Insights into the reaction mechanism of the prolyl - Acyl carrier protein oxidase involved in anatoxin-a and homoanatoxin-a biosynthesis

Article abstract of DOI:10.1021/bi200892a

Anatoxin-a and homoanatoxin-a are two potent cyanobacterial neurotoxins. We recently reported the identification of the gene cluster responsible for the biosynthesis of these toxins as well as the in-vitro reconstitution of the first steps of this biosynthesis. We now report experimental evidence supporting the proposed reaction mechanism of AnaB, a flavoprotein homologous to acyl-CoA dehydrogenase. AnaB catalyzes the two-electron oxidation of prolyl-AnaD, which is proline linked to the acyl carrier protein holo-AnaD, to dehydroprolyl-AnaD using oxygen as the second substrate. AnaB is thus an oxidase. By using liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS), we have identified and characterized dehydroprolyl-AnaD, the AnaB product. We estimated an apparent catalytic constant of 1 min^{- 1} for AnaB catalysis. We synthesized several deuterium-labeled prolines and enzymatically transformed them into their corresponding prolyl-AnaD. These deuterium-labeled prolyl-AnaDs were oxidized in the presence of AnaB, and the deuterium labeling in the remaining substrate and in the product was determined by LC-MS/MS. The data supported a reaction mechanism starting with a rapid enolization followed by a slow oxidation to give the conjugated imine, which in turn was isomerized to pyrroline-5-carboxyl-AnaD. We also showed that cis- and trans-4-fluoro-l-prolyl- AnaD and 3,4-dehydro-l-prolyl-AnaD were transformed into pyrrole-2-carboxyl-AnaD by AnaB. Thus, the 4-fluoro-analogues experienced a β-elimination supporting the AnaB-catalyzed aza - allylic isomerization. We identified by sequence alignment the AnaB active site base, Glu244. We produced, purified, and characterized the E244A AnaB mutant, which is inactive, supporting the catalytic role of E244 as a base.

Full text of DOI:10.1021/bi200892a

Article
pubs.acs.org/biochemistry
Insights into the Reaction Mechanism of the Prolyl−Acyl Carrier
Protein Oxidase Involved in Anatoxin-a and Homoanatoxin-a
Biosynthesis
Stephane Mann,†,‡ Berangere Lombard,∥ Damarys Loew,∥ Annick Mejean,†,‡,§ and Olivier Ploux*,†,‡
́
́
̀
́
†Laboratoire Charles Friedel, Chimie ParisTech, ENSCP, 11 rue Pierre et Marie Curie, 75231 Paris Cedex 05, France
‡CNRS, UMR 7223, 75005 Paris, France
§Universite
́
Paris Diderot-Paris 7, 75013 Paris, France
∥Laboratory of Proteomic Mass Spectrometry, Centre de Recherche, Institut Curie, 26 rue d'Ulm 75248, Paris Cedex 05, France
S
* Supporting Information
ABSTRACT: Anatoxin-a and homoanatoxin-a are two potent cyanobacterial
neurotoxins. We recently reported the identification of the gene cluster
responsible for the biosynthesis of these toxins as well as the in-vitro
reconstitution of the first steps of this biosynthesis. We now report experimental
evidence supporting the proposed reaction mechanism of AnaB, a flavoprotein
homologous to acyl-CoA dehydrogenase. AnaB catalyzes the two-electron
oxidation of prolyl-AnaD, which is proline linked to the acyl carrier protein
holo-AnaD, to dehydroprolyl-AnaD using oxygen as the second substrate. AnaB
is thus an oxidase. By using liquid chromatography coupled to tandem mass
spectrometry (LC-MS/MS), we have identified and characterized dehydro-
prolyl-AnaD, the AnaB product. We estimated an apparent catalytic constant of 1 min−1 for AnaB catalysis. We synthesized
several deuterium-labeled prolines and enzymatically transformed them into their corresponding prolyl-AnaD. These deuterium-
labeled prolyl-AnaDs were oxidized in the presence of AnaB, and the deuterium labeling in the remaining substrate and in the
product was determined by LC-MS/MS. The data supported a reaction mechanism starting with a rapid enolization followed by
a slow oxidation to give the conjugated imine, which in turn was isomerized to pyrroline-5-carboxyl-AnaD. We also showed that
cis- and trans-4-fluoro-L-prolyl-AnaD and 3,4-dehydro-L-prolyl-AnaD were transformed into pyrrole-2-carboxyl-AnaD by AnaB.
Thus, the 4-fluoro-analogues experienced a β-elimination supporting the AnaB-catalyzed aza−allylic isomerization. We identified
by sequence alignment the AnaB active site base, Glu244. We produced, purified, and characterized the E244A AnaB mutant,
which is inactive, supporting the catalytic role of E244 as a base.
yanobacteria produce a wide range of secondary
However, the discovery of the ana genes and of the
biosynthetic route leading to anatoxin-a and homoanatoxin-a
revealed some interesting enzymatic steps, as shown in Scheme
1. We proposed that proline is loaded on AnaD, an acyl carrier
protein (ACP), and oxidized to (S)-1-pyrroline-5-carboxyl-
AnaD (P5C-AnaD). This starter would then be transferred to
the polyketide synthase (PKS) AnaE for elongation. The
second PKS, AnaF, would then add another acetate unit and
catalyze a Mannich-type cyclization step to form anatoxic acid
linked to the ACP module of AnaF. The third PKS, AnaG,
would catalyze the formation of the β-oxo-thioester precursor
of anatoxin-a or, if a methylation had occurred on AnaG, that of
homoanatoxin-a. Hydrolysis of the β-oxo-thioester and
decarboxylation of the acid, to give the corresponding toxin,
would be catalyzed by the thioesterase, AnaA. We have recently
reconstituted in vitro the first steps of this pathway and showed
that AnaC is indeed a proline specific adenylation protein and
1
Cmetabolites including toxins to human and animals.
These oxygenic photosynthetic prokaryotes can be found in
almost every terrestrial or aquatic environment and thus pose a
threat to human and animal health. Indeed, cases of animal
death and human intoxication, due to cyanobacterial toxin
exposure, are regularly reported in different places around the
world.2−7 For instance, anatoxin-a and homoanatoxin-a, two
cyanobacterial neurotoxins, provoke the rapid death of animals
by acute asphyxia, when ingested, because these alkaloids are
potent agonists of the nicotinic acetylcholine receptor.8 We
have recently reported the identification of the gene cluster
responsible for the biosynthesis of these toxins in the
cyanobacterium Oscillatoria sp. PCC 6506 (PCC stands for
Pasteur culture collection of cyanobacteria) and proposed an
original biosynthetic scheme for these alkaloids (Scheme 1).9
The identification of the ana biosynthetic genes was an
important step because it provided a simple way to detect in
water the presence of anatoxin-a producing cyanobacteria by
specific PCR amplifications.10−12
Received: June 9, 2011
Revised: July 22, 2011
Published: July 25, 2011
© 2011 American Chemical Society
7184
dx.doi.org/10.1021/bi200892a|Biochemistry 2011, 50, 7184−7197

Biochemistry
Article
vis spectra and absorbance measurements were recorded on an
Uvikon 930 Kontron spectrophotometer. NMR spectra were
recorded at the ENSCP NMR laboratory. Protein purifications
were achieved using an integrated protein purification system
(AKTAprime, GE Healthcare). PCR amplifications and
electrophoretic separations (agarose and acrylamide) were
run as already described.9 Protein concentrations were
estimated by using the Bradford colorimetric assay (BioRad).28
Recombinant histidine-tagged, apo-AnaD, holo-AnaD, AnaC,
and wild-type AnaB were obtained as previously described.9,13
Scheme 1. The ana Gene Cluster Responsible for the
Biosynthesis of Anatoxin-a and Homoanatoxin-a in
Cyanobacteria and the Postulated Biosynthetic Route
Leading to These Neurotoxins
Synthesis of Deuterium-Labeled Proline. The follow-
ing synthesis have been adapted from previous works.29−32
(±)-[2-2H]-2-Carboxypyrrolidinium Chloride (1). L-Proline
(265 mg, 2.3 mmol) was dissolved in D2O (1.5 mL) to
substitute exchangeable hydrogen atoms by deuterium, and the
solution was concentrated under vacuum. The residue was
placed under an argon atmosphere, and pyridoxal-5′-phosphate
(PLP, 53 mg, 0.2 mmol) was added. Then the mixture was
dissolved in a solution of NaOD prepared from sodium (265
mg, 11.5 mmol) and D2O (2 mL) and refluxed under argon for
8 h. The resulting solution was acidified to pH 1 with aqueous 1
M HCl and loaded on a TEMEX 50W-x8 column, which was
eluted with a gradient of aqueous HCl (from 0 to 1 M). The
fractions containing proline were concentrated, and the
hydrochloride salt of deuterium-labeled proline was isolated
as white crystals; mp 158 °C, lit. 156 °C.30 Because the
deuterium incorporation after this process was only 58%, as
measured by integrating the NMR signal of the C2−H proton,
the reaction was repeated twice until incorporation of
provided initial experimental evidence supporting the formation
of dehydroprolyl-AnaD from prolyl-AnaD by the flavoprotein
AnaB.13 We also noted that these steps leading to P5C-AnaD
were reminiscent of those involved in the formation of pyrrole-
2-carboxyl-ACP, the starter of a number of biosyntheses of
pyrrole containing secondary metabolites.14−26 However, AnaB
is unique among these enzymes because it catalyzes a two-
electron oxidation reaction rather than a four-electron oxidation
as its homologues.
1
deuterium at that position reached a value of 97%. H NMR
(400 MHz, D2O): δ 2.05 (2H, m, CH2CH2N), 2.16 (1H, m,
CH2CHCO), 2.42 (1H, m, CH2CHCO), 3.40 (2H, m,
CH2N), 4.45 (0.03H, m, CH). 13C NMR (75 MHz, D2O): δ
23.3 (CH2CH2N), 28.2 (CH2CHCO), 46.2 (CH2N), 59.3
(CD, t), 171.9 (CO).
(2S)-5-(Piperidine-1-carbonyl)pyrrolidin-2-one (2). A mix-
ture of L-pyroglutamic acid (1 g, 7.7 mmol), 1-hydroxybenzo-
triazole (HOBT, 1 g, 7.7 mmol), and N,N′-dicyclohexylcarbo-
diimide (DCC, 1.6 g, 7.7 mmol) in tetrahydrofuran (THF, 25
mL) was stirred 10 min at room temperature under an argon
atmosphere. Piperidine (760 μL, 7.7 mmol) was then added
dropwise, and the solution was stirred for 4 h. After
concentrating the solution under vacuum, the crude mixture
was separated by flash chromatography (CH2Cl2/ethanol: 195/
5 to 180/20, v/v) to give compound 2 as a colorless oil (1.4 g,
95% yield). 1H NMR (300 MHz, CDCl3): δ 1.52−164 (6H, m,
CH2CH2CH2N), 2.02−2.12 (1H, m, CH2CH2CO), 2.28−2.48
(3H, m, CH2CH2CO), 3.31−3.62 (4H, m, CH2N), 4.48 (1H,
t, J = 8.0 Hz, CHCO), 6.82 (1H, broad, NH). 13C NMR (75
MHz, CDCl3): δ 24.3 (CH2CH2CO), 25.3 (CH2CH2N), 26.3
(CH2CH2CH2N), 29.5 (CH2CO), 43.3 and 46.0 (CH2N),
54.2 (CHN), 169.5 (CO), 178.3 (CO).
We now report the complete identification and character-
ization, by high performance liquid chromatography (HPLC)
and by liquid chromatography coupled to tandem mass
spectrometry (LC-MS/MS), of dehydroprolyl-AnaD, the
AnaB product. Using diverse deuterium-labeled prolines as
well as 4-fluoro-L-prolines and 3,4-dehydro-L-proline, we have
studied the reaction mechanism of this oxidation step. We
provide new evidence supporting a reaction mechanism, for this
flavoprotein, involving several steps to give P5C-AnaD as the
product. We have also identified, by site-directed mutagenesis,
the putative active site base of AnaB and showed that this
enzyme uses oxygen as the second substrate. AnaB is thus a
close homologue of acyl-CoA dehydrogenase (ACAD) like the
human isovaleryl-CoA dehydrogenase (hIVD), in terms of
sequence and reaction mechanism, but is rather an oxidase than
a dehydrogenase.
(2S)-1-(tert-Butoxycarbonyl)-5-(piperidine-1-carbonyl)-
pyrrolidin-2-one (3). A solution composed of 2 (600 mg, 3
EXPERIMENTAL PROCEDURES
■
mmol), di-tert-butyl dicarbonate (Boc2O, 1.3 g, 6 mmol), 4-
(dimethylamino)pyridine (DMAP, 37 mg, 0.3 mmol), and dry
acetonitrile (7 mL) under an argon atmosphere was stirred
overnight at room temperature. After concentration under
vacuum, the crude mixture was purified by flash chromatog-
raphy (cyclohexane/ethyl acetate: 5/5 to 0/1, v/v). The tert-
butyloxycarbonyl (Boc) derivative 3 was isolated as a white
General. Alignments of protein sequences were performed
using the Clustal X software.27 Chemicals and biochemicals
were purchased from Sigma-Aldrich. HPLC was run on a
Hewlett-Packard 1050 series apparatus equipped with a Jupiter
5 μm C18 300 Å Phenomenex column. Elutions were
performed using a linear gradient (from 60% to 80% (v/v) in
20 min) of methanol in aqueous 0.1% trifluoroacetic acid
(TFA) at 1 mL/min, and the detector was set at 210 nm. UV−
1
powder (638 mg, 72% yield). H NMR (300 MHz, CDCl3): δ
7185
dx.doi.org/10.1021/bi200892a|Biochemistry 2011, 50, 7184−7197

