Poly specific trans -Acyltransferase machinery revealed via engineered acyl-coa synthetases

Article abstract of DOI:10.1021/cb3003489

Polyketide synthases construct polyketides with diverse structures and biological activities via the condensation of extender units and acyl thioesters. Although a growing body of evidence suggests that polyketide synthases might be tolerant to non-natural extender units, in vitro and in vivo studies aimed at probing and utilizing polyketide synthase specificity are severely limited to only a small number of extender units, owing to the lack of synthetic routes to a broad variety of acyl-CoA extender units. Here, we report the construction of promiscuous malonyl-CoA synthetase variants that can be used to synthesize a broad range of malonyl-CoA extender units substituted at the C2-position, several of which contain handles for chemoselective ligation and are not found in natural biosynthetic systems. We highlighted utility of these enzymes by probing the acyl-CoA specificity of several trans-acyltransferases, leading to the unprecedented discovery of poly specificity toward non-natural extender units, several of which are not found in naturally occurring biosynthetic pathways. These results reveal that polyketide biosynthetic machinery might be more tolerant to non-natural substrates than previously established, and that mutant synthetases are valuable tools for probing the specificity of biosynthetic machinery. Our data suggest new synthetic biology strategies for harnessing this promiscuity and enabling the regioselective modification of polyketides.

Full text of DOI:10.1021/cb3003489

Articles
pubs.acs.org/acschemicalbiology
Poly Specific trans-Acyltransferase Machinery Revealed via
Engineered Acyl-CoA Synthetases
Irina Koryakina,† John McArthur,† Shan Randall,‡ Matthew M. Draelos,† Ewa M. Musiol,§
David C. Muddiman,‡ Tilmann Weber,§ and Gavin J. Williams*,†
†Department of Chemistry and ‡W. M. Keck FT-ICR Mass Spectrometry Laboratory, North Carolina State University, Raleigh, North
Carolina, United States
§Eberhard-Karls-Universitat Tubingen, Interfakultares Institut fur Mikrobiologie und Infektionsmedizin,
̈
̈
̈
̈
Mikrobiologie/Biotechnologie, Tubingen, Germany
̈
S
* Supporting Information
ABSTRACT: Polyketide synthases construct polyketides with
diverse structures and biological activities via the condensation of
extender units and acyl thioesters. Although a growing body of
evidence suggests that polyketide synthases might be tolerant to
non-natural extender units, in vitro and in vivo studies aimed at
probing and utilizing polyketide synthase specificity are severely
limited to only a small number of extender units, owing to the lack
of synthetic routes to a broad variety of acyl-CoA extender units.
Here, we report the construction of promiscuous malonyl-CoA
synthetase variants that can be used to synthesize a broad range of
malonyl-CoA extender units substituted at the C2-position, several
of which contain handles for chemoselective ligation and are not found in natural biosynthetic systems. We highlighted utility of
these enzymes by probing the acyl-CoA specificity of several trans-acyltransferases, leading to the unprecedented discovery of
poly specificity toward non-natural extender units, several of which are not found in naturally occurring biosynthetic pathways.
These results reveal that polyketide biosynthetic machinery might be more tolerant to non-natural substrates than previously
established, and that mutant synthetases are valuable tools for probing the specificity of biosynthetic machinery. Our data suggest
new synthetic biology strategies for harnessing this promiscuity and enabling the regioselective modification of polyketides.
olyketides are a large class of pharmaceutically relevant
propyl, hexyl, and isobutylmalonyl-CoA (8−11, respectively)
Pnatural products with wide-ranging biological activities that (Figure 1) are available to a subset of PKSs via the reductive
include antimicrobial, anticancer, and immunosuppressant.1
carboxylation of α,β-unsaturated acyl-CoA precursors by
crotonyl-CoA carboxylase/reductase (CCR) homologues.10−12
Moreover, polyketides constitute a significant fraction of
The selection and incorporation of extender units by PKSs
contributes to significant portions of polyketide structure, in
addition to the vast structural diversity across the polyketide
family, as illustrated by the structures of concanamycin A
aglycone (12), zwittermicin A (13), salinosporamides (14a−
14d) and the FK506 series (e.g., 15a and 15b) (Figure 1).
Moreover, variation of extender unit selection and incorpo-
ration at specific positions in polyketide backbones has been
shown to directly modulate biological activities of polyketides
produced by natural PKSs.9,13
Even though Nature provides a modest selection of potential
PKS extender units, most polyketide-producing organisms
provide biosynthetic routes to only a relatively small number of
suitable acyl-CoAs. Subsequently, there is a growing body of
evidence that PKSs can tolerate non-natural extender units that
are not normally provided by the producing host.14,15 However,
approved and top-selling drugs, annual sales of which total
$20 billion.2,3 The broad and potent biological activities of
polyketides are determined by their chemical structures, the
carbon backbones of which are biosynthesized via iterative
Claisen condensations between activated malonate extender
units and acyl thioesters, catalyzed by polyketide synthases
(PKSs).4 A modest variety of malonate extender units
substituted at the C2 position are cumulatively available to
polyketide biosynthetic pathways,5,6 although producing
organisms typically only possess biosynthetic routes to a
small number of extender units, depending on the structures of
the corresponding polyketides. The most common PKS
extender units include malonyl-Coenzyme A (CoA) (1),
methylmalonyl-CoA (2), and ethylmalonyl-CoA (3) (Figure
1). A smaller number of PKSs use specialized extender units
that are functionalized directly on standalone acyl carrier
proteins (ACPs) and include methoxymalonyl-ACP7 (4),
hydroxymalonyl-ACP8 (5), aminomalonyl-ACP8 (6), and
allylmalonyl-ACP9 (7) (Figure 1). Additionally, it has recently
emerged that several extender units, including chloroethyl,
Received: July 11, 2012
Accepted: October 19, 2012
Published: October 19, 2012
© 2012 American Chemical Society
200
dx.doi.org/10.1021/cb3003489 | ACS Chem. Biol. 2013, 8, 200−208

ACS Chemical Biology
Articles
involved in polyketide biosynthesis, revealing unprecedented
activities toward non-natural extender moieties. The ability to
design acyl-CoA synthetases with new specificities is a first step
toward using synthetic biology to create promiscuous in vitro
and in vivo enzymatic systems for generating non-natural
polyketide analogues.
