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

Article abstract of DOI:10.1021/bi9018785

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 in cyanobacteria and proposed a biosynthetic scheme starting from L-proline and involving three polyketide synthases for which the starter would be (S)-1-pyrroline-5-carboxylate bound to an acyl carrier protein, AnaD. We now report the in vitro reconstitution of the first steps of this biosynthesis in Oscillatoria PCC 6506. We identified in PCC 6506 the gene coding for an Sfp-like phosphopantetheinyl transferase and purified the gene product, OsPPT, that catalyzed the transfer of the phosphopantetheinyl arm to the serine 41 of AnaD. The pure adenylation protein AnaC loaded L-proline on holo-AnaD and was specific for L-proline (K_m = 0.97 mM, k_cat =68 min^-1) among the 20 natural amino acids. Among six close structural analogues of L-proline, including (S)-1-pyrroline-5-carboxylate, we only found 3,4-dehydro-L-proline to be an alternate substrate for AnaC (K_m=1.5 mM, k_cat=29min^-1). The putative prolyl-AnaD dehydrogenase, AnaB, purified to homogeneity as a histidine-tagged protein, showed an absorption spectrum characteristic of FAD-containing proteins. It oxidized prolyl-AnaD to dehydroprolyl-AnaD as shown by tryptic digestion of the protein followed by liquid chromatography coupled to tandem mass spectrometry. Alignment of the amino acid sequence of this dehydrogenase with related enzymes showed that AnaB belongs to the acyl-CoA dehydrogenase superfamily and thus probably catalyzes an α-β-dehydrogenation of the thioester-bound proline followed by an aza-allylic isomerization to yield (S)-pyrroline-5-carboxyl-AnaD, the proposed starter for the subsequent polyketide synthase, AnaE.

Full text of DOI:10.1021/bi9018785

Biochemistry 2010, 49, 103–113 103
DOI: 10.1021/bi9018785
In Vitro Reconstitution of the First Steps of Anatoxin-a Biosynthesis in Oscillatoria PCC
†
6
506: From Free -Proline to Acyl Carrier Protein Bound Dehydroproline
L
‡,§
‡
‡,§
,‡
Annick M ꢀe jean, St ꢀe phane Mann, Ga €e lle Vassiliadis, B ꢀe rang ꢁe re Lombard, Damarys Loew, and Olivier Ploux*
‡
Biochimie des micro-organismes, Laboratoire Charles Friedel, UMR CNRS 7223, ENSCP, 11 rue Pierre et Marie Curie, 75231 Paris
Cedex 05, France, Universit eꢀ Paris Diderot-Paris 7, 75013 Paris, France, and Laboratory of Proteomic Mass Spectrometry,
Centre de Recherche, Institut Curie, 26 rue d’Ulm, 75248 Paris, France
§
Received November 3, 2009; Revised Manuscript Received December 2, 2009
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 in cyanobacteria and
proposed a biosynthetic scheme starting from -proline and involving three polyketide synthases for which the
L
starter would be (S)-1-pyrroline-5-carboxylate bound to an acyl carrier protein, AnaD. We now report the
in vitro reconstitution of the first steps of this biosynthesis in Oscillatoria PCC 6506. We identified in PCC
6
506 the gene coding for an Sfp-like phosphopantetheinyl transferase and purified the gene product, OsPPT,
that catalyzed the transfer of the phosphopantetheinyl arm to the serine 41 of AnaD. The pure adenylation
protein AnaC loaded -proline on holo-AnaD and was specific for -proline (Km = 0.97 mM, kcat
8 min ) among the 20 natural amino acids. Among six close structural analogues of -proline, including
S)-1-pyrroline-5-carboxylate, we only found 3,4-dehydro- -proline to be an alternate substrate for AnaC
K = 1.5 mM, k = 29 min ). The putative prolyl-AnaD dehydrogenase, AnaB, purified to homogeneity
L
L
=
-
1
6
(
(
L
L
-
1
m
cat
as a histidine-tagged protein, showed an absorption spectrum characteristic of FAD-containing proteins. It
oxidized prolyl-AnaD to dehydroprolyl-AnaD as shown by tryptic digestion of the protein followed by liquid
chromatography coupled to tandem mass spectrometry. Alignment of the amino acid sequence of this
dehydrogenase with related enzymes showed that AnaB belongs to the acyl-CoA dehydrogenase superfamily
and thus probably catalyzes an R-β-dehydrogenation of the thioester-bound proline followed by an
aza-allylic isomerization to yield (S)-pyrroline-5-carboxyl-AnaD, the proposed starter for the subsequent
polyketide synthase, AnaE.
Cyanobacteria are oxygenic photosynthetic prokaryotes that
While a great deal of information is available on hepatotoxins,
such as microcystins, and the cyanobacteria that produce these
metabolites, our knowledge concerning the cyanobacterial neu-
rotoxins, anatoxin-a, homoanatoxin-a, and saxitoxin, and the
species that produce them is limited. For instance, the genomes of
two strains of the genus Microcystis, that produce microcystins,
have been sequenced (10, 11), and the biosynthetic gene cluster
responsible for the formation of microcystins has been identi-
fied (12, 13). On the other hand, the sequencing of the genome of
neurotoxic cyanobacteria has not yet been accomplished, and the
biosynthetic genes responsible for the production of saxitoxin,
anatoxin-a, and homoanatoxin-a have just been sequenced and
annotated (14, 15). These data will certainly be helpful for the
detection of cyanobacteria producing these neurotoxins, in water
reservoir and in the environment, by using specific PCR-based
methodologies.
To contribute to the knowledge on anatoxin-a-producing
cyanobacteria, we have initiated a project on this interesting
group of bacteria. We have reported the isolation of 15 new
filamentous cyanobacteria from dog poisoning sites in the south
of France (5) and showed that all of these isolated strains were
producing anatoxin-a and homoanatoxin-a and belonged to the
Oscillatoria genus. More recently, we have reported the identifi-
cation of the anatoxin-a and homoanatoxin-a biosynthetic gene
cluster, called ana cluster (15), and showed that a specific DNA
sequence could serve as a genetic marker for this class of
cyanobacteria (16).
can be found in almost every terrestrial or aquatic environment.
This group of bacteria is extremely diverse, and certain species
produce a wide range of secondary metabolites, some of which
are potent toxins for higher animals (1). Several classes of
cyanobacterial toxins have so far been identified including
hepatotoxins, neurotoxins, cytotoxins, and irritant toxins. Cases
of animal death, due to cyanobacterial toxin exposure, are
regularly reported in different places around the world (2-7).
For instance, anatoxin-a and homoanatoxin-a, two cyanobacter-
ial neurotoxins, provoke the rapid death of animals by acute
asphyxia, when ingested, because these alkaloids are potent
agonists of the nicotinic acetylcholine receptor (8). It is now
recognized that the release of cyanobacterial toxins in waters and
water supplies has major implications for public health and for
the environment (9). Therefore, several countries have developed
survey programs to assess the risk for humans of cyanobacterial
toxin exposure in drinking and recreational waters. In that
context, it is important to deepen our knowledge on toxic
cyanobacteria and particularly to develop accurate methods to
detect these toxic bacteria in the environment.
†
This work was supported in part by Aventis Pharma (Groupe Sanofi-
Aventis) and Bayer Pharma, grant “Multiorganismes” for the ENSCP
laboratory. The Laboratory of Proteomic Mass Spectrometry is sup-
ported by grants from Canc ꢀe rop o^ le Ile-de-France and from the Institut
National du Cancer, INCa.
*
To whom correspondence should be addressed. Tel/Fax: þ(33) 1 44
2
7 67 01. E-mail: olivier-ploux@enscp.fr.
r 2009 American Chemical Society
Published on Web 12/02/2009
pubs.acs.org/Biochemistry

1
04 Biochemistry, Vol. 49, No. 1, 2010
M ꢀe jean et al.
Annotation of the ana gene cluster, feeding experiments as well
Scheme 1: The ana Gene Cluster Responsible for Biosynthesis
of Anatoxin-a and Homoanatoxin-a in Cyanobacteria and the
Postulated Biosynthetic Route Leading to These Neurotoxins
as preliminary in vitro data, led us to propose a biosynthetic route
starting from -proline that involved three modular type I
L
1
polyketide synthases (PKS) (15). The ana cluster is represented
in Scheme 1 together with the putative function of the gene
products and with the proposed biosynthetic route to anatoxin-a
and homoanatoxin-a. There are two main stages in this proposed
biosynthesis, first, the steps that allow the formation of the actual
starter for the PKS and, second, the reactions that take place on
the three PKSs. In the first stage
activated by AnaC and then loaded on an acyl carrier protein
ACP), AnaD. In a second step we proposed that AnaB would
oxidize the pyrrolidine ring to form a dehydroproline, probably
S)-1-pyrroline-5-carboxylate (P5C). In the second stage, that is,
L-proline is thought to be
(
(
the steps that are catalyzed by the PKSs on the tethered substrate,
the proposed chemical events were deduced from bioinformatic
analysis of the PKS amino acid sequences. Using the colinearity
principle, we predicted that the starter should be elongated,
cyclized, methylated, and then released and decarboxylated to
form either anatoxin-a or homoanatoxin-a. We showed in a
preliminary report using in vitro experiments (15) that AnaD was
an ACP that was transformed in its holo form by the Bacillus
subtilis Sfp phosphopantetheinyl transferase (PPTase) and that
AnaC was capable of activating L-proline for acylation on holo-
AnaD. These preliminary data confirmed the putative functions
of these proteins predicted from the bioinformatic analysis.
