Harnessing the chemical activation inherent to carrier protein-bound thioesters for the characterization of lipopeptide fatty acid tailoring enzymes

Article abstract of DOI:10.1021/ja078081n

Here, we report a new experimental approach utilizing an amide ligation reaction for the characterization of acyl carrier protein (ACP)-bound reaction intermediates, which are otherwise difficult to analyze by traditional biochemical methods. To explore fatty acid tailoring enzymes of the calcium-dependent antibiotic (CDA) biosynthetic pathway, this strategy enabled the transformation of modified fatty acids, covalently bound as thioesters to an ACP, into amide ligation products that can be directly analyzed and compared to synthetic standards by HPLC-MS. The driving force of the amide formation is the thermodynamic activation inherent to thioester-bound compounds. Using this novel method, we were able to characterize the ACP-mediated biosynthesis of the unique 2,3-epoxyhexanoyl moiety of CDA, revealing a new type of FAD-dependent oxidase HxcO with intrinsic enoyl-ACP epoxidase activity, as well as a second enoyl-ACP epoxidase, HcmO. In general, our approach should be widely applicable for the in vitro characterization of other biosynthetic systems acting on carrier proteins, such as integrated enzymes from NRPS and PKS assembly lines or tailoring enzymes of fatty and amino acid precursor synthesis.

Full text of DOI:10.1021/ja078081n

Published on Web 02/01/2008
Harnessing the Chemical Activation Inherent to Carrier
Protein-Bound Thioesters for the Characterization of
Lipopeptide Fatty Acid Tailoring Enzymes
Florian Kopp, Uwe Linne, Markus Oberthu¨r, and Mohamed A. Marahiel*
Department of Chemistry, Philipps-UniVersity Marburg, Hans-Meerwein-Strasse, D-35032
Marburg, Germany
Received October 26, 2007; E-mail: marahiel@chemie.uni-marburg.de
Abstract: Here, we report a new experimental approach utilizing an amide ligation reaction for the
characterization of acyl carrier protein (ACP)-bound reaction intermediates, which are otherwise difficult to
analyze by traditional biochemical methods. To explore fatty acid tailoring enzymes of the calcium-dependent
antibiotic (CDA) biosynthetic pathway, this strategy enabled the transformation of modified fatty acids,
covalently bound as thioesters to an ACP, into amide ligation products that can be directly analyzed and
compared to synthetic standards by HPLC-MS. The driving force of the amide formation is the
thermodynamic activation inherent to thioester-bound compounds. Using this novel method, we were able
to characterize the ACP-mediated biosynthesis of the unique 2,3-epoxyhexanoyl moiety of CDA, revealing
a new type of FAD-dependent oxidase HxcO with intrinsic enoyl-ACP epoxidase activity, as well as a second
enoyl-ACP epoxidase, HcmO. In general, our approach should be widely applicable for the in vitro
characterization of other biosynthetic systems acting on carrier proteins, such as integrated enzymes from
NRPS and PKS assembly lines or tailoring enzymes of fatty and amino acid precursor synthesis.
Introduction
However, today we still lack a detailed understanding of the
biochemical mechanisms underlying fatty acid incorporation in
Among the multitude of medicinally relevant natural products,
nonribosomal lipopeptides such as daptomycin and the calcium-
dependent antibiotic (CDA) have received considerable attention
from basic and applied research. The last-resort antibiotic
daptomycin, for example, is highly active against multidrug
resistant pathogens and represents the first member of a new
structural class of natural antimicrobial agents to be approved
for clinical use in over 30 years.1,2 A structural key feature of
such lipopeptide antibiotics is the eponymous long chain fatty
acid, which is invariably attached to the N-terminus of the cyclic
peptide core. Moreover, straight and branched chain fatty acids
that can significantly differ in the degree of saturation and the
oxidation state (Figure 1) are frequently found and contribute
to the remarkable structural diversity of this class of compounds.
In particular, the lipid portion has a high impact on the biological
properties of these molecules, since antimicrobial behavior and
toxicity are dramatically affected by the nature of the incorpo-
rated fatty acid moiety.2,3
nonribosomal lipopeptides and the dedicated tailoring steps for
the generation of diverse fatty acid building blocks. Recently,
important contributions from the Walsh laboratory elucidated
the activation and tailoring processes involved in the formation
of the â-amino fatty acid moiety of the mixed NRP/PK
mycosubtilin.6-8 The daptomycin and CDA nonribosomal
peptide synthetases (NRPSs) do not contain integrated PKS
elements; instead, the organization of the CDA biosynthetic gene
cluster suggests a different type of fatty acid activation and
modification mediated by stand-alone enzymes working in trans
to the assembly line machinery.9,10 Cloning and sequencing of
the CDA biosynthesis genes revealed a putative fab operon
adjacent to the NRPS, comprised of five open reading frames,
which are predicted to be responsible for the synthesis and
modification of the unique trans-2,3-epoxyhexanoyl moiety of
CDA.10
Within this operon, the hxcO gene encodes for an enzyme
that shows strong sequence homology to FAD-dependent acyl-
Intensive studies on modular biosynthetic assembly line
machinery have provided researchers with a profound knowl-
edge of how nonribosomal peptides (NRP) and polyketides (PK)
used in different therapeutic areas are produced in nature.4,5
(5) Grunewald, J.; Marahiel, M. A. Microbiol. Mol. Biol. ReV. 2006, 70, 121-
146.
(6) Aron, Z. D.; Dorrestein, P. C.; Blackhall, J. R.; Kelleher, N. L.; Walsh, C.
T. J. Am. Chem. Soc. 2005, 127, 14986-14987.
(7) Hansen, D. B.; Bumpus, S. B.; Aron, Z. D.; Kelleher, N. L.; Walsh, C. T.
J. Am. Chem. Soc. 2007, 129, 6366-6367.
(1) Kirkpatrick, P.; Raja, A.; LaBonte, J.; Lebbos, J. Nat. ReV. Drug DiscoVery
2003, 2, 943-944.
(8) Aron, Z. D.; Fortin, P. D.; Calderone, C. T.; Walsh, C. T. ChemBioChem
2007, 8, 613-616.
(2) Baltz, R. H.; Miao, V.; Wrigley, S. K. Nat. Prod. Rep. 2005, 22, 717-
(9) Miao, V.; Coeffet-Legal, M. F.; Brian, P.; Brost, R.; Penn, J.; Whiting, A.;
Martin, S.; Ford, R.; Parr, I.; Bouchard, M.; Silva, C. J.; Wrigley, S. K.;
Baltz, R. H. Microbiol. 2005, 151, 1507-1523.
741.
(3) Debono, M.; Abbott, B. J.; Molloy, R. M.; Fukuda, D. S.; Hunt, A. H.;
Daupert, V. M.; Counter, F. T.; Ott, J. L.; Carrell, C. B.; Howard, L. C.;
Boeck, L. D.; Hamill, R. L. J. Antibiot. (Tokyo) 1988, 41, 1093-1105.
(4) Fischbach, M. A.; Walsh, C. T. Chem. ReV. 2006, 106, 3468-3496.
