a Library of Chiral Building Blocks
(A) The DEBS assembly line synthesizes 6-dEB; 9/10 stereocenters are set by KRs that are classified by the combination of stereocenters they produce. LDD,
loading didomain.
(B) GDH helps drive the reduction of diketide-SNACs, regenerating NADPH through the oxidation of D-glucose.
(C) A general synthesis of a-substituted, b-keto diketides was developed. Stereospecific KRs can convert these common intermediates into chiral diketide-
SNACs.
(D) Representatives of the library of diketide chiral building blocks that can be accessed through KR biocatalytic reactions.
See also Figure S1 and Table S1.
yielding D-b-hydroxyl groups are referred to as ‘‘B-type’’ (reduc- Reductions of 3 mediated by EryKR1, EryKR2, EryKR5, EryKR6,
tase-incompetent KRs are referred to as ‘‘C-type’’) (Reid et al., MycKRA, MycKRB, and TylKR1 revealed that EryKR1 (B2-type)
2003; Caffrey, 2003; Keatinge-Clay, 2007; Valenzano et al., and TylKR1 (B1-type) were completely stereospecific, whereas
2009). KRs that reduce D-a-substituted intermediates are the other KRs produced a mixture of diastereomers (isolated
denoted with a ‘‘1,’’ whereas KRs that reduce L-a-substituted domains are referred to by their PKS and module of origin, except
intermediates are denoted with a ‘‘2’’ (Keatinge-Clay, 2007). the mycolactone KRs, which are identified by KR type) (Siskos
Thus, an a-substituted, b-keto diketide intermediate can be et al., 2005; Bali and Weissman, 2006). That from the isolated
reduced to the ‘‘2R,3S’’ product by an A1-type KR, the ‘‘2S,3S’’ KRs examined for their stereospecificity two KRs were identified
product by an A2-type KR, the ‘‘2R,3R’’ product by a B1-type that catalyze stereospecific reduction suggested that other
KR, or the ‘‘2S,3R’’ product by a B2-type KR (Keatinge-Clay KRs that perform stereocontrolled reductions isolated from
and Stroud, 2006; Keatinge-Clay, 2007; Zheng et al., 2010).
their PKSs could be discovered. If all KR types were available
Because KRs can exert control over two stereocenters in one as biocatalysts, then the full stereochemical diversity naturally
reduction reaction and, as a family, operate on diverse organic achieved by KRs could be accessed.
substrates, they possess extraordinary potential as biocatalysts
Thus, to demonstrate the utility of PKS enzymes in natural
(Patel, 2008). At least two studies of KRs as biocatalysts have product and pharmaceutical synthesis, we aimed to develop
been reported; however, the substrates examined were not KRs as biocatalysts and access a library of diketide chiral
very similar to those naturally encountered and the stereochem- building blocks (Figure 1D). Required were both a facile synthetic
ical purities of the reaction products were not investigated route to a-substituted, b-keto diketide thioester analogs of the
(Bali et al., 2006; Bali and Weissman, 2006). Analytical-scale polyketide intermediates naturally reduced by KRs as well as
enzymological studies have focused on the substrate analog a scalable, economic biocatalytic process that does not require
(2RS)-methyl-3-oxopentanoyl-S-N-acetylcysteamine (-SNAC), 3. stoichiometric NADPH for KR-mediated reduction.
1332 Chemistry & Biology 18, 1331–1340, October 28, 2011 ª2011 Elsevier Ltd All rights reserved

Chemistry & Biology
Biocatalysis with Modular PKS Ketoreductases
A general synthesis of a-substituted, b-ketoacyl N-acetylcys- methyl-3-oxohexanoyl-SNAC (4), and (2RS)-ethyl-3-oxopenta-
teamine thioesters was developed and used to generate diketide noyl-SNAC (5) (Figure 2). Compared to 3, each of the other
substrates 1–5. An NADPH-regeneration system in which KR- diketides is one methylene shorter or longer at the a or g position.
mediated reductions are driven by glucose was designed. Selection of KRs
Eleven isolated KRs were incubated with 1–5 in biocatalytic To identify KRs capable of stereospecifically reducing a range of
reactions and the products were analyzed by chiral chromatog- diketide substrates, 11 diverse reductase-competent KRs were
raphy. Robust A1-type and A2-type KR biocatalysts were selected (see ‘‘Cloning’’ in Experimental Procedures). They
identified. Several biocatalytic reactions were scaled up, both were isolated from modules that do not possess DHs or ERs
with purified KRs as well as KR-containing lysate, to readily because KRs that cooperate with these enzymes are usually
provide 100 mg yields of stereopure chiral building blocks. In B1-type (Valenzano et al., 2010). MycKRA (A-type), TylKR1
addition to illustrating that building blocks for the synthesis of (B1-type), and EryKR1 (B2-type) were chosen, in part, to serve
natural products and pharmaceuticals can be accessed by as positive controls, as how they process 3 has already been
PKS enzymes, the described biocatalytic systems provide PKS determined (Siskos et al., 2005; Bali and Weissman, 2006). The
enzymologists access to polyketide substrates and enable selection of the other eight KRs was biased toward A-type KRs
experiments on PKS enzymes to be performed on a more readily both because isolated A1- or A2-type KRs that stereospecifically
analyzable scale.
reduce b-ketoacyl-SNAC substrates had previously not been
identified and because A1-type KRs are the most common in
modules without DHs or ERs. All but two of the KRs selected
naturally operate on a-substituted intermediates because we
aimed to set two stereocenters through a single KR-mediated
reduction reaction. KRs were chosen from diverse PKSs
RESULTS AND DISCUSSION
Chemoenzymatic Design
NADPH-Regeneration Scheme
To drive KR-mediated reduction reactions, an NADPH-regener- (amphotericin, erythromycin, mycolactone, oleandomycin, pik-
ation scheme was employed (Figure 1B). Glucose dehydroge- romycin, spinosyn, and tylosin) and from varying locations within
nase (GDH) cloned from Bacillus subtilis was used to couple PKS assembly lines (see Figure S1 available online). Thus, the
the oxidation of glucose to the reduction of diketides. GDH selection of reductase-competent KRs includes two A-type
generates NADPH from NADP+ as it oxidizes glucose to gluco- KRs that naturally reduce a-unsubstituted intermediates, four
nolactone (Wong et al., 1985; Kataoka et al., 2003). A high A1-type KRs, two A2-type KRs, one B1-type KR, and two B2-
concentration of glucose establishes a large NADPH:NADP+ type KRs. A reductase-incompetent C1-type KR was also
ratio that drives the reduction reaction toward completion. At selected as a negative control (Figure S2).
the pH of the biocatalytic reactions (pH 7.5), most of the gluco-
nolactone produced is hydrolyzed to gluconic acid, which de- Chiral Chromatography and Establishing Elution Orders
protonates to gluconate. Because decreases in pH result in To determine the stereocontrol exerted by KRs when reducing 3,
GDH and KRs precipitating out of solution, excess buffering Leadlay and coworkers synthesized standards that established
strength is supplied.
the 3a–3d elution order from a chiral column (Holzbaur et al.,
1999). We sought to establish the elution order of 2a–2d
and 4a–4d on the same chiral column to identify the reduced
General Synthetic Route to a-Substituted, b-Keto
N-Acetylcysteamine Thioesters
The route most commonly employed to generate 3 requires diketides produced by the 11 KRs in the biocatalytic reactions.