In the present report, we describe the reconstitution, in vitro, of
the first stage of the biosynthesis of anatoxin-a and homoana-
toxin-a in Oscillatoria PCC 6506, that is, the transformation of
were performed using the Clustal X software (17). Chemicals and
biochemicals were purchased from Sigma-Aldrich. High-perfor-
mance liquid chromatography (HPLC) was run on a Hewlett-
Packard 1050 series apparatus equipped with a Jupiter 5 μm C18
˚
300 A Phenomenex column. Elutions were performed using a
L
-proline into AnaD-bound dehydroproline. We have identified
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 220 nm. UV-visible spectra and absorbance
measurements were recorded on an Uvikon 930 Kontron spec-
trophotometer. NMR spectra were recorded at the ENSCP
NMR laboratory. Protein purifications were achieved using an
integrated protein purification system (AKTAprime; GE
Healthcare). PCR and electrophoresis separations (agarose and
acrylamide) were run as already described (15). Axenic strain
Oscillatoria PCC 6506 was obtained from the Pasteur Culture
Collection of Cyanobacteria (PCC) and was grown and main-
tained as already described (18, 19). Protein concentrations were
estimated by using the Bradford colorimetric assay (Bio-
Rad) (20).
the gene in Oscillatoria PCC 6506 that codes for the PPTase of
this organism and showed that the gene product is able to
catalyze the transformation of apo-AnaD into holo-AnaD. We
have also fully characterized the in vitro activity of AnaC, the
proline adenylation protein, and we have shown that pure AnaB
is indeed a prolyl-AnaD dehydrogenase. Thus, we present
experimental evidence that confirms the proposed biosynthetic
pathway for the formation of the neurotoxins anatoxin-a and
homoanatoxin-a in Oscillatoria PCC 6506.
EXPERIMENTAL PROCEDURES
General. Alignments of protein sequences were performed
using the basic local alignment search tool (BLAST) supported
by the National Center for Biotechnology Information or the
Magnifying Genome (MaGe) web servers. Specific alignments
Cloning, Expression, and Purification of the Recombi-
nant Histidine-Tagged Oscillatoria PCC 6506 PPTase
(
OsPPT). The Oscillatoria PCC 6506 PPTase gene (osppt) was
identified by automatic annotation of the draft genome of PCC
506 (7 Mbases, 419 contigs) using the MaGe system (21)
1
6
Abbreviations: ACAD, acyl-CoA dehydrogenases; ACP, acyl carrier
protein; BLAST, basic local alignment search tool; BsPPT, Sfp phospho-
pantetheinyl transferase from Bacillus subtilis; DCPIP, 2,6-dichloro-
(A. M ꢀe jean, O. Ploux, and C. M ꢀe digues, unpublished data).
The osppt gene was amplified from genomic DNA of PCC 6506
using the Pfu DNA polymerase (Promega) and standard condi-
tions, as described by the manufacturer, and the following
phenolindophenol; 3,4-Dhp, 3,4-dehydro-
spray ionization mass spectrometry; HPLC, high-performance liquid
chromatography; IPTG, isopropyl β- -thiogalactoside; LB, Luria-
Bertani; LC-MS/MS, liquid chromatography coupled to tandem mass
spectrometry; -ProSNAC, N-acetylcysteamine thioester of -proline;
L-proline; ESI-MS, electro-
D
0
set of primers (Eurogentec): ospptF, 5 -CGGAATTCAT-
L
L
0
GAATGTTTCTGATTGCCTC-3 , EcoRI site underlined, and
MALDI-TOF, matrix-assisted laser desorption/ionization time-of-flight
mass spectrometry; NRP, nonribosomal peptide; NRPS, nonribosomal
peptide synthase; oAB, ortho-aminobenzaldehyde; OsPPT, Oscillatoria
PCC 6506 phosphopantetheinyl transferase; P5C, (S)-1-pyrroline-5-car-
boxylic acid; PCC, Pasteur culture collection of cyanobacteria; PCP,
peptidyl carrier protein; PK, polyketide; PKS, polyketide synthase; PMS,
phenazine methosulfate; PPTase, phosphopantetheinyl transferase;
ProDh, proline dehydrogenase; SDS-PAGE, sodium dodecyl sulfa-
te-polyacrylamide gel electrophoresis in denaturing conditions; TCEP,
tris(2-carboxyethyl)phosphine; TFA, trifluoroacetic acid; TOF-MS, time-
of-flight mass spectrometry.
0
ospptR,
5 -GCCTCGAGTCAGCTTAAGATATAATTCT-
CTG-3 , XhoI site underlined. The amplified DNA, identified
0
as a single band on agarose gel electrophoresis, was purified
(
PCR Prep; Promega) and digested with the corresponding
restriction enzymes (Promega) and ligated (T4 DNA ligase from
Promega) into pET28a vector (Novagen) previously cut with the
same enzymes. After transformation in Escherichia coli JM109
competent cells (CaCl treated) and selection of positive clones,
2

Article
Biochemistry, Vol. 49, No. 1, 2010 105
the plasmid, pET28H6osppt, was purified (WizardPlus Miniprep;
Promega) and the insert sequenced (GATC company) for
verification. The recombinant protein was expressed in E. coli
BL21(DE3) cells (Novagen) transformed with pET28H6osppt.
Cells were grown in 1 L of Luria-Bertani (LB) medium at 37 °C
under agitation (200 rpm), supplemented with 50 μg/mL kana-
mycin, and the expression was induced at midlog phase
(Amicon Ultra-15 centrifugal filter unit 10k), and sorbitol was
added to the solution up to 7% (w/v) before storage at -20 °C.
Determination of AnaC Kinetic Parameters and Speci-
ficity. Histidine-tagged AnaC was produced and purified as
already described (15). The kinetic parameters and specificity of
AnaC were determined using the PP -ATP exchange reaction as
i
already described (15). Briefly, the assay (100 μL total volume)
(
absorbance at 600 nm = 0.6) with 0.250 μM isopropyl β-
D
-
consisted of 100 mM Tris-HCl, pH 7.7, 5 mM TCEP, 10 mM
3
2
thiogalactoside (IPTG). The cells were then left overnight at
MgCl , 5 mM ATP, 5 mM substrate, 1 mM [ P]pyrophosphate
2
2
4
5 °C with agitation and collected by centrifugation (4000g, at
°C for 20 min, Sorval RC5B centrifuge, SLA-3000 rotor). The
(specific radioactivity 1.2 Ci/mol; Amersham), and 0.2 μM
AnaC. The reactions were initiated by the addition of the enzyme
and incubated at 28 °C for 30 min. The reactions were quenched
by the addition of 1 mL of stop solution: 1.6% (w/v) activated
charcoal, 4.5% (w/v) tetrasodium pyrophosphate,and 3.5% (v/v)
perchloric acid in water. The samples were filtered on Whatman
paper filters and rinsed three times with 15 mL of water. After
drying, the filters were counted using 6 mL of Picofluor scintilla-
tion counting liquid on a scintillation counter (Beckman LS6500;
Beckman Coulter). Each reaction was run in triplicate, and
background counting (reaction without enzyme) was subtracted.
cell paste was resuspended in 10 mL of 500 mM NaCl and 100
mM sodium phosphate buffer, pH 8.0, and the cells were
disrupted by sonication (Misonix sonicator 500 W, 3 mm probe,
4
min at 4 °C with cycles of 15 s sonication followed by cooling
periods of 15 s). The resulting suspension was centrifuged
25000g, at 4 °C for 30 min, Sorval RC5B centrifuge, SS-34
rotor), and the supernatant was loaded on a Ni-affinity column
1 mL HisTrap column; GE Healthcare) previously equilibrated
(
(
with buffer A: 500 mM NaCl and 50 mM sodium phosphate
buffer, pH 8.0, containing 20 mM imidazole. After the column
was washed (20 mL of buffer A at pH 8.0) the protein was then
eluted using a linear gradient of imidazole from 20 mM to 0.5 M
in 500 mM NaCl and 100 mM sodium phosphate buffer, pH 8.0,
during 20 min at 1 mL/min. The collected fractions were analyzed
by polyacrylamide gel electrophoresis under denaturing condi-
tions (SDS-PAGE), and the fractions containing the pure
protein were pooled. The protein solution was desalted on a
PD10 column (GE Healthcare) using 50 mM Tris-HCl buffer, pH
The kinetic parameters for L-proline and 3,4-dehydro-L-proline
(3,4-Dhp) were determined using the standard assay but using
various concentrations of the substrate (from 0 to 6 mM). The
data points were fitted to the Michaelis-Menten equation using
nonlinear regression analysis (Kaleidagraph software).