(10) Hojati, Z.; Milne, C.; Harvey, B.; Gordon, L.; Borg, M.; Flett, F.; Wilkinson,
B.; Sidebottom, P. J.; Rudd, B. A.; Hayes, M. A.; Smith, C. P.; Micklefield,
J. Chem. Biol. 2002, 9, 1175-1187.
9
2656
J. AM. CHEM. SOC. 2008, 130, 2656-2666
10.1021/ja078081n CCC: $40.75 © 2008 American Chemical Society

Lipopeptide Fatty Acid Tailoring Enzymes
A R T I C L E S
Figure 1. Chemical structures of nonribosomal peptides containing modified fatty acid building blocks. Unsaturated fatty acids (blue) and Câ-modified
fatty acids (orange) are predominantly found. The calcium-dependent antibiotic (CDA) contains a trans-2,3-epoxyhexanoyl moiety (red) with an unknown
absolute configuration.
CoA oxidases and dehydrogenases, enzymes responsible for the
desaturation of their thioester substrates in fatty acid catabo-
lism.11 In close proximity, hcmO encodes a putative flavin-
dependent enoyl-CoA epoxidase, homologous to salicylate
hydoroxylase and zeaxanthin epoxidase,12 which was proposed
to be responsible for the addition of the epoxy function to the
enoyl-CoA substrate. Based on the sequence homologies, it was
assumed that both enzymes act on CoA substrates.10 In contrast
to this, our in vitro experiments reported here reveal that the
HxcO- and HcmO-catalyzed transformations occur on ACP-
bound substrates.
In order to examine enzymatic epoxide formation in detail,
we were confronted with several hurdles, since standard
methodology for the cleavage and analysis of carrier protein
(CP)-bound reaction products cannot be applied in this case.
First, acyl carrier protein (ACP)-bound compounds are not
recognized by the thioesterase II (TEII) for enzymatic cleav-
age,13 and the use of ACP hydrolases is restricted by their high
substrate specificity for native ACPs.14,15 Second, although ACP-
bound thioesters can be cleaved under alkaline conditions, the
carboxylic acids obtained are often difficult to analyze, e.g. by
HPLC, which makes additional derivatization steps necessary.
Finally, cleavage of an ACP-bound epoxy acid under rather
harsh basic conditions would obviously result in the racemiza-
tion or opening of the relatively unstable epoxy function. To
circumvent these problems, we developed a new experimental
approach, utilizing an amide ligation reaction that allows a
straightforward characterization of reaction intermediates and
products of ACP-mediated pathways. By applying this strategy,
we demonstrate that HxcO is a novel type of oxidase with
intrinsic epoxidase activity, which suggests a different overall
view of the synthesis of the 2,3-epoxyhexanoyl chain during
CDA assembly. This approach allowed us to successfully
characterize the biosynthesis of modified fatty acids in the CDA
biosynthetic pathway and can in general be useful for the in
vitro characterization of other biosynthetic processes occurring
on CPs, such as fatty and amino acid precursor formation or
modifications catalyzed by integrated enzymes from NRPS and
PKS assembly lines.
Experimental Procedures
Cloning of hxcO, hcmO, and ACP (SCO3249) Genes. The hxcO,
hcmO, and ACP (SCO3249) genes were amplified from genomic DNA
of Streptomyces coelicolor A3(2) by PCR using the primer pairs listed
in the Supporting Information (Table S1). PCRs were carried out with
Pfu polymerase (Finnzymes) following the instructions of the manu-
facturer for GC-rich DNA templates. The resulting amplicons were
digested with the enzymes indicated in Table S1 and cloned into
(11) Thorpe, C.; Kim, J. J. FASEB J. 1995, 9, 718-725.
(12) Buch, K.; Stransky, H.; Hager, A. FEBS Lett. 1995, 376, 45-48.
(13) Schwarzer, D.; Mootz, H. D.; Linne, U.; Marahiel, M. A. Proc. Natl. Acad.
Sci. U.S.A. 2002, 99, 14083-14088.
(14) Thomas, J.; Cronan, J. E. J. Biol. Chem. 2005, 280, 34675-34683.
(15) Mercer, A. C.; Burkart, M. D. Nat. Prod. Rep. 2007, 24, 750-773.
9
J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008 2657

A R T I C L E S
Kopp et al.
appropriate restriction sites of a pET-28a(+) (Novagen) for hxcO and
hcmO and a derivatized pQTev (Quiagen) for ACP, respectively. The
resulting overexpression plasmids were checked by DNA sequencing
and transformed into E. coli BL21(DE3) for protein production.
Heterologous Expression and Purification of HxcO, HcmO, and
ACP. His-tagged proteins were overexpressed in E. coli BL21 (DE3)
transformed with their respective overexpression plasmids. Cultures
(0.5 L) in 2 L Erlenmeyer flasks were inoculated with 5 mL of an
overnight culture and grown to an optical density of 0.5 at 37 °C. The
cultures were cooled to 16 °C, and protein production was induced by
the addition of IPTG to a final concentration of 0.1 mM. The cultures
were incubated for another 14 to 16 h and then harvested by
centrifugation.
For purification of ACP, HxcO, and HcmO proteins, cell pellets
from 5 L of culture were resuspended in 50 mL of buffer A (50 mM
Hepes, pH 8.0, 300 mM NaCl) and disrupted by the use of an
EmulsiFlex-C5 High Pressure Homogenizer (Avestin). After centri-
fugation at 27 000 g for 30 min, the supernatant was carefully re-
moved and applied to Ni-NTA chromatography on an FPLC system
(Amersham Pharmacia Biotech). Briefly, the protein raw extract was
run over a column filled with 500 µL of Ni-NTA superflow resin
(Qiagen) with a flow rate of 0.5 mL/min. The bound proteins were
washed with buffer A containing 2% buffer B (50 mM Hepes, pH 8.0,
300 mM NaCl, 250 mM imidazole) for 5 min and eluted by applying
a linear gradient of 2-50% buffer B with a flow rate of 0.7 mL/min
over 30 min.
Fractions containing the recombinant proteins were monitored by
SDS-PAGE, pooled, and dialyzed against buffer C (25 mM Hepes,
pH 7.0, 50 mM NaCl) using Hi-Trap desalting columns (GE Health-
care). For preparative expressions of HxcO and HcmO, buffer C was
supplied with a 5-fold molar excess of FAD cofactor. The recombinant
proteins could be stored at -20 °C for 3 months without significant
loss of activity.
HPLC Analysis of HxcO and HcmO Flavin Cofactor. A 50 µL
sample of purified HxcO (30 µM) or HcmO (30 µM) was boiled for
5 min, and denatured protein was removed by centrifugation. The flavin
present in the supernatant was analyzed by HPLC on a C18ec Nucleodur
column (Macherey and Nagel, 250/2, pore diameter of 100 Å, particle
size of 3 µm) with the following gradient: 0-70% acetonitrile,
0.1% TFA in water, 0.1% TFA from 0 to 30 min at a flow rate of 0.2
mL/min. Product elution was monitored at 445 nm (Figure S2).
