Báo cáo khoa học: Biochemical characterization of the minimal polyketide synthase domains in the lovastatin nonaketide synthase LovB pot

Biochemical characterization of the minimal polyketide
synthase domains in the lovastatin nonaketide synthase
LovB
Suzanne M. Ma and Yi Tang
Department of Chemical and Biomolecular Engineering, University of California, Los Angeles, CA, USA
Lovastatin is a polyketide metabolite produced by the
filamentous fungi Aspergillus terreus and is an efficient
inhibitor of 3-hydroxy-3-methylglutaryl coenzyme A
(CoA) reductase, the enzyme that catalyzes the rate-
limiting step in cholesterol biosynthesis [1]. Among the
enzymes that biosynthesize lovastatin are two poly-
ketide synthases (PKS) and numerous accessory
enzymes (Fig. 1) [2]. The two megasynthases are the
lovastatin nonaketide synthase (LovB, 335 kDa) and
lovastatin diketide synthase (LovF, 277 kDa), which
catalyze the assembly of the decalin core and the
2-methylbutyrate side chain, respectively (Fig. 1). Both
LovB and LovF are multidomain enzymes with
domain architectures and activities related to animal
fatty acid synthases (FAS) and bacterial type I PKS.
Central to PKS and FAS are the minimal catalytic
Keywords
filamentous fungi; ketosynthase; lovastatin;
megasynthase; polyketide
Correspondence
Y. Tang, Department of Chemical and
Biomolecular Engineering, 5531 Boelter Hall,
420 Westwood Plaza, UCLA, Los Angeles,
CA 90095, USA
Fax: +1 310 206 4107
Tel: +1 310 825 0375

E-mail:
(Received 1 March 2007, revised 29 March
2007, accepted 2 April 2007)
doi:10.1111/j.1742-4658.2007.05818.x
The biosynthesis of lovastatin in Aspergillus terreus requires two mega-
synthases. The lovastatin nonaketide synthase, LovB, synthesizes the inter-
mediate dihydromonacolin L using nine malonyl-coenzyme A molecules, and
is a reducing, iterative type I polyketide synthase. The iterative type I poly-
ketide synthase is mechanistically different from bacterial type I polyketide
synthases and animal fatty acid synthases. We have cloned the minimal
polyketide synthase domains of LovB as standalone proteins and assayed
their activities and substrate specificities. The didomain proteins ketosyn-
thase-malonyl-coenzyme A:acyl carrier protein acyltransferase (KS-MAT)
and acyl carrier protein-condensation (ACP-CON) domain were expressed
solubly in Escherichia coli. The monodomains MAT, ACP and CON were
also obtained as soluble proteins. The MAT domain can be readily labeled
by [1,2-
14
C]malonyl-coenzyme A and can transfer the acyl group to both
the cognate LovB ACP and heterologous ACPs from bacterial type I and
type II polyketide synthases. Using the LovB ACP-CON didomain as an
acyl acceptor, LovB MAT transferred malonyl and acetyl groups with
k
cat
⁄ K
m
values of 0.62 min
)1
Ælm
)1

and 0.032 min
)1
Ælm
)1
, respectively. The
LovB MAT domain was able to substitute the Streptomyces coelicolor
FabD in supporting product turnover in a bacterial type II minimal poly-
ketide synthase assay. The activity of the KS domain was assayed inde-
pendently using a KS-MAT (S656A) mutant in which the MAT domain
was inactivated. The KS domain displayed no activity towards acetyl
groups, but was able to recognize malonyl groups in the absence of ceru-
lenin. The relevance of these finding to the priming mechanism of fungal
polyketide synthase is discussed.
Abbreviations
ACP, acyl carrier protein; CoA, coenzyme A; CON domain, condensation domain; 6-DEBS, 6-deoxyerythronolide B synthase; FAS, fatty acid
synthase; KR, ketoreductase; KS, ketosynthase; MAT, malonyl-CoA:ACP acyltransferase; NSAS, norsolorinic acid synthase; PKS, polyketide
synthase.
2854 FEBS Journal 274 (2007) 2854–2864 ª 2007 The Authors Journal compilation ª 2007 FEBS
domains [3], consisting of the ketosynthase (KS) that
performs decarboxylative claisen condensation for
chain elongation [4]; the malonyl-CoA:ACP acyltrans-
ferase (MAT) that selects and transfers the extender
unit in the form of malonic esters; and the acyl carrier
protein (ACP) that serves as the tether for the extender
unit and the growing chain. In addition, tailoring
enzymes such as ketoreductase (KR), dehydratase,
methyltransferase and enoylreductase modify the car-
bon backbone and introduce structural diversity [5].
LovB was first identified by Hendrickson et al. [6]
and reconstituted in Aspergillus nidulans by Hutchinson

and Vederas [2]. It is a reduced iterative PKS that is
mechanistically different from the modular, bacterial
type I PKS, such as the well-characterized 6-deoxy-
erythronolide B synthase (6-DEBS). The minimal PKS
domains in LovB are used repeatedly to synthesize the
nonaketide decalin core from nine malonyl-CoA units
(Fig. 1). Varying degrees of polyketide tailoring modifi-
cations are performed after each condensation step by a
different combination of LovB catalytic domains (inclu-
ding the dissociated enoylreductase, LovC) [2] to afford
the key intermediate dihydromonacolin L. LovB also
contains a C-terminus domain with sequence similarity
to a nonribosomal-peptide synthase condensation
(CON) domain. The function of the CON domain in di-
hydromonacolin L biosynthesis is not known. Whereas
the colinearity rule of modular type I PKS allows pre-
diction of product structure from primary domain
arrangements [7], iterative PKS such as LovB use an
unknown set of programming rules and the sequence of
catalytic events are difficult to decipher from examining
domain sequences alone [8]. Elucidating the biochemi-
cal properties and substrate specificities of individual
domains are therefore important in understanding the
underlying mechanisms of fungal PKS megasynthases,
predicting polyketide products from protein sequences,
and engineering biosynthesis of novel polyketides.
Biochemical studies and engineering of the fungal
megasynthases are more difficult than the bacterial
counterparts for several reasons. First, the intrinsic size
of these megasynthases poses a significant barrier to

the isolation and characterization of these enzymes.
Heterologous expression of intact PKS are generally
performed in evolutionarily closely related fungal hosts
such as A. nidulans [2,9] and Aspergillus oryzae [10].
Second, the iterative nature of the PKS renders each
domain indispensable during chain elongation.
Approaches such as domain inactivation, mutasynthe-
sis and precursor directed biosynthesis are ineffective
in probing domain specificities and sequences of cata-
lytic events. Recent refinement of fungal genetic tech-
niques has allowed construction of PKS mutants and
hybrid synthases [11], with the notable example of KS
swapping between heterologous fungal megasynthases
[12]. Nevertheless, these manipulations are difficult to
perform and are time-consuming.
Reconstitution of the catalytic domains as individual
enzymes has been successful in analyzing the biochemi-
cal and structural features of animal FAS [13] and bac-
terial type I PKS, including the KS-AT [14,15], KR
[16], ACP [17], and thioesterase domains [18–20] from
6-DEBS. This strategy is especially attractive in study-
ing iterative PKS because the activities and substrate
specificities of individual domains can be probed sepa-
rately. For example, Crawford and Townsend [21] used
KS MAT DH MT KR ACP CON
KS MAT DH MT ER KR ACP
ER
KS MAT DH MT KR ACP CON
KS MAT DH MT KR ACP CON
KS MAT DH MT KR ACP CON