Preparation of Prolyl-AnaD. The loading of L-proline on
the holo-AnaD was achieved as already described (15). The assay
(100 μL total volume) consisted of 100 mM Tris-HCl, pH 7.7,
5 mM TCEP, 10 mM MgCl , 5 mM ATP, 5 mM
[U- C, N]-L-proline (Cambridge Isotope), 50 μM holo-AnaD,
L
-proline or
2
1
3
15
7.7, and 100 mM NaCl, as eluent. After the protein was
concentrated using a concentrator device (Amicon Ultra-15
centrifugal filter unit 10k), sorbitol was added to a final concen-
tration of 7% (w/v), and the pure protein was stored at -20 °C.
PPTase Activity Assay. Histidine-tagged recombinant apo-
AnaD was produced and purified as already described (15). The
histidine-tagged Sfp PPTase from B. subtilis (BsPPT) was
obtained as already described (15), using an overproducing strain
kindly provided by Peter F. Leadlay and S. Dickens (Department
of Biochemistry, University of Cambridge; strain number PFL-
C1043). A typical PPTase assay (100 μL total volume) consisted
and 0.2 μM AnaC, and the reaction was incubated at 28 °C for
30 min. For preparation of large quantities of prolyl-AnaC the
assay was adapted to 2 mL total volume, and the concentration of
AnaC was 2 μM. The prolyl-AnaD was then desalted on a PD-10
column (GE Healthcare) using 100 mM Tris-HCl, pH 7.0, as
eluent. The protein was then concentrated by ultrafiltration
(Amicon Ultra-15 centrifugal filter unit 10k) and used immedi-
ately, or the solution was supplemented with 7% (w/v) sorbitol
before storage at -20 °C. The purified prolyl-AnaD eluted at
13.8 min using the HPLC conditions described above.
of 100 mM Tris-HCl, pH 7.7, 10 mM MgCl , 5 mM tris-
Synthesis of (S)-1-Pyrroline-5-carboxylic Acid (P5C).
P5C was prepared as described by Gerratana et al. (23, 24) by the
deprotection in acidic conditions of (2S)-tert-butyl-N-Boc-
5-hydroxyprolinate. P5C was then purified by ion-exchange
chromatography and assayed using o-aminobenzaldehyde
(oAB) as already described (25, 26). Pure fractions of P5C were
concentrated to dryness under vacuum, solubilized in 0.1 M HCl
to a final concentration of 132 mM, and stored at 4 °C.
2
(
2-carboxyethyl)phosphine (TCEP), 50 μM CoASH, 0.5 μM
PPTase, either the BsPPT or the OsPPT, and 50 μM apo-AnaD.
The reaction was incubated at 28 °C and stopped by the addition
of 150 μL of a stop solution: 50% (v/v) methanol in 0.1% (v/v)
aqueous TFA. The solution was kept at 0 °C prior to analysis by
HPLC. The apo-AnaD (retention time = 14.2 min) and holo-
AnaD (retention time = 14.8 min) were separated and quantified
by HPLC. The holo-AnaD production, in this assay, was linear
with time, from 0 to 15 min, and linear with PPTase concentra-
tion, from 0 to 0.15 μM. The effect of pH on the OsPPT activity
was examined using the above assay using a complex buffer
mixture (0.1 M acetic acid, 0.1 M 2-(N-morpholino)ethanesulfo-
nic acid, 0.2 M Tris (22)) set at different pHs from 4.0 to 9.0.
Large quantities of holo-AnaD were prepared as follows: the
assay (1.5-5 mL final volume) consisted of 100 mM Tris-HCl,
Synthesis of the N-Acetylcysteamine Thioester of L-
Proline (L-ProSNAC).
by Ehmann et al. (27).
L-ProSNAC was prepared as described
Cloning, Expression, and Purification of the Recombi-
nant Histidine-Tagged Oscillatoria PCC 6506 AnaB.
The anaB gene (15) was PCR-amplified (GoTaq polymerase;
Promega) from genomic DNA from PCC 6506 using the
0
following primers: anaBF, 5 -CGCGCGAATTCGATTTTG-
0
pH 7.7, 10 mM MgCl , 5 mM TCEP, 60 μM CoASH, 2 μM
CCTGGAACAGTC-3 , EcoRI site underlined, and anaBR,
2
0
0
PPTase, either the BsPPT or the OsPPT, and 50 μM apo-AnaD.
The reaction mixture was incubated for 2 h at 20 °C, and the
holo-AnaD thus formed was desalted on a PD-10 column (GE
Healthcare) using 100 mM Tris-HCl, pH 7.7, and 1 mM TCEP as
eluent. The protein was then concentrated by ultrafiltration
5 -CGCGCCTCGAGTTACAATCCCAAGAATTTA-3 , XhoI
site underlined. The PCR product was purified and cloned into
a pET28a vector as described above for the osppt gene.
The recombinant AnaB histidine-tagged protein was produced
in E. coli BL21-CodonPlus (DE3)-RIPL strain (Stratagene).

1
06 Biochemistry, Vol. 49, No. 1, 2010
M ꢀe jean et al.
Cells were grown in LB medium (1 L) at 37 °C under agitation
250 rpm) in the presence of 50 μg/mL kanamycin and 34 μg/mL
chloramphenicol. When the culture reached an absorbance of
.6 at 600 nm, the medium was cooled to 25 °C, and the induction
described by Wilm and Mann (30). Analyst QS software 2.0
(Applied Biosystems, MDS Sciex) was used for the spectrum
acquisition and the data analysis. Prolyl-AnaD and its oxidation
product formed after incubation in the presence of AnaB, as
described above, were analyzed as follows. Two hundred pico-
moles of the HPLC-purified prolyl-AnaD and of the oxidized
product of prolyl-AnaD, prepared as described above, was dried
under vacuum (SpeedVac system) and then resuspended by
sonication in 100 μL of a solution of 80% (v/v) acetonitrile/
25 mM ammonium bicarbonate. Trypsin (Sigma, T6567) diges-
tion was then performed by the addition of 10 μL of a trypsin
solution (200 ng/mL) to the suspension, and the samples were
incubated on a rotator for 6 h at 37 °C. The trypsin-generated
mixture was dried under vacuum and suspended in 20 μL of
solvent A (5% (v/v) acetonitrile/water, 0.1% (v/v) formic acid),
and 1 μL of the solution was analyzed using an actively split
capillary HPLC system (Ultimate 3000; Dionex, Germering,
Germany) connected to the Q-TOF mass spectrometer. The
digest was loaded onto a C18 PepMap guard column (0.3 mm
(
0
was initiated with 0.2 mM IPTG. The agitation was maintained
overnight at 25 °C. Cells were harvested by centrifugation (4000g
at 4 °C for 20 min; SLA-3000 rotor) and resuspended in 5 mL of
20 mM Tris-HCl buffer, pH 7.4, containing 0.5 M NaCl, 20 mM
imidazole, 10% (v/v) glycerol, and 0.1 mM FAD. Lysis was
performed by sonication at 4 °C for 3 min with 10 s pulses spaced
by 20 s cooling periods. After centrifugation (two runs at 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 Tris-HCl
buffer, pH 8.0, containing 0.5 M NaCl and 10% (v/v) glycerol.
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, and
2
5
0% (v/v) glycerol. The overall yield of production of AnaB was
mg/L of culture. The UV-visible absorption spectrum of pure
AnaB was recorded on a pure sample of AnaB at 0.56 mg/mL
˚
i.d. ꢀ 5 mm long, 5 μm particle size, 100 A pore size; Dionex S.A.)
in solvent A. The sample was washed on the column for 3 min
with solvent A prior to separation using a LC Packings PepMap
C18 column (75 μm i.d. ꢀ 150 mm long, packed with 3 μm
(
12 μM) in 20 mM Tris-HCl, pH 7.4, containing 10% (v/v) glycerol.
Alternate Substrates for AnaB. Two assays were used to
˚
detect any AnaB activity using free
L-proline,
L-proline methyl
particles with 100 A pore size; Dionex S.A.) using a 60 min linear
ester, or -ProSNAC as alternate substrates. 2,6-Dichlorophe-
L
gradient from 5% to 50% (v/v) of solvent B (80% (v/v)
acetonitrile/water, 0.085% (v/v) formic acid) at 200 nL/min.
Data acquisition was performed using the Analyst QS Software,
set for the positive ion mode with a spray voltage of 2.3 kV.
