Crystallographic Trapping of Reaction Intermediates in Quinolinic Acid Synthesis by NadA

Article abstract of DOI:10.1021/acschembio.7b01104

NadA is a multifunctional enzyme that condenses dihydroxyacetone phosphate (DHAP) with iminoaspartate (IA) to generate quinolinic acid (QA), the universal precursor of the nicotinamide adenine dinucleotide (NAD(P)) cofactor. Using X-ray crystallography, we have (i) characterized two of the reaction intermediates of QA synthesis using a pH-shift approach and a slowly reacting Thermotoga maritima NadA variant and (ii) observed the QA product, resulting from the degradation of an intermediate analogue, bound close to the entrance of a long tunnel leading to the solvent medium. We have also used molecular docking to propose a condensation mechanism between DHAP and IA based on two previously published Pyrococcus horikoshi NadA structures. The combination of reported data and our new results provide a structure-based complete catalytic sequence of QA synthesis by NadA.

Full text of DOI:10.1021/acschembio.7b01104

Articles
Cite This: ACS Chem. Biol. XXXX, XXX, XXX−XXX
Crystallographic Trapping of Reaction Intermediates in Quinolinic
Acid Synthesis by NadA
Anne Volbeda,† Jaione Saez Cabodevilla,‡,$ Claudine Darnault,† Oceane Gigarel,† Thi-Hong-Lien Han,‡
́
Oriane Renoux,† Olivier Hamelin,γ Sandrine Ollagnier-de-Choudens,‡ Patricia Amara,†
and Juan C. Fontecilla-Camps*,†
†Univ. Grenoble Alpes, CEA, CNRS, IBS, Metalloproteins Unit, F-38000 Grenoble, France
‡Univ. Grenoble Alpes, CNRS, CEA, Laboratoire de Chimie et Biologie des Met
́
aux, BioCat, 38000, Grenoble, France
artement de Pharmacochimie Moleculaire, F-38041, Grenoble, France
γUniv. Grenoble Alpes, CNRS, CEA, Laboratoire de Chimie et Biologie des Met
aux, BioCE, 38000, Grenoble, France
$Univ. Grenoble Alpes, CNRS, ICMG FR 2607, Dep
́
́
́
S
* Supporting Information
ABSTRACT: NadA is a multifunctional enzyme that condenses dihydroxyacetone
phosphate (DHAP) with iminoaspartate (IA) to generate quinolinic acid (QA), the
universal precursor of the nicotinamide adenine dinucleotide (NAD(P)) cofactor.
Using X-ray crystallography, we have (i) characterized two of the reaction
intermediates of QA synthesis using a “pH-shift” approach and a slowly reacting
Thermotoga maritima NadA variant and (ii) observed the QA product, resulting from
the degradation of an intermediate analogue, bound close to the entrance of a long
tunnel leading to the solvent medium. We have also used molecular docking to
propose a condensation mechanism between DHAP and IA based on two previously
published Pyrococcus horikoshi NadA structures. The combination of reported data
and our new results provide a structure-based complete catalytic sequence of QA synthesis by NadA.
he synthesis of the essential nicotinamide adenine
Tdinucleotide (NAD(P)) cofactor by all organisms uses
quinolinic acid (QA) as its main precursor.1 The QA-derived
nicotinamide fragment of NAD(P) is the site of hydride
transfer of the cofactor, a fundamental process in many central
redox pathways such as the Krebs and Calvin cycles and
glycolysis.2 The first step in the de novo II pathway, used by
most prokaryotes to synthesize QA, is the production of
iminoaspartate (IA) from L-Asp.3 In Escherichia coli (Ec), IA is
produced by the L-Asp oxidase NadB, whereas in Thermotoga
maritima (Tm), this substrate is made by an aspartate
dehydrogenase called either NadB-II or NadX.4 IA can also
be produced by combining oxaloacetate with ammonium ions,
to be used by NadA in in vitro experiments.5 Although IA is
synthesized differently in E. coli and T. maritima, in both
microorganisms the remaining reactions leading to QA
synthesis are catalyzed by the same enzyme, the quinolinate
synthase NadA.4,6,7 The enzyme has three homologous
domains and appears to be extensively multifunctional (Scheme
1).8,9 Indeed, QA synthesis involves (i) the generation of the
nucleophilic C3 atom of IA;3,10 (ii) the attack of this
nucleophile to dihydroxyacetone phosphate (DHAP)a triose
phosphate sugar produced by the glycolytic pathwayresulting
in the condensed intermediate W; (iii) the keto-aldo isomer-
ization of W to form X; (iv) a first dehydration reaction upon
formation of a Schiff base and cyclization of the intermediate X
to generate Y; and (v) a second dehydration, reminiscent of the
reaction catalyzed by aconitase,11 to yield QA. Like aconitase,
NadA uses three cysteine ligands to bind a [4Fe-4S] cluster,
leaving an iron ion with a coordination site available for Lewis-
type chemistry.12 As shown by Mossbauer spectroscopy, this
̈
unique iron ion can interact with the thiolate groups of the QA
analogue and NadA inhibitor 4,5-dithiohydroxyphthalic acid.13
In 2014, we reported the first crystal structure of a [4Fe-4S]
cluster-containing NadA enzyme, the one from T. maritima.12
We showed then that the unique iron from the [4Fe-4S] cluster
is found at the end of a long tunnel that connects this metal ion
with the molecular surface.12 More recently, the structural
aspects of QA synthesis has been the subject of intense
scrutiny. Thus, crystal structures have been communicated by
Fenwick and Ealick for the wild type (wt) Pyrococcus horikoshii
(Ph)NadA complexes with three similar IA analogues and
DHAP14 and by us for the TmNadA-K219R/Y107F-W
intermediate and the TmNadA-K219R/Y107F phosphoglyco-
lohydroxamate (PGH) complex;15 PGH is a DHAP-like
molecule and a triose phosphate isomerase (TIM) inhibitor.
In addition, three very similar structures of cluster-bound QA in
NadA complexes have also been recently determined, one of
them by us.14−16 We will abbreviate “TmNadA-K219R” to
“TmNadA*” because it is functionally equivalent to the wt
enzyme.
Received: December 25, 2017
Accepted: April 11, 2018
Published: April 11, 2018
© XXXX American Chemical Society
A
DOI: 10.1021/acschembio.7b01104
ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology
Articles
a
Scheme 1. Mechanism of QA Synthesis by NadA
a
This mechanism is based on the X and Y complexes (this study) and previous propositions (see text). Atoms are numbered according to IUPAC
rules. W: 2-imino, 3-carboxy, 5-oxo, 6-hydroxy hexanoic acid, X: 2-(2-hydroxy-3-oxopropyl)-3-iminosuccinic acid or 2-amino-3-(2-hydroxy-3-
oxopropyl)maleic acid, Y: 5-hydroxy-4,5-dihydropyridine-2,3-dicarboxylic acid (these names correspond to the protonated states).