KS MAT DH MT KR ACP CON
KS MAT DH MT ER KR ACP
OH
HO
COOH
OH
Monacolin J acid
OH
HO
COOH
O
O
LovF
LovB
CYP P450
Lovastatin acid
Dihydromonacolin L
OH
HO
COOH
LovC
S
O
SH
S
S
SOH
O
OH
O

Diels-Alder
Reaction
SH
SH
4 X
Malonyl-CoA
2 X
Malonyl-CoA
3 X
Malonyl-CoA
O
2 X
Malonyl-CoA
PKS
Release
O
LovD
Fig. 1. Biosynthetic pathway for lovastatin
in A. terreus. LovB is the lovastatin nonake-
tide synthase that synthesizes dihydromo-
nacolin L together with the dissociated
enoylreductase LovC. Dihydromonacolin L is
oxidized to monacolin J by one or more
cytochrome P450 enzymes. LovF is the
lovastatin diketide synthase that synthesizes
2-methylbutyry-S-LovF. The side chain is
transferred by LovD to monacolin J to yield
lovastatin. Domain abbreviations: KS, keto-
synthase; MAT, malonyl-CoA:acyl carrier
protein transacylase; ACP, acyl carrier pro-

tein; DH, dehydratase; ER, enoylreductase;
MT, methyltransferase; KR, ketoreductase;
CON, possible condensation domain.
S. M. Ma and Y. Tang Minimal polyketide synthase domains in LovB
FEBS Journal 274 (2007) 2854–2864 ª 2007 The Authors Journal compilation ª 2007 FEBS 2855
the UWA algorithm to clone and identify the function
of the starter unit:ACP transacylase domain from
norsolorinic acid synthase (NSAS) involved in afla-
toxin biosynthesis [22]. Recently, Ma et al. [23] demon-
strated that the standalone NSAS ACP domain can
function as the ACP component of a bacterial minimal
PKS in vitro.
In this study, we have cloned the minimal PKS
domains from LovB and isolated them as standalone
mono- and didomains. We characterized the activities
of the KS and MAT domains in detail and examined
the interactions between the core catalytic units with
the cognate and heterologous ACP domains. Further-
more, we assayed the substrate specificities of the KS
and MAT domains to highlight similarities and funda-
mental differences between theses domains embedded
in bacterial PKS, fungal PKS, and animal FAS.
Results and Discussion
Expression of minimal PKS domains
To explore the activities of the KS, MAT and ACP
domains of LovB (Fig. 2), we proceeded to construct
expression vectors that contained excised mono- and
didomains. Putative domain boundaries were identified
through sequence alignment with bacterial type I PKS
as well as mammalian and fungal FAS, and are indi-

cated in Fig. 2(B). PCR primers were designed to
amplify the targeted domains based on the designated
boundaries. The template of the PCR reaction was the
complete, uninterrupted reading frame of LovB, which
was obtained by removing the seven introns using
splice by overlap extension PCR. We constructed
pET28a(+) derived expression vectors for the KS-
MAT didomain, the ACP monodomain, the CON
monodomain and the ACP-CON didomain (Table 1).
The KS-MAT (pSMA30) and CON (pSMA61) expres-
sion constructs were transformed into BL21(DE3)
for protein expression, whereas the ACP-containing
constructs were transformed into BAP1 for phospho-
pantetheine modification of the ACP domain with the
broad spectrum phosphopantetheinyl transferase gene
sfp from Bacillus subtilis [24].
The KS-MAT didomain can be expressed at very
high levels (approximately 20 mgÆL
)1
) when the His-tag
is appended at the C-terminus. The KS-MAT can be
purified to homogeneity by nickel affinity chromato-
graphy followed by anion exchange chromatography.
N-terminus His-tag fusion resulted in significant
AB
Fig. 2. (A) Domain organization of LovB and the standalone domains studied in this work. SDS ⁄ PAGE of purified proteins assayed: lane 1, LovB
KS-MAT (103 kDa); lane 2, LovB KS-MAT (S656A); lane 3, LovB KS-MAT (S657A); lane 4, LovB MAT (52 kDa); lane 5, LovB MAT (S208A);
lane 6, LovB ACP-CON (66 kDa); lane 7, LovB ACP (11 kDa); lane 8, DEBS M3 ACP (11 kDa); lane 9, DEBS M6 ACP (11 kDa); lane 10, ZhuN
(10 kDa). (B) Amino acid sequence of the LovB megasynthase and domain boundaries. Shaded regions indicate the individual catalytic domains
identified through sequence homology with the fatty acid synthase. The active site residues of KS, MAT and ACP are underlined.

Minimal polyketide synthase domains in LovB S. M. Ma and Y. Tang
2856 FEBS Journal 274 (2007) 2854–2864 ª 2007 The Authors Journal compilation ª 2007 FEBS
aggregation of the protein into inclusion bodies. Com-
pared to bacterial modular PKS, the N-terminal of
LovB does not contain the coiled-coil linkers crucial for
module–module communication. By contrast, the N-ter-
minal sequence (P10IVVVGSGCR19) of LovB KS is
well aligned with the residues (P40IAIVGMACR49)
that form the first b -sheet of DEBS M5 KS-MAT dimer
[14]. Therefore, it is plausible that an N-terminus
His-tag may interfere with the folding of the KS core.
The ACP domain of NSAS, an iterative, nonreducing
PKS from Aspergillus parasiticus, was recently expressed
as a standalone protein [23]. Although the ACP domain
boundary differed for the reducing LovB PKS, both
the ACP (1 mgÆL
)1
) and the ACP-CON didomain
(15 mgÆL
)1
) can be expressed solubly as N-terminal His-
Tag proteins (Fig. 2A). Complete pantetheinylation of
the active site serine can be confirmed by MALDI-TOF
analysis (Observed: 12225; Expected: 12221; Apo form:
11881). Appending the CON domain to the ACP
domain resulted in a 15-fold increase in the yield of the
soluble protein. Pantetheinylation of the ACP in ACP-
CON didomain was verified using MALDI-TOF of a
tryptic fragment of the purified protein (data not
shown). The LovB CON domain can also be expressed

as a standalone protein when cloned using the boundar-
ies shown in Fig. 2(B) (data not shown).
In addition to LovB ACP and LovB ACP-CON, we
also purified four heterologous ACP domains for bio-
chemical analysis of KS:ACP and MAT:ACP inter-
actions. We expressed the holo-forms of DEBS M3
ACP and M6 ACP (domains from bacterial type I PKS)
[15], as well as ZhuN [25] and OxyC [26] (standalone
ACPs from type II PKS) using BAP1 (Fig. 2A). Each of
these proteins was purified using Ni-nitrilotriacetic acid
chromatography and were verified by MALDI-MS.
Biochemical characterization of the LovB KS-MAT
didomain
Equipped with soluble forms of the LovB KS-MAT
didomain and standalone ACP domains, we assayed
the catalytic properties of the didomain, as well as
interactions between the minimal PKS components.
Figure 3A shows the labeling of LovB KS-MAT by
[1,2-
14
C]malonyl and [1-
14
C]acetyl-CoA. We expected
the radioactive acyl substrate to reside either on the
active site cysteine of the KS domain through thioester
exchange or on the active site serine of the MAT
domain through esterification. Under identical condi-
tions (10 lm KS-MAT, 0 °C, 10 min), more than 70%
of the didomain was labeled by [
14