A TOF-MS survey scan was acquired for 1 s over a mass range
of m/z 700-800 and MS/MS range of m/z 65-1700. An
information-dependent acquisition method was used to acquire
product ion scans on the three most intense ions per cycle over
a mass range of m/z 65-1700, excluding previously gated ions
for 60 s. A Smart setting of 20.0 was used.
nolindophenol (DCPIP) assay (28): the assay consisted of 50 mM
Tris-HCl buffer, pH 7.5, 50 μM DCPIP, 100 μM FAD, 1 mM
phenazine methosulfate (PMS), 1-5 mM substrate, and 1 μM
AnaB. The solution was incubated at 30 °C, and the absorbance
at 610 nm was monitored against time. oAB assay (29): the assay
consisted of 50 mM Tris-HCl buffer, pH 7.5, 1-5 mM substrate,
and 1 μM AnaB in a total volume of 210 μL. After various
incubation times (0, 5, 10, 20, 60 min) at 30 °C, 200 μL of
10% (v/v) aqueous trichloroacetic acid and 40 μL of 0.1 M oAB
in 40% (v/v) aqueous ethanol were added to the assay solution,
and the absorbance at 440 nm was measured.
Oxidation Reaction Catalyzed by AnaB. Prolyl-AnaD
RESULTS
(
25 μM), prepared as described above, was incubated in the
presence of pure AnaB (2 μM) in 100 mM Tris-HCl buffer, pH
.0, in the presence of PMS (1 mM) or in the absence of any
Identification and Functional Characterization of the
Oscillatoria PCC 6506 PPTase. We have undertaken the
sequencing of the genome of the neurotoxic cyanobacteria
Oscillatoria PPC 6506, and a draft genome has already been
obtained consisting of 7 Mbases on 419 contigs. This draft
genome has been automatically annotated using the MaGe
system (A. M ꢀe jean, O. Ploux, and C. M ꢀe digue, unpublished
data), and among the annotated genes we found only one
putative PPTase gene that we called osppt. Alignment of the
sequence of the gene product OsPPT with PPTase sequences that
gave the highest BLAST scores (Supporting Information Figure
S1 and Table S1) showed that this enzyme belonged to the Sfp-
like PPTase family (31, 32) and more specifically to the W/KEA
cyanobacterial PPTase subfamily as defined by Neilan and co-
workers (33). Indeed, we found the following consensus se-
7
electron acceptor except naturally present oxygen. The oxidation
reaction was incubated at 28 °C for 30-60 min. The reaction
mixture was analyzed using HPLC as described above and mass
spectrometry as described below. A control experiment was run
in parallel in the absence of AnaB.
Mass Spectrometry Analyses. OsPPT, AnaC, and AnaB
samples were desalted using ZipTipC4 pipet tips prior to analysis
by mass spectrometry. The MALDI-TOF mass spectra were
recorded on an ABI 4800 TOF/TOF mass spectrometer (Applied
Biosystems) in positive ion linear mode by using a saturated
sinapinic acid matrix (50% (v/v) acetonitrile/water, 0.1% (v/v)
TFA). The AnaD samples, apo-, holo-, and proly-AnaD, were
first desalted by reversed-phase HPLC under the conditions
described above. The samples were concentrated in vacuum
87
quences that are conserved in this class of enzymes: F ,
9
5
113
130
152
GKP , FNX(S/A)HS , (I/L)G(V/I)D , E , and F(F/Y)-
1
80
(
SpeeVac system) and analyzed on a QSTAR Elite quadrupole
X WXKEAX K (the numbering refers to the native OsPPT
2 2
time-of-flight (Q-TOF) mass spectrometer (Applied Biosystems/
MDS Sciex) equipped with an external nanoelectrospray ioniza-
tion (nanoES) ion source. NanoES was performed using a coated
medium borosilicate capillary (Proxeon Biosystems A/S,
Denmark). The electric field required to spray the solution was
established by applying a voltage of about 950 V. The solution
flow rate was estimated to be between 10 and 40 nL/min as
sequence). The closest PPTase, based on BLAST searches,
was the enzyme from Trichodesmium erythraeum IMS101 (46%
identity, 64% similarity, Supporting Information Table S1),
another cyanobacterium whose genome has been completely
sequenced.
We have cloned the osppt gene into an expression vector, and
we have purified the N-terminal histidine-tagged fusion protein

Article
Biochemistry, Vol. 49, No. 1, 2010 107
Table 1: Mass Spectrometry Analysis of the Purified Proteins Studied in
This Report
Table 2: AnaC Specificity toward Its Amino Acid Substrate As Probed by
the PP -ATP Exchange Assay
i
a
calcd average mass (Da)
amino acid
substrate
relative
amino acid
substrate
relative
activity (%)
activity (%)
intact
obsd
protein
protein
protein - Met
mass (Da)
Ala
Arg
Asn
Asp
Cys
Gln
Glu
Gly
His
Ile
0.6 ( 0.1
0.2 ( 0.1
0.0 ( 0.1
0.0 ( 0.1
0.5 ( 0.1
0.0 ( 0.1
0.0 ( 0.1
0.1 ( 0.1
0.2 ( 0.1
0.6 ( 0.1
Leu
Lys
Met
Phe
Pro
Ser
1.8 ( 0.2
0.4 ( 0.1
0.7 ( 0.1
0.3 ( 0.1
100.0 ( 0.3
0.3 ( 0.1
0.3 ( 0.1
0.7 ( 0.1
0.9 ( 0.1
0.4 ( 0.1
b
þ
þ
þ
OsPPT
32851.38
63693.03
46548.44
13941.03
14281.36
14378.48
32720.18
63561.83
46417.24
13809.82
14150.16
14247.28
14253.20
32720 (M þ H)
63513 (M þ H)
46519 (M þ H)
13809.6
b
AnaC
b
AnaB
c
apo-AnaD
holo-AnaD
c
14150.2
Thr
Trp
Tyr
Val
c
prolyl-AnaD
14247.2
13
15
c
[
U- C, N]-prolyl-AnaD 14384.40
14252.8
a
The calculated masses correspond to the intact recombinant protein
left) or to the protein lacking the N-terminal Met (right). Mass measured
by MALDI-TOF. Mass measured by ESI-MS after deconvolution. Minor
species, acetylated M þ 42, and gluconylated M þ 178 (37) were also
observed.
b
(
c
F
IGURE 2: Specificity of AnaC-catalyzed PP
ward proline analogues. The assay was run in the presence of
mM proline analogue, and the relative initial rates were determined.
i
-ATP exchange to-
5
Each data point represents the average of three independent experi-
ments (standard errors were less than 0.2%). Compounds: 1,
L-proline; 2, 3,4-dehydro-L-proline (3,4-Dhp); 3, trans-4-hydroxy-L-
proline; 4, (S)-1-pyrroline-5-carboxylic acid (P5C); 5, pyrrole-2-
carboxylic acid; 6, L-pyroglutamic acid; 7, D-proline.
F
IGURE 1: HPLCtracesofthe three different forms ofthe acylcarrier
protein AnaD. From top to bottom: apo-AnaD (b), holo-AnaD (1),
prolyl-AnaD (9), and the mixture of the three samples. The HPLC
analyses were run as described in the Experimental Procedures
section. Approximately 2.5 nmol (35 μg) of protein samples was
injected.
The pH profile of the specific activity of OsPPT, measured under
saturating conditions, was bell-shaped with an optimum pH at
7.0 as observed for similar enzymes (34-36).
Kinetics and Specificity of the Adenylation Protein
AnaC. We previously showed that the adenylation protein AnaC
to homogeneity by affinity chromatography. It showed one band
by SDS-PAGE and one peak by reversed-phase HPLC and
showed an average mass consistent with that predicted (Table 1).
The PPTase activity of this new enzyme was probed using an
HPLC-based assay and apo-AnaD as the substrate because we
could readily separate the apo-AnaD and holo-AnaD forms
using this technique (Figure 1). The OsPPT readily transformed
apo-AnaD into holo-AnaD at saturating concentrations of
substrates, that is, 50 μM apo-AnaD and 50 μM CoASH. The
specific activity measured under these conditions was 0.21 μmol
was capable of activating L-proline for acylation of holo-AnaD to
yield prolyl-AnaD (15). Using the classical PP -ATP exchange
i
assay, we tested the ability of this enzyme to discriminate among
the 20 proteinogenic amino acids. Table 2 shows that AnaC is
very specific for L-proline and does not activate the other 19
amino acids at a reasonable rate. However, slow activation was
observed for Leu (1.8% of relative activity). We also tested six
close proline analogues for activation by AnaC as shown in
Figure 2. We observed a sharp specificity with only 3,4-dehydro-
-
1
-1
-1
min mg (apparent kcat = 6.0 min ). The Sfp PPTase from
B. subtilis was slightly less effective under the same conditions for
the transformation of apo-AnaD into holo-AnaD: specific
L-proline, 2, being an alternate substrate (relative activity=
activity of 0.13 μmol min- mg (apparent kcat = 3.3 min ).
