Crystal structure of archaeal highly thermostable
L-aspartate dehydrogenase/NAD/citrate ternary complex
Kazunari Yoneda
1
, Haruhiko Sakuraba
2
, Hideaki Tsuge
3
, Nobuhiko Katunuma
3
and Toshihisa Ohshima
1
1 Institute of Genetic Resources, Faculty of Agriculture, Kyushu University, Fukuoka, Japan
2 Department of Biological Science and Technology, Faculty of Engineering, The University of Tokushima, Japan
3 Institute for Health Sciences, Tokushima Bunri University, Japan
In prokaryotes, de novo NAD biosynthesis generally
proceeds via a condensation reaction between l-aspar-
tate and dihydroxyacetone phosphate that is catalyzed
by two enzymes: l-aspartate oxidase (LAO; the nadB
gene product) and quinolinate synthase (QS; the nadA
gene product) [1]. LAO catalyzes the oxidation of
l-aspartate to iminoaspartate, after which QS catalyzes
the condensation of iminoaspartate with dihydroxyace-
tone phosphate to produce quinolinate. Quinolinate is
then converted to nicotinate mononucleotide by quino-
linate phosphoribosyltransferase (the nadC gene prod-
uct), which is followed by conversion to NAD via a
metabolic sequence involving two enzymes: nicotinate
mononucleotide adenylyltransferase and NAD syn-
thase [2].
We recently detected the presence of LAO in Pyro-

with a crystallographic R-factor of 21.7% (R
free
¼
22.6%). The structure indicates that each subunit consists of two domains
separated by a deep cleft containing an active site. Structural comparison
of the A. fulgidus l-aspDH ⁄ NAD ⁄ citrate ternary complex and the Thermo-
toga maritima l-aspDH ⁄ NAD binary complex showed that A. fulgidus
l-aspDH assumes a closed conformation and that a large movement of the
two loops takes place during substrate binding. Like T. maritima l-aspDH,
the A. fulgidus enzyme is highly thermostable. But whereas a large number
of inter- and intrasubunit ion pairs are responsible for the stability of
A. fulgidus l-aspDH, a large number of inter- and intrasubunit aro-
matic pairs stabilize the T. maritima enzyme. Thus stabilization of these
two l-aspDHs appears to be achieved in different ways. This is the first
detailed description of substrate and coenzyme binding to l-aspDH and
of the molecular basis of the high thermostability of a hyperthermophilic
l-aspDH.
Abbreviations
LAO,
L-aspartate oxidase; L-aspDH, L-aspartate dehydrogenase; MIRAS, multiple isomorphous replacement with anomalous scattering;
QS, quinolinate synthase.
FEBS Journal 274 (2007) 4315–4325 ª 2007 The Authors Journal compilation ª 2007 FEBS 4315
an operon with nadA, nadC and three other unknown
genes. We therefore proposed that a de novo NAD bio-
synthetic pathway functions in P. horikoshii under
anaerobic conditions [3]. More recently, a previously
unknown amino acid dehydrogenase, l-aspartate dehy-
drogenase (l-aspDH; the TM1643 gene product),
was identified in a hyperthermophilic bacterium Ther-
motoga maritima by Yang et al. based on the 3D struc-

only low similarity to aspartate semialdehyde dehydro-
genase, inositol 1-phosphate synthase and dihydrodipi-
colinate dehydrogenase. This enzyme thus appears to
represent a new class of amino acid dehydrogenase.
However, details about the molecular strategy underly-
ing its high thermal stability, as well as the manner in
which substrate and coenzyme are bound by the
enzyme, are still unclear. Therefore, our aim was to
determine the structure of the A. fulgidus l-aspDH in
complex with NAD and a substrate analog, citrate.
Factors that could stabilize the enzyme were then com-
pared with those in the T. maritima l-aspDH, and the
structural features that appear to be responsible for
the high thermostability of each enzyme are discussed.
Finally, we describe a substrate-induced conforma-
tional change in the ternary complex of A. fulgidus
l-aspDH.
Results and Discussion
Overall structure
The structure of A. fulgidus l-aspDH was determined
using multiple isomorphous replacement with anoma-
lous scattering (MIRAS) and was refined at a resolu-
tion of 1.9 A
˚
(Table 1). The asymmetric unit consisted
of one homodimer with a solvent content of 39.3%,
which corresponds to a Matthew’s coefficient [6] of
2.0 A
˚
3

