Structural and functional characterization of plant aminoaldehyde dehydrogenase from pisum sativum with a broad specificity for natural and synthetic aminoaldehydes

Article abstract of DOI:10.1016/j.jmb.2009.12.015

Aminoaldehyde dehydrogenases (AMADHs, EC 1.2.1.19) belong to the large aldehyde dehydrogenase (ALDH) superfamily, namely, the ALDH9 family. They oxidize polyamine-derived ω-aminoaldehydes to the corresponding ω-amino acids. Here, we report the first X-ray structures of plant AMADHs: two isoenzymes, PsAMADH1 and PsAMADH2, from Pisum sativum in complex with β-nicotinamide adenine dinucleotide (NAD⁺) at 2.4 and 2.15 A resolution, respectively. Both recombinant proteins are dimeric and, similarly to other ALDHs, each monomer is composed of an oligomerization domain, a coenzyme binding domain and a catalytic domain. Each subunit binds NAD⁺ as a coenzyme, contains a solvent-accessible C-terminal peroxisomal targeting signal (type 1) and a cation bound in the cavity close to the NAD⁺ binding site. While the NAD⁺ binding mode is classical for PsAMADH2, that for PsAMADH1 is unusual among ALDHs. A glycerol molecule occupies the substrate binding site and mimics a bound substrate. Structural analysis and substrate specificity study of both isoenzymes in combination with data published previously on other ALDH9 family members show that the established categorization of such enzymes into distinct groups based on substrate specificity is no more appropriate, because many of them seem capable of oxidizing a large spectrum of aminoaldehyde substrates. PsAMADH1 and PsAMADH2 can oxidize N,N,N-trimethyl-4-aminobutyraldehyde into γ-butyrobetaine, which is the carnitine precursor in animal cells. This activity highly suggests that in addition to their contribution to the formation of compatible osmolytes such as glycine betaine, β-alanine betaine and γ-aminobutyric acid, AMADHs might participate in carnitine biosynthesis in plants.

Full text of DOI:10.1016/j.jmb.2009.12.015

doi:10.1016/j.jmb.2009.12.015
J. Mol. Biol. (2010) 396, 870–882
Available online at www.sciencedirect.com
Structural and Functional Characterization of Plant
Aminoaldehyde Dehydrogenase from Pisum sativum
with a Broad Specificity for Natural and
Synthetic Aminoaldehydes
Martina Tylichová1, David Kopečný1 , Solange Moréra2 ,
⁎
⁎
Pierre Briozzo2,3, René Lenobel4, Jacques Snégaroff3
and Marek Šebela1
⁎
1Department of Biochemistry,
Faculty of Science, Palacký
University, Šlechtitelů 11,
CZ-783 71 Olomouc,
Aminoaldehyde dehydrogenases (AMADHs, EC 1.2.1.19) belong to the
large aldehyde dehydrogenase (ALDH) superfamily, namely, the ALDH9
family. They oxidize polyamine-derived ω-aminoaldehydes to the
corresponding ω-amino acids. Here, we report the first X-ray structures of
plant AMADHs: two isoenzymes, PsAMADH1 and PsAMADH2, from
Pisum sativum in complex with β-nicotinamide adenine dinucleotide
(NAD+) at 2.4 and 2.15 Å resolution, respectively. Both recombinant
proteins are dimeric and, similarly to other ALDHs, each monomer is
composed of an oligomerization domain, a coenzyme binding domain and
a catalytic domain. Each subunit binds NAD+ as a coenzyme, contains a
solvent-accessible C-terminal peroxisomal targeting signal (type 1) and a
cation bound in the cavity close to the NAD+ binding site. While the NAD+
binding mode is classical for PsAMADH2, that for PsAMADH1 is unusual
among ALDHs. A glycerol molecule occupies the substrate binding site and
mimics a bound substrate. Structural analysis and substrate specificity
study of both isoenzymes in combination with data published previously
on other ALDH9 family members show that the established categorization
of such enzymes into distinct groups based on substrate specificity is no
more appropriate, because many of them seem capable of oxidizing a large
spectrum of aminoaldehyde substrates. PsAMADH1 and PsAMADH2 can
oxidize N,N,N-trimethyl-4-aminobutyraldehyde into γ-butyrobetaine,
which is the carnitine precursor in animal cells. This activity highly
suggests that in addition to their contribution to the formation of compatible
osmolytes such as glycine betaine, β-alanine betaine and γ-aminobutyric
acid, AMADHs might participate in carnitine biosynthesis in plants.
© 2009 Elsevier Ltd. All rights reserved.
Czech Republic
2Laboratoire d'Enzymologie et
Biochimie Structurales (LEBS),
CNRS, F-91198 Gif-sur-Yvette
Cedex, France
3UMR206 Laboratoire
de Chimie Biologique,
INRA-AgroParisTech, F-78850
Thiverval-Grignon, France
4Laboratory of Growth
Regulators, Palacký University
and Institute of Experimental
Botany AS CR, CZ-783 71
Olomouc, Czech Republic
Received 29 October 2009;
received in revised form
8 December 2009;
accepted 10 December 2009
Available online
21 December 2009
Keywords: aminoaldehyde dehydrogenase; betaine aldehyde dehydrogenase;
Edited by G. Schulz
NAD+ complex; carnitine; protein crystallography
*Corresponding authors. E-mail addresses: kopecny_david@yahoo.co.uk; morera@lebs.cnrs-gif.fr;
marek.sebela@upol.cz.
Abbreviations used: AAS, atomic absorption spectrometry; ABAL, 4-aminobutyraldehyde; AHBAL,
4-amino-2-hydroxybutyraldehyde; ALDH, aldehyde dehydrogenase; AMADH, aminoaldehyde dehydrogenase;
APAL, 3-aminopropionaldehyde; BADH, betaine aldehyde dehydrogenase; BAL, betaine aldehyde; deamino NAD+,
nicotinamide hypoxanthine dinucleotide; GABA, 4-aminobutyric acid; GBAL, 4-guanidinobutyraldehyde; GHBAL,
4-guanidino-2-hydroxybutyraldehyde; GPAL, 3-guanidinopropionaldehyde; NAD+, β-nicotinamide adenine dinucleotide;
NADP+, β-nicotinamide adenine dinucleotide phosphate; NiR, nicotinamide riboside; PCAL, pyridine carboxaldehyde;
TMABAL, N,N,N-trimethyl-4-aminobutyraldehyde; TMAPAL, N,N,N-trimethyl-3-aminopropionaldehyde.
0022-2836/$ - see front matter © 2009 Elsevier Ltd. All rights reserved.

