MgATP Regulates Allostery
and Fiber Formation in IMPDHs
Gilles Labesse,1,2Thomas Alexandre,3,4,5Laure `ne Vaupre ´,3,4Isabelle Salard-Arnaud,3,4,8Jose ´phine Lai Kee Him,1,2
Bertrand Raynal,6,7Patrick Bron,1,2and He ´le `ne Munier-Lehmann3,4,*
1CNRS, UMR5048, Universite ´s Montpellier 1 et 2; Centre de Biochimie Structurale, 29 rue de Navacelles, F-34090 Montpellier, France
2INSERM, U1054, Centre de Biochimie Structurale, F-34090 Montpellier, France
3Institut Pasteur, Unite ´ de Chimie et Biocatalyse, De ´partement de Biologie Structurale et Chimie, 28 rue du Dr Roux, F-75015 Paris, France
4CNRS, UMR3523, F-75015 Paris, France
5Universite ´ Paris Diderot, Sorbonne Paris Cite ´, F-75205 Paris, France
6Institut Pasteur, Proteopole, Plateforme de Biophysique des Macromolecules et de Leurs Interactions, 25 rue du Dr Roux, F-75015 Paris,
7CNRS, UMR3528, F-75015 Paris, France
8Present address: Universite ´ d’Evry-Val-d’Essonne, Laboratoire Analyse et Mode ´lisation pour la Biologie et l’Environnement,
CNRS UMR8587, Evry, France
Inosine-50-monophosphate dehydrogenase (IMPDH)
is a rate-limiting enzyme in nucleotide biosynthesis
studied as an important therapeutic target and its
complex functioning in vivo is still puzzling and
debated. Here, we highlight the structural basis for
the regulation of IMPDHs by MgATP. Our results
demonstrate the essential role of the CBS tandem,
conserved among almost all IMPDHs. We found
that Pseudomonas aeruginosa IMPDH is an octa-
meric enzyme allosterically regulated by MgATP
and showed that this octameric organization is
widely conserved in the crystal structures of other
IMPDHs. We also demonstrated that human IMPDH1
adopts two types of complementary octamers that
can pile up into isolated fibers in the presence of
MgATP. The aggregation of such fibers in the auto-
somal dominant mutant, D226N, could explain the
onset of the retinopathy adRP10. Thus, the regulato-
ry CBS modules in IMPDHs are functional and they
can either modulate catalysis or macromolecular
Nucleotides are ubiquitous and essential biological molecules
that participate in many cellular processes as cofactors or effec-
tors, as a source of energy or as building blocks. Their synthesis
involves multiple reactions catalyzed by enzymes, often pre-
cisely regulated. This is especially true for DNA precursors,
namely dNTP, because base pairing implies properly balanced
biosynthesis to limit mis-incorporation that could lead to muta-
tions. In addition, biosynthetic intermediates should not accu-
mulate because they may represent misleading mimics.
Allosteric regulation has an important role in this context and
has beenconsidered asauniqueplayer. However,enzyme func-
tioning in living cells appears to be more complex because it
could involve additional levels of regulation and take advantage
of dedicated subcellular localization and, more recently, supra-
molecular organization (Narayanaswamy et al., 2009; Noree
et al., 2010). This adds new dimensions to the already tantalizing
annotations of genomes and it may also imply the reassessment
of supposedly well-characterized systems, especially when dis-
ease-related mutations are not fully understood using current
models. Purine metabolism may represent such an archetypal
example with the recent identification of the so-called purino-
some (An et al., 2008) regulated via a novel mechanism involving
its dynamic assembly and disassembly (An et al., 2008). This
complex is composed of six enzymes responsible for the
inosine-50-monophosphate (IMP) biosynthesis in eukaryotes
from phosphoribosyl pyrophosphate (Deng et al., 2012). Muta-
tions identified in patients with neurological diseases are located
in two out of the six enzymes and affect purinosome assembly
(Baresova et al., 2012). Surprisingly, this supramolecular assem-
bly does not contain the enzyme catalyzing the next step
in guanosine monophosphate biosynthesis, the ubiquitous
nicotinamide adenine dinucleotide (NAD)-dependent IMPDH
(EC 126.96.36.199). Indeed, this enzyme has been shown to cluster
into macrostructures induced by disturbance of the nucleotide
pools (Carcamo et al., 2011; Ji et al., 2006; Thomas et al.,
2012). While the purinosome complex may ensure optimal sub-
strate channeling and delivers only IMP to the cell, IMPDH may
be involved in other levels of regulation.
IMPDH controls the de novo synthesis of guanosine nucleo-
tides and its inhibition causes not only a reduction of the guanine
nucleotide pools, but also more importantly an imbalance be-
tweenadenineandguanine nucleotides, leadingto wide-ranging
repercussions (Hedstrom, 2009; Pankiewicz and Goldstein,
2003).As a consequence, IMPDH has emerged asa major target
for antiviral (Nair and Shu, 2007), antiparasitic (Umejiego et al.,
2008), antileukemic, and immunosuppressive therapies (Chen
and Pankiewicz, 2007; Ratcliffe, 2006); and more recently,
some antibacterial applications were proposed (Hedstrom
et al., 2011). Accordingly, it was the subject of various functional
Structure 21, 975–985, June 4, 2013 ª2013 Elsevier Ltd All rights reserved 975
and structural studies. The catalytic domain has been well char-
acterized with X-ray crystallography and biochemistry, and
potent inhibitors targeting IMPDHs bind into this domain (Gold-
stein et al., 2003; Shu and Nair, 2008). Conversely, the role of
the accessory CBS modules is less understood. Deletion of
these modules in the human or Escherichia coli IMPDHs did
not significantly affect the enzyme activity (Nimmesgern et al.,
1999; Pimkin and Markham, 2008). However, an E. coli strain
harboring a guaBDCBSgene is dramatically impaired in the ratio
between its purine nucleotide pools (Pimkin and Markham,
2008) and sensitizes the bacterium to adenosine and inosine
leading to growth arrest on minimal media (Pimkin et al., 2009).
In the case of human IMPDHs, contradictory results have been
reported regarding the role of the CBS modules in binding of
ATP (Mortimer and Hedstrom, 2005; Pimkin and Markham,
2008; Pimkin et al., 2009; Scott et al., 2004; Thomas et al.,
2012). Interestingly, point mutations of IMPDH1 responsible for
retinal degeneration and adRP10 (Aherne et al., 2004; Bowne
ules. These mutations were shown to have no effect on the
catalytic activity but to decrease the affinity for single-stranded
nucleic acids (Kozhevnikova et al., 2012; Mortimer and Hed-
strom, 2005) or to enhance protein aggregation (Aherne et al.,
2004). All these observations emphasized the importance of
the CBS modules, although the underlying molecular mecha-
nisms of action remain to be clarified.
Meanwhile, the structural and functional characterization of
other CBS modules has advanced in various other proteins.
These modules were rapidly recognized as potential 60-residue
regulatory elements, usually associated in tandem to form a
so-called Bateman domain (Bateman, 1997). They are found in
a wide variety of proteins with divergent functions and belonging
to all kingdoms (Ignoul and Eggermont, 2005). They were first
identified in cystathionine beta-synthase, where they play a reg-
ulatory role on the catalytic activity with S-adenosylmethionine
as the positive effector. A similar functional role could be attrib-
uted to the CBS modules of the AMP-activated protein kinase
(Hardie et al., 2011; Ignoul and Eggermont, 2005). In this case,
AMP is the allosteric activator, which triggers the kinase domain
in its active conformation through its binding to the CBS mod-
ules. By similarity, involvement of nucleotides in the regulation
of IMPDH has been postulated but is still under debate.