Table 1. X-ray Data and Refinement Statistics
crystal
1
2
3
4
TmNadA*-Y21F XY-
TmNadA* 5MPDC-derived QA-
TmNadA*-Y21F phthalate-
TmNadA*-Y21F PGH-
complex
complex
complex
complex
Data Collection
ID30A3
0.96770
P21
ESRF beamline
wavelength (Å)
space group
cell: a (Å)
ID30A3
0.96770
ID23−1
0.97242
ID30B
1.00000
C2
P21
P1
55.50
55.44
49.14
102.97
b (Å)
48.95
49.31
52.16
49.17
c (Å)
61.25
61.90
58.95
69.39
α (deg)
90
90
103.82
90
β (deg)
107.27
107.21
90.63
107.79
γ (deg)
90
90
90.37
90
resolution (Å)
(high-resolution shell)
measured reflections
unique reflections
redundancy
completeness (%)
Rmerge (%)
CC1/2
37.5−1.50
(1.55−1.50)
217368 (21491)
49728 (4844)
4.4 (4.4)
98.8 (99.3)
4.0 (78.3)
0.998 (0.719)
16.8 (1.7)
47.0−2.30
(2.38−2.30)
77646 (8006)
14259 (1412)
5.4 (5.7)
99.1 (99.7)
12.0 (157.4)
0.996 (0.375)
8.9 (1.0)
Refinement
47.0−2.30
13514
49.1−2.00
(2.07−2.00)
82582 (8437)
34665 (3465)
2.4 (2.4)
90.2 (91.9)
5.0 (24.1)
0.996 (0.898)
12.1 (3.2)
44.0−2.00
(2.07−2.00)
72297 (6484)
22406 (2193)
3.2 (3.0)
99.3 (99.1)
8.0 (79.6)
0.996 (0.469)
10.2 (1.4)
<I/σI>
resolution (Å)
used reflections
Rwork (%)
37.5−1.50
47225
13.7
43.4−2.00
32832
19.0
39.2−2.00
21244
19.2
20.2
reflections for Rfree
Rfree (%)
2499
720
1831
1156
17.9
22.1
23.4
24.1
number of atoms
average B-factor (Å2)
rms deviations
bond lengths (Å)
bond angles (deg)
2813
2468
5934
2662
27.2
55.1
31.3
37.0
0.011
1.5
0.011
1.4
0.011
1.4
0.011
1.5
Here we report the crystal structures of complexes of the
TmNadA*Y21F variant with (i) the substrate-derived products
X and Y (Scheme 1) at 1.5 Å resolution, (ii) PGH at 2.0 Å
resolution, (iii) the QA analogue phthalate at 2.0 Å resolution,
and (iv) of TmNadA* with a differently bound QA derived
from 5-mercaptopyridine-2,3-dicarboxylic acid (5MPDC) at 2.3
Å resolution (Table 1). We also discuss the structure of
TmNadA* with citrate bound in two conformations, which
combined with a series of docking experiments may explain an
unexpected result concerning our previously published W
complex.15 In addition, and based on the PhNadA-citraconate
complex,14 we have carried out another series of docking
experiments that generate a plausible model for NadA-bound
IA and DHAP substrates correctly poised to initiate the
condensation reaction that generates W. Taken together, our
results provide a structure-based sequence for all the catalytic
steps of QA synthesis by NadA.
RESULTS AND DISCUSSION
■
Trapping Experiments and the Orientation of the W
Intermediate. We already showed in our previous crystallo-
graphic analysis15 that a 20 mg mL−1 solution of the slowly
reacting TmNadA*Y21F variant can, after several days, produce
QA from DHAP and IA, the latter being obtained from
oxaloacetate and ammonium ions.5 Here an equivalent solution
was incubated 36 h in the glovebox at pH 7.5, a value that is
close to optimal for NadA activity.17 Next, the protein solution
was mixed with our standard crystallization solution to a final
B
DOI: 10.1021/acschembio.7b01104
ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology
Articles
Figure 1. X and Y reaction intermediates. (a) Stereo image of the electron density (Fobs − Fcalc) omit map calculated with diffraction data at 1.5 Å
resolution from a crystal grown from a TmNadA*Y21F protein solution first incubated with DHAP, oxaloacetate and ammonium sulfate for 36 h at
pH 7.5 and then set at pH 7.1 during the crystallization experiment (the “pH-shift” trapping experiment; see text). This electron density has been
modeled as a combination of the X and Y intermediates (40% each) and solvent (20%). The mutation of Tyr21 is indicated by a *. (b) Stereo image
of the X intermediate in a partial electron density (Fobs − Fcalc) omit map calculated after adding a 40% occupied Y intermediate and four 20%
occupied water molecules to the calculated phases and structure factors. c. Stereo pair of the Y intermediate in an equivalent (Fobs − Fcalc) omit map
obtained after adding instead 40% X and 20% occupied water to the model phases and structure factors. The modeled X, Y and water molecules (not
shown) fully explain the original map in a. All omit maps were contoured at 3.5σ.
pH of 7.1 where wt NadA displays very little activity17 (and its
TmNadA*Y21F variant should be even less efficient). An omit
(Fobs − Fcalc) electron density map calculated using X-ray data
from a crystal grown under these “pH-shifted” intermediate
trapping conditions17 and refined model phases displayed a
significant electron density peak at the active site (Figure 1a).
Examination of the shape and size of this peak suggested that it
corresponds to a mixture of different species. Indeed, by
adjusting their relative occupancies, it was possible to build the
X and Y intermediates (Scheme 1) into partial omit electron
density maps with a remarkably good match (Figure 1b,c and
Methods).
Our comparison of the previously obtained TmNa-
dA*Y107F-W intermediate15 with X (Figure 1b) indicates
that the former binds in an approximately 180°-rotated position
relative to the latter (Supplementary Figure 1). This likely
explains why catalysis stops at W in the TmNadA*Y107F
variant. However, in both cases, IA and DHAP must be
productively positioned for the condensation reaction to occur
because, as indicated by the comparison of the shape and
chirality of the imino forms of W and X (Supplementary Figure
1), the rotated stereochemically correct W intermediate is also
obtained with TmNadA*Y107F. Alternative positions for the W
condensation reaction product would be consistent with our
observation of citrate bound to TmNadA* in two conforma-
C
DOI: 10.1021/acschembio.7b01104
ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology
Articles
tions related by an approximate 140° rotation axis (Supple-
mentary Figure 2).15 We also note that in wt PhNadA, DHAP
binds in an orientation that is consistent with X14 and,
consequently, rotated relative to our previously reported W.15
Although we have not been able to crystallize DHAP complexes
with TmNadA*Y107F and TmNadA*Y21F to verify the
different relative orientations of the substrate molecule, we
observe this rotation effect in the crystal structures of the two
variants in complex with PGH (Supplementary Figure 3).