Crystal Structure of Plant Aminoaldehyde Dehydrogenase
871
conserved catalytic cysteine residue. A sequential
binding model for ALDHs catalysis in which the
cofactor binds prior to the aldehyde substrate has
been proposed.11 The reaction mechanism of ALDH
superfamily members including AMADHs involves
nucleophilic activation and attack on the aldehyde
substrate by the catalytic cysteine forming a
thiohemiacetal intermediate.12 In the following
step, a hydride is transferred to the nicotinamide
ring C4 atom of NAD(P)+ yielding NAD(P)H and a
thioester. In the last step, a conserved glutamate
residue functioning as a general base activates a
water molecule, which performs nucleophilic attack
at the thioester acyl–sulfur bond resulting in the
release of the corresponding acid.
PsAMADH1 and PsAMADH2 from pea (Pisum
sativum) are two AMADH isoforms of 503 amino
acids sharing 80% sequence identity.13,14 Although
both PsAMADHs possess the C-terminal peroxi-
some targeting sequence signal type 1 [S/A]KL,
their intracellular localization is still unknown. The
presence of two isoenzymes is common in plants.
Analysis of numerous sequences for plant
AMADHs suggests the coexistence of peroxisomal
and nonperoxisomal isoforms in dicots and
monocots.15 Biochemical characterization of native
PsAMADH1 and its substrate specificity showed
that it preferentially oxidized C3 and C4 aminoal-
dehydes and had no activity with BAL.13
To shed light on the substrate specificity of both
PsAMADHs, we performed kinetic and X-ray
crystallographic studies. We report the structures
of PsAMADH1 and PsAMADH2 in complex with
NAD+ coenzyme at 2.4 and 2.15 Å resolution,
respectively. The structures show that both enzymes
are dimeric and provide a detailed description of the
coenzyme binding site. Although cocrystallization
or soaking methods with substrates were attempted,
obtaining AMADHs in complex with substrates was
unsuccessful as commonly observed for other
ALDHs. However, a glycerol molecule close to the
catalytic cysteine mimics a substrate bound to the
substrate binding site. In addition, one metal ion
occupies a conserved binding site near the coen-
zyme site. Enzyme kinetics in combination with
structural analysis provides novel evidence that pea
AMADHs might be involved in the biosynthesis of
carnitine (3-hydroxy-4-N-trimethylaminobutyric
acid) via γ-butyrobetaine generated from TMABAL.
Introduction
Aminoaldehyde dehydrogenases (AMADHs) be-
long to the aldehyde dehydrogenase 9 family
(ALDH9) within the ALDH superfamily.1,2 They
catalyze the oxidation of a broad range of ω-
aminoaldehydes to the corresponding ω-amino
acids in an NAD(P)+-dependent manner. The
ALDH9 family interconnects three unrelated meta-
bolic pathways—metabolism of polyamines, choline
and lysine—which has led to a classification of
ALDH9 members based on substrate specificity into
distinct groups: 4-aminobutyraldehyde (ABAL)
dehydrogenases (ABALDHs, EC 1.2.1.19), 4-guani-
dinobutyraldehyde (GBAL) dehydrogenases
(GBALDHs, EC 1.2.1.54), betaine aldehyde (BAL)
dehydrogenases (BADHs, EC 1.2.1.8) and 4-tri-
methylaminobutyraldehyde dehydrogenases
(TMABALDHs, EC 1.2.1.47). Polyamines such as
spermine, spermidine and putrescine are oxidized
by amine oxidases, which results in ω-aminoalde-
hydes and hydrogen peroxide. Choline is oxidized
by choline oxidase or choline monooxygenase to
BAL. Metabolism of lysine through its methylation,
subsequent hydroxylation and aldolytic cleavage
leads to N,N,N-trimethyl-4-aminobutyraldehyde
(TMABAL). Thus, all three pathways end up with
aminoaldehyde compounds that are in turn oxi-
dized by enzymes from the ALDH9 family. Since the
enzymes have been shown to oxidize a wide range
of aminoaldehyde substrates (reviewed by Tyli-
chová et al.3) such as BAL, TMABAL, APAL (3-
aminopropionaldehyde), ABAL, GBAL and several
aromatic and aliphatic aldehydes, the current
classification system is no more appropriate. For
example, the Escherichia coli YdcW gene product,
called BADH or medium-chain ALDH,4 has been
found to accept ABAL as the best substrate.5 For that
reason, the term E. coli AMADH will be used further
in the text. Other BADHs from different species have
also been reported to prefer APAL or ABAL to
BAL.6–8 Although substrate specificity can vary
among species, the above-mentioned enzymes
should in general be considered as AMADHs.
To date, the crystal structures of numerous NAD
(P)+-dependent ALDH enzymes from different
species are available in the Protein Data Bank
(PDB). Among the ALDH9 family members, three
enzymes have been characterized by X-ray diffrac-
tion, resulting in a total of five structures including
four ligand-containing structures: BADH from cod
liver was solved in the apoenzyme form and in a
complex with β-nicotinamide adenine dinucleotide
(NAD+) (PDB codes 1A4S and 1BPW),9 E. coli
AMADH in the apo form and in a ternary complex
with NADH and BAL substrate (PDB codes 1WND
and 1WNB)4 and BADH from Pseudomonas aerugi-
nosa (PaBADH) in a complex with NAD+ phosphate
(NADP+) and glycerol (PDB code 2VE5).10 All of
these enzymes are tetrameric. Their subunits possess
the characteristic ALDH fold consisting of an
oligomerization domain, an NAD(P)+ coenzyme
binding domain and a catalytic domain with a
Results and Discussion
Structure of PsAMADH1 and PsAMADH2
The crystal structures of PsAMADH1 (PDB code
3IWK) and PsAMADH2 (PDB code 3IWJ) represent
the first examples of plant AMADHs. The asymmetric
unit of PsAMADH1 crystal contains six very similar
dimers with an average root mean square deviation
(RMSD) of 0.35 Å for all Cα atoms. The asymmetric
unit of PsAMADH2 crystal contains only one dimer

872
Crystal Structure of Plant Aminoaldehyde Dehydrogenase
Table 1. Data collection and refinement statistics of pea
that PsAMADHs resemble the numerous ALDH
structures from different species with RMSD values
ranging from 1.0 to 1.6 Å over 450–460 Cα atoms
corresponding to a sequence identity between 30% and
45%. PsAMADHs are structurally closer to human
mitochondrial ALDH (PDB code 1CW3),18 which was
used as a search model for molecular replacement,
than to the three available members of the ALDH9
family. Indeed, a superposition of PsAMADH1 with
BADH from cod liver (PDB code 1BPW),9 E. coli
AMADH (PDB code 1WNB)4 and Pseudomonas
aeruginosa BADH (PaBADH, PDB code 2VE5)10 led
to RMSD between 1.2 and 1.35 Å for 460 Cα atoms. The
dimer of PsAMADHs is formed in a fashion similar to
those of other ALDHs. The oligomerization domains
protrude over the surface of the neighboring subunits
and connect both monomers, while the NAD+ binding
site is on the opposite side of the dimer (Fig. 1). So far,
ALDHs are known as dimeric or tetrameric enzymes.
Dissociation of the tetrameric PaBADH into mono-
mers led to a loss of activity.19,20 Therefore, the
minimum functional oligomer amongst ALDH super-
family members is probably a dimer, as the dissocia-
tion into individual subunits seems to be incompatible
with the catalytic function.
AMADHs
Data set of
PsAMADH1
PsAMADH2
Space group
Asymmetric unit
Unit cell (Å)
a
P21
P212121
6 dimers
1 dimer
86.4
216.9
66.3
86.9
b
c
205.8
179.3
α (°)
90.0
90.0
β (°)
98.02
90.0
γ (°)
90.0
90.0
Resolution (Å)
Observed reflections
Unique reflections
Completeness (%)
I/σ (I)
50.0–2.40
1,941,734
290,858
99.7 (99.5)a
14.1 (3.7)
10.3 (62.2)
20.0
50.0–2.15
251,226
56,568
99.2 (99.5)
14.7 (3.4)
5.9 (44.2)
23.4
Rsym (%)
Rcryst (%)
Rfree (%)
24.4
28.9
RMSD
Bond lengths (Å)b
Bond angles (°)
Solvent molecules
Mean B value (Å2)
Overall
0.007
1.114
2017
0.008
1.168
297
34.7
34.3
70.7
67.0
47.5
35.2
39.3
39.1
—
37.3
39.8
43.3
Protein
NAD
ADP part
Na+ ion
Solvent
Cation binding site
Ramachandran statistics (%)b
Most favorable regions
Additional allowed regions
90.8
8.3
90.7
8.5
In both PsAMADHs, only one cation is present in a
conserved intrasubunit cavity close to the coenzyme
binding site. This ion site was proposed to be a
common feature of ALDHs10 and was clearly
identified in human ALDH2 with a Na+ ion (PDB
code 1O01)11 and in PaBADH with a K+ ion (PDB
code 2VE5).10 The site in ALDH2/PaBADH is
formed by the side chains of Thr39/Thr26 and
Asp109/Asp93, three carbonyl groups of Val40/
Ile27, Asp109/Asp93 and Gln196/Val180 and one
water molecule. The cation binding site of human
ALDH2 is shown in Fig. 2a. In both PsAMADHs, the
conserved cation cavity is formed by the three
equivalent carbonyl groups of Ile28, Asp99 and
Leu189 and the side chain of Asp99 (Fig. 2b and c).
The side chain of Thr39/Thr26 in ALDH2/PaBADH
is replaced by the polar side chain of Asn27 in
PsAMADH1 and by the nonpolar side chain of Val27
in PsAMADH2, which loses the interaction with the
cation. Both the nature of the protein ligands and the
distances of around 2.4–2.6 Å are typical for Na+
bound to proteins.21 Sodium ions were present in the
purification and crystallization buffers. Analysis of
pure PsAMADHs used for crystallization by atomic
absorption spectrometry (AAS) revealed a content of
2.5 mol of Na+ per mole of the dimeric enzyme,
which roughly gives one Na+ ion per monomer.
Although the role of this monovalent ion in ALDHs
is not clearly understood, it could stabilize, through
its coordination, the position of a loop involved in
NAD+ binding (Lys185, Ser187 and Glu188 in
PsAMADHs), contributing to the optimal active-site
conformation for the coenzyme binding.11. In PsA-
MADHs, the observed partial loss of activity after
dialysis of the enzyme during purification procedure
a
Numbers in parentheses represent values in the highest-
resolution shell: 2.53–2.40 Å for PsAMADH1, 2.26–2.15 Å for
PsAMADH2, last of 10 shells.
b
As evaluated by PROCHECK.16
(Table 1). This suggests that the active form in
solution is a dimer in line with the gel-filtration
measurement leading to molecular mass estimates of
120 and 117 kDa for PsAMADH1 and PsAMADH2,
respectively (Supplementary Figure S1). PsAMADH1
and PsAMADH2 dimers are very similar to each
other with an average RMSD of 0.68 Å for all Cα
atoms (Fig. 1). Omitting a very flexible loop (residues
59–66) observed in the PsAMADH1 structure and
helix αD (residues 219–230; see the Coenzyme
binding site section) in contact with the adenine part
of the coenzyme decreases the RMSD to 0.5 Å.
All monomers of both PsAMADHs are identical
and adopt the characteristic ALDH fold consisting of
three domains. The coenzyme binding domain in
PsAMADHs is formed by residues 1–131, 152–261
and 453–479. The catalytic domain spreads from
residue 262 to 452. Central parts of both domains
consist of five β strands and four α helices. The
oligomerization domaincomprises two long β strands
and one short β strand formed by residues 132–151
and 480–503. The C-terminus is slightly longer than C-
termini in other known ALDH structures because it
carries the type 1 peroxisomal targeting sequence.
A structural comparison of the PsAMADH mono-
mer to all entries in the PDB using SSM-EBI†17 shows
† http://www.ebi.ac.uk/mrd-srv/ssm