Biochemistry
Article
CH), 6.64 (1H, broad, NH). 13C NMR (75 MHz, CDCl3): δ
24.8 (CH2), 27.9 (CH3), 29.4 (CH2), 56.2 (CHN), 82.3
(C(CH3)3), 171.1 (CO), 178.2 (CO).
tert-Butyl (2S)-(1-tert-Butoxycarbonyl)-5-oxo-pyrrolidine-
2-carboxylate (7). The ester 6 (1.1 g, 5.9 mmol) was
solubilized in THF (23 mL) in the presence of DMAP (55
mg, 0.45 mmol) and Boc2O (1.96 g, 9 mmol) at 4 °C, under an
argon atmosphere. The stirring was maintained overnight, and
the temperature was raised to room temperature. Then the
mixture was concentrated and purified by flash chromatography
(cyclohexane/ethyl acetate: 9/1 to 7/3, v/v) to give the Boc-
derivative 7 as a yellow oil, which crystallized (1.6 g, 95% yield).
1H NMR (300 MHz, CDCl3): δ 1.49 (9H, s, CH3), 1.51 (9H,
s, CH3), 1.95−2.05 (1H, m, CH2), 2.22−2.36 (1H, m, CH2),
2.42−2.68 (2H, m, CH2), 4.48 (1H, dd, J = 2.6 Hz and J = 9.4
Hz, CH). 13C NMR (75 MHz, CDCl3): δ 21.6 (CH2), 27.9
(CH3), 31.1 (CH2), 59.6 (CH), 82.2 and 83.3 (C(CH3)3),
149.3 (CO), 170.3 (CO), 173.5 (CO).
1.44 (9H, s, CH3), 1.54−1.64 (6H, m, CH2), 1.79−1.84 (1H,
m, CH2CH2CO), 2.14−2.28 (1H, m, CH2CH2CO), 2.35−2.44
(1H, m, CH2CO), 2.56−2.69 (1H, m, CH2CO), 3.36−3.48
(3H, m, CH2N), 3.60−3.68 (1H, m, CH2N), 4.91 (1H, dd, J =
1.5 Hz and J = 9.0 Hz, CHN). 13C NMR (75 MHz, CDCl3): δ
21.6 (CH2CH2CO), 24.3 (CH2CH2CH2N), 25.5 and 26.4
(CH2CH2N), 27.9 (C(CH3)3), 31.2 (CH2CO), 43.3 and 46.1
(CH2N), 56.6 (CHN), 82.2 (C(CH3)3), 149.5 (CO), 168.5
(CO), 173.9 (CO).
(2S)-1-(tert-Butoxycarbonyl)-2-(piperidine-1-carbonyl)-
[5,5-2H2]-pyrrolidine (4). A solution of iodine (305 mg, 1.2
mmol) in dry THF (2 mL) was slowly added (in 3 h) dropwise
to a solution of 3 (296 mg, 1 mmol) in dry THF (2 mL) in the
presence of NaBD4 (84 mg, 2 mmol, 97% deuterium labeling)
at −12 °C, and the temperature was maintained constant over
the experiment. The reaction was stopped at −5 °C by adding
ethanol (1 mL), and the mixture was diluted with CH2Cl2 (5
mL) and brine (1 mL). The organic layer was separated, dried
over MgSO4, concentrated under vacuum, and separated by
flash chromatography (cyclohexane/ethyl acetate: 7/1 to 0/1,
v/v). The pyrrolidine 4 was isolated as a yellow oil (119 mg,
42% yield). NMR spectra showed the presence of two rotamers
in a 1/1 ratio. Estimated deuterium labeling at position C5 =
tert-Butyl (2S)-(1-tert-Butoxycarbonyl)-5-hydroxypyrroli-
dine-2-carboxylate (8). A solution of carbamate 7 (1 g, 3.5
mmol) in dry THF (30 mL) was cooled to −78 °C, and 5 mL
of a 1 M solution of diisobutylaluminum hydride (DIBAL-H)
in dry THF was added dropwise over 10 min. The mixture was
maintained under stirring at this temperature for 2 h, and the
reaction was quenched by adding a saturated aqueous solution
of NaHCO3. The temperature was slowly increased to room
temperature in 20 min. An aqueous solution of 35% w/v H2O2
(10 drops) was then added, and stirring was maintained at this
temperature for 30 min. The mixture was diluted with CH2Cl2
(200 mL), with a saturated aqueous solution of NaHCO3 (40
mL), and with water (40 mL). The organic layer was extracted
with CH2Cl2 (2 × 75 mL), and the combined fractions were
dried and concentrated over Na2SO4. The crude product was
then purified by flash chromatography (cyclohexane/ethyl
acetate: 9/1 to 7/3, v/v) to yield 8 as a mixture of
diastereoisomers and conformers (yellow oil, 950 mg, 95%
1
90%. H NMR (300 MHz, CDCl3): δ 1.36 and 1.41 (9H, s,
CH3), 1.50−1.70 (6H, m, CH2), 1.78−1.82 (2H, m, CH2),
1.86−1.99 (1H, m, CH2CHCO), 2.04−2.19 (1H, m,
CH2CHCO), 3.30−3.70 (4.2H, m, CH2N), 4.48 and 4.63
(1H, dd, J = 2.6 and 8.3 Hz, CHCO). 13C NMR (75 MHz,
CDCl3):
δ 23.2 and 23.8 (CH2CH2CO), 24.5
(CH2CH2CH2N), 25.7 and 26.4 (CH2CH2N), 28.4 and 28.5
(C(CH3)3), 29.8 and 30.5 (CH2CO), 43.1 and 46.2 (CH2N),
56.5 and 56.9 (CHN), 79.2 (C(CH3)3), 154.0 and 154.5 (CO),
170.4 and 170.5 (CO).
(2S)-[5,5-2H2]-2-Carboxypyrrolidinium Chloride (5). The
pyrrolidine 4 was solubilized in a (1/1, v/v) TFA/CH2Cl2
solution (3 mL) and stirred for 30 min at room temperature.
The solvents were removed by successive dilutions with
CH2Cl2 and concentrations under vacuum (3 times). The
crude product was then solubilized in an aqueous 0.5 M NaOH
solution (3 mL) and refluxed overnight. The mixture was
acidified to pH 3 with an aqueous 1 M HCl solution and
purified on a TEMEX 50W-x8 column that was eluted with an
aqueous HCl gradient (from 0 to 1 M). The fractions
containing proline were combined and evaporated to afford
the hydrochloride salt of the labeled proline 5 (54 mg, 85%
1
yield); Rf = 0.3 (cyclohexane−AcOEt). H NMR (300 MHz,
CDCl3): δ 1.36, 1.38, 1.41, 1.43 (18H, s, CH3), 1.74−2.05 (2H,
m CH2CH2), 2.10−2.60 (2H, m, CH2CH2), 2.75 (1H, broad,
OH), 4.05−4.45 (1H, m, CHCO), 5.35−5.58 (1H, m,
CHOH). 13C NMR (75 MHz, CDCl3): δ 27.9, 27.94, 28.3,
28.4 (CH3), 30.9, 31.1, 32.2 (CH2CH2), 59.6, 59.9, 60.0
(CHCO), 80.7, 80.9, 81.1 (C(CH3)3), 81.9, 82.2, 82.5, 88.3
(CHOH), 154.4 (CO), 171.6 (CO).
tert-Butyl (2S,5S)-(1-tert-Butoxycarbonyl)-[5-2H]-pyrroli-
dine-2-carboxylate (9). A solution of hemiaminal 8 (150
22
yield); mp 60 °C, lit. 115 °C (30); [α]D −33.1° (H2O, c =
1
2); deuterium labeling at position C5 = 91%. H NMR (300
mg, 0.52 mmol) in anhydrous toluene (5 mL) under an argon
atmosphere was cooled down to −78 °C, and freshly distilled
BF3−OEt2 (160 μL, 1.14 mmol) and Bu3SnD (280 μL, 1.06
mmol, 96% deuterium labeling) were added in two portions at
30 min intervals. The mixture was stirred at this temperature
for 4 h. Then, the reaction was quenched by adding saturated
aqueous NaHCO3 (0.5 mL). The mixture was warmed to room
temperature, diluted with 10% aqueous NaHCO3, and
extracted with CH2Cl2 (3 × 10 mL). The organic layer was
dried over Na2SO4, concentrated, and purified by flash
chromatography (cyclohexane/ethyl acetate: 9/1, v/v). The
labeled pyrrolidine 9 was obtained as a colorless oil (20 mg,
MHz, D2O): δ 2.02 (2H, t, J = 7.0 Hz, CH2CD2), 2.08−2.20
(1H, m, CH2CH), 2.35−2.47 (1H, m, CH2CH), 3.35 and 3.40
(0.09H, t, J = 7.5 Hz, CHDNH), 4.38 (1H, t, J = 7.7 Hz,
CHN). 13C NMR (75 MHz, D2O): δ 23.1 and 28.3
(CH2CH2), 45.9 (CD2), 59.6 (CH), 172.0 (CO).
tert-Butyl (2S)-5-Oxo-pyrrolidine-2-carboxylate (6). A sol-
ution of L-pyroglutamic acid (2 g, 15.5 mmol) in tert-butyl
acetate (8 mL, 60 mmol) and in aqueous 70% w/v HClO4 (0.5
mL) was stirred under argon for 24 h. The mixture was then
poured into a saturated aqueous NaHCO3 solution (50 mL)
and extracted with diethyl ether (4 × 100 mL). The organic
layer was then dried over MgSO4 and concentrated under
1
15% yield). H NMR (300 MHz, CDCl3): δ 1.41 and 1.44
1
(18H, s, CH3), 1.76−1.96 and 2.10−2.24 (3H + 1H, m,
CH2CH2), 3.39 and 3.50 (1H, m, CHD), 4.06−4.17 (1H, m,
CHCO). 13C NMR (75 MHz, CDCl3): δ 23.3 and 24.1
vacuum to give 6 as a white powder (1.18 g, 31% yield). H
NMR (300 MHz, CDCl3): δ 1.41 (9H, s, CH3), 2.10 (1H, m,
CH2), 2.31(2H, m, CH2), 2.37(1H, m, CH2), 4.08 (1H, m,
7186
dx.doi.org/10.1021/bi200892a|Biochemistry 2011, 50, 7184−7197

