Characterization of the initial enzymatic steps of barbamide biosynthesis

Article abstract of DOI:10.1021/np050523q

Barbamide is a mixed polypeptide-polyketide natural product that contains an unusual trichloromethyl group. The origin of the trichloromethyl group was previously shown to be through chlorination of the pro-R methyl group of L-leucine. Trichloroleucine is subsequently decarboxylated and oxidized to trichloroisovaleric acid and then extended with an acetate unit to form the initial seven carbons of barbamide. In this study we used a combination of biosynthetic feeding experiments and enzymatic analysis to characterize the initial steps required for formation of trichloroleucine and its chain-shortened product, trichloroisovaleric acid. Results from isotope-labeled feeding experiments showed that both dichloroleucine and trichloroleucine are readily incorporated into barbamide; however, monochloroleucine is not. This suggests that halogenation of the pro-R methyl group of leucine occurs as two discrete reactions, with the first involving incorporation of at least two halogen atoms and the second converting dichloroleucine to trichloroleucine. Additionally, the initial tandem dichlorination must occur before substrate can be further processed by the remaining bar pathway enzymes. In vitro analysis of the first five open reading frames (ORFs; barA, barB1, barB2, bar C, barD) of the barbamide gene cluster has yielded new insights into the processing of leucine to form the trichloroisovaleryl-derived unit in the final product.

Full text of DOI:10.1021/np050523q

938
J. Nat. Prod. 2006, 69, 938-944
Characterization of the Initial Enzymatic Steps of Barbamide Biosynthesis
Patricia M. Flatt,† Susan J. O’Connell,‡ Kerry L. McPhail,† Gloria Zeller,† Christine L. Willis,‡ David H. Sherman,§ and
William H. Gerwick*,†,
College of Pharmacy, Oregon State UniVersity, CorVallis, Oregon 97331, School of Chemistry, UniVersity of Bristol, Bristol, BS8 ITS U.K.,
Life Sciences Institute, Department of Medicinal Chemistry, Department of Chemistry, and Department of Microbiology and Immunology,
UniVersity of Michigan, Ann Arbor, Michigan 48109, and Scripps Institution of Oceanography and Skaggs School of Pharmacy and
Pharmaceutical Sciences, UniVersity of California San Diego, La Jolla, California 92093
ReceiVed December 12, 2005
Barbamide is a mixed polypeptide-polyketide natural product that contains an unusual trichloromethyl group. The
origin of the trichloromethyl group was previously shown to be through chlorination of the pro-R methyl group of
L-leucine. Trichloroleucine is subsequently decarboxylated and oxidized to trichloroisovaleric acid and then extended
with an acetate unit to form the initial seven carbons of barbamide. In this study we used a combination of biosynthetic
feeding experiments and enzymatic analysis to characterize the initial steps required for formation of trichloroleucine
and its chain-shortened product, trichloroisovaleric acid. Results from isotope-labeled feeding experiments showed that
both dichloroleucine and trichloroleucine are readily incorporated into barbamide; however, monochloroleucine is not.
This suggests that halogenation of the pro-R methyl group of leucine occurs as two discrete reactions, with the first
involving incorporation of at least two halogen atoms and the second converting dichloroleucine to trichloroleucine.
Additionally, the initial tandem dichlorination must occur before substrate can be further processed by the remaining
bar pathway enzymes. In vitro analysis of the first five open reading frames (ORFs; barA, barB1, barB2, bar C, barD)
of the barbamide gene cluster has yielded new insights into the processing of leucine to form the trichloroisovaleryl-
derived unit in the final product.
Over the past 30 years, cyanobacteria have emerged as extraor-
dinarily prolific producers of biologically active secondary me-
tabolites.1,2 Several compounds deriving from cyanobacteria are
currently in clinical trials for the treatment of human cancer,
including analogues of dolastatin 10, dolastatin 15, and cryptophycin
A.3-6 A number of other metabolites have been identified as
promising preclinical lead compounds for cancer treatment, includ-
ing curacin A and scytophycin A.1,6 Numerous cyanobacterial
metabolites have been reported to possess other biological effects
as well, including antifungal, antibacterial, and anti-HIV proper-
ties.1,2
peptidyl carrier (PCP) and adenylation (A) domain of one NRPS
module being encoded on separate ORFs, (2) a PKS module that
is encoded on two separate ORFs, (3) an oxidative decarboxylation
of a cysteine-derived residue at the end of the assembly to form a
terminal thiazole ring, and (4) the chlorination, R-oxidation, and
decarboxylation of leucine to form a trichloroisovaleric acid-derived
moiety in the final product. This latter chlorinated residue is
subsequently ketide extended, and the enolized form of the â-keto
acid is O-methylated before extension with additional amino acid
residues.