RESULTS AND DISCUSSION
■
Structure-Guided Mutagenesis. Recently, following
preliminary screening of small MatB saturation mutagenesis
libraries against methylmalonate (17) and allylmalonate (20)
(Figure 2b), we reported18 the identification of two MatB
single mutants, T207S and M306I, which displayed improve-
ments in catalytic efficiency toward two extenders, 17 and 20.
In addition, both mutants showed improvements toward
ethylmalonate (18), which was not originally screened for.
Here, in order to expand the potential scope of MatB-based
extender unit biosynthesis and to further test our ability to
engineer the promiscuity of MatB, we designed a panel of seven
additional malonates substituted at the C2-position with diverse
functionality (Figure 2b). This panel included members that
were reported15 to be either very poor substrates for the wild-
type enzyme (18, 21, 22; which represent <10% relative
reaction rate with malonate 16) or moderately good substrates
(17, 20; 10−20% relative rate with 16) or were not previously
examined (19, 23−26). This panel represents broad structural
diversity and includes analogues functionalized with chemical
handles for downstream chemoselective ligation (19, 20, 26),
which could prove useful for further diversification strategies
should these novel extender units become incorporated into
polyketides. In addition, three members (20, 24, 25) represent
extender units that are provided to PKSs as ACP-linked
substrates in natural biosynthetic systems,5,9 and acyl-CoA
routes to such extenders are not currently known. Moreover,
three members (19, 23, 26) are currently completely unknown
to natural biosynthetic systems. Panel members were either
commercially available (16−18, 20−24), prepared from the
corresponding commercially available diester (19, 25), or
synthesized from diethyl cyclopropyl-1,1-dicarboxylate in the
case of 26. To assess promiscuity of wild-type MatB and the
mutants T207S and M306I, kinetic parameters (kcat and
apparent Km) for each analogue were determined (Table 1,
Supplementary Figure S1 and Supplementary Table S1) using a
robust HPLC assay for acyl-CoA formation.18 In each case,
acyl-CoA product identity was confirmed by LC−MS analysis
(Supplementary Table S2). Of the 11 malonates tested (16−
26), only five (16−20) resulted in formation of the desired
acyl-CoA with wild-type MatB (Table 1). Furthermore, the
wild-type enzyme clearly favors malonate 16 out of the set and
is 2.4-, 93-, 10000-, and 38-fold less efficient (kcat/Km) with 17,
18, 19, and 20, respectively. Thus, although wild-type MatB
clearly has some capacity to turnover malonate analogues,
promiscuity is insufficient to process the entire set. Kinetic
parameters for our previously described T207S and M306I
were also determined using the entire panel of analogues
(Table 1, Supplementary Figure S1 and Supplementary Table
S1). As we previously described,18 both mutants display
significant shifts in specificity toward 18 and 20. Notably,
compared to wild-type MatB, the butyl (22), methoxy (25),
and azidoethyl (26) analogues are additional substrates for
T207S, albeit very poor ones, whereas azidoethyl (26) is also a
substrate for M306I. Analogues 21 and 23 are non-substrates
for both wild-type MatB and the mutants T207S and M306I.
Figure 1. Variety of naturally occurring PKS extender units and
contribution to polyketide structure. (a) Extender units are
biosynthesized as acyl-CoAs (1−3), ACP-linked acyl thioesters (4−
7), or from CRR-mediated reductive carboxylation of α,β-unsaturated
acyl-CoA precursors (8−11). (b) Contribution of extender unit C2-
side chains to the structures representative polyketides is highlighted
red. Positions derived from C2-H of malonate extender units are
shown as red dots.
in vitro and in vivo studies aimed at investigating PKS specificity
are limited in scope and number,14,16 owing at least in part to a
lack of easily synthesized extender units. We proposed to build
on the previously described15,17 partial promiscuity of the
malonyl-CoA synthetase MatB from Rhizobium trifolii (Figure
2). This enzyme displays low-level activity toward several
alkylmalonates, and we sought to produce a robust biocatalyst
for the activation of malonates modified at the C2-position with
a broad range of substituents. We postulated that a truly
promiscuous synthetase would prove useful for probing the in
vitro acyl-CoA specificity of PKSs and related biosynthetic
machinery. Consequently, sufficiently promiscuous MatB
mutants could also form the basis of prototype bacterial strains
for in vivo extender unit generation and subsequent installation
into polyketides.
Here, we report structure-based engineering of MatB to
produce variant enzymes capable of synthesizing a broad range
of acyl-CoAs, including several that are not produced by the
wild-type enzyme. This panel of extender units was used to
probe the specificity of trans-acyltransferases (trans-ATs)
201
dx.doi.org/10.1021/cb3003489 | ACS Chem. Biol. 2013, 8, 200−208

ACS Chemical Biology
Articles
Figure 2. Engineering the acyl-CoA specificity of MatB. (a) Active site architecture of MatB. The Streptomyces coelicolor MatB crystal structure (cyan
carbons, PDB code 3NYQ) is shown overlaid with a homology model of the Rhizobium trifolii MatB (pale green, SWISS-MODEL38 using 3NYQ as
template), illustrating the intimate role T207 and M306 play in extender side-chain specificity. Residue labels are those of S. coelicolor/R. trifolii,
respectively. Methylmalonyl-CoA is shown with green carbons, and the C2-methyl is indicated with an asterisk. (b) Reactions catalyzed by wild-type
and mutant MatB. The natural substrate for MatB is 16.
