Characterization of L-aspartate oxidase and quinolinate
synthase from Bacillus subtilis
Ilaria Marinoni
1
, Simona Nonnis
2
, Carmine Monteferrante
1
, Peter Heathcote
3
, Elisabeth Ha
¨rtig
4
,
Lars H. Bo
¨ttger
5
, Alfred X. Trautwein
5
, Armando Negri
2
, Alessandra M. Albertini
1
and
Gabriella Tedeschi
2
1 Department of Genetics and Microbiology, University of Pavia, Italy
2 D.I.P.A.V., Section of Biochemistry, University of Milano, Italy
3 School of Biological and Chemical Sciences, Queen Mary College, University of London, UK
4 Institute of Microbiology, Technical University of Braunschweig, Germany
5 Institute of Physics, University of Lu
¨beck, Germany
NAD is a ubiquitous and essential molecule in all
living organisms. In addition to its well-established
role in redox biochemistry and energetic metabolism,
NAD can function as a signaling molecule in a variety
of cellular processes [1]. In eubacteria, NAD is pro-
duced by a de novo pathway or starting from pre-
formed nicotinic acid. Quinolinic acid is the precursor
for the de novo pathway; in most eukaryotes, it is pro-
duced via degradation of tryptophan, whereas in many
eubacteria, including several pathogens, it is synthe-
sized from l-aspartate and dihydroxyacetone phos-
phate (DHAP). This reaction involves the so-called
quinolinate synthase complex: the first enzyme,
l-aspartate oxidase (NadB, EC 1.4.3.16), encoded by
the gene nadB, catalyzes the oxidation of l-aspartate
to iminoaspartate; the second enzyme, quinolinate syn-
thase (NadA), is encoded by the gene nadA and cata-
lyzes the condensation between iminoaspartate and
DHAP, resulting in quinolinic acid production
(Scheme 1) [2]. Quinolinic acid is then converted to
Keywords
L-aspartate oxidase; NAD biosynthesis;
NadA; NadB; quinolinate synthase
Correspondence
G. Tedeschi, D.I.P.A.V., Section of
Biochemistry, University of Milano, Via
Celoria 10, 20133 Milano, Italy.
Fax: +39 02 50318123
Tel: +39 02 50318127
E-mail: gabriella.tedeschi@unimi.it
(Received 4 July 2008, revised 1 August
2008, accepted 12 August 2008)
doi:10.1111/j.1742-4658.2008.06641.x
NAD is an important cofactor and essential molecule in all living organ-
isms. In many eubacteria, including several pathogens, the first two steps in
the de novo synthesis of NAD are catalyzed by l-aspartate oxidase (NadB)
and quinolinate synthase (NadA). Despite the important role played by
these two enzymes in NAD metabolism, many of their biochemical and
structural properties are still largely unknown. In the present study, we
cloned, overexpressed and characterized NadA and NadB from Bacil-
lus subtilis, one of the best studied bacteria and a model organism for low-
GC Gram-positive bacteria. Our data demonstrated that NadA from
B. subtilis possesses a [4Fe–4S]
2+
cluster, and we also identified the cysteine
residues involved in the cluster binding. The [4Fe–4S]
2+
cluster is coordi-
nated by three cysteine residues (Cys110, Cys230, and Cys320) that are
conserved in all the NadA sequences reported so far, suggesting a new non-
canonical binding motif that, on the basis of sequence alignment studies,
may be common to other quinolinate synthases from different organisms.
Moreover, for the first time, it was shown that the interaction between
NadA and NadB is not species-specific between B. subtilis and Escherichia
coli.
Abbreviations
DHAP, dihydroxyacetone phosphate; GST, glutathione S-transferase; GST–NadA, quinolinate synthase fused to glutathione S-transferase
(GST) at its N-terminus; IPTG, isopropyl thio-b-D-galactoside; NadA, quinolinate synthase; NadA–His, quinolinate synthase with a His
6
-tag at
the N-terminus; NadB, L-aspartate oxidase.