C]malonyl-CoA, but
only approximately 5% of the protein was labeled with
[
14
C]acetyl-CoA.
To examine the LovB KS–MAT interaction with the
dissociated ACP domain, we incubated the LovB ACP
and LovB ACP-CON didomain with the KS-MAT
(5 lm) in the presence of either [
14
C]malonyl-CoA
or [
14
C]acetyl-CoA (Fig. 3B). When malonyl-CoA
was used as the acyl donor, both the ACP and
the ACP-CON didomain can be readily labeled, con-
firming the interaction between the dissected KS-MAT
and ACP domains remained intact (Fig. 3B). Inter-
estingly, although the ACP and ACP-CON didomain
were supplied at the same molar amounts (10 lm), the
steady state labeling of the ACP-CON domain was
approximately six-fold higher than that of the ACP
domain. To probe whether there are additional sites
for acylation in the CON domain, we performed
an acyltransfer assay using the CON domain alone.
No labeling can be observed when the CON domain
or the apo-ACP-CON didomain was used in the
Table 1. PCR primers used for amplification and mutation of DNA encoding the LovB KS-MAT didomain, LovB ACP, and LovB ACP-CON
didomain. The introduced restriction sites and mutation sites are underlined.
Gene Plasmid Primers (5¢) to 3¢)

LovB KS-MAT pSMa30 AA
CATATGGCTCAATCTATGTATCCTAATG
AATCAGGGGCGAGTTGC
TCTAGATTCCACCCAGTAGCGACGAGAG
LovB ACP pSMa7 AA
CATATGCTCTTGCAGGCAGACGAACCTG
AATTAGGAGCTAGGTGGCAATCGCGCAGCAG
LovB ACP-CON pSMa8 AA
CATATGCTCTTGCAGGCAGACGAACCTG
AATCATGCCAGCTTCAGGGCGGGATTC
LovB CON pSMa61 AA
CATATGGTCGCAGCCACCGACGGGGG
AATCATGCCAGCTTCAGGGCGGGATTC
LovB MAT pSMa73 TT
CCATGGAGCCAGAGCAAAACCAG
AATCAGGGGCGAGTTGC
TCTAGATTCCACCCAGTAGCGACGAGAG
LovB MAT (S208A) ⁄ pSMa70 ⁄ GGCTCTTGTTGGACAT
GCTAGCGGCGAAATTGCTTG
LovB KS-MAT (S656A) pSMa36 CAAGCAATTTCGCCGCTA
GCATGTCCAACAACAGCC
LovB MAT (S209A) pSMa71 CTGTTGTTGGACATAGC
GCTGGCGAAATTGCTTGCG
CGCAAGCAATTTCGCC
AGCGCTATGTCCAACAACAG
S. M. Ma and Y. Tang Minimal polyketide synthase domains in LovB
FEBS Journal 274 (2007) 2854–2864 ª 2007 The Authors Journal compilation ª 2007 FEBS 2857
acyltransfer assay, confirming the malonyl group was
indeed only transferred to the active site phosphopant-
etheine thiol of the ACP domain. The increase in labe-

ling of the ACP-CON didomain may be a result of
improved folding of the ACP domain when attached
to the larger CON domain, which can also explain the
increase in soluble protein yield; or enhanced acyl-
transfer rate as a result of more extensive protein–pro-
tein interactions between the dissociated didomains.
The LovB KS-MAT didomain displayed broad speci-
ficity towards heterologous ACPs in the acyltransfer
reaction and transferred the malonyl group to all
ACPs to a comparable extent (Fig. 3B). The KS-MAT
didomain also transferred [
14
C]acetyl to both ACP
and ACP-CON after 30 min of incubation (Fig. 3B,
lanes 6 and 7). The steady state amount of acetyl-
ACP-CON was approximately 20% that of malonyl-
ACP-CON.
Isolation and cloning of standalone MAT domain
To understand the relative contributions of the KS
and MAT domains in selecting and transferring the
acyl substrates shown in Fig. 3, we aimed to further
dissect the KS-MAT didomain into individual cons-
tituents. During nickel-affinity purification of the
103 kDa LovB KS-MAT didomain, a significant con-
taminant protein with an apparent molecular mass of
45 kDa was copurified (Fig. 4A). We suspected the
smaller protein may be a truncated MAT (C-term His-
tag) domain that resulted from proteolytic cleavage at
the boundary separating the KS and MAT domains.
Proteolytic acyltransferase fragments have been previ-

ously obtained from the rat FAS [27]. The 45 kDa
protein can be readily labeled by [
14
C]malonyl-CoA
(Fig. 4A), demonstrating that it can form a covalent
66
55
42
116
kDa
66
55
42
116
kDa
1
LovB KS-MAT
LovB MAT
212
158
97
kDa
LovB
“KS-less” LovB
LovB MAT
WT S208A S209A
23
12
3
123

4
A
B
C
Fig. 4. (A) Identification of a MAT proteolytic fragment. Lane 1,
SDS ⁄ PAGE of LovB KS-MAT(C-term His-tag) after Ni-nitrilotriacetic
purification; lane 2, autoradiography of proteins in lane 1 after labe-
ling with [
14
C]malonyl-CoA; lane 3, SDS ⁄ PAGE of purified, cloned
standalone MAT domain; lane 4, autoradiography of protein in
lane 3 after labeling with [
14
C]malonyl-CoA. (B) Loss of the KS frag-
ment in the full length LovB. Lane 1, SDS ⁄ PAGE of LovB (C-term
His-tag) after Ni-nitrilotriacetic showing two large proteins; lane 2,
autoradiography of proteins in lane 1 after labeling with [
14
C]malo-
nyl-CoA; lane 3, autoradiography of proteins in lane 1 after labeling
with [
14
C]acetyl-CoA. (C) Examination of the active site serine of
standalone MAT. [
14
C]malonyl-CoA labeling of wild-type (lane 1),
S208A (lane 2) and S209A (lane 3); 50 m
M NaH
2
PO