The trichloroisovalerate-derived moiety found in barbamide (1)
is very rare in natural products chemistry and provides a unique
opportunity to dissect and understand the biochemical mechanism-
(s) responsible for its formation. Stable isotope feeding experiments
using [2-13C]leucine (3) or [2-13C]5,5,5-trichloroleucine (4) suggest
that L-leucine or an R-keto derivative of L-leucine is the actual
substrate for chlorination (Figure 1).17 Although none of the ORFs
in the bar cluster were initially identified as sharing homology with
known halogenases, it became evident through BLAST analysis
that barB1 and barB2 encoded for proteins showing homology to
oxidoreductases that share high sequence homology with the
syringomycin protein, SyrB2, and the coronamic acid protein,
CmaB.16,18 Enzymes within this family have an obligate requirement
for iron(II) and molecular oxygen, and all but two also require
2-oxoglutarate as a cofactor.19,20 Alignment with the phytanoyl-
CoA hydrolase (PAHX) protein, SyrB2, and CmaB shows that
BarB1 and BarB2 retain the conserved HXD motif required for
coordination with the iron(II).21 Recently, the SyrB2 and CmaB
proteins were demonstrated to be a new class of 2-oxoglutarate,
iron-dependent halogenases.22,23 In the case of syringomycin
biosynthesis, SyrB2 was shown to carry out the monochlorination
of the methyl group of L-Thr bound as a thioester to the SyrB1
protein.22 In the case of coronamic acid, the CmaB protein has been
shown to chlorinate the gamma-position of L-alloisoleucine.23 This
chlorinated intermediate is then suitably activated for cyclopropyl
ring formation with consequent loss of chlorine (i.e., a cryptic
chlorination). Each of these enzymes requires molecular oxygen,
iron, 2-oxoglutarate, and chloride ion for activity.22,23 In the current
The metabolic and biomedical potential of marine cyanobacteria
is well exemplified by a Curac¸ao collection of Lyngbya majuscula,
which yielded nine biologically active compounds that are predicted
to be the products of at least five distinct biosynthetic gene
clusters.7-12 In addition to having an exceptional capacity to
generate diverse natural product frameworks by a rich utilization
of interdigitated non-ribosomal peptide synthetase (NRPS) and
polyketide synthase (PKS) pathways, marine cyanobacterial com-
pounds show evidence of many unique tailoring modifications,
including multiple N-, O-, and C-methylations, oxidations, hetero-
cyclic ring formation, and an exceptional ability to create diverse
halogen-containing functionalities. Included in this latter group are
vinylic and acetylenic bromides, as found in phormidolide13 and
jamaicamide A,14 and vinyl chlorides and trichloromethyl groups,
as present in malyngamide I15 and barbamide (1).11
Previously, we have reported on the isolation and characterization
of the gene cluster encoding barbamide (1) biosynthesis (bar).16
The genetic architecture and catalytic domain organization of the
bar cluster is generally collinear and contains 12 open reading
frames (ORFs). However, barbamide is produced by a mixed PKS/
NRPS system with several unusual features, which include (1) a
* To whom correspondence should be addressed. Tel: (858) 534-0578.
Fax: (858) 534-0529. E-mail: wgerwick@ucsd.edu.
† Oregon State University.
‡ University of Bristol.
§ University of Michigan.
University of California San Diego.
10.1021/np050523q CCC: $33.50
© 2006 American Chemical Society and American Society of Pharmacognosy
Published on Web 06/09/2006

Initial Enzymatic Steps of Barbamide Biosynthesis
Journal of Natural Products, 2006, Vol. 69, No. 6 939
Figure 1. Incorporation of biosynthetic precursors into barbamide. (A) Summary of leucine-derived precursors that are incorporated into
barbamide and structures of dechlorobarbamide (2) and didechlorobarbaleucamide B (7). Comparison of the 13C NMR spectra for (B)
natural abundance barbamide and (C) barbamide produced during supplementation with [6-13C](2S,4S)-5,5-dichloroleucine.
study, we have employed a combination of biosynthetic feeding
experiments and recombinant technologies to characterize the initial
steps of barbamide biosynthesis.
molecular ion cluster for 2, as well as for the other chlorinated
metabolites from these experiments, prevented calculation of 13C
incorporation rates from ESIMS data alone because 13C-enriched
standards were not available, and therefore the TIC averages
measured for each metabolite showed considerable variability in
isotopomeric composition. Incorporation of dichloroleucine (6) but
not monochloroleucine (5) into barbamide (1) suggests either that
the halogenation of leucine occurs initially as a tandem reaction
that incorporates two chlorine atoms or that only dichlorinated or
trichlorinated intermediates are recognized and processed into the
remaining steps of the barbamide/dechlorobarbamide biosynthetic
pathway. This is fully consistent with the recently published work
by Walsh et al., who have worked in vitro with the barbamide
halogenases BarB1 and BarB2 as prepared from synthetic genes,25
and demonstrates that these halogenation reactions occur in vivo
in the tandem stepwise manner predicted from the in vitro studies.
Results and Discussion
Biosynthetic Feeding Experiments. Previous isotope-labeled
precursor feeding experiments have demonstrated that both leucine
and trichloroleucine can be incorporated into the barbamide (1)
biosynthetic pathway;16 however, monochloroleucine was not
incorporated.24 These results suggest that multiple halogenations
of leucine or a leucine-derived intermediate occur as an initial event
and that this is a prerequisite for further processing of the leucine-
derived substrate. Since a dichlorinated form of barbamide, dechlo-
robarbamide (2), has also been isolated from the producing
organism,17 we hypothesized that dichloroleucine would be incor-
porated into the pathway and at least produce compound 2.
However, it was uncertain whether dichloroleucine could be a
substrate for further chlorination and thus lead to barbamide (1),
or if dichloroleucine is only incorporated into dechlorobarbamide
(2).