5090 FEBS Journal 275 (2008) 5090–5107 ª2008 The Authors Journal compilation ª2008 FEBS
nicotinic acid and, finally, to NAD by a biosynthetic
pathway common to all organisms. As NadA and
NadB are absent in mammals, they are considered to
be ideal targets for the development of novel prophy-
lactic and therapeutic agents [3]. Moreover, very
recently, Pruinier et al. [4] reported that the pathogenic
bacterium Shigella is a nicotinic acid auxotroph,
unable to synthesize NAD via the de novo pathway,
due to nadA and nadB gene mutations. When the func-
tionality of nadA ⁄Bin Shigella was restored, a consis-
tent loss of virulence and inability to invade host cells
were observed. On the basis of this result, they defined
NadA and NadB as antivirulence loci.
Besides being important in bacteria, NadA and
NadB analogs seem to be involved in NAD biosynthe-
sis also in plants. Many experimental findings, together
with the apparent absence of genes encoding enzymes
involved in other possible routes to quinolinate, sug-
gest that several plants may obtain this key precursor
via the aspartate pathway, like many bacteria [5–7].
Therefore, because of the growing amount of evi-
dence indicating the importance in several organisms
of de novo NAD biosynthesis through the reaction
catalyzed by NadA and NadB, it is of the utmost
importance to gain a thorough knowledge of the bio-
chemical and structural properties of these two
enzymes.
The gene nadB is present in several microorganisms
and in plants, but the protein has been purified only
from Escherichia coli,Pyrococcus horikoshii and Sulfol-
obus tokadaii, and characterized from a biochemical
and structural point of view only from E. coli and
S. tokadaii [8–17]. It is a flavoprotein containing 1 mol
of noncovalently bound FAD ⁄mol of protein. This
enzyme presents several peculiarities that distinguish it
from all other flavo-oxidases: (a) in vitro, it is able to
use different electron acceptors such as oxygen, fuma-
rate, cytochrome cand quinones [9], suggesting that it
is involved in NAD biosynthesis in anaerobic as well
as aerobic conditions; and (b) the primary and tertiary
structures are not similar to those of other flavo-oxid-
ases, but to those of the flavoprotein subunit of the
succinate dehydrogenase ⁄fumarate reductase class of
enzymes. As a consequence, NadB shares with these
proteins most of the active site features, including the
presence of an arginine playing an acid–base role in
catalysis [11–15]. Accordingly, NadB can reduce fuma-
rate, but it is unique in that it is able to stereospecifi-
cally oxidize l-aspartate and is unable to oxidize
succinate.
Interestingly, in 2003, Yang et al. [18] described
another enzyme, from Thermotoga maritima, that is
involved in the de novo biosynthesis of NAD and that
plays the same role as NadB, although it does not
share any recognizable sequence similarity to NadB. It
is described as NADP-dependent l-aspartate dehydro-
genase, and is strictly specific for l-aspartate. This
enzyme produces iminoaspartate, which is then con-
verted to quinolinate through the condensation with
DHAP catalyzed by NadA.
The second enzyme involved in the de novo biosyn-
thesis of NAD, NadA, is extremely sensitive to oxygen;
therefore, it has been poorly characterized so far, and
very little is known regarding its biochemical and
structural properties. The enzyme has only been puri-
fied from E. coli [19–21] and P. horikoshii [22]. Recent
studies on the enzyme from E. coli have demonstrated
that the protein harbors a [4Fe–4S]
2+
cluster [20,21]
that, as it is very sensitive to oxygen, probably explains
why NadA is identified as the site of oxygen poisoning
of NAD synthesis in anaerobic bacteria [23]. The 3D
structure has been obtained for the enzyme from
P. horikoshii [22]. The protein shows a triangular
architecture in which conserved amino acids determine
three structurally homologous domains. Unfortunately,
the structure lacks any data on the [Fe–S] center, and
the three surface loops that contain two highly
conserved cysteine residues are disordered. Moreover,
Scheme 1. Reaction catalyzed by the ‘quinolinate synthase complex’. The first enzyme, NadB, catalyzes the oxidation of L-aspartate to
iminoaspartate using either oxygen or fumarate as electron acceptor for FAD reoxidation; the second enzyme, NadA, catalyzes the conden-
sation between iminoaspartate and DHAP, resulting in quinolinic acid production.