a
Table 1. Specificity Constants (kcat/Km) (mM−1 s−1) of Wild-Type and Mutant MatB with 16−26
malonate analogue
b
enzyme
16
17
18
19
20
21
22
23
24
25
26
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
WT
50.3
3.3
21.0
16.6
13.3
35.5
0.9
0.54
21.4
26.9
43.6
4.2
0.005
0.5
1.3
49.5
11.3
40.8
0.8
ND
ND
1.5
3.5
1.4
ND
ND
1.5
2.7
2.3
2.5
0.6
1.6
ND
0.4
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
T207G
0.03
21.1
5.5
c
c
c
c
c
T207A
16.5
44.4
4.5
0.9
ND
ND
ND
ND
4.8
ND
ND
ND
ND
c
c
c
T207S
0.4
0.004
2.2
c
c
c
c
M306V
ND
ND
0.06
0.08
58
c
M306I
13.4
4.7
29.0
30.9
53.9
14.3
26.5
22.5
219
0.01
0.02
0.9
2
6.2
ND
c
c
T207G/M306V
T207G/M306I
T207A/M306V
T207A/M306I
T207S/M306V
T207S/M306I
ND
ND
1.7
3.6
1.6
2.8
0.31
0.78
5.5
15.8
48
0.32
0.6
ND
c
8.7
0.004
0.09
0.1
24.5
6.8
c
2.4
0.2
ND
ND
ND
ND
c
c
c
5.2
0.1
0.02
2.6
4.0
0.02
22
c
c
1.0
2.1
ND
ND
1.2
10.9
92
0.02
0.08
22.7
a
Steady state kinetic parameters were determined as described in the Experimental Section, using a fixed concentration of CoA and ATP, and varied
concentration of malonate analogue. See Supplementary Figure S1 and Supplementary Table S1 for Michaelis−Menten plots, numerical data, and
b
standard deviations. See Supplementary Table S2 for product characterization by MS analysis. Activity toward 24 was detected, but only under
c
unique assay conditions and only when wild-type MatB was used. Nondetected (ND) refers to an estimated detection limit <0.004 mM−1 s−1.
Cumulatively, this data provides evidence that mutation of the
MatB active site could provide activity toward malonate
analogues that are not detectable substrates for the wild-type
enzyme.
type or previously identified single mutants (Table 1,
Supplementary Figure S1 and Supplementary Table S1). In
contrast, the substrate specificity of T207A and M306V were
little changed in comparison to the original mutants T207S and
M306I (Table 1, Supplementary Figure S1 and Supplementary
Table S1). Not completely satisfied that none of these single
mutants were capable of utilizing the entire panel of analogues,
nor satisfied with the overall level of catalytic proficiency of
these single mutants with some substrates, we next constructed
every possible double mutant combination of T207G, T207S,
T207A, M306I, and M306V and determined kinetic parameters
for the purified enzymes. These double mutants were
significantly improved in comparison to the wild-type and
single mutant MatB for several reasons. First, both double
mutants that contained glycine at position 207 displayed
significantly improved activity with 21, 23, and 26 (Table 1,
Supplementary Figure S1 and Supplementary Table S1). For
example, T207G/M306V was 160- and ∼3-fold improved
To fulfill our goal of generating MatB variants for in vivo and
in vitro production of a broad range of extender units, we set
out to further expand the specificity of MatB toward the entire
malonate panel (18−26). Accordingly, we sought to broaden
our search by considering other T207X and M306X members
that were improved compared to the wild-type enzyme with
18/20 but also less active than T207S and M306I with these
substrates. Subsequent re-analysis of our original screening
data,18 DNA sequencing, enzyme purification, and HPLC assay
revealed three additional mutants (T207A, T207G, and
M306V) that were improved compared to wild-type MatB
toward 18/20. Gratifyingly, kinetic analysis revealed that one of
these mutants (T207G) displayed activity with two analogues
(21, 23) that were not detectable substrates for either the wild-
202
dx.doi.org/10.1021/cb3003489 | ACS Chem. Biol. 2013, 8, 200−208

ACS Chemical Biology
Articles
Figure 3. Conversion rates of wild-type MatB and mutant T207G/M306I toward a panel of malonate analogues. Absence of bar indicates no
detectable conversion (estimated detection limit 0.2% conversion). Numbers above bars indicate fold-improvement from wild-type enzyme, using an
estimated minimal detection limit of 0.2% conversion. See Experimental Section for assay conditions and detection method.
toward 21 and 26, respectively, in terms of kcat/Km, compared
to the most promiscuous single mutant, T207G (Table 1 and
Supplementary Table S1). In addition, T207G/M306I was 290-
and 2.5-fold improved toward 21 and 18, respectively, in terms
of kcat/Km, compared to the parent T207G (Table 1). In
contrast, double mutants that contained T207S (the most
active position-207 mutant from saturation screening) dis-
played poorer activity toward 21, 23, and 26 compared to
T207G/M306I (Table 1, Supplementary Figure S1 and
Supplementary Table S1). Gratifyingly though, T207S/M306I
was fully 405- and 71-fold improved toward 18 and 20,
respectively, compared to wild-type MatB (Table 1). These
data suggest that significant changes in substrate specificity can
be obtained via relatively subtle alterations to the MatB active
site. Accordingly, even relatively low levels of activity (e.g.,
toward 19 and 23) may provide suitable starting points for
further improvement by directed evolution. Moreover,
identification of a set of MatB mutants that can utilize every
substrate tested underscores the effectiveness of originally
screening MatB mutant libraries against only 1 or 2 substrates.
An additional benefit revealed through site-directed recombi-
nation is the concomitant reduction in catalytic efficiency with
16 and 17. In particular, the double mutants T207G/M306V
and T207G/M306I do not display detectable activity toward 16
and are fully 114- and 46-fold poorer, in terms of kcat/Km,
toward 17, compared to the best mutant from our earlier
preliminary study, T207S (Table 1 and Supplementary Table
S1). This could be highly significant for future in vivo
applications of the engineered MatB, given that endogenous
16/17 could otherwise be rapidly consumed by MatB,
providing little flux through non-natural malonate analogues.
To illustrate the synthetic capabilities of the wild-type and
MatB T207G/M306I, a simple end-point HPLC assay was used
to compare product conversion after incubation with each
malonate analogue. Most notably, even after extended
incubation, wild-type MatB was unable to convert 21−23 and
26 to the corresponding acyl-CoA (Figure 3, Supplementary
Figure S2 and Supplementary Table S2). In contrast, MatB
T207G/M306I was able to utilize 21−23 and 26 with synthetic
conversions ranging from 62% to 92%, in addition to
conversion of 16−20 and 25, with similarly high efficiency.