Laboratory cultures of L. majuscula were incubated with [6-13C]-
(2S,4S)-dichloroleucine (6) for 7 days and harvested, and lipid
natural products were extracted by standard methods. Barbamide
and related metabolites were subsequently isolated by successive
chromatography via solid-phase extraction and two rounds of
HPLC. Isolated barbamide (0.5 mg) was analyzed by ESIMS and
high-field 1H and 13C NMR and showed a strongly enriched (604%
over natural abundance) 13C NMR signal for the C-1 methyl group
at δ 15.2 (Figure 1; Supporting Information). This experiment
clearly revealed that the dichloromethyl group of dichloroleucine
is indeed a recognized and acceptable substrate for additional
chlorination. However, dechlorobarbamide, while observed by
DAD-HPLC and LC-MS, was produced in too small quantity in
this experiment for NMR analysis (see Supporting Information data
S3). Nevertheless, these UV and MS data, taken in conjunction
with retention times for authentic dechlorobarbamide, strongly
support the identity of this material. However, the complicated
Intriguingly, a new compound (7) was produced in small yield
(0.1 mg) from the [6-13C](2S,4S)-dichloroleucine feeding experiment
and purified by recycling HPLC. By LC-MS analysis, this new
material was consistent with a didechloro analogue of barbaleuca-
mide B, a compound previously obtained from a Philippine
collection of the sponge Dysidea sp.26 [We have recently shown
that the metabolic capacity to produce halogenated natural products
in Dysidea resides within its symbiotic cyanobacterium Oscillatoria
spongeliae.21] The 1H NMR spectrum for 7 clearly resembled that
of barbamide except for the conspicuous absence of aromatic signals
for the monosubstituted phenyl group. Additionally, it possessed
new resonances for an additional methyl doublet (δ 1.40) as well
as several methylene and methine signals (δ 2.10-3.10, Supporting
Information data S5). Finally, HRFABMS data were consistent with
a formula of C17H2435Cl4N2O2S for 7 (Supporting Information data
S4). Most significantly, the 13C NMR spectrum for this minor
metabolite exhibited two 13C-enriched methyl signals at the
appropriate chemical shifts for the two leucine-derived fragments.
However, no natural abundance signals were detected in the 13C
NMR spectrum of 7, consistent with a very high level of
13C-enrichment from exogenous [6-13C](2S,4S)-dichloroleucine (see
Supporting Information data S5). Hence, it appears that the

940 Journal of Natural Products, 2006, Vol. 69, No. 6
Flatt et al.
Figure 2. His-fusion proteins of BarA, BarB1, BarB2, BarC, and
BarD and the SFP protein were purified using Ni+ agarose affinity
chromatography and visualized using a 15% SDS-PAGE stained
with Coomassie. Bovine serum albumin (BSA) was used as a
standard (0.5, 1, 2, 4, 6, 8, and 10 µg).
adenylation domain of BarG, which normally specifies for L-
phenylalanine,11 has considerable substrate tolerance and can accept
halogenated leucine derivatives when they are available in abun-
dance.
Recombinant Protein Expression. To reconstitute the initial
steps involved in the biosynthesis of the barbamide starter unit,
we recombinantly expressed the first five open reading frames of
the pathway (barA, barB1, barB2, barC, barD). We predicted that
the BarD protein was required to activate leucine and load it onto
the small PCP protein, BarA. BarB1 and BarB2 were envisioned
to catalyze the multiple chlorinations of the BarA-leucyl intermedi-
ate. Release and recovery of the halogenated product was then
proposed using the type II thioesterase protein, BarC. Thus, the
five genes encoding BarA-BarD were subcloned in frame with a
6XHis fusion tag into the pET20b expression vector and expressed
in the BL21∆(DE3) protein expression strain of E. coli. All of the
expressed proteins were soluble and could be readily purified using
nickel agarose chromatography (Figure 2).
Figure 3. To demonstrate the formation of a thioester bond
between BarA and leucine, BarA, BarD, and the SFP protein were
incubated with 0.5 mM CoASH and 1 mM TCEP for 45 min at 37
°C to load phosphopantethiene onto BarA. Thioester formation
activity was initiated by the addition of 5 mM ATP in the presence
of 1 µCi [UL-14C]leucine. The reaction was incubated at 37 °C for
30 min and terminated by the addition of 10% TCA and 400 mg
of BSA. Precipitated proteins were pelleted by centrifugation and
washed three times with 10% TCA. Incorporation of radiolabled
leucine was determined by (A) SDS-PAGE-Coomassie staining with
subsequent autoradiography and (B) liquid scintillation counting.
Thioester Formation Assay. Previous experiments using the
ATP/PPi conversion assay have demonstrated that BarD is capable
of activating leucine to its aminoacyl adenylate.16 To determine if
BarD is also competent to transfer the activated form of leucine to
BarA and thus form a thioester linkage, we incubated the BarA
protein at 5 µM with Bacillus subtilis phosphopantetheinyl trans-
ferase (SFP), the BarD protein, and [UL-14C]leucine. Incorporation
of [UL-14C]leucine into BarA was visualized using SDS-PAGE
followed by autoradiography. Results demonstrated that leucine was
incorporated into the BarA protein and that this reaction was
dependent on the presence of both BarD and ATP (Figure 3). In
addition, this set of experiments demonstrated that the thioesterase,
BarC, was able to release the bound 14C-labeled intermediate (Figure
3).
Figure 4. To demonstrate that BarB1/B2 mediate conversion of
2-oxoglutarate to succinate, the release of [14C]CO2 from 2-oxo-
glutarate was measured with increasing BarB1/B2 protein concen-
trations (0, 5, 10, 15, and 20 µg) in the presence or absence of the
substrate, leucyl-BarA.
2-Oxoglutarate Conversion Assay. The activity of iron-de-
pendent, 2-oxoglutarate-dependent enzymes can be monitored by
the conversion of 2-oxoglutarate to succinate. Thus, to determine
if BarB1 and BarB2 can utilize 2-oxoglutarate as a cofactor, we
incubated BarB1 and BarB2 with [1-14C]2-oxoglutarate and mea-
sured the release of radiolabeled CO2. The rate of 2-oxoglutarate
turnover increased with increasing concentrations of the proposed
substrate, leucyl-BarA (Figure 4). However, it has previously been
demonstrated that many iron-dependent, 2-oxoglutarate oxidoreduc-
tases can mediate the modest, but significant, conversion of the
cofactor 2-oxoglutarate to succinate even in the absence of
substrate.27,28 In agreement with these observations, an increased
release of CO2 from [1-14C]2-oxoglutarate was also seen in the
absence of leucyl-BarA, albeit at a reduced rate over the former
experiments (Figure 4). These results indicate that BarB1 and BarB2
are indeed 2-oxoglutarate, iron-dependent enzymes that are most
likely involved in catalyzing chlorination of leucine, as shown in
Figure 5.29
min at 30 °C and then subjected to treatment with 5 µM of the
BarC thioesterase. Amino acids in the crude lysate were separated
from the protein slurry using a Microcon centrifugal device with a
MW cutoff of 3000, derivatized using dansyl chloride and analyzed
by ESIMS. However, despite three attempts, including one with
isolation of the BarB1 and BarB2 proteins in the absence of oxygen
(it has been shown that the purification of 2-oxoglutarate, iron-
dependent enzymes in the presence of oxygen can limit or
completely inhibit the activity of the recombinant protein22,23,30),
we were unable to detect a halogenated product. Because the assay
is a single-turnover reaction that requires the activity of six
reconstituted recombinant proteins (BarA, BarB1, BarB2, BarC,
BarD, and the SFP protein), it has inherent low detectability.