Crystal Structure of Plant Aminoaldehyde Dehydrogenase
873
Fig. 1. Structure of PsAMADHs. (a) Ribbon representation of PsAMADH1 dimer. The catalytic domain, the coenzyme
binding domain and the oligomerization domain are shown in pale grey, black and blue, respectively. The NAD+
coenzyme (in atom-coded colors) and metal ion (in green) are shown in spheres. (b) Ribbon representation of PsAMADH2
dimer. The catalytic domain, the coenzyme binding domain and the oligomerization domain are shown in beige, brown
and violet, respectively. The peroxisomal signal sequence (S/A)KL at the C-terminus is shown in red.
could result from a loss of Na+ ion accompanied by
loosening of structural integrity.
PsAMADH2 structure. A disordered flexible nico-
tinamide riboside (NiR) moiety of NAD+ is quite
common in crystal structures of NAD(P)+-depen-
dent ALDHs.10,22 PsAMADH2 possesses a NAD+
binding mode similar to that described for ALDH
enzymes. The adenine base is inserted in a hydro-
phobic pocket located between helix αD and helix αE
Coenzyme binding site
Only the ADP moiety of the NAD+ coenzyme is
well defined in density in both subunits of the
Fig. 2. Cation binding site. Na+ in orange sphere is shown bound to (a) human ALDH2 (PDB code 1O01), (b)
PsAMADH1, and (c) PsAMADH2. Metal coordination is shown with dashed lines and distances are indicated in
angstroms.

874
Crystal Structure of Plant Aminoaldehyde Dehydrogenase
Fig. 3. NAD+ binding site.
NAD+ binding in the coenzyme
cavity of PsAMADH2 (a) and of
PsAMADH1 (b). PsAMADH1 is
pale grey and PsAMADH2 is light
brown. NAD+ molecules are shown
in yellow and atom-coded color
sticks in their Fo −Fc omit maps
calculated from the deposited struc-
tures and contoured at 2.5σ. In
PsAMADH2, only the ADP moiety
was ordered. Side-chain residues
and secondary structure are la-
beled. Hydrogen bonds are shown
as broken lines (for distances less
than 3.3 Å). (c) Superposition of
coenzyme conformations. NAD+ in
PsAMADH1, PsAMADH2 and cod
liver BADH (PDB code 1BPW) is
shown as pale grey, light brown
and yellow sticks, respectively.
NADH in human ALDH (PDB
code 1O01) is shown as blue sticks.
The catalytic cysteine and gluta-
mate are shown in their respective
colors and labeled.

Crystal Structure of Plant Aminoaldehyde Dehydrogenase
875
being sandwiched between the main-chain group of
Gly218 and the Thr242 side chain (Fig. 3a). It does not
make a polar contact with the enzyme as observed
for the cod liver enzyme and PaBADH. The O2B and
O3B oxygen atoms of the ribose establish a hydrogen
bond with the side chain of Glu188 and the backbone
oxygen of Thr159, respectively, and they both
interact with the Lys185 side chain. Both α-phos-
phate oxygen atoms O1A and O2A bind to the OG1
atom of Thr242. O2A makes additional interactions
with the main-chain NH and the OG atom of Ser239.
One oxygen atom of β-phosphate is bound to NE1 of
Trp161.
the NiR moiety only, which could undergo large
conformational changes during catalysis.11,24,25
Several NAD+ analogs were tested with both
PsAMADHs (Table 2). NADP+ is a poor electron
acceptor compared to NAD+, reducing the activity
of both enzymes by 85%. In PsAMADH2, an
NADPH model based on bound NAD+ shows that
its 2′-phosphate clashes with the side chain of
Glu188. A rearrangement of the Glu188 side chain
as observed in PaBADH,10 which could permit the
2′-phosphate binding, is not possible due to the
three close bulky side chains of Ile25, Ile39 and
Leu217. For PsAMADH1, the 2′-phosphate clashes
with the Lys245 side chain, but a rearrangement of
its side chain is possible and it might interact with
the phosphate group, suggesting that no or little
binding difference between NAD+ and NADP+
would be expected. Therefore, the possibility that
this unusual NAD+ conformation observed in the
PsAMADH1 crystal structure represents a nonpro-
ductive binding conformation cannot be ruled out.
Among three analogs with substitutions at the
nicotinamide ring, 3-acetylpyridine-NAD+ is the
best electron acceptor and leads to 40% activity
compared to that with NAD+, while thio-NAD+
drastically and especially affects PsAMADH1 activ-
ity. 3-Pyridinealdehyde-NAD+ hardly functions as a
coenzyme for both PsAMADHs, which is consistent
with studies on glyceraldehyde 3-phosphate dehy-
drogenase, indicating that its aldehyde group might
form a thiohemiacetal linkage with the catalytic
cysteine preventing its reduction.28 Interestingly, a
hypoxanthine derivative of NAD+ (deamino NAD+)
is a better coenzyme than NAD+ and increases the
reaction rate by about 50%. In comparison with
NAD+, deamino NAD+ might form an additional
hydrogen bond between its hypoxanthine oxygen
and Lys245 side chain.
The whole NAD+ molecule was observed in 11 of
the 12 subunits of the PsAMADH1 structure. The
ADP moiety is well defined, while the electron
density is weaker for the NiR moiety. Although all
residues interacting with the bound coenzyme are
strictly conserved with PsAMADH2, the ADP part
of NAD+ adopts a striking conformation that has
never been observed in any ALDH structure. In
comparison with the usual ADP conformation, the
adenine base is flipped by 180° so that the N3–N9
side is exposed to the solvent. As a result, the ribose
and both phosphates lie outside the binding pocket
(Fig. 3b). The interactions between PsAMADH1
and NAD+ are almost equivalent in all 11 subunits
and NAD+ is not involved in crystal packing
contacts. The adenine base is stacked between the
main chains of Gly218 and Ala223 on one side and
the side chains of Thr242 and Ile246 on the other
side. The O2B and O3B oxygen atoms of ribose
interact with the Lys245 side chain and O2B can
make an additional hydrogen bond with NE2 of
Gln341. While the α-phosphate shows no polar
contact, β-phosphate oxygen atoms form hydrogen
bonds to Thr242 and Ser239 side chains and to a
water molecule that is within hydrogen-bonding
distance to both main-chain NH and OG of Ser239
and to the O4D atom of ribose from the NiR moiety.
The NiR moiety makes extensive interactions. The
ribose O3D atom is H-bonded to NE2 of Gln341.
The amide group of the nicotinamide ring estab-
lishes hydrogen bonds with OE1 and OE2 of
Glu393 and its carbonyl group interacts with the
main-chain oxygen of Gly238 and the side chain of
catalytic Cys294.
Both PsAMADHs exhibit similar reaction rates
and affinity for NAD+ (Km =40 μM for PsAMADH1
and Km =55 μM for PsAMADH2) in favor of a
similar NAD+ binding mode. Despite the unusual
ADP conformation of the coenzyme in PsAMADH1,
the NiR moiety occupies a position similar to that in
the so-called hydrolysis conformation or contracted
conformation of NADH (Fig. 3c).11,23 NAD(P)+
generally prefers an extended conformation. This
suggests that a reduction to NAD(P)H could have
occurred during crystallization due to the presence
of aldehydes from the polyethylene glycol (PEG)
used for crystallization as proposed by Perez-Miller
and Hurley.11 The different binding mode of the
ADP part in PsAMADH1 could reflect a large
flexibility of the whole NAD+ molecule and not of
Active-site residues
The NAD+ binding cavity is contiguous with the
substrate channel located on the opposite side of the
monomer. The catalytic Cys294 residue lies between
the NAD+ and the substrate binding sites (Fig. 4a
and b). Two more residues involved in catalysis,
namely, Asn162 and Glu260, are located at the
substrate channel bottom and together form the
active site. The catalytic Cys294 points toward the
NAD+ cavity in all 12 subunits of PsAMADH1 and
both subunits of PsAMADH2 adopting the resting
conformation. The backbone NH of Cys294 and
ND2 of the Asn162 side chain constitute the
oxyanion hole that stabilizes through hydrogen
bonds the thiohemiacetal and thioacyl intermediates
formed during the enzyme reaction.30
The side chain of the active-site base Glu is mobile
in many ALDH structures, and this is also the case
for the equivalent Glu260 of both PsAMADHs. In
PsAMADH1, Glu260 is in attacking conformation
except for the subunits C and D where it adopts the
resting conformation also observed in both subunits
of PsAMADH2. In the attacking conformation, its