Biochemistry
Article
(CH2CHCO), 27.9 and 28.0 (CH3), 29.7 and 30.9
(CH2CHD), 46.5 (CHD), 58.4 and 59.7 (CHCO), 79.6 and
80.8 (C(CH3)3), 154.0 (CO), 172.3 (CO).
26000g, SS-34 rotor, 15 min) the supernatant was loaded on a 1
mL HisTrap column (GE Healthcare). The desired protein was
eluted with a linear gradient of imidazole (20−500 mM) in 20
mM Na-phosphate buffer, pH 7.4, 0.5 M NaCl. Pure fractions
of AnaB were pooled, and the protein was desalted and
concentrated by ultrafiltration (Amicon Ultra-15 Centrifugal
Filter Unit 30k) and stored in 20 mM Tris-HCl buffer, pH 7.4,
20% (v/v) glycerol at −80 °C. A sample of pure recombinant
E244A AnaB (30 μg) was injected and purified by HPLC,
under the conditions described above. The single peak at 15.5
min was collected, and the solution was concentrated under
vacuum and analyzed by electrospray ionization mass
spectrometry (ESI-MS), as already described.13 The protein
showed an experimental mass of 46 358.4 Da, in agreement
with the calculated mass of 46 359.07 Da for the protein lacking
the first Met. Recombinant wild-type AnaB, analyzed using the
same conditions, showed an experimental mass of 46 417.7 Da
for a calculated mass of 46 417.1 Da for the protein lacking the
first Met.
Dithionite Reduction of AnaB and Oxidation by Air. A
solution of either recombinant histidine-tagged wild-type AnaB
(7.9 μM) or its E244A mutant (12.8 μM) in 20 mM Tris-HCl
buffer, pH 7.7, containing 10% v/v glycerol was introduced in a
cuvette fitted with a rubber septum. The electronic absorption
spectrum of the solutions was then recorded at room
temperature. Then, freshly prepared dithionite (52 mM sodium
dithionite aqueous solution) was added to the enzyme solution
to a final concentration of 0.52 mM through the septum using a
gastight syringe, the solution was gently mixed by manual
inverting, and the electronic absorption spectrum was recorded.
The solution was then exposed to air and gently mixed by
manual inverting, and the electronic absorption spectrum was
recorded again.
Preparation of Prolyl-AnaD and Analogues. Prolyl-
AnaD was prepared as previously described.9,13 The assay
consisted (100 μL total volume) of 100 mM Tris-HCl, pH 7.7,
1 mM tris(2-carboxyethyl)phosphine (TCEP), 10 mM MgCl2,
5 mM ATP, 5 mM L-proline, 80 μM holo-AnaD, and 1 μM
AnaC, and the reaction was incubated at 28 °C for 30 min. For
preparation of large quantities of prolyl-AnaD the assay was
adapted to 2 mL total volume, and the concentration of AnaC
was 2 μM. The prolyl-AnaD was not further purified because of
its instability (hydrolysis of the thioester bond) and used as
such for oxidation by AnaB. The analogues were produced
using the same procedure except that L-proline was replaced by
either 10 mM [2-2H]-DL-proline 1 or 5 mM of either [5,5-
2H2]-L-proline 5, 5-(S)-[5-2H]-L-proline 10, [2,3,3,4,4,5,5-
2H7]-L-proline (Cambridge Isotope, proline-d7, 97−98%), 4-
(R)-4-fluoro-L-proline (Bachem), 4-(S)-4-fluoro-L-proline
(Bachem), or 3,4-dehydro-L-proline (Sigma). The reaction of
formation of each prolyl-AnaD analogue was followed by
HPLC using the conditions described above to ensure that the
conversion was complete. The retention times were the
followings: holo-AnaD 19.3 min; prolyl-AnaD and deuterium-
labeled analogues 18.3 min; 4-fluoro-prolyl-AnaD 17.2 min;
3,4-dehydro-L-prolyl-AnaD 18.0 min.
(2S,5S)-[5-2H]-2-Carboxypyrrolidinium Chloride (10).
Compound 9 (20 mg, 0.073 mmol) was stirred in a (1/1, v/
v) TFA/CH2Cl2 solution (2 mL) for 20 min at room
temperature. After concentration of the mixture, the excess of
TFA was removed by successive dilution−concentration cycles
with CH2Cl2 (three times). The resulting oil was solubilized in
aqueous 2 M HCl (2 mL) and refluxed for 3 h. After
concentrating under vacuum, the hydrochloride salt of proline
22
10 was obtained (11 mg, quantitative yield). [α]D −21°
1
(H2O, c = 1.5). Integration of the H NMR spectrum of 10
showed that the C5-Hpro-R and C5-Hpro-S integrated for 0.85
and 0.15, respectively. 1H NMR (400 MHz, D2O): δ 1.90−1.95
(2H, m, CH2CHD), 1.99−2.07 (1H, m, CH2CHCO), 2.26−
2.33 (1H, m, CH2CHCO), 3.25 (0.85H, t, J = 8.0 Hz, (5R)-
CHD), 3.30 (0.15H, t, J = 8.0 Hz, (5S)−CHD), 4.25 (1H, t, J
= 7.8 Hz, CHCO). 13C NMR (75 MHz, D2O): δ 23.2 and 28.2
(CH2CH2), 45.9 (t, CHD), 59.6 (CHCO), 172.0 (CO).
Cloning, Expression, and Purification of the Recombi-
nant Histidine-Tagged Oscillatoria PCC 6506 E244A
AnaB. The entire plasmid, containing the wild-type anaB gene
for production of the histidine-tagged recombinant protein,13
was PCR-amplified (Phusion polymerase, New England
BioLabs) using the following primers (Eurogentec): anaBPstIF,
5′-CTA-GCC-GCT-GCA-GTG-GGA-ACA-ATG-GAA-CG-3′,
containing a silent mutation (bold typed) that creates a unique
PstI site (underlined), and anaBE244AR, 5′-GCC-TGC-AGC-
GGC-TAG-AAT-AAA-TCC−CCG-TGC−CCA-CTC-CAT-
AC-3′, containing a GC clamp (italicized), a single silent
mutation (bold typed) that creates a unique PstI site
(underlined), and a single mutation (bold typed) that creates
the E244A amino acid replacement. The overlapping sequence
(5′-CTA-GCC-GCT-GCA-G-3′) of the two primers contains
the unique new PstI restriction site (underlined). The resulting
6.4 kb PCR product was subjected to electrophoretic separation
on a 0.8% (w/v) agarose gel, and the band was excised and
extracted (PCR purification Kit, Promega). After digestion with
PstI the PCR product was circularized using T4 DNA ligase
(Promega), and the ligation product was used to transform
chemically competent E. coli JM109 cells. Positive clones
(lysogeny broth (LB) agar containing 50 μg/mL of kanamycin)
were analyzed and the plasmid purified, and the E244A anaB
insert sequence was verified by DNA sequencing (GATC Co.).
The plasmid pETanaBE244A was used for production of the
recombinant E244A AnaB histidine-tagged protein in E. coli
BL21-CodonPlus (DE3)-RIPL strain (Stratagene), essentially
as described for the wild-type enzyme13 with slight
modifications. Briefly, the resulting strain was grown in LB
medium supplemented with chloramphenicol (35 μg/mL) and
kanamycin (50 μg/mL) at 37 °C with shaking until it reached
an absorbance of 1 at 600 nm. The culture medium was then
cooled down to 16 °C; isopropyl-β-D-thiogalactopyranoside
(IPTG) was then added to a final concentration of 250 μM,
and the culture was left overnight at 16 °C with shaking. The
resulting cells were harvested by centrifugation (4000g, at 4 °C
for 20 min, Sorval RC5B centrifuge, SLA-3000 rotor) and lysed
by sonication (Misonix sonicator 500 W, 3 mm probe, 3 min
with 10 s pulses spaced by 20 s cooling periods at 4 °C) in 20
mM Na−phosphate buffer, pH 7.4, 0.5 M NaCl, 20 mM
imidazole, 100 μM FAD. After centrifugation (two runs at
AnaB-Catalyzed Oxidation Reaction of Prolyl-AnaD
and Analogues. Prolyl-AnaD, prepared as described above,
was oxidized in the presence of 10 μM purified recombinant
histidine-tagged wild-type AnaB. The enzyme was directly
added to the prolyl-AnaD solution (60 μM final concentration
of prolyl-AnaD, 100 μL final volume), and the reaction mixture
7187
dx.doi.org/10.1021/bi200892a|Biochemistry 2011, 50, 7184−7197