Taken together, these observations indicate that the NadA
active site can lodge the same molecule in different orientations
and that in the Y107F variant W binds in an unproductive
configuration. In order to try and elucidate the origin of the
Y107F phenotype, we have used PGH and DHAP to perform a
series of docking experiments and structural analyses in the
closed form of NadA (see Methods for details). The results
may be interpreted as follows: (i) in the wt PhNadA and
TmNadA*Y21F variant the −OH group of Tyr107 (Tyr109 in
PhNadA) forms an H-bond with the “proximal” carbonyl O
atom from the carboxylate group of Glu195 (Glu198 in
PhNadA). The other, “distal” O atom, is protonated and in the
PhNadA-DHAP structure it interacts with the terminal iron-
bound O of the substrate (Supplementary Figure 4a).14
Because in the TmNadA*Y107F variant this −OH group is
missing, docking results indicate that in this case it is the
proximal O that should be protonated (Supplementary Figure
4b and Methods). We conclude that in the Y107F variant, the
loss of the partial positive charge from the −OH group switches
the binding affinity of the carboxylate group of Glu195 from the
terminal O (Supplementary Figures 3a and 4a) to the carbonyl
O of the substrate (Supplementary Figure 4b). In addition, in
the TmNadA*Y107F-PGH structure (Supplementary Figure
3b), the rotated iron-bound terminal O also interacts with
Asn109, which requires the conformational change of Phe58
also observed in the closed TmNadA*Y107F-W (Supplemen-
tary Figure 1), TmNadA*-citrate (Supplementary Figure 2) and
PhNadA-citraconate and -maleate structures (PDB codes 5KTS
and 5KTR, respectively).14 Based on the rotated W structure,
we mistakenly postulated that Tyr21 was involved in the last
dehydration reaction.15 In the productive W model, Tyr21 still
has acid/base properties except that its postulated role is in the
first dehydration reaction with the deprotonation of the Schiff
base and the protonation of the O atom that will become the
leaving water molecule upon cyclization (Supplementary
Scheme 1).
The unproductive binding of the condensation product in
the TmNadA*Y107F-W structure induces a partially closed
form of NadA. In this structure, the −OH group of Ser36 forms
an H-bond with the N atom of W. This H-bond may help
keeping TmNadA*-Y107F in its partially closed form because
Ser36 is located at the N-terminal end of a helix that
significantly shifts positions between open and closed
structures.12,15 Conversely, the active TmNadA*Y21F-X/Y
structure is in its open form like in ligand-free TmNadA*.12
In this structure, the main chain N of Ser36 interacts with one
of the carboxylate groups in both ligands. Because the only
difference between W and X is the keto-aldo isomerization
(Scheme 1) it seems reasonable to conclude that the partially
closed form in the TmNadA*Y107F-W complex is also the
consequence of the mutation. To test this hypothesis, we have
also docked W in open TmNadA* and found that, indeed, it
preferentially adopts a conformation similar to the one
observed for X (Supplementary Figure 5).
New QA Binding Mode in NadA. Another surprising
result was obtained when, as part of our ongoing efforts to
characterize intermediate analogue binding to NadA, we solved
the structure of TmNadA* cocrystallized with 5MPDC. Indeed,
in this structure we could clearly see QA bound in a position
completely different from the one previously observed.14−16 In
addition, the amide fragment of Asn109 was missing and the
[4Fe-4S] cluster appeared to be partially disrupted (Figure 2a).
Figure 2. (a) NadA-QA structure from 5MPDC. Electron density
(Fobs − Fcalc) omit map in the crystal structure of the TmNadA*-
5MPDC-derived QA complex. The positive black mesh at the top of
the figure has been modeled as QA, whereas the one at the bottom,
and based on a coinciding peak in the anomalous scattering difference
map (shown as a pink mesh), is thought to correspond to a Fe ion
displaced from the [4Fe-4S] cluster. The negative red mesh in the
center of the figure indicates the missing amide group of Asn109 (see
text). The omit, the Fo-Fc around Asn109 and the Δanom maps were
contoured at 3.3σ, −3.3σ, and 4σ, respectively. (b) Electron density
(Fobs − Fcalc) omit map in the crystal structure of the TmNadA*Y21F-
phthalate complex. The QA analogue binds the enzyme as the reaction
product does. The omit map is contoured at 4σ.
A plausible explanation for the desulfurization reaction that
must have generated QA from 5MPDC would be the homolytic
cleavage of the 5MPDC’s S−C bond followed by H atom
abstraction from the amide group of Asn109 by the nascent QA
radical (Supplementary Scheme 2). The resulting radical amide
group will rearrange causing the cleavage of the Asn109 Cβ-Cγ
bond and the release of isocyanate. Finally, the generated
Ala109 radical will abstract an H atom from a component of the
solution. Conversely, the thiyl radical probably oxidizes the
[4Fe-4S] cluster causing its observed changes that, however,
cannot be unambiguously modeled. An example of the
homolytic cleavage of an aromatic C−S bond while bound to
a sulfur−metal cluster is given by Curtiss & Druker.18
The new QA position makes catalytic sense because the final
reaction product now points at the tunnel that leads to the
solvent medium12 and coincides with the equivalent fragment
in Y (Figures 1c and 2a). Although the desulfurization of
5MPDC and the dehydration of Y are not chemically related
reactions both involve the loss of a ring substituent to produce
D
DOI: 10.1021/acschembio.7b01104
ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology
Articles
Figure 3. Active site cavities in NadA. (a) Active site cavity I in structure A (60%) of citraconate-bound PhNadA (with ligand and structural water
molecules removed).14 (b) Active site cavities I and II in structure B (40%) of citraconate-bound PhNadA (with ligand and structural water
molecules removed). The probe radius used to contour the cavity maps is 0.9 Å.
QA. We have also detected the slow synthesis of QA from
5MPDC and TmNadA* in the crystallization buffer (see
Methods). For the first 25 days we observed no peak on the
HPLC chromatogram at 12.8 min, indicative of QA (dashed
line in Supplementary Figure 6a). Conversely, at day 35 there
was a small peak at 12.8 min (black line in Supplementary
Figure 6a), which UV−visible spectrum matched that of
authentic QA (Supplementary Figure 6b). After quantitation,
the QA concentration corresponded to 1/100 of the NadA
concentration and 1/330 of the initial 5MPDC equivalents.
The control experiment without NadA did not produce any QA
after 35 days (gray line in Supplementary Figure 6a and
Supporting Information). This result from the reaction between
NadA and 5MPDC in solution agrees well with our
observation, after five months, of QA in a crystal of TmNadA*
initially complexed with 5MPDC (Figure 2a). The functional
relevance of this new QA position is also supported by the
crystal structure of a TmNadA*Y21F-phthalate complex where
this QA analogue occupies the same position as the NadA
reaction product and establishes the same interactions with the
protein, with only one exception due to the missing N atom in
the phthalate ring (Figure 2b).
binding to the same site is sequential. We already have
discussed possibilities (ii) and (iii).15 A new element to be
considered in this respect is the observation by Fenwick and
Ealick that in the PhNadA citraconate and maleate complexes
about 40% of the protein molecules in the crystal (structure B)
have alternative positions for residues Tyr109, Asn111 and
Phe60 (Tyr107, Asn109 and Phe58 in Tm).14 The concerted
movement of these three residues generates a cavity (cavity II
in Figure 3b) that in the reported structures is occupied by
three water molecules. The position and shape of that cavity,
next to the DHAP binding site (cavity I in Figure 3), and its
rather unusual generation by the concerted movement of three
buried residues suggests that it could serve to bind IA
productively. Indeed, after removal of the three waters, it was
possible to dock IA with its reactive C3 atom3,10 oriented
toward, and not far from, the C1 atom of DHAP (Figure 4 and
Supplementary Figure 7).
