Crystal Structure of the ATPPase Subunit and Its Substrate-Dependent Association with the GATase Subunit: A Novel Regulatory Mechanism for a Two-Subunit-Type GMP Synthetase from Pyrococcus horikoshii OT3

Article abstract of DOI:10.1016/j.jmb.2009.10.053

Guanosine 5′-monophosphate synthetase(s) (GMPS) catalyzes the final step of the de novo synthetic pathway of purine nucleotides. GMPS consists of two functional units that are present as domains or subunits: glutamine amidotransferase (GATase) and ATP pyrophosphatase (ATPPase). GATase hydrolyzes glutamine to yield glutamate and ammonia, while ATPPase utilizes ammonia to convert adenyl xanthosine 5′-monophosphate (adenyl-XMP) into guanosine 5′-monophosphate. Here we report the crystal structure of PH-ATPPase (the ATPPase subunit of the two-subunit-type GMPS from the hyperthermophilic archaeon Pyrococcus horikoshii OT3). PH-ATPPase consists of two domains (N-domain and C-domain) and exists as a homodimer in the crystal and in solution. The N-domain contains an ATP-binding platform called P-loop, whereas the C-domain contains the xanthosine 5'-monophosphate (XMP)-binding site and also contributes to homodimerization. We have also demonstrated that PH-GATase (the glutamine amidotransferase subunit of the two-subunit-type GMPS from the hyperthermophilic archaeon P. horikoshii OT3) alone is inactive, and that all substrates of PH-ATPPase except for ammonia (Mg²⁺, ATP and XMP) are required to stabilize the active complex of PH-ATPPase and PH-GATase subunits.

Full text of DOI:10.1016/j.jmb.2009.10.053

doi:10.1016/j.jmb.2009.10.053
J. Mol. Biol. (2010) 395, 417–429
Available online at www.sciencedirect.com
Crystal Structure of the ATPPase Subunit and Its
Substrate-Dependent Association with the GATase
Subunit: A Novel Regulatory Mechanism for a
Two-Subunit-Type GMP Synthetase from
Pyrococcus horikoshii OT3
Shintaro Maruoka, Shoichiro Horita, Woo Cheol Lee,
Koji Nagata and Masaru Tanokura
⁎
Department of Applied
Guanosine 5′-monophosphate synthetase(s) (GMPS) catalyzes the final step
of the de novo synthetic pathway of purine nucleotides. GMPS consists of
two functional units that are present as domains or subunits: glutamine
amidotransferase (GATase) and ATP pyrophosphatase (ATPPase). GATase
hydrolyzes glutamine to yield glutamate and ammonia, while ATPPase
utilizes ammonia to convert adenyl xanthosine 5′-monophosphate (adenyl-
XMP) into guanosine 5′-monophosphate. Here we report the crystal
structure of PH-ATPPase (the ATPPase subunit of the two-subunit-type
GMPS from the hyperthermophilic archaeon Pyrococcus horikoshii OT3).
PH-ATPPase consists of two domains (N-domain and C-domain) and exists
as a homodimer in the crystal and in solution. The N-domain contains an
ATP-binding platform called P-loop, whereas the C-domain contains the
xanthosine 5'-monophosphate (XMP)-binding site and also contributes to
homodimerization. We have also demonstrated that PH-GATase (the
glutamine amidotransferase subunit of the two-subunit-type GMPS from
the hyperthermophilic archaeon P. horikoshii OT3) alone is inactive, and
that all substrates of PH-ATPPase except for ammonia (Mg2+, ATP and
XMP) are required to stabilize the active complex of PH-ATPPase and
PH-GATase subunits.
Biological Chemistry, Graduate
School of Agricultural and Life
Sciences, University of Tokyo,
1-1-1 Yayoi, Bunkyo-ku, Tokyo
113-8657, Japan
Received 9 July 2009;
received in revised form
17 October 2009;
accepted 23 October 2009
Available online
10 November 2009
© 2009 Elsevier Ltd. All rights reserved.
Keywords: ATP pyrophosphatase; crystal structure; glutamine amidotransferase;
GMPS; purine nucleotide biosynthesis
Edited by G. Schulz
sine 5′-monophosphate (GMP), a precursor of DNA
and RNA. GMPS is also involved in chromatin
regulation1 and axon guidance,2 as shown in
Drosophila. In addition, GMPS has been studied as
an important chemotherapeutic target for immuno-
suppressive agents.3
GMPS belongs to the class I glutamine amidotrans-
ferase (GATase) family and consists of two catalytic
units, GATase and ATP pyrophosphatase (ATPPase):
Introduction
Guanosine 5′-monophosphate synthetase(s)
(GMPS) is a widespread enzyme seen in all domains
of life. GMPS is required for the final step of the de
novo synthesis of guanine nucleotides, converting
xanthosine 5′-monophosphate (XMP) into guano-
*Corresponding author. E-mail address:
amtanok@mail.ecc.u-tokyo.ac.jp.
Abbreviations used: GMPS, guanosine
GATase : L glutamine þ H2O →
ꢀ
L glutamate þ NH3
ð1Þ
ð2Þ
ꢀ
5′-monophosphate synthetase(s); GATase, glutamine
amidotransferase; ATPPase, ATP pyrophosphatase; XMP,
xanthosine 5′-monophosphate; GMP, guanosine 5′-
monophosphate; PDB, Protein Data Bank; GDH, glutamate
dehydrogenase; EDTA, ethylenediaminetetraacetic acid;
adenyl-XMP, adenyl xanthosine 5′-monophosphate.
ATPPase : XMP þ MgQATP →
adenyl XMP intermediate
ꢀ
þ pyrophosphate þ Mg2þ
0022-2836/$ - see front matter © 2009 Elsevier Ltd. All rights reserved.