Biochemistry
Article
was incubated for 5 min at 28 °C. The sample was then
immediately analyzed by HPLC or frozen at −20 °C for
subsequent analysis by HPLC or LC-MS/MS. Dehydroprolyl-
AnaD eluted at 18.9 min, slightly after prolyl-AnaD under the
HPLC conditions described above. AnaB eluted at 15.5 min.
The same procedure was used for all prolyl-AnaD analogues
and for control experiments in the absence of AnaB or in the
presence of E244A AnaB (10 μM). Oxidized products always
eluted slightly after the corresponding reduced substrate, but
the peaks overlapped and the separation was not optimized.
sequence of the mutated gene was verified. The recombinant
E244A AnaB was purified to homogeneity as judged by
separation on gel electrophoresis (Figure S2), and we verified
by ESI-MS the integrity of the mutated protein (calculated 46
359.07 Da for the protein lacking the first Met, found 46 359.4
Da). The mutant protein behaved as the wild-type enzyme
during chromatographic separation and showed a characteristic
electronic absorption spectrum for FAD-containing proteins
(Figure 1).
Mass Spectrometry Analyses. Proly-AnaD, analogues,
and AnaB-oxidized products were analyzed using two different
strategies. The first one consisted in digesting the samples by
trypsin followed by LC-MS/MS analysis as already described.13
The mass and isotopic distributions of the ion corresponding to
the dephosphorylated phosphopantetheinyl arm with proline
(m/z 358.18) or dehydroproline (m/z 356.17) or analogues
attached to it were then determined. The second strategy,
which allowed rapid analysis to avoid degradation or possible
nonenzymatic exchange of deuterium, is based on LC
separation of prolyl-AnaD or analogues and its oxidized
product. LC-MS/MS analyses of the intact proteins were
done as described13 with minor modifications. Proteins (2
pmol) were loaded on a C18 pepMap300 guard column (0.3
mm i.d. × 5 mm long, 5 μm particle size, 300 Å pore size,
Dionex S. A.) in 40% solvent B (solvent B is 2% (v/v) water/
acetonitrile, 0.1% (v/v) formic acid and solvent A is 2% (v/v)
acetonitrile/water, 0.05% (v/v) TFA for loading, and 0.1% (v/
v) formic acid for elution). Elution from the LC Packing
PepMap300 column (75 μm i.d. × 150 mm long, packed with 5
μm particle size, 300 Å pore size, Dionex S. A.) was performed
by using a 17 min-linear gradient from 40% to 100% (v/v) of
solvent B at 300 nL/min. The time-of-flight mass spectrometry
(TOF-MS) survey scan was acquired for 1 s over a mass range
of 800−2000 m/z to cover the multiple charge states of the
intact proteins and 2 s for MS/MS range of 50−1000 m/z. An
information-dependent acquisition method was used with
maximum collision energy set to 50 V and with include ion
list corresponding to the 13+, 14+, 15+, and 16+ charge states
of prolyl-AnaD or its analogues. For these highly charge states,
a collision energy of 50 V was critical for the detection of the
pantetheine signature fragment ion at mass m/z 261.13 and the
dephosphorylated phosphopantetheinyl arm with proline (m/z
358.19) or dehydroproline (m/z 356.17) or analogues attached
to it. As already noted, intact prolyl-AnaD and analogues
showed minor covalent adducts at M + 178 (gluconoylation)
and at M + 97 (proline linked to the N-terminal amine).13,33
Figure 1. Reduction by dithionite and reoxidation by air of wild-type
and E244A AnaB. (A) Electronic absorption spectrum of 7.9 μM
recombinant wild-type AnaB before reduction by dithionite (solid
line), after reduction by 0.52 mM dithionite (dashed line), and after
reoxidation by air (dotted line). (B) Electronic absorption spectrum of
12.8 μM recombinant E244A AnaB before reduction by dithionite
(solid line), after reduction by 0.52 mM dithionite (dashed line), and
after reoxidation by air (dotted line). The spectra were recorded at
room temperature in 20 mM Tris-HCl buffer, pH 7.7, containing 10%
v/v glycerol.
RESULTS
■
Construction, Production, and Purification of E244A
AnaB. We reported that AnaB, which catalyzes the oxidation
Dithionite-Reduced Wild-Type and Mutant AnaB Are
Reoxidized by Air. Figure 1A shows the electronic
absorption spectrum of recombinant wild-type AnaB in the
visible region before addition of dithionite (solid line). This
absorption spectrum, in the 350−600 nm range, is character-
istic of FAD-containing enzymes. We used excess dithionite
(0.52 mM) to reduce the wild-type enzyme (7.9 μM), in a
sealed cuvette, and we obtained a characteristic bleaching of the
visible spectrum (dashed line). We observed a complete and
immediate reoxidation of the enzyme by simply opening the
sealed cuvette to air and by inverting the cuvette twice. The
spectrum of the reoxidized enzyme (dotted line) was identical
of prolyl-AnaD in the biosynthesis of anatoxin-a, is a member of
the ACAD family of enzymes, as evidenced by protein sequence
alignments,13 and that it is homologous to hIVD, an enzyme
involved in the degradation of fatty acids in humans, for which
a three-dimensional structure is available.34 We could thus
identify the putative active site base of AnaB, a glutamyl
residue, E244, which perfectly aligned with the known active
site base of hIVD, E254 (Figure S1). We have now constructed
a mutant, E244A AnaB, to study the role of this residue in
AnaB-catalyzed reactions. The mutant was constructed using a
PCR-based site-directed mutagenesis methodology, and the
7188
dx.doi.org/10.1021/bi200892a|Biochemistry 2011, 50, 7184−7197

Biochemistry
Article
to that of the native enzyme. Thus, AnaB is a flavoprotein that
probably uses oxygen as the second substrate, and this enzyme
is likely an oxidase rather than a dehydrogenase. We observed
the same reduction/reoxidation behavior with E244A AnaB,
confirming the integrity of this mutant enzyme (Figure 1B).
Identification of the AnaB-Oxidized Product by HPLC
and LC-MS/MS. In our previous study we reported that AnaB
catalyzed the two-electron oxidation of prolyl-AnaD, but we
were unable to directly observe, by HPLC, the oxidation
product because prolyl-AnaD and its oxidation product are
unstable due to rapid hydrolysis of their thioester bond,
yielding to holo-AnaD.13 To facilitate the observation of the
oxidized product, we have now optimized the preparation of
prolyl-AnaD and its oxidation by using short incubation times
and high concentrations of AnaB. Prolyl-AnaD was thus
prepared and immediately incubated, without further purifica-
tion, in the presence of recombinant wild-type AnaB: 60 μM
prolyl-AnaD was oxidized in the presence of 10 μM AnaB for 5
min. Because we showed that AnaB did not oxidize free proline
and that AnaC did not accept free P5C or pyrrole-2-carboxylate
as substrates,13 any oxidation reaction observed must occur on
proline tethered to AnaD. The reaction was analyzed by HPLC,
from 0.5 to 7.5 min incubation times (Figure 2A). A new peak
at 18.9 min was clearly visible on the HPLC chromatogram,
and its intensity increased with incubation time. At longer
incubation times, over 7 min, holo-AnaD appeared as a
shoulder of the new peak at 18.9 min and became the major
product after 10 min incubation time. Although quantification
of the product was not very accurate due to overlapping of the
peaks, the relative concentration of the product over that of the
substrate increased linearly against time, up to 5 min and 50%
transformation (Figure 2B). From these data an apparent rate
Figure 2. HPLC analysis of the AnaB-catalyzed oxidation of prolyl-
AnaD. A solution of prolyl-AnaD (60 μM) was incubated in the
absence or in the presence of 10 μM AnaB at 28 °C. The reaction was
stopped at different time points, and 20 μL aliquots were analyzed by
HPLC. (A) From bottom to top: chromatogram of the reaction in the
absence of AnaB, and portion of the chromatograms of the reactions
run in the presence of 10 μM wild-type recombinant AnaB at 0.5, 2.5,
5.0, and 7.5 min incubation time. The oxidation product is marked by
an arrow. AnaB eluted at 15.5 min but is not shown here for clarity.
The absorption in the void volume (4 min) is mainly due to the ATP
used in the loading of proline on holo-AnaD. (B) The relative amount
of product was plotted against incubation time. The first data points
were fitted to a straight line (slope =11% per min, r = 0.99). (C) A
control experiment was run in the presence of 10 μM E244A
recombinant AnaB (7.5 min incubation time, 20 μL injected). E244A
AnaB eluted at 15.5 min (marked with a triangle).
and an apparent catalytic constant could be estimated: Vapp
=
11 μM/min and kcat;app = 1 min−1. Thus, AnaB catalyzed the
reaction with multiple turnovers and uses oxygen as the second
substrate, as suggested above. We did not detect the formation
of any new peak in control experiments without AnaB or in the
presence of E244A AnaB, as shown in Figure 2A and Figure
2C, respectively. Thus, the E244A AnaB mutant was inactive, as
expected.
AnaD, that could be formed after two successive AnaB-
catalyzed oxidation reactions. Two control experiments, one
without AnaB and the other one in the presence of 10 μM
E244A AnaB, were run (Figures S4 and S5), and in both cases
we did not observe any oxidation product, as we observed using
classical HPLC (Figure 2).
We then unambiguously identified, by LC-MS/MS, the two
species detected by classical HPLC. Figure 3A shows the LC-
MS chromatogram, as the total ion current, of the oxidation
reaction after 5 min incubation time in the presence of 10 μM
wild-type AnaB, revealing two overlapping peaks. Analysis of
the ESI-MS spectra at 15.7 and 16.1 min (Figure S31) gave the
following mass: 14 246.6 and 14 244.8 Da, respectively, which
corresponded to the calculated average mass for prolyl-AnaD
(14 247.2 Da) and dehydroprolyl-AnaD (14 245.2 Da). The
MS/MS spectrum of the 16+ charge states at 15.7 and 16.1 min
elution times are shown in Figures 3B and 3C, respectively. In
these MS/MS spectra, the ion corresponding to the
dephosphorylated phosphopantetheine arm acylated with
proline, m/z 358.18 (Figure 3B), or dehydroproline, m/z
356.17 (Figure 3C), are clearly visible. This facile fragmentation
allowing direct identification of the phosphopantetheine arm,
acylated or not, has been coined the “Ppant ejection assay” by
its inventors.35−37 Moreover, the typical m/z 261.13 ion, which
is formed after loss of the acyl moiety of the thioester of the
dephosphorylated arm, is also clearly visible in the MS/MS
spectra. We did not observe, as we previously noted,13 the
putative four-electron oxidation product, pyrrole-2-carboxyl-
Synthesis of Deuterium-Labeled Proline and the
Corresponding Prolyl-AnaD. In our previous study, we
proposed that the oxidation reaction catalyzed by AnaB took
place in two steps: first, formation of the conjugated double
bond (C2N) with hydride transfer to the flavin and, then, an
aza−allylic isomerization to form the C5N double bond.13
Indeed, the proposed AnaB product, in the biosynthesis of
anatoxin-a and homoanatoxin-a, is P5C-AnaD.9,13 To challenge
this proposition, we have now prepared three deuterium-
labeled analogues of proline as depicted in Scheme 2. The
reaction sequences are based on published procedures or are
adaptation of these procedures.29−32 Labeling at position C2
was achieved by racemization of proline in the presence of PLP
in D2O.29,30 We obtained rac-[2-2H]-proline 1, containing 97%
deuterium, as determined by 1H NMR, only after three rounds
of this step. The doubly labeled [5,5-2H2]-L-proline 5 was
7189
dx.doi.org/10.1021/bi200892a|Biochemistry 2011, 50, 7184−7197