I. Marinoni et al. NadA and NadB from B. subtilis
FEBS Journal 275 (2008) 5090–5107 ª2008 The Authors Journal compilation ª2008 FEBS 5091
the canonical binding motif for [4Fe–4S]
2+
clusters
(CXXCXXC) that is found in the C-terminal regions
of most quinolinate synthases from bacteria, including
E. coli [19–21], is absent in NadA from P. horikoshii,
and in this case the cofactors remain to be identified
[22]. The absence of the consensus sequence for the
binding of a [4Fe–4S]
2+
cluster is observed also in
NadA from several plants. On the other hand, very
recently Murthy et al. [7] described a new SufE-like
protein from Arabidopsis thaliana chloroplasts that
contains two domains, one SufE-like domain and one
with similarity to the bacterial NadA carrying a highly
oxygen-sensitive [4Fe–4S]
2+
cluster. Therefore, two
important areas have to be clarified: (a) the nature of
the cofactor for quinolinate synthase, in particular for
NadA proteins that do not contain a canonical binding
motif for a [4Fe–4S]
2+
cluster; and (b) the identifica-
tion of the residues involved in the binding of the
[4Fe–4S]
2+
cluster, if present.
In an attempt to resolve some of these issues, we
cloned, overexpressed and characterized NadA and
NadB from Bacillus subtilis, one of the best studied
bacteria and a model for low-GC Gram-positive bacte-
ria, including pathogens. Our data add new informa-
tion regarding the NadA cofactor and the interaction
between NadA and NadB. In particular, it is demon-
strated that the cofactor for NadA from B. subtilis is a
[4Fe–4S]
2+
cluster, even though the sequence does not
show a canonical binding motif. Moreover, for the first
time, the cysteines involved in the cluster binding are
identified. Taken together, our data suggest that in
NadA from B. subtilis, the [4Fe–4S]
2+
cluster is coor-
dinated by three strictly conserved cysteine residues
(Cys110, Cys230, and Cys320). Thus, NadA presents a
new noncanonical binding motif that, on the basis of
sequence alignment studies, may be common to other
quinolinate synthases from different sources. More-
over, the results show for the first time that the inter-
action between NadA and NadB is not species-specific
between the proteins from B. subtilis and E. coli.
Results and Discussion
NadA cloning and protein purification and
characterization
In order to optimize the heterologous production and
purification of B. subtilis NadA, several expression
vectors with different tags were utilized: NadA with a
His
6
-tag at the N-terminus, NadA with a His
6
-tag at
the C-terminus (NadA–His), and NadA fused to gluta-
thione S-transferase (GST) at its N-terminus (GST–
NadA). The best results in terms of soluble protein
yield were obtained by cloning the nadA gene in
pET28-a with the His-tag at the C-terminal region.
Upon purification in a glove box under anaerobic con-
ditions, a soluble pure protein, brown in color, was
obtained with a yield of 10 mg of pure protein from
1L of E. coli culture expressing B. subtilis NadA
(Fig. 1A). Therefore, this protein was utilized for
further studies. As determined by gel filtration, it is a
trimer of 124 kDa (expected molecular mass for the
monomer 41 kDa) under both aerobic and anaerobic
conditions (data not shown).