876
Crystal Structure of Plant Aminoaldehyde Dehydrogenase
Table 2. Substrate specificity of pea AMADHs
Relative reaction rate (%)
Ligand
PsAMADH1
PsAMADH2
Coenzyme
NAD+
100
14
100
15
NADP+
3-Acetylpyridine-NAD+
Thio-NAD+
33
48
4.5
1.9
152
21
2.1
172
3-Pyridinealdehyde-NAD+
Deamino NAD+
Substrate
APAL
100
108
100
99
CPAL
TMAPAL
54
12
35
17
GPAL
ABAL
82
20
36
13
AHBAL
DMABAL
TMABAL
57
45
80
48
GBAL
29
25
30
22
GHBAL
BAL
1.3
0.3
0.8
0.1
Acetaldehyde
Propionaldehyde
1.4
1.8
Butyraldehyde
Valeraldehyde
Capronaldehyde
Enanthaldehyde
2.6
2.7
2.7
1.6
2.1
3.5
3.8
0.9
2-PCAL
3-PCAL
0
0.1
1.9
2.5
4-PCAL
2.9
0.5
Activities with substrates at 1 mM final concentration were measured in 0.15 M Tris–HCl buffer (pH 9.0). The rate of APAL oxidation was
arbitrarily taken as 100%. The specific activities of PsAMADH1 and 2 with 1 mM APAL were 41 and 106 nkat mg−1, respectively.
Coenzyme efficiency was measured in the same buffer with 1 mM APAL and a saturating concentration of 500 μM of the tested
compounds. For calculations, absorption maxima and extinction coefficients of reduced forms of NAD+ analogs were taken from the
literature.26,27 The rate of NAD+-mediated reaction was arbitrarily taken as 100%. CPAL, 3-cyanopropionaldehyde; DMABAL, N,N-
dimethyl-4-aminobutyraldehyde.

Crystal Structure of Plant Aminoaldehyde Dehydrogenase
877
Fig. 4. Substrate channel. (a and b) The substrate channel and coenzyme binding sites of PsAMADH1 and
PsAMADH2. The total surface of the cavities (shown in red for PsAMADH1 and in green for PsAMADH2) was calculated
using Hollow29 with 0.8 Å grid spacing and 1.1 Å interior probe . Both cavities join at a narrow pore around the catalytic
Cys294 shown in stick representation. Bound NAD+ coenzyme and glycerol molecule (GOL) are shown as sticks and
labeled. PsAMADH1 is pale grey, PsAMADH2 is light brown. (c and d) A close-up view of the substrate channel of
PsAMADH1 and PsAMADH2. Hydrogen bonds between GOL and enzyme are shown as dashed lines.
side chain, which points toward the nicotinamide
moiety, is at a distance of 3.5 Å from the SG atom of
Cys294 as frequently observed in ALDH structures
in complex with coenzyme in hydrolysis/contracted
conformation. Indeed, the side chain of Glu260
would make steric clashes with the NiR moiety in an
extended/hydride transfer conformation. In the
resting conformation, which is most commonly
found in ALDH structures, the carboxylate of
Glu260 is 7–8 Å away from the catalytic cysteine
and interacts with the main-chain carbonyl oxygen
of Trp459 and main-chain nitrogen of Glu470. A
proton relay mechanism involved in proton abstrac-
tion from the hydrolytic water and with participa-
tion of the two conserved glutamates Glu260 and
Glu470 and a close lysyl residue (Lys171) has been
proposed to operate in the ALDH family.10
Substrate binding site and specificity
Soaking or cocrystallization of substrate did not
result in any electron density map at the active site.