1
-1
-1
39.6 ( 0.2%) and to a lesser extent trans-4-hydroxy- -proline,
L

1
08 Biochemistry, Vol. 49, No. 1, 2010
M ꢀe jean et al.
3
(relative activity = 10.4 ( 0.2%). A similar specificity profile
the presence of 0.2 μM AnaC. The mass spectrometry data were
in perfect agreement with the formation of a prolyl thioester on
the phosphopantetheinyl arm of the holo-AnaD using either
among these three compounds, 1, 2, and 3, was observed in the
case of the homologous CloN4 and CouN4 enzymes (38), that
catalyze the activation of proline in the biosynthesis of clorobio-
cin and coumermycin A, respectively. The four other compounds
tested were very poor to null substrate for AnaC. P5C, 4 (relative
activity = 2.5 ( 0.1%), the putative structure formed on AnaD
after AnaB oxidation, is a very poor substrate. The aromatic
pyrrole-2-carboxylate, 5, and the pyroglutamate, 6, were not
better substrates for AnaC (relative activities = 1.1 ( 0.1% and
1
3
15
unlabeled proline or [U- C, N]-L-proline (Table 1). This prolyl-
AnaD was rather unstable at pH 7.0 with decomposition into the
holo-AnaD form. It is not clear at the moment what reaction
takes place, but preliminary data (S. Mann and O. Ploux,
unpublished data) showed that the thioester bond is hydrolyzed
to form the thiol and free proline. Thus, the prolyl-AnaD could
only be kept for 1 or 2 days before its use (estimated half-life of 2 h
in buffered solution at pH 7.0 and 28 °C).
0.0 ( 0.1%, respectively). Interestingly,
D-proline, the enantio-
mer of the substrate, is not activated by AnaC (relative activity =
Given the fact that AnaC is very specific for L-proline, we have
1
.7 ( 0.1%), and thus this adenylation protein is rather en-
aligned the amino acid sequence of this adenylation protein with
those of nine other adenylation proteins that activate proline and
that are involved in the biosynthesis of pyrrole containing
secondary metabolites: pyoluteorin (42), coumermycin (43), pro-
diginine (44), clorobiocin (45), prodigiosin (46), pyrrolomycin
and dioxapyrrolomycin (49), Dkxanthene (50), and indanomy-
cin (51). As expected, these enzymes are closely related with
sequence identity to the AnaC sequence ranging from 31% to
40% (Supporting Information Table S2). Because several
authors have proposed a code to predict the specificity of any
adenylation protein or domain (40, 52), based on its amino acid
sequence and on the three-dimensional structure of GrsA,
we have aligned the GrsA and AnaC sequences together
with the nine proline adenylation protein sequences quoted
above (42-51). Inspecting the alignment of the ten proline-
specific adenylation proteins (Supporting Information Figure S2)
showed that the residues corresponding to the GrsA residues
implicated in the amino acid substrate binding pocket were
almost strictly conserved within the proline-specific enzymes
antioselective, unlike other homologous adenylation domains or
proteins (39-41).
Using the PP -ATP exchange reaction, we have also deter-
i
mined the kinetic parameters for activation of L-proline and 3,
4
-Dhp (Figure 3): K (Pro) = 0.97 ( 0.17 mM; kcat(Pro) =
m
-
1
-1 -1
6
8.5 ( 3.8 min ; kcat/K (Pro) = 1 177 M
s ; K (3,4-Dhp) =
m
m
-1
1.54 ( 0.29 mM; k (3,4-Dhp) = 29.4 ( 2.1 min ; kcat/K (3,4-
cat
-1
m
-
1
Dhp) = 318 M
s
. Thus, the discrimination of Pro versus 3,
4-Dhp is largely determined by the catalytic constant. The kinetic
parameters determined for -proline activation by AnaB were in
close agreement with those obtained for the homologous CloN4
and CouN4 enzymes (38).
Using the HPLC-based assay (Figure 1), we showed that
AnaC readily transformed holo-AnaD into prolyl-AnaD. The
conversion of 50 μM holo-AnaD was completed in 15-30 min in
L
(
Table 3). However, the codes previously proposed for the
prediction of adenylation of proline by adenylation proteins were
235
not fully respected in AnaC-related proteins. Besides Asp and
5
17
Lys
(numbering of GrsA) which cannot be used for the
prediction because they are almost strictly conserved in all
adenylation proteins, we only found three positions out of eight
that conformed to the previous codes.
Production and Characterization of the Prolyl-AnaD
Dehydrogenase, AnaB. In the cluster of genes responsible for
the biosynthesis of anatoxin-a and homoanatoxin-a we identified
a putative prolyl-AnaD dehydrogenase coded by the gene anaB
(Scheme 1, Supporting Information Figure S3 and Table S3). We
speculated that this enzyme would oxidize the prolyl-AnaD to the
F
IGURE 3: Determination of the kinetic parameters of AnaC for
L-proline (O) and 3,4-Dhp (b) using the PP -ATP exchange assay.
i
The assay was run in the presence of various concentrations of the
substrate, and the initial rates were determined. Each data point
represents the average of three independent experiments. The data
points were fitted to the Michaelis-Menten equation using nonlinear
regression analysis.
(
S)-pyrroline-5-carboxyl-AnaD (P5C-AnaD) based on similar
biosynthetic schemes (42-51) and on reasonable chemistry that
could lead to anatoxin-a and homoanatoxin-a (15). We have
cloned the anaB gene in a suitable expression vector to produce a
Table 3: A Plausible Code for Predicting Proline-Specific Adenylation Proteins
sequence/consensus
amino acid found on the selected positions
a
235
236
239
278
299
301
322
330
331
517
GrsA
D
D
D
A
V
V
W
Q
Q
T
L
I
I
A
A
A
A
I
C
V
K
K
K
b
c
code for Pro
code for Pro
H
H
V
V
V
V
S/Y/V/F
I/A
-
d
219
220
223
262
289
291
315
323
324
527
AnaC
D
D
L
L
F
Y
L
A
A
L
C
K
K
e
consensus found in present study
F/L
Y
L/I
L/W
C
a
b
c
These ten residues constitute the Phe binding site in GrsA, as shown by Marahiel et al. (39, 40). GrsA numbering. As defined by Marahiel et al. (40). As
d
defined by Townsend et al. (52). Residues of AnaC corresponding to the ten residues of GrsA implicated in the amino acid binding site, as determined by
sequence alignment. AnaC numbering. Residues found in the sequence of the nine proline-specific adenylation proteins under study, at the selected positions
e
thought to be involved in proline binding, as determined by multiple alignment. Strictly conserved residues in the different codes are in bold type.

Article
Biochemistry, Vol. 49, No. 1, 2010 109
F
0
(
IGURE 4: UV-visible absorption spectrum of pure AnaB at
.56 mg/mL (12 μM) in 20 mM Tris-HCl, pH 7.4, containing 10%
v/v) glycerol.
recombinant N-terminal histidine-tagged protein. The enzyme
was readily produced in the E. coli BL21-CodonPlus (DE3)-
RIPL strain and purified to homogeneity by affinity chromatog-
raphy. The pure enzyme, as judged by SDS-PAGE and HPLC,
had a mass in agreement with that predicted from the sequence
(
Table 1). The enzyme exhibited an electronic absorption spec-
trum in the UV-visible range with bands at 272, 345, and 438 nm
and shoulders at 407 and 466 nm (Figure 4). The ratio of
absorbance at 438 nm and at 272 nm was 0.175. This spectrum
is characteristic of FAD-dependent enzymes and is consistent
with the fact that a BLAST search with the protein sequence of
AnaB showed that this enzyme belonged to the acyl-CoA
dehydrogenase (ACAD) superfamily (Supporting Information
Figure S3 and Table S3).
F
IGURE 5: Mass spectrometry analysis by LC-MS/MS of the reac-
tion catalyzed by AnaB. The spectra were recorded after tryptic
digestion and liquid chromatography separation of either prolyl-
AnaD before oxidation (A) or after AnaB-catalyzed oxidation (B).
(
A) Mass spectrum (the spectrum was zoomed twice along the y axis
To detect the AnaB-catalyzed oxidation product, we incubated
prolyl-AnaD (25 μM) in the presence of pure AnaB (2.5 μM) at
pH 7.0 using either PMS or oxygen as the electron acceptor.