To evaluate its enzymatic activity, quinolinate for-
mation was measured by a discontinuous enzymatic
assay that couples the production of iminoaspartate by
NadB with the condensation between DHAP and
iminoaspartate to form quinolinic acid catalyzed by
NadA [19] (Scheme 1). As described below, NadB is
able to use both molecular oxygen and fumarate as
electron acceptors for FAD reoxidation. Therefore, to
better evaluate NadA activity, the assays were per-
formed under aerobic and anaerobic conditions (in the
presence of fumarate), using recombinant B. subtilis
NadA plus B. subtilis NadB, overexpressed and puri-
fied as detailed below. Different concentrations of
NadA, NadB and fumarate (under anaerobic condi-
tions) were utilized in order to set up a suitable assay
to be used to check NadA activity. The data showed
that: (a) the assay is linear up to 0.25 mg of NadA;
(b) 10 lg of NadB is the lowest amount suitable to
measure NadA activity; and (c) under anaerobic condi-
tions, NadA activity becomes independent of fumarate
concentration, starting from 1 mmfumarate, but
decreases at concentrations higher than 2 mmfuma-
rate, due to inhibition of NadB by fumarate [9]. There-
fore, to evaluate quinolinate formation, the assay
routinely used contained 70 lg of NadA, 30 lgof
NadB and 1 mmfumarate under anaerobic conditions.
AB
Fig. 1. Production and purification of NadA from Bacillus subtilis.
(A) 11% SDS ⁄PAGE of NadA–His before and after purification in a
glove box. Std, molecular markers; P, pellet; S, soluble fraction;
NadA–His, purified protein. (B) Visible absorption spectrum of
NadA–His purified under anaerobic conditions (
_______
) and after 2 h
of exposure to air (- - -).
NadA and NadB from B. subtilis I. Marinoni et al.
5092 FEBS Journal 275 (2008) 5090–5107 ª2008 The Authors Journal compilation ª2008 FEBS
An apparent K
m
of 0.36 ± 0.05 mmwas calculated for
DHAP at 25 C, using oxygen as electron acceptor.
The specific activity of NadA from B. subtilis was
0.05 ± 0.01 lmolÆmin
)1
Æmg
)1
in the presence of fuma-
rate as electron acceptor for NadB, and 0.027 ±
0.01 lmolÆmin
)1
Æmg
)1
using oxygen to reoxidize
NadB. These values were more than two times
higher than that reported for NadA from E. coli
(0.015 lmolÆmin
)1
Æmg
)1
using fumarate) [20,21],
and comparable to the results described for SufE3
purified from A. thaliana, which catalyzes the for-
mation of quinolinate with a specific activity of
0.05 lmolÆmin
)1
Æmg
)1
using fumarate as electron
acceptor for NadB [7]. Similar data were obtained if
NadA without tags or GST–NadA was used in the
assay mixture instead of NadA–His, ruling out
the possibility that the presence of a tag at either the
N-terminus or C-terminus had any effect on the
enzymatic activity.
NadA from B. subtilis contains an oxygen-labile
[4Fe–4S]
2+
cluster as a cofactor
Figure 1B shows the absorbance spectrum of NadA
from B. subtilis purified under anaerobic conditions.
The shoulder at 420 nm in the spectrum suggests the
presence of an [Fe–S] cluster in the protein. This clus-
ter appears to be oxygen-sensitive, because absorption
in the visible region was altered after exposure to air,
with a progressive decrease of the absorption in the
420 nm region (Fig. 1B). Using the protein purified in
the glove box, it was possible to determine that the
protein contained 3.8 ± 0.2 mol iron ⁄mol NadA and
3.3 ± 0.01 mol inorganic sulfide ⁄mol protein, suggest-
ing the presence of one [4Fe–4S]
2+
cluster per mono-
mer of NadA. In accordance with this finding,
complete loss of activity was detected for NadA from
B. subtilis purified under anaerobic conditions and
exposed to oxygen overnight, as the cluster integrity is
compromised in such conditions. These results are in
agreement with the data reported for NadA from
E. coli and A. thaliana, which contain one highly oxy-
gen-sensitive [4Fe–4S]
2+
cluster per monomer of NadA
[7,20,21]. The data are in keeping with the hypothesis
proposed by Sun & Setlow [24], who suggested that
NadA from B. subtilis may contain an [Fe–S] cluster,
on the basis of the observation that, like E. coli,
B. subtilis iscS
)
strains are auxotrophic for nicotinic
acid and are unable to synthesize NAD de novo.