(Fig. 2). The internal structure of the A. fulgidus
enzyme was basically the same as that of the T. mari-
tima enzyme (r.m.s.d. ¼ 1.9 A
˚
for the Ca atoms of 223
residues), although a marked difference was observed
in the position of three loops in the catalytic domain.
When the Rossmann-fold domain of the A. fulgidus
l-aspDH structure was superimposed on that of the
T. maritima l-aspDH structure, we observed a large
shift in the positions of two loops (loop 1: R133–
G143, loop 2: V180–I188 in the A. fulgidus l-aspDH)
toward the active site cavity (Fig. 3). The average
movements of loops 1 and 2 were estimated to be 3.2
and 3.6 A
˚
, and the largest movements of loops 1 and 2
were estimated to be 6.7 A
˚
(K142) and 5.8 A
˚
(E183), respectively. As described below, this large
Crystal structure of L-aspDH from A. fulgidus K. Yoneda et al.
4316 FEBS Journal 274 (2007) 4315–4325 ª 2007 The Authors Journal compilation ª 2007 FEBS
conformational change may be essential for substrate
binding. In addition, loop 3 (P208–S216), which is dis-
ordered in the T. maritima enzyme, was clearly obser-
vable in the structure of A. fulgidus enzyme and
formed a flap over the active site cavity (Fig. 3). It has
been proposed that the structure of T. maritima l-as-

boxylate are situated within hydrogen- and ionic-bond-
ing distance of the side chains of N187, H189 and
S216, and the main chain amide proton of S216. In
addition, the oxygen atoms of the Cb carboxyl group
Table 1. Statistics on data collection, phase determination and refinement. The crystal belongs to space group P2
1
2
1
2 with a ¼ 47.52 A
˚
,
b ¼ 89.58 A
˚
, c ¼ 100.49 A
˚
.
Native Hg1
a
Hg2
b
Hg3
c
Data collection
Maximum resolution (A
˚
) 1.9 2.0 2.0 1.79
Total reflections 237826 212964 202467 289592
Unique reflections 34 001 55997 54558 77004
Redundancy 7.0 3.8 3.7 3.8
Completeness

RMSD bond angles (°) 1.5
Average B-factors (A
˚
2
)
Protein 29.5
NAD 26.9
Citrate 34.2
Water 33.4
a
Hg1, ethyl mercuric phosphate.
b
Hg2, 1,4-diacetoxymercuri-2,3-dimethoxybutane.
c
Hg3, phenylmercury acetate.
d
Values in parentheses
are for the last resolution shell.
e
R
sym
¼ S
h
S
i
| I
i
(h) –<I (h) >|⁄S
h
S

are within hydrogen-bonding distance of the side chain
of K134, the main chain amide proton of L161, and a
water molecule (WAT33). All of the residues are con-
served in T. maritima l-aspDH, except for L161, which
is replaced with Ile. Within the active site of the
T. maritima enzyme, however, we found that the side
chains of residues corresponding to K134 in loop 1,
N187 in loop 2 and S216 in loop 3 are far removed
from the l-aspartate molecule, and it does not appear
that hydrogen bonds are formed with the carboxyl
groups of the substrate (Figs 3A and 4B). This
suggests that the large movement of loops 1 and 2, in
addition to the formation of a flap by loop 3, may
induce the suitable positioning of the three residues
(K134, N187 and S216) for substrate binding and the
expression of l-aspDH’s catalytic activity.
In general, NAD(P)H-dependent dehydrogenases
show pro-R or pro-S stereospecificity for hydrogen
removal from the C4 position of the nicotinamide moi-
ety of NAD(P)H. In our binding model, the re-face of
the nicotinamide ring is in front of the a-hydrogen
atom of substrate (Fig. 4B), and that is in good agree-
ment with our earlier finding that A. fulgidus l-aspDH
belongs to the dehydrogenase group with pro-R-spe-
cific hydrogen transfer [5].
Cofactor binding
The electron density corresponding to the NAD coen-
zyme bound within the active site was very clear,
which enabled us to place the ligand with reasonable
accuracy (Fig. S1). The map enabled clear positioning