Furthermore, turnover is also dependent on the efficiency of leucine
incorporation into BarA. While some refinement of the reagents
Additional incubations of BarA appropriately charged with
leucine were reacted with the BarB1 and BarB2 constructs for 60

Initial Enzymatic Steps of Barbamide Biosynthesis
Journal of Natural Products, 2006, Vol. 69, No. 6 941
Figure 5. Barbamide halogenation is believed to occur via a modified 2-oxoglutarate, Fe(II)-dependent dioxygenase mechanism.29 The
catalytic scheme is based on the crystal structure and mechanistic data from the BarB1/BarB2 homologue human phytanoyl-CoA hydroxylase
(PAHX).37 We propose that coordination of chlorine with iron will displace either a water molecule or one of the enzyme coordination
points (e.g., histidine).
involved in forming leucyl-BarA was accomplished, this is the only
step in the overall transformation for which it was possible to
perform any optimization. Further experiments are planned with a
genetically modified strain of E. coli capable of overexpressing
the first five ORFs of the bar pathway in addition to the B. subtilis
SFP protein. Alternatively, it may be that our inability to observe
a halogenated product arises because leucyl-BarA is not the optimal
substrate for the halogenase reaction (however, Walsh et al. were
recently able to demonstrate chlorination of this substrate using
BarB2 isolated in the absence of O2).25
Adenylation Domain Activity. During the biosynthesis of
barbamide, it is clear that several tailoring modifications to the
leucine starter unit must occur, including halogenation, oxidation,
and decarboxylation, to convert leucine to the trichloroisovaleryl-
derived moiety found in barbamide. Our previous adenylation
domain experiments suggested that the first step in leucine
modification is the conversion of leucine to the trichlorinated
intermediate. BarD was shown to activate both leucine and
trichloroleucine, whereas BarE demonstrated moderate activity only
toward trichloroleucine. These initial results suggested that a leucyl-
BarA intermediate was the likely substrate for the halogenase
Figure 6. ATP/PPi exchange activity of (A) BarD and (B) the
enzymes, BarB1 and BarB2. However, the scope of these initial
BarE adenylation domain were evaluated using the following
experiments did not rule out the possibility that the oxidation and/
substrates: no amino acid (No a.a.), glycine (Gly), valine (Val),
or decarboxylation of leucine may precede halogenation of the
leucine-derived substrate.
leucine (Leu), R-keto derivative of leucine (R-keto), isovaleric acid
(IVA), trichloroisovaleric acid (TClIVA), trichloroleucine (TClLeu),
To address this issue, additional experiments were conducted
using isovaleric acid, trichloroisovaleric acid, and the R-keto
derivatives of leucine and trichloroleucine as substrates for the BarD
and BarE adenylation domains. The results from these experiments
supported our previous findings that BarD can activate both leucine
and trichloroleucine with high affinity (Figure 6A). However, BarE
demonstrated only modest activity toward trichloroleucine in
comparison with the R-keto derivatives of leucine and/or trichlo-
roleucine (Figure 6B). Neither of the enzymes demonstrated
significant activity toward isovaleric acid or the trichloroisovaleric
acid substrates (Figure 6). Together, these data suggest that leucine
is converted to the R-keto derivative prior to its transfer to BarE
and that decarboxylation occurs after incorporation of the R-keto
intermediate into BarE. A candidate enzyme from the barbamide
and the R-keto derivative of trichloroleucine (R-keto TCl). Results
are representative of three independent experiments.
gene cluster predicted to convert leucine to the R-keto derivative
is BarJ, as it shares high homology with L-amino acid oxidases.
However, several attempts at recombinant expression of BarJ
yielded an unstable protein product that is cleaved into two major
fragments. We are currently developing an alternate expression
system using the freshwater cyanobacterium Nostoc punctiforme
to try to overcome these and related heterologous expression
problems. While these experiments lend insight into the timing of
leucine oxidation, the precise timing of halogenation remains
unclear.

942 Journal of Natural Products, 2006, Vol. 69, No. 6
Flatt et al.
CmaB proteins function as halogenase enzymes involved in
syringomycin and coronamic acid biosynthesis, respectively, lends
strong support that BarB1 and BarB2 are indeed representative of
a new class of halogenase proteins belonging to the 2-oxoglutarate
superfamily.22,23 Further mechanistic studies with these novel
tailoring enzymes will reveal the timing and optimal substrate
intermediates for the halogenation reactions, offering a unprec-
edented opportunity to explore the utility of this new class of
halogenases in biotechnological applications.
Experimental Section
General Experimental Procedures. NMR data were obtained on
a Bruker DRX300 in DMSO-d6 using the solvent as an internal
reference (δC 39.51). 13C NMR (75 MHz) experiments on both natural
abundance and labeled barbamide samples were performed using an
inverse-gated pulse program (zgig) with a preparation delay of 500
ms. Low-resolution MS data were acquired on a ThermoFinnigan LCQ
Advantage Max spectrometer. High-resolution MS data were acquired
on a VG micromass 3DS/2RS MS9 double-focusing mass spectrometer.