418
Two-Subunit-Type GMP Synthetase
and Pyrococcus horikoshii (PDB ID 1WL8; Lokanath
and Kunishima, unpublished results) (PDB ID 2D7J;
Maruoka et al.22), and the ATPPase subunit from P.
horikoshii (PDB ID 2DPL; Asada and Kunishima,
unpublished results) (PDB ID 3A4I; this study)—
have been deposited in the PDB. In the crystal
structures of two-domain-type GMPS, the two active
sites of the GATase and ATPPase domains are
exposed to the solvent, and it has not been revealed
yet how the putative ammonia channel, which
efficiently couples the GATase and ATPPase reac-
tions, is formed.21
To gain insights into the catalytic and regulatory
mechanisms of GMPS, particularly two-subunit-
type GMPS, we have determined the crystal struc-
tures of the GATase subunit [PH-GATase (the
glutamine amidotransferase subunit of the two-
subunit-type GMPS from the hyperthermophilic
archaeon P. horikoshii OT3); PDB ID 2D7J]22 and the
ATPPase subunit [PH-ATPPase (the ATPPase sub-
unit of the two-subunit-type GMPS from the
hyperthermophilic archaeon P. horikoshii OT3); this
study] of PH-GMPS (the two-subunit-type GMPS
from the hyperthermophilic archaeon P. horikoshii
OT3) and analyzed the regulatory mechanisms of
PH-GMPS. Our crystal structure has revealed that
the ATPPase subunit of PH-GMPS (PH-ATPPase) is
very similar to the ATPPase domains of EC-GMPS
and TT-GMPS in terms of backbone fold, ATP-
binding and XMP-binding residues, and dimeriza-
tion mode, but is dissimilar with respect to the
relative orientation of N-domain and C-domain.
PH-ATPPase is also similar to the ATPPase domain
of HS-GMPS in terms of backbone fold and ligand-
binding residues, but is dissimilar in oligomerization
mode because HS-GMPS exists as a monomer
in solution.12 We also performed activity assays
adenyl XMP intermediate þ NH3 → GMP þ AMP
ꢀ
ð3Þ
GATase hydrolyzes glutamine to generate gluta-
mate and ammonia (Reaction (1)), and ATPPase
adenylates XMP to form an adenyl xanthosine
5′-monophosphate (adenyl-XMP) intermediate in
the presence of Mg2+, ATP, and XMP (Reaction
(2)). The highly reactive adenyl-XMP intermediate
is then aminated by ammonia provided by GATase
(Reaction (3)).4–6 Overall, GMP is generated by the
concerted reactions of these two catalytic units.7,8
The two catalytic units are either encoded by a
single gene (two-domain type) in eukaryotes, bacte-
ria, and some archaea, or encoded by two separate
genes (two-subunit type) in other archaea. In two-
domain-type GMPS, the GATase domain is located
in the N-terminal half, and the ATPPase domain is
located in the C-terminal half; in two-subunit-type
GMPS, these two units exist as separate polypeptides
(Fig. 1a). Almost all structural and functional studies
on GMPS have been carried out for two-domain-
type GMPS from humans,11–13 rats,14 mice,15 Escher-
ichia coli,6,16–19 and Plasmodium falciparum.20 So far,
the crystal structures of three two-domain-type
GMPS—EC-GMPS (the two-domain-type GMPS
from E. coli) [Protein Data Bank (PDB) ID 1GPM],21
TT-GMPS (the two-domain-type GMPS from Ther-
mus thermophilus) (PDB IDs 2YWB and 2YWC; Baba
et al., unpublished results), and HS-GMPS (the two-
domain-type GMPS from Homo sapiens) (PDB ID
2VXO; Welin et al., unpublished results)—and two
two-subunit-type GMPS—the GATase subunit from
Thermoplasma acidophilum (PDB ID 2A9V; Joint
Center for Structural Genomics, unpublished results)
Fig. 1. (a) Gene and protein structure of GMPS from P. horikoshii and E. coli. P. horikoshii GMPS is encoded by two
genes, ph1346 (encoding GATase) and ph1347 (encoding ATPPase), located 7 bp apart. E. coli GMPS is encoded by a single
gene guaA, which encodes the GATase domain in the N-terminal half and the ATPPase domain in the C-terminal half. (b)
Multiple sequence alignment of ATPPase subunits and domains. The proteins are derived from P. horikoshii, T.
thermophilus, and E. coli. The secondary structure elements of PH-ATPPase are shown above the alignment. Sequence
alignment was obtained with CLUSTAL W.9 The figure was prepared with ESPript.10 α-Helices, 310-helices, and β-strands
are denoted as α, η, and β, respectively. Completely and highly conserved residues are boxed and indicated by white
letters on a red background and red letters on a white background, respectively. Blue and green triangles indicate the
putative binding residues for ATP and XMP, respectively. Red triangles indicate residues forming intermolecular
hydrogen bonds in homodimers of PH-ATPPase, EC-GMPS, and TT-GMPS.