Only 16, 17, and 24 were used more efficiently by wild-type
MatB. These conversions are in agreement with that expected
from the kinetic parameters described above.
Cumulatively, this set of mutagenesis afforded MatB variants
that can utilize every malonate analogue tested. The robust
synthesis of a broad panel of malonate analogues prompted us
to investigate whether these acyl-CoAs could be utilized to
discover novel promiscuities displayed in vitro by polyketide
synthetic systems.
Extender Unit Specificity of trans-Acyltransferases.
Acyltransferase (AT) domains of typical type I PKSs are
responsible for selecting extender units for incorporation into
polyketides and are an integral part of each PKS module. A
variant of this arrangement involves a freestanding AT enzyme
that is physically separated from the PKS and operates in trans.
Because several such “standalone” trans-ATs utilize small
molecule acyl-CoAs, versus ACP-linked extender units, they
present interesting possibilities for combinatorial biosynthesis
of regioselectively modified polyketide analogues.19,20 To date,
however, biochemical studies have only identified three unique
trans-AT acyl-CoA specificities (1−3), while substrate specific-
ity studies of trans-ATs and the related transacyltransferases
from fatty acid biosynthesis20−22 that use acyl-CoAs as extender
units23−26 have been limited entirely to the commerically
available extender substrates 1/2, and chemically synthesized 3.
These studies have suggested that trans-ATs are highly
stringent to their cognate acyl-CoA (e.g., 1-3). Subsequently,
in vivo combinatorial biosynthesis efforts involving trans-ATs
have been restricted to only conservative alterations to
203
dx.doi.org/10.1021/cb3003489 | ACS Chem. Biol. 2013, 8, 200−208

ACS Chemical Biology
Articles
Figure 4. Natural function and assay scheme of DSZS, MCAT, and KirCII. (a) Natural products 35−37 involve the trans-ATs DSZS, MCAT, and
KirCII, respectively. Contribution of each trans-AT to the final product structure is highlighted red. (b) Reaction scheme for each representative
trans-AT with the corresponding natural acyl-CoA extender unit.
polyketide structure,20 and the true potential of trans-ATs for
regioselective polyketide modification is unknown. We there-
fore used our mutant MatB synthesized acyl-CoA panel to
more fully probe the specificity of three representative trans-
ATs, the malonyl-CoA:ACP transacylase (MCAT) from
Streptomyces coelicolor,22 the malonyl-CoA specific DSZS-AT
from the disorazole synthase (DSZS),27 and the ethylmalonyl-
CoA specific KirCII from kirromycin biosynthesis25 (Figure
4a). We hypothesized that KirCII would display some
promiscuity toward acyl-CoAs that have C2-substitutions larger
than the ethyl side chain of 3 for two reasons: First, both
MCAT and DSZS-AT utilize 1 and discriminate significantly
against 2 and are therefore unlikely to accept extender units
with even larger side chains. Second, in the natural kirromycin
producer strain, KirCII does not accept 1/2 as substrates, but
there is presumably no requirement for KirCII to discriminate
against malonate extender units with C2-substituents larger that
of 3.
To test this hypothesis, each trans-AT was expressed in E. coli
as hexahistidine-tagged proteins and purified by nickel-affinity
chromatography. For MCAT and DSZS-AT, ACP6 from the 6-
deoxyerythronolide B synthase (ACP6DEBS) was used as the
acceptor substrate,27 while ACP5 from kirromycin biosynthesis
(ACP5Kir) was used as the cognate carrier protein for KirCII
(Figure 4b).25 Carrier proteins were converted to their holo-
ACPs in vivo by Sfp-mediated phosphopantetheinylation and
subsequently purified by metal chelation chromatography,
ensuring removal of Sfp, which is not His6-tagged. Complete
conversion to the holo enzyme was confirmed by LC−MS
analysis (Supplementary Figure S3). The extender unit
specificity of each trans-AT was determined using an end-
point assay. Briefly, each trans-AT was incubated with the
corresponding fully activated holo-ACP and acyl-CoA (1−3,
27−34). Acylation of the ACPs and percent conversions were
determined by Fourier-transform ion cyclotron resonance (FT-
ICR) mass spectrometry-based assay of intact proteins present
in the reaction mixtures (Figure 5a,b and Supplementary Table
S3). Negative controls that lacked each trans-AT were also
analyzed in the same way, in order to identify possible
background acylation (Figure 5a,b and Supplementary Table
S3). A high level of stringency was chosen for the confirmation
of trans-AT substrates, whereby acyl-CoAs were considered
substrates only when the conversion from holo-ACP was
completely dependent on the presence of trans-AT (Figure 5a,b
and Supplementary Table S3). Subsequently, out of 11
extender units tested, only malonyl-CoA 1 led to the
identification of the corresponding acylated-ACP (Figure 5a,b
and Supplementary Table S3) when MCAT or DSZS were
used in these assays. Although the specificity of MCAT or
DSZS has not previously been probed beyond 1/2, this
stringent malonyl specificity is not completely surprising, given
the requirement of these enzymes to at least discriminate
against methylmalonyl-CoA. In stark contrast to MCAT and
DSZS, KirCII utilized 4 out of 11 acyl-CoA extender units,
including the propargyl (27), allyl (28), and azidoethyl (34)
analogues (Figure 5a,b), in addition to the established natural
substrate (3). For each successful reaction (involving acyl-CoAs
1, 3, 27, 28, and 34), control reactions that lacked trans-AT
failed to produce an intact mass corresponding to acyl-transfer
(Figure 5a,b, Supplementary Figure S4 and Supplementary
Table S3), ruling out self-acylation by the ACP or trans-
acylation by a contaminant. Although the KirCII-catalyzed
acylation of ACP5Kir with 31 was greater than background, we
cannot rule out the possibility that 31 does not require the
trans-AT, given the error of these measurements (Figure 5b).