878
Crystal Structure of Plant Aminoaldehyde Dehydrogenase
However, a glycerol molecule from the cryoprotec-
tant solution occupies the substrate binding site of 9
among 12 subunits of PsAMADH1 [subunits A and
E were refined with a water molecule; D was refined
with 2-methyl-2,4-pentanediol (MPD)] and of both
subunits of PsAMADH2, as previously observed
with PaBADH (PDB code 2VE5). The glycerol
molecule binds at the bottom of the substrate
channel. The channel of PsAMADH1, which is
about 5–8 Å wide and 14 Å deep, is formed by
Ala109, Asp113, Pro452 and Ser453 at the entrance
and by the side chains of Asp110, Asn162, Tyr163,
Leu166, Met167, Trp170, Phe284, Phe288, Ile293,
Cys294, Ser295, Gln451 and Trp459 in the interior
(Fig. 4c). The bound glycerol is surrounded by
Asn162, Tyr163, Trp170, Ile293, Cys294, Ser295,
Ser453 and Trp459. Its O1 atom interacts with the
main-chain NH of the catalytic Cys294 and ND2 of
Asn162 in both PsAMADHs. The substrate channel
of PsAMADH2 is comparable to that of PsAMADH1
in shape and size. It differs only in three residues,
Ala/Trp109, Phe/Trp288 and Ser/Cys453 (Fig. 4d),
in agreement with a substrate-screening study
performed in this work, which shows a similar
substrate preference for both isoenzymes (Table 2).
The screening study of numerous natural and
synthetic aldehydes was performed at a concentra-
tion of 1 mM and the enzymatic conversion of
several substrates into the corresponding carboxylic
acids was verified by electrospray ionization mass
spectrometry (ESI-MS, Supplementary Figure S2).
At this concentration, PsAMADHs are more active
with C3 and C4 aminoaldehydes (APAL and ABAL)
than with their methyl and guanidino derivatives
(Table 2). Hydroxylated derivatives are converted
more slowly than the original aminoaldehydes but
much better than C2–C8 n-alkyl aldehydes, which
are very poor substrates and known as substrates of
nonspecific ALDHs.31 Among this group, butyral-
dehyde, valeraldehyde and capronaldehyde are the
best substrates and are oxidized at a rate around 3%
of that with APAL, indicating that the presence of a
nitrogen atom in the aldehyde molecule is compul-
sory for enzyme activity. Branched n-alkyl alde-
hydes such as methyl-2-butanal or 3-methyl-2-
butenal are not substrates at all. A structural
comparison with human ALDH2, which preferen-
tially oxidizes n-alkyl aldehydes, in complex with
crotonaldehyde (PDB code 1O01)11 shows that
Phe459 interacting through strong van der Waals
contacts with the aldehyde chain corresponds to
Ser/Cys453 in PsAMADH1/2, which might be
responsible for the poor activity of PsAMADHs
with n-alkyl aldehydes. 3- and 4-Pyridine carbox-
aldehydes (3-, 4-PCAL) are surprisingly oxidized at
rates similar to those of n-alkyl aldehydes, while
benzaldehyde is not. Because the single electron pair
at the nitrogen atom is not delocalized, pyridine
behaves as a tertiary amine and PCAL derivatives
may function as AMADH substrates. Interestingly,
BAL, which is the representative of C2 aminoalde-
hydes, is also poorly oxidized. This could be
explained by the absence of negatively charged
residues (Asp or Glu) or aromatic residues in close
vicinity to the quaternary amino group of the BAL
molecule modeled at the active site.
Natural C3 and C4 aminoaldehydes were therefore
analyzed further and the Km and Vmax values were
determined for four selected substrates (Table 3).
TMABAL is a slightly better substrate for PsA-
MADH1 than for PsAMADH2, as opposed to
GBAL. TMABAL and GBAL are similarly oxidized
by both PsAMADHs, although excess substrate
inhibition at higher concentrations was observed.
The side chain of Phe288 in PsAMADH1 overlaps that
of Phe436, which interacts with the trimethylamino
group of BAL in the structure of E. coli AMADH in
complex with BAL (PDB code 1WNB).4 However,
BAL rather resembles the binding of a C3-C4
branched chain because its aldehyde oxygen is 4.5 Å
far from the catalytic cysteine. Thus, Phe288 in
PsAMADH1 should be responsible for specific inter-
actions with the trimethylamino group of TMABAL.
In contrast, Trp288 in PsAMADH2, which is bulkier,
could account for the lower Km value of TMABAL
compensated for by a higher Vmax. GBAL, which is a
long substrate, would interact with Asp113 through
its guanidino group in both PsAMADHs. An addi-
tional interaction with Trp109 in PsAMADH2 (Ala in
PsAMADH1) might explain its lower Km value. A
bound glycerol molecule in the upper part of the
substrate channel in PsAMADH2 surrounded by
Asp113, Trp109, Trp288 and Leu266 occupies the
modeled position of the guanidino group of GBAL.
APAL and ABAL display higher Km values
(between 7- and 17-fold) and average Vmax values
Table 3. Kinetic parameters of PsAMADHs for the natural substrates and coenzyme NAD+
PsAMADH1
PsAMADH2
Vmax
Vmax/Km
(relative)
Vmax
Vmax/Km
(relative)
Ligand
Km (μM)
(nmol s−1 mg−1
)
Km (μM)
(nmol s−1 mg−1
)
TMABAL
GBAL
10
11
87
68
57
50
—
1
21
7
10
29
55
140
78
190
57
—
0.35
0.59
1
0.71
0.09
0.03
APAL
75
ABAL
170
40
0.10
NAD+
Activities were measured in 0.15 M Tris–HCl buffer (pH 9.0) at 20 °C. Kinetic constants (Km, Vmax) of substrates were measured using a
saturating NAD+ concentration of 500 μM and calculated from the Lineweaver–Burk plots of initial rates. The Km value for NAD+ was
measured at a fixed 1 mM concentration of APAL.

Crystal Structure of Plant Aminoaldehyde Dehydrogenase
879
compared to TMABAL and GBAL in PsAMADH1
and are therefore weaker substrates. In PsA-
MADH2, the Trp288 side chain reduces the volume
of the substrate channel compared to Phe288 in
PsAMADH1. This may increase π electron interac-
tions of the Trp288 ring with the substrate amino
group and strengthen the binding of linear ami-
noaldehydes such as APAL and ABAL, leading to
an increase of Vmax and decrease of Km. The high
Vmax determined for APAL makes it the best
substrate of PsAMADH2.
suggesting the existence of a carnitine cycle. In
mammals, the resulting product of the reaction of
AMADH with TMABAL, γ-butyrobetaine, is hydrox-
ylated at the β-position by β-butyrobetaine hydrox-
ylase (EC 1.14.11.1) yielding carnitine. Homologous
sequences coding for this enzyme have not been
found in plants. The hydroxylation might be mediat-
ed by a different hydroxylase or performed even prior
to the AMADH reaction. This could be possible
because 2-hydroxyputrescine has been found in
plants42 and hydroxylated derivatives AHBAL (4-
amino-2-hydroxybutyraldehyde) and GHBAL (4-
guanidino-2-hydroxybutyraldehyde) are PsAMADH
substrates. Moreover, recent results in Arabidopsis
show that plant and animal cells might share similar
features regarding carnitine biosynthesis because the
same metabolites trimethyllysine and γ-butyrobe-
taine are present (Y. Perrin, University of Technology
of Compiègne, France, personal communication).
Physiological aspects of AMADH in plants
The physiological role of plant AMADHs has been
studied because they catalyze the terminal step in
polyamine catabolism. Plant BADHs have been
studied because of their role in the production of
the compatible osmolyte glycine betaine from BAL
and the participation in cellular response to salinity
and drought stress.32 In glycine betaine nonaccumu-
lating plants such as pea or Limonium, compatible
osmolytes β-alanine betaine, that is, propionobetaine
from APAL (via β-alanine), and GABA (4-aminobu-
tyric acid; directly from ABAL or indirectly from
GBAL) are produced as an alternative to glycine
betaine.32–34 Recently, the loss of AMADH gene
function in rice has been found to be responsible for
the fragrance of aromatic rice varieties,8 which
appears due to the accumulation of a volatile
compound, 2-acetyl-1-pyrroline, arising from the
acetylation of ABAL normally converted to GABA.35
A majority of AMADHs/BADHs carry the C-
terminal peroxisomal targeting sequence,15 but the
existence of nonperoxisomal forms such as chloro-
plastic or cytosolic ones has also been reported.32
When the pH optima of both pea isoenzymes are
compared, PsAMADH2 had the highest rate with
APAL at pH 10.2, while PsAMADH1 showed a
maximum at pH 9.7. These values are in agreement
with those previously reported for other AMADH/
BADH enzymes, for example, from rice,8 and they
are likely related to the basic pH of peroxisomes.36,37
The catalytic Cys294 (pKa =8.0)14 exists in its thiolate
form when pH is significantly above 8.0, which
facilitates an efficient nucleophilic attack at the
substrate. Finally, since determined isoelectric
points are acidic (pI=5.70 and 5.45) and protein
solubility reaches its minimum at the isoelectric
point, it is unlikely that PsAMADHs operate in an
acidic environment.
Conclusions
The reported structural data together with the
substrate specificity values revealed interesting
results regarding the physiological role of PsA-
MADHs. APAL, the best substrate of PsAMADH2
and one of the best substrates of PsAMADH1, is
oxidized to β-alanine, which is an important
metabolic intermediate known as a precursor of
the plant osmoprotectant β-alanine betaine.33 Sur-
prisingly, PsAMADH1 preferentially oxidizes TMA-
BAL to the carnitine precursor in animal cells, γ-
butyrobetaine. Thus, in addition to the currently
discussed AMADH functions, the enzyme might be
involved in carnitine biosynthesis in plants, which
remains to be elucidated.
Materials and Methods
Cloning, expression and purification of
recombinant enzymes
PsAMADH1 open reading frame (1510 bp; EMBL/
GenBank accession no. AJ315852) was cloned into a
pET28b His tag vector as described.43 PsAMADH2 open
reading frame (1512 bp; EMBL/GenBank accession no.
AJ315853)14 was cloned into a pCDFDuet His-tag vector
(Novagen, La Jolla, CA) digested by BamHI and XhoI
restriction endonucleases. Primers used for amplification of
PsAMADH2 cDNA contained restriction sites (in bold) for
BamHI (upstream primer, 5′-CAGGATCCAGATATTCC-
GATCCCAACTCGT-3′) and XhoI (downstream primer, 5′-
CGCTCGAGTTACAGTTTTGCAGGAGGCT-3′). DNA
sequencing of the cloned PsAMADH1 and PsAMADH2
cDNAs revealed several differences in comparison with the
database sequences and, consequently, these were
updated. Expression, cell lysis and purification were
performed as described for PsAMADH1.43
The observed TMABAL oxidation might have
further physiological implications. Human and rat
liver ALDH9 enzymes have been shown to convert
TMABAL as the best substrate to γ-butyrobetaine,
thus participating in the biosynthesis of carnitine.38
Similarly, AMADHs might be involved in carnitine
production in plants. In comparison with mammals,
little is known about carnitine biosynthesis and
function in plants and its abundance is about 100–
1000 times lower.39 However, an enzymatic activity
of mitochondrial and chloroplastic acylcarnitine
transferases has been detected in pea and spinach,40,41
PsAMADH2 was purified by the same procedure on a
column filled with cobalt- or nickel-charged IDA-Sephar-
ose 6B (Sigma-Aldrich, Steinheim, Germany). The recom-
binant enzymes were further purified by ion-exchange
chromatography on a Resource Q column (GE Healthcare,