a costly chiral auxiliary to generate an a-substituted, b-ketoimide Thus, well-established diastereospecific aldol reactions were
that is subsequently reacted with excess N-acetylcysteamine performed to obtain synthetic ‘‘2R,3R,’’ ‘‘2R,3S,’’ and ‘‘2S,3R’’
(even more expensive than the chiral auxiliary) (Gilbert diketide-SNACs and the elution orders of 2a–2d and 4a–4d
et al., 1995). Instead, we access a-substituted, b-keto N-acetyl- were determined (Evans et al., 1981; Raimundo and Heathcock,
cysteamine thioesters through an efficient and cost-effective 1995) (Figure 3). The products of 2, 3, and 4 elute in the
extension of a described route to b-keto N-acetylcysteamine same order—first the two anti products, then the two syn prod-
thioesters (Figure 1C) (Gilbert et al., 1995). After the desired ucts: ‘‘2R,3R’’ (B1-type product), ‘‘2S,3S’’ (A2-type product),
b-keto N-acetylcysteamine thioester has been obtained from ‘‘2R,3S’’ (A1-type product), and ‘‘2S,3R’’ (B2-type product)
Meldrum’s acid, an acyl chloride, and N-acetylcysteamine (Holzbaur et al., 1999).
(generated from cysteamine and acetic anhydride), an a substit-
Syntheses of reduced diketide-SNACs were initiated through
uent is installed via an alkylation reaction. Thus, the g substituent aldol reactions between (4S)-4-benzyl-3-propionyl-2-oxazolidi-
of a diketide-SNAC is selected through the acyl chloride, and the none and the appropriate aldehyde (Figure 3A). In keeping
a substituent is selected through the alkyl halide. That the with the classic Evans syn-selective reaction protocol, one
syntheses yield racemic mixtures of a-substituted, b-ketoacyl- equivalent of dibutylboron triflate was used to generate the
SNACs is a nonissue because the enantiomers rapidly intercon- ‘‘2S,3R’’ product (Evans et al., 1981). Heathcock and coworkers
vert in the aqueous environment of the subsequent biocatalytic demonstrated that if two equivalents of dibutylboron triflate are
reaction.
supplied to this reaction, the anti product with the a stereocenter
For isolated KRs to be considered robust biocatalysts, they opposite in configuration to the Evans syn product as well
should catalyze the reduction of a variety of substrates. Thus, as some quantity of anti-Evans syn product are generated
a panel of five diketide substrates was synthesized to test (Raimundo and Heathcock, 1995). We utilized this protocol to
the KRs: 3-oxopentanoyl-SNAC (1), (2RS)-methyl-3-oxobuta- generate a mixture of ‘‘2R,3R’’ and ‘‘2R,3S’’ products. Chemistry
noyl-SNAC (2), (2RS)-methyl-3-oxopentanoyl-SNAC (3), (2RS)- detailed by Boddy and coworkers was applied to each of the
Chemistry & Biology 18, 1331–1340, October 28, 2011 ª2011 Elsevier Ltd All rights reserved 1333

Chemistry & Biology
Biocatalysis with Modular PKS Ketoreductases
Figure 2. KR Biocatalytic Assays
Representative chromatograms show the reactions of five of the most active KRs with each of the diketide substrates 1–5 on the left (representative chro-
matograms of the other KRs are illustrated in Figure S2). Chiral building block products are displayed in the middle. The turnover efficiencies of each KR for each
substrate are graphed on the right (‘‘turnover efficiency’’ is defined as the quantity of a product divided by quantities of the products and remaining substrate). The
1334 Chemistry & Biology 18, 1331–1340, October 28, 2011 ª2011 Elsevier Ltd All rights reserved

Chemistry & Biology
Biocatalysis with Modular PKS Ketoreductases
Analyzing KR Reactions
After reactions in which each of the 11 reductase-competent
KRs was paired with each of 1–5 had been incubated for 1 day
at 22ꢀC, they were extracted with ethyl acetate and analyzed
by chiral chromatography (Figure 2; Figure S2). Some reactions
may not have gone to completion due to substrate incompati-
bility, slow kinetics, protein stability, or many other factors.
Because the overall fitness of KRs as biocatalysts was of
interest, a metric of ‘‘turnover efficiency’’ was constructed,
defined as the quantity of a product formed divided by the quan-
tities of all products and unreacted substrate. Relative quantities
were measured by integrating peaks from the 235 nm absor-
bance traces of chiral chromatography runs. Because all reac-
tions were performed in duplicate, the reported turnover efficien-
cies are averages from two runs (in no case was there a deviation
larger than 10%) (Table S2). As 1 could not be accurately quan-
tified (it migrates as an extremely broad peak), the turnover
efficiencies of 1a and 1b were calculated relative to the TylKR1
reaction, which produced the largest product peak. The elution
order for 1a and 1b was assigned as first the ‘‘3R’’ product
then the ‘‘3S’’ product because reactions of active B-type
KRs generated the first peak and reactions of all of the active
A-type KRs (with the exception of AmpKR2) generated the
second peak. The elution order of 5a–5d was assigned in
the same order as the products of 2–4. Added validation for
this assignment comes from the observation that the syn and
anti products of 2–4 have different maximal absorbances (the
anti products maximally absorb at ꢁ232 nm, whereas the syn
products maximally absorb at ꢁ233 nm): the first two bio-
catalytic products of 5 that elute from the chiral column absorb
as anti products, whereas the last two absorb as syn products.
KR Biocatalysis Results
KRs Usually Maintain Natural Stereocontrol
Active KRs usually maintained the stereocontrol anticipated from
their KR types (Figure 2). Most of these KRs even maintained
stereocontrol toward substrate 5, which harbors an uncommon
a-ethyl group. AmpKR2 was identified as a general A1-type KR
biocatalyst, and AmpKR1 was identified as a general A2-type
KR biocatalyst. Because TylKR1 proved to be a general B1-
type KR biocatalyst and EryKR1 proved to be a general B2-
type KR biocatalyst (PikKR1 is even more robust), the desired
spectrum of KR activities was realized. That the KR reactions
are as stereocontrolled as they were determined to be is note-
worthy, because the natural substrates of these enzymes are
polyketide intermediates bound to the pantetheinyl arm pros-
thetic group of holo-ACPs.
Figure 3. Determining Elution Orders
(A) Aldol reactions helped synthesize chiral chromatography standards. After
(i) the propionyl group was added to the chiral auxiliary, then either (ii) the
Heathcock route using two equivalents of dibutylboron triflate or (iii) the
standard Evans route using one equivalent yielded diketide oxazolidinones
that were (iv) hydrolyzed and coupled to NAC.
(B and C) Standards were compared to products from MycKRA (A-type),
TylKR1 (B1-type), and EryKR1 (B2-type) to establish the elution order of
products from KR reactions with 2 and 4. See also Table S3.
KRs that Naturally Reduce Smaller Intermediates
Are Most Active toward Diketide-SNACs
In general, KRs that reduce small polyketides within their native
diketide-oxazolidinones to hydrolyze the diketides from their PKSs, such as AmpKR1, AmpKR2, EryKR1, PikKR1, and
chiral auxiliaries and couple them to NAC (Sharma and Boddy, TylKR1, were most active on the panel of diketides. KRs that
2007). The ‘‘2R,3R’’ and ‘‘2R,3S’’ diketide-SNACs were sepa- reduce larger polyketides within their native PKSs, such as
rated by reverse-phase HPLC.