Unfortunately, the HPLC traces did not reveal the presence of a
new product, either because the substrate and the product
coeluted or because the reaction yield was too low. We thus
decided to use mass spectrometry to detect this oxidation
product. Because a loss of 2 Da over 14247 Da, the expected
decrease in mass after oxidation of prolyl-AnaD by AnaB would
be impossible to detect as already noted by others (38), we
decided to detect the reaction product by digesting the protein
and by analyzing the fragments using liquid chromatography
coupled to tandem mass spectrometry (LC-MS/MS). When the
substrate of the reaction, prolyl-AnaD, was analyzed using this
strategy (Figure 5A), we detected the tryptic fragment F1
containing the conserved serine at position 41 of the apo-AnaD
for clarity) of the prolyl-AnaD tryptic peptide, from positions 38 to
46 of AnaD that contains the phosphopantetheinyl arm. The molec-
2þ
ular ion is symbolized as F1 m/z 773.8744 for (M þ 2H) . (B) Mass
spectrum (the spectrum was zoomed twice along the y axis for clarity)
of the tryptic peptide, from positions 38 to 46 of AnaD that contains
the phosphopantetheinyl arm, of the AnaB-catalyzed oxidation
product. The molecular ion is symbolized as F7 m/z 772.8639 for
2þ
(M þ 2H) . See Table 4 and text for full interpretation of the
fragmentations.
interpretation of the y and b fragments formed (Supporting
Information Figure S5).
After oxidation of prolyl-AnaD catalyzed by AnaB in the
presence of oxygen, the reaction product was analyzed using the
same strategy (Figure 5B). We clearly observed fragments F2, F3,
F7, F8, and F9. The tryptic fragment F7 had lost 2 Da when
compared to the corresponding fragment F1. The same reduction
of 2 Da was observed on fragments F8 and F9, corresponding to
fragments F4 and F5, respectively, of the mass spectrum shown in
Figure 5A but not on fragments F2, F3, and F6 that were present
in the two mass spectra. No oxidation product was oberved in the
control experiment without AnaB. This unambiguously shows
that the two-electron oxidation took place on the proline moiety,
as expected. Thus AnaB is a prolyl-AnaD dehydrogenase
(Scheme 1). The oxidation reaction was not complete because
in the oxidation mixture we observed the presence of the tryptic
fragment that came from prolyl-AnaD. The yield was low based
on the ratio of the tryptic fragments that came from the substrate
and from the product in the LC-MS chromatogram (total ion
current). Interestingly, we did not detect, in the reaction mixture,
(
native sequence; see Supporting Information Figure S4 and
Table S4), with the phosphopantetheinyl arm attached to it.
Fragmentation of this ion, F1, was observed, and we could
attribute the structure of the different major fragments formed,
F2 to F6 (Figure 5A and Table 4). Fragment F1 is the intact
tryptic fragment which gives fragments F2 and F3 with the
modified serine at position 41 as a dehydroalanine or a phos-
phoserine, respectively. Fragment F4 corresponds to the entire
phosphopantetheinyl arm with the prolyl thioester attached
to it. This fragment gives rise to fragment F5 by the loss of
H PO , which in turn loses the prolyl moiety to yield fragment
3
4
F6. The sequence of the tryptic fragment together with
the localization of the pantotetheinyl arm was confirmed by

1
10 Biochemistry, Vol. 49, No. 1, 2010
M ꢀe jean et al.
Table 4: Fragments Observed after Tryptic Digestion and LC-MS/MS Analysis of Prolyl-AnaD and Its Oxidation Product after AnaB Catalysis
the presence of pyrrolyl-AnaD, the putative four-electron oxida-
tion product.
In an attempt to perform kinetic analysis of the AnaB-
catalyzed reaction we searched for alternate substrates for this
enzyme because of the inherent difficulties in performing the
these toxins have been reported, dramatically highlighting the
risk that represents the presence of these neurotoxic cyanobacteria
in the environment and in water reservoirs. The biosynthesis of
these harmful neurotoxins remained unclear until we published
recently the identification of the gene cluster responsible for the
biosynthesis of anatoxin-a and homoanatoxin-a in Oscillatoria
PCC 6506 (15). Using different approaches we concluded that the
reaction using prolyl-AnaD. Unfortunately, free
L
-proline,
-proline methyl ester, or the N-acetylcysteamine thioester of
-proline (ProSNAC) were not substrates for AnaB. We used two
L
L
biosynthesis started from free L-proline and involved three PKSs
different assays to detect any AnaB activity: the reduction of
DCPIP or the formation of an adduct with the putative pyrroline
product and oAB. As noted by Stachelhaus and co-workers (27)
the ProSNAC is rather unstable at neutral or slightly basic pH.
We found that this molecule decomposed into proline and
N-acetylcysteamine probably via a hydrolysis catalyzed by the
free amine of proline (S. Mann and O. Ploux, unpublished
observation). Nevertheless, using optimized conditions, we could
not observe the oxidation of this prolyl-AnaD surrogate by
AnaB.
to yield anatoxin-a or homoanatoxin-a.
In this study, we have reconstituted the first steps of the
anatoxin-a/homoanatoxin-a biosynthesis in Oscillatoria PCC
6506. We have purified the four enzymes implicated in the
transformation of free L-proline to acyl carrier protein bound
dehydroproline that will serve as the actual starter for the PKS-
mediated following steps (Scheme 1).
The PPTase responsible for the attachment of the phospho-
pantetheinyl arm on the ACP apo-AnaD was identified by
annotation of the genome of PCC 6506. The osppt gene was
cloned in an expression vector, the enzyme was purified, and its
activity was clearly demonstrated using an HPLC-based assay.
The catalytic properties of this newly described PPTase, the
OsPPT, were similar to those reported for the PPtase from
Synechocystis PCC 6803 (35) and from B. subtilis, the standard
PPTase (34). Alignment of the sequence of this OsPPT with other
DISCUSSION
Anatoxin-a and its higher homologue homoanatoxin-a are two
potent cyanobacterial neurotoxins. Cases of animal death after
ingestion of water contaminated by cyanobacteria producing

Article
Biochemistry, Vol. 49, No. 1, 2010 111
PPTases showed that it belonged to the Sfp superfamily, as
expected, and the closest enzyme found was the PPTase from
T. erythreaum IMS101. As for this cyanobacterium, whose
genome sequence determination has been completed, we only
found one PPTase in the draft genome of Oscillatoria PCC 6506.
It cannot be excluded that another PPTase enzyme exists in
Oscillatoria PCC 6506 since the genome sequencing is not
finished, but it seems reasonable to postulate that this OsPPT
serves as the transferase for all ACPs in this cyanobacterium,
including ACPs involved in fatty acid metabolism.
Scheme 2: Proposed Mechanism for the AnaB Oxidation
Yielding to P5C-AnaD
a
The second step of anatoxin-a biosynthesis that we studied in
this report was the loading of
catalyzed by the adenylation protein AnaC. We showed that this
enzyme was very specific for its substrate -proline and that even
L-proline to the holo-AnaD,
a
L
2
Theelectronacceptor(Ox/Red)isO intheinvitroexperiment.
close structural analogues, such as P5C, were not substrates for
AnaD. The 3,4-Dhp analogue was the only alternate substrate
with a kcat/Km 4 times lower than that for L-proline, essentially
responsible for the loading of proline on an ACP followed by the
oxidation of the tethered proline ring. Thus, these clusters, or at
least the genes coding for the enzymes catalyzing the first steps of
the biosynthesis of the secondary metabolites, are very likely
evolutionarily related. Alignments of the amino acid sequence of
AnaB with related prolyl-ACP dehydrogenases and representa-
tives of the ACAD superfamily (Supporting Information Figure
S3 and Table S3) showed that these FAD-containing prolyl
dehydrogenases were homologous to the isovaleryl-CoA
dehydrogenase (53-55) and that the active site base (E254 in
human isovaleryl-CoA dehydrogenase) is conserved in AnaB and
related prolyl-ACP dehydrogenases while the active site base of
the pig medium chain acyl-CoA dehydrogenase (E376) was not
conserved in this class of enzymes. Thus, the prolyl-AnaD
dehydrogenase are likely evolutionarily related to isovaleryl-
CoA dehydrogenases.