HPLC analyses were carried out using a Shimadzu SPD-10A dual-
wavelength UV-vis detector with LC610 pumps or a Waters 996
photodiode array detector with Waters 515 pumps.
Synthesis of [6-13C](2S,4S)-5-Monochloroleucine (5), [6-13C]-
(2S,4S)-5,5-Dichloroleucine (6), and 4-Trichloromethyl-2-oxopen-
tanoic Acid (18). [6-13C](2S,4S)-5-Chloroleucine (5) was prepared from
L-glutamic acid.24 The key step was a stereoselective alkylation31 of
the protected glutamate (8) using 13CH3I to give 9, which was
transformed to the target compound 5 in a further five steps and 40%
yield.A similar strategy was used to introduce the isotopic label into
[6-13C](2S,4S)-5,5-dichloroleucine (6). Rodriguez and co-workers32
have used a modification of Takeda’s procedure33 involving hydrazine
monohydrate in anhydrous MeOH to form the hydrazone followed by
CuCl2 and Et3N for the conversion of unlabeled aldehyde 11 to the
geminal dichloride 12 in 40% yield. These conditions were used to
transform [13C]aldehyde 13 to 14, which, after deprotection in refluxing
6 N HCl, gave the required [6-13C](2S,4S)-dichloroleucine (6) as a white
solid. 4-Trichloromethyl-2-oxopentanoic acid (18) was prepared from
the known34 3-trichloromethylbutanoic acid (15). The one carbon
homologation of acid 15 was achieved via ozonolysis of the â-keto-
cyanophosphorane 16 using conditions analogous to those reported for
the synthesis of [5-13C]L-leucine.35 Hydrolysis of methyl ester 17 under
mild conditions using 1 equiv of aqueous 2 N NaOH gave the novel
R-keto acid 18 as a yellow oil.
Figure 7. Three possible mechanisms for the modification of
leucine during barbamide biosynthesis: (A) halogenation of leucyl-
BarA precedes R-oxidation, (B) R-oxidation of leucyl-BarA pre-
cedes halogenation, or (C) the leucyl-BarA intermediate is R-ox-
idized prior to transfer onto BarE, where it undergoes halogenation.
We propose three possible routes to the formation of a trichloro
intermediate, as depicted in Figure 7. In the first two hypotheses
(Figure 7A and 7B), both the halogenation and oxidation of leucine
are proposed to occur while leucine is bound as a thioester
intermediate to BarA. In pathway A, the halogenation of leucine is
proposed to precede oxidation, whereas in pathway B oxidation
occurs prior to halogenation. In pathway C, an alternate hypothesis
is proposed where leucyl-BarA is oxidized to the R-keto acid
derivative prior to being incorporated into BarE. The BarE
intermediate is then proposed to be the substrate for the halogenation
reaction (Figure 7C).
[6-13C](2S,4S)-5,5-Dichloroleucine (6): mp 208-210 °C; [R]D -24
1
(c 1, D2O), (lit.32 [R]D unlabeled material -23 (c 0.16, 1 N HCl); H
NMR (400 MHz, D2O) 1.2 (3H, dd, J 128, 6, 13CH3), 1.9 (1H, ddd, J
15, 9, 5, 3-HH), 2.13 (1H, ddd, J 15, 10, 5, 3-HH), 2.35 (1H, m, 4-H),
3.8 (1H, dd, J 10, 5, 2-H), 6.11 (1H, t, J 3, 5-H); 13C NMR (100 MHz,
D2O) 14.4 (13CH3 enriched), 33.7 (C-3), 40.0 (C-4, d, J 19.2), 52.8
(C-2), 78.3 (C-5), 174.5 (C-1); m/z (CI) 201 (MH+, 23%), 155 (65),
74 (100); 13CC5H12O2N35Cl2 requires 201.0277, found 201.0278.
4-Trichloromethyl-2-oxopentanoic acid (18): 1H NMR (300 MHz,
CDCl3) 1.36 (3H, d, J 6.5, 5-H3), 3.12 (1H, dd, J 19, 9, 3-HH), 3.26
(1H, m, 4-H), 3.60 (1H, dd, J 19, 2.5, 3-HH), 7.57 (1H, br s, CO2H);
13C NMR (75 MHz, CDCl3) 17.0 (C-5), 42.3 (C-3), 50.9 (C-4), 104.2
(CCl3), 160.5 (C-1), 192.6 (C-2); m/z (CI) 186.9500 (MH+ - CO2H,
C5H6OCl3 requires 186.9484), 233/235/237/239 (MH+, 8/10/5/3), 187/
189/191/193 (34/34/22/6), 151/153/155 (60/46/15), 123 (100).
Conclusions. Barbamide biosynthesis is mediated by a mixed
NPRS-PKS biosynthetic pathway that incorporates a unique
trichloroisovaleryl starter unit. To elucidate the reactions that lead
to the formation of this unique starter unit, we used a combination
of biosynthetic feeding experiments and enzymatic analysis. Bio-
synthetic feeding experiments demonstrate the incorporation of
dichloroleucine as well as trichloroleucine into the barbamide
structure; monochloroleucine is not incorporated.24 These results
suggest that the barbamide halogenases act in tandem to incorporate
at least two halogen atoms in the first reaction and a third in a
second reaction, or that only di- or trichlorinated intermediates can
be recognized by enzymes downstream in the barbamide biosyn-
thetic pathway. Similar conclusions were recently reached by the
Walsh group working with the cloned and overexpressed BarB1
and BarB2 proteins.25 Because our studies reported herein were
conducted with the native barbamide-producing strain of L.
majuscula, it confirms that the deduced sequence of tandem
halogenations observed in these latter in vitro studies is relevant to
natural metabolic transformations within the cyanobacterium.
