Helices on Interdomain Interface Couple Catalysis in the ATPPase Domain with Allostery in Plasmodium falciparum GMP Synthetase

Article abstract of DOI:10.1002/cbic.202000158

GMP synthetase catalyses the conversion of XMP to GMP through a series of reactions that include hydrolysis of Gln to generate ammonia in the glutamine amidotransferase (GATase) domain, activation of XMP to adenyl-XMP intermediate in the ATP pyrophosphatase (ATPPase) domain and reaction of ammonia with the intermediate to generate GMP. The functioning of GMP synthetases entails bidirectional domain crosstalk, which leads to allosteric activation of the GATase domain, synchronization of catalytic events and tunnelling of ammonia. Herein, we have taken recourse to the analysis of structures of GMP synthetases, site-directed mutagenesis and steady-state and transient kinetics on the Plasmodium falciparum enzyme to decipher the molecular basis of catalysis in the ATPPase domain and domain crosstalk. Our results suggest an arrangement at the interdomain interface, of helices with residues that play roles in ATPPase catalysis as well as domain crosstalk enabling the coupling of ATPPase catalysis with GATase activation. Overall, the study enhances our understanding of GMP synthetases, which are drug targets in many infectious pathogens.

Full text of DOI:10.1002/cbic.202000158

Combining Chemistry and Biology
Accepted Article
Title: Helices on Interdomain Interface Couple Catalysis in the
ATPPase Domain With Allostery in Plasmodium falciparum GMP
Synthetase
Authors: Santosh Shivakumaraswamy, Nivedita Pandey, Lionel Ballut,
Sébastien Violot, Nushin Aghajari, and Hemalatha Balaram
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemBioChem 10.1002/cbic.202000158
Link to VoR: https://doi.org/10.1002/cbic.202000158
01/2020

10.1002/cbic.202000158
ChemBioChem
Helices on Interdomain Interface Couple Catalysis in the
ATPPase Domain With Allostery in Plasmodium falciparum
GMP Synthetase
Santosh Shivakumaraswamy,[a] Nivedita Pandey,[a] Lionel Ballut,[b] Sébastien Violot,[b]
Nushin Aghajari,[b] and Hemalatha Balaram*[a]
[a] Dr. S. Shivakumaraswamy, N. Pandey, Prof. Dr. H. Balaram
Molecular Biology and Genetics Unit,
Jawaharlal Nehru Centre for Advanced Scientific Research,
Bengaluru, Karnataka 560064, India.
E-mail: hb@jncasr.ac.in
[b] Dr. L. Ballut, Dr. S. Violot, Prof. Dr. N. Aghajari
Biocrystallography and Structural Biology of Therapeutic Targets, Molecular
Microbiology and Structural Biochemistry,
UMR 5086 CNRS-University of Lyon 1,
7 passage du Vercors, 69367 Lyon Cedex 07, France.
Abstract
GMP synthetase catalyzes the conversion of XMP to GMP through a series of reactions that
include hydrolysis of Gln to generate ammonia in the glutamine amidotransferase (GATase)
domain, activation of XMP to adenyl-XMP intermediate in the ATP pyrophosphatase (ATPPase)
domain and reaction of ammonia with the intermediate to generate GMP. The functioning of
GMP synthetases entails bidirectional domain crosstalk, which leads to allosteric activation of
the GATase domain, synchronization of catalytic events and tunnelling of ammonia. Herein, we
have taken recourse to the analysis of structures of GMP synthetases, site-directed mutagenesis
and, steady-state and transient kinetics on the Plasmodium falciparum enzyme to decipher the
molecular basis of catalysis in the ATPPase domain and domain crosstalk. Our results suggest an
arrangement at the inter-domain interface, of helices with residues that play roles in ATPPase
1
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
catalysis as well as domain crosstalk enabling the coupling of ATPPase catalysis with GATase
activation. Overall, the study enhances our understanding of GMP synthetases, which are drug
targets in many infectious pathogens.