Biochemistry
Article
Scheme 2. Outline of the Synthesis of Deuterium Labeled
a
Prolines
a
Keys: a: NaOD, PLP, D2O, purification on ion-exchange resin. This
process was repeated three times. b: HOBT, DCC, piperidine, THF. c:
DMAP, Boc2O, CH3CN. d: NaBD4, I2, THF. e: TFA, CH2Cl2, then
neutralization with NaOH and purification on ion-exchange resin. f:
tert-Butyl acetate, HClO4. g: DMAP, Boc2O, THF. h: DIBAL-H, THF.
i: Bu3SnD, toluene. j: TFA, CH2Cl2, then 2 M HCl.
C5-pro-R position, as determined by careful examination of the
1H NMR spectrum of compound 10. Thus, compound 10 is a
mixture of 82% of (5R)-[5-2H]-L-proline and 14% of (5S)-[5-
2H]-L-proline and 4% of unlabeled proline. We also used the
perdeuterated proline, [2,3,3,4,4,5,5-2H7]-L-proline, containing
97−98% deuterium, obtained from commercial source. All
deuterated prolines were substrate of AnaC, the adenylation
protein that transforms holo-AnaD into prolyl-AnaD,9,13 and
we verified by HPLC that the reactions were all complete.
Because we showed in our previous report that D-proline is not
a substrate for AnaC,13 we obtained [2-2H]-L-prolyl-AnaD
using compound 1, which is a racemate. All these deuterium-
labeled prolyl-AnaD were then directly used in the AnaB-
catalyzed oxidation reaction, as described above for unlabeled
prolyl-AnaD.
LC-MS/MS Analysis of the AnaB-Catalyzed Oxidation
Reaction Using Deuterium-Labeled Prolyl-AnaDs. The
oxidation reactions, and control experiments either in the
absence of AnaB or in the presence of E244A AnaB, were
analyzed by LC-MS/MS. We could measure the mass of the
substrates and of the products (Table 1) and also detect, in the
MS/MS spectra, the pantetheine thioester ion ejected from the
substrates or from the products and analyze its isotopic
distribution (Table 2 and Figure S17). The substrates were all
labeled according to the deuterium labeling of the proline used
to prepare them, and the observed isotopic distributions of the
acylated pantetheine ion matched that of calculated ones
(Figure S17).
Figure 3. LC-MS/MS analysis of the AnaB-catalyzed oxidation of
prolyl-AnaD. The oxidation reaction mixture containing 60 μM prolyl-
AnaD and 10 μM wild-type AnaB and incubated at 28 °C for 5 min
was analyzed by LC-MS/MS. (A) The total ion current chromatogram
is shown. AnaB was not observed under these separation conditions
because it was not retained on the precolumn. (B) MS/MS spectrum
at 15.79 min of the 16+ charge state of the substrate. (C) MS/MS
spectrum at 16.11 min of the 16+ charge state of the product. (D)
Structure and calculated mass for the detected ions.
We observed a partial loss (∼50%) of the deuterium labeling
on position C2 of the remaining substrate, at ∼50%
transformation, in the presence of wild-type AnaB (Table 2,
entry 4). This deuterium loss was absent in the untreated
substrate (Table 2, entry 2) and in the substrate that was
treated with the E244A mutant (Table 2, entry 3).
Furthermore, we also observed this partial loss, at the substrate
level, when we used [2H7]-L-prolyl-AnaD (Table 2, entries 7−
9). Indeed, in the absence of AnaB or in the presence of E244A
AnaB (Table 2, entries 7 and 8), we observed the expected ion
m/z 365.22 and the ion m/z 364.22, as a minor species, arising
from of all species containing six deuteriums (Figure S17). This
isotopic distribution was reversed in the remaining substrate
when treated with wild-type AnaB (Table 2, entry 9). This loss
obtained using an interesting regioselective reduction of Boc-L-
pyroglutamic piperidine amide 3 by NaBD4 in the presence of
iodine. This reaction has been applied to diverse amides and
lactams,31 but we have used it for the first time on Boc-
pyroglutamic amide. The overall deuterium content of
compound 5 was 91% with an equal labeling of both C5
positions, as determined by 1H NMR. We used a stereo-
selective reduction of hemiaminal 8, the tert-butyl ester of Boc-
L-glutamate semialdehyde, by Bu3SnD containing 96%
deuterium, to specifically labeled the C5-pro-S position of
proline.32 However, the reduction was not fully stereoselective
and gave 85% labeling at the C5-pro-S position and 15% at the
7190
dx.doi.org/10.1021/bi200892a|Biochemistry 2011, 50, 7184−7197

Biochemistry
Article
digest. However, we considered the data presented in Table 2
more reliable because nonspecific deuterium exchanges could
occur during the trypsin digestions.
LC-MS/MS Analysis of the AnaB-Catalyzed Oxidation
Reaction Using Analogues of Prolyl-AnaD. In our
Table 1. Summary of the Calculated and Observed Mass for
the Different AnaD Forms Characterized in This Report
calcd av
obsd av mass Figure in Supporting
a
protein
holo-AnaD
mass (Da)
(Da)
Information
b
c
c
c
14 150.12
14 247.28
14 245.22
14 248.25
14 150.2
14 246.6
14 244.8
14 247.2
previous report we showed that AnaC, the adenylation protein,
was very specific for L-proline but that it accepted some close
structural analogues of proline as substrates.13 We took
advantage of this narrow substrate promiscuity to load onto
holo-AnaD, 3,4-dehydro-L-proline and the two diaster-
eoisomers of 4-fluoro-L-proline. We obtained the three
corresponding prolyl-AnaD analogues, that were characterized
by HPLC and LC-MS/MS (Tables 1 and 3). These prolyl-
AnaD analogues were used in typical AnaB-catalyzed oxidation
reactions and analyzed as described for prolyl-AnaD (Table 3).
When 3,4-dehydro-L-prolyl-AnaD was oxidized by AnaB we
observed on the LC-MS chromatogram (Figure S14) a large
transformation of the substrate, suggesting that 3,4-dehydro-L-
prolyl-AnaD is a very good substrate for AnaB. As expected, we
observed, on the MS/MS spectrum of the product, the m/z
354.16 ion, which is most likely the daughter ion arising from
pyrrole-2-carboxyl-AnaD (Table 3, entry 1). (4S)- and (4R)-4-
fluoro-L-prolyl-AnaD, neatly detected by LC-MS/MS (m/z
376.18), were substrates of AnaB and gave, as the only
detectable product, pyrrole-2-carboxyl-AnaD (m/z 354.16,
Table 3, entries 2 and 3). However, the (4S)-4-fluoro-L-
prolyl-AnaD (cis-isomer) was a poor substrate compared to the
trans-isomer (Figures S15 and S16). Thus, both diaster-
eoisomers were oxidized by AnaB and experienced rapid
elimination of HF because we could not detect the
intermediate, 4-fluoro-dehydroprolyl-AnaD. This elimination
is most likely explained by formation of a conjugated carbanion
on position C5, during the AnaB-catalyzed aza−allylic isomer-
ization, as discussed below. The E244A AnaB mutant was
inactive toward these fluoro- or dehydroprolyl-AnaD substrates,
as determined in control experiments analyzed by HPLC.
prolyl-AnaD
S3
dehydroprolyl-AnaD
[2-2H]-L-prolyl-
AnaD
S3
S6, S7, S8
c
c
c
[5,5-2H2]-L-prolyl-
AnaD
(5S)-[5-2H]-L-prolyl-
AnaD
14 249.25
14 248.25
14 246.23
14 248.3
14 247.1
14 245.4
S9
S10
[5-2H]-
S9, S10
dehydroprolyl-
AnaD
[2,3,3,4,4,5,5-2H7]-L-
prolyl-AnaD
[2H5]-dehydroprolyl-
AnaD
c
14 254.28
14 250.25
14 245.22
14 265.23
14 265.23
14 243.21
14 253.5
14 250.0
14 244.7
14 264.4
14 264.6
14 242.7
S11, S12
S13
c d
;
c
c
c
c
3,4-dehydro-L-prolyl-
AnaD
S14
(4S)-4-fluoro-L-
prolyl-AnaD
S15
(4R)-4-fluoro-L-
prolyl-AnaD
S16
pyrrole-2-carboxyl-
AnaD
S14−S16
a
The calculated masses correspond to the recombinant protein lacking
the N-terminal Met1. Taken from our previous reports.9,13 A minor
species corresponding to the formation of an amide, probably on the
N-terminal amine, with the proline or proline analogue used during
the AnaC-catalyzed loading on holo-AnaD, was also observed. The
b
c
minor gluconoylated M+178 species was always present.33 This
d
sample is in fact a mixture of several isotopologues.
of labeling at position C2 is likely due to a rapid enolization
catalyzed by AnaB, as discussed below. The experiment using
[2H7]-L-prolyl-AnaD and the inactive E244A AnaB mutant
(Table 2, entry 8) showed that no deuterium loss was observed
on any position of the substrate, thus serving as a control
experiment for all other experiments described in Table 2.
DISCUSSION
■
On the basis of sequence alignments, we showed that AnaB
belongs to the ACAD superfamily.13 Using more specific
alignments, we showed that AnaB and the prolyl-ACP
dehydrogenases that are involved in the pyrrole ring formation
of secondary metabolites, such as pyoluteorin,14,19 prodigi-
nine,16 clorobiocin,17,20 prodigiosin,18,21 or leupyrrin,26 for
instance, have all a strictly conserved glutamyl residue that
corresponds to the hIVD active site base (E254). In contrast,
the active site base of the pig medium chain acyl-CoA
dehydrogenase (MCAD), E376, is not conserved in AnaB.
We have mutated the conserved glutamyl residue in AnaB into
an alanyl residue and showed that this E244A AnaB mutant
contains a FAD cofactor and that it retains the flavin redox
properties but is completely inactive toward prolyl-AnaD.
Hence, this residue is probably the AnaB active site base.
However, after reduction by dithionite, AnaB is completely and
very rapidly reoxidized by molecular oxygen. This is in sharp
contrast to the very slow reactivity of ACAD with molecular
oxygen.38,39 Thus, AnaB, and probably the other prolyl-ACP
dehydrogenases, are oxidases rather than dehydrogenases. The
structural basis for this different reactivity toward oxygen must
await the solution of the three-dimensional structure of AnaB.
Thus, AnaB and hIVD share sequence identity and strong
We clearly observed a complete loss of deuterium in the MS/
MS spectrum of the product obtained after AnaB-catalyzed
oxidation of [2-2H]-L-prolyl-AnaD (Table 2, entry 4). The
isotopic distribution of the ion ejected from the product
corresponded to an unlabeled ion (m/z 356.17, Figure S17).
When the substrate, either doubly labeled on position C5 or
labeled on position C5-pro-S, was oxidized, we observed a
major ion at m/z 357.18 (Table 2, entries 5 and 6; see Scheme
3 for identification of the prochiral hydrogens of proline). This
is consistent with a stereospecific retention of deuterium on
position C5-pro-S after oxidation and a loss of the deuterium on
position C5-pro-R, as discussed below. The isotopic distribu-
tion of the ion detected in the MS/MS spectrum of the product
when using [2H7]-L-prolyl-AnaD was more complex than
expected (Table 2, entry 9). The m/z 358.18, 359.21, and
360.20 ions were observed in approximate equal intensity
together with the m/z 356.15, 357.19, and 361.18 ions as minor
species. These data revealed incomplete exchanges on other
positions than the C2 and C5-pro-R positions, as discussed
below. The data presented in Table 2 were confirmed using an
alternate strategy that we previously described,13 consisting in a
trypsin digestion of the substrate and product of AnaB-
catalyzed reactions, followed by LC-MS/MS analysis of the
7191
dx.doi.org/10.1021/bi200892a|Biochemistry 2011, 50, 7184−7197