Biosynthetic Feeding Experiments. Three 1 L flasks were seeded
with 4 g of L. majuscula strain 3L in 500 mL of BG-11 medium.36
The cultures were established and allowed to grow for 5 days prior to
the start of feedings. Each flask was fed a total of 20 mg of [6-13C]-
(2S,4S)-5,5-dichloroleucine (6) in three 6.6 mg doses on days 1, 4,
and 8. The cultures were harvested on day 10 by removing the filaments
from the culture medium, pooling, blotting briefly on Whatman filter
paper, and storing at -20 °C until the samples were extracted. The
wet weight of the algae prior to freezing was 18.15 g. Primary extraction
was performed by repetitive steeping (4×) in 2:1 CH2Cl2/MeOH. Water
was removed from the CH2Cl2 layer in a separatory funnel, which was
then filtered over Whatman #1 paper, and dried using rotary evaporation
to yield 98 mg of crude extract. Initial fractionation utilized a stepped
gradient of EtOAc in hexanes (5% to 100%) and a Varian Mega Bond
Elute SI column (1.5 × 2 cm; normal phase) to yield seven fractions.
Initial characterization of the putative halogenases, BarB1 and
BarB2, demonstrates that they share high homology with a
superfamily of 2-oxoglutarate-dependent oxygenases. Results from
the current study also demonstrate that BarB1 and BarB2 can utilize
2-oxoglutarate as a cofactor. Recent discovery that the SyrB2 and

Initial Enzymatic Steps of Barbamide Biosynthesis
Journal of Natural Products, 2006, Vol. 69, No. 6 943
Scheme 1. Chemical Synthesis of [6-13C](2S,4S)-5-Chloroleucine 5
Scheme 2. Chemical Synthesis of [6-13C](2S,4S)-5,5-Dichloroleucine 6a
a Reagents: (i) NH2NH2‚H2O, MeOH, Et3N, CuCl2; (ii) 6 N HCl.
Scheme 3. Chemical Synthesis of 4-Trichloromethyl-2-oxopentanoic Acid 18a
a Reagents: (i) Ph3PCHCN, EDCI, DMAP; (ii) O3, MeOH, CH2Cl2; (iii) NaOH.
Normal-phase TLC analysis (1:1 EtOAc/hexanes) revealed UV-active
materials in fractions 4 and 5 (50% and 70% EtOAc, respectively).
These two fractions were pooled and further purified using a Varian
Mega Bond Elute C18 cartridge and a stepped gradient of 60%, 80%,
and 100% MeOH. Normal-phase TLC (1:1 EtOAc/hexanes) indicated
that the 80% and 100% fractions contained a UV-active compound
consistent with barbamide. These fractions were pooled and further
purified using exhaustive reversed-phase HPLC (YMC-ODS-AQ-323,
250 × 10 mm; 3 mL/min, 80% aqueous MeOH; Waters Symmetry,
250 × 4.6 mm, 1 mL/min, 70% aqueous CH3CN) to yield 0.5 mg of
barbamide. In addition, purification of barbamide (1) by a second round
of HPLC led to the concomitant isolation of the new metabolite
didechlorobarbaleucamide B (7, 0.1 mg). Dechlorobarbamide (2) was
detected by LC-MS in the fraction preceding that containing barbamide
from the initial round of HPLC.
cells as follows. One liter of LB culture was inoculated with a 1:100
dilution of an overnight culture of B. subtilis and grown in a 3 L
Fernbach flask at 18 °C overnight. The E. coli were pelleted by
centrifugation and resuspended in 10 mL of lysis buffer (20 mM Tris,
pH 8.0 with 500 mM NaCl and 20 mM imidazole) per liter of culture.
The lysate was sonicated five times at 70 W for 10 s/burst. The cellular
debris was removed by centrifugation and the protein isolated by nickel
agarose chromatography as previously described.16 Protein purity and
concentration were determined by SDS-polyacrylamide gel electro-
phoresis (SDS-PAGE) and Coomassie staining.
Thioester Formation Assay. The assay was carried out in 100 µL
total volume containing 75 mM Tris, pH 8.0, 1 mM TCEP, 10 mM
MgCl2, 5 mM ATP, 0.5 mM CoASH, 550 nM SFP protein, 5 µM BarD,
5 µM BarA, and 800 µM [UL-14C]leucine. The reaction was initiated
by the addition of 5 mM ATP. Prior to the initiation of the reaction,
the sample was incubated in the absence of ATP and leucine for 45
min at 37 °C to convert the apo PCP domain in BarA to the
homoenzyme (phosphopantetheine containing). Once the reaction was
initiated by the addition of ATP, it was incubated for an additional 30
min at 37 °C. The reaction was quenched by the addition of 0.8 mL of
10% TCA containing 375 µg of BSA. The precipitated proteins were
pelleted by centrifugation at 12600g and washed three times with 10%
TCA to remove any free [UL-14C]leucine from the assay mixture. After
the final wash, the protein pellet was resuspended in 150 µL of 1 M
Tris, diluted with scintillation fluid (3.5 mL), and counted for
radioactivity. Alternatively, the samples were resuspended in 20 µL of
1× Laemmli sample buffer and analyzed by SDS-PAGE, Coomassie
staining, and autoradiography.