Introduction
Glutamine amidotransferases are enzymes that operate in different biosynthetic pathways where
they catalyze the substitution of an oxo group with an amino group.[1] Guanosine monophosphate
synthetase (GMPS), a glutamine amidotransferase is a two domain (or two subunit) enzyme with
modular catalytic units that catalyzes the substitution of the C2 oxo group of XMP with an amino
group, producing GMP. This constitutes the last step in the de novo production of GMP. This
conversion of XMP to GMP involves the coordinated functioning of the glutamine
amidotransferase (GATase) domain that catalyzes the hydrolysis of the amide side chain of
glutamine (Gln) producing ammonia and the ATP pyrophosphatase (ATPPase) domain that
catalyzes the synthesis of adenyl-XMP intermediate from ATP.Mg2+ and XMP (Scheme 1).[1,2]
The ammonia generated in the GATase domain is tunnelled to the ATPPase domain where it
attacks the adenyl-XMP intermediate generating GMP. In vitro, GMP synthetases can utilize
externally available ammonia, although this is not physiologically relevant as at cellular pH,
ammonia would be present largely as ammonium ion, which cannot serve as a nucleophile
required for the reaction.[3] The GATase domain is allosterically activated by the binding of
substrates to the ATPPase domain, thus ensuring that Gln hydrolysis occurs only when the
ATPPase domain is primed to receive ammonia.[4–6] Further, the GATase and the ATPPase
domains synchronize their reaction rates to avoid any futile utilization of Gln and ATP.[7] In
some archaea, the GATase and the ATPPase domains are expressed separately (two-subunit type
GMPS), and the two subunits interact during catalysis.[8,9] However, in most organisms,
2
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
including archaea, the GATase domain (N-terminus) is connected to the ATPPase domain
through a linker to form the two-domain type GMPS. Since GMP is involved in critical cellular
functions, a functional GMPS is indispensable for the growth and survival of many organisms
including infectious pathogens where the enzyme is pursued as drug targets.[6,10–13]
Scheme 1. Reaction catalyzed by GMPS (A), GATase domain (B) and ATPPase domain (C).
The causative agent of cerebral malaria, Plasmodium falciparum, expresses a two-domain
type GMPS (PfGMPS).[14] Unlike other GMP synthetases, PfGMPS is unique in displaying basal
GATase activity in the absence of ATPPase substrates.[15] However, binding of substrates to the
ATPPase domain upregulates GATase activity by 8-12-fold.[15,16] A comparison of the kinetic
parameters of the GATase reaction in the absence and presence of ATPPase substrates shows
that GATase activation is enabled by lowering the Km(app) value for Gln by 360-fold.[16] Apart
from information flow from ATPPase to GATase domain, the inhibition of ammonia dependent
3
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
GMP formation by Gln binding to a GATase inactive PfGMPS_C89A mutant suggests
bidirectional interdomain crosstalk in PfGMPS.[7] This bidirectional domain crosstalk enables
synchronization of the GATase and ATPPase reaction rates as evidenced by the 1:1
stoichiometry of the products of the two domains.[7]
The GATase domain in GMP synthetases has an α/β hydrolase fold and a Cys, His, Glu
triad catalysing the hydrolysis of Gln.[17] The mechanism of substrate binding and catalysis in the
GATase domain is well explored,[3,5,18] however, insights into the functioning of the ATPPase
domain are lacking. Further, unlike in a majority of amidotransferases which possess either pre-
formed or transient ammonia tunnels,[19–21] the crystal structures of GMP synthetases in apo and
substrate-bound states have a solvent-exposed GATase active site and do not reveal a
tunnel.[16,17,22] Regardless of the absence of a tunnel in the available structures, biochemical
studies provide evidence for GATase activation and ammonia tunnelling in PfGMPS.[7] We
recently reported the crystal structure of Gln bound PfGMPS_C89A mutant which revealed an
85° rotation of the GATase domain.[16] Upon reconstruction of the path traversed during domain
rotation, a tunnel for the transport of ammonia is evident at an intermediate state of GATase
rotation.
To understand the molecular basis of catalysis in the ATPPase domain and crosstalk with
the GATase domain, we analyzed the structures of PfGMPS and other available GMP
synthetases. Based on this analysis, eighteen site-directed mutants of PfGMPS were generated
and subjected to enzyme assays to examine the activity of the individual domains, GATase
activation and GMP formation. These studies have enabled the identification of residues that are
involved in the catalysis of adenyl-XMP formation. In addition, our results highlight the
involvement of residues on interface helices in the catalysis of adenyl-XMP formation and
4
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
domain crosstalk. By tracing the connectivity of catalytic residues in the ATPPase domain with
GATase residues, we present the molecular basis of active site synchronization. Our results on a
dimer-interface mutant provide a functional rationale for the obligate dimeric nature of GMP
synthetases from prokaryotes and parasitic protozoa, including PfGMPS, a feature distinct from
most eukaryotic homologs that are monomers.
Results
Examination of the available structures of GMP synthetases in conjunction with an alignment of
GMPS sequences from diverse species enabled the identification of residues that could be
involved in the catalysis of adenyl-XMP formation and interdomain crosstalk between the
ATPPase and GATase domains. These observations were validated by generating eighteen site-
directed mutants of PfGMPS and the positions of the mutated residues are shown in Figure 1A.