Using a multidisciplinary approach, we describe the catalytic
and structural properties of the recombinant IMPDH from
Pseudomonas aeruginosa (IMPDHpa) as well as the ultrastruc-
tural characterization of the human IMPDH1 in its wild-type
and one of its pathogenic mutant forms. We report IMPDH in
an octameric form that is allosterically regulated by MgATP
through its CBS modules. Subsequently, we have found similar
octameric states in other IMPDH deposited crystal structures.
Moreover we show by cryo-electron microscopy that human
IMPDH1 exist as octamers, which polymerize in the presence
of MgATP in the case of the WT enzyme and form aggregating
fibers in that of the D226N mutant. Our findings shed light on
the functional and structural role of the CBS domains in IMPDH
regulation and provideamechanism for theadRP10onset open-
ing the road for future treatments.
Recombinant IMPDHpa Is an Octamer in Solution
as the gene coding for IMPDHpa (Stover et al., 2000; Winsor
et al., 2011). The recombinant enzyme was expressed in E. coli
(around 30% of soluble proteins) and purified using affinity chro-
matography at over 95% purity as indicated by SDS-PAGE. In
mer) eluted in a single peak, at the same volume as ferritin
(440 kDa), indicating that the protein may be an octamer.
IMPDHpa apo form revealed only one species with a sedimenta-
tion coefficient (S20,w) of 14.5 ± 0.2 S (Figure 1) at all the concen-
tration investigated (1.8–120 mM). The detected species had a
frictional ratio of 1.4, corresponding to an octamer in good
agreement with size exclusion chromatography (Table 1). There
was no evidence for the existence of a tetrameric arrangement in
any of the investigated conditions, neither in the absence nor in
the presence of a reducing agent.
The size and shape of IMPDHpa were also investigated using
SAXS measurements at 22 and 58 mM (Figure S1 available
online). The radius of gyration (Rg) and the molecular weight
(MW), which were calculated using the Guinier law, match those
expected for an octamer (Rg ?6.19 nm and MW ?430 kDa).
Figure 1. Analytical Ultracentrifugation Analysis of IMPDHpa at
1 mg/ml (18 mM)
(A and B) Sedimentation profiles (dots) and direct boundary modeling of data
analyzed with a sedimentation coefficient distribution c(s) of Lamm equation
solutions (plain line) in the absence (A) or in the presence of MgATP (B).
(C and D) Continuous sedimentation coefficient distribution analysis in the
absence (C) or in the presence of MgATP (D). Sedimentation coefficients are
expressed in Svedberg units (1S = 10?13s) at 20?C in water.
MgATP Modulates IMPDHs
976 Structure 21, 975–985, June 4, 2013 ª2013 Elsevier Ltd All rights reserved
Therefore, IMPDHpa is natively an octamer in solution in the
large range of protein concentrations studied (from low concen-
trations used in kinetic experiments to high concentrations used
Allosteric Regulation of IMPDHpa by MgATP
IMPDHpa was fully active at pH 8.0 and in the presence of a
minimal concentration of 100 mM K+as found for other
IMPDHs. Other monovalent cations such as Na+did not yield
an active enzyme. At saturating concentrations of IMP (6 mM),
the reaction rate of IMPDHpa as a function of NAD+concentra-
tion followed the Michaelis-Menten equation with an average
value of KmNAD+= 139 ± 14 mM and Vm= 2.26 ± 0.07 U/mg
(Figure S2 and Table 2). The plot of IMPDHpa activity at variable
concentrations of IMP and saturating concentrations of
NAD+(2 mM) was sigmoidal, indicating cooperative kinetics
(Figure 2A). This behavior has not been reported so far for any
IMPDH. Three separate experiments using different prepara-
tions of IMPDHpa yielded the following average values:
Vm = 2.13 ± 0.07 U/mg, K0.5 = 1,760 ± 109 mM, and
nH= 1.55 ± 0.08 (Table 3).
Natural 50-mono-, di-, and triphosphate nucleotides were
explored as potential effectors. At saturating concentration of
IMP and NAD+(10 mM and 3 mM, respectively) and 5 mM
MgCl2, MgATP increased the reaction rate by a factor of 2.7,
with half-saturation at 0.08 mM (Figure S3). The presence of a
divalent cation was necessary to attain full activation (yet not
necessary to get activation), manganese and magnesium being
equally efficient. The activation was not due to a phosphate
transfer because nonhydrolyzable ATP such as AMPPCP or
AMPPNP was also effective. On the other hand, while 30-dATP
was able to mimic the effect of ATP, 20-dATP, ADP, or AMP
and all the other natural nucleotides tested were ineffective.
MgATP was found to be a positive effector, which increased
both the maximal rate (Figure S3) and the affinity for IMP (Fig-
ure 2A and Table 3), but decreased the affinity for NAD+(Fig-
ure S2; Table 2) with substrate inhibition as commonly observed
in different IMPDHs (Hedstrom, 2009).
Crystal Structure of IMPDHpa in Its apo Form
First, the crystal structure of the apoenzyme was solved by mo-
lecular replacement at a 4 A˚resolution (Table 4). Six monomers
were observed in each asymmetric unit. Each monomer assem-
bled around the 4-fold axis to form a tetramer, which was similar
to those previously described for other IMPDHs. Despite the
rather low resolution, the catalytic domain was clearly visible.
Conversely, little or no electron density appeared for the CBS
modules, which were thus not modeled in the structure. A similar
conclusion was drawn from an independent structure determi-
nation at much higher resolution (2.23 A˚; Protein Data Bank
[PDB] 4AVF) by the AEROPATH project (Moynie et al., 2013).
Surprisingly, in our structure, no ligand (IMP, ATP, or NAD) was
bound despite their presence in the protein crystallization buffer.
Interestingly, each of the six independent tetramers observed
in the asymmetric unit appeared to be related to another
perpendicular to the 4-fold symmetry axis. Accordingly, in this
apo structure, we have detected three octamers built up from
the six independent monomers and the various symmetries.
interface involving the long loop 371–427 connecting the sec-
ondary structure b8 and a8 and protruding from each monomer.
This loop is rather well conserved among IMPDHs and highly
similar interactions can be observed in the crystal structures of
the apo forms of Streptococcus pyogenes (IMPDHsp; PDB
1ZFJ) and Bacillus anthracis (IMPDHba; PDB 3TSB) IMPDHs
(see subsequent information).
Crystal Structure of IMPDHpa in Complex with MnATP
Better diffracting crystals were obtained with Mn2+as a surro-
gate of Mg2+. The structure of the MnATP-bound IMPDHpa
was solved by molecular replacement and refined to a
2.49 A˚resolution (Figure 3 and Table 4).
The overall structure of the catalytic domain seemed only
slightly affected by the binding of MnATP in the CBS modules.
It resembled the catalytic domain of other IMPDHs (Goldstein
et al., 2003; Hedstrom, 2009). However, the b8-a8 loop (residues
371–427) was no longer visible in the electron density. Two
monomers were present in the asymmetric unit (Figure 3A). As
in the apo form, two independent tetramers could be deduced
from the crystal symmetry 4-fold axis, while an octamer built
up from these two tetramers thanks to a noncrystallographic
2-fold pseudosymmetry. This octamer appeared to be stabilized
by a larger interface because now it involved the clearly visible
CBS tandems. Each of these modules appeared to bind one
molecule of ATP and one manganese ion (Figures 3A and 3B).
The two CBS modules (residues 96–156 and 158–200, respec-
tively) were connected to the catalytic domain by two short seg-
ments (residues 90–92 and 201–208). The latter adopted one
helical turn (residues 197–201) and an extended conformation
(residues 201–208), while the first segment was shorter and
irregular (Figure 3A).