It is now possible to postulate an initial sequence where IA,
synthesized by NadB (or NadX), binds to the site where
citraconate and maleate bind (Supplementary Figure 8),
possibly facilitating and/or stabilizing the concerted conforma-
tional shift of Tyr107, Asn109, and Phe58, which then
The previously observed QA position15 bound as a bidentate
ligand to the [4Fe-4S] cluster in a closed form of NadA was
puzzling because it provided no clues for the release of the
reaction product from the enzyme. Based on our new result, we
believe that that binding does not belong in the catalytic
sequence; instead it is an in vitro artifact caused by either an
excess of added QA or crystal packing effects.14−16
Condensation Reaction of DHAP with IA. Next, we
turned our attention to the first reaction, which involves the
condensation of DHAP with IA (Scheme 1). According to
previous work, the phosphate group of DHAP (and PGH15)
and the carboxylate groups of the IA analogues malate
citraconate, itaconate, and maleate bind to the same site in
NadA.14 Furthermore, X, Y, and the QA reported here, as well
as W,15 also bind to that site with their two carboxylate groups.
This precludes the simultaneous binding of DHAP and IA there
and suggests that either (i) at least one of these two binding
modes is artifactual, (ii) the condensation reaction does not
take place at the identified active site of NadA, or (iii) their
Figure 4. Structure B of PhNadA with docked DHAP and IA in
cavities I and II (Figure 3b), respectively (see Methods for details). In
this configuration, the distance between the atoms C1 (DHAP) and
C3 (IA) is 4.4 Å.
E
DOI: 10.1021/acschembio.7b01104
ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology
Articles
galactopyranoside (Eurogentec) and cultures were shaken for 4 h.
Purification of proteins was performed using an affinity Ni-NTA
column. After extensive washing with 100 mM Tris-HCl, 500 mM
NaCl, 10 mM imidazole pH 8 buffer, TmNadA* , and TmNadA*Y21F
were eluted with the above buffer containing 0.2 M imidazole.
Imidazole was removed using a desalting PD-25 column (Pharmacia).
Chemical reconstitution was performed anaerobically as described
in ref 15. Briefly, purified TmNadA* proteins (150 μM) were
incubated with 5 mM DTT for 15 min. Then, 6 mol equiv of L-
cysteine with catalytic amount of IscS cysteine desulfurase (1.5 μM)
and 6 mol equiv ferrous iron ((NH4)2Fe(SO4)2) were added. After a 5
h incubation, proteins were heated for 30 min at 65 °C in the presence
of 50 mM DTT. Monomeric [4Fe-4S]-TmNadA* proteins were
obtained after Superdex-75 chromatography and concentrated to 20
mg mL−1.
generates cavity II. In the next step, IA is displaced to cavity II
by the arrival of DHAP that occupies cavity I with higher
affinity (due to its additional interaction with the [4Fe-4S]
cluster relative to IA binding) in a productive position that
differs from the one observed in the PhNadA-DHAP complex
(Figure 4 and Supplementary Figure 7);14 both substrates are
now well poised to form a pentavalent transition state. The
condensation reaction ends when the C1−O bond of DHAP is
broken, being replaced by a C1−C3 bond thus forming W
while concomitantly phosphate leaves the active site through
the long protein channel.12 During this process the nascent W
intermediate rearranges occupying either the productive
orientation equivalent to the one observed for X in its complex
with TmNadA*Y21Fand mimicked by DHAP in its complex
with PhNadA14or the unproductive one found in the
TmNadA*Y107F-W complex, which is imitated by our
TmNadA*Y107F-PGH complex.15 Based on the docking
experiments reported above, we believe that these orientations
depend, at least partially, on which of the carboxylate O atoms
of Glu195 (Glu198 in PhNadA) is protonated. This
protonation should differ in the wt enzyme and Y107F variant.
We have tried to cocrystallize different combinations of
citraconate, IA, DHAP, and PGH with TmNadA, in order to
get a structure where the two cavities are occupied, but have
not yet succeeded. In the future, we will consider the possibility
of synthesizing a transition-state analogue that could occupy
both cavities and will prepare new site-directed variants to test
this hypothesis. For the time being, docking experiments
performed to explore its validity indicate that the proposed
reaction is clearly plausible (Figure 4).
5MPDC Synthesis. 1. 5-Chloropyridine-2,3-dicarboxylic Acid
Dimethyl Ester. 5-Chloropyridine-2,3-dicarboxylic acid (1.67 g) was
synthesized using a reported procedure19 starting from 3-chloroquino-
line (3.50 g, 21.4 mmol). The crude product was engaged in the next
step without further purification. To a solution of the obtained 5-
chloropyridine-2,3-dicarboxylic acid in methanol (7 mL) at 0 °C,
H2SO4 (0.47 mL, 8.81 mmol) was added dropwise within 5 min. The
resulting mixture was refluxed for 24 h until completion (monitored by
TLC (2% methanol in CH2Cl2)). After concentration in vacuo, the
resulting residue was basified with saturated aqueous sodium
bicarbonate to pH ≈ 8, and the product was extracted three times
with ethyl acetate. The combined organic fractions were dried with
sodium sulfate and concentrated under reduced pressure. The
resulting residue was purified by silica gel column chromatography
(2% methanol in CH2Cl2) to afford 1.16g (24% over two steps) of the
1
desired product as a white powder. H NMR (300 MHz, CDCl3), δ
CONCLUSIONS
■
8.73 (d, J = 2.4 Hz, 1H), 8.15 (d, J = 2.4 Hz, 1H), 4.02 (s, 3H), 3.97
(s, 3H). 13C NMR (75 MHz, CDCl3), δ 165.7 (C), 164.7 (C), 150.6
(CH), 148.1 (C), 137.01 (CH), 133.8 (C), 127.9 (C), 53.2 (2 CH3).
MS-ESI: calculated m/z for [C9H8ClNO4+H+]+, 230.0; found, 230.2.
2. 5-((4-Methoxybenzyl)thio)pyridine-2,3-dicarboxylic Acid Di-
methyl Ester. A solution of dimethyl-5-chloropyridine-2,3-dicarbox-
We first note that although in our previously published
structure W was not bound in a productive orientation, it was
the correct intermediate species. More importantly, here we
present a series of significant results that allow us to describe
fully the structural aspects of QA synthesis by NadA. A rather
convincing model for the condensation of IA and DHAP is
based on cavity II, which is generated by the concerted shift of
residues Tyr109, Asn111, and Phe60 in 40% in the PhNadA
citraconate and maleate complex structures.14 The presence of
this cavity explains why both substrates can sequentially bind to
the same site at the top of cavity I. Indeed, this site may
represent an entry point for the unstable IA before it is
displaced to cavity II by DHAP. It also provides a plausible
model for the condensation reaction because the phosphate
group of DHAP is properly positioned to leave through the
protein tunnel. Next in the catalytic sequence, our trapping
experiment yielded the key structures of X and Y. In addition,
we were able to solve the structure of an intermediate analogue-
derived QA bound to TmNadA* in an orientation that favors
its release from the active site and is likely to represent the
physiologically relevant one. This is also supported by the
TmNadA*Y21F with phthalate. Taken together, our results
provide a complete structure-based description of QA synthesis
by NadA.