Didechlorobarbaleucamide B (7): colorless oil; 1H NMR (DMSO-
d6, 300 MHz) δ 7.74 (2H, m, H-14/H-15), 6.22 (2H, m, H-11/H-12),
5.50 (1H, s, H-5), 5.35 (1H, m, H-7), 3.66 (3H, s, OCH3), 3.10-2.10
(6H, m, H-2, H-9, H2-3, H2-8), 2.82 (3H, s, N-CH3) 1.39 (3H, d, J )
6.4 Hz, H3-10), 0.96 (3H, d, J ) 6.0 Hz, H3-1); 13C NMR partial data
(DMSO-d6, 75 MHz) δ 15.9 (CH3, C-1), 15.3 (CH3, C-10); TOF MS/
MS 463.0 ES+ m/z 247/246/245/244 (18/100/19/35), 211/210/209/208
(6/74/27/78), 182/181/180/179 (41/9/44/7); HRESIMS obsd M+ m/z
460.0300 (calcd for C17H2435Cl4N2O2S, 460.0313).
Recombinant Protein Expression. The genes barA, barB1, barB2,
and barC were each subcloned and recombinantly expressed in frame
with the 6XHis fusion tag into the pET20b expression vector. Full-
length barA, barB1, barB2, and barC were each PCR amplified and
subcloned into the pET20b expression vector at the Nde1/XhoI
restriction sites in frame with the C-terminal 6XHis fusion tag. Full-
length barD and the barE adenylation domain were subcloned into
pET20b as previously described.16 The Bacillus subtilis SFP gene,
kindly provided by C. Walsh (Harvard), was PCR amplified from the
pUC-SFP vector and subcloned in frame with the 6XHis tag into the
pET29 expression vector using the following primers: SFP-F1, 5′-
GGAATTCCATATGAAGATTTACGGAATTTATATG-3′; SFP-R1,
5′-CCGCTCGAGTAAAAGCTCTTCGTACGAGAC-3′.
2-Oxoglutarate Conversion Assay. To measure basal levels of
2-oxoglutarate activity in the absence of substrate, increasing concentra-
tions of BarB1 and BarB2 were incubated with [1-14C]2-oxoglutarate
in 50 mM Tris, pH 7.5, 1 mM FeSO4, 100 µM TCEP, 10 mM ascorbate,
4 mM ATP, and 50 mM (NH4)2SO4. The reaction mixture was initiated
by the addition of the [1-14C]2-oxoglutarate in an open 1.5 mL
Eppendorf tube. The tube was then incubated open, suspended over
200 µL of hyamine hydroxide in a sealed 50 mL conicle tube. Hyamine
hydroxide absorbs [14C]CO2 that is released during the conversion of
[1-14C]2-oxoglutarate to succinate. To measure the effect of leucyl-
BarA on the conversion of [1-14C]2-oxoglutarate to [14C]CO2 and
Recombinant proteins BarA, BarB1, BarB2, BarC, BarD, and BarE
(adenylation domain) were expressed and purified from BL21∆(DE3)

944 Journal of Natural Products, 2006, Vol. 69, No. 6
Flatt et al.
(11) Orjala, J.; Gerwick, W. H. J. Nat. Prod. 1996, 59, 427-430.
(12) Orjala, J.; Nagle, D.; Gerwick, W. H. J. Nat. Prod. 1995, 58, 764-
768.
(13) Williamson, R. T.; Boulanger, A.; Vulpanovici, A.; Roberts, M. A.;
Gerwick, W. H. J. Org. Chem. 2002, 67, 7927-7936.
(14) Edwards, D. J.; Gerwick, W. H. J. Am. Chem. Soc. 2004, 126,
11432-11433.
(15) Todd, J. S.; Gerwick, W. H. Tetrahedron Lett. 1995, 36, 7837-
7840.
succinate, we generated the BarA-leucyl intermediate as above in the
thioester formation assay, using 2 mM unlabeled leucine instead of
the [UL-14C]leucine. A total of 100 µL of this reaction mixture was
added to the 2-oxoglutarate protocol given above for a final volume of
200 µL. All of the reactions were incubated for 2 h at 28 °C and stopped
by the addition of 800 µL of 10% TCA. The hyamine hydroxide was
transferred to 3 mL of scintillation cocktail and analyzed by liquid
scintillation counting (LSC).
Adenylation Domain Exchange Assay. BarD and the BarE ade-
nylation domains were prepared as previously described and assayed
in the ATP-PPi exchange assay. Briefly, the assay was initiated by
the addition of 0.3 µCi of [32P] pyrophosphate to 2 µM protein, 2 mM
ATP, and 2 mM substrate (glycine, leucine, trichloroleucine, valine,
isovaleric acid, trichloroisovaleric acid, or the R-keto derivatives of
leucine and trichloroleucine), in 75 mM Tris, pH 8.0, with 100 mM
NaCl, 10 mM MgCl2, and 5 mM DTT. The samples were incubated at
30 °C for 2 h and the reactions quenched by the addition of 0.5 mL of
Stop Buffer (1.2 wt %/vol activated charcoal, 0.1 M tetrasodium
pyrophosphate, and 0.35 M HClO4). Trapped ATP was washed three
times using 0.1 M tetrasodium pyrophosphate in 0.35 M HClO4. The
charcoal pellet was resuspended in 0.5 mL of H2O, added to 3.5 mL
of scintillation fluid, and analyzed by LSC.
(16) Chang, Z.; Flatt, P.; Gerwick, W. H.; Nguyen, V. A.; Willis, C. L.;
Sherman, D. H. Gene 2002, 296, 235-247.
(17) Sitachitta, N.; Marquez, B. L.; Williamson, R. T.; Rossi, J.; Roberts,
M. A.; Gerwick, W. H.; Nguyen, V.-A.; Willis, C. L. Tetrahedron
2000, 56, 9103-9114.
(18) Bender, C. L.; Arcon-Chaidez, F.; Gross, D. C. Microbiol. Mol. Biol.
ReV. 1999, 63, 266-292.