In this structure, the two CBS modules adopted the topology
usually observed among the other known CBS structures, called
Table 1. Hydrodynamic Characteristics of IMPDHpa in the
Absence or Presence of MgATP
AUCS020, w(10?13s?1)14.5 ± 0.214.4 ± 0.1
longest distance (A˚)
The frictional ratio f/f0is characteristic of the elongation and hydration
when compared to an anhydrous sphere. Theoretical sedimentation co-
efficients, radius of gyration, and distance of the complexes were calcu-
lated from the crystal structures (see under Experimental Procedures).
These values were also calculated from ab initio models based on
SAXS or EM data.
AUC, analytical ultracentrifugation; ND, not determined.
Also see Figure S1.
MgATP Modulates IMPDHs
Structure 21, 975–985, June 4, 2013 ª2013 Elsevier Ltd All rights reserved 977
a Bateman tandem. Conversely, the two Bateman tandems in
the IMPDHpa structure appeared slightly tilted, while a common
in-plane assembly has been previously described in most other
bound Bateman tandems. This original feature might correlate
with the binding of two MnATP per tandem, while most Bateman
tandems bind only one nucleos(t)ide. Indeed, the four ATP mol-
ecules all brought their phosphate groups into the interface.
These large groups might prevent a perfect closure of the Bate-
man tandem. Otherwise, the interactions between each CBS
module and the adenosine moiety matched those already
observed in other CBS complexes. The recognition of the
adeninerings wasmainlyduetohydrogenbonds fromtwoback-
bone carbonyl atoms to their N6 nitrogen (V98 and G120 for ATP
no. 1 and V159 and K181 for ATP no. 2), while the nucleobases
were sandwiched by two hydrophobic residues (F118 and
L194 for ATP no. 1, see Figure 3C; and I132 and I179 for ATP
no. 2, see Figure 3D). The ribose moieties interacted tightly
with two aspartic acids (D137 and D199) through hydrogen
bonds involving both O20and O30hydroxyl groups. Ionic interac-
tions were observed between the basic side chains of R136,
K181, and R198 and the phosphate groups, while E180 pointed
toward two manganese ions. One of these ions stabilized the
triphosphate groups protruding from one Bateman tandem,
while the second is chelated by the third phosphate of the four
ATP molecules at the tandem interface.
These structural features perfectly matched the strict require-
ment for a divalent cation and the specificity for ATP (versus
other nucleotides including ADP). The observed complex sheds
light on the role of the CBS modules, conserved in most IMPDHs
as regulatory domains. It also confirmed that CBS tandem
dimerized in IMPDHpa, with an interface of roughly 920 A˚2(per
monomer) sufficient to stabilize the formation of large octamers
Role of the CBS Modules in IMPDHpa Regulation
We assessed how the CBS modules were involved in the regula-
tion of IMPDHpa by constructing a variant deleted of these two
modules by linking alanine 92 directly to lysine 202. The DCBS
variant was properly expressed in E. coli and retrieved in the sol-
uble fraction. It was purified following the same procedure as
described for the wild-type IMPDHpa. In size exclusion chroma-
tography, the purified DCBS variant (39,486 Da per monomer)
eluted as aldolase (158 kDa), compatible with a tetramer. Sedi-
mentation velocity experiments showed that, at a concentration
of 2 mM, only one species could be detected with a sedimenta-
tion coefficient (S20,w) of 8.3S compatible with a tetramer. On
be detected as well as a species with a S20,wof 12.7S corre-
gested the existence of an equilibrium between a tetramer and a
dimer of tetramers. Therefore the data were analyzed globally
ization constant at 74 mM.
As expected, the specific activity of the DCBS variant was not
sensitive to MgATP (Figure S3), except at high concentrations
where a slight inhibition was observed. Moreover, the DCBS
variant exhibited Michaelian kinetics for both substrates (Fig-
ure S4). Surprisingly, its affinity for IMP as well as its specific
activity corresponded to the ones of the wild-type IMPDHpa in
the presence of MgATP, i.e., in its activated form.
Therefore, the CBS modules appear to mediate the allosteric
regulation by MgATP and stabilize the octameric state.
In-Depth Characterization of the MgATP Binding Sites
within the Two CBS Modules
We focused on two residues (D137 and D199) of the CBS mod-
ules of IMPDHpa involved in the interactions with the effector
because (1) these aspartic acids were highly conserved among
IMPDHs; (2) they were hydrogen-bonded to the 20-hydroxyl
group of ATP (Figures 3C and 3D), which is critical in the
activation mechanism as demonstrated by the absence of effect
of 20-dATP on the catalytic activity; and (3) D199 corresponded
to D226 in human IMPDH1, which is frequently substituted by
asparagine in the RP10 form of adRP (Bowne et al., 2002).
Two single mutants (D137N and D199N) as well as the double
esis and overexpressed in E. coli. All the variants eluted from the
gel filtration column as octamers as did the wild-type IMPDHpa.
Their kinetic parameters were determined as a function of the
concentration of either NAD+(Table 2) or IMP (Table 3 and Fig-
ure 2) at saturating concentrations of the other substrate, and
in the absence or presence of MgATP. All the variants were still
activated by MgATP, yet at higher concentrations than for the
wild-type enzyme (half-saturation concentrations of 0.3, 1.2,
Figure S3). The specific activity of all the variants was only
slightly modified, with the D137N D199N double mutation
showing the most detrimental effects (specific activity of
Table 2. Kinetic Parameters of Wild-Type IMPDHpa and Variants, with NAD+as Variable Substrate
No Effector With MgATP
Wild-type 2.26 ± 0.07139 ± 14– 9.0 ± 0.5498 ± 444,067 ± 550
DCBS10.40 ± 0.50580 ± 474,445 ± 5668.9 ± 0.9499 ± 78 3,967 ± 912
D137N1.18 ± 0.03105 ± 11–6.0 ± 0.3505 ± 377,490 ± 1106
D199N1.09 ± 0.0293 ± 7–8.1 ± 0.5450 ± 5311,789 ± 3685
D137N D199N 0.28 ± 0.0181 ± 8–2.1 ± 0.4513 ± 1492,599 ± 1233
MgATP (10 mM for D137N D199N and 3 mM for the others), and fitted according to the Michaelis-Menten equation (see under Experimental
Also see Figures S2 and S4.
MgATP Modulates IMPDHs
978 Structure 21, 975–985, June 4, 2013 ª2013 Elsevier Ltd All rights reserved
10%–20%,compared to that of the wild-type enzyme). Similarly,
theKmNAD+of all thevariants was inthe same rangeasthatof the
wild-type enzyme. On the other hand, the dependence of the
activity on the concentration of the second substrate, namely
IMP, revealed striking differences. The only parameter signifi-
cantly affected by the D137N mutation was its affinity for IMP
in the presence of MgATP, which was decreased 5.8-fold. In
contrast, the D199N variant has lost its cooperativity for IMP,
displaying a hyperbolic curve in the absence of MgATP, with
nHclose to 1. Nonetheless, it has retained the allosteric regula-
tion by MgATP (100-fold decrease of the K0.5 for IMP and
4-fold increase of its specific activity).
These results indicated that the two CBS modules are not
equivalent and act in a cooperative manner to regulate the cata-
lytic activity via MgATP binding.
Figure 2. IMPDHpa Activity versus IMP
Enzyme activity was determined at a fixed con-
centration of NAD+(2 mM), and in the absence
(filled square) or in the presence of 3 mM MgATP
(filled circle) except for D137N D199N variant with
10 mM MgATP. The curves correspond to the fit of
the experimental data to the Hill equation and the
calculated parameters are displayed in Table 3.