Biochemistry
Article
Table 2. Summary of the LC-MS/MS Analysis of the AnaB-Catalyzed Oxidation Reaction Using the Deuterium-Labeled Prolyl-
a
AnaD as Substrates
a
Footnotes: a: P represents the holo-AnaD protein. b: The isotopic distributions of the detected ions are shown in Figure S17 and were compared to
simulated distributions. The relative intensities are only given here as a qualitative estimate: maj: major ion detected; min: minor ion detected; iso:
ions of approximate equal abundance. c: ND: not detected. d: In this experiment we used 5 μM wild-type AnaB rather than 10 μM to reduce the
substrate transformation for an easier LC-MS/MS analysis of the remaining substrate. e: As indicated by the manufacturer. f: This ion arose from the
molecules containing six deuterium rather than seven. See Figure S17 for further interpretation of the isotopic distributions.
mechanistic similarities, as discussed below, but differ in their
reactivity toward molecular oxygen.
Scheme 3. Proposed Reaction Mechanism for the Oxidation
Reaction of Prolyl-AnaD Catalyzed by AnaB
a
By using controlled incubation conditions, we have been able
to observe, by HPLC and LC-MS/MS, the AnaB product,
dehydroprolyl-AnaD. We could also observe multiple turnovers
by AnaB, but unfortunately, due to intrinsic instability of the
substrate and low sensitivity of the detection of the product by
HPLC coupled to a UV detector, we have not yet been able to
conduct thorough kinetic analysis of the AnaB-catalyzed
reaction but estimated an apparent catalytic constant of 1
min−1. This value is much lower than that reported for
MCAD,38 for instance, and it is not clear at the moment why
AnaB catalysis is so slow. It might be possible that AnaB
efficiency depends on some specific protein−protein inter-
actions with the PKS AnaE, to which AnaB product is thought
to be transferred.
To study in detail the reaction mechanism of AnaB, we have
prepared several deuterium-labeled prolines and their corre-
sponding prolyl-AnaD proteins. These proteins were substrates
of AnaB, and we could analyze by LC-MS/MS, thanks to the
so-called “Ppant ejection assay”,35−37 the fate of the deuterium
a
P represents the holo-AnaD protein.
7192
dx.doi.org/10.1021/bi200892a|Biochemistry 2011, 50, 7184−7197

Biochemistry
Article
Table 3. Summary of the LC-MS/MS Analysis of the AnaB-Catalyzed Oxidation Reaction Using 3,4-Dehydro-L-Prolyl-AnaD and
a
cis- and trans-4-Fluoro-L-Prolyl-AnaD
a
P represents the holo-AnaD protein.
labeling in the substrates and products. First, we conclusively
observed a partial loss of the C2-D deuteron, and hence the
C2−H proton in the unlabeled substrate, on prolyl-AnaD in the
presence of wild-type AnaB. This is most consistent by
assuming that AnaB catalyzes a fast enolization, as depicted
in Scheme 3, and that its protonated active site base (E244)
exchanges rapidly its proton with the solvent. Observation of
this exchange suggests that all physical and chemical steps prior
to hydride transfer to the FAD cofactor are faster than the
redox chemical step. This, again, is in contrast to the situation
in ACAD, which do not show, during turnover of good
substrates, such exchange on the C2−H proton of the acyl-CoA
substrate.38
Second, we observed, in the two electron-oxidized product, a
complete loss of the labeling at positions C2, as revealed by the
isotopic distribution of the ejected ion of the product (Figure
S17). This is consistent with the formation of the P2C-AnaD
imine as depicted in Scheme 3. Third, we observed a loss of
labeling at position C5 that could, in principle, be explained by
either an exchange of the C5-deuteron with solvent proton on
the P2C-AnaD imine or a loss of deuteron during the P2C-
AnaD to P5C-AnaD isomerization. By using a doubly labeled
substrate on C5 with symmetrical labeling on that position and
a singly labeled substrate, which is mainly labeled on the C5-
pro-S position, we could discriminate among these possibilities.
Figure 4 shows the predicted isotopic distribution of the ion
ejected from the product, for all these possibilities, together
with the experimental isotopic patterns obtained using both
substrates. Clearly, the experimental data fitted very well to the
prediction for a stereospecific loss of the C5-pro-R deuteron.
However, these experiments cannot discriminate between a
stereospecific loss of the C5-pro-R deuteron to form P5C-AnaD
and a complete stereospecific exchange of the same deuteron
on P2C-AnaD. In other words, we cannot at the moment
obtain any information on the position of the equilibrium
between P2C-AnaD and P5C-AnaD. Attempts to quench the
product with specific reagents (NaBD4, 2-aminobenzaldehyde,
or 2,4-dinitrophenylhydrazine) for further spectroscopic
analysis have so far been unsuccessful.
are consistent with the proposed AnaB reaction mechanism
involving formation of an imine at position C2N in the first
step, followed by an aza−allylic isomerization in the second
step (Scheme 3). This isomerization is very likely catalyzed by
the same active site base, E244, because the two deuterons or
protons that are lost or exchanged are on the same side of the
5-membered cycle of the substrate, intermediate or product
(the Re−Re prochiral face of the imines). There are precedents
for this catalyzed isomerization in ACAD and homologous
enzymes. Indeed, this isomerase activity has been observed in
MCAD on various substrates or inhibitors.40−42 Furthermore,
Liu and colleagues suggested that this isomerization, observed
in ACAD, was catalyzed by one base, the actual glutamyl active
site base.40,41 Interestingly, in the biosynthesis of the antibiotic
friulimicin in Actinoplanes friuliensis, a flavoprotein, LipB,
homologous to ACAD, has been identified, and it was shown
that it catalyzed the formation of the Δcis3 double bond (C3
C4 Z-double bond) of the acyl-ACP starter.43,44 Thus, in this
ACAD, the aza−allylic isomerization is fully operative and
yields to the unconjugated less stable product.
The abstraction of the C5−H-pro-R was also substantiated by
observation of the HF elimination when cis- or trans-4-fluoro-L-
prolyl-AnaD was used as a substrate. The cis-isomer was a much
poorer substrate than the trans-isomer, but both gave, as unique
product, the pyrrole-2-carboxyl-AnaD. This is consistent with
the formation of a conjugated carbanion on position C5
followed by a rapid elimination of fluorine (Scheme 4). This
elimination is thus of the E1cB type although the relative rates
of both steps (proton abstraction and elimination) cannot at
this point be determined.
We also observed some exchange of deuterium at other
positions than C2 and C5-pro-R, in the AnaB-catalyzed
oxidation of the perdeuterated [2H7]-L-prolyl-AnaD, because
the isotopic distribution of the ejected ion showed the presence
of ions from m/z 356 to m/z 361 (Figure S17). Because the
deuterium on position C5-pro-S is retained, there are four
positions left for possible exchanges that are positions at C3
and C4. This is chemically sound because these exchanges
could occur during enamine−imine tautomerisms. We have
calculated the isotopic distribution of the ion ejected from the
product if these four different positions would independently
exchange 50% of their deuterium (at the same rate) with the
solvent protons (Figure S17). This oversimplified calculation
The fact that the loss of deuterium at position C5 is
stereospecific strongly suggests that we are observing an
enzyme-catalyzed reaction rather than a nonspecific exchange
occurring in solution after release of the product. These data
7193
dx.doi.org/10.1021/bi200892a|Biochemistry 2011, 50, 7184−7197