4
,pH¼ 7.4,
20 l
M KS-MAT, 2 mM acyl-CoA, 0 °C , 10 min.
A
1 2
KS-AT
B
1 2 3 4 5
KS-MAT
ACP-CON
ACP
6 7
Fig. 3. (A) [1,2-
14
C]malonyl-CoA (lane 1) and [1-
14
C]acetyl-CoA labe-
ling (lane 2) of the LovB KS-MAT, 50 m
M NaH
2
PO
4
,pH¼ 7.4,
10 l
M KS-MAT, 2 mM acyl-CoA, 0 °C, 10 min 70% of KS-MAT is
labeled by malonyl-CoA, whereas only 5% is labeled by acetyl-CoA.
(B) Acyltransfer of [
14
C]malonyl and [

14
C]acetyl to ACPs in the pres-
ence of LovB KS-MAT. Each ACP is performed in duplicate.
[
14
C]Malonyl-CoA: lane set 1, LovB ACP-CON; lane set 2, LovB
ACP; lane set 3, DEBS M3 ACP; lane set 4, DEBS M6 ACP; lane set
5, ZhuN. [
14
C]Acetyl-CoA: lane set 6, LovB ACP-CON; lane set 7,
LovB ACP; 50 m
M NaH
2
PO
4
,pH¼ 7.4, 5 lM KS-MAT, 2 mM
acyl-CoA, 10 lM ACP. 25 °C, 30 min.
Minimal polyketide synthase domains in LovB S. M. Ma and Y. Tang
2858 FEBS Journal 274 (2007) 2854–2864 ª 2007 The Authors Journal compilation ª 2007 FEBS
adduct with the malonyl acyl group. An additional
anion exchange chromatography step was used to
separate the two proteins in pure forms. N-terminal
sequencing of the 45 kDa protein indicated it is indeed
the standalone MAT domain. The truncation site is at
EY448|M449EPEQ (N-terminal sequencing results are
indicated in bold), which corresponds to a region that
is highly conserved among fungal and bacterial type I
PKS [28]. Examining the recently elucidated crystal
structure of the 6-DEBS module 5 KS-AT suggested
that this proteolytic site identified for LovB KS-MAT

is at the beginning of a structurally highly ordered lin-
ker region (approximately 100 amino acids) connecting
and interacting with the KS and MAT domains [14].
Intriguingly, LovB KS-MAT is highly susceptible to
proteolytic cleavage during cell lysis at this domain
junction (Fig. 4A), but does not undergo further deg-
radation after affinity and anion chromatography
steps. By contrast, no apparent proteolysis of the
DEBS M5 KS-AT [14] or the animal FAS KS-MAT
[29] didomains were observed during purification.
We then cloned the MAT domain using the
sequence information obtained above and expressed
it as a standalone protein from Escherichia coli. The
recombinant MAT domain can be expressed at very
high levels (approximately 50 mgÆL
)1
), be purified to
homogeneity (Fig. 4A, lane 3) and can be labeled
with [
14
C]malonyl-CoA (Fig. 4A, lane 4). By con-
trast, the standalone KS region cannot be isolated
when expressed in E. coli. Expression of the LovB
KS domain devoid of the MAT domain resulted in
the formation of inclusion bodies, which was recently
noted for the KS domains of other megasynthases
[23].
Proteolytic cleavage of the KS domain was also
observed for the entire LovB protein. When expressed
with a C-terminal His-tag from BAP1, the entire

megasynthase (335 kDa) can be expressed solubly at
surprisingly high levels (Fig. 4B; 5 mgÆL
)1
). However,
during Ni-nitrilotriacetic affinity purification, we
observed an additional high molecular weight protein
(approximately 290 kDa) coeluted with LovB. Incuba-
tion of the purified proteins with [
14
C]malonyl-CoA
yielded strong labeling of both bands (Fig. 4B), sug-
gesting the smaller fragment may contain an intact
MAT domain, and is likely a ’KS-less’ LovB. The
recent crystal structures of the animal FAS showed
the importance of the dimeric KS in stabilizing the
parallel FAS structure [30]. Our results with LovB,
along with the observation of a recombinant ’KS-less’
FAS protein by Witkowski et al. [29], confirm that the
individual domains of the megasynthase may remain
folded despite the lack of the KS dimer.
Characterization of the standalone MAT protein
The LovB MAT active site region contains two con-
secutive serine residues (GHS656S657G). The serine
dyad occurs frequently among fungal PKS and is also
found in the active site of the lovastatin diketide syn-
thase LovF. This differs from FAS MAT and PKS
AT, where the active site contains a single serine in a
highly conserved GHSXG motif [31]. Although S656 is
the putative nucleophile, it is unknown whether the
second serine in the dyad can also participate in the

acyltransfer reaction in LovB MAT. We constructed
two mutant forms of the standalone MAT, S208A and
S209A, to probe the relative contribution of both ser-
ine residues. The amount of soluble protein obtained
for the S208A mutant is comparable to the yield of the
wild-type protein (approximately 50 mgÆL
)1
). By con-
trast, the S209A mutant expressed poorly (2 mgÆL
)1
),
suggesting the second serine contributes substantially
to the overall folding and structural integrity of
the catalytic domain. The wild-type MAT and the
two mutants were incubated in the presence of
[
14
C]malonyl-CoA and protein labeling was visualized
with autoradiography (Fig. 4C). The S208A mutant
was devoid of any detectable labeling, whereas the
S209A mutant was labeled, albeit weaker than the
wild-type enzyme. We therefore concluded that S656 in
LovB is absolutely essential for the catalytic activity of
the LovB MAT domain and is the site of extender unit
esterification.
We then repeated the MAT:ACP acyltransfer experi-
ment using the standalone LovB MAT domain. We
incubated the MAT domain (5 lm) and various holo-
ACP proteins (10 lm) with [
14

C]malonyl-CoA for
30 min and the levels of acyltransfer were measured
using autoradiography. The MAT domain displayed
broad substrate specificity towards ACP domains and
is capable of acylating all the heterologous ACP pro-
teins examined (data not shown).
Figure 5A compares the initial rates of acyltransfer
catalyzed by the MAT domain using LovB ACP-CON
as the acyl acceptor. MAT was able to transfer the
malonyl group (K
m
¼ 5.4 ± 1.5 lm; k
cat
¼ 3.3 ± 0.16
min
)1
; k
cat
⁄ K
m
¼ 0.62 min
)1
Ælm
)1
) almost 20-fold fas-
ter than towards the acetyl group (K
m
¼ 18.7 ± 1.6 lm;
k
cat