an anti conformation to the glycosidic bond linking
the nicotinamide ring and its associated ribose moiety.
N7, N1 and N6 of the adenine base, respectively,
formed hydrogen bonds with the side chains of S59
and Y66 and, via a water molecule (WAT26), with the
side chain of D65. The hydroxyl groups of the adenine
ribose interact with R33 using side-on geometry. A
glycine-rich motif, GXGXXG, which lies close to the
adenine ribose, dictates the nature of the hydrogen
bonding between the main chain and the adenine
ribose moiety [7]. In A. fulgidus l-aspDH, this motif is
recognized at the position of the 7–12th amino acids
from the N-terminus (Fig. 2). It is known that the
occurrence of an aspartic or glutamic acid residue at
the C- or N-terminus of the second b strand of the bab
fold is a common feature of NAD(P)-dependent dehy-
drogenases [8–12]. The acidic residue plays an impor-
tant role in the formation of hydrogen bonds with the
adenine ribose hydroxyl groups of the cofactor [13,14].
In A. fulgidus l-aspDH, D31, which is at the N-termi-
nal end of b2, forms hydrogen bonds with the 2¢ and
3¢ hydroxyl groups of the adenine ribose of NAD
(Fig. 5). The adenine phosphate interacts with A10,
I11, a water molecule (WAT119) and, via water mole-
cules, with R33 (WAT125) and A57 (WAT5). In addi-
tion, the nicotinamide phosphate interacts with N212,
T215 and a water molecule (WAT119) and, via a water
molecule (WAT5), with A57. The hydroxyl groups of
the nicotinamide ribose interact with the side chain of
S81, N162 and backbone oxygen of A58. The hydro-

Fig. 3. Comparison of the structures of A. fulgidus L-aspDH and
T. maritima
L-aspDHs. (A) The superimposed Ca-traces of A. fulgi-
dus
L-aspDH and T. maritima L-aspDH; the structures of the two
enzymes are shown in green and magenta, respectively. NAD
(magenta) and citrate (yellow) molecules are shown as sphere mod-
els. (B) Scheme around the substrate binding loop in the catalytic
domain. The citrate molecule is shown as a stick model in yellow.
Oxygen and nitrogen atoms are shown in red and blue, respec-
tively.
L-AspDHs are colored as in (A).
K. Yoneda et al. Crystal structure of
L-aspDH from A. fulgidus
FEBS Journal 274 (2007) 4315–4325 ª 2007 The Authors Journal compilation ª 2007 FEBS 4319
dimer, 21 381 A
˚
2
). Likewise, the total interface area
between A. fulgidus l-aspDH subunits A–B (2322 A
˚
2
)
is similar to that between the T. maritima l-aspDH
subunits (2201 A
˚
2
), and the total hydrophobic areas
of the interfaces are about the same (972–973 A
˚

the
L-aspartate. The C4 atom of the pyridine
ring (a hydride acceptor site), and si- and
re-faces are labeled. The
L-aspartate mole-
cule is shown as a stick model in cyan. The
structures of A. fulgidus and T. maritima
L-aspDH are shown in green and white,
respectively. Atoms are colored as
described for Fig. 3.
Fig. 5. Stereo representation of NAD bound
to A. fulgidus
L-aspDH. Residues that inter-
act with NAD are labeled. The networks of
hydrogen bonds are shown as dotted lines.
Atoms are colored as described for Fig. 3.
Crystal structure of
L-aspDH from A. fulgidus K. Yoneda et al.
4320 FEBS Journal 274 (2007) 4315–4325 ª 2007 The Authors Journal compilation ª 2007 FEBS
the A. fulgidus and T. maritima l-aspDH monomers
were determined to be 149 and 131, respectively, and
the number of intersubunit hydrogen bonds was 26
and 28, respectively. Again, the two enzymes showed
considerable similarity.
Comparison of the amino acid compositions
The higher numbers of Pro at the N1 position or Ala
and lower numbers of b-branched residues (Val, Thr
and Ile) are known to correlate significantly with the
thermostability of proteins by stabilizing their a helices
[16–18]. The total numbers of Pro at the N1 position

Using a cut-off distance of 4.0 A
˚
between oppositely
charged residues, we calculated that A. fulgidus
l-aspDH contained 16 intrasubunit ion pairs, whereas
T. maritima l-aspDH contained only nine (Table 2).
In addition, three major intersubunit ion-pair interac-
tions were observed in A. fulgidus l-aspDH: E201–
R203¢, R203–E201¢ and R95–E232¢ (the prime indi-
cates the neighboring subunit in the dimer). An ion
pair between E232 and R95¢ could not be observed
because of the poor electron density of the side chains.
Four (E201, R203, E201¢, and R203¢) of the residues
are located within b11 and b11¢, and form a four-resi-
due ion-pair network between the A and B subunits
(Fig. 6). In the T. maritima l-aspDH, the charged R95
is replaced by aromatic F93, which is involved in the
largest aromatic pair network in T. maritima enzyme
(see below). In addition, T. maritima l-aspDH con-
tains no ion-pair networks (Table 2).
Aromatic pair interaction
Aromatic interactions are also known to participate in
stabilizing protein structure [22,23]. A pair of aromatic
Table 2. Comparison of ion-pair interactions, aromatic pair interac-
tions, accessible surface areas, and hydrogen bonding in A. fulgidus
and T. maritima
L-aspDHs.
A. fulgidus T. maritima
PDB code 2DC1 1J5P
Resolution (A