The PfGMPS wild type (WT) and mutants were expressed in Escherichia coli and purified to
greater than 95 % homogeneity (Figure S1). The absence of any significant difference in the far-
UV circular dichroism spectra and elution times on analytical size-exclusion chromatography
between the WT and mutants confirmed that the overall secondary structure and oligomeric state
are retained (Figure S1). The mutants were then subjected to enzymatic assays to assign the role
of the mutated residue to one of the three key steps in catalysis (Figure 1B) viz., adenyl-XMP
formation, activation of GATase domain and GMP synthesis using either Gln or NH4Cl as
nitrogen source with the former entailing ammonia tunnelling. The rate (k’) of adenyl-XMP
formation was measured with ATP.Mg2+ and XMP as substrates using a stopped-flow
spectrophotometer. The enhanced rate of Gln hydrolysis due to allosteric activation of the
GATase domain by the binding of substrates to the ATPPase domain was measured using L-
glutamate dehydrogenase as the coupling enzyme to detect the liberated glutamate. The same
5
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
assay was also employed to determine the binding affinities (Kd values) of the inactive mutants
for ATP.Mg2+ and XMP. The results of these experiments are discussed in the following
sections.
6
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
Figure 1. Overview of residues mutated and enzyme assays performed. (A) Schematic showing
the location of the conserved motifs that are examined in this study. The interface helices are
shown as squiggles. The sequence of the conserved motifs of α1, α11 and α12 helices, the lid-
loop and the C-terminal loop in PfGMPS, as well as the WebLogo representation of the multiple
sequence alignment (Figures S2 and S3) are shown. The location and conservation (in WebLogo
format) of α1 residue His20, interface residues Lys160, Glu213, Lys376 from the 85° GATase
rotated form, and dimer interface residue Arg539 are also shown. The residues investigated by
site-directed mutagenesis in this and an earlier study[16] are shown as stars and circles,
respectively. The location of the GATase catalytic triad residues is shown as squares. (B)
Scheme illustrating the enzymatic assays performed using PfGMPS WT and mutants. In the plot
for adenyl-XMP formation, data is shown as points and the fit to the equation is shown as a line.
The dashed line in the plot for GATase activation represents basal GATase activity and the solid
line represents allosteric activation of GATase activity by substrates binding to the ATPPase
domain.
Catalysis in the ATPPase domain
Catalysis in the ATPPase domain involves the binding of substrates ATP.Mg2+, XMP and the
subsequent formation of the adenyl-XMP intermediate through a phosphoryl transfer reaction.
Here, the β-γ phosphoanhydride bond of ATP is cleaved and the AMP so generated is transferred
to XMP.[23,24] The crystal structure of E.coli GMPS (EcGMPS) is complexed with AMP and PPi
(PDB ID 1GPM)[17] while the structures of human (PDB ID 2VXO),[22] T. thermophilus (PDB ID
2YWC) and P. falciparum GMPS (PDB ID 3UOW) have a bound XMP. A 25-residue long lid-
loop which connects the helix α11 with the strand β12 is disordered in all available structures
except for human GMPS[22] and PfGMPS_C89A/Gln (PDB ID 4WIO)[16]. The lid-loop is
mapped away from the ATPPase active site in human GMPS; however, in PfGMPS_C89A/Gln,
except for the eight intervening residues, the remaining residues of the lid-loop are mapped and
positioned in the ATPPase active site (Figure 2A). Superposition of AMP-PPi and XMP from the
structures of EcGMPS and PfGMPS, respectively, onto PfGMPS_C89A/Gln structure, revealed
that residues that interact with ATP are primarily located in three regions viz., helices α11 and
α12, the lid-loop and a conserved motif called the P-loop motif. Although residues from both the
sub-domains interact with XMP, a majority of them are located on the C-terminal loop of the
7
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
ATPPase domain. Considering that the role of the P-loop residues in catalyzing phosphoryl
transfer reactions has been well studied,[25,26] we mutated the residues on helices α11 and α12,
the lid-loop and the C-terminal loop to evaluate their role.
Invariant residues on the lid-loop.