880
Crystal Structure of Plant Aminoaldehyde Dehydrogenase
Uppsala, Sweden). Equilibration buffers were as follows:
20 mM Bis–Tris–HCl, pH 6.0 (PsAMADH1), or 20 mM
Tris–HCl, pH 8.0 (PsAMADH2), containing 5% (v/v)
glycerol and 1 mM EDTA (ethylenediaminetetraacetic
acid). Elution buffers additionally contained 1 M NaCl.
Active fractions were pooled and desalted by diafiltration
using 10 -kDa-cutoff Centricon filters (Millipore, Bedford,
MA). Complete purification from a 400-mL culture
yielded 5 mg of PsAMADH1 with a specific activity of
about 41 nkat mg−1 and 15 mg of PsAMADH2 with a
specific activity of 106 nkat mg−1 (measured at 20 °C with
1 mM APAL as a substrate). The identity of purified
proteins was confirmed by peptide mass fingerprinting,
which was performed on a Microflex LRF20 matrix-
assisted laser desorption/ionization time-of-flight mass
spectrometer (Bruker Daltonik, Bremen, Germany) after
SDS-PAGE and in-gel digestion.44 NCBInr database
searches for PsAMADH1 and PsAMADH2 provided
sequence coverage and probability-based score values of
49%/224 (accession no. gi|206715498) and 47%/187
(accession no. gi|206715496), respectively. Isoelectric
focusing (pH range 3–10) confirmed an agreement
between predicted and experimental isoelectric points.
bicarbonate was incubated with 10 μg of PsAMADH2 at
room temperature overnight. The enzyme was then
removed using a Microcon 10-kDa-cutoff filter (Millipore).
Filtrates were diluted 1:100 (v/v) by methanol or
acetonitrile and analyzed on a Q-TOF micro mass
spectrometer equipped with a nanoelectrospray ion
source (Waters-Micromass, Manchester, UK). Positive
ionization mode was used to produce protonated quasi-
molecular ions [M+H]+. Each sample was directly
introduced to the mass spectrometer by a capillary
borosilicate emitter (Proxeon Biosystems). Parameters of
the nanoelectrospray were as follows: capillary voltage,
800 V; cone voltage, 25 V; extraction cone, 2 V; and source
temperature, 80 °C. Full mass spectra were recorded in the
mass range m/z 50–800. Tandem mass spectra (MS/MS;
daughter ion scan mode) were recorded in the mass range
from m/z 50 to that of appropriate parent ion with
variable collision cell energy in the range 10–20 V using
argon as a collision gas.
Crystallization and data collection
Purified PsAMADHs were crystallized by the hanging-
drop vapor-diffusion method at 20 °C. Prior to crystalli-
zation, both enzymes were devoid of imidazole and salt
and concentrated to 12–18 mg ml−1 by ultrafiltration using
a Centricon 30 -kDa-cutoff device (Millipore). Preliminary
crystallization conditions of both enzymes with 5 mM
NAD+ were found using Crystal Screen and Crystal
Screen 2 from Hampton Research (Laguna Niguel, CA).
PsAMADH1 was crystallized as described43 over reser-
voirs containing 0.1 M Hepes (pH 7.5), 13% (w/v) PEG
6000 and 5% (v/v) MPD. PsAMADH2 was crystallized
similarly over reservoirs containing 0.1 M Hepes (pH 7.5),
18% (w/v) PEG 4000, 10% (v/v) isopropanol and 0.5%
(w/v) n-octyl β-D-glucopyranoside. The crystals were
transferred to a cryoprotectant solution [the mother liquor
supplemented with 20% (v/v) glycerol] for 3 min and
flash-frozen in liquid nitrogen. Diffraction data for crystals
of PsAMADH1 (2.4 Å resolution) and PsAMADH2
(2.15 Å) were collected at 100 K on beamlines ID29 and
ID14-2, respectively, at the European Synchrotron Radi-
ation Facility (Grenoble, France). Diffraction intensities
were integrated with the program XDS46 and further
processed using the CCP4 program suite.47 Data collection
and processing statistics are given in Table 1.
Assays of activity, protein concentration and
ion content
AMADH activity was measured spectrophotome-
trically by monitoring the formation of NADH (ɛ340
=
6.62 mM−1 cm−1) at 20 °C.14 The reaction mixture (2 ml)
contained 1–2 nkat amounts of pure enzymes. Diethyla-
cetals of AHBAL, GBAL, GHBAL and GPAL (3-guanidi-
nopropionaldehyde) were synthetic preparations.13 BAL
dimethyl acetal iodide, N,N,N-trimethyl-3-aminopropio-
naldehyde (TMAPAL) diethyl acetal iodide and TMABAL
diethyl acetal iodide are crystalline compounds synthe-
sized according to a procedure described for the latter
compound.38 Free aminoaldehydes were prepared by
heating their respective diethyl acetals with 0.1 M HCl for
10 min in a plugged test tube.6 NAD+ analogs were
assayed as coenzymes at a concentration of 500 μM with
1 mM APAL as a substrate. The following absorption
maxima and extinction coefficients of reduced forms of
these analogs26,27 were used for calculations: NADPH,
ɛ340 =6.62 mM−1 cm−1; thio-NADH, ɛ400 =11.9 mM−1
cm−1; 3-pyridinealdehyde-NADH, ɛ358 =9.3 mM−1 cm−1
,
3-acetylpyridine-NADH, ɛ363 =9.1 mM−1 cm−1; deamino
NADH, ɛ338 =6.2 mM−1 cm−1. Tris–HCl buffers (150 mM)
in the pH range of 7.0–9.0 and glycine–NaOH buffers
(150 mM) in the pH range of 8.8–10.6 were used to
determine pH optimum using 1 mM APAL and 0.5 mM
NAD+. Protein concentration was determined by the
Bradford method45 or with the following calculated
extinction coefficients: PsAMADH1, ɛ280 =82,390 M−1
cm−1; PsAMADH2, ɛ280 =90,410 M−1 cm−1. Sodium and
magnesium content in enzyme preparations was analyzed
by AAS on an Avanta Sigma spectrometer (GBC Scientific
Equipment, Dandenong, Victoria, Australia) using flame
atomization.
Molecular replacement, refinement and final model
Structure determination of PsAMADH1 was performed
by molecular replacement with PHASER48, using the
monomer structure of the human mitochondrial ALDH18
(PDB code 1CW3) as a model search. Twelve monomers
were placed in the asymmetric unit forming six dimers.
Model refinement was performed with REFMAC49 using
the data set at 2.4 Å resolution. Electron density maps
were evaluated using COOT.50 They revealed clear
density for the ADP moiety of the NAD+ coenzyme.
NAD+ and water molecules were added during refine-
ment. The final structure (Rcryst =20.0%; Rfree =24.4%)
contains 12 polypeptidic chains A to L forming six dimers,
11 NAD+ molecules and the ADP part of one NAD+ in
chain D, 1997 water molecules, 19 glycerol molecules, 12
sodium ions and one MPD molecule. The N- and C-
terminal parts are disordered. The chains start at Ser7
except for three chains (H, J, L) starting at Ser6. Terminal
Leu503, which belongs to the SKL peroxisomal signal, is
missing, as well as Lys502 in chains C, F and L. Each
ESI-MS and MS/MS analysis of reaction products
Because many substrates were synthetic preparations
or, as in the case of PCAL, which for the first time was
identified as an AMADH substrate, the enzymatic
conversion to the corresponding carboxylic acids was
verified by ESI-MS. A reaction mixture containing 1 mM
substrate and 1 mM NAD+ in 50 mM ammonium