¼ 0.6 ± 0.014 min
)1
; k
cat
⁄ K
m
¼ 0.032 min
)1
Ælm
)1
).
The stronger preference of MAT towards malonyl-
CoA is consistent with the differential extents of labeling
observed in Figs 3(A) and 4(B). We then compared the
substrate specificity of LovB MAT to the Streptomyces
coelicolor malonyl-CoA:ACP transacylase (scFabD)
[32]. Using ACP-CON as an acceptor, scFabD dis-
played extremely rapid activity towards malonyl-CoA
S. M. Ma and Y. Tang Minimal polyketide synthase domains in LovB
FEBS Journal 274 (2007) 2854–2864 ª 2007 The Authors Journal compilation ª 2007 FEBS 2859
(k
cat
¼ 396 ± 14 min
)1
, K
m
not determined), whereas
it remained completely inactive towards acetyl-CoA.
The substrate selectivity of LovB MAT towards
malonyl-CoA over acetyl-CoA therefore lies between

that of bacterial FabD (malonyl only) [33] and that of
the animal FAS MAT domain, which transacylates
acetyl and malonyl unit with comparable kinetic
parameters [13]. The FAS MAT employs acetyl unit
selected by the MAT domain as the starter unit in
fatty acid biosynthesis. During chain elongation, the
FAS MAT uses a ’self-editing’ mechanism to rapidly
remove the incorrectly charged acetyl moiety to an
acyl acceptor, such as free CoA, to avoid stalling of
the megasynthase [34]. When the FAS MAT domain
was charged with the cognate malonyl unit, electro-
static interaction between an arginine residue and the
carboxylate anion of malonyl unit anchors the malonyl
unit in the active site [35]. The conserved arginine can
be found in the LovB MAT domain (AYLR681G) and
may play a similar role in conferring substrate specifi-
city towards malonyl-CoA and proofread against ace-
tyl-CoA. Nevertheless, the poor activity of LovB MAT
towards acetyl-CoA is indicative of a different initi-
ation mechanism of polyketide biosynthesis in which
the acetate starter unit may be derived from the
decarboxylation of malonyl-CoA instead o f acetyl-ACP
(see below).
The broad ACP-specificity of the LovB MAT
domain hints that the standalone protein may act as a
general acyltransferase between malonyl-CoA and any
holo-ACP domain. We assayed the ability of LovB
MAT to substitute for scFabD in a type II PKS turn-
over assay. The role of MAT in this assay is to gene-
rate the malonyl-ACP extender unit in situ [36]. As

shown in Fig. 5B, when the oxytetracycline minimal
PKS was incubated with LovB MAT, the expected
polyketide products, SEK15 and SEK15B, were syn-
thesized. The decreased amounts of products reflect
the slower turnover rate of LovB MAT is limiting the
level of malonyl-OxyC available for polyketide biosyn-
thesis (lane 2). Increasing the amount of LovB MAT
in the reaction mixture led to an corresponding
increase in the amount of product recovered (lane 3).
These results confirm that the LovB MAT can indeed
interact with heterologous PKS and reconstitute the
activities of the type II minimal PKS, and strengthen
the recent suggestion that iterative type I PKS can be
considered as covalently linked type II PKS enzymes
[23].
Biochemical characterization of the LovB KS-MAT
(S656A) domain
In order to isolate and study the property of the LovB
KS domain, we constructed the LovB KS-MAT°
S656A point mutant, which completely inactivates the
MAT domain (Fig. 4C). The mutant was expressed
and purified from BL21(DE3) ⁄ pSMA36 as described
in the Experimental procedures section.
The LovB KS-MAT° mutant was assayed for its
specificity towards different acyl units. Acetyl-CoA
has been proposed to be the starter unit for numerous
fungal PKS and may acetylate the KS domain directly
through a KS-mediated thioester exch ange. Alternatively,
the PKS can generate a cetyl-KS through decarboxylation
of malonyl-CoA or malonyl-ACP. No labeling of

the KS-MAT° mutant by [
14
C]malonyl-CoA and
[
14
C]acetyl-CoA can be observed (data not shown).
To assay whether any of the acyl substrates are only
activated by the KS domain transiently followed by
0
1
2
3
4
0
Concentration [
14
C]acyl CoA (µ
M
)
fo revonruT
4
1
n
im
(
NO
C
-
P
C

A
-
lyca C
1
–
)
Malonyl
Acetyl
123
SEK15/15b
Baseline
250200
150100
50
A
B
Fig. 5. (A) Michaelis–Menten plot of the rate of acylation catalyzed
by LovB MAT domain. LovB ACP-CON didomain was used as an
acyl acceptor (50 l
M). LovB MAT displayed a 20-fold improvement
in kinetic properties (k
cat
⁄ K
m
) in transacylating malonyl-CoA (s)
when compared to acetyl-CoA (e). (B) Radio-TLC analysis of poly-
ketides produced by minimal oxy PKS (2 l
M OxyA-OxyB (KS-CLF),
LovB MAT or scFabD, 2 m
M

14
C-malonyl-CoA (1.6 mCiÆmmol
)1
),
50 l
M holo-OxyC), 25 °C, 30 min. Lane 1, 0.7 lM scFabD. The
products were identified as SEK15 and SEK15b by comparison with
standards; lane 2, 0.7 l
M LovB MAT; lane 3, 7 lM LovB MAT. LovB
MAT was able to support the turnover of oxy minimal PKS through
the in situ generation of malonyl-OxyC. No products were detected
in the presence of OxyA, OxyB and OxyC alone.
Minimal polyketide synthase domains in LovB S. M. Ma and Y. Tang
2860 FEBS Journal 274 (2007) 2854–2864 ª 2007 The Authors Journal compilation ª 2007 FEBS
interthiol acyltransfer to an ACP acceptor protein [37],
we coincubated the KS-MAT° didomain with stand-
alone ACP domains in the presence of the different
acyl-CoAs. Surprisingly, we detected transfer of radio-
labeled acyl unit to the ACP domains only in the pres-
ence of [
14
C]malonyl-CoA (Fig. 6). Addition of
cerulenin to the reaction mixture significantly attenu-
ated the rate of acyl transfer to ACP, validating that
the detected thioester exchange is indeed facilitated by
the active site cysteine of the KS domain [38]. The rate
of label transfer by the KS domain is, however, signifi-
cantly slower (k
cat
⁄ K

m
< 0.05 min
)1
Ælm
)1
) compared
to that of the MAT domain, and hence unlikely to
be a physiologically important mechanism of ACP
malonylation during polyketide biosynthesis. However,
our results clearly show that KS domain possesses
interthiol acyltransferase activities and can distinguish
between malonyl and acetyl-CoA. Furthermore, the
KS domain also processes broad ACP specificity and
is able to transfer the malonyl unit to the different
ACP domains examined.
Our results with the MAT and the KS-MAT°
domains suggest that the fungal PKS may initiate
polyketide synthesis through decarboxylation of malo-
nyl-ACP as the predominant pathway. The KS domain
does not display any detectable interthiol acyltransfer
activity towards the acetyl unit in our assay; hence,
it is unlikely to be directly primed by acetyl-CoA or
acetyl-ACP. This is in contrast to the priming mechan-
ism observed for both bacterial type I and type II
PKS, where polyketide synthesis can be initiated by
short chain alkylacyl-CoAs. The proposed priming
mechanism for LovB KS-MAT° is consistent with our
ongoing work with the bikaverin PKS (pks4) from
Gibberella fujikuroi (S. M. Ma, J. Zhan & Y. Tang,
unpublished data) and may represent a general mech-