Two-Subunit-Type GMP Synthetase
419
Fig. 1 (legend on previous page)
and gel-filtration experiments for PH-GATase and
PH-ATPPase under various ligand conditions. The
results indicate that PH-GATase is inactive when
alone and becomes active upon association with PH-
ATPPase, and that all the substrates of PH-ATPPase
except for ammonia (Mg-ATP and XMP) are
required to stabilize the active PH-GMPS complex
formed by PH-GATase and PH-ATPPase, suggest-
ing the pivotal role of the adenyl-XMP intermediate
in promoting active complex formation.
Results and Discussion
Crystal structure of PH-ATPPase
The crystal structure of the 34-kDa PH-ATPPase,
which consists of 308 amino acid residues, was
solved at 1.80 Å resolution (Figs. 2 and 3a; PDB ID
3A4I). Data collection and refinement statistics are
given in Table 1. The structure was refined to Rwork

420
Two-Subunit-Type GMP Synthetase
Fig. 2. Stereoview of the PH-ATPPase protomer (molecule A). The termini and secondary structures are labeled. N-
domains and C-domains are shown in green and blue, respectively. The Pro-rich interdomain loop is shown in pink. The
22-residue disordered peptide is depicted with an orange dotted line. The figure was generated using the program
PyMOL.
and Rfree values of 21.0% and 24.6%, respectively.
There are two protomers (referred to as chains A and
B) of PH-ATPPase that form a homodimer in the
asymmetric unit (Fig. 3a). The final model of PH-
ATPPase consists of residues 1–131 and 154–308
(286 of 308 residues, covering 92.9% of the sequence)
for both chains A and B. No electron densities are
observed for the remaining residues. The Rama-
chandran plot shows that all the φ–ψ pairs lie in the
most favored and additionally allowed regions
(Table 1). The two protomers in the PH-ATPPase
homodimer are very similar to each other, as shown
by the small root-mean-square deviation (RMSD) of
0.34 Å for 263 Cα atom pairs. Major structural
differences between them are caused by crystal
contacts and are located in two regions: residues
240–248 and 261–263.
The protomer of PH-ATPPase consists of two
domains: the N-domain (residues 1–186) and the C-
domain (residues 187–308) (Fig. 2). The larger N-
domain is composed of a five-stranded parallel β-
sheet and α-helical layers on both sides of the sheet.
Electron densities are not observed for 22 continu-
ous residues (residues 132–153) located between β4
and β5 (Fig. 2). The N-domain contains a glycine-
rich ATP-binding motif called the “P-loop motif”
(residues 28–34; SGGVDSS) located after β1. This P-
loop motif is strictly conserved in the ATPPase
subunits/domains of GMPS (Fig. 1b); thus, the N-
domain is considered to be responsible for ATP
binding. The C-domain, consisting of three α-helices
and five β-strands, serves as the dimerization
domain. The dimer interface is mainly formed by
the α10 helix and a three-stranded parallel β-sheet
consisting of β7, β8, and β9.
In the dimer interface, approximately 11% (1500 Å2)
of the surface area per protomer is buried, as indicated
by the PISA server.23 The gel-filtration chromatogram
of PH-ATPPase indicates that the protein also behaves
as a homodimer (ca 70 kDa) in solution (Fig. 7). The
dimer interface contains 28 hydrogen bonds and eight
salt bridges, which contribute to homodimerization.
In particular, two cis-prolines (Pro301-Pro302) located
after β10 enable the formation of a tetrad of Asp296,
Thr304, Glu306, and Arg292′ (the prime denotes a
residue from the other subunit) whose side chains
contribute to homodimerization by electrostatic inter-
actions. The residues Arg292, Val293, Tyr295, Asp296,
and Thr304, which form intermolecular hydrogen
bonds, are conserved in PH-ATPPase, TT-GMPS, and
EC-GMPS (Fig. 4). These residues are also conserved
in human GMPS (referred to as HS-GMPS), with the
exception of Val293 (Ile in HS-GMPS), but they are not
involved in dimerization in the crystal or in solution.
HS-GMPS exists as a homodimer in the crystal (PDB
ID 2VXO), but the mode of dimerization is different
from those of PH-ATPPase, EC-GMPS, and TT-GMPS
(Fig. 4). On the other hand, HS-GMPS exists as a
monomer in solution, as shown by density gradient
centrifugation.12
In addition to our crystal structure (PDB ID
3A4I), another crystal structure of PH-ATPPase at
1.43 Å resolution was deposited in the PDB (PDB
ID 2DPL) by Asada et al. (unpublished results).

Two-Subunit-Type GMP Synthetase
421
Fig. 3. The overall structure of the PH-ATPPase homodimer (a) compared to those of the TT-GMPS homodimer (b),
the EC-GMPS homodimer (c), and the HS-GMPS homodimer (d). Each protomer is shown in yellow and orange. Each
active-site triad of the GATase domain is shown with a blue sphere. The ATP-binding motif, called the “P-loop motif,” is
indicated with a green sphere.