The lid-loop has an invariant motif IK(T/S)HHN (residues 385-390 in PfGMPS) (Figures 1A, S2
and S3) and except for Asn390 which is disordered, other residues of the motif are in contact
with AMP (superposed from EcGMPS) (Figure 2B). The residues KTHHN of the invariant motif
were mutated one at a time and the purified mutants subjected to enzymatic assays. The Gln and
NH4Cl dependent GMP formation by the PfGMPS mutants K386L, H388A and H389A were
drastically impaired, ranging from 65-fold to 731-fold drop compared to WT (Table 1). Stopped-
flow experiments revealed that the k’ for adenyl-XMP formation in PfGMPS_H388A was 24-
fold lower than WT while the progress curves for K386L and H389A mutants did not display
any drop in absorbance suggesting that these mutants were severely compromised in their ability
to synthesize adenyl-XMP (Figures 2C and 3B).
8
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
9
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
Figure 2. Role of lid-loop residues in adenyl-XMP formation. (A) Superposition of the ATPPase
domain of XMP bound PfGMPS (magenta, PDB ID 3UOW) on the ATPPase domain of
PfGMPS_C89A/Gln (green, PDB ID 4WIO). The structures superpose with an RMSD of 0.55 Å
over 240 Cα atoms. The lid-loop in PfGMPS_C89A/Gln is highlighted for reference. (B)
Interactions of residues on the lid-loop with AMP in the structure of PfGMPS_C89A/Gln.
Shown in dashed line are only salt bridge interactions. In both (A) and (B), AMP and PPi are
superposed from EcGMPS (PDB ID 1GPM). The ATPPase domains of EcGMPS and
PfGMPS_C89A/Gln superpose with an RMSD of 1.51 Å over 244 Cα atoms. Protein backbone
is in ribbon and ligands are in ball and stick representation. (C) Stopped-flow progress curves for
the formation of adenyl-XMP intermediate. Shown are absorbance values at 290 nm as a
function of time (points) and fit of the data (solid lines) to one-phase exponential decay equation.
The concentrations used were 7.5 µM enzyme, 2 mM ATP.Mg2+ and 0.15 mM XMP. Structural
superposition was performed using PyMOL.
To rule out impairment in substrate(s) binding being the cause for the lowered rate of
adenyl-XMP/ GMP formation, we used GATase activity assay to measure affinities for
ATP.Mg2+ and XMP. The levels of GATase activation in H388A and H389A mutants of
PfGMPS were 51 and 43% of the WT (Figure 3A), respectively, while PfGMPS_K386L showed
a massive drop to 19% that is only marginally greater than background basal (without binding of
ATP.Mg2+ and XMP) glutaminase activity of PfGMPS. The Kd values for the complex of
PfGMPS mutants H388A and H389A with ATP.Mg2+ and XMP are in the low micromolar range
(Table S1) while the low level of activation in PfGMPS_K386L precluded Kd measurement.
These results demonstrate that Lys386 is essential for substrate binding, whereas His388 and
His389 play a catalytic role in the synthesis of adenyl-XMP intermediate.
10
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
Table 1. Gln and NH4Cl dependent GMP formation by PfGMPS WT, lid-loop and C-terminal
loop mutants.
Specific activity (nmol min-1 mg-1)
Enzyme
Gln dependent GMP
formation
NH4Cl dependent
GMP formation
WT
658 ± 19
8.1 ± 0.2
526 ± 14
0.69 ± 0.09
3.1 ± 0.2
No Activity
486 ± 7
700 ± 22
10.5 ± 0.1
525 ± 1
K386L
T387A
H388A
H389A
N390A
E555L
2.08 ± 0.65
3.8 ± 0.1
No Activity
496 ± 3
The substrate concentrations used were 2 mM ATP.Mg2+, 150 µM XMP and 0.5 mM Gln or 50
mM NH4Cl. The values are expressed as mean ± SEM of three measurements.
11
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
Figure 3. GATase activation and adenyl-XMP formation in PfGMPS WT and mutants. (A)
Magnitude of GATase activation estimated using a coupled enzyme assay. GATase activity was
measured using 20 mM Gln, 3 mM ATP.Mg2+ and 150 µM XMP. The specific activity values
are normalized to the GATase activity of the WT. The error bars represent the SEM of data from
three measurements. (B) Rate (k’) of adenyl-XMP formation measured using a stopped-flow
spectrophotometer. The error bars represent the SEM of data from three independent
measurements of k’. The concentration of the enzyme used was 7.5 µM and the concentration of
ATP.Mg2+ and XMP were 2 mM and 0.15 mM, respectively. PfGMPS mutant is indicated by the
residue, location in sequence and the residue to which it is mutated. In both A and B, the results
of unpaired Student’s t test are indicated as follows, ns, not significant; *, p < 0.05; **, p < 0.01;