Walsh and co-workers clearly showed that CloN3, the flavo-
protein involved in cloromycin biosynthesis, oxidized the prolyl-
ACP to the four-electron oxidation state, that is, pyrrolyl-
ACP (38). Based on early feeding experiments (56), reasonable
chemistry (15), and the putative functions of the PKSs involved in
anatoxin-a biosynthesis, we speculated that AnaB would only
oxidize the prolyl-AnaD to dehydroprolyl-AnaD and possibly
would form the (S)-pyrroline-5-carboxyl-AnaD (P5C-AnaD). As
noted by others (38), it is not easy to detect a two-electron
oxidation product bound to an ACP. Furthermore, prolyl-
AnaD, the AnaB substrate, is unstable and decomposes into
holo-AnaD and presumably proline. We thus used mass spectro-
metry to analyze the product formed after oxidation by AnaB,
using oxygen as the electron acceptor because alternate oxidant
such as PMS were not superior in the conditions tested. Digestion
by trypsin followed by LC-MS/MS analysis unambiguously
showed that the two-electron oxidation reaction took place on
the proline moiety. Therefore, while AnaB is homologous to
CloN3 and the other prolyl-dehydrogenases that oxidize the
prolyl-ACP to pyrrolyl-ACP, it is the first prolyl-ACP dehydro-
genase that only oxidizes prolyl-ACP to the pyrroline oxidation
state. However, we could not determine, in this study, the
pyrroline regioisomer that was formed.
due to a reduction in kcat. Thus, AnaD discriminates among the
amino acid substrates on the catalytic constant, that is, on
transition state stabilization rather than on the substrate recogni-
tion. Because there are numerous adenylation proteins or do-
mains involved in NRPS, PKS, or NRPS/PKS hybrids, several
authors have tried to find a simple code that could predict the
specificity of a particular adenylation protein or domain from its
amino acid sequence (40, 52). These codes were based on the
three-dimensional structure of GrsA (39), a Phe-specific adenyla-
tion domain, and on the amino acid sequences of several
adenylation proteins or domains for which the specificity was
known. The two different proposed codes for proline-specific
adenylation proteins, shown in Table 3, were similar with only
one position (position 278) occupied by different residues. We
thus aligned the amino acid sequences of AnaC homologues
together with GrsA (Supporting Information Figure S2 and
Table S2) and found that none of the proposed codes were
respected in these proline-specific adenylation proteins. Another
putative code (Table 3) actually emerged from these alignments.
Of course, we compared quite similar proteins (Supporting
Information Table S2), but the fact is that the previous codes
are not predictive. Determination of the three-dimensional
structure of AnaC will certainly help in understanding the
structural determinants for the amino acid discrimination in this
class of enzymes. However, as we showed for the discrimination
of proline versus 3,4-Dhp by AnaC, the driving force is the
stabilization of the transition state rather than the stabilization of
the substrate. Thus, it is perhaps impossible to predict the
specificity just by modeling alternate substrate in the active site
and by looking at favorable or unfavorable interactions. It is not
surprising that simple codes based on primary sequences are thus
not very helpful in predicting specificities.
BLAST searches have shown that the anaB gene product
belonged to the ACAD superfamily and was related to prolyl-
ACP dehydrogenases implicated in the biosynthesis of pyrrole-
containing secondary metabolites: pyoluteorin (42), coumermy-
cin (43), prodiginine (44), clorobiocin (45), prodigiosin (46),
pyrrolomycin and dioxapyrrolomycin (49), Dkxanthene (50),
and indanomycin (51). These genes, coding for prolyl dehydro-
genases, are all found in clusters of genes responsible for the
complete biosynthesis of the corresponding secondary metabo-
lites. It is quite interesting to note that the genetic context is
conserved in some of these clusters of genes (ana (15), cou (43),
clo (45), and idm (51)), with the following gene order, prolyl-ACP
dehydrogenase gene, adenylation protein gene, and ACP gene,
Our mechanistic proposal is depicted in Scheme 2. This
proposition is based on the homology of AnaB with ACAD
and on the proposed biosynthesis of anatoxin-a in which the (S)-
pyrroline-5-carboxylate thioester is believed to be the electrophile
in the cyclization reaction catalyzed by the PKS AnaF
(Scheme 1). The isomerization proposed here is reminiscent
of the allylic isomerization catalyzed by the ACAD enzymes
(57, 58). It is interesting to note that oxidation of proline to P5C is

1
12 Biochemistry, Vol. 49, No. 1, 2010
M ꢀe jean et al.
a common reaction in the catabolism of proline and that the
enzymes catalyzing this oxidation reaction are flavoenzymes (29).
However, this class of enzymes, the so-called proline dehydro-
genase (ProDh) family, shares no sequence similarity with AnaB
and related prolyl-ACP dehydrogenases. Thus, AnaB and related
enzymes would be evolutionarily related to ACAD rather than to
ProDhs. The difference between AnaB and the related prolyl-
ACP dehydrogenases would be that the former stops at the
dehydroproline oxidation state and that the latter fully oxidize
the substrate to pyrrolyl-ACP. We are currently trying to identify
the isomer formed after AnaB oxidation of prolyl-AnaD and also
to find the best electron acceptor in this reaction. Determination
of the three-dimensional structure of this interesting enzyme is
also underway.
genomic structure of the bloom-forming toxic cyanobacterium Mi-
crocystis aeruginosa NIES-843. DNA Res. 14, 247–256.
1. Frangeul, L., Quillardet, P., Castets, A. M., Humbert, J. F., Matthijs,
H. C., Cortez, D., Tolonen, A., Zhang, C. C., Gribaldo, S., Kehr, J.
C., Zilliges, Y., Ziemert, N., Becker, S., Talla, E., Latifi, A., Billault,
A., Lepelletier, A., Dittmann, E., Bouchier, C., and de Marsac, N. T.
1
(2008) Highly plastic genome of Microcystis aeruginosa PCC 7806,
a ubiquitous toxic freshwater cyanobacterium. BMC Genomics 9, 274.
2. Tillett, D., Dittmann, E., Erhard, M., von D o€ hren, H., B o€ rner, T.,
and Neilan, B. A. (2000) Structural organization of microcystin
biosynthesis in Microcystis aeruginosa PCC7806: an integrated pep-
tide-polyketide synthetase system. Chem. Biol. 7, 753–764.
3. Noguchi, T., Shinohara, A., Nishizawa, A., Asayama, M., Nakano,
T., Hasegawa, M., Harada, K., Nishizawa, and T., Shirai, M. (2009)
Genetic analysis of the microcystin biosynthesis gene cluster in
Microcystis strains from four bodies of eutrophic water in Japan.
J. Gen. Appl. Microbiol. 55, 111–123.
1
1
1
4. Kellmann, R., Mihali, T. K., Jeon, Y. J., Pickford, R., Pomati, F., and
Neilan, B. A. (2008) Biosynthetic intermediate analysis and functional
homology reveal a saxitoxin gene cluster in cyanobacteria. Appl.
Environ. Microbiol. 74, 4044–4053.
ACKNOWLEDGMENT
15. Mejean, A., Mann, S., Maldiney, T., Vassiliadis, G., Lequin, O., and
ꢀ
We thank Y. Raoul des Essarts for technical help during the
purification and characterization of the OsPPT, L. Giorgi for
Ploux, O. (2009) Evidence that biosynthesis of the neurotoxic alka-
loids anatoxin-a and homoanatoxin-a in the cyanobacterium Oscilla-
toria PCC 6506 occurs on a modular polyketide synthase initiated by
help in performing the PP -ATP exchange reaction, C. M ꢀe digue
i
and her team (G ꢀe noscope, Evry) for help during bioinformatic
L-proline. J. Am. Chem. Soc. 131, 7512–7513.
1
1
1
6. Cadel-Six, S., Iteman, I., Peyraud-Thomas, C., Mann, S., Ploux, O.,
and M ꢀe jean, A. (2009) Identification of a polyketide synthase coding
sequence specific for anatoxin-a-producing Oscillatoria cyanobacter-
ia. Appl. Environ. Microbiol. 75, 4909–4912.
7. 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.
analysis of the PCC 6506 genome data, and P. F. Leadlay
(
Cambridge, U.K.) for providing the Sfp overproducing strain.
SUPPORTING INFORMATION AVAILABLE
Sequence alignments and BLAST scores for OsPPTase, AnaC,
AnaB, and AnaD (Figures S1-S4 and Tables S1-S4) and
determination of the sequence of the tryptic fragment of AnaD
containing serine 41 (Figure S5). This material is available free of
charge via the Internet at http://pubs.acs.org.
8. Herdman, M., Iteman, I., and Rippka, R. (2005) Catalogue of
cyanobacterial strains , 2nd ed., Institut Pasteur, Paris.
19. Rippka, R. (1988) Isolation and purification of cyanobacteria. Meth-
ods Enzymol. 167, 3–27.
2
0. 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.
REFERENCES
2
1. Vallenet, D., Labarre, L., Rouy, Z., Barbe, V., Bocs, S., Cruveiller, S.,
Lajus, A., Pascal, G., Scarpelli, C., and M ꢀe digue, C. (2006) MaGe: a
microbial genome annotation system supported by synteny results.
Nucleic Acids Res. 34, 53–65.
1. van Apeldoorn, M. E., van Egmond, H. P., Speijers, G. J., and
Bakker, G. J. (2007) Toxins of cyanobacteria. Mol. Nutr. Food Res.
5
1, 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.
22. Ellis, K. E., and Morrison, J. F. (1982) Buffers of constant ionic
strength for studying pH-dependent processes. Methods Enzymol. 87,
405–426.
2
3. Gerratana, B., Arnett, S. O., Stapon, A., and Townsend, C. A. (2004)
Carboxymethylproline synthase from Pectobacterium carotorova: a
multifaceted member of the crotonase superfamily. Biochemistry 43,
3
4
. Gunn, G. J., Rafferty, A. G., Rafferty, G. C., Cockburn, N., Edwards,
C., Beattie, K. A., and Codd, G. A. (1992) Fatal canine neurotoxicosis
attributed to blue-green algae (cyanobacteria). Vet. Rec. 130, 301–
15936–15945.