Additionally, the identity of the cofactor was proven by mass
spectroscopy (data not shown).
N-(Hex-2,3-epoxyhexanoyl)-D-Phe-OMe (8 (2R,3S) and 9 (2S,3R)).
The stereoselective preparation of 8 from (2S,3S)-2,3-epoxyhexanol (4),
obtained by Sharpless epoxidation of hex-2-enol,16 is described. (2S,3S)-
Epoxy alcohol 4 (115 mg, 1.0 mmol) was dissolved in acetonitrile/
carbon tetrachloride (1:1, 5 mL). To this solution, water (5 mL),
ruthenium trichloride monohydrate (5 mg), and sodium periodate
(1.0 g, 4.7 mmol) were added, and the mixture was stirred vigorously
for 16 h at room temperature. Following the addition of dichloromethane
(50 mL) and brine (50 mL), the emulsion was filtered through Celite,
upon which phase separation occurred. The organic phase was dried
(MgSO4), filtered, and concentrated. Flash chromatography (SiO2,
dichloromethane/methanol/AcOH 10:1:0.1) of the residue gave the
crude (2R,3S)-acid 6 (100 mg) as a dark, amorphous solid, which was
used directly for the next step. To a solution of crude (2R,3S)-acid 6,
DIPEA (0.42 mL, 2.4 mmol), and HOBt monohydrate (185 mg,
1.2 mmol) in dichloromethane (3 mL) was added EDCI hydrochloride
(230 mg, 1.2 mmol) at 0 °C with stirring. After 5 min, D-Phe-OMe
hydrochloride (2, 260 mg, 1.2 mmol) was added, and stirring was
continued at room temperature for 16 h. The mixture was diluted with
dichloromethane, washed with 1 N HCl, satd. aq. NaHCO3 solution,
and brine, dried (MgSO4), filtered, and concentrated. Flash chroma-
tography (SiO2, petroleum ether/ethyl acetate 2:1 f 1:1) of the residue
gave amide 8 (125 mg, 43% over 2 steps) as a colorless syrup. Rf )
0.44 (petroleum ether/ethyl acetate 2:1); 1H NMR (300 MHz, CDCl3):
δ 7.16-7.24, 6.98-7.01 (2m, 3H and 2H, arom. H), 6.43 (d, 1H, JR,NH
) 7.9 Hz, NH), 4.76 (ddd, 1H, JR,âa ) 5.8, JR,âb ) 7.0, JR,NH ) 7.9 Hz,
R-H), 3.68 (s, 3H, OMe), 3.13 (dd, 1H, JR,â ) 5.8, Jâ,â ) 14.0 Hz,
â-Ha), 3.08 (d, 1H, J2,3 ) 2.1 Hz, 2-H), 2.94 (dd, 1H, JR,â ) 7.0, Jâ,â
) 14.0 Hz, â-Hb), 2.52 (ddd, 1H, J2,3 ) 2.1, J3,4a ) 4.7, J3,4b ) 5.8 Hz,
3-H), 1.30-1.48 (m, 4H, 4-H2, 5-H2), 0.88 (t, 3H, J5,6 ) 7.1 Hz, 6-H3);
13C NMR (75 MHz, CDCl3): δ 171.4 (COOMe), 168.3 (CONH), 135.7,
129.2, 128.6, 127.2 (arom. C), 59.4 (C-3), 55.1 (C-2), 52.4 (OMe),
52.2 (C-R), 37.8 (C-â), 33.5 (C-4), 18.9 (C-5), 13.6 (C-6); ESI-MS
calcd for C16H21NO4: 291.15, found 291.1 [M]+.
HPLC retention time (gradient: 0-60% acetonitrile, 0.1% TFA in
water, 0.1% TFA from 0 to 20 min at a flow rate of 0.4 mL/min on a
ChiraDex Gamma column (Merck KGaA, Darmstadt, Germany, 250/
4, particle size of 5 µm): t ) 15.4 min.
For the synthesis of the diastereomeric mixture 8/9, racemic
2,3-epoxyhexanol (4/5), obtained by mCPBA oxidation of hex-2-enol,17
was used in an analogous fashion. As expected, HPLC analysis using
the conditions described above showed two peaks of equal intensity
(8: t ) 15.4 min, 9: t ) 14.4 min). NMR data for 9: 1H NMR (300
MHz, CDCl3): 1H NMR (300 MHz, CDCl3): δ 7.16-7.24, 7.00-
7.04 (2 m, 3H and 2H, arom. H), 6.42 (d, 1H, JR,NH ) 8.2 Hz, NH),
4.77 (dt, 1H, JR,â ) 6.0, JR,NH ) 8.2 Hz, R-H), 3.63 (s, 3H, OMe), 3.13
(d, 1H, J2,3 ) 2.0 Hz, 2-H), 3.03 (t, 2H, JR,â ) 6.0 Hz, â-H2), 2.87 (m,
1H, 3-H), 1.35-1.52 (m, 4H, 4-H2, 5-H2), 0.88 (t, 3H, J5,6 ) 7.0 Hz,
6-H3); 13C NMR (75 MHz, CDCl3): δ 171.5 (COOMe), 168.4 (CONH),
135.5, 129.1, 128.7, 127.2 (arom. C), 59.4 (C-3), 55.0 (C-2), 52.3 (2x,
OMe and C-R), 37.9 (C-â), 33.6 (C-4), 18.9 (C-5), 13.8 (C-6).
Synthesis of Fatty Acid-CoA Substrates. Coenzyme A trilithium
salt (0.05 mmol), the fatty acid (0.10 mmol), PyBOP (0.08 mmol),
and K2CO3 (0.20 mmol) were dissolved in 4 mL of THF/water (1:1)
and incubated for 2 h at room temperature. After lyophilization, the
resulting white solids were dissolved in water and purified by HPLC
(Agilent, 1100 series) on a preparative Nucleodur C18ec column
(Macherey and Nagel, 250/2, pore diameter of 100 Å, particle size of
3 µm) with a gradient of 5-70% acetonitrile, 0.1% TFA in water, 0.1%
TFA over 30 min at a flow rate of 20.0 mL/min. Product elution was
monitored at 215 nm, and the identity was confirmed by MALDI-TOF
MS analysis (Table S2).