(A) wild-type, (B) D137N, (C) D199N, and (D)
Table 3. Kinetic Parameters of Wild-Type IMPDHpa and Variants, with IMP as Variable Substrate
No EffectorWith MgATP
Wild-type2.13 ± 0.07 1760 ± 109 1.55 ± 0.085.58 ± 0.2136 ± 4 0.99 ± 0.12
DCBS5.75 ± 0.1134 ± 2 0.89 ± 0.07 4.78 ± 0.0724 ± 2 1.03 ± 0.10
D137N 2.98 ± 0.11 2030 ± 108 1.87 ± 0.114.38 ± 0.17209 ± 21 1.35 ± 0.16
D199N1.11 ± 0.07 1968 ± 2871.09 ± 0.103.72 ± 0.15 44 ± 51.09 ± 0.13
Reaction rateswere determinedat aconstant concentration ofNAD+(2mM),and intheabsenceor presenceofMgATP(10mMforD137N andD199N;
3 mM for the others), and fitted according to the Hill equation (see under Experimental Procedures).
Also see Figure S4.
0.32 ± 0.04 1599 ± 489 0.87 ± 0.151.35 ± 0.04378 ± 30 0.94 ± 0.01
Structural Rearrangements of
IMPDHpa upon Effector Binding
ments were performed for IMPDHpa in
the presence of MgATP, in the same con-
ditions as for the apoenzyme. Again only
one species could be detected with a
sedimentation coefficient (S20,w= 14.4 ±
0.1 S), compatible with an octamer (Fig-
ure 1 and Table 1). This is in excellent
agreement with the sedimentation coeffi-
cient of 14.5 S computed from the crystal
structure of the MgATP-bound form.
Similarly, SAXS data yielded a similar
(425 kDa) in the MgATP-bound state
than for the apoenzyme. Ab initio shape
modeling confirmed the proposed domain organization, with a
large central core and small peripheral modules (Figure S1).
Furthermore, significant changes were observed between the
SAXS curves of the apoenzyme and of the activated form (Fig-
ure S1). The observed changes qualitatively matched those
observed with X-ray crystallography with the overall shape
becoming more spherical in the presence of MgATP (instead of
a more elongated ellipsoid in the apoenzyme). Indeed, the com-
parison of the two crystal structures of IMPDHpa (in its apo-
and MnATP-bound forms) showed that the octameric structures
underwent a dramatic increase in height (from 80 to 100 A˚along
the 4-fold axis) upon effector binding.
To get a clearer view of the oligomeric state and structural re-
arrangements of IMPDHpa in solution, single particle cryo-EM
analysis was performed (Figure 4). Octameric particles were
MgATP Modulates IMPDHs
Structure 21, 975–985, June 4, 2013 ª2013 Elsevier Ltd All rights reserved 979
easily detected and cryo-EM maps were computed at 1.5 and
1.6 nm resolution for the apo- and Mg-bound states, respec-
tively. The hydrodynamic characteristics of both IMPDHpa
states were calculated from the cryo-EM maps, giving a
sedimentation coefficients of 14.7 S in the absence or in the
presence of MgATP, which is in agreement with the analytical
ultracentrifugation experimental values of 14.5 ± 0.2 S and
14.4 ± 0.1 S for IMPDHpa without and with MgATP, respectively.
The dimensions and shape of IMPDHpa deduced from cryo-EM
experiments are therefore consistent with those previously
measured for the enzyme in solution. In these experiments, the
CBS modules were clearly visible, forming four appendices
pointing outward from the roughly cubic octameric core in
both the apo and the activated conformations. Most interest-
ingly, the CBS modules were well defined in the apo cryo-EM
map of IMPDHpa, while they were absent in the X-ray density
map. The addition of MnATP induced a stretching of densities
along the four-fold axis. In addition, the CBS modules appeared
in closer contacts with the catalytic domain than in their apo
conformation. Our cryo-EM study of IMPDHpa confirmed the
overall structural rearrangements deduced from SAXS and
X-ray diffraction and clearly established that IMPDHpa adopts
an octameric architecture in solution at a concentration as low
as 0.1 mg/ml.
Comparisons with Other IMPDHs Structures
As illustrated previously, we have confirmed the octameric na-
ture of IMPDHpa in various conditions, for a wide range of con-
centrations and in different states (unbound and MgATP bound;
see Figure 1 and Figure S1). In the apoenzyme, a long and
371–427 in IMPDHpa) in the catalytic domain appears to link
the two facing tetramers. Conversely, in the MgATP-bound
form, the dimer of tetramers is mainly stabilized by interactions
between the CBS tandems. This is in agreement with the now
well-established dimeric state of most known CBS modules
studied up to now (Ignoul and Eggermont, 2005). Surprisingly,
these two types of octameric states were in marked contrast
with the tetrameric state reported for all the other IMPDHs stud-
ied so far. The octameric organization has been observed in
crystals of the Tritichomonas fetus IMPDH (Whitby et al., 1997)
and the human IMPDH2 (Colby et al., 1999), but was considered
an artifact and non-biologically relevant.
crystallized IMPDHs, reported as tetrameric, we used the server
PISA (Krissinel and Henrick, 2007) to analyze the crystal packing
and search for higher assemblies. Surprisingly, all IMPDH crystal
structures gave rise to prediction of (at least) marginally stable
octameric states whatever the species or whether a ligand was
present. These octameric assemblies were collected and super-
corresponded to the dimer of tetramers observed for the
IMPDHpa. Indeed, similar convex octamers appeared to be
predominant in all the other crystal structures (Figures 5A
and 5B). Furthermore, the structural variations observed among
the convex octamers (between all known crystal structures)
matched the structural rearrangements observed upon effector
binding in the case of IMPDHpa. Indeed, the human IMPDH2
octamer adopts a fairly spherical structure in the presence of
ribavirin monophosphate and c2-mycophenolic adenine dinu-
cleotide (PDB 1NF7), while the IMPDHsp octamer (PDB 1ZFJ)
forms an elongated ellipsoid (Figure 4A). The interface between
two tetramers involved the CBS modules (when present and
structured as in IMPDHpa; IMPDH2/PDB 1NF7; IMPDHsp/PDB
1ZFJ, and IMPDHba/PDB 3TSD) or long hairpins (corresponding
to the b8-a8 loops) protruding from the catalytic domains. These
extended loops were well conserved, as were the CBS modules.
All these features suggested that the octameric state was a
conserved IMPDH characteristic that had been underappreci-
ated until now.
Surprisingly, human IMPDH1 (PDB 1JCN) seemed to adopt a
distinct architecture because the dimerization of tetramers
involved an alternative interface. The resulting concave octamer
was predicted to be stabilized bya uniqueand largely hydropho-
bic interface. In this configuration, the tandem of CBS modules
remained isolated and protruded into the solvent.
Table 4. Data Collection and Refinement Statistics for the
Space group I4 I4
a, b, c (A˚) 131.3, 131.3, 393.0110.0, 110.0, 194.4
No. molecules in a.u
Resolution (A˚)– 2.49
No. protein atoms–6,035
No. water molecules–205
–4 Mn + 4 ATP
Bond lengths (A˚)
Bond angles (?)–1.634
Rmsd, root-mean-square deviation.
aRmerge = ShklSirIhkl,i ? Iaverage,hklr/rShklSirIhkl,ir3100.
bRwork = ShklrFobs ? Fcalcr/Shkl rFobsr3100.
used in the refinement (5%).
dDeviation from ideal values.
MgATP Modulates IMPDHs
980 Structure 21, 975–985, June 4, 2013 ª2013 Elsevier Ltd All rights reserved
These results prompted us to study the oligomeric state of the
or in the presence of MgATP.
Electron Microscopy Reconstruction of the Wild-Type
and D226N Human IMPDH1
The wild-type human IMPDH1 was overexpressed and purified
as described in the Experimental Procedures section. At nonsa-
of MgATP (at a concentration up to 3 mM) was observed on
human IMPDH1 catalytic activity (kcat?0.4 s?1), in agreement
with previous reports (Carr et al., 1993; Mortimer and Hedstrom,
2005; Thomas et al., 2012).