anism for the initial steps of fungal PKS biosynthesis.
In conclusion, we have examined the minimal PKS
components of the lovastatin nonaketide synthase by
obtaining dissociated mono- and didomain proteins.
We have shown that most of the domains examined
can be expressed as standalone proteins, except the KS
domain. We have shown that the LovB MAT displays
strong substrate selectivity towards malonyl-CoA over
acetyl-CoA and can functionally interact with bacterial
PKS components in synthesizing a polyketide product,
further strengthening the converging theory that iterat-
ive type I PKS is a linear juxtaposition of bacterial
PKS components. The LovB MAT displayed signifi-
cantly different properties when compared to the mam-
malian FAS MAT domain, especially in acyl-CoA
substrate specificity. Our work with LovB also sets the
stage for a more systematic exploration of the catalytic
mechanisms of reduced fungal PKS, including the bio-
chemical basis for iterative condensation and the logic
of differential tailoring steps during each cycle of chain
elongation. The broad specificity of the KS and MAT
domains toward heterologous ACP domains may also
provide immense opportunities for combinatorial bio-
synthesis of novel polyketide entities.
Experimental procedures
Strains
Genomic DNA from A. terreus strain ATCC 20542 was
used as the template for PCR amplification. E. coli XL1-
blue and E. coli BL21(DE3) were used as cloning and
expression strains, respectively. For recombinant expres-

sion of holo-ACPs in which the proteins are functionalized
with the phosphopantetheine prosthetic group, the BAP1
strain [24], which contains a chromosomal copy of the
B. subtilis phosphopantetheinyl transferase gene sfp, was
used.
Cloning of domains
The lovB gene was amplified from A. terreus genomic DNA
using splice by overlap extension PCR to delete the seven
introns. Engineered restriction sites were introduced at
putative domain boundaries via silent mutations to facili-
tate domain cloning. The sites are KpnI (KS–AT bound-
ary); XbaI (AT–dehydratase boundary) and SpeI (KR–ACP
boundary). The boundaries were identified through multiple
sequence alignment of LovB with individual modules from
6-DEBS, NSAS and rat FAS. The entire lovB gene was
12 34 5 6
KS-MAT
0
ACP-CON
ACP
cerulenin (200
µ
M)
-+ -+ -+
Fig. 6. LovB KS-MAT° mediated malonyl transfer from [
14
C]malo-
nyl-CoA to different ACP acceptors. Cerulenin is added to inhibit
the KS functions. Lanes 1 and 2, LovB ACP-CON didomain; lanes 3
and 4, LovB ACP; lanes 5 and 6, DEBS M3 ACP; 50 m

M NaH
2
PO
4
,
pH ¼ 7.4, 3 l
M KS-MAT°,2mM[
14
C]malonyl-CoA, 10 lM ACP,
25 °C, 30 min. No transfer of [
14
C]acetyl-CoA was observed.
S. M. Ma and Y. Tang Minimal polyketide synthase domains in LovB
FEBS Journal 274 (2007) 2854–2864 ª 2007 The Authors Journal compilation ª 2007 FEBS 2861
inserted into pET21a(+) between NdeI and EcoRI to create
pSMa33, the expression plasmid for LovB with a C-ter-
minal polyhistidine tag. The LovB KS-MAT didomain,
MAT, ACP, ACP-CON and CON were each amplified
from pSMa33 using the primer pairs shown in Table 1 and
were cloned into pET28a(+) to yield the corresponding
expression vectors. The KS-MAT didomain contained a
C-terminal His-tag, whereas all the other constructs con-
tained N-terminal His-tags. Design of the forward primer
used to amplify t he recombinant MAT domain was facilitated
through N-terminal sequencing of the truncated MAT
domain, which was recovered from expression of the LovB
KS-MAT didomain. The construction of DEBS M3 ACP
[15], DEBS M6 ACP [15] and R1128 ZhuN [25] previously
described.
Site-directed mutagenesis

Site-directed mutagenesis was used to generate the point
mutants of the LovB MAT monodomain and KS-MAT
didomain. The mutagenic primers are shown in Table 1.
The expression plasmids pSMA73 and pSMa30 were used
as the templates for MAT and KS-MAT mutagenesis,
respectively. Restriction analysis of the plasmid products
was first performed to identify mutants, followed by DNA
sequencing of the entire gene to verify the intended
mutations.
Expression and purification of ACP proteins
All ACP proteins (including ACP-CON) were expressed
and purified using the following procedure. The expres-
sion plasmids were transformed into E. coli BAP1 strain
for holo-ACP expression and into E. coli BL21(DE3)
strain for apo-ACP expression. The cells were grown at
37 °C in LB medium with 35 lgÆmL
)1
kanamycin to an
absorbance of 0.4 at 600 nm. The cells were incubated
on ice for 10 min, and induced with 0.1 mm isopropyl
thio-b-d-galactoside for 16–24 h at 16 °C. The cells were
harvested by centrifugation (2500 g, Beckman Coulter
Allegra X-12R, rotor SX4750, 10 min), resuspended in
buffer A (20 mm Tris ⁄ HCl, 10 mm imidazole, 500 mm
NaCl, pH ¼ 7.9), lysed by sonication, and cellular debris
was removed by centrifugation (27 200 g, Beckman Coul-
ter J2-21, rotor JA-20, 1 h). Ni-nitrilotriacetic agarose
resin was added to the supernatant (3 mLÆ L
)1
of culture)

and the proteins were purified using affinity chromatogra-
phy with increasing concentration of the imidazole in
buffer A. Purified proteins were concentrated, and buffer
exchanged into buffer B (50 mm Tris ⁄ HCl, pH ¼ 7.9,
2mm EDTA, 2 mm dithiothreitol, and 10% glycerol), ali-
quoted and flash frozen. Protein concentrations were
determined with the Bradford assay using BSA as a
standard.
Expression and purification of the LovB KS-MAT,
LovB MAT, LovB MAT (S208A), LovB MAT
(S209A) and LovB KS-MAT (S656A)
The expression plasmids for each of the mono- and dido-
mains were transformed into BL21(DE3). Cell growth, pro-
tein expression and cell lysis were performed as described
for the ACP proteins. Ni-nitrilotriacetic purification was
used as a first chromatography step to purify all proteins.
The MAT proteins were sufficiently pure (> 95%) after
elution with buffer A + 250 mm imidazole. The MAT and
mutants were then buffer exchanged into buffer B, ali-
quoted and flash frozen. The KS-MAT didomains required
a second purification step, largely due to a major contamin-
ant protein subsequently identified as the truncated, MAT
protein. After buffer exchange of the Ni-nitrilotriaceic puri-
fied protein into buffer B, an anion-exchange chromatogra-
phy was performed using a 5 mL HiTrap-Q column and a
Biologic LP Chromatography system (Bio-Rad, Hercules,
CA, USA). The following gradient was used with a flow
rate of 2 mLÆmin
)1
: buffer B, 8 CV; a linear gradient from