3
02.
2
2
2
4. Gerratana, B., Stapon, A., and Townsend, C. A. (2003) Inhibition and
alternate substrate studies on the mechanism of carbapenam synthe-
tase from Erwinia carotovora. Biochemistry 42, 7836–7847.
5. Williams, I., and Frank, L. (1975) Improved chemical synthesis and
enzymatic assay of delta-1-pyrroline-5-carboxylic acid. Anal. Bio-
chem. 64, 85–97.
6. Mezl, V. A., and Knox, W. E. (1976) Properties and analysis of a
stable derivative of pyrroline-5-carboxylic acid for use in metabolic
studies. Anal. Biochem. 74, 430–440.
. 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,
9
19–928.
5
. Cadel-Six, S., Peyraud-Thomas, C., Brient, L., Tandeau de Marsac,
N., Rippka, R., and M ꢀe jean, A. (2007) Different genotypes of
anatoxin-producing cyanobacteria coexist in the Tarn River, France.
Appl. Environ. Microbiol. 73, 7605–7614.
6
. 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.
27. Ehmann, D. E., Trauger, J. W., Stachelhaus, T., and Walsh, C. T.
(2000) Aminoacyl-SNACs as small-molecule substrates for the con-
densation domains of nonribosomal peptide synthetases. Chem. Biol.
7, 765–772.
7
8
9
. Puschner, B., Hoff, B., and Tor, E. R. (2008) Diagnosis of anatoxin-a
poisoning in dogs from North America. J. Vet. Diagn. Invest. 20,
28. Becker, D. F., and Thomas, E. A. (2001) Redox properties of the PutA
protein from Escherichia coli and the influence of the flavin redox state
on PutA-DNA interactions. Biochemistry 40, 4714–21.
29. White, T. A., Krishnan, N., Becker, D. F., and Tanner, J. J. (2007)
Structure and kinetics of monofunctional proline dehydrogenase
from Thermus thermophilus. J. Biol. Chem. 282, 14316–14327.
30. Wilm, M., and Mann, M. (1996) Analytical properties of the nanoe-
lectrospray ion source. Anal. Chem. 68, 1–8.
31. Lambalot, R. H., Gehring, A. M., Flugel, R. S., Zuber, P., LaCelle,
M., Marahiel, M. A., Reid, R., Khosla, C., and Walsh, C. T. (1996) A
new enzyme superfamily;the phosphopantetheinyl transferases.
Chem. Biol. 3, 923–936.
32. Weissman, K. J., Hong, H., Oliynyk, M., Siskos, A. P., and Leadlay,
P. F. (2004) Identification of a phosphopantetheinyl transferase for
8
9–92.
. Wonnacott, S., and Gallagher, T. (2006) The chemistry and pharma-
cology of anatoxin-a and related homotropanes with respect to
nicotinic acetylcholine receptors. Mar. Drugs 4, 228–254.
. Chorus, I., and Bartram, J. (1999) Toxic Cyanobacteria in Water: A
Guide to their Public Health Consequences, Monitoring and Manage-
ment, E&FN Spon, London.
1
0. Kaneko, T., Nakajima, N., Okamoto, S., Suzuki, I., Tanabe, Y.,
Tamaoki, M., Nakamura, Y., Kasai, F., Watanabe, A., Kawashima,
K., Kishida, Y., Ono, A., Shimizu, Y., Takahashi, C., Minami, C.,
Fujishiro, T., Kohara, M., Katoh, M., Nakazaki, N., Nakayama, S.,
Yamada, M., Tabata, S., and Watanabe, M. M. (2007) Complete

Article
Biochemistry, Vol. 49, No. 1, 2010 113
erythromycin biosynthesis in Saccharopolyspora erythraea. ChemBio-
Chem 5, 116–125.
46. 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.
47. 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.
3
3
3. Copp, J. N., and Neilan, B. A. (2006) The phosphopantetheinyl
transferase superfamily: phylogenetic analysis and functional impli-
cations in cyanobacteria. Appl. Environ. Microbiol. 72, 2298–2305.
4. Quadri, L. E., Weinreb, P. H., Lei, M., Nakano, M. M., Zuber, P., and
Walsh, C. T. (1998) Characterization of Sfp, a Bacillus subtilis
phosphopantetheinyl transferase for peptidyl carrier protein domains
in peptide synthetases. Biochemistry 37, 1585–1595.
3
3
3
5. Copp, J. N., Roberts, A. A., Marahiel, M. A., and Neilan, B. A. (2007)
Characterization of PPTNs, a cyanobacterial phosphopantetheinyl
transferase from Nodularia spumigena NSOR10. J. Bacteriol. 189,
48. 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.
49. Zhang, X., and Parry, R. J. (2007) Cloning and characterization of the
pyrrolomycin biosynthetic gene clusters from Actinosporangium vita-
minophilum ATCC 31673 and Streptomyces sp. strain UC 11065.
Antimicrob. Agents Chemother. 51, 946–957.
50. Meiser, P., Weissman, K. J., Bode, H. B., Krug, D., Dickschat, J. S.,
Sandmann, A., and M u€ ller, R. (2008) DKxanthene biosynthesis:
understanding the basis for diversity-oriented synthesis in myxobac-
terial secondary metabolism. Chem. Biol. 15, 771–781.
51. 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.
3
133–3139.
6. Roberts, A. A., Copp, J. N., Marahiel, M. A., and Neilan, B. A. (2009)
The Synechocystis sp. PCC6803 Sfp-type phosphopantetheinyl trans-
ferase does not possess characteristic broad-range activity. ChemBio-
Chem 10, 1869–1877.
7. 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.
3
8. Garneau, S., Dorrestein, P. C., Kelleher, N. L., and Walsh, C. T.
52. Challis, G. L., Ravel, R., and Townsend, C. A. (2000) Predictive,
structure-based model of amino acid recognition by nonribosomal
peptide synthetase adenylation domains. Chem. Biol. 7, 211–224.
53. Kim, J. J., Wang, M., and Paschke, R. (1993) Crystal structures of
medium-chain acyl-CoA dehydrogenase from pig liver mitochondria
with and without substrate. Proc Natl. Acad. Sci. U.S.A. 90, 7523–
7527.
(
2005) Characterization of the formation of the pyrrole moiety during
clorobiocin and coumermycin A1 biosynthesis. Biochemistry 44,
770–2780.
2
3
4
4
4
4
4
9. Conti, E., Stachelhaus, T., Marahiel, M. A., and Brick, P. (1997)
Structural basis for the activation of phenylalanine in the non-
ribosomal biosynthesis of gramicidin S. EMBO J. 16, 4174–4183.
0. Stachelhaus, T., Mootz, H. D., and Marahiel, M. A. (1999) The
specificity-conferring code of adenylation domains in non-ribosomal
peptide synthetases. Chem. Biol. 6, 493–505.
1. Villiers, B. R., and Hollfelder, F. (2009) Mapping the limits of
substrate specificity of the adenylation domain of TycA. ChemBio-
Chem 10, 671–682.
2. Nowak-Thompson, B., Chaney, N., Wing, J.S., Gould, S.J., and Loper,
J. E. (1999) Characterizationof the pyoluteorin biosynthetic gene cluster
of Pseudomonas fluorescens Pf-5. J. Bacteriol. 181, 2166–2174.
3. Wang, Z. X., Li, S. M., and Heide, L. (2000) Identification of the
coumermycin A(1) biosynthetic gene cluster of Streptomyces rishir-
iensis DSM 40489. Antimicrob. Agents Chemother. 44, 3040–3048.
4. Cerde n~ o, 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.
54. 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 A resolution: structural basis for substrate
specificity. Biochemistry 36, 8455–8464.
55. Battaile, K. P., Nguyen, T. V., Vockley, J., and Kim, J. J. (2004)
Structures of isobutyryl-CoA dehydrogenase and enzyme-product
complex: comparison with isovaleryl- and short-chain acyl-CoA
dehydrogenases. J. Biol. Chem. 279, 16526–16534.
56. Hemscheidt, T., Rapala, J., Sivonen, K., and Skulberg, O. M. (1995)
Biosynthesis of anatoxin-a in Anabaena flos-aquae and homoanatox-
in-a in Oscillatoria formosa. J. Chem. Soc., Chem. Commun., 1361–
1362.
57. 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.
4
5. 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.
58. Dakoji, S., Shin, I., Becker, D. F., Stankovich, M. T., and Liu, H.-w.
(1996) Studies of acyl-CoA dehydrogenase catalyzed allylic isomer-
ization: a one-base or two-base mechanism? J. Am. Chem. Soc. 118,
10971–10979.