Notably, this discovery stands as the first example of a
polyspecific trans-AT that can utilize extender unit acyl-CoAs
other than 1−3 and demonstrates for the first time trans-AT
based utilization of extender units not found in natural
biosynthetic pathways (27, 28, 34).
These data suggest that the tolerance of some trans-ATs
toward non-natural extender units may be considerably higher
than previously considered. Presumably, given that KirCII
utilizes ethylmalonyl-CoA extender units with the largest C2-
204
dx.doi.org/10.1021/cb3003489 | ACS Chem. Biol. 2013, 8, 200−208

ACS Chemical Biology
Articles
Figure 5. In vitro analysis of MCAT, DSZS and KirCII extender unit specificity. (a) LC-FTICR-MS analysis of successful trans-AT catalyzed
conversions of holo-ACP to acyl-ACP. Each spectrum represents the 13C distribution of the indicated charge state. Each boxed set of spectra shows
reactions in the presence (upper) and absence (lower) of each trans-AT with the indicated acyl-CoA. Reactions that did not produce the expected
mass ion are not shown for brevity. (b) Relative specificity of DSZS, MCAT, and KirCII toward each extender unit. Activity of each wild-type trans-
AT toward the natural extender unit is set to 100%. Each conversion is corrected using the background acylation rate determined in the absence of
trans-AT. Acyl-CoAs are considered substrates for the trans-AT only when the background rate in the absence of trans-AT is zero (see
Supplementary Table S3 for complete MS data). The estimated error of these measurements is 25% of the conversion based on the reaction of
KirCII with 27, which was performed in triplicate. MMA = mass measurement accuracy. See Experimental Section for assay and detection
conditions.
205
dx.doi.org/10.1021/cb3003489 | ACS Chem. Biol. 2013, 8, 200−208

ACS Chemical Biology
Articles
substituent available to the producing host, this enzyme does
not require discrimination against extender units with larger
side chains. Consequently, probing the extender unit specificity
of other trans-ATs and PKS modules that transfer malonyl-CoA
derivatives with C2-substituents that are the largest provided by
the host organism might also reveal similar promiscuity.
Whereas such candidate trans-ATs are currently very rare,23,28
PKSs that, for example, use the extender ethylmalonyl-
CoA7,29−31 are unlikely to require discrimination against larger
extender units (such ATs would nevertheless require
discrimination against 1/2). Any such trans/cis-ATs and the
corresponding PKSs might provide valuable promiscuous
machinery for future combinatorial biosynthesis efforts.
Further, several newly identified substrates for the engineered
MatB mutants contain chemical handles for chemoselective
ligation (19, 20, 26) and are not found in natural extender unit
biosynthetic pathways (19, 23, 25, 26). Beyond these results,
we expect to be able to extend our enzyme engineering strategy
to further improve activity of MatB toward analogues that
remain relatively poor substrates, such as the phenyl analogue
23. Additionally, the enzyme engineering strategy described
here is expected to overcome the stringent substrate specificity
of other naturally occurring acyl-CoA ligases.34 The creation of
promiscuous acyl-CoA synthetases for the chemo-enzymatic
synthesis of natural and non-natural extender units represents a
considerable advance in our ability to probe the specificity and
activity of a broad range of polyketide synthases and related
machinery. For example, the availability of a panel of diverse
extender units enabled unprecedented poly specificity of a
trans-AT (KirCII) to be discovered. In summary, our results
suggest that mutant synthetases are valuable tools for probing
the specificity of polyketide synthase biosynthetic machinery
and for future synthetic biology efforts aimed at engineering
PKSs for the synthesis of non-natural analogues.
Implications for in Vivo Combinatorial Biosynthesis
Strategies. Although acyl-CoA extender units can be chemi-
cally synthesized via expensive and inefficient synthetic routes,
acyl-CoAs are not cell-permeable32,33 and are therefore
ultimately not suitable in vivo substrates for polyketide
biosynthesis. Yet, natural biosynthetic routes to extender unit
acyl-CoAs can be lengthy, sometimes involve several
cofactors,5,6 and have yet to be shown to be suitable targets
for protein engineering. Accordingly, beyond the clear
demonstration of MatB mutants as important tools for probing
in vitro PKS specificity, these mutant acyl-CoA synthetases are
likely to prove useful for providing extender units for in vivo
polyketide diversification. Indeed, MatB T207G/M306I shows
improvements in both kcat and Km toward non-natural extender
units, and the wild-type MatB-synthesized products are utilized
by the erythronolide PKS machinery without the requirement
of a suitable promiscuous epimerase.15 The newly discovered
extender unit specificities described here indicate that trans-
ATs, and possibly by extension cis-ATs of the canonical PKSs,
could be more versatile with respect to extender unit utilization
than previously suggested, and if effectively exploited could
offer new strategies to diversify polyketide structure. Thus, the
promiscuity of KirCII could be harnessed to produce
regioselectively modified kirromycin analogues if endogenous
ethylmalonyl-CoA biosynthesis can be knocked out and
replaced with our mutant MatB. Moreover, if KirCII displays
a level of carrier protein promiscuity similar to that of other
biochemically characterized trans-ATs,20,26,27 this enzyme or
mutants thereof may provide routes to regioselective
modification of other polyketides. In addition, trans-ATs, as
opposed to their cis-acting counterparts, may be good targets
for directed evolution studies aimed at further broadening or
altering extender unit specificity given their free-standing
nature. Indeed, our demonstration that KirCII can utilize
extender units modified with handles for chemoselective
ligation suggests a potential strategy for screening or selecting
novel enzyme activities from large libraries of trans-AT variants.
Conclusions. In an effort to overcome the limited
specificity of acyl-CoA synthetases, we have used structure-
guided saturation mutagenesis and site-directed recombination
to improve the promiscuity of the malonyl-CoA synthetase
MatB. Notably, the substrate specificity of MatB was
considerably expanded by only a small number of amino acid
mutations to include five substrates not utilized by the wild-
type enzyme (Table 1). Improvements in catalytic efficiency of
mutants compared to the wild-type MatB were as high as
14,500-fold (e.g., with 26), and catalytic efficiencies of the
mutant MatBs with several of the analogues were as high as that
of the wild-type enzyme with the natural substrate 16 and in
some cases were even higher (e.g., T207S/M306I with 18, 20).