ylate (1.15 g, 5.01 mmol) in anhydrous dimethyl sulfoxide (DMSO; 25
mL) was deoxygenated by argon bubbling for 10 min. Then, Cs2CO3
(1.80 g, 5.51 mmol) and 4-methoxy-α-toluenethiol (0.73 mL, 5.26
mmol) were added, and the resulting solution was stirred at RT under
an argon atmosphere for 48 h (monitored by TLC (2% methanol in
CH2Cl2)). The reaction mixture was quenched by addition of water
(25 mL) and extracted three times with diethyl ether (Et2O). The
organic layers were combined, dried across magnesium sulfate, and the
solvent evaporated. The obtained residue was purified by silica gel
column chromatography (5% methanol in CH2Cl2) to afford 1.06 g
(57%) of the desired product as a pale yellow powder. 1H NMR (300
MHz, CDCl3), δ 8.55 (s, 1H), 7.89 (s, 1H), 7.23 (d, J = 8.4 Hz, 2H),
6.84 (d, J = 8.4 Hz, 2H), 4.16 (s, 2H), 3.97 (s, 3H), 3.92 (s, 3H), 3.79
(s, 3H). 13C NMR (75 MHz, CDCl3), δ 165.9 (C), 163.0 (C), 159.3
(C), 150.3 (CH), 146.2 (C), 137.4 (C), 136.3 (CH), 130.0 (2 CH),
127.5 (C), 127.3 (C), 114.3 (2 CH), 55.3 (CH3), 53.1 (CH3), 53.0
(CH3), 37.3 (CH2). ESI-MS: calculated m/z for [C17H17NO5S+H+]+,
348.1; found, 348.1.
METHODS
■
Purification and Chemical Reconstitution of NadA Proteins.
Expression and purification of TmNadA* and the TmNadA*Y21F
variant were performed as in ref 15. Briefly, BL21DE3 E. coli cells were
grown aerobically at 37 °C in LB medium containing 100 μg/mL
amplicillin. Expression was induced with 0.5 mM Isopropyl 1-thio-β-d-
3. Dimethyl 5-((4-Methoxybenzyl)thio)pyridine-2,3-dicarboxylic
Acid. To a solution of 5-((4-methoxybenzyl)thio)pyridine-2,3-
dicarboxylic acid dimethyl ester (0.90 g, 2.59 mmol) in diethyl ether
(4.5 mL), a 5 M aqueous solution of NaOH (3.2 mL) was added. The
mixture was refluxed for 24 h and then acidified to pH ≈ 2 by
F
DOI: 10.1021/acschembio.7b01104
ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology
Articles
TmNadA*Y21F-X/Y Complex. Initially, 50 μL of a 20 mg mL−1
protein solution in 50 mM KCl, 125 mM Tris HCl pH 7.5 were
incubated anaerobically 36 h at 20 °C after the addition of 0.5 μL of
0.5 M (NH4)2SO4, 1 μL of 0.5 M oxaloacetic acid, and 2 μL of 50 mM
DHAP. One microliter of this solution was then mixed with the same
volume of a 21% PEG3350, 200 mM NaF 100 mM Bis Tris propane
pH 7.1 crystallization solution and set up as indicated in the previous
paragraph. Several brown crystals appeared in the drop after 9 days.
One of these crystals was transferred to a cryo-protecting solution
composed of 28% PEG3350, 200 mM NaF, 20% glycerol in 100 mM
Bis Tris propane buffer pH 7.1 and flash-cooled using liquid ethane
inside the glovebox.
dropwise addition of a concentrated HCl solution. The resulting
mixture was extracted three times with ethyl acetate, and the combined
organic layers were dried with Na2SO4 and concentrated in vacuo.
Reextraction of the product from the resulting residue with acetone,
drying with sodium sulfate, and evaporation of the solvent under
reduce pressure afforded 0.55g (62%) of the title compound as a
yellow powder. 1H NMR (300 MHz, acetone d-6), δ 8.62 (s, 1H), 8.13
(s, 1H), 7.36 (d, J = 8,3 Hz, 2H), 6.89 (d, J = 8,3 Hz, 2H), 4.40 (s,
2H), 3.77 (s, 3H). 13C NMR (75 MHz, acetone-d6), δ 165.6 (C),
165.5 (C), 159.3 (C), 149.2 (C), 146.3 (C), 137.5 (CH), 136.4 (C),
130.2 (2 CH), 128.0 (C), 127.8 (CH), 114.0 (2 CH), 54.6 (CH3),
36.1 (CH2). ESI-MS: calculated m/z for [C15H13NO5S−H+]−, 318.1;
found, 317.9.
TmNadA*-5MPDC Complex. First, 50 μL of a 44.6 mg mL−1
protein solution in 50 mM KCl, 100 mM Tris HCl pH 8.0 were
incubated anaerobically overnight at 20 °C after the addition of 27 μL
7.5 mM 5MPDC in 25 mM Tris HCl pH 8.0 and 23 μL of a solution
containing 50 mM KCl and 125 mM Tris HCl pH 7.2. One μL of the
resulting solution, containing 22.3 mg mL−1 protein, was initially
mixed with 1 μL of a solution containing 22% PEG3350, 200 mM
Na2HPO4 in 100 mM CHES pH 9.1. Prior to mixing the latter
solution was left to sit in order to deposit some precipitated Na2HPO4.
After 37 days, the crystallization drop was transferred to a well
containing 1 mL of a solution that before phosphate precipitation
contained 34% PEG3350, 200 mM Na2HPO4 in 100 mM CHES pH
9.1. Twelve days later, the drop was again transferred to a well
containing 1 mL of a solution composed of 36% PEG3350, 200 mM
Na2HPO4 in 100 mM CHES pH 9.1 before phosphate precipitation.
Several brown crystals were observed in the transferred drop 105 days
later. A fragment of one of these crystals was transferred to a cryo-
protecting solution composed of 40% PEG3350, 200 mM Na2HPO4 in
100 mM CHES pH 9.5 and flash-cooled using liquid propane inside
the glovebox.
4. 5-Mercaptopyridine-2,3-dicarboxylic Acid. To a mixture of 5-
((4 methoxybenzyl)thio)pyridine-2,3-dicarboxylic acid (0.472 g, 1.48
TmNadA*Y21F-Phthalate Complex. In the initial step, 100 μL of a
20 mg mL−1 protein solution in 50 mM KCl, 125 mM Tris HCl pH
7.2 were incubated anaerobically overnight with 2,4 μL 50 mM
Phtalate in 50 mM Hepes pH 7.0 at 20 °C. One microliter of this
solution was mixed with the same volume of a 26% w/v PEG3350, 200
mM Na2HPO4, 100 mM Hepes pH 7.8 crystallization solution on a
siliconized coverslip. The resulting drop was then equilibrated in the
vapor phase against 1 mL of the crystallization solution at 20 °C. A
brown crystalline plate was obtained after 3 days. The crystal was
subsequently transferred to a cryo-protecting solution composed of
40% PEG3350, 200 mM Na2HPO4 in 100 mM Hepes pH 7.8 buffer,
and flash-cooled using liquid propane inside the glovebox as previously
described.