PikKR5, OleKR6, AmpKR10, and AmpKR11, were nearly inactive
heights of the bars can be interpreted as relative activities, whereas the ratio of colors within a bar can be interpreted as a measure of the stereocontrol exerted by
a KR on a substrate. Diverse KR types were used in these assays: two A types that naturally operate on a-unsubstituted intermediates, four A1 types, two A2
types, one B1 type, two B2 types, and one C1 type (as a negative control). See also Figure S2 and Table S2.
Chemistry & Biology 18, 1331–1340, October 28, 2011 ª2011 Elsevier Ltd All rights reserved 1335

Chemistry & Biology
Biocatalysis with Modular PKS Ketoreductases
form of a thioester. This moiety can be converted to a carbox-
ylic acid or an aldehyde or used to couple the diketide to
other fragments through an ester, amide, or ketone linkage,
making thioester-containing building blocks very versatile in the
syntheses of natural products (Bruice et al., 1963; Tokuyama
et al., 1998). The installation of a second synthetic handle on
the other end of the diketide chain is also desirable. Because
the biocatalytic assays reported here revealed that variations
at the g position of the diketide substrates are tolerated by
many KRs, it may be possible to incorporate synthetic handles
such as a terminal chlorine, alcohol, double bond, or triple
bond.
Figure 4. Accessing Diverse Building Block Libraries through PKS
Biocatalysis
Biocatalytic reactions using DHs, ERs, and Module+TEs could enable the
generation of libraries of other building blocks useful in the syntheses of natural
products, such as a-substituted diketides and triketide lactones containing
up to four contiguous stereocenters (representative molecules are shown).
toward 1–5. This could indicate that KRs that naturally reduce
Additional building block libraries may be accessed using
short-chain polyketide intermediates within their native PKSs other PKS enzymes as biocatalysts. Because many DHs form
obtain binding energy for catalysis from interactions with the trans double bonds from (2R,3R)-a-alkyl-b-hydroxy intermedi-
polyketide.
ates, preparative quantities of a-substituted, a/b-unsaturated di-
MycKRA was one of the most active KRs in these studies. ketides may be accessed by incubating a-substituted, b-keto-
Three modules of the mycolactone PKS contain 99% identical diketide-SNACs with appropriate KRs and DHs (Keatinge-Clay,
MycKRA enzymes as a result of duplication events; thus, within 2008). As ERs stereospecifically reduce such double bonds,
the synthase MycKRA enzymes reduce a diketide, a triketide, a-substituted, b-keto-diketide-SNACs could be converted,
and a hexaketide (Stinear et al., 2004). That both A1-type and through the driving force of an NADPH-regeneration system, to
A2-type products were produced when MycKRA was incubated chiral a-substituted, b-methylene-diketide-SNACs by employing
with a-substituted diketides 2–5 is understandable in that this a combination of KR, DH, and ER biocatalysts (Kwan et al.,
enzyme naturally operates on a-unsubstituted intermediates. 2008). Diketide-SNACs can also be enzymatically converted
MycKRA did not produce B-type products. Broad in its substrate to triketide lactones containing four contiguous stereocenters
selectivity, MycKRA is a robust, stereocontrolled A-type KR (Figure 4). Such molecules have been generated by Module+TEs
biocatalyst.
Deviations from Expected
(PKS modules C-terminally fused to EryTE) in quantities analyz-
able by radio-TLC or GC-MS as readout of enzymological
A few KRs were sensitive to slight differences within the panel of studies (Chen et al., 2006). Recently, we have developed an
diketides. TylKR1, which is in general a robust, active B1-type economical route to supply extender units accepted by ATs to
KR, produced mainly the A2-type product 2b when incubated in vitro PKS reactions (Hughes and Keatinge-Clay, 2011). Thus,
with 2, which is one methylene shorter than its natural substrate. the scaled-up incubation of reduced, chiral diketide-SNACs
AmpKR2, an A1-type KR, converted the a-unsubstituted dike- with diverse Module+TEs in the presence of NADPH may yield
tide 1 mainly to a B-type product. Research is ongoing to deter- libraries of triketide building blocks.
mine whether KRs are more active and/or stereocontrolled
Currently, most experiments on modular PKS enzymes are
toward pantetheine-bound diketides than NAC-bound diketides. performed either in vivo, by expressing PKSs heterologously
and extracting polyketides from the growth medium, or in vitro,
by combining the appropriate enzymes, cofactors, and sub-
Scaled-Up Reactions
Because KR biocatalysts representing each KR type were iden- strates and then analyzing the reaction products by radio-TLC,
tified that could generate the desired library of diketide chiral GC-MS, or LC-MS (reduced polyketides are usually invisible
building blocks, we sought to demonstrate that useful quantities on UV absorbance traces). An experimental platform in which
of chiral building blocks could be obtained. AmpKR2 (A1-type), PKS enzymes can perform catalysis in a controlled setting to
AmpKR1 (A2-type), TylKR1 (B1-type), and EryKR1 (B2-type) produce quantities of polyketides analyzable by NMR or crystal-
were each incubated in scaled-up reactions with 3. Over lography is highly desired (the stereocenters of products would
100 mg of each of 3a, 3b, 3c, and 3d was obtained from these be much more readily determined, for example). Expense has
scaled-up reactions. Because the reactions with lower turnover been the principal deterrent of larger in vitro reactions because
efficiencies required significant KR concentrations and the polyketide extender units such as methylmalonyl-CoA and
nickel column purification step only captures a small percentage cofactors like NADPH are costly. However, such biocatalytic
of KR in the lysate from overexpressing cells, reactions were processes as the MatB-catalyzed production of extender units
also attempted using lysate. Reactions using the lysate were and GDH-catalyzed regeneration of NADPH should enable
complete within 6 hr and also provided more than 100 mg of PKS enzymologists to perform in vitro reactions more affordably
chiral product. Both for reactions using pure KRs and KR-con- under controlled conditions, generating multimilligram quantities
taining lysate, KR reactions were as stereocontrolled as in the of polyketide products.
smaller-scale reactions.
SIGNIFICANCE
Building Block Libraries, PKS Biocatalysis,
and PKS Enzymology
The scalable production of reduced diketide-SNACs (-S-N-
As well as containing valuable combinations of stereocenters, acetylcysteamines) has been realized through the described
the described building blocks harbor a synthetic handle in the general chemoenzymatic route. A screen of 11 diverse,
1336 Chemistry & Biology 18, 1331–1340, October 28, 2011 ª2011 Elsevier Ltd All rights reserved

Chemistry & Biology
Biocatalysis with Modular PKS Ketoreductases
Scaled-Up Reactions
isolated ketoreductases (KRs) toward diketides 1–5 re-
vealed that KRs isolated from modules that naturally reduce
shorter polyketide intermediates are most active on dike-
tide-SNACs. AmpKR2 and AmpKR1 were identified as
robust A1- and A2-type KR biocatalysts, and TylKR1 and
EryKR1 were shown to be robust B1- and B2-type KR bio-
catalysts. More than 100 mg of each diketide 3a–3d was
produced to demonstrate that quantities of chiral diketide
building blocks useful for natural product and pharmaceu-
tical syntheses could be generated economically and
greenly. Other polyketide synthase (PKS) enzymes such as
dehydratases, enoylreductases, and PKS modules C-termi-
nally fused to EryTE can now be investigated as biocatalysts
in the generation of an even greater diversity of diketide and
triketide building blocks. The described biocatalysis will
also benefit PKS enzymology because it is no longer cost-
prohibitive to perform in vitro reactions that yield significant
quantities of polyketide products as readout.