Cryo-EM experiments on freshly purified enzyme were per-
formed (Figures 6A and 6B). Human IMPDH1 isolated particles
had a size similar to that of IMPDHpa. Image analysis revealed
that it forms two types of octamers in solution. The major popu-
lation of particles (69%) displayed the association of two tetra-
mers present in the crystal state (PDB 1JCN), with the eight
CBS modules pointing outward (concave octamer). In parallel,
the minor population of particles (31%) adopted an octameric
architecture similar to that observed for IMPDHpa in its apo
form (convex octamer). These two octameric architectures are
compatible and their superimposition strongly suggested they
could pile up. Indeed, the two types of octamers appeared to
interact and initiate a few nascent fibers. In the presence of
MgATP, particles were nearly all associated in thin fibers ranging
from 50 to 200 nm long (Figures 6C and 6D). Class averages re-
vealed that fibers were actually formed by a superimposition of
octamers. The small CBS modules were clearly visible protrud-
ing out perpendicularly to the long axis of the fiber. Apart from
demonstrating that MgATP could bind to the human IMPDH1,
these results showed that the dimerization of CBS modules
was a limiting step in fiber formation.
This behavior prompted us to study the effect of the conserva-
tive but pathogenic mutation D226N in human IMPDH1. This
Figure 3. Structure of IMPDHpa with Its
(A) The crystal structure of IMPDHpa is shown in
red and blue ribbon (catalytic domains) and pink
and purple ribbons (CBS tandems). The electron
density corresponding to the four ATP molecules
is shown as green meshes and surfaces. The
manganese atoms are shown as orange spheres.
(B) Zoom on the dimerized CBS modules. Same
color codes as in panel (A).
(C and D) Zoom on the ATP binding cavities no. 1
and no. 2, respectively. ATP molecules and the
side chain of interacting residues are shown in
sticks. The figure was prepared using Pymol
mutation corresponds to the D199N
variant of IMPDHpa described previously
in which allosteric regulation is impaired.
The D226N variant consisted of large ag-
gregates. At higher magnification, these
aggregates appeared to be formed by
numerous intertwined fibers (Figure 6E).
No isolated fiber could be visualized, suggesting that fibers
strongly interacted laterally. Consequently, the D226N variant
could form fibers even in the absence of ATP. The fact that fibers
were highly intertwined suggested that the mutated CBS mod-
ules presented a certain degree of freedom or of misfolding,
allowing cross-connections between neighboring fibers.
IMPDH (Hedstrom, 2009; Pankiewicz and Goldstein, 2003) is an
ubiquitous enzyme, with the exception of two protozoan para-
sites (Giardia lamblia and Trichomonas vaginalis). It controls
the guanine nucleotides pool and occupies a key position in
the purine nucleotide metabolism. Our biochemical character-
ization of the bacterial IMPDHpa has revealed peculiar kinetic
properties: (1) a cooperativity toward IMP and (2) MgATP as
the positive effector. Scott and colleagues (Scott et al., 2004)
have also reported ATP as an allosteric activator of the human
IMPDH2, but these results were controversial, as others have
not been able to reproduce them (Mortimer and Hedstrom,
2005; Pimkin and Markham, 2008). The crystal structure of the
MnATP/IMPDHpa complex described herein further confirmed
the allosteric regulation of this bacterial IMPDH by MgATP and
sheds light on its interactions with ATP. Each CBS module was
liganded to one MnATP, globally resorting the same residues
to interact with both the ribose and the phosphate moieties of
ATP and the manganese ion. The main differences between
the two CBS modules lie in their relative position with respect
to the catalytic site. This translates in different effects of aspara-
gine substitutions of aspartic acid 137, distant from the catalytic
domain, and aspartic acid 199, which initiates the short segment
(residues 200–212 in IMPDHpa) linking the second CBS module
to the catalytic domain. Indeed, while mutation D137N did not
affect the cooperativity (although it affected the affinity for
ATP), mutation D199N resulted in a total loss of IMPDHpa
MgATP Modulates IMPDHs
Structure 21, 975–985, June 4, 2013 ª2013 Elsevier Ltd All rights reserved 981
A further analysis revealed that the octameric state of
IMPDHpa we report here is in fact well conserved in all other
IMPDH crystal structures and had been overlooked up to now.
This oligomeric state is promoted by CBS module dimerization.
The connection of the octameric state with the allosteric regula-
tion of IMPDHs may allow us to better understand the discrep-
ancies between the reported in vitro and the in vivo effects of
terized IMPDHs was not affected in vitro by the deletion of the
fine-tuning of the enzymatic activity in an in vivo context would
be lost. The conserved ATP-binding pocket found in the CBS
modules of all IMPDHs sequenced so far suggested that ATP
or related adenosine-containing compounds were likely to be
IMPDH1 for which ATP binding has a drastic impact on the over-
all architecture as demonstrated by our study. Moreover, the key
aspartic acid 199 in IMPDHpa corresponds to D226 in human
IMPDH1, whose substitution by asparagine is one of the most
frequent mutations leading to adRP. Aherne and colleagues
(Aherne et al., 2004) have reported that two IMPDH1 variants
(R224P and D226N) were mostly found in the insoluble fraction.
Our cryo-EM images revealed that the D226N IMPDH1 variant
formed aggregated fibers, which could explain the in vivo cyto-
toxic effect. Aggregation of the R224P mutant was recently
shown to be prevented when inducing chaperone overexpres-
sion using 17-AAG (Tam et al., 2010). It can thus be foreseen
that drugs activating cellular chaperones could be of general
use to counter the onset of IMPDH-related severe retinopathies.
IMPDHpa (WT and variants) and human IMPDH1 were expressed in E. coli
(Table S1). The recombinant proteins were purified using affinity and/or size
exclusion chromatography. Analytical ultracentrifugation was carried out to
determine the oligomeric state of IMPDHpa in different conditions. Kinetic
parameters were determined at 30?C following the synthesis of NADH at
340 nm. Detailed information is described in the Supplemental Experimental
Activity Assay of IMPDHpa
IMPDHpa activity was determined at 30?C by monitoring the synthesis of
NADH. The reaction medium contained 50 mM Tris-HCl pH 8, 100 mM KCl
and 1 mM DTT, and various concentrations of IMP, NAD+, ATP, and MgCl2,
to which IMPDHpa (final concentration range: 0.18–1.8 mM) diluted in 20 mM
potassium phosphate pH8and0.1MKClwasthenadded.Oneunitofenzyme
activity corresponds to 1 mmole of the product formed in 1 min at 30?C and
pH 8. Experimental data were fitted using the Kaleidagraph software accord-
ing to the Michaelis-Menten equation v = Vm[S]/(Km+ [S]), the substrate
inhibition equation v = Vm[S]/(Km+ [S] + [S]2/KI), or to the Hill equation
v = Vm[S]nH/(K0.5nH+ [S]nH), where v is the reaction rate, Vmthe maximal
rate, [S] the NAD+or IMP concentration, Kmthe Michaelis-Menten constant,
KIthe inhibitory constant, K0.5the IMP concentration at half-saturation, and
nHthe Hill number index.
Crystallization and X-Ray Data Collection
Crystallization was first obtained for the apoenzyme (despite the presence of
2 mM ATP, 1 mM IMP, and 1 mM NAD) in PEG2000 (10%), 30 mM sodium
citrate at pH 5.1, in the presence of heptanetriol. Subsequently, the MnATP-
bound form was obtained in PEG1500 (12%), 40 mM potassium citrate and
100 mM Tris at pH 8.0, in the presence of 2 mM ATP, 1 mM IMP, 1 mM
NAD, and 20 mM MnCl2. X-ray diffraction data sets were collected from frozen
single crystals at the European Synchrotron Radiation Facility (Grenoble,
France, beamline ID14-4) and processed with the programs MOSFLM,
SCALA, and TRUNCATE from the CCP4 program suite (CCP4, 1994).