buffer B to buffer C (buffer B + 1 m NaCl), 20 CV; buf-
fer C: 8 CV. The truncated MAT and the intact KS-MAT
eluted between approximately 200–250 mm NaCl with a
majority of MAT eluted first as pure proteins. Pure MAT
and KS-MAT fractions were separately collected, buffer
exchanged into buffer B, aliquoted and flash frozen with
liquid nitrogen. N-terminus Edman sequence analysis of
MAT was performed at the Molecular Structure Facility at
the University of California, Davis, CA, USA.
Labeling of enzymes by [
14
C]-malonyl- and
[
14
C]-acetyl-CoA
Labeling of enzymes were performed in buffer L (100 mm
NaH
2
PO
4
,pH¼ 7.4, 2 mm dithiothreitol, 10% glycerol).
Enzymes (1–10 lm) and radiolabeled acyl-CoA (180 lm,
55 mCiÆmmol
)1
) were added to buffer L (final volume
10 lL) and were incubated either at room temperature or
on ice for 10 min. The reaction was quenched with one vol-
ume of SDS ⁄ PAGE loading buffer lacking any reducing
reagents such as dithiothreitol or b-mercaptoethanol. Sam-
ples were directly loaded onto a 6% or 12% SDS ⁄ PAGE

gel and electrophoresis was performed at 100 V for 90 min.
The gel was dried and analyzed using a phosphoimager
(Packard Instant Imager(tm), Packard Instrument Com-
pany, Meriden, CT, USA).
Transacylation activity of enzymes
The transfer of acyl groups from acyl-CoA to acceptor
ACPs was performed in buffer L. Holo-ACP domain
was each added to a final concentration of 20–100 lm
[
14
C]-Malonyl- and [
14
C]-Acetyl-CoA were added to a final
Minimal polyketide synthase domains in LovB S. M. Ma and Y. Tang
2862 FEBS Journal 274 (2007) 2854–2864 ª 2007 The Authors Journal compilation ª 2007 FEBS
concentration of 180 lm (55 mCiÆmmol
)1
). LovB MAT,
LovB MAT (S208A), LovB KS-MAT and LovB KS-MAT
(S656A) were added to initiate the reaction. The final con-
centration of the enzymes was between 500 nm and 1 lm.
The reaction was performed at room temperature for
30 min. Samples were quenched, separated on SDS ⁄ PAGE
and the radiolabels were visualized by autoradiography.
To perform the kinetic studies, the final acyl-CoA con-
centrations were varied between 10 lm and 200 lm. The
ACP-CON concentration was kept at 50 lm. In the mal-
onylation reaction, LovB MAT and scFabD were added to
final concentrations of 500 nm and 5 nm, respectively. In
the acetylation reaction, both LovB MAT and scFabD

were added to a final concentration of 1 lm. At each time
point (1, 2 and 3 min), an aliquot of the labeling reaction
(10 lL) was quenched with an equal volume of SDS ⁄ PAGE
loading dye and kept on ice. The amount of acyltransfer
was visualized by autoradiography as described above.
Polyketide turnover assay
The polyketide turnover assay was performed as described
previously. The oxytetracycline minimal PKS, oxyA-oxyB
(2 lm) and oxyC (50 lm), we re m ixed with 2 mm [1,2-
14
C]
malonyl-CoA (1.6 mCiÆmmol
)1
) in buffer L (final volume
10 lL). The reaction was initiated by the addition of either
scFabD (0.7 lm) or LovB MAT (0.7 lm or 2.7 lm). A neg-
ative control without either scFabD or LovB MAT was
also performed. After 30 min of incubation at 30 °C, the
reaction mixture was extract twice by 300 lL ethyl acetate
(EA). The combined EA phases were evaporated to
dryness, redissolved in 10 lL of EA. The reaction products
were separated by thin-layer chromatography (TLC)
(EA ⁄ acetic acid, 99 : 1) and quantified with a phos-
phoimager.
Acknowledgements
This work was supported by the American Heart
Association (YT: 0535069N), the UCLA engineering
school faculty start-up grant. We thank Chaitan
Khosla for the expression plasmids for DEBS M3 and
M6 ACPs. YT thanks Ms Alice Chen for helpful

discussions.
References
1 Tobert JA (2003) Lovastatin and beyond: the history of
the HMG-CoA reductase inhibitors. Nat Rev Drug
Discov 2, 517–526.
2 Kennedy J, Auclair K, Kendrew SG, Park C, Vederas
JC & Hutchinson CR (1999) Modulation of polyketide
synthase activity by accessory proteins during lovastatin
biosynthesis. Science 284, 1368–1372.
3 Hopwood DA & Sherman DH (1990) Molecular gene-
tics of polyketides and its comparison to fatty acid bio-
synthesis. Annu Rev Genet 24, 37–66.
4 Heath RJ & Rock CO (2002) The Claisen condensation
in biology. Nat Prod Rep 19, 581–596.
5 Staunton J & Weissman KJ (2001) Polyketide biosyn-
thesis: a millennium review. Nat Prod Rep 18, 380–416.
6 Hendrickson L, Davis CR, Roach C, Nguyen DK,
Aldrich T, McAda PC & Reeves CD (1999) Lovastatin
biosynthesis in Aspergillus terreus: characterization of
blocked mutants, enzyme activities and a multifunc-
tional polyketide synthase gene. Chem Biol 6, 429–439.
7 Donadio S, Staver MJ, McAlpine JB, Swanson SJ &
Katz L (1991) Modular organization of genes required
for complex polyketide biosynthesis. Science 252,
675–679.
8 Keller NP, Turner G & Bennett JW (2005) Fungal sec-
ondary metabolism ) from biochemistry to genomics.
Nat Rev Microbiol 3, 937–947.
9 Moriguchi T, Ebizuka Y & Fujii I (2006) Analysis of
subunit interactions in the iterative type I polyketide

synthase ATX from Aspergillus terreus. Chembiochem 7,
1869–1874.
10 Fujii I, Yoshida N, Shimomaki S, Oikawa H &
Ebizuka Y (2005) An iterative type I polyketide
synthase PKSN catalyzes synthesis of the decaketide
alternapyrone with regio-specific octa-methylation.
Chem Biol 12, 1301–1309.
11 Yu F, Zhu X & Du L (2005) Developing a genetic sys-
tem for functional manipulations of FUM1, a poly-
ketide synthase gene for the biosynthesis of fumonisins
in Fusarium verticillioides. FEMS Microbiol Lett 248,
257–264.
12 Zhu X, Yu F, Li XC & Du L (2007) Production of
dihydroisocoumarins in Fusarium verticillioides by swap-
ping ketosynthase domain of the fungal iterative poly-
ketide synthase Fum1p with that of lovastatin diketide
synthase. J Am Chem Soc 129, 36–37.
13 Rangan VS & Smith S (1996) Expression in Escherichia
coli and refolding of the malonyl- ⁄ acetyltransferase
domain of the multifunctional animal fatty acid syn-
thase. J Biol Chem 271, 31749–31755.
14 Tang Y, Kim CY, Mathews II, Cane DE & Khosla C
(2006) The 2.7-Angstrom crystal structure of a 194-kDa
homodimeric fragment of the 6-deoxyerythronolide B
synthase. Proc Natl Acad Sci USA 103, 11124–11129.
15 Kim CY, Alekseyev VY, Chen AY, Tang Y, Cane DE
& Khosla C (2004) Reconstituting modular activity
from separated domains of 6-deoxyerythronolide B
synthase. Biochemistry 43, 13892–13898.
16 Keatinge-Clay AT & Stroud RM (2006) The structure