Reactions were performed on 50 mM 3 in 10% (w/v) glycerol, 300 mM HEPES
(pH 7.5), 100 mM NaCl, 300 mM D-glucose, 100 mM NADP+, 1 mM GDH, 50 mM
purified KR in a total volume of 15 ml. All reactions were incubated at 22ꢀC for
1 day. Each was determined to be complete by reverse-phase HPLC. After
EtOAc extraction and silica gel chromatography (EtOAc), yields of ꢁ75%
were obtained. Reactions were also performed using lysate (50 ml from 6 l
culture) that was not poured over a nickel column but rather dialyzed overnight
at 4ꢀC in 10 kDa MWCO cellophane dialysis tubing against 10% (w/v) glycerol,
250 mM NaCl, 30 mM HEPES (pH 7.5). In these reactions, 50 mM 3 was incu-
bated in 10% (w/v) glycerol, 300 mM HEPES (pH 7.5), 100 mM NaCl, 300 mM
D-glucose, 100 mM NADP+, 1 mM GDH in a total volume of 15 ml, but with 5 ml
of the reaction volume being supplied from dialyzed lysate. All reactions were
incubated at 22ꢀC for 6 hr. Each was determined to be complete by reverse-
phase HPLC. After EtOAc extraction and silica gel chromatography (EtOAc),
yields of ꢁ85% were obtained. Background thioester hydrolysis occurs with
a half-time of ꢁ3 days for both reactions with purified KRs and reactions
with KR-containing lysate.
Chiral Chromatography
Extracts of reactions from the biocatalytic screens were separated using
a ChiralCel OC-H column (250 3 4.6 mm) on a Beckman Coulter HPLC system
with a 20 ml loop. Solvent systems and flow rates were optimized for each set of
reactions with a particular diketide-SNAC (Table S3). Substrates and products
were observed at a wavelength of 235 nm.
EXPERIMENTAL PROCEDURES
Cloning
All KRs, with the exception of EryKR1, and GDH were cloned from genomic
DNA (gDNA) (Table S1). Regions of DNA encoding KRs from amphotericin
(Amp), oleandomycin (Ole), pikromycin (Pik), spinosyn (Spn), and tylosin (Tyl)
were amplified from Streptomyces nodosus, Streptomyces antibioticus,
Streptomyces venezuelae, Saccharopolyspora spinosa, and Streptomyces
fradiae gDNA, respectively. N- and C-terminal boundaries were chosen using
the boundaries determined from the structures of EryKR1 and TylKR1
(Keatinge-Clay and Stroud, 2006; Keatinge-Clay, 2007). The region encoding
the KR from the mycolactone (Myc) PKS was amplified from a bacterial artifi-
cial chromosome containing DNA from pMUM001 (Stinear et al., 2004). The
regions encoding the KRs from the erythromycin (Ery) PKS were amplified
from a synthetic gene (Kodumal et al., 2004). GDH was cloned from B. subtilis
gDNA. NdeI and EcoRI restriction sites were designed to flank the amplicons
encoding AmpKR2, AmpKR10, AmpKR11, AmpKR13, EryKR2, OleKR6,
PikKR1, PikKR5, and SpnKR3; NdeI and XhoI restriction sites were designed
to flank the amplicons encoding AmpKR1, EryKR1, TylKR1, and GDH; NheI
and NotI restriction sites were designed to flank the amplicon encoding
MycKRA. All restriction sites were engineered so that histidine tags encoded
by pET28b (Novagen) would be N-terminally fused to the proteins, facilitating
purification.
Synthetic Protocols
General Synthetic Route to a-Substituted, b-Keto Diketide-SNAC
Substrates
N-Acetylcysteamine. Cysteamine hydrochloride (3.05 g, 53.6 mmol, 1 eq.),
potassium hydroxide (1.50 g, 53.6 mmol, 1 eq.), and sodium bicarbonate
(6.75 g, 160.8 mmol, 3 eq.) were added to 50 ml H2O. Acetic anhydride
(2.28 ml, 48.2 mmol, 0.9 eq.) was then added dropwise over 5 min, and the
solution was allowed to stir at 22ꢀC for 15 min. The reaction was quenched
with 1 M HCl, and the product was extracted with ethyl acetate (3 3 50 ml),
dried with MgSO4, filtered, and concentrated in vacuo, yielding a clear, viscous
liquid.
Meldrum’s Acid Derivatives. Pyridine (2 eq.) was added to Meldrum’s acid
(1 eq.) in 40 ml dry dichloromethane (DCM) at 0ꢀC. Next, the acyl chloride
(1 eq.) was added dropwise over 15 min, turning the solution a dark orange
color, and the reaction was stirred overnight at 22ꢀC. The reaction was washed
with 0.1 M HCl (3 3 50 ml) and the organic layer was dried with MgSO4, filtered,
and concentrated in vacuo.
a-Unsubstituted Diketide-SNACs. The Meldrum’s acid derivative (1 eq.) and
N-acetylcysteamine (1 eq.) were refluxed in dry toluene for 5 hr. The resulting
solution was then concentrated in vacuo.
Protein Expression and Purification
a-Substituted Diketide-SNACs. The a-unsubstituted diketide-SNAC (1 eq.)
was dissolved in 50 ml dry THF at 0ꢀC. Potassium tert-butoxide (1.2 eq.)
was added. After the alkyl iodide (5 eq.) was injected, the reaction was allowed
to warm to 22ꢀC and stir overnight. After the reaction was quenched with
0.1 M HCl, the product was extracted with ethyl acetate (3 3 50 ml), dried
with MgSO4, filtered, and concentrated in vacuo, resulting in a clear, viscous
liquid.
All plasmids were transformed into Escherichia coli BL21(DE3). The cells were
grown in Luria broth with 25 mg/l kanamycin at 37ꢀC until they reached an
OD600 of 0.5. They were then cooled to 15ꢀC, induced with 1 mM IPTG, and
grown for an additional 16 hr. Cells were spun down, resuspended in lysis
buffer (10% [w/v] glycerol, 0.5 M NaCl, 30 mM Tris [pH 7.5]), and lysed by soni-
cation. After centrifugation at 30,000 3 g for 30 min, lysate was poured over
nickel-NTA resin (QIAGEN) equilibrated with lysis buffer. Bound protein was
washed with 15 mM imidazole in lysis buffer and eluted with 150 mM imidazole
in lysis buffer. Proteins were then polished using a Superdex 200 gel filtration
column that was equilibrated with 10% (w/v) glycerol, 150 mM NaCl, 10 mM
Tris (pH 7.5) and concentrated to 200 mM. Protein purity was evaluated by
Coomassie-stained SDS-PAGE gels (Figure S1).