mmol) and anisole (9.5 mL, 87.27 mmol) and cooled to −5 °C,
trifluoroacetic acid (95 mL) and mercury(II) acetate (0.49 g, 1.53
mmol) were added.20 The reaction mixture was stirred for 1 h at −5
°C and then concentrated in vacuo to afford an orange oil. Addition of
a 1:1 mixture of water-Et2O (150 mL) resulted in the precipitation of a
part of the product. The precipitated solid and the water phase were
mixed together, and the resulting suspension was concentrated in
vacuo and taken up in methanol (40 mL). Next, the methanolic
solution was bubbled with hydrogen sulfide for 2 h. The black
suspension was filtered through Celite with methanol, and the
obtained filtrate was concentrated in vacuo and purified by C18 column
chromatography to afford 0.053 g (9%) of the tittle compound as a
1
disulfide dimer. H NMR (300 MHz, acetone-d6), δ 8.48 (d, J = 2.4
Hz, 1H), 8.94 (d, J = 2.4 Hz,1H). 1H NMR (300 MHz, 1 M Tris-D2O,
pH = 7.5) δ 8.55 (s, 1H), 8.16 (s, 1H) . 13C NMR (75 MHz, acetone-
d6), δ 165.9 (C), 165.0 (C), 150.2 (C), 149.8 (CH), 136.8 (CH),
135.0 (C), 127.2 (C). ESI-MS: calculated m/z for [C14H8N2O8S2−
H+]−, 395.0; found, 395.0 (88); calculated m/z for [C14H8N2O8S2-
2H+]2−, 197.2; found, 197.2 (100).
X-ray Structure Determination. All diffraction data (Table 1)
were collected on a Pilatus detector, using the program MXCuBe22 at
beamlines ID30A3, ID30B, and ID23-1 of the European Synchrotron
Radiation Facility in Grenoble, France. Data indexation, integration
and initial scaling was performed with the XDS package,23 followed by
a final scaling and merging step with AIMLESS24 of the CCP4
package.25 Structures were solved by molecular replacement with the
PHASER program26 using previously published TmNadA struc-
tures.12,15 The obtained models were refined with REFMAC5.27 The
resolution of the Y21F-XY model was good enough to refine individual
anisotropic B-factors. For the three other models, isotropic B-factors
were refined along with TLS parameters.27
Because of the fast oxidation of the title compound when exposed
to air, it was stored as a dimer and reduced in the glovebox just before
further use with 1 equiv of DTT as reducing agent. This procedure
permitted to obtain the desired monomeric form in a quantitative
1
manner. H NMR (300 MHz, 1 M Tris-D2O, pH = 7.5), δ 8.29 (s,
1H), 7.73 (s, 1H). 13C NMR (75 MHz,), δ. ESI-MS: calculated m/z
Local manual model corrections were performed with COOT.28
Except for MXCuBe, the programs mentioned above were compiled by
SBGrid.29
for [C7H4NO4S−H+]−, 198.0; found, 198.2 (100).
Crystallization Conditions. TmNadA*Y21F-PGH Complex. First,
100 μL of a 20 mg mL−1 protein solution in 50 mM KCl, 10 mM Tris
HCl pH 8.0 were incubated anaerobically overnight with 7 μL 20 mM
PGH in water at 20 °C. One microliter of this solution was mixed with
the same volume of a 24% w/v PEG3350, 200 mM Na2HPO4, 100
mM Tris base pH 8.5 crystallization solution on a siliconized coverslip.
The resulting drop was then equilibrated in the vapor phase against 1
mL of the crystallization solution at 20 °C. A brown crystalline plate
was obtained after 3 days. The crystal was subsequently transferred to
a cryo-protecting solution composed of 40% PEG3350, 200 mM
Na2HPO4 in 100 mM Tris base pH 8.4 buffer, and it was flash-cooled
using liquid propane inside the glovebox as previously described.21
A series of 1.5 Å resolution (Fobs − Fcalc) omit maps were used to
resolve the mixture of reaction intermediates observed in crystal 1,
which resulted from the pH shift experiment described in the main
text. The first omit map (Figure 1a) was calculated after removing all
atoms inside cavity I (Figure 3a) from the protein model. This map
showed a continuous electron density leading to the unique Fe ion of
the [4Fe-4S] cluster that could be assigned to a partially occupied
intermediate Y. After including this in the atomic model with an
occupation of 40%, the next omit map revealed clear electron density
for the X intermediate, which was also included at 40% occupation.
G
DOI: 10.1021/acschembio.7b01104
ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology
Articles
Different relative occupations of X and Y resulted in worse refinement
statistics. Remaining features in the following (Fobs − Fcalc) map could
be satisfactorily explained by four water molecules at H-bonding
distances from each other, with an occupation of 20% each. Including
these waters with 40% X and 0% Y or 40% Y and 0% X improved the
quality of the respective omit maps for Y and X, allowing to obtain
well-defined structures for their bound forms (Figure 1b,c). Dictionary
files needed for the refinement of the X and Y intermediates were
developed in the same way as previously described for intermediate
W.15 Because their structures are very similar, replacing 40% of X by
40% of properly rotated W in the protein model gave the same
refinement statistics, with a final Rfree of 17.9% (Table 1). The
assignment to X is based on the observed binding mode to the cluster
(an alkoxide oxygen ligand at C5 should be preferred over a carbonyl
one, see also Scheme 1), combined with the presence of Y (the next
intermediate in the catalytic mechanism) and our previous finding that
the Y21F variant is able to produce QA.15
Unexpectedly, the crystal structure of the complex of TmNadA*
with 5MPDC revealed that the QA product can be obtained from the
latter molecule. Refinement of the structure revealed the cleavage of
the Cβ-Cγ bond of Asn109 (Figure 2a). In addition, a large electron
density peak next to the [4Fe-4S] cluster coinciding with one of the
highest peaks in the anomalous difference map (Figure 2a) indicated a
significant movement of the unique Fe ion of the cluster (Fe3 in the
PDB file). Because of the limited 2.3 Å resolution of the diffraction
data and possible disorder, we were not able to resolve the new cluster
structure, which could also be part of a mixture of different forms. The
observed structural changes are compatible with the formation of
reactive radical species during the production of QA from 5MPDC
(Supplementary Scheme 2). Examination of the data obtained for the
phthalate complex indicated that the crystal contained a mixture of
open and closed NadA conformations, which were modeled starting
from previous PDB depositions 4P3X12 and 5F33,15 respectively.
Their occupations were manually refined to 59% and 41%,
respectively, which gave the lowest Rfree factor. The phthalate molecule
binds to the open form (Figure 2b) in almost the same way as the QA
product obtained from 5MPDC. Finally, the 2.0 Å resolution omit
electron density of PGH in complex with TmNadA*-Y21F
(Supplementary Figure 3a) could be explained by a mixture of 70%
PGH and 30% phosphate from the crystallization mixture, the position
of the latter coinciding with the phosphate moiety of PGH. The four
crystal structures are deposited in the Protein Data Bank (PDB). Their
refinement statistics are given in Table 1.