422
Two-Subunit-Type GMP Synthetase
Table 1. Data collection and refinement statistics
GMPS, and a difference of 5° between PH-ATPPase
and TT-GMPS. This suggests the presence of a hinge
motion between the N-domain and the C-domain
(Fig. 5; Figs. S1 and S2, and Tables S1 and S2), as also
suggested by the reaction mechanism of ATPPase
(described later in this section). The relative orienta-
tions of the N-domain and C-domain of PH-
ATPPase are almost the same among the four chains
of PDB IDs 3A4I and 2DPL (Table S3), suggesting
that domain orientation would be most prevalent in
unliganded PH-ATPPase, although the crystal pack-
ing effect cannot be denied.
Data collection
Space group
Unit cell parameters
a (Å)
P212121
69.9
82.2
111.6
b (Å)
c (Å)
Resolutiona (Å)
Observed reflections
Unique reflections
Completenessa (%)
50.0–1.79 (1.85–1.79)
537,148
60,634
98.4 (100.0)
0.052 (0.076)
28.1 (3.3)
a
Rmerge
〈I〉/〈σ(I)〉a
Since the crystal structures of EC-GMPS (PDB ID
1GPM)21,24 and TT-GMPS (PDB ID 2YWC)
contained bound AMP and PPi21 and bound XMP,
respectively, the putative ATP-binding site and
XMP-binding site of PH-ATPPase can be readily
identified by superimposing the structures. The
putative ATP-binding site is located in the N-
domain, and all the putative ATP-recognizing
amino acid residues (Fig. 6a) are conserved between
PH-ATPPase and EC-GMPS, except for Ala26/
Gly233 and Ile125/Ala333, whose main-chain
atoms are involved in ATP recognition. Thus, PH-
ATPPase would probably recognize ATP as EC-
GMPS recognizes AMP and PPi in the crystal
structure. The purine base, ribose ring, and triphos-
phate of ATP would be recognized by Val53 and
Phe108; Ala26-Leu27 and Ile125; and Ser28, Asp32-
Ser33, Lys166, and Phe188, respectively. On the
other hand, all residues of the putative XMP-binding
site are located in the C-domain, except for Arg101,
and all of the putative XMP-recognizing amino acid
residues (Fig. 6b) are conserved between PH-
ATPPase and TT-GMPS. Thus, PH-ATPPase would
probably recognize XMP as does TT-GMPS. The
purine base, ribose ring, and phosphate of XMP
would be recognized by Arg101 and Pro189-Gly190-
Pro-191; Gln229 and Glu306; and Lys300, Ile305-
Gln306, and Glu308, respectively.
Refinement
Resolution range (Å)
Rwork (%)
50.0–1.79
21.0
Rfree (%)
24.6
Average B-factors of protein
21.4
heavy atoms (Å2)
Number of protein heavy atoms
Number of water molecules
RMSD from ideal geometry
Bond lengths (Å)
4528
282
0.015
1.361
Bond angles (°)
Ramachandran plot
Most favored region (%)
Additionally favored regions (%)
95.1
4.9
a
The data of the highest-resolution shell are shown in
parentheses.
The space group and unit cell dimensions of our
crystals and those of Asada et al. are essentially the
same: PDB ID 3A4I, P212121, a=69.9 Å, b=82.2 Å,
and c=111.6 Å; PDB ID 2DPL, P212121, a=70.4 Å,
b=83.3 Å, and c=112.3 Å. Moreover, our crystals
and their crystals were obtained at very similar pH
values, although the temperatures were different:
PDB ID 3A4I at pH 8.4 and 278 K, and PDB ID
2DPL at pH 8.5 and 291 K. Since the two kinds of
crystals obtained at very similar pH values belong
to the same space group and have almost the same
unit cell dimensions, the two crystal structures of
PH-ATPPase are essentially the same, with a small
RMSD of 0.446 Å for 564 Cα atom pairs in the
homodimer between PDB IDs 3A4I and 2DPL. We
use our crystal structure (PDB ID 3A4I) for the
following discussions.
We have modeled bound ATP and XMP on the
crystal structures of PH-ATPPase, EC-GMPS, and
PH-GMPS. The two atoms Pα of ATP and O2 of
XMP, which will form a covalent bond (1.5 Å) to
yield the adenyl-XMP intermediate,4 are separated
by 3.3 Å, 4.4 Å, and 5.5 Å in the crystal structures of
TT-GMPS, PH-ATPPase, and EC-GMPS, respective-
ly. These observations suggest that further move-
ment between the N-domain and the C-domain has
to occur for the reaction catalyzed by the ATPPase
subunits/domains to proceed.
Structural comparison of PH-ATPPase with
ATPPase domains in two-domain-type GMPS
The structure of PH-ATPPase was compared with
the reported structures of ATPPase domains in the
two-domain-type GMPS from E. coli (PDB ID
1GPM)21,24 and T. thermophilus (PDB ID 2YWC;
Baba et al., unpublished results). PH-ATPPase is
structurally similar to the corresponding domains of
EC-GMPS and TT-GMPS (55% and 53% identical in
sequence), with RMSDs of 1.04 Å and 1.24 Å (for 264
and 240 Cα atom pairs), respectively. Remarkable
differences, however, are observed in the relative
orientation between the N-domain and the C-
domain among these ATPPase subunits/domains:
a difference of 8° between PH-ATPPase and EC-
Substrate-induced stable complex formation of
PH-ATPPase with PH-GATase
To examine the effects of the substrates of PH-
ATPPase on the interaction between PH-GATase
and PH-ATPPase, we performed gel-filtration chro-
matography with a Superdex 75 10/30 HR column
(GE Healthcare) at room temperature (Fig. 7). The
two subunits form a stable complex exclusively in
the presence of 10 mM MgCl2, 0.1 mM ATP, and
0.1 mM XMP (buffer f). The stable complex

Two-Subunit-Type GMP Synthetase
423
Fig. 4 (legend on next page)