Syntheses of a-Methyl, b-Hydroxyl Diketide-SNAC Standards
(4S)-4-Benzyl-3-Propionyl-2-Oxazolidinone (6). (4S)-benzyl-2-oxazolidinone
(250 mg, 1.41 mmol, 1 eq.) was dissolved in 5 ml dry THF at ꢂ78ꢀC under
nitrogen. n-butyllithium (0.64 ml of a 2.5 M stock in hexanes, 1.59 mmol,
1.13 eq.) was added dropwise to the cooled solution by syringe and stirred
for 5 min. Propionyl chloride (0.158 ml, 1.81 mmol, 1.28 eq.) was added drop-
wise by syringe, and the solution was stirred at ꢂ78ꢀC for 30 min. The reaction
was warmed to room temperature and stirred for an additional 30 min before
being quenched with 3 ml saturated NH4Cl (aq.). The product was extracted
with DCM (2 3 20 ml), washed with 10 ml saturated NaHCO3 (aq.) and then
10 ml brine, dried with MgSO4, filtered, and concentrated in vacuo to produce
a clear oil (260 mg, 80%). Rf 0.40 (20% EtOAc in hexanes).
KR Reductions
Biocatalytic Screen
Reactions were performed on 10 mM 1–5 in 10% (w/v) glycerol, 100 mM NaCl,
150 mM HEPES (pH 7.5), 200 mM D-glucose, 100 mM NADP+, 1 mM GDH,
and 5 mM KR in a total volume of 200 ml. Following incubation at 22ꢀC
for 1 day, reactions were extracted with 2 3 500 ml ethyl acetate and evapo-
rated. The extracts were resuspended in ethanol prior to analysis by chiral
chromatography.
(4S,20S,30R)-3-(30-Hydroxy-20-Methylbutanoyl)-4-Benzyl-2-Oxazolidinone (7).
(4S)-4-benzyl-3-propionyl-2-oxazolidinone (6) (560 mg, 2.40 mmol, 1 eq.) was
added to 10 ml dry DCM under nitrogen and cooled to 0ꢀC. Dibutylboron triflate
Chemistry & Biology 18, 1331–1340, October 28, 2011 ª2011 Elsevier Ltd All rights reserved 1337

Chemistry & Biology
Biocatalysis with Modular PKS Ketoreductases
(Bu2BOTf) (2.83 ml of a 1 M solution in DCM, 2.83 mmol, 1.18 eq.) was supplied
via syringe. Diisopropylethylamine (0.56 ml, 3.19 mmol, 1.33 eq.) was then
added dropwise over 2 min, and the solution was cooled to ꢂ78ꢀC. After acet-
aldehyde (0.151 ml, 270 mmol, 1.12 eq.) was added, the solution was stirred at
ꢂ78ꢀC for 30 min, allowed to warm to 0ꢀC, and further stirred for 2 hr. A solution
of 3.3 ml Na2HPO4 (1 M, pH 7.4) and 8.3 ml MeOH was added to the reaction,
followed by the dropwise addition of 8.5 ml 2:1 MeOH:30% H2O2 (aq.). After
1.5 hr, organic solvents were evaporated in vacuo and the remaining material
was resuspended in 10 ml 5% NaHCO3 (aq.). The product was extracted in
DCM (3 3 20 ml), washed with 20 ml 5% NaHCO3 (aq.) and then 20 ml brine,
dried with MgSO4, filtered, and concentrated in vacuo to yield a clear oil
(350 mg, 53% after flash chromatography). Flash chromatography was per-
formed with a 15%–30% EtOAc in hexanes gradient. Rf 0.38 (50% EtOAc in
hexanes).
(4S,20S,30R)-3-(30-Hydroxy-20-Methylhexanoyl)-4-Benzyl-2-Oxazolidinone (8).
The synthesis for 7 was followed with the exception that butyrylaldehyde
(0.213 ml, 2.41 mmol, 1.12 eq.) was substituted for acetaldehyde. Flash chro-
matography was performed in the same manner, yielding 241 mg (33%) of the
desired product. Rf 0.67 (50% EtOAc in hexanes).
(4S,20R,30R)-3-(30-Hydroxy-20-Methylbutanoyl)-4-Benzyl-2-Oxazolidinone (9)
and (4S,20R,30S)-3-(30-Hydroxy-20-Methylbutanoyl)-4-Benzyl-2-Oxazolidinone
(10). Synthesis followed that of 7, with the exception that two equivalents of
Bu2BOTf were used (2.4 ml of a 1 M solution in DCM, 2.4 mmol, 2 eq.).
Rf 0.56 and 0.60 (50% EtOAc in hexanes).
tem equipped with an electrospray-ionization source. A 5%–95% B gradient
over 12 min at a flow rate of 0.7 ml/min was run in which the mobile
phases were (A) H2O with 0.1% formic acid and (B) acetonitrile with 0.1% formic
acid.
3-Oxopentanoyl-N-Acetylcysteamine Thioester (1). 1H NMR (400 MHz,
CDCl3) d 1.08 (t, 3H, J = 7.4 Hz, CH3), 1.96 (s, 3H, CH3-C=O), 2.55 (q, 2H,
J = 7.4 Hz C-CH2-C=O), 3.00–3.10 (m, 2H, S-CH2), 3.4–3.5 (m, 2H, N-CH2),
3.7 (s, 2H, O=C-CH2-C=O), 5.88 (br s, 1H, NH). ESI-MS expected mass:
218.3; observed mass: 218.2.
3-Oxobutanoyl-N-Acetylcysteamine Thioester. 1H NMR (400 MHz, CDCl3)
d 1.92 (s, 3H, CH3-C(=O)-N), 2.20 (s, 3H, CH3-C=O), 2.93–3.16 (m, 2H,
S-CH2), 3.36–3.42 (m, 2H, N-CH2), 3.62 (s, 2H, O=C-CH2-C=O), 5.99 (br s,
1H, NH).
(2RS)-Methyl-3-Oxobutanoyl-N-Acetylcysteamine Thioester (2). 1H NMR
(400 MHz, CDCl3) d 1.33 (d, 3H, J = 7.5 Hz, CH3-C-(C=O)2), 1.90 (s, 3H,
CH3-C(=O)-N), 2.19 (s, 3H, CH3-C=O), 2.95–3.08 (m, 2H, S-CH2), 3.30–3.46
(m, 2H, N-CH2), 3.71 (q, 1H, J = 7.5 Hz, O=C-CH-C=O), 5.87 (br s, 1H, NH).
ESI-MS expected mass: 218.3; observed mass: 218.4.
(2RS)-Methyl-3-Oxopentanoyl-N-Acetylcysteamine Thioester (3). 1H NMR
(400 MHz, CDCl3) d 1.08 (t, 3H, J = 7.5 Hz, CH3), 1.38 (d, 3H, J = 7.5 Hz,
CH3-C-(C=O)2), 1.96 (s, 3H, CH3-C=O), 2.48–2.68 (m, 2H, C-CH2-C=O),
3.00–3.14 (m, 2H, S-CH2), 3.36–3.52 (m, 2H, N-CH2), 3.79 (q, 1H, J = 7.5 Hz,
O=C-CH-C=O), 5.92 (br s, 1H, NH). ESI-MS expected mass: 232.3; observed
mass: 232.4.
3-Oxohexanoyl-N-Acetylcysteamine Thioester. 1H NMR (400 MHz, CDCl3)
d 0.90–1.00 (m, 3H, CH3), 1.55–1.70 (m, 2H, C-CH2-C), 1.96 (s, 3H, CH3-C=O),
2.48–2.60 (m, 2H, C-CH2-C=O), 3.00–3.14 (m, 2H, S-CH2), 3.40–3.52 (m, 2H,
N-CH2), 3.70 (s, 2H, O=C-CH2-C=O), 5.92 (br s, 1H, NH).