and magenta, respectively. Atoms are colored as described for
Fig. 3.
K. Yoneda et al. Crystal structure of
L-aspDH from A. fulgidus
FEBS Journal 274 (2007) 4315–4325 ª 2007 The Authors Journal compilation ª 2007 FEBS 4321
interactions contributes between )0.6 and )1.3 kcalÆ
mol
)1
to protein stability [24]. Using a cut-off distance
between the aromatic ring centers of 7.0 A
˚
, we deter-
mined that there are six intrasubunit aromatic pairs in
A. fulgidus l-aspDH (Table 2). By contrast, there are
11 aromatic pairs in T. maritima l-aspDH. The most
extensive aromatic pair network (seven residues; Y75,
F83, F88, F92, F93, F105, and F240¢) is located in the
Rossmann-fold domain (Fig. S2), and two of those
residues (F105 and F240¢) form an intersubunit aro-
matic pair. By contrast, the largest aromatic pair net-
work identified in A. fulgidus l-aspDH is composed of
only four residues (F14, W18, F24, and Y217) and is
located on the surface of the Rossmann-fold domain
(Fig. S2). It is noteworthy that whereas five major
intersubunit aromatic pair interactions (F105–F240¢,
F122–Y225¢, F206–F206¢, Y225–F122¢, and F240–
F105¢) were observed in T. maritima l-aspDH, there
are no intersubunit aromatic pairs in the A. fulgidus
enzyme (Fig. S2). Structure-based sequence alignment
showed that with the exception of F83 and F88, the

, c ¼ 100.49 A
˚
and
a ¼ b ¼ c ¼ 90°. The crystals were grown in sitting drops
in which 1 lL of enzyme solution (14.5 mgÆmL
)1
) contain-
ing 1 mm NAD was mixed with 1 lL of mother liquor con-
taining 100 mm phosphate-citrate buffer pH 4.2 (60.5 mm
Na
2
HPO
4
, 39.5 mm citric acid), 5% (v ⁄ v) polyethylene gly-
col 3000 (PEG 3000), 10% (v ⁄ v) glycerol and 22% (v ⁄ v)
1.2-propanediol.
Data collection
Crystals were coated with a layer of viscous oil (Paratone-N)
and transferred into a stream of nitrogen gas for data col-
lection at 100 K. Diffraction data were collected at a reso-
lution of 1.9 A
˚
on beamline KEK-NW12 at the Photon
Factory (Tsukuba, Japan) using monochromatized radia-
tion at k ¼ 1.0 A
˚
and an ADSC Quantum 210 CCD detec-
tor (Area Detector Systems, San Diego, CA, USA). The
oscillation angle per image was set to 1°, and the data were
processed using HKL 2000 [25]. Heavy atom derivatives

olution using refmac [29] and cns [30]. The model was
adjusted in xtal view using both |F
o
| ) |F
c
| and 2|F
o
| ) |F
c
|
maps. NAD and citrate molecules were clearly visible in
both the r
A
-weighted |F
o
| ) |F
c
| and 2|F
o
| ) |F
c
| maps, and
these molecules were included in the latter part of the refine-
ment. The current model contains 472 residues (A 1–236
and B 1–236), 136 water molecules, two NAD and two
citrate molecules. The model geometry was analyzed using
procheck [31], and 91.2% of the nonglycine residues were
in the most favored region of the Ramachandran plot and
8.8% in the additionally allowed region. Molecular graphics
were created using pymol (http://pymol.sourceforge.net/).