Synthesis of D-Phe-OMe Amides 3, 8, and 9. N-(Hex-2-enoyl)-
D-Phe-OMe (3). To a solution of trans-hex-2-enoic acid (1, 230 mg,
2.0 mmol), DIPEA (0.77 mL, 4.4 mmol), and HOBt monohydrate
(398 mg, 2.6 mmol) in dichloromethane (7 mL) was added EDCI (500
mg, 2.6 mmol) at 0 °C with stirring. After 5 min, D-Phe-OMe
hydrochloride (2, 430 mg, 2.0 mmol) was added, and stirring was
continued at room temperature for 16 h. The mixture was diluted with
dichloromethane, washed with 1 N HCl, satd. aq. NaHCO3 solution,
and brine, dried (MgSO4), filtered, and concentrated. Flash chroma-
tography (SiO2, petroleum ether/ethyl acetate 2:1 f 1:1) of the residue
gave amide 3 (510 mg, 92%) as a colorless syrup. Rf ) 0.44 (petroleum
1
ether/ethyl acetate 2:1); H NMR (300 MHz, CDCl3): δ 7.16-7.24,
7.00-7.04 (2m, 3H and 2H, arom. H), 6.78 (dt, 1H, J2,3 ) 15.3, J3,4
7.0 Hz, 3-H), 5.88 (bd, 1H, JR,NH ) 7.7 Hz, NH), 5.70 (dt, 1H, J2,3
)
)
15.3, J2,4 ) 1.5 Hz, 2-H), 4.89 (dt, 1H, JR,âa ) JR,âb ) 5.6, JR,NH ) 7.7
Hz, R-H), 3.65 (s, 3H, OMe), 3.11 (dd, 1H, JR,â ) 5.6, Jâ,â ) 13.9 Hz,
â-Ha), 3.06 (dd, 1H, JR,â ) 5.6, Jâ,â ) 13.9 Hz, â-Hb), 2.07 (dq, 2H,
J2,4 ) 1.5, J3,4 ) J4,5 ) 7.0 Hz, 4-H2), 1.40 (sext, 2H, J4,5 ) J5,6 ) 7.0
Hz, 5-H2), 0.85 (t, 3H, J5,6 ) 7.0 Hz, 6-H3); 13C NMR (75 MHz,
CDCl3): δ 172.1 (COOMe), 165.4 (CONH), 145.6 (C-3), 135.9, 129.3,
128.5, 127.1 (arom. C), 123.1 (C-2), 53.1 (C-R), 52.3 (OMe), 37.9 (C-
â), 34.1 (C-4), 21.4 (C-5), 13.6 (C-6); ESI-MS calcd for C16H21NO3:
275.15, found 275.1 [M]+.
(16) Gao, Y.; Hanson, R. M.; Klunder, J. M.; Ko, S. Y.; Masamune, H.;
Sharpless, K. B. J. Am. Chem. Soc. 1987, 109, 5765-5780.
(17) Thurner, A.; Faigl, F.; To¨ke, L.; Mordini, A.; Vallachi, M.; Reginato, G.;
Czira, G. Tetrahedron 2001, 57, 8173-8180.
9
2658 J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008

Lipopeptide Fatty Acid Tailoring Enzymes
A R T I C L E S
Figure 2. In vitro phosphopantetheinylation of apo-ACP. (A) Coomassie stained SDS-PAGE (left) and in-gel fluorescence (right) of the ACP laoding
mixture with (+) and without (-) Sfp, the phosphopantetheine transferase, loading from B. subtilis. (B) FTMS broad band spectrum of ACP (c.m.: 11464.7)
before and after loading with hexanoic acid. (c.m.: 11901.9).
Chemoenzymatic Synthesis of CDA Analogues. The chemoenzy-
matic synthesis of hexanoyl-CDA and hex-2-enoyl-CDA was carried
out as described previously.18 Briefly, linear peptidyl substrates were
prepared on a 0.1 mmol scale via Fmoc solid-phase peptide synthesis
and C-terminally activated with a thiophenol leaving group. After
deprotection and preparative HPLC purification, the linear substrates
were enzymatically cyclized by CDA TE domain in a semipreparative
scale and purified by HPLC yielding macrocyclic hexanoyl-CDA and
hex-2-enoyl-CDA. The identity of the products was confirmed by
MALDI-TOF (Table S3).
HcmO Epoxidation Assay. Typical incubations (150 µL) for the
detection of HcmO epoxidation products were prepared using 100 µM
loaded acyl-ACP substrate and 10 µM HcmO, 250 µM FAD, and
250 µM NAD(P)H in assay buffer. After 30 min at 25 °C, 10 µg of
trypsin (Promega) were added and incubated for another 5 min at
30 °C. The tryptic digest was stopped by the addition of 15 µL of
formic acid, and the reaction volumes were analyzed by ESI-FTICR-
MS as reported for HxcO assays.
Analysis of ACP-Bound Products by Direct Amide Ligation. To
further characterize the ACP-bound products of HxcO and HcmO
reactivity the complete reaction mixtures were incubated with a 1000-
fold excess of D-Phe-OMe (2) at 60 °C for 2 h. The precipitated proteins
were removed by centrifugation. Protein pellets were washed twice with
25 µL of assay buffer, and the combined fractions were concentrated
to 50 µL prior to HPLC-MS analysis.
For epoxide product detection, the reaction volumes were compared
to the synthesized standard using the following gradient: 0-60%
acetonitrile, 0.1% TFA in water, 0.1% TFA from 0 to 20 min at a flow
rate of 0.4 mL/min on a ChiraDex Gamma column (Merck KGaA,
Darmstadt, 250/4, particle size of 5 µm).
The HPLC-MS analysis of the 2,3-unsaturated fatty acid was carried
out on a standard C18ec Nucleodur column (Macherey and Nagel, 250/
2, pore diameter of 100 Å, particle size of 3 µm) with the following
gradient: 0-60% acetonitrile, 0.1% TFA in water, 0.1% TFA from 0
to 20 min at a flow rate of 0.2 mL/min.
In Vitro 4′-Phosphopantetheinylation of ACP. A reaction mixture
containing 200 µM fluorescein-CoA/acetyl-CoA, 200 µM ACP, 10 mM
MgCl2, and 50 µM recombinant Bacillus subtilis 4′-phosphopantetheine
transferase (PPTase) Sfp in assay buffer (buffer C) was incubated at
30 °C for 30 min and analyzed directly by LC-ESI-MS. The loading
of fluorescein-CoA, which was prepared as previously reported,19 onto
the ACP was monitored by measuring the in gel fluorescence.
Assays with ACP-bound Acyl Substrates. Acyl-phosphopanteth-
einylation of the ACP was achieved by incubating the carrier protein
with Sfp and acyl-CoA substrates. The reaction mixture was incubated
at 30 °C for 30 min and contained 200 µM ACP, 200 µM Acyl-CoA
substrate, 2 mM MgCl2, and 50 µM Sfp in a 25 mM HEPES buffer
pH 7.5 (assay buffer). The solution was desalted using a Micro Bio-
Spin 6 column (Bio-Rad).
HxcO Oxidation/Epoxidation Assay. A typical reaction mixture
(150 µL) for the detection of HxcO epoxidation product was prepared
using 100 µM loaded acyl-ACP substrate, 5 µM HxcO, and 250 µM
FAD in assay buffer. After incubation at 25 °C for 30 min, 10 µg of
trypsin (Promega) were added to the reaction mixture and incubated
for 5 min at 30 °C. Finally, 15 µL of formic acid were added, and the
samples were subjected to HPLC and high-resolution ESI-FTICR-MS
analysis carried out with an LTQ-FT mass spectrometer (Finnigan,
Bremen) using the following gradient: 25-50% acetonitrile, 0.045%
formic acid in water, 0.05% formic acid over 30 min at a flow rate of
0.2 mL/min on a Jupiter C4 column (Phenomenex, 150 mm × 2 mm,
5 µm) at 40 °C. To achieve maximum sensitivity of the described ESI-
FTICR-MS method, the ms experiment was carried out at single-
reaction monitoring (SRM) mode of the LTQ mass analyzer. Masses
selected for MS2 were 1046.5 ([M + H]2+ of hexenoyl-S-Ppan-frag-
ment) and 1054.5 ([M + H]2+ of 2,3-epoxy-hexenoyl-S-Ppan-fragment).