EXPERIMENTAL SECTION
■
General. Unless otherwise stated, all materials and reagents were of
the highest grade possible and purchased from Sigma (St. Louis, MO).
Isopropyl β-D-thiogalactoside (IPTG) was from Calbiochem. Bacterial
strain Escherichia coli (E. coli) BL21(DE3) pLysS competent cells was
from Promega. Primers were ordered from Integrated DNA
Technologies. Plasmid pET28a-MatB was as previously described.18
MatB Synthetic Assay. In vitro enzyme assays were performed in
50 μL of reaction mixture containing 100 mM sodium phosphate (pH
7), MgCl2 (2 mM), ATP (4 mM), coenzyme A (4 mM), malonate or
analogue 16−26 (16 mM), and wild-type or mutant MatB (10 μg) at
25 °C. Aliquots were removed after 3 h of incubation, quenched with
an equal volume of ice-cold methanol, and centrifuged at 10,000g for
10 min and cleared supernatants were used for HPLC analysis on a
Varian ProStar HPLC system. A series of linear gradients was
developed from 0.1% TFA (A) in water to methanol (HPLC grade, B)
using the following protocol: 0−32 min, 80% B; 32−35 min, 100% A.
The flow rate was 1 mL/min, and the absorbance was monitored at
254 nm using Pursuit XRs C18 column (250 mm × 4.6 mm, Varian
Inc.). The malonate analogue and the acyl-CoA product HPLC peak
areas were integrated, and the conversion (%) was calculated as a
percent of the total peak area. Product identity was confirmed by LC−
MS (Supplementary Table S2).
Mass Spectrometry Analysis of Acyl-CoAs. Samples were
subjected to negative-ESI LC−MS on a Thermo TSQ Quantum
Discovery MAX connected to a UV−vis diode array detector with a
Waters BEH C18, 2.1 mm × 50 mm, 1.7 μm particle column. A series
of linear gradients was developed from water/1 mM ammonium
formate (pH 5.3) (A) to methanol (B) using the following protocol:
0−10 min, 3−80% B; 10−11 min, 80−95% B; 11−13 min, 95% B;
13−14 min, 95−5% B; 14−17.5 min, 5% B.
trans-AT Assays. DEBS ACP6 and Kir ACP5 were expressed in E.
coli and purified to homogeneity as completely phosphopantetheiny-
lated holo-ACPs using the B. subtilis PPTase Sfp. An aliquot of ACP
(30 μL) was added to 45 μL of 50 mM sodium phosphate (pH 6.8)
(MCAT) or 50 mM sodium phosphate (pH 7.2) (DSZS) or 50 mM
Tris-HCI (pH 7.5) containing 50 mM MgCI2 (KirCII) containing 10
μM KirCII or 10 μM DSZS or 10 μM MCAT, and 5 mM acyl-CoA
(generated by suitable mutant MatB). Reactions were incubated at 25
°C for 15 min and analyzed by LC-FTICR-MS analysis as described
below.
LC-FTICR-MS Analysis of trans-AT Reactions. Reaction
mixtures were separated using an Eksigent 1D+ nano-LC system
utilizing a vented column configuration.35 Analytical and trap columns
were packed in-house with Magic C8 (5 μm particle, 300 Å pore size;
206
dx.doi.org/10.1021/cb3003489 | ACS Chem. Biol. 2013, 8, 200−208

ACS Chemical Biology
Articles
Microm BioResources) stationary phase to approximately 15 and 5
cm, respectively. PicoFrit (75 μm i.d.) and IntegraFrit (100 μm i.d.)
capillaries were purchased from New Objective (Woburn, MA). LC
solvents were purchased from Burdick and Jackson (Muskegon, MI).
Mobile phase A consisted of 98% water, 2% acetonitrile, and 0.2%
formic acid, and mobile phase B consisted of 98% acetonitrile, 2%
water, and 0.2% formic acid. A 50 ng portion of total protein was
injected and desalted at 2 μL/min before switching in-line with the
analytical column at a flow rate of 250 nL/min via a 10-port valve. The
gradient was held at 30% B for 1 min before ramping to 37% B over
the next 19 min. The gradient was then adjusted to 95% B in 1 min,
holding there for the next 9 min. Finally, the gradient was adjusted to
30% in 1 min and held for an additional 9 min for re-equilibration.
Mass measurements were performed on a 7T LTQ-FT-ICR Ultra
(ThermoFisher Scientific) mass spectrometer operating at 50,000fwhm
resolving power at m/z = 400. Data acquisition was set to one
broadband scan event in the ICR cell. Spectra were acquired with an
automatic gain control setting of 1 × 106 ions using one microscan
with a maximum ionization time of 500 ms. Tube lens voltage was set
to 150 V, and the capillary temperature was set to 200 °C.
Determination of monoisotopic masses for intact proteins has been
previously described.36,37 Using a similar approach in this study,
monoisotopic masses of identified proteins were determined by
overlaying the theoretical isotopic distribution of the most abundant
charge state with the experimental isotopic distribution obtained from
summed spectra. A Java-based algorithm (Isotopic Pattern Calculator,
v1.0) provided as freeware from Pacific Northwest National
Laboratory (Richland, WA) was used in conjunction with the
molecular formulas of the analytes to generate the theoretical isotopic
distributions. Multiply charged spectra were mass transformed to a
neutral monoisotopic mass to determine mass measurement accuracy.
Although the authors are aware of possible electrospray ionization
biases due to differences in hydrophobicity, extracted ion chromato-
grams corresponding to the holo-ACP and acyl-CoA product were
generated and integrated for estimating percent conversion using the
following formula: (Areaacyl‑ACP/Areaholo‑ACP + Areaacyl‑ACP).
(4) Hertweck, C. (2009) The biosynthetic logic of polyketide
diversity. Angew. Chem., Int. Ed. 48, 4688−4716.
(5) Chan, Y. A., Podevels, A. M., Kevany, B. M., and Thomas, M. G.
(2009) Biosynthesis of polyketide synthase extender units. Nat. Prod.