424
Two-Subunit-Type GMP Synthetase
using a coupled assay with glutamate dehydroge-
nase (GDH) in the presence or in the absence of
PH-ATPPase and its substrates, which are required
for the complex formation (Mg2+, ATP, and XMP)
(Fig. 8a). The results revealed that PH-GATase
alone does not show any measurable glutaminase
activity (Condition (2)). PH-GATase is activated by
the copresence of PH-ATPPase (Condition (4)).
Moreover, the activity of PH-GATase is enhanced
approximately fivefold by the copresence of PH-
ATPPase and its substrates (Mg2+, ATP, and XMP)
(Condition (5)). Because these PH-ATPPase sub-
strates show no direct effect on PH-GATase
(Condition (3)), they are considered to enhance
PH-GATase activity by stabilizing the PH-GMPS
complex, as shown in Fig. 7. Assuming that PH-
GMPS complex formation and PH-GATase activa-
tion are coupled, the results of the GATase activity
assay suggest that the substrates of PH-ATPPase
are not necessarily required for complex formation
and that the PH-GMPS complex without all
ATPPase ligands would be unstable. Thus, PH-
GATase consumes much less glutamine, unless all
the components required for GMPS activity are
present. This regulation mechanism is effective in
avoiding glutamine wasting.
A similar phenomenon has been observed for EC-
GMPS, a two-domain-type GMPS; that is, the GATase
activity of EC-GMPS depends on the presence of the
substrates for ATPPase (Mg2+, ATP, and XMP),25 but
the molecular mechanism for the substrate depen-
dency of EC-GMPS remains unknown. Here we show
that the substrates for ATPPase promote the associ-
ation of the GATase and ATPPase subunits of PH-
GMPS and also activate PH-GATase. Based on the
high sequence similarity of these two subunits (50%
identity), the same regulatory mechanism would
likely be used in EC-GMPS.
Fig. 5. Structural comparisons of PH-ATPPase, EC-
GMPS (PDB ID 1GPM), and TT-GMPS (PDB ID 2YWC).
The C-terminal domain of PH-ATPPase is superimposed
onto the C-terminal domain of EC-GMPS (cyan) com-
plexed with AMP (cyan), and onto the C-terminal domain
of TT-GMPS (green) complexed with XMP (magenta). The
superimposed C-terminal domains are transparent to
better visualize the bent conformations of the N-terminal
domains. α-Helices are depicted as cylinders, and β-
strands are depicted as arrows. The AMPs bound to PH-
ATPPase and TT-GMPS (depicted as yellow and green
sticks, respectively) are modeled based on the superim-
position of the N-terminal domains of the three proteins.
formation, however, is not observed when any one
of the ATPPase substrates is omitted from the
buffer (buffers a–e). The results clearly show that all
the substrates of PH-ATPPase except for ammonia
(Mg2+, ATP and XMP) are required for stabilizing
the PH-GMPS complex formed by PH-GATase and
PH-ATPPase subunits. The ATP cannot be replaced
by AMP and PPi (buffer g). Since Mg-ATP and
XMP are required to yield the adenyl-XMP inter-
mediate, these results suggest that the formation of
the adenyl-XMP intermediate would be important
in stabilizing the PH-GMPS complex.
Effect of ammonia on PH-ATPPase activity
To examine the effect of ammonia on the activity of
PH-ATPPase, we measured ATPPase activity (Fig.
8b). PH-ATPPase is capable of utilizing ammonia
derived from ammonium chloride (Condition (2)), as
well as ammonia derived from glutamine by PH-
GATase in the manner of EC-GMPS (Condition (1)).26
The copresence of ammonium chloride and gluta-
mine does not promote ATPPase activity additively,
but decreases ATPPase activity to a level similar to
that observed in Condition (1) (Condition (3)). Thus,
the glutamine-dependent and ammonium-chloride-
dependent reactions are not independent. Interest-
ingly, the ATPPase activity increases approximately
1.5-fold when PH-GATase is omitted from the
reaction mixture of Condition (3) (Condition (4)).
This comparison (Conditions (3) and (4)) clearly
Effect of PH-GMPS complex formation on
PH-GATase activity
To examine the effect of PH-GMPS complex
formation on the activity of PH-GATase, we
measured the glutaminase activity of PH-GATase
Fig. 4. The dimer interface of PH-ATPPase (a), TT-GMPS (b), and EC-GMPS (c). The amino acid residues involved in
hydrogen bonds between protomers are represented by sticks and labeled. Hydrogen bonds are shown by dotted lines.
The color scheme is the same as that in Fig. 3. (i) The dimer interface of PH-ATPPase. (ii) The dimer interface of TT-GMPS.
(iii) The dimer interface of EC-GMPS. The conserved residues are shown in magenta.

Two-Subunit-Type GMP Synthetase
425
Fig. 6. Putative ligand recognition by PH-ATPPase. (a) Structural superposition of the ATP-binding site of PH-
ATPPase (yellow) and E. coli GMPS (blue). AMP (magenta) and PPi (orange) from E. coli GMPS are also depicted with
stick models. (b) Structural superposition of the XMP-binding site of PH-ATPPase (yellow) and T. thermophilus GMPS
(blue). XMP (light magenta) from T. thermophilus GMPS is depicted with a stick model.
shows that the complex formation by PH-GATase
and PH-ATPPase partly blocks the direct access of
ammonium-chloride-derived ammonia to the active
site of the PH-ATPPase and suggests the existence of
an ammonia channel that prevents glutamine-derived
ammonia from being released into the free ammonia
pool prior to incorporation into XMP. Moreover, the
results (Conditions (1), (2), and (3)) suggest that
glutamine binding to GATase may complement the
full shielding of the ammonia channel of the GMPS
complex.
archaea. The distribution of the two-subunit-type
GMPS in archaea, however, is not correlated with
the phylogenetic tree based on ribosomal proteins27
because both two-domain-type GMPS and two-
subunit-type GMPS are distributed in Euryarchaeota
and Crenarchaeota. Two-subunit-type archaeal
GMPS can be divided into two groups: one group
in which the genes encoding GATase and ATPPase
subunits are colocalized (Pyrococcus, Thermococcus,
Methanobacterium, Methanobrevibacter, and Methanos-
paera) and the other group in which the two genes are
located apart. The separate GMPS genes may
contribute to the compact genome size of many
archaeal species by enabling the GATase subunit to
be shared by various enzymes that utilize ammonia
as substrate.
Two-subunit-type GMPS and two-domain-type
GMPS
To examine the distribution of the two-subunit-
type and two-domain-type GMPS, we collected
DNA sequence information for 565 GMPS in the
three domains—41 in archaea, 482 in bacteria, and
42 in eukarya—from the KEGG database†. All
GMPS from eukarya and bacteria are two-domain-
type GMPS consisting of GATase and ATPPase
domains. In contrast, the GMPS from archaea
include 6 two-domain-type GMPS and 34 two-
subunit-type GMPS, indicating that two-subunit-
type GMPS are unique and major properties of
Materials and Methods
Expression and purification of PH-ATPPase for
structure determination
The gene encoding PH-ATPPase (ph1347) was amplified
by PCR using the genomic DNA of P. horikoshii OT3 as
template. The PCR product was digested with NdeI and
EcoRI, and ligated into the expression vector pET-26b(+)
(Novagen). The E. coli strain Rosetta(DE3) (Novagen) was
used as the host for protein expression. The cells were
† http://www.kegg.jp/kegg/kegg2.html