Assays with Acyl-CoA Substrates and Chemoenzymatically
Derived CDA Analogues. HxcO and HcmO assays with acyl-CoA
substrates and chemoenzymatically derived CDA analogs were carried
out as reported in the Supporting Information.
Results
Characterization of SCO3249 as an Acyl Carrier Protein.
Similar to the synthesis of NRPs and PKs, the assembly of fatty
acids catalyzed by Fatty Acid Synthases (FASs) occurs on CPs,
to which the substrates, reaction intermediates, and products
are covalently tethered during biosynthesis.20 SCO3249 of the
CDA cluster, in the following referred to as ACP, showed strong
sequence homology to CPs from NRPS and PKS and was likely
to be involved in the biosynthesis of the CDA lipid portion.
The N-terminally His7-tagged ACP was overproduced in E.
coli BL21(DE3) and purified as described in the Experimental
(18) Gru¨newald, J.; Sieber, S. A.; Marahiel, M. A. Biochemistry 2004, 43, 2915-
2925.
(19) La Clair, J. J.; Foley, T. L.; Schegg, T. R.; Regan, C. M.; Burkart, M. D.
Chem. Biol. 2004, 11, 195-201.
(20) Finking, R.; Marahiel, M. A. Annu. ReV. Microbiol. 2004, 58, 453-488.
9
J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008 2659

A R T I C L E S
Kopp et al.
Figure 3. HxcO assay. (A) Experimental setup. HxcO reaction products were digested with trypsin and subjected to HPLC prior to tandem FTMS spectrometry.
(B) Single reaction monitoring (SRM; 1046.5 f [1732.30-1737.30]) of hexanoyl-S-ACP (Lys57-Arg71) and MS2 data of the ppan ejection assay resulting
from a reference assay without HxcO. (C) SRM (1054.5 f [1732.30-1737.30]) of epoxyhexanoyl-S-ACP (Lys57-Arg71) and MS2 data of the ppan ejection
assay resulting from a HxcO assay. (D) SRM (1046.5 f [1732.30-1737.30]) of hex-2-enoyl-S-ACP (Lys57-Arg71) and MS2 data of the ppan ejection assay
resulting from a HxcO assay. This HxcO side product is formed in ∼15% compared to epoxyhexanoyl-S-ACP (indicated by areas in C and D).
Procedures yielding 8 mg of protein per liter of culture. The
identity of the expressed protein was proven by mass spec-
trometry and SDS-PAGE (Figure S1). To test whether ACP
could be loaded in vitro, the recombinant protein was incubated
with the 4′-phosphopantetheine transferase (PPTase) (Sfp) from
Bacillus subtilis and fluorescein-CoA. The 4′-phosphopanteth-
einylation of ACP was monitored by the in-gel fluorescence of
the loading mixture (Figure 2 A); shown are positive and
negative control experiments. Additionally, the loading with
fatty acid CoA-substrates to the ACP was verified by mass
spectroscopy (Figure 2 B).
Characterization of HxcO as a Fatty Acid-S-ACP Oxidase
with Intrinsic Epoxidase Activity. HxcO shows sequence
homology in the range of 30% to several members of the acyl-
CoA dehydrogenase/oxidase superfamily that catalyze double
bond formation between C-2 and C-3 of their thioester
substrates.11 Based on the sequence homology, it was also
postulated that HcmO is responsible for the oxidation of a
hexanoyl-CoA substrate to 2,3-hexenoyl-CoA.10 Alternatively,
different scenarios with an ACP-bound HxcO substrate or the
possibility that the oxidation occurs at the stage of the already
assembled CDA molecule are conceivable. Accordingly, in vitro
assays to examine all three possible routes were conducted.
The N-terminally His7-tagged HxcO was overproduced in
E. coli BL21(DE3) and purified yielding ∼1.0 mg of protein
per liter of culture. The identity of the protein was confirmed
by mass spectrometry and SDS-PAGE (Figure S1). HxcO was
colorless and showed no spectral absorbance typical for flavin
cofactors. However, HPLC analysis of HxcO supernatant after
heat denaturation revealed the presence of FAD as the electron
accepting cofactor (Figure S2). This observation suggests that
HxcO utilizes FAD as a free coenzyme.
To test if acyl-CoAs are the physiological substrates of HxcO,
the purified enzyme was assayed with synthetic hexanoyl-CoA
9
2660 J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008

Lipopeptide Fatty Acid Tailoring Enzymes
A R T I C L E S
Scheme 1. Synthesis of Chemical Standardsa
a
EDCI ) 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride, HOBt ) 1-hydroxybenzotriazole monohydrate.
under a variety of assay conditions typical for acyl-CoA
dehydrogenases and oxidases.21,22 No oxidation products were
detected by HPLC-MS. Additional substrates with a chain length
from 4 to 10 C-atoms (Figure S3) did not reveal the enzymatic
introduction of a double bond. Analogously, experiments with
chemoenzymatically generated hexanoyl-CDA showed no fatty
acid oxidation, indicating that HxcO does not act on the mature
CDA molecule.
Next, Sfp, which had previously been shown to be promiscu-
ous for the in vitro phosphopantetheinylation of various acyl-
and peptidyl-CoA substrates onto carrier proteins,23 was em-
ployed to assay the alternative ACP-mediated pathway. The
required hexanoyl-ACP substrate was prepared and incubated
with HxcO. After tryptic digest, the samples were subjected to
HPLC high-resolution Fourier-transform mass spectrometry
(FTMS) to analyze product formation (Figure 3).
FADH2, catalase was added to the assay in different concentra-
tions. Further, reaction buffers with different pH values in the
range from 5 to 9 were tested to decompose potentially formed
H2O2. No influence on product formation was observed.
Therefore, epoxyhexanoyl-S-ACP was clearly identified to be
the main product of HxcO reactivity. HxcO displayed epoxide
formation (Figure S4) with a kobs ) 0.3 min-1, whereas double
bond formation did not increase over the examined period of
time.
The substrate specificity of HxcO was evaluated with
additional linear fatty acids (4-10 C-atoms) loaded onto an ACP
(Figure S3). As with the physiological substrate, these reactions
revealed the formation of the expected enoyl product paired
with the most apparent species (∼85%), a fragment ion with a
mass of +16 compared to the enoyl product.