(19) Prescott, A. G. J. Exp. Bot. 1993, 44, 849-861.
(20) Prescott, A. G.; Lloyd, M. D. Nat. Prod. Rep. 2000, 17, 367-383.
(21) Flatt, P. M.; Gautschi, J. T.; Thacker, R. W.; Musafija-Girt, M.;
Crews, P.; Gerwick, W. H. Mar. Biol. 2005, 147, 761-774.
(22) Vaillancourt, F. H.; Yin, J.; Walsh, C. T. Proc. Natl. Acad. Sci. U.S.A.
2005, 102, 10111-10116.
(23) Vaillancourt, F. H.; Yeh, E.; Vosburg, D. A.; O’Connor, S. E.; Walsh,
C. T. Nature 2005, 436, 1191-1194.
Acknowledgment. We thank C. Walsh (Harvard) for providing the
B. subtilis SFP expression construct (pUC-SFP). Financial support for
this project was through NIH GM 83155 and the EPSRC (to S.J.O.).
We thank the country of Curac¸ao and the Carmabi Research Institute
for permission to make collections of marine cyanobacteria.
(24) Gerwick, W. H.; Leslie, P.; Long, G. C.; Marquez, B. L.; Willis, C.
L. Tetrahedron Lett. 2003, 44, 285-288.
(25) Galonic, D. P.; Vaillancourt, F. H.; Walsh, C. T. J. Am. Chem. Soc.
2006, 128, 3900-3901.
(26) Harrigan, G. G.; Goetz, G. H.; Luesch, H.; Yang, S.; Likos, J. J.
Nat. Prod. 2001, 64, 1133-1138.
(27) McDonough, M. A.; Kavanagh, K. L.; Butler, D.; Searls, T.;
Oppermann, U.; Schofield, C. J. J. Biol. Chem. 2005, 280, 41101-
41110.
(28) Mukherji, M.; Chien, W.; Kershaw, N. J.; Clifton, I. J.; Schofield,
C. J.; Wierzbicki, A. S.; Lloyd, M. D. Hum. Mol. Genet. 2001, 10,
1971-1982.
(29) Flatt, P. M.; Chang, Z.; Zeller, G.; Sherman, D. H.; Gerwick, W. H.
Poster: SIM 7th International Conference on the Biotechnology of
Microbial Products, 2002.
(30) Yeh, E.; Garneau, S.; Walsh, C. T. Proc. Natl. Acad. Sci. U.S.A.
2005, 102, 3960-3965.
(31) Hannesian, S.; Margarita, R. Tetrahedron Lett. 1998, 39, 5887-5890.
(32) Arda, A.; Jimenez, C.; Rodriguez, J. Tetrahedron Lett. 2004, 45,
3241-3243.
(33) Takeda, T.; Sasaki, R.; Yamauchi, S.; Fujiwara, T. Tetrahedron 1997,
53, 557-566.
(34) Williard, P. G.; deLaszlo, S. E. J. Org. Chem. 1984, 49, 3489-3493.
(35) Fletcher, M. D.; Harding, J. R.; Hughes, R. A.; Kelly, N. M.; Schmalz,
H.; Sutherland, A.; Willis, C. L. J. Chem. Soc., Perkin Trans. 1 2000,
1, 43-51.
(36) Kaku, N.; Hibino, T.; Tanaka, Y.; Ishikawa, H.; Araki, E.; Takabe,
T.; Takabe, T. Plant Sci. 2000, 159, 281-288.
Supporting Information Available: Panels of selected carbon atom
intensities in 13C-enriched barbamide, and LC-MS data for 13C-labeled
barbamide and didechlorobarbaleucamide B. This material is available
free of charge via the Internet at http://pubs.acs.org.
References and Notes
(1) Gerwick, W. H.; Tan, L. T.; Sitachitta, N. Alkaloids Chem. Biol.
2001, 57, 75-184.
(2) Burja, A. M.; Banaigs, B.; Abou-Mansour, E.; Burgess, J. G.; Wright,
P. C. Tetrahedron 2001, 57, 9347-9377.
(3) Schwartsmann, G.; Da Rocha, A. B.; Mattei, J.; Lopes, R. Expert.
Opin. InVest. Drugs 2003, 12, 1367-1383.
(4) Perez, E. A.; Hillman, D. W.; Fishkin, P. A.; Krook, J. E.; Tan, W.
W.; Kuriakose, P. A.; Alberts, S. R.; Dakhil, S. R. InVest. New Drugs
2005, 23, 257-261.
(5) Bayes, M.; Rabasseda, X.; Prous, J. R. Methods Find. Exp. Clin.
Pharmacol. 2004, 26, 723-753.
(6) Simmons, T. L.; Andrianasolo, E.; McPhail, K.; Flatt, P.; Gerwick,
W. H. Mol. Cancer Ther. 2005, 4, 333-342.
(7) Hooper, G. J.; Orjala, J.; Schatzman, R. C.; Gerwick, W. H. J. Nat.
Prod. 1998, 61, 529-533.
(8) Orjala, J.; Nagle, D. G.; Hsu, V.; Gerwick, W. H. J. Am. Chem. Soc.
1995, 117, 8281-8282.
(37) Mukherji, M.; Chien, W.; Kershaw, N. J.; Clifton, I. J.; Schofield,
C. J.; Wierzbicki, A. S.; Lloyd, M. D. Hum. Mol. Genet. 2001, 10,
1971-1982.
(9) Blokhin, A. V.; Yoo, H. D.; Geralds, R. S.; Nagle, D. G.; Gerwick,
W. H.; Hamel, E. Mol. Pharmacol. 1995, 48, 523-531.
(10) Marquez, B.; Verdier-Pinard, P.; Hamel, E.; Gerwick, W. H.
Phytochemistry 1998, 49, 2387-2389.
NP050523Q