100 °C, after which the residual activity of the enzymes was
determined at appropriate intervals using a standard assay
as described previously [5].
Acknowledgements
Data collection was performed at the Photon Factory
(Tsukuba, Japan). We thank Drs K. Demura, N. Mats-
ugaki, N. Igarashi and S. Wakatsuki for their kind assis-
tance with the data collection. This work was supported
in part by the ‘National Project on Protein Structural
and Functional Analysis’ promoted by the Ministry of
Education, Science, Sports, Culture, and Technology of
Japan and by a Grant-in-Aid for Scientific Research (C)
from the Japan Society for the Promotion of Science.
References
1 Nasu S, Wicks FD & Gholson RK (1982) The mamma-
lian enzyme which replaces B protein of E. coli quinoli-
nate synthetase is d-aspartate oxidase. Biochim Biophys
Acta 704, 240–252.
2 Rizzi M & Schindelin H (2002) Structural biology of
enzymes involved in NAD and molybdenum cofactor
biosynthesis. Curr Opin Struct Biol 12, 709–720.
3 Sakuraba H, Satomura T, Kawakami R, Yamamoto S,
Kawarabayasi Y, Kikuchi H & Ohshima T (2002)
l-Aspartate oxidase is present in the anaerobic hyper-
thermophilic archaeon Pyrococcus horikoshii OT-3:
characteristics and role in the de novo biosynthesis of
nicotinamide adenine dinucleotide proposed by genome
sequencing. Extremophiles 6, 275–281.
4 Yang Z, Savchenko A, Yakunin A, Zhang R, Edwards
A, Arrowsmith C & Tong L (2003) Aspartate dehydroge-

6401.
12 Nishiyama M, Birktoft JJ & Beppu T (1993) Alteration
of coenzyme specificity of malate dehydrogenase from
Thermus flavus by site-directed mutagenesis. J Biol
Chem 268, 4656–4660.
13 Baker PJ, Sawa Y, Shibata H, Sedelnikova SE & Rice
DW (1998) Analysis of the structure and substrate bind-
ing of Phormidium lapideum alanine dehydrogenase.
Nat Struct Biol 5, 561–567.
14 Wierenga RK, de Maeyer MCH & Hol WGJ (1985)
Interaction of pyrophosphate moieties with a-helixes in
dinucleotide-binding proteins. Biochemistry 24, 1346–
1357.
15 Chan MK, Mukund S, Kletzin A, Adams MW & Rees
DC (1995) Structure of a hyperthermophilic tungsto-
pterin enzyme, aldehyde ferredoxin oxidoreductase.
Science 267, 1463–1469.
16 Yun RH, Anderson A & Hermans J (1991) Proline in
a-helix: stability and conformation studied by dynamics
simulation. Proteins 10, 219–228.
17 Padmanabhan S, Marqusee S, Ridgeway T, Laue TM &
Baldwin RL (1990) Relative helix-forming tendencies of
nonpolar amino acids. Nature 344, 268–270.
18 Facchiano AM, Colonna G & Ragone R (1998)
Helix-stabilizing factors and stabilization of thermo-
philic proteins: an X-ray based study. Protein Eng 11,
753–760.
19 Hennig M, Darimont B, Sterner R, Kirschner K &
Jansonius JN (1995) 2.0 A
˚

849–861.
27 Terwilliger TC (2000) Maximum-likelihood density
modification. Acta Crystallogr D56, 965–972.
28 McRee DE (1999) XtalView ⁄ Xfit – A versatile program
for manipulating atomic coordinates and electron den-
sity. J Struct Biol 125, 156–165.
29 Murshudov GN, Vagin AA & Dodson EJ (1997)
Refinement of macromolecular structures by the maxi-
mum-likelihood method. Acta Crystallogr D53, 240–255.
30 Brunger AT, Adams PD, Clore GM, DeLano WL, Gros
P, Grosse-Kunstleve RW, Jiang JS, Kuszewski J, Nilges
M, Pannu NS et al. (1998) Crystallography & NMR
system: a new software suite for macromolecular struc-
ture determination. Acta Crystallogr D54, 905–921.
31 Laskowski RA, MacArthur MW, Moss DS & Thornton
JM (1993) PROCHECK: a program to check the stereo-
chemical quality of protein structure. J Appl Crystallogr
26, 283–291.
32 Rodriguez R, Chinea G, Lopez N, Pons T & Vriend G
(1998) Homology modeling, model and software evalua-
tion: three related resources. Bioinformatics 14, 523–528.
33 Barlow DJ & Thornton JM (1983) Ion-pairs in proteins.
J Mol Biol 168, 867–885.
34 Nicholls A, Sharp KA & Honig B (1991) Protein fold-
ing and association: insights from the interfacial and
thermodynamic properties of hydrocarbons. Proteins 11,
281–296.
35 Thompson JD, Higgins DG & Gibson TJ (1994) CLUS-
TAL W: improving the sensitivity of progressive multi-
ple sequence alignment through sequence weighting,

for the content or functionality of any supplementary
materials supplied by the authors. Any queries (other
than missing material) should be directed to the corre-
sponding author for the article.
K. Yoneda et al. Crystal structure of L-aspDH from A. fulgidus
FEBS Journal 274 (2007) 4315–4325 ª 2007 The Authors Journal compilation ª 2007 FEBS 4325