426
Two-Subunit-Type GMP Synthetase
Fig. 7. Substrate-dependent complex formation of PH-ATPPase with PH-GATase. The reaction buffers used
contained: no substrate (a); 10 mM MgCl2 (b); 10 mM MgCl2 and 0.1 mM ATP (c); 10 mM MgCl2 and 0.1 mM XMP (d);
1 mM EDTA, 0.1 mM ATP, and 0.1 mM XMP (e); 10 mM MgCl2, 0.1 mM ATP, and 0.1 mM XMP (f); and 10 mM MgCl2,
0.1 mM AMP, 0.1 mM XMP, and PPi (g). The fractions obtained by gel-filtration chromatography with a Superdex 75 10/
30 HR column were subjected to SDS-PAGE. PH-GATase (monomer) and PH-ATPPase (dimer) were eluted at elution
volumes of 9.6 ml and 12.1 ml, respectively. Their complex was eluted at 9.2 ml. Aldolase (158k), conalbumin (75k), and
carbonic anhydrase (29k) were used as markers.
grown in LB medium containing 30 μg/ml kanamycin at
310 K. Protein expression was induced with 1 mM IPTG.
After 12 h, the cells were harvested by centrifugation at
5000g and resuspended in 20 mM Tris–HCl (pH 8.0). The
cells were disrupted by sonication and centrifuged at
40,000g for 30 min. The supernatant was heated at 353 K for
30 min and centrifuged at 40,000g for 30 min. Then, 0.01%
(final concentration) polyethyleneimine was added to the
supernatant and stirred for 30 min on ice. After centrifu-
gation at 40,000g for 30 min, the supernatant was loaded
onto a RESOURCE Q 6-ml column (GE Healthcare)
equilibrated with 20 mM Tris–HCl (pH 8.0). The GMPS
subunit was eluted at 80–100 mM NaCl in 20 mM Tris–HCl
(pH 8.0). Then the fractions containing the GMPS subunit
were pooled, concentrated, dialyzed against 20 mM Tris–
HCl (pH 8.0) and 100 mM NaCl, and loaded onto a
Superdex 75 10/30 column (GE Healthcare) equilibrated
with 20 mM Tris–HCl (pH 8.0) and 100 mM NaCl. The

Two-Subunit-Type GMP Synthetase
427
with unit cell dimensions a=69.9 Å, b=82.2 Å, and c=
111.6 Å, and an asymmetric unit contained two molecules
of PH-ATPPase.
Structure solution and refinement of PH-ATPPase
The structure of PH-ATPPase was determined by
molecular replacement with the program MOLREP,29
using the atomic coordinates of the ATPPase domain of
the GMPS from E. coli (PDB ID 1GPM) as initial model.
Further model building was performed with the program
Xfit,30 and refinements were performed with XtalView30
and Refmac5.31 Solvent atoms were initially built using
the program ARP/wARP32 and later added or removed
by manual inspection. The refined structure was validated
with PROCHECK33 and visualized with PyMOL‡.
Expression and purification of PH-GATase for activity
assay and gel-filtration analysis
PH-GATase was prepared as previously described,22
with the following modifications. To obtain a tagless PH-
GATase, we cloned the gene ph1346 into pET26b (not
pET28a) using the NdeI and BamHI sites. E. coli strain
BL21(DE3) was used as the host for protein expression.
Gel-filtration chromatography
Fig. 8. Activity assays of PH-GATase and PH-ATP-
Pase. (a) GATase activity was measured in a coupled
enzymatic assay with GDH. The base reaction mixtures
(Condition (1)) contained 60 mM Hepes buffer (pH 8.0),
0.8 mM EDTA, 20 mM Gln, 1 mM NADP+, and 10 nM
GDH. A concentration of 1 μM was used for the enzyme(s)
(PH-GATase and/or PH-ATPPase), and the following
concentrations were used for the ATPPase substrates:
1 mM ATP, 0.3 mM XMP, and 20 mM MgCl2. (b) ATPPase
activity was measured using glutamine and ammonium
chloride as donors of ammonia. The following concentra-
tions were used: 2 μM enzyme(s) (PH-ATPPase and/or
PH-GATase) and 20 mM ammonia donor(s) (glutamine
and/or ammonium chloride).
To analyze the interaction of PH-ATPPase and PH-
GATase, we performed gel-filtration chromatography
under five buffer conditions. Each buffer contained
50 mM Tris–HCl (pH 8.0) and 100 mM NaCl, with the
addition of the following substrates:
buffer a: no substrate
buffer b: 10 mM MgCl2
buffer c: 10 mM MgCl2 and 0.1 mM ATP
buffer d: 10 mM MgCl2 and 0.1 mM XMP
buffer e: 1 mM ethylenediaminetetraacetic acid
(EDTA), 0.1 mM ATP, and 0.1 mM XMP
buffer f: 10 mM MgCl2, 0.1 mM ATP, and 0.1 mM XMP
buffer g: 10 mM MgCl2, 0.1 mM AMP, 0.1 mM XMP,
and 0.1 mM PPi
The mixture of 0.01 mM PH-GATase and 0.01 mM PH-
ATPPase dissolved in five kinds of buffers was loaded
onto a Superdex 75 10/30 column (GE Healthcare) at a
flow rate of 0.5 ml/min at room temperature. The 0.5-ml
fractions were collected and analyzed by 15% SDS-PAGE.
fractions containing the GMPS subunit were pooled,
dialyzed against 5 mM Tris–HCl (pH 8.0), and concentrat-
ed to 30 mg/ml with a Centriprep-3 concentrator
(Amicon). Purity was ascertained by SDS-PAGE, and
concentration was determined by measuring absorbance at
280 nm with an extinction coefficient of 30,940 M− 1 cm−1
.
GATase activity assay
Crystallization and data collection of PH-ATPPase
The glutaminase activity of PH-GATase was measured
in a coupled enzymatic assay with GDH (Wako, Japan).
The glutamate produced by PH-GATase was oxidized
by GDH in the presence of NADP+, yielding 2-oxo-
glutarate and NADPH+H+ +NH3. The reaction was
monitored by the increase in absorbance at 340 nm
using the equation: Δɛ340 [NADPH−NADP+]=6220 M−1
cm−1. The reaction mixtures contained 60 mM Hepes
buffer (pH 8.0), 0.8 mM EDTA, 20 mM Gln, 1 mM
NADP+, and 10 nM GDH. The substrates of PH-
Crystallization was performed at 278 K using sitting-
drop vapor diffusion. Crystals of PH-ATPPase (0.4 mm×
0.3 mm×0.1 mm) were obtained in 3 weeks by mixing 1 μl
of the protein solution and 1 μl of the reservoir solution
containing 30% (vol/vol) polyethylene glycol 400, 100 mM
Tris–HCl (pH 8.4), and 200 mM MgCl2. A drop was
equilibrated against 100 μl of the reservoir solution. The
crystals were flash-cooled to 100 K with a nitrogen
cryostream. Diffraction data were collected at beamline
NW12 of the Photon Factory Advanced Ring (Tsukuba,
Japan). The diffraction data were indexed and scaled with
HKL2000.28 The crystal belonged to space group P212121
‡ http://pymol.sourceforge.net/