The structure was solved by molecular replacement using the program Mol-
Rep (Vagin and Teplyakov, 2010) and the crystal structure of the IMPDH from
Streptococcus pyogenes (PDB 1ZFJ) (Zhang et al., 1999) as a search model.
The apo IMPDHpa structure was solved at 4.0 A˚while the MnATP-bound
form was refined at 2.5 A˚resolution. Model refinement was performed using
the program COOT (Emsley and Cowtan, 2004) and the program REFMAC5
(Murshudov et al., 1997), using a translation/liberation/screw model (Winn
et al., 2001). In the final model, short segments of the protein were not clearly
(last 28 residues) lying over the NAD-binding site could not be modeled in the
SAXS Experiments and Data Analysis
Synchrotron X-ray scattering data from solutions of IMPDHpa were collected
ontheSWINGbeamline oftheSoleil Synchrotron (Saint-Aubin, France) usinga
PCCD-170170 detector at a wavelength of 1.03 A˚. The scattering patterns
were measured by merging 10–20 data recordings with 1 s exposure time
each, for two solute concentrations (1.14–3.03 mg/ml). To check for radiation
damage, all successive exposures were compared, and no changes were
Figure 4. Electron Microscopy of IMPDHpa without and with MgATP
(A and B) Three-dimensional EM maps of frozen-hydrated particles of
IMPDHpa without (APO) and in the presence of MgATP. IMPDHpa adopts an
octameric architecture in solution with CBS modules of two IMPDH tetramers
interplaying together in, thereby forming a well-defined external domain.
MgATP induces structural rearrangements: the octamer stretches along the
4-fold axis and the CBS modules move closer to the central core of the
(C and D) Docking of the apo and MgATP-bound IMPDHpa structures into
corresponding cryo-EM maps. In the apoenzyme, the CBS modules appear
clearly in the cryo-EM map while they are absent in our X-ray structure. The
small dome, observed in the center of the square domain, corresponds to
interactions between the four His-tag extremities. Because this dome is
absent in the cryo-EM map of the MgATP-bound form, these extremities are
supposed to be more flexible after IMPDHpa structural rearrangements upon
MgATP Modulates IMPDHs
982 Structure 21, 975–985, June 4, 2013 ª2013 Elsevier Ltd All rights reserved
detected. Using the sample-detector distance of 1.8 m, a range of momentum
transfer of 0.0065 < s < 0.35 A˚?1was covered (s = 4p sin(q)/l, where 2q is the
scattering angle, l is the X-ray wavelength). The data were processed using
standard procedures and extrapolated to infinite dilution using the program
PRIMUS (Konarev et al., 2003). The forward scattering, I(0), and the radius of
gyration, Rg, were evaluated using the Guinier approximation, assuming that
at very small angles (s < 1.3/Rg) the intensity is represented as I(s) = I(0)
exp(?s2Rg2)/3). The values of I(0) and Rg, as well as the maximum dimension,
Dmax, and the interatomic distance distribution functions, [p(r)], were also
computed using the program GNOM (Svergun, 1992). GASBOR (Svergun
et al., 2001) was used to compute ab initio shapes using a P42 symmetry as
a constraint and a variable Dmax (170, 175, and 180 A˚). Ten independent
shapes for each Dmax were aligned and averaged for computing the final
shape using DAMAVER (Volkov and Svergun, 2003).
Electron Microscopy and Image Processing
Three microliters of IMPDH samples at a final concentration ranging from 0.03
to 1.0 mg/ml were applied to glow discharged Quantifoil R 2/2 grids (Quantifoil
Micro Tools GmbH, Jena, Germany), blotted for 1 s, and then flash-frozen in
liquid ethane. Cryo-EM was carried out on a JEOL 2200FS FEG operating at
200 kV under low-dose conditions (total dose of 20 electrons/A˚2) in the zero-
energy-loss mode with a slit width of 20 eV. Images were recorded on a
4K 3 4K slow-scan CCD camera (Gatan inc.) at a nominal magnification of
100,0003 for single particles and 50,0003 for fibers with defocus ranging
from 1.4 to 2.5 mm.
Particles were semi-automatically extracted using boxer (Ludtke et al.,
1999). The defocus and astigmatism values of each micrograph were deter-
mined using CTFFIND3 (Mindell and Grigorieff, 2003). IMAGIC-5 (van Heel
et al., 1996) was used to flip phases and for the subsequent image processing
In case of cryo-EM single particle analysis, particles were binned to
2.3 A˚/pixel, band-pass filtered between 300 and 12 A˚, normalized, and
centered by iteratively alignments against their rotationally averaged sum.
Images were first aligned iteratively against best class averages using MSA
(multi-statistical analysis) and MRA (multi-references alignment) programs.
Images presented into unresolved class averages were excluded for further
analysis. The three-dimensional reconstruction and structure refinement was
iteratively done using FREALIGN program (Grigorieff, 2007) and applying
D4 symmetry. For IMPDHpa, initial models were derived from our resolved
atomic structures, which were converted to electron density map and low-
pass filtered at 50 A˚.
For human IMPDH1, the final cryo-EM map of apo IMPDHpa was filtered to
50 A˚resolution and used as starting model for structure refinement.
Fourier shell correlation (van Heel and Harauz, 1986) was used to estimate
the resolution of the final structures at 1.5 nm and 1.6 nm for apo and
Mg-bound IMPDHpa states with a 0.5 correlation cut-off without masking.
Visualization of the models and fitting of atomic structures into EM maps
were done using CHIMERA (Pettersen et al., 2004).
With the MgATP-bound human IMPDH1 state, images recorded at 50,0003
magnification, corresponding to a pixel size of 2.3 A˚, were aligned iteratively
against best class averages using MSA and MRA programs.
The Protein Data Bank accession number for the coordinates of the MgATP/
IMPDHpa complex reported in this paper is 4DQW. The EM Data Bank acces-
sion codes for the EM maps reported in this paper are EMD-2069 and
EMD-2070 for the apo- and MgATP-bound IMPDHpa states, respectively.
Supplemental Information includes four figures, one table, and Supplemental
Experimental Procedures and can be found with this article online at http://dx.
Emilie Dufour is acknowledged for performing cloning of IMPDHpa and first
crystallization trials, and Marine Tomazi dit Dassonville for participation in
the mutagenesis of IMPDHpa. We thank Fre ´de ´ric Allemand for performing
IMPDH1 site-directed mutagenesis and Yannick Bessin for participation in
IMPDH1 purification. We thank Patrick England for critical reading of the
manuscript. We thank the ESRF staff for their help in recording diffraction
data on beamline ID14-2 and ID14-4. SAXS experiments were recorded on
thebeamline SWING inSOLEIL (Saint-Aubin,France) withthekindhelp of Jav-
ier Perez. G.L., P.B., B.R., and H.M.-L. wrote the paper. G.L. and H.M.-L.
contributed to experimental design. G.L. collected X-ray data, solved IM-
PDHpa structures, and performed SAXS experiments and analysis. L.V.,
I.S.-A., T.A., and H.M.-L. performed cloning and site-directed mutagenesis,
provided purified proteins, and did all the biochemical experiments. J.L.K.H.
andP.B.carriedoutcryo-EMobservations,image processing and dockingex-
periments. B.R. conducted analytical ultracentrifugation experiments and
analyzed the hydrodynamic characteristics of IMPDHpa. This work was sup-
ported in part by the Centre National de la Recherche Scientifique (CNRS),
the Institut National de la Sante ´ Et de la Recherche Me ´dicale (INSERM), the
Conseil Re ´gional d’Ile-de-France (Chemical Library Project, grant nos. I 06-
222/R and I 09-1739/ R, which included a postdoctoral fellowship for I.S.-A.)
and the French Infrastructure for Integrated Structural Biology (FRISBI) ANR-
10-INSB-05-01. T.A. was a recipient of a PhD fellowship from the Conseil
Re ´gional d’Ile-de-France and the ‘‘D.I.M. maladies infectieuses, parasitaires
et nosocomiales e ´mergentes’’ 2011.