Figure 4a) and PGH-containing TmNadA*-Y107F15 (PDB code 5F33,
Supplementary Figure 4b) correspond to the closed form of the
protein. The ligand-free, open form of TmNadA* was also used12
(PDB code 4P3X and Supplementary Figure 5a). Hydrogen atoms
were added to the models and protonation states were assigned using
the standard protein preparation protocol within the Schrodinger
̈
suite. The positions of the built hydrogen atoms were then optimized,
and water molecules present in the X-ray model were removed. A
special attention was payed to the protonation state of Glu195
(Glu198 in PhNadA) that was set to be either deprotonated or
protonated on the distal or proximal oxygen atom relative to PhNadA
Tyr109 (Figure 4). A receptor docking grid was defined by the ligand
position for the three closed forms of NadA and by the water
molecules filling the cavity in the case of the open form; ligands and
water molecules were removed. The DHAP, PGH, and W molecules
were prepared using the Ligprep program and then docked using the
Glide program with the standard (SP) and the extra precision (XP)
docking procedures. The docked poses were then compared to the
available X-ray models. Next, IA was first docked in the B structure of
the citraconate-containing PhNadA in a grid centered at the
citraconate position (without the ligand); in these conditions IA
binds as citraconate except that the amino group of the former is
opposite to the methyl group of the latter (Supplementary Figure 8).
To try and investigate the first condensation reaction between IA and
DHAP (Scheme 1), we performed a multiple ligand docking
procedure using Glide.34 IA was docked in a grid defined by the
three water molecules present in cavity II (Figure 3b) while keeping
citraconate present in cavity I. Analyzing the IA docked poses, we
found only one family where the C3 atom involved in the
condensation reaction (see main text) points toward cavity I (Figure
4). We then used this IA-containing model to generate a grid defined
by the citraconate position and we docked DHAP. The DHAP docked
pose shown in Figure 4 is the most suitable one we found for
condensation, that is, with the C1 atom from DHAP and the C3 atom
from IA at a distance of 4.4 Å. Further rearrangements, such as
substrate displacements and the change from a closed to an open form
of NadA will be necessary for the reaction to proceed (see main text).
Figure 4 and Supplementary Figures 4, 5, and 8 were prepared with
Maestro from Schrodinger.33
̈
ASSOCIATED CONTENT
* Supporting Information
■
S
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acschem-
bio.7b01104.
Figures 1 and 2 and Supplementary Figure 3a were prepared with
BOBSCRIPT,30 and Supplementary Figures 1, 2, and 3b with
MOLSCRIPT.31 All these figures were rendered with Raster3D.32
Figure 3 and Supplementary Figure 7 were prepared with COOT,28
using the CAVENV program of the CCP4 package25 to calculate the
protein cavities.
Additional X-ray structure superpositions, docking
experiments, solution experiment, and schemes for the
role of Y21 and QA formation (PDF)
QA Production by 5MPDC and NadA in Solution. TmNadA*
(430 μM) was mixed with 3.5 mol equiv of 5MPDC in 125 mM Tris,
50 mM KCl pH 8.0 buffer. After an overnight incubation, 100 mM
CHES pH 9.5 crystallization buffer was added to a 1:1 ratio (v/v) to
the TmNadA*-5MPDC solution. Formation of QA was monitored by
HPLC by withdrawing 20 μL from this reaction mixture daily for the
first 2 days and every 5 days until the 35th day. At that time, 2 μL of 2
M H2SO4 was added to the withdrawn 20 μL in order to stop the
reaction. Then, samples were diluted by the addition of 80 μL of
0.03% trifluoroacetic acid, pH 2.4 buffer and centrifuged at 13000 rpm
for 15 min to remove all insoluble materials. The supernatant (80 μL)
was then injected to a HPLC system (Agilent 1100, column HPLC
Kinetex reversed-phase 5u C18 100A, 250 × 4.6 mm). QA formation
was detected and quantified as previously described.13 A control
experiment with 5MPDC but without TmNadA*, was carried out in
parallel and a sample from this solution was analyzed by HPLC at day
35 (Supplementary Figure 6a).
AUTHOR INFORMATION
Corresponding Author
*E-mail: juan.fontecilla@ibs.fr.
ORCID
Juan C. Fontecilla-Camps: 0000-0002-3901-1378
Notes
The authors declare no competing financial interest.
The atomic coordinates and structure factors have been
deposited to the PDB (ID code 6F48, 6F4D, 6F4L, and 6G74)
■
ACKNOWLEDGMENTS
■
We thank the CEA and the CNRS for institutional support. We
also thank the Agence Nationale pour la Recherche for contract
ANR-16-CE18-0026 (NADIN). S.O.d.C. also acknowledges
partial financial support from the ARCANE Labex (ANR-11-
LABX-0003-01) and GRAL Labex (ANR-10-LABEX-04 GRAL
Molecular Docking. All calculations were performed with the
Schrodinger suite.33 The used X-ray strutures from citraconate-
̈
containing PhNadA14 (B structure from PDB code 5KTS and Figure
4), DHAP-containing PhNadA14 (PDB code 5KTN, Supplementary
H
DOI: 10.1021/acschembio.7b01104
ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology
Articles
Structure of quinolinate synthase from Pyrococcus horikoshii in the
presence of its product, quinolinic acid. J. Am. Chem. Soc. 138, 7224−
7227.
(17) Suzuki, N., Carlson, J., Griffith, G., and Gholson, R. K. (1973)
Studies on the de novo biosynthesis of NAD in Escherichia coli. V.
Properties of the quinolinic acid synthetase system. Biochim. Biophys.
Acta, Gen. Subj. 304, 309−315.
(18) Curtis, M. D., and Druker, S. H. (1997) Homolytic C−S bond
scission in the desulfurization of aromatic and aliphatic thiols mediated
by a Mo/Co/S cluster: mechanistic aspects relevant to HDS catalysis.
J. Am. Chem. Soc. 119, 1027−1036.
Labex, Grenoble Alliance for Integrated Structural Cell
Biology). Part of this work used the platforms of the Grenoble
Instruct center (ISBG; UMS 3518 CNRS-CEA-UJF-EMBL)
with support from FRISBI (ANR-10-INSB-05-02) and GRAL
(ANR-10-LABX-49-01) within the Grenoble Partnership for
Structural Biology (PSB). We appreciate the help from the staff
of the ESRF beamlines (Grenoble) with X-ray data collection
and of the computing facility provided by the Commissariat a
l’Energie Atomique et aux Energies Alternatives (CEA/DRF/
GIPSI), Saclay.
̀
(19) Obafemi, C. A., and Pfleiderer, W. (1994) Permanganate
oxidation of quinoxaline and its derivatives. Helv. Chim. Acta 77,
1549−1556.
REFERENCES
■
(1) Foster, J. W., and Moat, A. G. (1980) Nicotinamide adenine
dinucleotide biosynthesis and pyridine nucleotide cycle metabolism in
microbial systems. Microbiol. Rev. 44, 83−105.
(20) Holler, T. P., Ruan, F., Spaltenstein, A., and Hopkins, P. B.
(1989) Total synthesis of marine mercaptohistidines: ovothiols A, B,
and C. J. Org. Chem. 54, 4570−4575.
(2) Pollak, N., Dolle, C., and Ziegler, M. (2007) The power to
̈
̀
(21) Vernede, X., and Fontecilla-Camps, J. C. (1999) A method to
reduce: pyridine nucleotides − small molecules with a multitude of
stabilize reduced and or gas-treated protein crystals by flash-cooling
under a controlled atmosphere. J. Appl. Crystallogr. 32, 505−509.