Crystal Structure of Plant Aminoaldehyde Dehydrogenase
881
monomer and every second chain in the dimer contain a
very flexible loop (residues 365–370 and residues 59–64,
respectively), both having higher B values.
4. Gruez, A., Roig-Zamboni, V., Grisel, S., Salomoni, A.,
Valencia, C., Campanacci, V. et al. (2004). Crystal
structure and kinetics identify Escherichia coli YdcW
gene product as a medium-chain aldehyde dehydro-
genase. J. Mol. Biol. 343, 29–41.
5. Samsonova, N. N., Smirnov, S. V., Novikova, A. E. &
Ptitsyn, L. R. (2005). Identification of Escherichia coli
K12 YdcW protein as a gamma-aminobutyraldehyde
dehydrogenase. FEBS Lett. 579, 4107–4112.
6. Trossat, C., Rathinasabapathi, B. & Hanson, A. D.
(1997). Transgenically expressed betaine aldehyde
dehydrogenase efficiently catalyzes oxidation of
dimethylsulfoniopropionaldehyde and ω-aminoalde-
hydes. Plant Physiol. 113, 1457–1461.
7. Incharoensakdi, A., Matsuda, N., Hibino, T., Meng,
Y. L., Ishikawa, H., Hara, A. et al. (2000). Overpro-
duction of spinach betaine aldehyde dehydrogenase
in Escherichia coli. Structural and functional proper-
ties of wild-type, mutants and E. coli enzymes. Eur. J.
Biochem. 267, 7015–7023.
The structure of PsAMADH2 was solved by molecular
replacement performed at 2.5 Å with PHASER using the
PsAMADH1 dimer as a template. Assuming that the
asymmetric unit contains one dimer, the calculated
Matthews coefficient51 was 2.38 Å3 Da−1, corresponding
to a solvent content of 47.9%. Refinement was performed up
to 2.15 Å resolution. The final model (Rcryst =23.4%;
Rfree =28.9%) contains two chains A and B, the ADP parts
of two NAD+ molecules, two sodium ions, seven glycerol
and 290 water molecules. The first residues 1–3 are missing,
while the C-terminus is complete up to Leu503, which is a
part of the AKL peroxisomal signal sequence. Refinement
details of the two structures are shown in Table 1. Molecular
graphics images were generated using PyMOL.52
PDB accession codes
8. Bradbury, L. M. T., Gillies, S. A., Brushett, D. J.,
Waters, D. L. E. & Henry, R. J. (2008). Inactivation of
an aminoaldehyde dehydrogenase is responsible for
fragrance in rice. Plant Mol. Biol. 68, 439–449.
The atomic coordinates and structure factors have been
deposited in the PDB with the following accession codes:
3IWK for PsAMADH1 and 3IWJ for PsAMADH2.
9. Johansson, K., El-Ahmad, M., Ramaswamy, S.,
Hjelmqvist, L., Jörnvall, H. & Eklund, H. (1998).
Structure of betaine aldehyde dehydrogenase at 2.1 Å
resolution. Protein Sci. 7, 2106–2117.
10. González-Segura, L., Rudiño-Piñera, E., Muñoz-Clares,
R. A. & Horjales, E. (2009). The crystal structure of a
ternary complex of betaine aldehyde dehydrogenase
from Pseudomonas aeruginosa provides new insight into
the reaction mechanism and shows a novel binding
mode of the 2′-phosphate of NADP+ and a novel
cation binding site. J. Mol. Biol. 385, 542–557.
11. Perez-Miller, S. J. & Hurley, T. D. (2003). Coenzyme
isomerization is integral to catalysis in aldehyde
dehydrogenase. Biochemistry, 42, 7100–7109.
12. Wymore, T., Hempel, J., Cho, S. S., Mackerell, A. D.,
Jr., Nicholas, H. B., Jr. & Deerfield, D. W., II (2004).
Molecular recognition of aldehydes by aldehyde
dehydrogenase and mechanism of nucleophile acti-
vation. Proteins, 57, 758–771.
13. Šebela, M., Brauner, F., Radová, A., Jacobsen, S.,
Havliš, J., Galuszka, P. & Peč, P. (2000). Characterisa-
tion of a homogeneous plant aminoaldehyde dehy-
drogenase. Biochim. Biophys. Acta, 1480, 329–341.
14. Brauner, F., Šebela, M., Snégaroff, J., Peč, P. &
Meunier, J. C. (2003). Pea seedling aminoaldehyde
dehydrogenase: primary structure and active site
residues. Plant Physiol. Biochem. 41, 1–10.
Acknowledgements
This work was supported by grant 522/08/0555
from the Czech Science Foundation, grant MSM
6198959215 from the Ministry of Education, Youth
and Sports of the Czech Republic and grants from
the Centre National de la Recherche Scientifique
(UPR 3082, UPR 9063, URA 2185 and 2171), and the
Institut National de la Recherche Agronomique
(UMR 206). Dr. Petr Hekera from Palacký University
is acknowledged for AAS analysis. Prof. David
Morris from the University of Southampton is
thanked for his help with manuscript preparation.
Dr. Michaela Wimmerová from Masaryk University
contributed to data processing and interpretation.
Supplementary Data
Supplementary data associated with this article
can be found, in the online version, at doi:10.1016/
j.jmb.2009.12.015
15. Reumann, S., Ma, C., Lemke, S. & Babujee, L. (2004).
AraPerox. A database of putative Arabidopsis proteins
from plant peroxisomes. Plant Physiol. 136, 2587–2608.
16. Laskowski, R. A., MacArthur, M. W., Moss, D. S. &
Thornton, J. M. (1993). PROCHECK: a program to
check the stereochemical quality of protein structures.
J. Appl. Crystallogr. 26, 283–291.
References
1. Perozich, J., Nicholas, H., Wang, B. C., Lindahl, R. &
Hempel, J. (1999). Relationships within the aldehyde
dehydrogenase extended family. Protein Sci. 8,
137–146.
2. Sophos, N. A., Pappa, A., Ziegler, T. L. & Vasiliou, V.
(2001). Aldehyde dehydrogenase gene superfamily:
the 2000 update. Chem.-Biol. Interact. 130–132, 323–337.
3. Tylichová, M., Kopečný, D., Snégaroff, J. & Šebela, M.
(2007). Aminoaldehyde dehydrogenases: has the time
now come for new interesting discoveries? Curr.
Topics Plant Biol. 8, 45–70.
17. Krissinel, E. & Henrick, K. (2004). Secondary-structure
matching (SSM), a new tool for fast protein structure
alignment in three dimensions. Acta Crystallogr., Sect.
D: Biol Crystallogr. 60, 2256–2268.
18. Ni, L., Zhou, J., Hurley, T. D. & Weiner, H. (1999).
Human liver mitochondrial aldehyde dehydrogenase:
three-dimensional structure and the restoration of
solubility and activity of chimeric forms. Protein Sci. 8,
2784–2790.
19. Valenzuela-Soto, E. M., Velasco-García, R.,
Mújica-Jiménez, C., Gaviria-González, L. L. &