Biochemistry
Article
Scheme 4. Proposed Reaction Mechanism for AnaB-
Catalyzed Half-Reaction with 3,4-Dehydro-L-Prolyl-AnaD
a
and cis- or trans-4-Fluoro-L-Prolyl-AnaD
Figure 4. Simulated and experimental isotopic distributions of the
acylated pantetheine ion ejected from the product after AnaB-
catalyzed oxidation of [5,5-2H2]-L-prolyl-AnaD (A−C) and (5S)-[5-
2H1]-L-prolyl-AnaD (D−H). (A) Simulated isotopic distribution of
the ion from the oxidized product of [5,5-2H2]-L-prolyl-AnaD,
assuming a nonstereoselective partial exchange (50% on each position
and independently) of deuterium with proton at position C5 of the
P2C-AnaD imine. (B) Simulated isotopic distribution of the ion from
the oxidized product of [5,5-2H2]-L-prolyl-AnaD, assuming a stereo-
selective loss (at C5-pro-R or C5-pro-S) or nonstereoselective loss of
deuterium at position C5, to form P5C-AnaD. The same pattern is
obtained for these three possibilities because the labeling is
symmetrical in this substrate. (C) Experimental isotopic distribution
of the ion from the AnaB oxidized product of [5,5-2H2]-L-prolyl-
AnaD. (D) Simulated isotopic distribution of the ion from the oxidized
product of (5S)-[5-2H1]-L-prolyl-AnaD, assuming a nonstereoselective
exchange (50% on each position and independently) of deuterium
with proton at position C5 of the P2C-AnaD imine. (E) Simulated
isotopic distribution of the ion from the oxidized product of (5S)-[5-
2H1]-L-prolyl-AnaD, assuming a nonstereoselective loss of deuterium
at position C5, to form P5C-AnaD. (F) Simulated isotopic distribution
of the ion from the oxidized product of (5S)-[5-2H1]-L-prolyl-AnaD,
assuming a stereoselective loss of deuterium at position C5-pro-S, to
form P5C-AnaD. (G) Simulated isotopic distribution of the ion from
the oxidized product of (5S)-[5-2H1]-L-prolyl-AnaD, assuming a
stereoselective loss of deuterium at position C5-pro-R, to form P5C-
AnaD. (H) Experimental isotopic distribution of the ion from the
AnaB oxidized product of (5S)-[5-2H1]-L-prolyl-AnaD. The methods
used to simulate the isotopic distributions are described in Figure S17.
a
P represents the holo-AnaD protein.
Scheme 5. Proposed Mechanism for the Observed Exchange
of Deuterium at Positions C3 and C4 of [2H7]-L-Prolyl-
AnaD
a
a
P represents the holo-AnaD protein.
shows that the major species observed will be the D2, D3, and
D4 species (at 65%, 100%, and 75% normalized relative
intensities). We observed ions at m/z 358, 359, and 360 with
approximately the same abundance in the MS/MS spectrum of
the ion ejected form the oxidation product of [2H7]-L-prolyl-
AnaD (Table 2 and Figure S17). Thus, it is reasonable to
propose that the C3 and C4 positions, in the P5C-AnaD and
P2C-AnaD imines could experience exchanges via the
enamine−imine tautomerism (Scheme 5). However, these
exchanges have to be nonstereospecific to explain the loss of
deuterium on four different positions. Thus, the E244 base
could abstract all deuteron/proton on the lower face of the
molecule, but the reprotonation on position C3 and C4 should
be nonstereospecific. Perhaps a water molecule present in the
active site or another base could catalyze these reprotonations
on the upper face. There are indeed water molecules present in
the active site of ACAD, although binding of the acyl-CoA
substrate expels some of them.38 We cannot exclude that the
observed exchanges could occur in solution after the product is
released from the active site of AnaB, but several studies have
7194
dx.doi.org/10.1021/bi200892a|Biochemistry 2011, 50, 7184−7197

Biochemistry
Article
shown that the isomerization of P5C into P2C and the
enamine−imine equilibrium in these imines are very slow at pH
7.45−47 We thus believe that all exchanges observed in this
report are catalyzed by AnaB and that the loss of deuteron/
proton on positions C2 and C5-pro-R must be in the reaction
path since they are complete while the other exchanges are not
part of the reaction path since they are not complete.
ABBREVIATIONS
■
ACAD, acyl-CoA dehydrogenase; ACP, acyl carrier protein;
Boc, tert-butyloxycarbonyl; Boc2O, di-tert-butyl dicarbonate;
DCC, N,N′-dicyclohexylcarbodiimide; DIBAL-H, diisobutylalu-
minum hydride; DMAP, 4-(dimethylamino)pyridine; ESI-MS,
electrospray ionization mass spectrometry; hIVD, human
isovaleryl-CoA dehydrogenase; HOBT, 1-hydroxybenzotria-
zole; HPLC, high performance liquid chromatography; IPTG,
isopropyl-β-D-thiogalactopyranoside; LB, lysogeny broth; LC-
MS/MS, liquid chromatography coupled to tandem mass
spectrometry; MCAD, medium chain acyl-CoA dehydrogenase;
P2C, 1-pyrroline-2-carboxylic acid; P5C, (S)-1-pyrroline-5-
carboxylic acid; PCC, Pasteur culture collection of cyanobac-
teria; PKS, polyketide synthase; PLP, pyridoxal-5′-phosphate;
TCEP, tris(2-carboxyethyl)phosphine; TFA, trifluoroacetic
acid; THF, tetrahydrofuran; TOF-MS, time-of-fligth mass
spectrometry.
The fact that 3,4-dehydro-L-prolyl-AnaD is a very good
substrate for AnaB and that it gave pyrrole-2-carboxyl-AnaD
excludes this molecule as a possible intermediate in the reaction
path of AnaB. Walsh and co-workers had already observed that
the prolyl-ACP dehydrogenase, CloN3, also transformed 3,4-
dehydro-L-prolyl-ACP into pyrrole-2-carboxyl-ACP.20−22 They
interpreted that result by invoking that 3,4-dehydroprolyl-ACP
could be an intermediate in the reaction path of this four-
electron oxidation reaction, and this would be the mechanistic
difference between AnaB and CloN3. However, it could also be
possible that CloN3 utilizes the same reaction path than AnaB
but displaces the isomerizations toward the C4C5 enamine
that would be then oxidized in a second round to yield the
pyrrole aromatic ring. The difference between AnaB and CloN3
reaction paths would then be the concentration and lifetime in
the active site of the C4C5 enamine: low concentration and
short lifetime in AnaB versus high concentration and long
lifetime in CloN3. This interesting issue might only be solved
when we will have in hand the three-dimensional structure of
these enzymes to compare their active site structures and
detailed kinetic studies on these enzymes.
REFERENCES
■
(1) vanApeldoorn, M. E., van Egmond, H. P., Speijers, G. J., and
Bakker, G. J. (2007) Toxins of cyanobacteria. Mol. Nutr. Food Res. 51,
7−60.
(2) Edwards, C., Beattie, K., Scrimgeour, C., and Codd, G. (1992)
Identification of anatoxin-a in benthic cyanobacteria (blue-green algae)
and in associated dog poisoning at Loch Insh, Scotland. Toxicon 30,
1165−1175.
(3) Gugger, M., Lenoir, S., Berger, C., Ledreux, A., Druart, J. C.,
Humbert, J. F., Guette, C., and Bernard, C. (2005) First report in a
river in France of the benthic cyanobacterium Phormidium favosum
producing anatoxin-a associated with dog neurotoxicosis. Toxicon 45,
919−928.
In conclusion, we report here evidence that support the
proposed reaction mechanism for AnaB, involving the rapid
formation of an enolate followed by hydride transfer to the
flavin, to form the P2C-AnaD imine, which in turn is
isomerized to P5C-AnaD, the product. We have identified the
active site base, glutamyl residue E244, which probably serves
as a base during enolization and isomerization.
(4) Cadel-Six, S., Peyraud-Thomas, C., Brient, L., Tandeau de
Marsac, N., Rippka, R., and Mejean, A. (2007) Different genotypes of
́
anatoxin-producing cyanobacteria coexist in the Tarn River, France.
Appl. Environ. Microbiol. 73, 7605−7614.
(5) Wood, S. A., Selwood, A. I., Rueckert, A., Holland, P. T., Milne, J.
R., Smith, K. F., Smits, B., Watts, L. F., and Cary, C. S. (2007) First
report of homoanatoxin-a and associated dog neurotoxicosis in New
Zealand. Toxicon 50, 292−301.
(6) Puschner, B., Hoff, B., and Tor, E. R. (2008) Diagnosis of
anatoxin-a poisoning in dogs from North America. J. Vet. Diagn. Invest.
20, 89−92.
(7) Puschner, B., Pratt, C., and Tor, E. R. (2010) Treatment and
diagnosis of a dog with fulminant neurological deterioration due to
anatoxin-a intoxication. J. Vet. Emerg. Crit. Care (San Antonio) 20,
518−522.
ASSOCIATED CONTENT
■
S
* Supporting Information
Figure S1: sequence alignments of AnaB and hIVD; Figure S2:
gel electrophoresis of AnaB proteins; Figures S3−S16: LC-MS/
MS analysis of the oxidation reactions; Figure S17: analysis of
the isotopic distribution of the detected ions for deuterium-
labeled molecules. This material is available free of charge via
the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
(8) Wonnacott, S., and Gallagher, T. (2006) The chemistry and
pharmacology of anatoxin-a and related homotropanes with respect to
nicotinic acetylcholine receptors. Mar. Drugs. 4, 228−254.
■
Corresponding Author
*Tel: +(33) 1 44 27 67 01. E-mail: olivier-ploux@enscp.fr.
́
(9) Mejean, A., Mann, S., Maldiney, T., Vassiliadis, G., Lequin, O.,
Funding
and Ploux, O. (2009) Evidence that biosynthesis of the neurotoxic
alkaloids anatoxin-a and homoanatoxin-a in the cyanobacterium
Oscillatoria PCC 6506 occurs on a modular polyketide synthase
initiated by L-proline. J. Am. Chem. Soc. 131, 7512−7513.
This work was supported by the CNRS and the ENSCP. The
Laboratory of Proteomic Mass Spectrometry is supported by
grants from “Canceropôle Ile-de-France” and from the “Institut
́
National du Cancer, INCa”.
(10) Cadel-Six, S., Iteman, I., Peyraud-Thomas, C., Mann, S., Ploux,
O., and Mejean, A. (2009) Identification of a polyketide synthase
́
coding sequence specific for anatoxin-a-producing Oscillatoria
cyanobacteria. Appl. Environ. Microbiol. 75, 4909−4912.
(11) Wood, S. A., Heath, M. W., Kuhajek, J., and Ryan, K. G. (2010)
Fine-scale spatial variability in anatoxin-a and homoanatoxin-a
concentrations in benthic cyanobacterial mats: implication for
monitoring and management. J. Appl. Microbiol. 109, 2011−2018.
ACKNOWLEDGMENTS
■
We thank Santebala Sivacoumarane for technical help during
the preparation of prolyl-AnaD and its analogues as well as
during the HPLC analysis of the AnaB-catalyzed oxidation
reactions.
7195
dx.doi.org/10.1021/bi200892a|Biochemistry 2011, 50, 7184−7197