428
Two-Subunit-Type GMP Synthetase
ATPPase (1 mM ATP, 0.3 mM XMP, and 20 mM MgCl2)
were either added or omitted from the reaction mixture.
The enzyme(s), 1 μM PH-ATPPase, 1 μM PH-GATase, or
both 1 μM PH-ATPPase and 1 μM PH-GATase (final
concentration(s)) were added to start the reaction. All
measurements were performed in triplicate.
2. Long, H., Cameron, S., Yu, L. & Rao, Y. (2006). De novo
GMP synthesis is required for axon guidance in
Drosophila. Genetics, 172, 1633–1642.
3. Ishikawa, H. (1999). Mizoribine and mycophenolate
mofetil. Curr. Med. Chem. 6, 575–597.
4. Lagerkvist, U. (1958). Biosynthesis of guanosine 5′-
phosphate. J. Biol. Chem. 233, 143–149.
5. Fukuyama, T. & Moyed, H. S. (1964). A separate
antibiotic-binding site in xanthosine-5′-phosphate
aminase: inhibitor- and substrate-binding studies.
Biochemistry, 3, 1488–1492.
6. von der Saal, W., Crysler, C. S. & Villafranca, J. J.
(1985). Positional isotope exchange and kinetic experi-
ments with Escherichia coli guanosine-5′-monopho-
sphate synthetase. Biochemistry, 24, 5343–5350.
7. Raushel, F. M., Thoden, J. B. & Holden, H. M. (1999).
The amidotransferase family of enzymes: molecular
machines for the production and delivery of ammo-
nia. Biochemistry, 38, 7891–7899.
8. Zalkin, H. & Smith, J. L. (1998). Enzymes utilizing
glutamine as an amide donor. Adv. Enzymol. Relat.
Areas Mol. Biol. 72, 87–144.
9. Thompson, J. D., Higgins, D. G. & Gibson, T. J.
(1994). CLUSTAL W: improving the sensitivity of
progressive multiple sequence alignment through
sequence weighting, position-specific gap penalties
and weight matrix choice. Nucleic Acids Res. 22,
4673–4680.
ATPPase activity assay
The assays for the ATPPase activity were performed
using glutamine and ammonium chloride as donors of
ammonia, in accordance with a previously described
method,26,34 with some modifications. To a reaction
mixture containing 2 μM PH-ATPPase, 2 μM PH-GATase,
60 mM Hepes buffer (pH 8.0), 1 mM ATP, 0.3 mM XMP,
20 mM MgCl2, 0.1 mM DTT, 0.8 mM EDTA, and the
ammonia donor(s) (20 mM glutamine, 20 mM ammonium
chloride, or both) were added to start the reaction. The
reaction mixture was incubated at 323 K for 3 min. GMPS
activity was monitored by a decrease in absorbance at
290 nm: Δɛ290 [XMP−(GMP+AMP)]=1500 M−1 cm−1. To
examine the effect of PH-GATase on the activity of PH-
ATPPase, we also performed the assays without PH-
GATase. All measurements were performed in triplicate.
PDB accession number
10. Gouet, P., Courcelle, E., Stuart, D. I. & Métoz, F.
(1999). ESPript: analysis of multiple sequence align-
ments in PostScript. Bioinformatics, 15, 305–308.
11. Lou, L., Nakamura, J., Tsing, S., Nguyen, B., Chow, J.,
Straub, K. et al. (1995). High-level production from a
baculovirus expression system and biochemical char-
acterization of human GMP synthetase. Protein
Expression Purif. 6, 487–495.
Coordinates and structural factors for the structure of
PH-ATPPase have been deposited with the Research
Collaboratory for Structural Bioinformatics PDB with
accession code 3A4I.
12. Nakamura, J. & Lou, L. (1995). Biochemical charac-
terization of human GMP synthetase. J. Biol. Chem.
270, 7347–7353.
Acknowledgements
13. Nakamura, J., Straub, K., Wu, J. & Lou, L. (1995). The
glutamine hydrolysis function of human GMP syn-
thetase. Identification of an essential active site
cysteine. J. Biol. Chem. 270, 23450–23455.
14. Hirai, K., Matsuda, Y. & Nakagawa, H. (1987).
Purification and characterization of GMP synthetase
from Yoshida sarcoma ascites cells. J. Biochem. 102,
893–902.
15. Spector, T. (1975). Studies with GMP synthetase from
Ehrlich ascites cells. Purification, properties, and
interactions with nucleotide analogs. J. Biol. Chem.
250, 7372–7376.
16. Sakamoto, N., Hatfield, G. W. & Moyed, H. S. (1972).
Physical properties and subunit structure of xantho-
sine 5′-phosphate aminase. J. Biol. Chem. 247,
5880–5887.
We thank the beamline staff at the Photon Factory
for their kind help with data collection. Synchrotron
radiation experiments were performed with the
approval of the Photon Factory, KEK (2003G130,
2003S2-002, and 2008S2-001). This work was sup-
ported, in part, by grants from the National Project
on Protein Structural and Functional Analyses of the
Ministry of Education, Culture, Sports, Science, and
Technology of Japan; the Targeted Proteins Research
Program; and Grants-in-Aid for Scientific Research
from the Ministry of Education, Culture, Sports,
Science, and Technology of Japan.
17. Patel, N., Moyed, H. S. & Kane, J. F. (1975).
Xanthosine-5′-phosphate amidotransferase from
Escherichia coli. J. Biol. Chem. 250, 2609–2613.
18. Deras, M. L., Chittur, S. V. & Davisson, V. J. (1999). N2-
Hydroxyguanosine 5′-monophosphate is a time-de-
pendent inhibitor of Escherichia coli guanosine mono-
phosphate synthetase. Biochemistry, 38, 303–310.
19. Chittur, S., Klem, T., Shafer, C. & Davisson, V. (2001).
Mechanism for acivicin inactivation of triad glutamine
amidotransferases. Biochemistry, 40, 876–887.
20. Bhat, J. Y., Shastri, B. G. & Balaram, H. (2008).
Kinetic and biochemical characterization of Plasmo-
dium falciparum GMP synthetase. Biochem. J. 409,
263–273.
Supplementary Data
Supplementary data associated with this article
can be found, in the online version, at doi:10.1016/
j.jmb.2009.10.053
References
1. Knaap, J. A., Kumar, B. R., Moshkin, Y. M., Langen-
berg, K., Krijgsveld, J., Heck, A. J. et al. (2005). GMP
synthetase stimulates histone H2B deubiquitylation
by the epigenetic silencer USP7. Mol. Cell, 17, 695–707.