To further characterize the ACP-bound product formed by
HxcO, we envisioned an HPLC-based comparison with synthetic
standards. We therefore had to establish conditions that would
allow the efficient transformation of the modified fatty acid-S-
ACPs into derivatives of smaller size, which would then in turn
be analyzed by HPLC. Initial experiments using an excess of a
simple amine showed that this approach was indeed valid. For
example, after incubation of an HxcO reaction with benzyl-
amine, LC-MS analysis of the reaction mixture showed the
formation of the expected mass for the amide ligation products
(203.2 and 219.2 Da, data not shown). However, we were unable
to determine the stereochemistry of the epoxide formed by HxcO
by comparison with a synthetic standard as the two enantiomeric
benzyl amides (2S,3R and 2R,3S) showed the same retention
time using a chiral cyclodextrin-based column (data not shown).
Therefore, we turned our attention to chiral ligation partners.
In this case, the possible products are diastereomeric instead of
enantiomeric and therefore hopefully easier to distinguish by
HPLC analysis. Accordingly, we oxidized a racemic mixture
of 2,3-epoxyhexanols 4/5 with RuCl3/NaIO425 and coupled the
resulting carboxylic acids 6/7 with different chiral amines using
EDCI/HOBt. Best results in terms of separation of the reaction
products were obtained using the amino acid derivative D-Phe-
OMe (2, Scheme 1) as a ligation partner and a chiral column
for the HPLC analysis of the diastereomeric mixture of amides
8 and 9 (Scheme 1). Phenylalanine derivative 2 could also be
used successfully for the analysis of the HxcO reaction. Under
optimized conditions, HxcO reaction volumes containing the
reactive fatty acid-S-ACP thioester products were incubated with
a 1000-fold molar excess of D-Phe-OMe (2), which afforded
The ppan ejection assay24 of samples obtained from negative
control reaction volumes without the enzyme resulted in the
observation of a 1734.8 Da signal corresponding to the active
site ACP fragment (Lys57-Arg71) without the phosphopantetheine
(ppan) arm that was cleaved off during gas-phase fragmentation.
Additionally, a fragment ion of 359.2 Da was detected, which
is consistent with the calculated mass of the hexanoyl-ppan
MS-MS fragmentation product (Figure 3B). After incubation
with HxcO, a fragment ion of 373.2 Da (Figure 3C) and a
second signal of 357.2 Da (Figure 3D) with ∼15% intensity
compared to the 373.2 signal could be detected. These masses
are consistent with the calculated masses of the hex-2-enoyl-
ppan (357.2 Da) and the epoxyhexanoyl-ppan (373.2 Da)
fragmentation products. HxcO showed no conversion of the
unsaturated fatty acid-S-ACP in the presence of FAD and
NAD(P)H cofactors, which are commonly used for the genera-
tion of FADH2 by flavin dependent monooxygenases. This
suggests that HxcO generates FADH2 in situ by the oxidation
of the saturated fatty acid substrate. The reduced cofactor is
then the reactive species for the epoxidation reaction.
To prove whether the unexpected epoxyhexanoyl-product,
which was predominantly found (∼85%), is of enzymatic origin
or caused by H2O2 formed in vitro during the reoxidation of
(21) Lea, W.; Abbas, A. S.; Sprecher, H.; Vockley, J.; Schulz, H. Biochim.
Biophys. Acta 2000, 1485, 121-128.
(22) Kieweg, V.; Krautle, F. G.; Nandy, A.; Engst, S.; Vock, P.; Abdel-Ghany,
A. G.; Bross, P.; Gregersen, N.; Rasched, I.; Strauss, A.; Ghisla, S. Eur. J.
Biochem. 1997, 246, 548-556.
(23) Sieber, S. A.; Walsh, C. T.; Marahiel, M. A. J. Am. Chem. Soc. 2003, 125,
10862-10866.
(24) Dorrestein, P. C.; Kelleher, N. L. Nat. Prod. Rep. 2006, 23, 893-918.
9
J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008 2661

A R T I C L E S
Kopp et al.
Figure 4. Characterization of HxcO and HcmO reaction products via direct carrier protein amide ligation. (A) Cleavage of ACP-bound thioesters with a
nucleophile (D-Phe-OMe) allows the comparison of enzymatic reaction products bound to an ACP with synthetic standards by HPLC-MS. (B) Amide
ligation products of HxcO reactions coelute with the (2R,3S)-2,3-epoxyhexanoyl-D-Phe-OMe standard 8 ([M + H]+ ) 291.1). The HxcO side product was
proven to be identical with the synthetic hexenoyl amide 3 ([M + H]+ ) 275.1). (C) The reaction product of HcmO coelutes with the (2S,3R)-2,3-epoxyhexanoic
acid 9 ([M + H]+ ) 291.1). Shown are extracted ion chromatograms (EICs).
amide ligation products suitable for LC-MS analysis. The
chromatograms clearly evidenced that the epoxide generated
by HxcO is formed as a single enantiomer (Figure 4B). In order
to establish the exact stereochemistry of the enzyme product,
we next synthesized one of the stereoisomers, i.e., amide 8,
selectively. For this purpose, enantiomer 4, obtained by Sharp-
less epoxidation of hex-2-enol,16 was oxidized25 and coupled
with D-Phe-OMe 2 to provide the (2R,3S)-configured 8 (>90%
ee, Scheme 1). Comparison of the HPLC retention time of
stereoisomer 8 with the two peaks obtained from the diastere-
omeric mixture 8/9 then led to the structural assignments shown
in Figure 4B and C. The HPLC trace of a HxcO reaction
established that the HxcO reaction product has the (2R,3S)-
configuration (Figure 4B). The formation of the 2,3-hexenoic
acid side product, which occurred only to a minor extent, was
also proven by comparison with the synthetic standard 3, which
was obtained by EDCI/HOBt-mediated coupling of hex-2-enoic
acid (1) and D-Phe-OMe (2, Scheme 1).
Characterization of HcmO as a Fatty Acid-S-ACP Ep-
oxidase. HcmO shows amino acid homology to the diverse class
of flavin-dependent monooxygenases, particularly salicylate
hydroxylases26 and the zeaxanthin epoxidases12 in the range of
40% and 30%, respectively. To determine whether HcmO acts
on ACP-bound hexenoyl-substrates or on coenzyme A thioesters,
both possibilities were assayed.
The N-terminally His7-tagged HcmO was produced in E. coli
BL21(DE3) and purified with a yield of ∼1.0 mg protein per
liter of culture. The identity of the protein was proven by mass
(25) Vorkonov, M. V.; Gontcharov, A. V.; Wang, Z.; Richardson, P. F.; Kolb,
(26) Ortiz-Maldonado, M.; Ballou, D. P.; Massey, V. Biochemistry 1999, 38,
8124-8137.
H. C. Tetrahedron 2004, 60, 9043-9048.
9
2662 J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008

Lipopeptide Fatty Acid Tailoring Enzymes
A R T I C L E S
Figure 5. HcmO assay. (A) Experimental setup. HcmO reaction products were digested with trypsin and subjected to HPLC and tandem FTMS spectrometry.