882
Crystal Structure of Plant Aminoaldehyde Dehydrogenase
Muñoz-Clares, R. A. (2003). Monovalent cations
requirements for the stability of betaine aldehyde
dehydrogenase from Pseudomonas aeruginosa, porcine
kidney and amaranth leaves. Chem. Biol. Interact.143-
144,, 139–148.
35. Yoshihashi, T., Huong, N. T. T. & Inatomi, H. (2002).
Precursors of 2-acetyl-1-pyrroline, a potent flavor
compound of an aromatic rice variety. J. Agric. Food
Chem. 50, 2001–2004.
36. Dansen, T. B., Wirtz, K. W. A., Wanders, R. J. A. &
Pap, E. H. W. (2000). Peroxisomes in human fibro-
blasts have a basic pH. Nat. Cell. Biol. 2, 51–53.
37. Theodoulou, F. L., Job, K., Slocombe, S. P., Footitt, S.,
Holdsworth, M., Baker, A. et al. (2005). Jasmonic acid
levels are reduced in COMATOSE ATP-binding
cassette transporter mutants. Implications for trans-
port of jasmonate precursors into peroxisomes. Plant
Physiol. 137, 835–840.
38. Vaz, F. M., Fouchier, S. W., Ofman, R., Sommer, M. &
Wanders, R. J. A. (2000). Molecular and biochemical
characterization of rat γ-trimethylaminobutyralde-
hyde dehydrogenase and evidence for the involve-
ment of human aldehyde dehydrogenase 9 in
carnitine biosynthesis. J. Biol. Chem. 275, 7390–7394.
39. Wood, C., Masterson, C. & Thomas, D. R. (1992). The
role of carnitine in plant cell metabolism. In Plant
Organelles, Society for Experimental Biology Seminar
Series (Tobin, A, K., ed), vol. 50, pp. 229–263, Cam-
bridge University Press, Cambridge, UK.
20. González-Segura, L., Velasco-García, R., Rudiño-
Piñera, E., Mújica-Jiménez, C. & Muñoz-Clares, R. A.
(2005). Site-directed mutagenesis and homology
modeling indicate an important role of cysteine 439
in the stability of betaine aldehyde dehydrogenase
from Pseudomonas aeruginosa. Biochimie, 87, 1056–1064.
21. Zheng, H., Chruszcz, M., Lasota, P., Lebioda, L. &
Minor, W. (2008). Data mining of metal ion environ-
ments present in protein structures. J. Inorg. Biochem.
102, 1765–1776.
22. Lamb, A. L. & Newcomer, M. E. (1999). The structure
of retinal dehydrogenase type II at 2.7 Å resolution:
implications for retinal specificity. Biochemistry, 38,
6003–6011.
23. Hammen, P. K., Allali-Hassani, A., Hallenga, K.,
Hurley, T. D. & Weiner, H. (2002). Multiple con-
formations of NAD and NADH when bound to
human cytosolic and mitochondrial aldehyde dehy-
drogenase. Biochemistry, 41, 7156–7168.
24. D'Ambrosio, K., Pailot, A., Talfournier, F., Didierjean,
C., Benedetti, E., Aubry, A. et al. (2006). The first crystal
structure of a thioacylenzyme intermediate in the
ALDH family: new coenzyme conformation and
relevance to catalysis. Biochemistry, 45, 2978–2986.
25. Di Costanzo, L., Gomez, G. A. & Christianson, D. W.
(2007). Crystal structure of lactaldehyde dehydroge-
nase from Escherichia coli and inferences regarding
substrate and cofactor specificity. J. Mol. Biol. 366,
481–493.
26. Siegel, J. M., Montgomery, G. A. & Bock, R. M. (1959).
Ultraviolet absorption spectra of DPN and analogs of
DPN. Arch. Biochem. Biophys. 82, 288–299.
27. Stein, A. M., Lee, J. K., Anderson, C. D. & Anderson, B.
M. (1963). The thionicotinamide analogs of DPN and
TPN. I. Preparation and analysis. Biochemistry, 2,
1015–1017.
40. Burgess, N. & Thomas, D. R. (1986). Carnitine acetyltrans-
ferase in pea cotyledon mitochondria. Planta, 167, 58–65.
41. Masterson, C. & Wood, C. (2000). Pea chloroplast
carnitine acetyltransferase. Proc. Biol. Sci. 267, 1–6.
42. Samborski, D. J. & Rohringer, R. (1970). Abnormal
metabolites of wheat: occurrence, isolation and bio-
genesis of 2-hydroxyputrescine amides. Phytochemistry,
9, 1939–1945.
43. Tylichová, M., Briozzo, P., Kopečný, D., Ferrero, J.,
Moréra, S., Joly, N. et al. (2008). Purification, crystal-
lization and preliminary X-ray diffraction study of a
recombinant plant aminoaldehyde dehydrogenase
from Pisum sativum. Acta Crystallogr. F, 64, 88–90.
44. Šebela, M., Štosová, T., Havliš, J., Wielsch, N.,
Thomas, H., Zdráhal, Z. & Shevchenko, A. (2006).
Thermostable trypsin conjugates for high-throughput
proteomics: synthesis and performance evaluation.
Proteomics, 6, 2959–2963.
45. 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.
46. Kabsch, W. (1993). Automatic processing of rotation
diffraction data from crystals of initially unknown
symmetry and cell constants. J. Appl. Crystallogr. 26,
795–800.
28. Hill, E. J., Chou, T. H., Shih, M. C. & Park, J. H. (1975).
Covalent binding of 3-pyridinealdehyde nicotinamide
adenine dinucleotide and substrate to glyceraldehyde
3-phosphate dehydrogenase. J. Biol. Chem. 250,
1734–1740.
29. Ho, B. K.
&
Gruswitz, F. (2008). HOLLOW:
generating accurate representations of channel and
interior surfaces in molecular structures. BMC
Struct. Biol. 8, 49.
30. Steinmetz, C. G., Xie, P., Weiner, H. & Hurley, T. D.
(1997). Structure of mitochondrial aldehyde dehydro-
genase: the genetic component of ethanol aversion.
Structure, 5, 701–711.
31. Klyosov, A. A. (1996). Kinetics and specificity of
human liver aldehyde dehydrogenases toward ali-
phatic, aromatic, and fused polycyclic aldehydes.
Biochemistry, 35, 4457–4467.
47. Collaborative Computational Project, Number 4.
(1994). The CCP4 suite: programs for protein crystal-
lography. Acta Crystallogr., Sect. D: Biol. Crystallogr. 50,
760–763.
48. Storoni, L. C., McCoy, A. J. & Read, R. J. (2004).
Likelihood-enhanced fast rotation functions. Acta
Crystallogr., Sect. D: Biol. Crystallogr. 60, 432–438.
49. Murshudov, G. N., Vagin, A. A. & Dodson, E. J.
(1997). Refinement of macromolecular structures by
the maximum-likelihood method. Acta Crystallogr.,
Sect. D: Biol. Crystallogr. 53, 240–255.
32. Weigel, P., Weretilnyk, E. A. & Hanson, A. D. (1986).
Betaine aldehyde oxidation by spinach chloroplasts.
Plant Physiol. 82, 753–759.
33. Duhazé, C., Gouzerh, G., Gagneul, D., Larher, F. &
Bouchereau, A. (2002). The conversion of spermidine
to putrescine and 1,3-diaminopropane in the roots of
Limonium tataricum. Plant Sci. 163, 639–646.
34. Hasegawa, P. M., Bressan, R. A., Zhu, J. K. & Bohnert,
H. J. (2000). Plant cellular and molecular responses to
high salinity. Annu Rev. Plant Physiol. Plant Mol. Biol.
51, 463–499.
50. Emsley, P. & Cowtan, K. (2004). Coot: model-building
tools for molecular graphics. Acta Crystallogr., Sect. D:
Biol. Crystallogr. 60, 2126–2132.
51. Matthews, B. W. (1968). Solvent content of protein
crystals. J. Mol. Biol. 33, 491–497.
52. DeLano, W. L. (2002). The PyMOL Molecular Gra-
phics System on the World Wide Web. http://www.
pymol.org/.