(2RS)-Methyl-3-Oxohexanoyl-N-Acetylcysteamine Thioester (4). 1H NMR
(400 MHz, CDCl3) d 0.85 (t, 3H, J = 7.6 Hz, CH3), 1.33 (d, 3H, J = 7.9 Hz,
CH3-C-(C=O)2), 1.50–1.60 (m, 2H, C-CH2-C), 1.98 (s, 3H, CH3-C=O), 2.45
(m, 2H, C-CH2-C=O), 2.95–3.12 (m, 2H, S-CH2), 3.34–3.52 (m, 2H, N-CH2),
3.74 (q, 1H, J = 7.9 Hz, O=C-CH-C=O), 5.92 (br s, 1H, NH). ESI-MS expected
mass: 246.3; observed mass: 246.4.
(2RS)-Ethyl-3-Oxopentanoyl-N-Acetylcysteamine Thioester (5). 1H NMR
(400 MHz, CDCl3) d 0.94 (t, 3H, J = 7.3 Hz, CH3), 1.17 (t, 3H, J = 8.4 Hz,
CH3-C-C=O), 1.88–1.98 (m, 2H, C-CH2-C-(C=O)2), 2.15 (s, 3H, CH3-C=O),
2.48–2.66 (m, 2H, C-CH2-C=O), 3.01–3.20 (m, 2H, S-CH2), 3.41–3.58 (m, 2H,
N-CH2), 3.68–3.74 (m, 1H, O=C-CH-C=O), 6.27 (br s, 1H, NH). ESI-MS
expected mass: 246.3; observed mass: 246.4.
(4S)-4-Benzyl-3-Propyl-2-Oxazolidinone (6). 1H NMR (400 MHz, CDCl3)
d 1.22 (t, 3H, J = 8.0 Hz, CH3), 2.76–2.82 (m, 1H, one of CH2-Ph), 2.90–3.05
(m, 2H, CH2-C=O), 3.31 (dd, 1H, J = 15.2 Hz, J = 5.5 Hz, one of CH2-Ph),
4.16–4.23 (m, 2H, CH2-O), 4.66–4.70 (m, 1H, CH-N).
(4S,20S,30R)-3-(30-Hydroxy-20-Methylbutanoyl)-4-Benzyl-2-Oxazolidinone (7).
1H NMR (400 MHz, CDCl3) d 1.23 (d, 3H, J = 7.4 Hz, CH3), 1.28 (d, 3H, J = 6.1Hz,
CH3-C-C=O), 2.76–2.82 (m, 1H, one of CH2-Ph), 2.93 (br s, 1H, OH), 3.26 (dd,
1H, J = 12.3 Hz, J = 3.3 Hz, one of CH2-Ph), 3.75 (dq, 1H, J = 9.8 Hz,
J = 3.1 Hz, CH), 4.16–4.23 (m, 2H, CH2-O), 4.65–4.69 (m, 1H, CH-N).
(4S,20S,30R)-3-(30-Hydroxy-20-Methylhexanoyl)-4-Benzyl-2-Oxazolidinone(8).
1H NMR (400 MHz, CDCl3) d 0.94 (t, 3H, J = 7.2 Hz, CH3), 1.30 (d, 3H, CH3-C-
C=O), 1.35–1.45 (m, 1H, one of C-CH2-C), 1.45–1.58 (m, 1H one of C-CH2-
C), 2.75–2.81 (m, 1H, one of CH2-Ph), 2.84 (d, 1H, OH), 3.26 (dd, 1H,
J = 13.7 Hz, J = 3.1 Hz, one of CH2-Ph), 3.76 (dq, 1H, J = 7.2 Hz, J = 2.6 Hz,
CH), 3.95–4.00 (m, 1H, CH-C=O), 4.17–4.27 (m, 2H, CH2-O), 4.68–4.74
(m, 1H, CH-N).
(4S,20R,30R)-3-(30-Hydroxy-20-Methylbutanoyl)-4-Benzyl-2-Oxazolidinone(9).
1H NMR (400 MHz, CDCl3) d 1.22 (d, 3H, J = 5.6 Hz, CH3), 1.32 (d, 3H, J = 6.2Hz,
CH3-C-C=O), 2.56 (d, 1H, J = 7.4 Hz one of CH2-Ph), 2.82 (br s, 1H, OH),
3.28–3.35 (m, 1H, one of CH2-Ph), 3.90–3.95 (m, 1H, CH), 3.96–4.00 (m, 1H,
CH-C=O), 4.15–4.25 (m, 2H, CH2-O), 4.67–4.72 (m, 1H, CH-N).
(4S,20R,30R)-3-(30-Hydroxy-20-Methylhexanoyl)-4-Benzyl-2-Oxazolidinone (11).
1H NMR (400 MHz, CDCl3) d 0.94 (t, 3H, CH3), 1.20 (d, 3H, J = 7.2 Hz, CH3-C-
C=O), 1.35–1.45 (m, 1H, one of C-CH2-C), 1.45–1.58 (m, 1H one of C-CH2-
C), 2.75–2.82 (m, 1H, one of CH2-Ph), 2.91 (d, 1H, J = 7.2 Hz, OH), 3.31 (dd,
1H, J = 7.9 Hz, J = 3.1 Hz, one of CH2-Ph), 3.85 (dq, 1H, J = 7.2 Hz, J =
2.7 Hz, CH), 3.95–4.04 (m, 1H, CH-C=O), 4.15–4.26 (m, 2H, CH2-O), 4.66–
4.74 (m, 1H, CH-N).
(4S,20R,30R)-3-(30-Hydroxy-20-Methylhexanoyl)-4-Benzyl-2-Oxazolidinone(11)
and (4S,20R,30S)-3-(30-Hydroxy-20-Methylhexanoyl)-4-Benzyl-2-Oxazolidinone
(12). Synthesis followed that of 8, with the exception that two equivalents of
Bu2BOTf was added (4.32 ml of a 1 M solution in DCM, 4.32 mmol, 2 eq.). Rf
0.73 and 0.77 (50% EtOAc in hexanes).
(2S,3R)-3-Hydroxy-2-Methylbutanoyl-N-Acetylcysteamine Thioester (13).
Aldol product 7 (187 mg, 0.68 mmol) was dissolved in 3 ml of a 4:1 mixture of
THF:H2O and cooled to 0ꢀC. Slowly,
3 ml of 30% H2O2 was added,
followed by 1 ml saturated LiOH (0.5 g LiOH in 1 ml H2O). The reaction was
left to stir at 0ꢀC for 2 hr, at which point Na2SO3 was added until H2O2 was
quenched (bubbling ceased). The aqueous phase was washed with DCM
(2 3 15 ml), and the DCM layers were back-extracted with H2O (1 3 15 ml).
The combined aqueous layers were cooled to 0ꢀC and the pH was lowered
to 1 with 6 M HCl. The aldol acid was extracted into EtOAc (3 3 20 ml), dried
with MgSO4, filtered, and concentrated in vacuo (as the aldol acid is somewhat
volatile, the solution was only concentrated to 1 ml). 1-ethyl-3-(3-dimethylami-
nopropyl)carbodiimide (211.1 mg, 1.36 mmol, 2 eq.) was added to the solution
under nitrogen at 0ꢀC. NAC (96.8 mg, 0.81 mmol, 1.2 eq.) and a few crystals
of 4-dimethylaminopyridine were added. After 4 hr at 0ꢀC, the reaction
was quenched with 5 ml H2O. Product was extracted into EtOAc (3 3 15 ml),
washed with 5 ml brine, dried with MgSO4, and filtered. A yellowish oil was
produced after concentration in vacuo (50 mg, 47%). Rf 0.20 (EtOAc).