Figure 5. Oligomeric State Comparison
The possible octameric states of already crystal-
lized IMPDHs were computed using the server
PISA (Krissinel and Henrick, 2007). The arrange-
ment of contacting tetramers (in blue and red
ribbons) is shown in various crystal structures
varying in sequence and bound state (apo-, sub-
strate-bound; inhibited or activated).
(A) In four distinct structures (human IMPDH2:
PDB 1NF7; IMPDHpa: PDB 4DQW; IMPDHsp:
PDB 1ZFJ; and IMPDHba: PDB 3TSB) the CBS
modules are almost completely visible.
(B) In most other crystal structures (IMPDHtf: PDB
horikoshii IMPDH]: PDB 2CU0 and IMPDHcp
mainly involved a long loop protruding from the
catalytic domain. The figure was prepared using
MgATP Modulates IMPDHs
Structure 21, 975–985, June 4, 2013 ª2013 Elsevier Ltd All rights reserved 983
Received: December 14, 2012
Revised: March 11, 2013
Accepted: March 14, 2013
Published: May 2, 2013
Aherne, A., Kennan, A., Kenna, P.F., McNally, N., Lloyd, D.G., Alberts, I.L.,
Kiang, A.S., Humphries, M.M., Ayuso, C., Engel, P.C., et al. (2004). On the
molecular pathology of neurodegeneration in IMPDH1-based retinitis pigmen-
tosa. Hum. Mol. Genet. 13, 641–650.
An, S., Kumar, R., Sheets, E.D., and Benkovic, S.J. (2008). Reversible
compartmentalization of de novo purine biosynthetic complexes in living cells.
Science 320, 103–106.
Baresova, V., Skopova, V., Sikora, J., Patterson, D., Sovova, J., Zikanova, M.,
and Kmoch, S. (2012). Mutations of ATIC and ADSL affect purinosome assem-
bly in cultured skin fibroblasts from patients with AICA-ribosiduria and ADSL
deficiency. Hum. Mol. Genet. 21, 1534–1543.
Bateman, A. (1997). The structure of a domain common to archaebacteria and
the homocystinuria disease protein. Trends Biochem. Sci. 22, 12–13.
Bowne, S.J., Sullivan, L.S., Blanton, S.H., Cepko, C.L., Blackshaw, S., Birch,
D.G., Hughbanks-Wheaton, D., Heckenlively, J.R., and Daiger, S.P. (2002).
Mutations in the inosine monophosphate dehydrogenase 1 gene (IMPDH1)
cause the RP10 form of autosomal dominant retinitis pigmentosa. Hum. Mol.
Genet. 11, 559–568.
Bowne, S.J., Sullivan, L.S., Mortimer, S.E., Hedstrom, L., Zhu, J., Spellicy,
C.J., Gire, A.I., Hughbanks-Wheaton, D., Birch, D.G., Lewis, R.A., et al.
somal dominant retinitis pigmentosa and leber congenital amaurosis. Invest.
Ophthalmol. Vis. Sci. 47, 34–42.
Yao, B., Tamayo, S., Covini, G., von Mu ¨hlen, C.A., and Chan, E.K. (2011).
Induction of cytoplasmic rods and rings structures by inhibition of the CTP
and GTP synthetic pathway in mammalian cells. PLoS ONE 6, e29690.
Carr, S.F., Papp, E., Wu, J.C., and Natsumeda, Y. (1993). Characterization of
human type I and type II IMP dehydrogenases. J. Biol. Chem. 268, 27286–
Chen, L., and Pankiewicz, K.W. (2007). Recent development of IMP dehydro-
genase inhibitors for the treatment of cancer. Curr. Opin. Drug Discov. Devel.
Colby, T.D., Vanderveen, K., Strickler, M.D., Markham, G.D., and Goldstein,
B.M. (1999). Crystal structure of human type II inosine monophosphate dehy-
drogenase: implications for ligand binding and drug design. Proc. Natl. Acad.
Sci. USA 96, 3531–3536.
CCP4 (Collaborative Computational Project, Number 4). (1994). The CCP4
suite: programs for protein crystallography. Acta Crystallogr. D Biol.
Crystallogr. 50, 760–763.
Deng, Y., Gam, J., French, J.B., Zhao, H., An, S., and Benkovic, S.J. (2012).
Mapping protein-protein proximity in the purinosome. J. Biol. Chem. 287,
Emsley, P., and Cowtan, K. (2004). Coot: model-building tools for molecular
graphics. Acta Crystallogr. D Biol. Crystallogr. 60, 2126–2132.
Goldstein, B.M., Risal, D., and Strickler, M. (2003). IMPDH structure and ligand
binding. In Inosine Monophosphate Dehydrogenase: a Major Therapeutic
Target, K.W. Pankiewicz and B.M. Goldstein, eds. (Washington DC: ACS),
Grigorieff, N. (2007). FREALIGN: high-resolution refinement of single particle
structures. J. Struct. Biol. 157, 117–125.
Hardie, D.G., Carling, D., and Gamblin, S.J. (2011). AMP-activated protein
kinase: also regulated by ADP? Trends Biochem. Sci. 36, 470–477.
Hedstrom, L. (2009). IMP dehydrogenase: structure, mechanism, and inhibi-
tion. Chem. Rev. 109, 2903–2928.
Hedstrom, L., Liechti, G., Goldberg, J.B., and Gollapalli, D.R. (2011). The anti-
biotic potential of prokaryotic IMP dehydrogenase inhibitors. Curr. Med.
Chem. 18, 1909–1918.
Ignoul, S., and Eggermont, J. (2005). CBS domains: structure, function, and
pathology in human proteins. Am. J. Physiol. Cell Physiol. 289, C1369–C1378.
Ji, Y., Gu, J., Makhov, A.M., Griffith, J.D., and Mitchell, B.S. (2006). Regulation
of the interaction of inosine monophosphate dehydrogenase with mycophe-
nolic Acid by GTP. J. Biol. Chem. 281, 206–212.
Kennan, A., Aherne, A., Palfi, A., Humphries, M., McKee, A., Stitt, A., Simpson,
D.A., Demtroder, K., Orntoft, T., Ayuso, C., et al. (2002). Identification of an
IMPDH1 mutation in autosomal dominant retinitis pigmentosa (RP10) revealed
following comparative microarray analysis of transcripts derived from retinas
of wild-type and Rho(-/-) mice. Hum. Mol. Genet. 11, 547–557.
Konarev, P.V., Volkov, V.V., Sokolova, A.V., Koch, M.H.J., and Svergun, D.I.
(2003). PRIMUS: a Windows PC-based system for small-angle scattering
data analysis. J. Appl. Cryst. 36, 1277–1282.
W.F., Moshkin, Y.M., and Verrijzer, C.P. (2012). Metabolic enzyme IMPDH is
also a transcription factor regulated by cellular state. Mol. Cell 47, 133–139.
Figure 6. Electron Microscopy of Wild-Type and D226N Human
(A and B) Image processing of frozen-hydrated human IMPDH1 images reveal
two conformational states. The minor state corresponds to an octameric
architecture similar to that of IMPDHpa, while the major state corresponds to
two back-to-back tetramers with eight CBS domains pointing outward from
the core of the particle.
(C and D) Human IMPDH1 particles in the presence of MgATP form fibers,
which correspond to a linear stacking of octameric IMPDH1. CBS domains
point out perpendicular to the fiber axis.
(E and F) The D226N human IMPDH1 variant forms aggregates consisting of
intertwined fibers and involving strong interactions between adjacent fibers.