(B) Single reaction monitoring (SRM; 1046.5 f [1732.30-1737.30]) corresponding to the hex-2-enoyl-S-ACP (Lys57-Arg71) and MS2 data of the ppan
ejection assay from a reference without HcmO. (C) SRM (1054.5 f [1732.30-1737.30]) of the HcmO reaction product 2,3-epoxyhexanoyl-S-ACP (Lys57-
Arg71) and MS2 data of the ppan ejection assay.
spectrometry and SDS-PAGE (Figure S1). In contrast to HxcO,
HcmO is bright yellow in color, and shows absorption maxima
at 377 and 450 nm, typical of a flavin cofactor (Figure S2).
HPLC analysis of HcmO supernatant after heat denaturation
identified FAD as the flavin cofactor.
The substrate specificity of HcmO was assessed with ad-
ditional enoyl-ACP substrates (Figure S3). Whereas variations
of the position and the configuration of the double bond were
not tolerated, a crotonyl moiety with a chain length shorter than
that of the physiological C6 substrate was accepted. HcmO was
able to epoxidize the hexenoyl-substrate utilizing both NADH
and NADPH cofactors. In contrast, the hexenoyl-CoA substrate
and chemoenzymatically generated hex-2-enoyl-CDA could not
be epoxidized by incubation with HcmO.
The putative physiological substrate of HcmO, a hexenoyl-
moiety, was loaded onto the ACP with Sfp and subsequently
incubated with HcmO at a 25-fold molar excess of NAD(P)H
and FAD. Subsequent FTMS of the proteolytic mixture resulting
from tryptic digest was used to characterize HcmO reactivity
(Figure 5A). The ppan ejection assay24 of samples obtained from
negative control reaction volumes without the enzyme resulted,
in addition to the 1734.8 Da signal corresponding to the active
site ACP fragment (Lys57-Arg71) without the ppan arm, in a
fragment ion of 357.2 Da, which is consistent with the calculated
mass of the hex-2-enoyl-ppan MS-MS fragmentation product
(Figure 5B). The HcmO reaction revealed the formation of the
fragment ions of the active site ACP(Lys57-Arg71) without the
ppan arm (1734.8) accompanied by the epoxy acid ppan product
(373.2 Da), both resulting from MS2 fragmentation (Figure 5C).
A hexanoyl-S-ACP substrate could not be converted into the
epoxide product by HcmO that shows only 10% amino acid
sequence similarity with HxcO. Further, the ensuing application
of the described amide ligation strategy enabled the comparison
of the epoxy fatty acid with a synthetic standard (Figure 4C)
and proved the identity of the HcmO reaction product to be
(2S,3R)-2,3-epoxyhexanoic acid.
Discussion
Herein, we report the biochemical characterization of three
enzymes of the CDA biosynthetic gene cluster, HxcO, HcmO,
and ACP (SCO3249), and their mechanistic role in the biosyn-
thesis of the unique 2,3-epoxyhexanoyl moiety of CDA.
The obtained results are surprising in two aspects: First,
predictions based on sequence similarities favored HxcO and
HcmO to be a fatty acid oxidase and epoxidase, respectively,
both acting on acyl-CoA substrates. However, our experiments
with a set of acyl-CoA, acyl-ACP, and chemoenzymatically
derived CDA substrates clearly demonstrate that only fatty acids
loaded onto the recombinant ACP (SCO3249) are accepted by
HxcO and HcmO. In addition to the physiological substrate,
hexanoyl-S-ACP, HxcO accepted a range of ACP-bound
substrate analogues, namely linear fatty acids of 4-10 C-atoms.
HcmO epoxidation activity, instead, showed a more restricted
substrate specificity and was limited to hex-2-enoyl- and
9
J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008 2663

A R T I C L E S
Kopp et al.
Figure 6. Proposed mechanism of epoxidation catalyzed by HxcO and HcmO. During fatty acid oxidation (HxcO) or by the consumption of NAD(P)H
(HcmO) FADH2 is generated. The reaction with molecular oxygen affords the flavin C4a-hydroperoxide intermediate, which acts as a nucleophile in the
attack of the 2,3-hexenoyl-S-ACP substrate originating from either HxcO reactivity or FAS enzymes. Finally, the C4a-hydroxy flavin intermediate dehydrates
to regenerate the oxidized FAD.
crotonyl-S-ACP substrates. Second, whereas HcmO showed the
expected epoxidase activity, the characterization of the putative
dehydrogenase HxcO showed the time dependent formation of
2,3-epoxyhexanoyl-S-ACP as the main product and only minor
amounts of the expected hex-2-enoyl-S-ACP product.
propylamine, or amino acid methyl esters, e.g., D-Phe-OMe (2).
The rationale for this change was the now diastereomeric
relationship of the two possible epoxyhexanoyl amide ligation
products, which was anticipated to lead to a better separation
by HPLC. Synthesis of 2,3-epoxyhexanoyl amides using racemic
2,3-epoxyhexanoic acid (6/7, Scheme 1) and the amines
mentioned above and analysis of the diastereomeric product
mixtures by HPLC using a chiral column showed that, only in
the case of the N-acylated phenylalanine esters 8 and 9 (Scheme
1), a sufficient peak separation was obtained. Gratifyingly,
phenylalanine derivative 2 could also successfully be used as a
nucleophile in the amide ligation reaction, where it efficiently
cleaved the ACP-bound thioesters. Using this strategy, we could
clearly establish that the epoxide generated by HxcO has the
(2R,3S)-configuration (Figure 4B). Interestingly, the 2,3-epoxy-
hexanoyl moiety generated by the second putative tailoring
enzyme studied here, HcmO, was shown to be of opposite
configuration (2S,3R) based on HPLC analysis of the ligation
product. This experimental approach was therefore essential for
this study, since it enabled the detailed analysis of the CP-bound
HxcO and HcmO enzyme products as well as the unambiguous
assignment of product stereochemistry by comparison to
synthetic standards.
To further characterize the identity of HxcO and HcmO
enzymatic products by comparison to synthetic standards, we
were confronted with the problem that standard methodology
for the cleavage and analysis of CP-bound substances could
not be applied in this case. Enzymatic cleavage by TEII is only
possible for peptidyl carrier protein (PCP)-bound substrates,13
and the use of ACP hydrolases is restricted by their high
specificity for native ACPs.14,15 Relatively harsh alkaline
treatment, on the other hand, could hydrolyze the epoxy function
and lead to racemization. To establish alternative conditions that
would allow the direct transformation of the ACP-bound en-
zyme products into derivatives of smaller size appropriate for
standard HPLC-MS analysis, we envisioned cleaving the reac-
tive acyl-S-ACP thioesters by incubation with amine nucleo-
philes, thereby taking advantage of the chemical activation
already inherent to thioester-bound substrates. Initial experiments
with hydrophobic amines, e.g., benzylamine, showed indeed the
formation of the desired amide ligation products suitable for
HPLC analysis and comparison with synthetic standards.
However, we were unsuccessful in determining the absolute
stereochemistry of the epoxides generated by HxcO and HcmO,
respectively, because enantiomeric synthetic standards were not
well resolved by HPLC using a chiral column.