of a ketoreductase determines the organization of the
beta-carbon processing enzymes of modular polyketide
synthases. Structure 14
, 737–748.
S. M. Ma and Y. Tang Minimal polyketide synthase domains in LovB
FEBS Journal 274 (2007) 2854–2864 ª 2007 The Authors Journal compilation ª 2007 FEBS 2863
17 Chen AY, Schnarr NA, Kim CY, Cane DE & Khosla
C (2006) Extender unit and acyl carrier protein specifi-
city of ketosynthase domains of the 6-deoxyerythrono-
lide B synthase. J Am Chem Soc 128, 3067–3074.
18 Tsai SC, Lu H, Cane DE, Khosla C & Stroud RM
(2002) Insights into channel architecture and substrate
specificity from crystal structures of two macrocycle-
forming thioesterases of modular polyketide synthases.
Biochemistry 41, 12598–12606.
19 Giraldes JW, Akey DL, Kittendorf JD, Sherman DH,
Smith JL & Fecik RA (2006) Structural and mechanistic
insights into polyketide macrolactonization from poly-
ketide-based affinity labels. Nat Chem Biol 2, 531–536.
20 Tsai SC, Miercke LJ, Krucinski J, Gokhale R, Chen
JC, Foster PG, Cane DE, Khosla C & Stroud RM
(2001) Crystal structure of the macrocycle-forming
thioesterase domain of the erythromycin polyketide
synthase: versatility from a unique substrate channel.
Proc Natl Acad Sci USA 98, 14808–14813.
21 Udwary DW, Merski M & Townsend CA (2002) A
method for prediction of the locations of linker regions
within large multifunctional proteins, and application to
a type I polyketide synthase. J Mol Biol 323, 585–598.
22 Crawford JM, Dancy BC, Hill EA, Udwary DW &

Townsend CA (2006) Identification of a starter unit
acyl-carrier protein transacylase domain in an iterative
type I polyketide synthase. Proc Natl Acad Sci USA
103, 16728–16733.
23 Ma Y, Smith LH, Cox RJ, Beltran-Alvarez P, Arthur
CJ & Simpson FRST (2006) Catalytic relationships
between type I and type II iterative polyketide
synthases: the Aspergillus parasiticus norsolorinic acid
synthase. Chembiochem 7, 1951–1958.
24 Pfeifer BA, Admiraal SJ, Gramajo H, Cane DE &
Khosla C (2001) Biosynthesis of complex polyketides in
a metabolically engineered strain of E. coli. Science 291,
1790–1792.
25 Tang Y, Lee TS, Kobayashi S & Khosla C (2003)
Ketosynthases in the initiation and elongation modules
of aromatic polyketide synthases have orthogonal acyl
carrier protein specificity. Biochemistry 42, 6588–6595.
26 Findlow SC, Winsor C, Simpson TJ, Crosby J & Crump
MP (2003) Solution structure and dynamics of oxytetra-
cycline polyketide synthase acyl carrier protein from
Streptomyces rimosus. Biochemistry 42, 8423–8433.
27 Rangan VS, Witkowski A & Smith S (1991) Isolation of
a functional transferase component from the rat fatty
acid synthase by limited trypsinization of the subunit
monomer. Formation of a stable functional complex
between transferase and acyl carrier protein domains.
J Biol Chem 266, 19180–19185.
28 Rangan VS, Serre L, Witkowska HE, Bari A & Smith S
(1997) Characterization of the malonyl- ⁄ acetyltrans-
acylase domain of the multifunctional animal fatty acid

synthase by expression in Escherichia coli and refolding
in vitro. Protein Eng 10, 561–566.
29 Witkowski A, Ghosal A, Joshi AK, Witkowska HE,
Asturias FJ & Smith S (2004) Head-to-head coiled
arrangement of the subunits of the animal fatty acid
synthase. Chem Biol 11, 1667–1676.
30 Maier T, Jenni S & Ban N (2006) Architecture of mam-
malian fatty acid synthase at 4.5 A resolution. Science
311, 1258–1262.
31 Mikkelsen J, Hojrup P, Rasmussen MM, Roepstorff P
& Knudsen J (1985) Amino acid sequence around the
active-site serine residue in the acyltransferase domain
of goat mammary fatty acid synthetase. Biochem J 227,
21–27.
32 Keatinge-Clay AT, Shelat AA, Savage DF, Tsai SC,
Miercke LJ, O’Connell JD III, Khosla C & Stroud RM
(2003) Catalysis, specificity, and ACP docking site of
Streptomyces coelicolor malonyl-CoA:ACP transacylase.
Structure 11, 147–154.
33 Serre L, Verbree EC, Dauter Z, Stuitje AR & Dere-
wenda ZS (1995) The Escherichia coli
malonyl-CoA:acyl
carrier protein transacylase at 1.5-A resolution. Crystal
structure of a fatty acid synthase component. J Biol
Chem 270, 12961–12964.
34 Stern A, Sedgwick B & Smith S (1982) The free coen-
zyme A requirement of animal fatty acid synthetase.
Participation in the continuous exchange of acetyl and
malonyl moieties between coenzyme a thioester and
enzyme. J Biol Chem 257, 799–803.

35 Rangan VS & Smith S (1997) Alteration of the substrate
specificity of the malonyl-CoA ⁄ acetyl-CoA:acyl carrier
protein S-acyltransferase domain of the multifunctional
fatty acid synthase by mutation of a single arginine resi-
due. J Biol Chem 272, 11975–11978.
36 Summers RG, Ali A, Shen B, Wessel WA & Hutchin-
son CR (1995) Malonyl-coenzyme A:acyl carrier protein
acyltransferase of Streptomyces glaucescens: a possible
link between fatty acid and polyketide biosynthesis.
Biochemistry 34, 9389–9402.
37 Witkowski A, Joshi AK & Smith S (1997) Characteriza-
tion of the interthiol acyltransferase reaction catalyzed
by the beta-ketoacyl synthase domain of the animal
fatty acid synthase. Biochemistry 36, 16338–16344.
38 Moche M, Schneider G, Edwards P, Dehesh K &
Lindqvist Y (1999) Structure of the complex between
the antibiotic cerulenin and its target, beta-ketoacyl-acyl
carrier protein synthase. J Biol Chem 274, 6031–6034.
Minimal polyketide synthase domains in LovB S. M. Ma and Y. Tang
2864 FEBS Journal 274 (2007) 2854–2864 ª 2007 The Authors Journal compilation ª 2007 FEBS