(2S,3R)-3-Hydroxy-2-Methylhexanoyl-N-Acetylcysteamine Thioester (14).
14 was synthesized in the same manner as 13 from 8 (81%). Rf 0.35 (in EtOAc).
(2R,3R)-3-Hydroxy-2-Methylbutanoyl-N-Acetylcysteamine Thioester (15)
and (2R,3S)-3-Hydroxy-2-Methylbutanoyl-N-Acetylcysteamine Thioester (16).
15 and 16 were synthesized in the same manner as 13 from a mixture of 9
and 10 (33%, 15 + 16). They were then separated by reverse-phase HPLC
on a semipreparative C18 column (BondClone, 300 3 7.8 mm; Phenomenex)
connected to a Waters 1525 HPLC system through a gradient of 15%–
100% B over 30 min, where the mobile phases were (A) H2O with 0.1% TFA
and (B) MeOH with 0.1% TFA, pumped at a flow rate of 2 ml/min. Rf 0.15
and 0.20 (in EtOAc).
(2R,3R)-3-Hydroxy-2-Methylhexanoyl-N-Acetylcysteamine Thioester (17)
and (2R,3S)-3-Hydroxy-2-Methylhexanoyl-N-Acetylcysteamine Thioester (18).
17 and 18 were synthesized in the same manner as 13 from a mixture of 11
and 12 (85%, 17 + 18). They were then separated as described for 15 and
16. Rf 0.30 and 0.35 (in EtOAc).
Characterization via NMR and LC-MS
NMR was performed either on a Varian Mercury 400 MHz or a Varian INOVA 500
MHz instrument. LC-MS was performed on an Agilent Technologies 1200
Series HPLC with a Gemini C18 column (5 mm, 2 3 50 mm; Phenomenex)
coupled to an Agilent Technologies 6130 quadrupole mass spectrometer sys-
1338 Chemistry & Biology 18, 1331–1340, October 28, 2011 ª2011 Elsevier Ltd All rights reserved

Chemistry & Biology
Biocatalysis with Modular PKS Ketoreductases
(2S,3R)-3-Hydroxy-2-Methylbutanoyl-N-Acetylcysteamine Thioester (13).
1H NMR (500 MHz, CDCl3) d 1.22 (d, 3H, J = 7.6 Hz, CH3), 1.24 (d, 3H, CH3-
C-C=O), 2.01 (s, 3H, CH3-C=O), 2.70–2.75 (m, 1H, CH-C=O), 3.00–3.10
(m, 2H, S-CH2), 3.42–3.53 (m, 2H, N-CH2), 3.70–3.79 (m, 1H, CH), 5.93 (br s,
1H, NH). ESI-MS expected mass: 220.3; observed mass: 220.2.
Hilterhaus, L., and Liese, A. (2007). Building blocks. Adv. Biochem. Eng.
Biotechnol. 105, 133–173.
Holzbaur, I.E., Harris, R.C., Bycroft, M., Cortes, J., Bisang, C., Staunton, J.,
Rudd, B.A., and Leadlay, P.F. (1999). Molecular basis of Celmer’s rules: the
role of two ketoreductase domains in the control of chirality by the erythro-
mycin modular polyketide synthase. Chem. Biol. 6, 189–195.
(2S,3R)-3-Hydroxy-2-Methylhexanoyl-N-Acetylcysteamine Thioester (14).
1H NMR (500 MHz, CDCl3) d 0.94 (t, 3H, J = 7.7 Hz, CH3), 1.22 (d, 3H, J =
7.7 Hz, CH3-C-C=O), 1.50–1.60 (m, 1H, one of C-CH2-C), 1.60–1.70 (m, 1H
one of C-CH2-C), 1.95 (s, 3H, CH3-C=O), 2.70–2.75 (m, 1H, CH-C=O), 3.00–
3.08 (m, 2H, S-CH2), 3.40–3.52 (m, 2H, N-CH2), 3.92–3.98 (m, 1H, J =
5.4 Hz, CH), 5.86 (br s, 1H, NH). ESI-MS expected mass: 248.3; observed
mass: 248.4.
Hughes, A.J., and Keatinge-Clay, A. (2011). Enzymatic extender unit genera-
tion for in vitro polyketide synthase reactions: structural and functional show-
casing of Streptomyces coelicolor MatB. Chem. Biol. 18, 165–176.
Kataoka, M., Kita, K., Wada, M., Yasohara, Y., Hasegawa, J., and Shimizu, S.
(2003). Novel bioreduction system for the production of chiral alcohols. Appl.
Microbiol. Biotechnol. 62, 437–445.
(2R,3R)-3-Hydroxy-2-Methylbutanoyl-N-Acetylcysteamine Thioester (15).
1H NMR (500 MHz, CDCl3) d 1.22 (d, 3H, J = 7.7 Hz, CH3), 1.24 (d, 3H, CH3-
C-C=O), 1.96 (s, 3H, CH3-C=O), 2.70–2.76 (m, 1H, CH-C=O), 3.01–3.14
(m, 2H, S-CH2), 3.47–3.58 (m, 2H, N-CH2), 3.79–3.83 (m, 1H, CH), 6.26 (br s,
1H, NH). ESI-MS expected mass: 220.3; observed mass: 220.2.
Keatinge-Clay, A.T. (2007). A tylosin ketoreductase reveals how chirality is
determined in polyketides. Chem. Biol. 14, 898–908.
Keatinge-Clay, A.T. (2008). Crystal structure of the erythromycin polyketide
synthase dehydratase. J. Mol. Biol. 384, 941–953.
(2R,3R)-3-Hydroxy-2-Methylhexanoyl-N-Acetylcysteamine Thioester (17).
1H NMR (500 MHz, CDCl3) d 0.94 (t, 3H, J = 7.2 Hz, CH3), 1.22 (d, 3H, J =
6.9 Hz, CH3-C-C=O), 1.38–1.46 (m, 1H, one of C-CH2-C), 1.46–1.56 (m, 1H
one of C-CH2-C), 2.01 (s, 3H, CH3-C=O), 2.72–2.79 (m, 1H, CH-C=O),
3.01–3.10 (m, 2H, S-CH2), 3.43–3.55 (m, 2H, N-CH2), 3.72–3.78 (m, 1H, CH),
6.07 (br s, 1H, NH). ESI-MS expected mass: 248.3; observed mass: 248.4.
Keatinge-Clay, A.T., and Stroud, R.M. (2006). The structure of a ketoreductase
determines the organization of the b-carbon processing enzymes of modular
polyketide synthases. Structure 14, 737–748.
Khosla, C., Tang, Y., Chen, A.Y., Schnarr, N.A., and Cane, D.E. (2007).
Structure and mechanism of the 6-deoxyerythronolide B synthase. Annu.
Rev. Biochem. 76, 195–221.
Kodumal, S.J., Patel, K.G., Reid, R., Menzella, H.G., Welch, M., and Santi, D.V.
(2004). Total synthesis of long DNA sequences: synthesis of a contiguous
32-kb polyketide synthase gene cluster. Proc. Natl. Acad. Sci. USA 101,
15573–15578.
SUPPLEMENTAL INFORMATION
Supplemental Information includes two figures and three tables and can be
found with this article online at doi:10.1016/j.chembiol.2011.07.021.