Taken together, it turned out that HxcO is a new type of FAD-
dependent acyl-ACP oxidase with intrinsic epoxidase activity,
responsible for the desaturation and subsequent epoxidation of
the initially formed enoyl-ACP product. Regarding the oxidation
mechanism, it is likely that, similar to the case of related acyl-
CoA oxidases, flavin C4a-hydroperoxide (FADH-OOH) is
formed during the reoxidation of FADH2 by molecular oxygen
Accordingly, we turned our attention to chiral ligation
partners, such as D-R-methyl benzylamine, (S)-2-phenyl-1-
9
2664 J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008

Lipopeptide Fatty Acid Tailoring Enzymes
A R T I C L E S
Figure 7. Proposed model for the ACP-mediated biosynthetic pathway of the unique trans-2,3-epoxyhexanoyl moiety found in CDA. The two epoxidases
HxcO and HcmO are responsible for the generation of the two shown 2,3-epoxyhexanoyl enantiomers. Either FAS enzymes from primary metabolism
together with FabH3 and FabH4, or HxcO are likely to produce the hex-2-enoyl substrate for HcmO. The absolute configuration of the epoxyhexanoyl side
chain in the CDA antibiotics is unknown up to now. CDA PSI ) CDA Peptide Synthetase I.
(Figure 6). Instead of decomposing to FAD and H2O2, FADH-
OOH could be deprotonated and serve as a nucleophile in the
subsequent epoxidation of the electron deficient double bond.
Therefore, the enoyl substrate can undergo conjugated addition
with the reactive FAD-OO- species following a mechanism that
has been proposed for the Baeyer-Villiger type oxidation of
cyclic ketones by FAD-dependent epoxidases (Figure 6).27 The
biochemical characterization of HcmO as a flavin-dependent
epoxidase revealed the formation of the 2,3-epoxyhexanoyl
residue while acting on the ACP-bound hexenoyl-substrate. In
this case, FADH2 is generated by the oxidation of NAD(P)H,
and as described for HxcO, epoxidation is also likely to occur
with FAD-OO- as the nucleophilic species (Figure 6).
lipopeptides, e.g., enduracidin,28 ramoplanin,29 and friulimi-
cin,30,31 also contain stand-alone fatty acid tailoring enzymes.
Most particularly, gene knock-outs in the friulimicin NRPS
system confirmed the role of an acyl-CoA-dehydrogenase to
be involved in the formation of the cis double bond of the
N-terminal lipid moiety.30 In analogy to the examined CDA fatty
acid epoxidation, ACP-mediated tailoring pathways seem now
to be likely for the above-mentioned examples. This hypothesis
is additionally supported by the fact that in close proximity to
the genes of the tailoring enzymes ACP encoding segments were
found for each biosynthetic system.
Conclusions
The existence of two epoxidases within the CDA gene cluster
that display different stereoselectivity gives rise to speculations
on the molecular logic underlying the CDA epoxyhexanoic acid
biosynthesis. A total of five genes comprise the putative fab
operon that is involved in fatty acid biosynthesis and tailoring
of the CDA epoxyhexanoyl moiety. The hexanyol-ACP substrate
for HxcO is likely to be produced by the interplay of FabH4
and FabH3, putative â-ketoacyl-ACP synthases encoded within
the fab operon, together with enzymes from primary metabolism.
Subsequently, fatty acid tailoring by HxcO occurs on the ACP-
bound hexanoyl moiety and results in the (2R,3S)-2,3-epoxy-
hexanoyl product and minor amounts of the hex-2-enoyl-ACP,
as demonstrated in this study. This “side product” of HxcO could
then serve as the substrate for subsequent epoxidation by HcmO.
Alternatively, FAS enzymes from primary metabolism could
directly produce the hex-2-enoyl-ACP substrate for HcmO
during fatty acid synthesis on the ACP. Based on the finding
that two epoxidases exist in the CDA cluster, one might
speculate that HxcO has gained its epoxidation activity during
evolution and produces now together with HcmO the two
possible 2,3-epoxyhexanoyl enantiomers. However, it remains
to be seen whether both enantiomers are then transferred from
the ACP to the first module of the CDA peptide synthetases
(Figure 7). NMR studies allied with a total synthesis approach
are currently underway to investigate the stereochemistry of the
2,3-epoxyhexanoyl moiety in the natural product.
Here, we have shown that the described amide ligation
strategy is a robust system for the rapid and sensitive detection
of CP-bound substances. Using this powerful approach new
tailoring enzymes of CP-mediated pathways could be stereo-
chemically characterized, which was not possible via established
methodology. Within the CDA biosynthetic system, HxcO was
shown to be a novel type of enzyme with a dual function as an
FAD-dependent fatty acid-S-ACP oxidase paired with intrinsic
epoxidase activity. A second epoxidase HcmO was shown to
be responsible for the production of another 2,3-epoxyhexanoic
enantiomer during CDA biosynthesis. It remains to be shown
whether both enantiomers are then incorporated into the natural
product. In addition to this, fatty acid tailoring during CDA
biosynthesis was determined to occur on ACP-bound substrates.
Taken together, the obtained biochemical data provide new
insights into fatty acid tailoring of nonribosomal lipopeptides
and may facilitate future projects aiming to manipulate the lipid
portion of these compounds. Moreover, the herein presented
approach holds great potential for the detailed analysis of other
biochemical processes involving CP-bound intermediates, a
central paradigm in secondary metabolism.
Acknowledgment. We would like to thank Antje Scha¨fer for
her excellent technical assistance. Financial support was pro-
(28) Yin, X.; Zabriskie, T. M. Microbiol. 2006, 152, 2969-2983.
(29) McCafferty, D. G.; Cudic, P.; Frankel, B. A.; Barkallah, S.; Kruger, R. G.;
Li, W. Biopolymers 2002, 66, 261-284.
Overall, the data presented here support the view that fatty
acid tailoring during NRPS synthesis occurs on ACP-bound
substrates. The biosynthetic gene clusters of other nonribosomal
(30) Heinzelmann, E.; Berger, S.; Muller, C.; Hartner, T.; Poralla, K.; Wohlleben,
W.; Schwartz, D. Microbiol. 2005, 151, 1963-1974.
(31) Muller, C.; Nolden, S.; Gebhardt, P.; Heinzelmann, E.; Lange, C.; Puk,
O.; Welzel, K.; Wohlleben, W.; Schwartz, D. Antimicrob. Agents Chemoth-
er. 2007, 51, 1028-1037.
(27) Sheng, D.; Ballou, D. P.; Massey, V. Biochemistry 2001, 40, 11156-11167.
9
J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008 2665

A R T I C L E S
Kopp et al.
vided by DFG (M.A.M.) and the Fonds der Chemischen
Industrie (M.A.M. and F.K.).
analysis, synthesized acyl-CoA substrates, experimental details
and kinetic data. This information is available free of charge
via the Interment at http://pubs.acs.org.
Supporting Information Available: Oligonucleotide primers,
SDS-PAGE of studied enzymes, HxcO and HcmO cofactor
JA078081N
9
2666 J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008