Further characterization of the [Fe–S] cofactor was
performed by Mo
¨ssbauer and EPR spectroscopy
(Figs 2 and 3, respectively). The Mo
¨ssbauer spectrum
of an NadA sample, recorded at 77 K, is shown in
Fig. 2A. At first glance, it is appropriate to fit this
spectrum with one quadrupole doublet, representing
100% of the iron in the NadA sample. The resulting
fit parameters (isomer shift d= 0.44 mmÆs
)1
, quadru-
pole splitting DE
Q
= 1.05 mmÆs
)1
, and line width
G= 0.48 mmÆs
)1
) are characteristic for [4Fe–4S]
2+
clusters. The [4Fe–4S]
2+
clusters in other biological
systems exhibit similar Mo
¨ssbauer parameters
[21,25,26]. The Mo
¨ssbauer spectrum of NadA that was
exposed to air at room temperature (for 30 min), mea-
sured at 77 K, reveals that the [4Fe–4S]
2+
clusters are
oxygen-sensitive and are decomposed (Fig. 2B).
About 55% of the iron in that spectrum still repre-
sents [4Fe–4S]
2+
clusters; the remaining 45% of
the absorption pattern appears as a quadrupole
doublet with Mo
¨ssbauer parameters (d= 0.27 mmÆs
)1
,
DE
Q
= 0.53 mmÆs
)1
and G= 0.35 mmÆs
)1
) that are
A
B
C
Fig. 2. NadA contains a [4Fe–4S] cluster: Mo
¨ssbauer spectra mea-
sured at 77 K. (A) The quadrupole doublet represents [4Fe–4S]
2+
clusters. (B) NadA exposed to air for 30 min. The two quadrupole
doublets represent [4Fe–4S]
2+
clusters (dashed line) and, in addi-
tion, high-spin (S=5⁄2) tetrahedral-sulfur-coordinated iron sites
(dotted line) (see text). (C) Reanalysis of the measured spectrum
from (A) with two quadrupole doublets representing the 3 : 1 bind-
ing motif of the [4Fe–4S]
2+
clusters (see text). The solid line is the
envelope of the dashed and dotted lines in (B) and (C).
I. Marinoni et al. NadA and NadB from B. subtilis
FEBS Journal 275 (2008) 5090–5107 ª2008 The Authors Journal compilation ª2008 FEBS 5093
characteristic for high-spin (S=5⁄2) tetrahedral-sul-
fur-coordinated iron sites as observed in [2Fe–2S]
2+
clusters [21,27]. Oxygen sensitivity has been observed
for [4Fe–4S]
2+
clusters in other proteins, where this
sensitivity leads to partial or total degradation of these
clusters. Such proteins have in common that their
[4Fe–4S]
2+
clusters are ligated by only three cysteines,
whereas the fourth iron is coordinated by a nonprotein
ligand [28,29]. It was thus tempting to reanalyze the
Mo
¨ssbauer spectrum of NadA (which was not exposed
to air), but now assuming two different kinds of iron
sites in the [4Fe–4S]
2+
cluster, which is coordinated
with a 3 : 1 ratio of cysteine to noncysteine. This situa-
tion requires two quadrupole doublets with an area
ratio of 3 : 1, instead of one doublet only. A tetrahe-
dral-coordinated Fe
2.5+
site, with the cysteine ligand
replaced by a nonsulfur ligand, i.e. nitrogen or oxygen,
is expected to exhibit an increase of isomer shift by
around 0.05–0.1 mmÆs
)1
in comparison with the three
tetrahedral-sulfur-coordinated Fe
2.5+
sites. Visualiza-
tion of this specific 3 : 1 binding motif in a Mo
¨ssbauer
spectrum was provided before for the [4Fe–4S]
2+
clus-
ters in the ferredoxin of the anaerobic ribonucleotide
reductase from E. coli [30], in the ferredoxin from the
hyperthermophilic archeon Pyrococcus furiosus [31], in
the (E)-4-hydroxy-3-methylbut-2-enyl diphosphate syn-
thase from A. thaliana [25], and in the radical S-adeno-
sylmethionine enzyme coproporphyrinogen III oxidase
HemN [26]. A corresponding fit of the Mo