Rep. 26, 90−114.
(6) Wilson, M. C., and Moore, B. S. (2011) Beyond ethylmalonyl-
CoA: The functional role of crotonyl-CoA carboxylase/reductase
homologs in expanding polyketide diversity. Nat. Prod. Rep. 29, 72−86.
(7) Wu, K., Chung, L., Revill, W. P., Katz, L., and Reeves, C. D.
(2000) The FK520 gene cluster of Streptomyces hygroscopicus var.
ascomyceticus (ATCC 14891) contains genes for biosynthesis of
unusual polyketide extender units. Gene 251, 81−90.
(8) Chan, Y. A., Boyne, M. T., 2nd, Podevels, A. M., Klimowicz, A.
K., Handelsman, J., Kelleher, N. L., and Thomas, M. G. (2006)
Hydroxymalonyl-acyl carrier protein (ACP) and aminomalonyl-ACP
are two additional type I polyketide synthase extender units. Proc. Natl.
Acad. Sci. U.S.A. 103, 14349−14354.
(9) Mo, S., Kim, D. H., Lee, J. H., Park, J. W., Basnet, D. B., Ban, Y.
H., Yoo, Y. J., Chen, S. W., Park, S. R., Choi, E. A., Kim, E., Jin, Y. Y.,
Lee, S. K., Park, J. Y., Liu, Y., Lee, M. O., Lee, K. S., Kim, S. J., Kim, D.,
Park, B. C., Lee, S. G., Kwon, H. J., Suh, J. W., Moore, B. S., Lim, S. K.,
and Yoon, Y. J. (2011) Biosynthesis of the allylmalonyl-CoA extender
unit for the FK506 polyketide synthase proceeds through a dedicated
polyketide synthase and facilitates the mutasynthesis of analogues. J.
Am. Chem. Soc. 133, 976−985.
(10) Eustaquio, A. S., McGlinchey, R. P., Liu, Y., Hazzard, C., Beer, L.
L., Florova, G., Alhamadsheh, M. M., Lechner, A., Kale, A. J.,
Kobayashi, Y., Reynolds, K. A., and Moore, B. S. (2009) Biosynthesis
of the salinosporamide A polyketide synthase substrate chloroethyl-
malonyl-coenzyme A from S-adenosyl-L-methionine. Proc. Natl. Acad.
Sci. U.S.A. 106, 12295−12300.
(11) Liu, Y., Hazzard, C., Eustaquio, A. S., Reynolds, K. A., and
Moore, B. S. (2009) Biosynthesis of salinosporamides from alpha,beta-
unsaturated fatty acids: implications for extending polyketide synthase
diversity. J. Am. Chem. Soc. 131, 10376−10377.
(12) Wilson, M. C., Nam, S. J., Gulder, T. A., Kauffman, C. A.,
Jensen, P. R., Fenical, W., and Moore, B. S. (2011) Structure and
biosynthesis of the marine streptomycete Ansamycin ansalactam a and
its distinctive branched chain polyketide extender unit. J. Am. Chem.
Soc. 133, 1971−1977.
(13) Macherla, V. R., Mitchell, S. S., Manam, R. R., Reed, K. A., Chao,
T. H., Nicholson, B., Deyanat-Yazdi, G., Mai, B., Jensen, P. R., Fenical,
W. F., Neuteboom, S. T., Lam, K. S., Palladino, M. A., and Potts, B. C.
(2005) Structure-activity relationship studies of salinosporamide A
(NPI-0052), a novel marine derived proteasome inhibitor. J. Med.
Chem. 48, 3684−3687.
ASSOCIATED CONTENT
■
S
* Supporting Information
Figures S1−S7, Tables S1−S4, and supplemental methods
including site-directed mutagenesis, protein expression and
purification of MatB, DSZS, MCAT, DEBS holo-ACP6, Kir
holo-ACP5, and KirCII, determination of MatB kinetic
parameters, and synthesis of 19, 25, and 26. This material is
available free of charge via the Internet at http://pubs.acs.org.
(14) Bonnett, S. A., Rath, C. M., Shareef, A. R., Joels, J. R., Chemler,
J. A., Hakansson, K., Reynolds, K., and Sherman, D. H. (2011) Acyl-
CoA subunit selectivity in the pikromycin polyketide synthase PikAIV:
Steady-state kinetics and active-site occupancy analysis by FTICR-M.
Chem. Biol. 18, 1075−1081.
(15) Pohl, N. L., Hans, M., Lee, H. Y., Kim, Y. S., Cane, D. E., and
Khosla, C. (2001) Remarkably broad substrate tolerance of malonyl-
CoA synthetase, an enzyme capable of intracellular synthesis of
polyketide precursors. J. Am. Chem. Soc. 123, 5822−5823.
(16) 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.
(17) Hughes, A. J., and Keatinge-Clay, A. (2011) Enzymatic extender
unit generation for in vitro polyketide synthase reactions: structural
and functional showcasing of Streptomyces coelicolor MatB. Chem.
Biol. 18, 165−176.
(18) Koryakina, I., and Williams, G. J. (2011) Mutant malonyl-CoA
synthetases with altered specificity for polyketide synthase extender
unit generation. ChemBioChem 12, 2289−2293.
(19) Nguyen, T., Ishida, K., Jenke-Kodama, H., Dittmann, E., Gurgui,
C., Hochmuth, T., Taudien, S., Platzer, M., Hertweck, C., and Piel, J.
(2008) Exploiting the mosaic structure of trans-acyltransferase
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: gavin_williams@ncsu.edu.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors would like to thank the Mass Spectrometry Facility
at NC State. This study was supported by an NSF CAREER
Award to G.J.W. (CHE-1151299) and the W. M. Keck
Foundation.
REFERENCES
■
(1) Staunton, J., and Weissman, K. J. (2001) Polyketide biosynthesis:
a millennium review. Nat. Prod. Rep. 18, 380−416.
(2) Weissman, K. J., and Leadlay, P. F. (2005) Combinatorial
biosynthesis of reduced polyketides. Nat. Rev. Microbiol. 3, 925−936.