MgATP Modulates IMPDHs
984 Structure 21, 975–985, June 4, 2013 ª2013 Elsevier Ltd All rights reserved
Krissinel, E., and Henrick, K. (2007). Inference of macromolecular assemblies Download full-text
from crystalline state. J. Mol. Biol. 372, 774–797.
Ludtke, S.J., Baldwin, P.R., and Chiu, W. (1999). EMAN: semiautomated soft-
ware for high-resolution single-particle reconstructions. J. Struct. Biol. 128,
Mindell,J.A.,and Grigorieff,N.(2003).Accurate determination oflocaldefocus
and specimen tilt in electron microscopy. J. Struct. Biol. 142, 334–347.
tosa mutations in inosine 50-monophosphate dehydrogenase type I disrupt
nucleic acid binding. Biochem. J. 390, 41–47.
Moynie, L., Schnell, R., McMahon, S.A., Sandalova, T., Boulkerou, W.A.,
Schmidberger, J.W., Alphey, M., Cukier, C., Duthie, F., Kopec, J., et al.
(2013). The AEROPATH project targeting Pseudomonas aeruginosa: crystallo-
graphic studies for assessment of potential targets in early-stage drug discov-
ery. Acta Crystallogr. Sect. F Struct. Biol. Cryst. Commun. 69, 25–34.
Murshudov, G.N., Vagin, A.A., and Dodson, E.J. (1997). Refinement of macro-
molecular structures by the maximum-likelihood method. Acta Crystallogr. D
Biol. Crystallogr. 53, 240–255.
Nair, V., and Shu, Q. (2007). Inosine monophosphate dehydrogenase as a
probe in antiviral drug discovery. Antivir. Chem. Chemother. 18, 245–258.
Narayanaswamy, R., Levy, M., Tsechansky, M., Stovall, G.M., O’Connell, J.D.,
Mirrielees, J., Ellington, A.D., and Marcotte, E.M. (2009). Widespread reorga-
nization of metabolic enzymes into reversible assembliesupon nutrient starva-
tion. Proc. Natl. Acad. Sci. USA 106, 10147–10152.
Nimmesgern, E., Black, J., Futer, O., Fulghum, J.R., Chambers, S.P.,
Brummel, C.L., Raybuck, S.A., and Sintchak, M.D. (1999). Biochemical anal-
ysis of the modular enzyme inosine 50-monophosphate dehydrogenase.
Protein Expr. Purif. 17, 282–289.
Noree, C., Sato, B.K., Broyer, R.M., and Wilhelm, J.E. (2010). Identification of
novel filament-forming proteins in Saccharomyces cerevisiae and Drosophila
melanogaster. J. Cell Biol. 190, 541–551.
Pankiewicz, K.W., and Goldstein, B.M. (2003). Inosine Monophosphate
Dehydrogenase: A Major Therapeutic Target (Washington DC:: ACS).
Pettersen, E.F., Goddard, T.D., Huang, C.C., Couch, G.S., Greenblatt, D.M.,
Meng, E.C., and Ferrin, T.E. (2004). UCSF Chimera—a visualization system
for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612.
Pimkin, M., and Markham, G.D. (2008). The CBS subdomain of inosine
50-monophosphate dehydrogenase regulates purine nucleotide turnover.
Mol. Microbiol. 68, 342–359.
Pimkin, M., Pimkina, J., and Markham, G.D. (2009). A regulatory role of the
Bateman domain of IMP dehydrogenase in adenylate nucleotide biosynthesis.
J. Biol. Chem. 284, 7960–7969.
Ratcliffe, A.J. (2006). Inosine 50-monophosphate dehydrogenase inhibitors for
the treatment of autoimmune diseases. Curr. Opin. Drug Discov. Devel. 9,
Scott, J.W., Hawley, S.A., Green, K.A., Anis, M., Stewart, G., Scullion, G.A.,
Norman, D.G., and Hardie, D.G. (2004). CBS domains form energy-sensing
modules whose binding of adenosine ligands is disrupted by disease muta-
tions. J. Clin. Invest. 113, 274–284.
Shu, Q., and Nair, V. (2008). Inosine monophosphate dehydrogenase (IMPDH)
as a target in drug discovery. Med. Res. Rev. 28, 219–232.
Stover, C.K., Pham, X.Q., Erwin, A.L., Mizoguchi, S.D., Warrener, P., Hickey,
M.J., Brinkman, F.S., Hufnagle, W.O., Kowalik, D.J., Lagrou, M., et al.
(2000). Complete genome sequence of Pseudomonas aeruginosa PAO1, an
opportunistic pathogen. Nature 406, 959–964.
Svergun, D.I. (1992). Determination of the regularization parameter in indirect-
transform methods using perceptual criteria. J. Appl. Cryst. 25, 495–503.
Svergun, D.I., Petoukhov, M.V., and Koch, M.H.J. (2001). Determination of
domain structure of proteins from X-ray solution scattering. Biophys. J. 80,
Tam, L.C., Kiang, A.S., Campbell, M., Keaney, J., Farrar, G.J., Humphries,
M.M., Kenna, P.F., and Humphries, P. (2010). Prevention of autosomal domi-
nant retinitis pigmentosa by systemic drug therapy targeting heat shock pro-
tein 90 (Hsp90). Hum. Mol. Genet. 19, 4421–4436.
Thomas, E.C., Gunter, J.H., Webster, J.A., Schieber, N.L., Oorschot, V.,
Parton, R.G., and Whitehead, J.P. (2012). Different characteristics and nucle-
otide binding properties of inosine monophosphate dehydrogenase (IMPDH)
isoforms. PLoS ONE 7, e51096.
Umejiego, N.N., Gollapalli, D., Sharling, L., Volftsun, A., Lu, J., Benjamin, N.N.,
Stroupe, A.H., Riera, T.V., Striepen, B., and Hedstrom, L. (2008). Targeting a
prokaryotic protein ina eukaryotic pathogen: identification of lead compounds
against cryptosporidiosis. Chem. Biol. 15, 70–77.
Vagin, A., and Teplyakov, A. (2010). Molecular replacement with MOLREP.
Acta Crystallogr. D Biol. Crystallogr. 66, 22–25.
van Heel, M., and Harauz, G. (1986). Resolution criteria for three dimensional
reconstruction. Optik (Stuttg.) 73, 119–122.
van Heel, M., Harauz, G., Orlova, E.V., Schmidt, R., and Schatz, M. (1996). A
new generation of the IMAGIC image processing system. J. Struct. Biol.
Volkov, V.V., and Svergun, D.I. (2003). Uniqueness of ab initio shape determi-
nation in small-angle scattering. J. Appl. Cryst. 36, 860–864.
Whitby, F.G., Luecke, H., Kuhn, P., Somoza, J.R., Huete-Perez, J.A., Phillips,
J.D., Hill, C.P., Fletterick, R.J., and Wang, C.C. (1997). Crystal structure of
Tritrichomonas foetus inosine-50-monophosphate dehydrogenase and the
enzyme-product complex. Biochemistry 36, 10666–10674.
to model anisotropic displacements in macromolecular refinement. Acta
Crystallogr. D Biol. Crystallogr. 57, 122–133.
Winsor, G.L., Lam, D.K., Fleming, L., Lo, R., Whiteside, M.D., Yu, N.Y.,
Hancock,R.E., andBrinkman, F.S.
for Pseudomonas genomes. Nucleic Acids Res. 39(Database issue), D596–
Zhang, R., Evans, G., Rotella, F.J., Westbrook, E.M., Beno, D., Huberman, E.,
Joachimiak, A., and Collart, F.R. (1999). Characteristics and crystal structure
of bacterial inosine-50-monophosphate dehydrogenase. Biochemistry 38,
MgATP Modulates IMPDHs
Structure 21, 975–985, June 4, 2013 ª2013 Elsevier Ltd All rights reserved 985