Two-Subunit-Type GMP Synthetase
429
21. Tesmer, J. J., Klem, T. J., Deras, M. L., Davisson, V. J. &
Smith, J. L. (1996). The crystal structure of GMP
synthetase reveals a novel catalytic triad and is a
structural paradigm for two enzyme families. Nat.
Struct. Biol. 3, 74–86.
22. Maruoka, S., Lee, W. C., Kamo, M., Kudo, N., Nagata,
K. & Tanokura, M. (2005). Crystal structure of
glutamine amidotransferase from Pyrococcus horikoshii
OT3. Proc. Jpn. Acad. 81, 459–462.
23. Krissinel, E. & Henrick, K. (2007). Inference of
macromolecular assemblies from crystalline state. J.
Mol. Biol. 372, 774–797.
24. Tesmer, J. J., Stemmler, T. L., Penner-Hahn, J. E.,
Davisson, V. J. & Smith, J. L. (1994). Preliminary X-ray
structure of GMP synthetase: determination of anom-
alous scattering factors for a cysteinyl mercury
derivative. Proteins, 18, 394–403.
28. Otwinowski, Z. & Minor, W. (1997). Processing of X-
ray diffraction data collected in oscillation mode.
Methods Enzymol. 276, 307–326.
29. Vagin, A. & Teplyakov, A. (1997). MOLREP: an
automated program for molecular replacement. J.
Appl. Crystallogr. 30, 1022–1025.
30. McRee, D. E. (1999). XtalView/Xfit—a versatile
program for manipulating atomic coordinates and
electron density. J. Struct. Biol. 125, 156–165.
31. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997).
Refinement of macromolecular structures by the maxi-
mum-likelihood method. Acta Crystallogr. Sect. D, 53,
240–255.
32. Perrakis, A., Morris, R. & Lamzin, V. S. (1999).
Automated protein model building combined with
iterative structure refinement. Nat. Struct. Biol. 6,
458–463.
25. Zalkin, H. & Truitt, C. (1977). Characterization of the
glutamine site of Escherichia coli guanosine 5′-mono-
phosphate synthetase. J. Biol. Chem. 252, 5431–5436.
26. Zalkin, H. (1985). Glutamine amidotransferases.
Methods Enzymol. 113, 263–264.
27. Matte-Tailliez, O., Brochier, C., Forterre, P. & Philippe,
H. (2002). Archael phylogeny based on ribosomal
proteins. Mol. Biol. Evol. 19, 631–639.
33. Laskowski, R. A., MacArthur, M. W., Moss, D. S. &
Thornton, J. M. (1993). PROCHECK: a program to
check the stereochemical quality of protein structures.
J. Appl. Crystallogr. 26, 283–291.
34. Abbott, J., Newell, J., Lightcap, C., Olanich, M.,
Loughlin, D., Weller, M. et al. (2006). The effects of
removing the GAT domain from E. coli GMP
synthetase. Protein J. 25, 483–491.