(22) Gabadinho, J., Beteva, A., Guijarro, M., Rey-Bakaikoa, V.,
Spruce, D., Bowler, M. W., Brockhauser, S., Flot, D., Gordon, E. J.,
Hall, D. R., Lavault, B., McCarthy, A. A., McCarthy, J., Mitchell, E.,
Monaco, S., Mueller-Dieckmann, C., Nurizzo, D., Ravelli, R. B. G.,
Thibault, X., Walsh, M. A., Leonard, G. A., and McSweeney, S. M.
(2010) MxCuBE: a synchrotron beamline control environment
customized for macromolecular crystallography experiments. J.
Synchrotron Radiat. 17, 700−707.
functions. Biochem. J. 402, 205−218.
(3) Nasu, S., Wicks, F. D., and Gholson, R. K. (1982) L-Aspartate
oxidase, a newly discovered enzyme of Escherichia coli, is the B protein
of quinolinate synthetase. J. Biol. Chem. 257, 626−632.
(4) Yang, Z., Savchenko, A., Yakunin, A., Zhang, R., Edwards, A.,
Arrowsmith, C., and Tong, L. (2003) Aspartate dehydrogenase, a
novel enzyme identified from structural and functional studies of
TM1643. J. Biol. Chem. 278, 8804−8808.
(5) Nasu, S., and Gholson, R. K. (1981) Replacement of the B
protein requirement of the E. coli quinolinate synthetase system by
chemically-generated iminoaspartate. Biochem. Biophys. Res. Commun.
101, 533−539.
(6) Nasu, S., Wicks, F. D., Sakakibara, S., and Gholson, R. K. (1978)
Synthesis of quinolinate from D-aspartate in the mammalian liver-
Escherichia coli quinolinate synthetase system. Biochem. Biophys. Res.
Commun. 84, 928−935.
(7) Begley, T. P., Kinsland, C., Mehl, R. A., Osterman, A., and
Dorrestein, P. (2001) The biosynthesis of nicotinamide adenine
dinucleotides in bacteria. Vitam. Horm. 61, 103−119.
(8) Ollagnier-de Choudens, S., Loiseau, L., Sanakis, Y., Barras, F., and
Fontecave, M. (2005) Quinolinate synthetase, an iron-sulfur enzyme
in NAD biosynthesis. FEBS Lett. 579, 3737−3743.
(9) Cicchillo, R. M., Tu, L., Stromberg, J. A., Hoffart, L. M., Krebs,
C., and Booker, S. J. (2005) Escherichia coli quinolinate synthetase
does indeed harbor a [4Fe-4S] cluster. J. Am. Chem. Soc. 127, 7310−
7311.
(23) Kabsch, W. (2010) XDS. Acta Crystallogr., Sect. D: Biol.
Crystallogr. 66, 125−132.
(24) Evans, P. R., and Murshudov, G. N. (2013) How good are my
data and what is the resolution? Acta Crystallogr., Sect. D: Biol.
Crystallogr. 69, 1204−1214.
(25) Winn, M. D., Ballard, C. C., Cowtan, K. D., Dodson, E. J.,
Emsley, P., Evans, P. R., Keegan, R. M., Krissinel, E. B., Leslie, A. G.
W., McCoy, A., McNicholas, S. J., Murshudov, G. N., Pannu, N. S.,
Potterton, E. A., Powell, H. R., Read, R. J., Vagin, A., and Wilson, K. S.
(2011) Overview of the CCP4 suite and current developments. Acta
Crystallogr., Sect. D: Biol. Crystallogr. 67, 235−242.
(26) McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M.
D., Storoni, L. C., and Read, R. J. (2007) Phaser crystallographic
software. J. Appl. Crystallogr. 40, 658−674.
́
(27) Murshudov, G. N., Skubak, P., Lebedev, A. A., Pannu, N. S.,
Steiner, R. A., Nicholls, R. A., Winn, M. D., Long, F., and Vagin, A. A.
(2011) REFMAC5 for the refinement of macromolecular crystal
structures. Acta Crystallogr., Sect. D: Biol. Crystallogr. 67, 355−367.
(28) Emsley, P., Lohkamp, B., Scott, W. G., and Cowtan, K. (2010)
Features and development of Coot. Acta Crystallogr., Sect. D: Biol.
Crystallogr. 66, 486−501.
(10) Flint, D. H., and Allen, R. M. (1996) Iron-sulfur proteins with
nonredox functions. Chem. Rev. 96, 2315−2334.
(11) Beinert, H., Kennedy, M. C., and Stout, C. D. (1996) Aconitase
as iron-sulfur protein, enzyme, and iron-regulatory protein. Chem. Rev.
96, 2335−2374.
(29) Morin, A., Eisenbraun, B., Key, J., Sanschagrin, P. C., Timony,
M. A., Ottaviano, M., and Sliz, P. (2013) Collaboration gets the most
out of software. eLife 2, e01456.
(30) Esnouf, R. M. (1999) Further additions to MolScript version
1.4, including reading and contouring of electron-density maps. Acta
Crystallogr., Sect. D: Biol. Crystallogr. 55, 938−940.
(12) Cherrier, M. V., Chan, A., Darnault, C., Reichmann, D., Amara,
P., Ollagnier de Choudens, S., and Fontecilla-Camps, J. C. (2014) The
crystal structure of Fe4S4 quinolinate synthase unravels an enzymatic
dehydration mechanism that uses tyrosine and a hydrolase-type triad.
J. Am. Chem. Soc. 136, 5253−5256.
(13) Chan, A., Clemancey, M., Mouesca, J.-M., Amara, P., Hamelin,
O., Latour, J.-M., and Ollagnier de Choudens, S. (2012) Studies of
inhibitor binding to the [4Fe-4S] cluster of quinolinate synthase.
Angew. Chem., Int. Ed. 51, 7711−7714.
(14) Fenwick, M. K., and Ealick, S. E. (2016) Crystal structures of the
iron-sulfur cluster-dependent quinolinate synthase in complex with
dihydroxyacetone phosphate, iminoaspartate analogues, and quinoli-
nate. Biochemistry 55, 4135−4139.
(31) Kraulis, P. J. (1991) MOLSCRIPT: a program to produce both
detailed and schematic plots of protein structures. J. Appl. Crystallogr.
24, 946−950.
(32) Merritt, E. A., and Murphy, M. E. (1994) Raster3D Version 2.0.
A program for photorealistic molecular graphics. Acta Crystallogr., Sect.
D: Biol. Crystallogr. 50, 869−873.
(33) Schrodinger Release 2016-1: Jaguar; Schrodinger, LLC: New
̈
̈
York, NY, 2016.
(15) Volbeda, A., Darnault, C., Renoux, O., Reichmann, D., Amara,
P., Ollagnier de Choudens, S., and Fontecilla-Camps, J. C. (2016)
Crystal structures of quinolinate synthase in complex with a substrate
analogue, the condensation intermediate, and substrate-derived
product. J. Am. Chem. Soc. 138, 11802−11809.
́
(34) Vass, M., Tarcsay, A., and Keseru, G. M. (2012) Multiple ligand
docking by Glide: implications for virtual second-site screening. J.
̋
Comput.-Aided Mol. Des. 26, 821−834.
(16) Esakova, O. A., Silakov, A., Grove, T. L., Saunders, A. H.,
McLaughlin, M. I., Yennawar, N. H., and Booker, S. J. (2016)
I
DOI: 10.1021/acschembio.7b01104
ACS Chem. Biol. XXXX, XXX, XXX−XXX