¨ssbauer
spectrum of NadA using two quadrupole doublets
(Fig. 2C) yields the following: doublet I (dashed
line; the relative absorption area 75% was fixed
in the fit; d
I
= 0.42 mmÆs
)1
,DE
QI
= 1.05 mmÆs
)1
,
G
I
= 0.43 mmÆs
)1
) represents tetrahedral-sulfur-coordi-
nated Fe
2.5+
sites, and doublet II (dotted line; the rela-
tive area 25% was fixed in the fit; d
II
= 0.52 mmÆs
)1
,
DE
QII
= 1.09 mmÆs
)1
,G
II
= 0.42 mmÆs
)1
) represents
the tetrahedral-coordinated Fe
2.5+
site with the cyste-
ine ligand replaced by a noncysteine ligand. The
Mo
¨ssbauer parameters of the two doublets are in rea-
sonable agreement with those reported for [4Fe–4S]
2+
clusters in other proteins with this specific 3 : 1 bind-
ing motif [25,26,30,31].
The EPR spectrum of the ‘as isolated’ protein (not
presented) showed a trace contribution from a
[3Fe–4S] center at g= 2.031 and a relatively minor
amount of free iron at g= 4.2. However, the spec-
trum did contain a relatively significant contribution
from Cu
2+
, so the ‘as isolated’ spectrum was sub-
tracted from the spectra of the reduced samples to pre-
vent this baseline signal distorting the EPR spectra of
the [Fe–S] center at low fields. The difference
(reduced )oxidized) EPR spectra of the NadA protein
reduced at pH 8 and pH 10 are presented in Fig. 3.
The sharp derivative signal around g= 2.00 arises
from the radical of methyl viologen, which was added
to the samples as a redox mediator. The EPR spec-
trum of the reduced NadA protein produces an EPR
spectrum in the pH 10 sample at 15 K, which is typical
of a [4Fe–4S]
1+
center with g
1
= 2.054 and
g
2,3
= 1.932. Interestingly, the sample at pH 8 indi-
cates that the [Fe–S] center exists in two slightly differ-
ent forms, with this difference being indicated by a
split in the high field feature with features at g= 1.94
and g= 1.89. This could be caused by slight differ-
ences in folding of the protein, or a charged residue
close to the [Fe–S] center that has a pK
a
close to that
of the sample at pH 8 so that it is only charged in a
fraction of the samples (about 50%). The shift to
pH 10 clearly favors the g= 1.93 ⁄g= 1.94 fea-
ture ⁄conformation ⁄state. Given that it is thought from
studies reported in this article that the fourth ligand to
this [4Fe–4S]
2+
center is not a cysteine, it is tempting
to speculate that the two different forms of the [Fe–S]
cluster detected at pH 8 may reflect differences in the
fourth noncysteine ligand. We estimate that 70–80%
of the maximal [Fe–S] content of these samples is
contributing to the EPR spectra recorded.
Identification of [Fe–S] cluster-binding residues
of NadA
Recent studies on NadA from E. coli and on SufE3
from A. thaliana demonstrated that these enzymes
harbor a [4Fe–4S]
2+
cluster that is essential for the
A
B
Fig. 3. NadA contains a [4Fe–4S] cluster: EPR spectra of NadA
reduced at pH 8 and pH 10. NadA was reduced with sodium dithio-
nite and methyl viologen, as described in Experimental procedures.
The two spectra presented represent the difference between the
reduced sample and the unreduced control. The experimental con-
ditions for acquisition of the spectrum were: microwave power,
2 mW; modulation amplitude, 0.1 mT; temperature, 15 K.
NadA and NadB from B. subtilis I. Marinoni et al.
5094 FEBS Journal 275 (2008) 5090–5107 ª2008 The Authors Journal compilation ª2008 FEBS