Biochemistry
Article
(12) Ballot, A., Fastner, J., Lentz, M., and Wiedner, C. (2010) First
report of anatoxin-a-producing cyanobacterium Aphanizomenon
issatschenkoi in northeastern Germany. Toxicon 56, 964−971.
(29) Fujihara, H., and Schowen, R. L. (1984) Facile, economical
synthesis of L-[alpha-2H]-alpha-amino acids. J. Org. Chem. 49, 2819−
2820.
(13) Mej
́
ean, A., Mann, S., Vassiliadis, G., Lombard, B., Loew, D.,
(30) Hamilton, P. B. (1952) Proline: synthesis from ornithine,
citrulline, or arginine. J. Biol. Chem. 198, 587−597.
and Ploux, O. (2010) In vitro reconstitution of the first steps of
anatoxin-a biosynthesis in Oscillatoria PCC 6506: from free L-proline
to acyl carrier protein bound dehydroproline. Biochemistry 49, 103−
113.
(14) Nowak-Thompson, B., Chaney, N., Wing, J. S., Gould, S. J., and
Loper, J. E. (1999) Characterization of the pyoluteorin biosynthetic
gene cluster of Pseudomonas fluorescens Pf-5. J. Bacteriol. 181, 2166−
2174.
(31) Prasad, A. S. B., Kanth, J. V. B., and Periasamy, M. (1992)
Convenient methods for the reduction of amides, nitriles, carboxylic
esters, acids and hydroboration of alkenes using NaBH4/I2 system.
Tetrahedron 48, 4623−4628.
(32) Oba, M., Terauchi, T., Hashimoto, J., Tanaka, T., and
Nishiyama, K. (1997) Stereoselective synthesis of (2S,3S,4R,5S)-
proline-3,4,5-d3. Tetrahedron Lett. 38, 5515−5518.
(15) Wang, Z. X., Li, S. M., and Heide, L. (2000) Identification of the
coumermycin A(1) biosynthetic gene cluster of Streptomyces rishiriensis
DSM 40489. Antimicrob. Agents Chemother. 44, 3040−3048.
(33) Geoghegan, K. F., Dixon, H. B., Rosner, P. J., Hoth, L. R.,
Lanzetti, A. J., Borzilleri, K. A., Marr, E. S., Pezzullo, L. H., Martin, L.
B., LeMotte, P. K., McColl, A. S., Kamath, A. V., and Stroh, J. G.
(1999) Spontaneous alpha-N-6-phosphogluconoylation of a “His tag”
in Escherichia coli: the cause of extra mass of 258 or 178 Da in fusion
proteins. Anal. Biochem. 267, 169−184.
(34) Tiffany, K. A., Roberts, D. L., Wang, M., Paschke, R., Mohsen,
A. W., Vockley, J., and Kim, J. J. (1997) Structure of human isovaleryl-
CoA dehydrogenase at 2.6 Å resolution: structural basis for substrate
specificity. Biochemistry 36, 8455−8464.
(35) Dorrestein, P. C., Bumpus, S. B., Calderone, C. T., Garneau-
Tsodikova, S., Aron, Z. D., Straight, P. D., Kolter, R., Walsh, C. T., and
Kelleher, N. L. (2006) Facile detection of acyl and peptidyl
intermediates on thiotemplate carrier domains via phosphopante-
theinyl elimination reactions during tandem mass spectrometry.
Biochemistry 45, 12756−12766.
(36) Dorrestein, P. C., and Kelleher, N. L. (2006) Dissecting non-
ribosomal and polyketide biosynthetic machineries using electrospray
ionization Fourier-Transform mass spectrometry. Nat. Prod. Rep. 23,
893−918.
(37) Meluzzi, D., Zheng, W. H., Hensler, M., Nizet, V., and
Dorrestein, P. C. (2008) Top-down mass spectrometry on low-
resolution instruments: characterization of phosphopantetheinylated
carrier domains in polyketide and non-ribosomal biosynthetic
pathways. Bioorg. Med. Chem. Lett. 18, 3107−3111.
(38) Ghisla, S., and Thorpe, C. (2004) Acyl-CoA dehydrogenases. A
mechanistic overview. Eur. J. Biochem. 271, 494−508.
(39) Kim, J. J., and Miura, R. (2004) Acyl-CoA dehydrogenases and
acyl-CoA oxidases. Structural basis for mechanistic similarities and
differences. Eur. J. Biochem. 271, 483−493.
(40) Dakoji, S., Shin, I., Becker, D. F., Stankovich, M. T., and Liu, H-
w. (1996) Studies of Acyl-CoA Dehydrogenase Catalyzed Allylic
Isomerization: A One-Base or Two-Base Mechanism? J. Am. Chem.
Soc. 118, 10971−10979.
(41) Dakoji, S., Shin, I., Battaile, K. P., Vockley, J., and Liu, H. W.
(1997) Redesigning the active-site of an acyl-CoA dehydrogenase: new
evidence supporting a one-base mechanism. Bioorg. Med. Chem. 5,
2157−2164.
(16) Cerdeno, A. M., Bibb, M. J., and Challis, G. L. (2001) Analysis
̃
of the prodiginine biosynthesis gene cluster of Streptomyces coelicolor
A3(2): new mechanisms for chain initiation and termination in
modular multienzymes. Chem. Biol. 8, 817−829.
(17) Pojer, F., Li, S. M., and Heide, L. (2002) Molecular cloning and
sequence analysis of the clorobiocin biosynthetic gene cluster: new
insights into the biosynthesis of aminocoumarin antibiotics. Micro-
biology 148, 3901−3911.
(18) Harris, A. K., Williamson, N. R., Slater, H., Cox, A., Abbasi, S.,
Foulds, I., Simonsen, H. T., Leeper, F. J., and Salmond, G. P. (2004)
The Serratia gene cluster encoding biosynthesis of the red antibiotic,
prodigiosin, shows species- and strain-dependent genome context
variation. Microbiology 150, 3547−3560.
(19) Thomas, M. G., Burkart, M. D., and Walsh, C. T. (2002)
Conversion of L-proline to pyrrolyl-2-carboxyl-S-PCP during un-
decylprodigiosin and pyoluteorin biosynthesis. Chem. Biol. 9, 171−84.
(20) Garneau, S., Dorrestein, P. C., Kelleher, N. L., and Walsh, C. T.
(2005) Characterization of the formation of the pyrrole moiety during
clorobiocin and coumermycin A1 biosynthesis. Biochemistry 44, 2770−
2780.
(21) Garneau-Tsodikova, S., Dorrestein, P. C., Kelleher, N. L., and
Walsh, C. T. (2006) Protein assembly line components in prodigiosin
biosynthesis: characterization of PigA,G,H,I,J. J. Am. Chem. Soc. 128,
12600−12601.
(22) Walsh, C. T., Garneau-Tsodikova, S., and Howard-Jones, A. R.
(2006) Biological formation of pyrroles: nature’s logic and enzymatic
machinery. Nat. Prod. Rep. 23, 517−531.
(23) Zhang, X., and Parry, R. J. (2007) Cloning and characterization
of the pyrrolomycin biosynthetic gene clusters from Actinosporangium
vitaminophilum ATCC 31673 and Streptomyces sp. strain UC 11065.
Antimicrob. Agents Chemother. 51, 946−957.
(24) Meiser, P., Weissman, K. J., Bode, H. B., Krug, D., Dickschat, J.
S., Sandmann, A., and Muller, R. (2008) DKxanthene biosynthesis:
̈
understanding the basis for diversity-oriented synthesis in myxobacte-
rial secondary metabolism. Chem. Biol. 15, 771−781.
(42) Zeng, J., and Li, D. (2005) Intrinsic isomerase activity of
medium-chain acyl-CoA dehydrogenase. Biochemistry 44, 6715−6722.
(25) Li, C., Roege, K. E., and Kelly, W. L. (2009) Analysis of the
indanomycin biosynthetic gene cluster from Streptomyces antibioticus
NRRL 8167. ChemBioChem 10, 1064−1072.
(26) Kopp, M., Irschik, H., Gemperlein, K., Buntin, K., Meiser, P.,
Weissman, K. J., Bode, H. B., and Muller, R. (2011) Insights into the
complex biosynthesis of the leupyrrins in Sorangium cellulosum So
ce690. Mol. Biosyst. 7, 1549−1563.
(27) Thompson, J. D., Gibson, T. J., Plewniak, F., Jeanmougin, F.,
and Higgins, D. G. (1997) The ClustalX windows interface: flexible
strategies for multiple sequence alignment aided by quality analysis
tools. Nucleic Acids Res. 25, 4876−4882.
(43) Muller, C., Nolden, S., Gebhardt, P., Heinzelmann, E., Lange,
̈
C., Puk, O., Welzel, K., Wohlleben, W., and Schwartz, D. (2007)
Sequencing and analysis of the biosynthetic gene cluster of the
lipopeptide antibiotic Friulimicin in Actinoplanes friuliensis. Anti-
microb. Agents Chemother. 51, 1028−1037.
(44) Heinzelmann, E., Berger, S., Muller, C., Hartner, T., Poralla, K.,
Wohlleben, W., and Schwartz, D. (2005) An acyl-CoA dehydrogenase
is involved in the formation of the Delta cis3 double bond in the acyl
residue of the lipopeptide antibiotic friulimicin in Actinoplanes
friuliensis. Microbiology 151, 1963−1974.
(28) Bradford, M. M. (1976) A rapid and sensitive method for the
quantitation of microgram quantities of protein utilizing the principle
of protein-dye binding. Anal. Biochem. 72, 248−254.
(45) Keenan, M. V., and Alworth, W. L. (1974) The inibition of
proline racemase by a transition state analogueue: Δ-1-pyrroline-2-
carboxylate. Biochem. Biophys. Res. Commun. 57, 500−504.
7196
dx.doi.org/10.1021/bi200892a|Biochemistry 2011, 50, 7184−7197

Biochemistry
Article
(46) Lewis, M. L., Martin, S. L., Rowe, C. J., Sutherland, J. D.,
Wilson, E. J., and Wright, M. C. (1993) Reduction of Δ-1-pyrroline-2-
carboxylic acid to proline by an Escherichia coli proline auxotroph.
Bioorg. Med. Chem. Lett. 3, 1189−1192.
(47) Lewis, M. L., Rowe, C. J., Sewald, N., Sutherland, J. D., Wilson,
E. J., and Wright, M. C. (1993) The effect of pH on the solution
structure of Δ-1-pyrroline-2-carboxylic acid as revealed by NMR and
electrospray mass spectroscopy. Bioorg. Med. Chem. Lett. 3, 1193−
1196.
7197
dx.doi.org/10.1021/bi200892a|Biochemistry 2011, 50, 7184−7197