Kwan, D.H., Sun, Y., Schulz, F., Hong, H., Popovic, B., Sim-Stark, J.C.C.,
Haydock, S.F., and Leadlay, P.F. (2008). Prediction and manipulation of the
stereochemistry of enoylreduction in modular polyketide synthases. Chem.
Biol. 15, 1231–1240.
ACKNOWLEDGMENTS
The authors thank the Robert A. Welch Foundation for funding (F-1712 for
A.T.K.-C. and F-1694 for D.R.S.).
Masamune, S., Hirama, M., Mori, S., Ali, S.A., and Garvey, D.S. (1981). Total
synthesis of 6-deoxyerythronolide B. J. Am. Chem. Soc. 103, 1568–1571.
Received: July 7, 2010
Revised: July 18, 2011
Accepted: July 19, 2011
Published: October 27, 2011
Myles, D.C., Danishefsky, S.J., and Schulte, G. (1990). Development of a fully
synthetic stereoselective route to 6-deoxyerythronolide B by reiterative appli-
cations of the Lewis acid catalyzed diene aldehyde cyclocondensation reac-
tion: a remarkable instance of diastereofacial selectivity. J. Org. Chem. 55,
1636–1648.
REFERENCES
Patel, R.N. (2008). Synthesis of chiral pharmaceutical intermediates by bio-
catalysis. Coord. Chem. Rev. 252, 659–701.
Bali, S., and Weissman, K.J. (2006). Ketoreduction in mycolactone biosyn-
thesis: insight into substrate specificity and stereocontrol from studies of
discrete ketoreductase domains in vitro. ChemBioChem 7, 1935–1942.
Raimundo, B.C., and Heathcock, C.H. (1995). Further studies on the anti-
selective aldol reaction of chiral imides. Synlett 12, 1213–1214.
Reid, R., Piagentini, M., Rodriguez, E., Ashley, G., Viswanathan, N., Carney, J.,
Santi, D.V., Hutchinson, C.R., and McDaniel, R. (2003). A model of structure
and catalysis for ketoreductase domains in modular polyketide synthases.
Biochemistry 42, 72–79.
Bali, S., O’Hare, H.M., and Weissman, K.J. (2006). Broad substrate
specificity of ketoreductases derived from modular polyketide synthases.
ChemBioChem 7, 478–484.
Bruice, T.C., Bruno, J.J., and Chou, W. (1963). Nucleophilic displacement
reactions at the thiol-ester bond of d-thiolvalerolactone. J. Am. Chem. Soc.
85, 1659–1669.
Sharma, K.K., and Boddy, C.N. (2007). The thioesterase domain from the
pimaricin and erythromycin biosynthetic pathways can catalyze hydrolysis of
simple thioester substrates. Bioorg. Med. Chem. Lett. 17, 3034–3037.
Caffrey, P.(2003). Conservedamino acidresiduescorrelatingwithketoreductase
Sherman, D.H., and Smith, J.L. (2006). Clearing the skies over modular polyke-
stereospecificity in modular polyketide synthases. ChemBioChem 4, 654–657.
tide synthases. ACS Chem. Biol. 1, 505–509.
Chen, A.Y., Schnarr, N.A., Kim, C.Y., Cane, D.E., and Khosla, C. (2006).
Extender unit and acyl carrier protein specificity of ketosynthase domains of
the 6-deoxyerythronolide B synthase. J. Am. Chem. Soc. 128, 3067–3074.
Siskos, A.P., Baerga-Ortiz, A., Bali, S., Stein, V., Mamdani, H., Spiteller, D.,
Popovic, B., Spencer, J.B., Staunton, J., Weissman, K.J., and Leadlay, P.F.
(2005). Molecular basis of Celmer’s rules: stereochemistry of catalysis by iso-
lated ketoreductase domains from modular polyketide synthases. Chem. Biol.
12, 1145–1153.
Crimmins, M.T., and Slade, D.J. (2006). Formal synthesis of 6-deoxyerythrono-
lide B. Org. Lett. 8, 2191–2194.
Evans, D.A., Bartroli, J., and Shih, T.L. (1981). Enantioselective aldol conden-
sations. 2. Erythro-selective chiral aldol condensations via boron enolates.
J. Am. Chem. Soc. 103, 2127–2129.
Smith, S., and Tsai, S.C. (2007). The type I fatty acid and polyketide synthases:
a tale of two megasynthases. Nat. Prod. Rep. 24, 1041–1072.
Stang, E.M., and White, M.C. (2009). Total synthesis and study of 6-deoxyer-
Evans, D.A., Kim, A.S., Metternich, R., and Novack, V.J. (1998). General strat-
egies toward the syntheses of macrolide antibiotics. The total syntheses of 6-
deoxyerythronolide B and oleandolide. J. Am. Chem. Soc. 120, 5921–5942.
ythronolide B by late-stage C-H oxidation. Nat. Chem. 1, 547–551.
Staunton, J., and Weissman, K.J. (2001). Polyketide biosynthesis: a millennium
review. Nat. Prod. Rep. 18, 380–416.
Gilbert, I.H., Ginty, M., O’Neill, J.A., Simpson, T.J., Staunton, J., and Willis, C.L.
(1995). Synthesis of b-keto and a,b-unsaturated N-acetylcysteamine thioest-
ers. Bioorg. Med. Chem. 5, 1587–1590.
Stinear, T.P., Mve-Oblang, A., Small, P.L., Frigui, W., Pryor, M.J., Brosch, R.,
Jenkin, G.A., Johnson, P.D., Davies, J.K., Lee, R.E., et al. (2004). Giant
Chemistry & Biology 18, 1331–1340, October 28, 2011 ª2011 Elsevier Ltd All rights reserved 1339

Chemistry & Biology
Biocatalysis with Modular PKS Ketoreductases
plasmid-encoded polyketide synthases produce the macrolide toxin of
Valenzano, C.R., You, Y.O., Garg, A., Keatinge-Clay, A., Khosla, C., and Cane,
D.E. (2010). Stereospecificity of the dehydratase domain of the erythromycin
polyketide synthase. J. Am. Chem. Soc. 132, 14697–14699.
Mycobacterium ulcerans. Proc. Natl. Acad. Sci. USA 101, 1345–1349.
Tokuyama, H., Yokoshima, S., Yamashita, T., Lin, S., Li, L., and Fukuyama, T.
(1998). Facile palladium-mediated conversion of ethanethiol esters to alde-
hydes and ketones. J. Braz. Chem. Soc. 9, 381–387.
Wong, C., Drueckhammer, D., and Sweers, H.M. (1985). Enzymatic vs.
fermentative synthesis: thermostable glucose dehydrogenase catalyzed
regeneration of NAD(P)H for use in enzymatic synthesis. J. Am. Chem. Soc.
107, 4028–4031.
Valenzano, C.R., Lawson, R.J., Chen, A.Y., Khosla, C., and Cane, D.E. (2009).
The biochemical basis for stereochemical control in polyketide biosynthesis.
J. Am. Chem. Soc. 131, 18501–18511.
Zheng, J., Taylor, C.A., Piasecki, S.K., and Keatinge-Clay, A.T. (2010).
Structural and functional analysis of A-type ketoreductases from the ampho-
tericin modular polyketide synthase. Structure 18, 913–922.
1340 Chemistry & Biology 18, 1331–1340, October 28, 2011 ª2011 Elsevier Ltd All rights reserved