(3) Butler, M. S. (2008) Natural products to drugs: natural product-
derived compounds in clinical trials. Nat. Prod. Rep. 25, 475−516.
207
dx.doi.org/10.1021/cb3003489 | ACS Chem. Biol. 2013, 8, 200−208

ACS Chemical Biology
Articles
polyketide synthases for natural product discovery and pathway
dissection. Nat. Biotechnol. 26, 225−233.
transthyretin by liquid chromatography-dual electrospray ionization
fourier-transform ion-cyclotron-resonance mass spectrometry. Clin.
Chem. 50, 1535−1543.
(38) Kiefer, F., Arnold, K., Kunzli, M., Bordoli, L., and Schwede, T.
(2009) The SWISS-MODEL Repository and associated resources.
Nucleic Acids Res. 37, D387−392.
(20) Kumar, P., Koppisch, A. T., Cane, D. E., and Khosla, C. (2003)
Enhancing the modularity of the modular polyketide synthases:
transacylation in modular polyketide synthases catalyzed by malonyl-
CoA:ACP transacylase. J. Am. Chem. Soc. 125, 14307−14312.
(21) Koppisch, A. T., and Khosla, C. (2003) Structure-based
mutagenesis of the Malonyl-CoA: Acyl carrier protein transacylase
from Streptomyces coelicolor. Biochemistry 42, 11057−11064.
(22) Keatinge-Clay, A. T., Shelat, A. A., Savage, D. F., Tsai, S. C.,
Miercke, L. J., O’Connell, J. D., 3rd, Khosla, C., and Stroud, R. M.
(2003) Catalysis, specificity, and ACP docking site of Streptomyces
coelicolor malonyl-CoA:ACP transacylase. Structure 11, 147−154.
(23) Cheng, Y. Q., Coughlin, J. M., Lim, S. K., and Shen, B. (2009)
Type I polyketide synthases that require discrete acyltransferases.
Methods Enzymol. 459, 165−186.
(24) Cheng, Y. Q., Tang, G. L., and Shen, B. (2003) Type I
polyketide synthase requiring a discrete acyltransferase for polyketide
biosynthesis. Proc. Natl. Acad. Sci. U.S.A. 100, 3149−3154.
(25) Musiol, E. M., Hartner, T., Kulik, A., Moldenhauer, J., Piel, J.,
Wohlleben, W., and Weber, T. (2011) Supramolecular templating in
kirromycin biosynthesis: the acyltransferase KirCII loads ethylmalonyl-
CoA extender onto a specific ACP of the trans-AT PKS. Chem. Biol.
18, 438−444.
(26) Lopanik, N. B., Shields, J. A., Buchholz, T. J., Rath, C. M.,
Hothersall, J., Haygood, M. G., Hakansson, K., Thomas, C. M., and
Sherman, D. H. (2008) In vivo and in vitro trans-acylation by BryP,
the putative bryostatin pathway acyltransferase derived from an
uncultured marine symbiont. Chem. Biol. 15, 1175−1186.
(27) Wong, F. T., Chen, A. Y., Cane, D. E., and Khosla, C. (2010)
Protein-protein recognition between acyltransferases and acyl carrier
proteins in multimodular polyketide synthases. Biochemistry 49, 95−
102.
(28) Menche, D., Arikan, F., Perlova, O., Horstmann, N., Ahlbrecht,
W., Wenzel, S. C., Jansen, R., Irschik, H., and Muller, R. (2008)
Stereochemical determination and complex biosynthetic assembly of
etnangien, a highly potent RNA polymerase inhibitor from the
myxobacterium Sorangium cellulosum. J. Am. Chem. Soc. 130, 14234−
14243.
(29) Omura, S., Takeshima, H., Nakagawa, A., Miyazawa, J., Piriou,
F., and Lukacs, G. (1977) Studies on the biosynthesis of 16-membered
macrolide antibiotics using carbon-13 nuclear magnetic resonance
spectroscopy. Biochemistry 16, 2860−2866.
(30) Inoue, A., Deguchi, T., Yoshida, M., and Shirahata, K. (1983)
Biosynthesis and the metabolic-fate of C-14-labeled spiramycin-1. J.
Antibiot. 36, 442−444.
(31) Kakavas, S. J., Katz, L., and Stassi, D. (1997) Identification and
characterization of the niddamycin polyketide synthase genes from
Streptomyces caelestis. J. Bacteriol. 179, 7515−7522.
(32) Jackowski, S., Rock, C. O. (1996) Escherichia coli and
Salmonella typhimurium, in Cell. Mol. Biol. (Neidhart, F. C., Ed.),
pp 687−694, American Society of Microbiology, Washington, DC.
(33) Clarke, K. M., Mercer, A. C., La Clair, J. J., and Burkart, M. D.
(2005) In vivo reporter labeling of proteins via metabolic delivery of
coenzyme A analogues. J. Am. Chem. Soc. 127, 11234−11235.
(34) Go, M. K., Chow, J. Y., Cheung, V. W. N., Lim, Y. P., and Yew,
W. S. (2012) Establishing a toolkit for precursor-directed polyketide
biosynthesis: Exploring substrate promiscuities of acid-CoA ligases.
Biochemistry 51, 4568−4579.
(35) Andrews, G. L., Shuford, C. M., Burnett, J. C., Jr., Hawkridge, A.
M., and Muddiman, D. C. (2009) Coupling of a vented column with
splitless nanoRPLC-ESI-MS for the improved separation and
detection of brain natriuretic peptide-32 and its proteolytic peptides.
J. Chromatogr. B. Analyt. Technol. Biomed. Life. Sci. 877, 948−954.
(36) Senko, M., Beu, S., and McLafferty, F. (1995) Determination of
monoisotopic masses and ion populations for large biomolecules from
resolved isotopic distributions. J. Am. Soc. Mass Spectrom. 6, 229−233.
(37) Nepomuceno, A. I., Mason, C. J., Muddiman, D. C., Bergen, H.
R., 3rd, and Zeldenrust, S. R. (2004) Detection of genetic variants of
208
dx.doi.org/10.1021/cb3003489 | ACS Chem. Biol. 2013, 8, 200−208