***, p < 0.001; ****, p < 0.0001.
Of the two other lid-loop mutants, PfGMPS_T387A displayed only a minor reduction in
the specific activity for Gln and NH4Cl dependent GMP formation compared to WT while
12
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
PfGMPS_N390A was completely inactive (Table 1). Both the mutants could synthesize adenyl-
XMP intermediate albeit at reduced rates as the k’ for adenyl-XMP formation was 3-fold and 7-
fold lower in the T387A and N390A mutants of PfGMPS, respectively (Figures 2C and 3B). The
activation of the GATase domain in PfGMPS_T387A was reduced by 23% while it was
enhanced to 200% of WT in PfGMPS_N390A (Figure 3A). These results establish that Thr387 is
not important for PfGMPS catalysis while Asn390 plays an essential role in a step subsequent to
adenyl-XMP formation as it can synthesize the adenyl-XMP intermediate but not the final
product GMP.
Residues on inter-domain interface helices α11 and α12
In the crystal structures where the GATase domain is in 0° rotated state, the domain interface is
composed of helix α1 (GATase) and the helices α11 and α12 (ATPPase) (discussed below).
These helices harbour conserved motifs (Figures 1A, S2 and S3), of which the side chains of
some of the residues directed at the interface, engage in interdomain interactions. The side chains
of the other residues viz., Asp371 (α11 motif) and Lys411 and Lys415 (α12 motif) are directed
into the ATPPase catalytic pocket (Figure 4). The residue Asp371 has its side chain placed above
the expected location of the adenyl-XMP bond. The side chain of Lys411 is involved in salt
bridge like interactions with the phosphate of PPi that is distal to AMP (corresponding to γ-
phosphate of ATP) while the side chain of Lys415 is devoid of contacts with the ligand or with
other residues in PfGMPS structures (PDB IDs 3UOW, 4WIM).
13
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
Figure 4. Interaction of Lys411 on the interface helix α11 with PPi in the crystal structure of apo
PfGMPS (PDB ID 4WIM). The ligands AMP, PPi (ball and stick) and Mg2+ (green sphere) are
superposed from the crystal structure of EcGMPS (PDB ID 1GPM). The ATPPase domains of
EcGMPS and apo PfGMPS superpose with an RMSD of 1.73 Å over 246 Cα atoms. The helices
α11 and α12 are coloured red and purple, respectively. Structural superposition was performed
using PyMOL.
The almost complete impairment of adenyl-XMP formation upon mutation of Asp371 to
Ala, reported in our previous study, indicates a key catalytic role for this residue on the interface
helix α11.[16] Mutation of Lys411 on helix α12 to Leu resulted in a complete loss of Gln and
NH4Cl dependent GMP formation, arising from the complete absence of adenyl-XMP formation
in stopped-flow assays (Figure 3B). However, unlike in PfGMPS_D371A, where GATase
activation was higher than WT,[16] the 70% drop seen in PfGMPS_K411L indicates that
substrate(s) binding is highly compromised (Figure 3A). Mutation of Lys415 to Leu altered only
the Km(app) value for ATP that was 4.2-fold higher compared to WT (Table 2), and this is
reflected in 63% lower k’ for adenyl-XMP formation (Figure 3B). These results highlight the
14
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
role of residues on the interface helices in substrate binding (Lys411, Lys415) and adenyl-XMP
formation (Asp371). It should be noted that although the side chain of Lys415 (α12 motif) is
devoid of any interactions in PfGMPS structures, in most other GMPS structures, the side chain
of this residue is involved in H-bonding interactions with main chain atoms of residues lining the
ATPPase pocket.
15
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
Table 2. Steady-state kinetic parameters of interdomain interface mutants.
Varied
substrate
WT
K24L R25L K160L E213A K376L K411L D412A D413A K415L
0.33
1.75
0.47
0.363
1.29
1.46
2.61
1.19
0.25
Km(app)
± 0.02 ± 0.06 ± 0.01 ± 0.004 ± 0.04 ± 0.07
± 0.22 ± 0.10 ± 0.01
(mM)
**
*
ns
**
**
**
*
ns
0.766 0.296 0.777 0.832
0.794
0.804
0.657 0.584
± 0.004 ± 0.008 ± 0.004
0.707
kcat
(s-1)
Gln
± 0.005 ± 0.007 ± 0.031 ± 0.007 ± 0.026 ± 0.022
***
169
ns
*
ns
ns
**
**
*
kcat /Km(app)
(s-1 M-1)
2339
1657
2292
614
553
252
492
2780
119
± 2
67
± 3
117
± 4
148
± 3
162
± 2
122
± 0.1
96
± 5
102
± 1
500
± 11
K m(app)
(µM)
**
ns
*
**
ns
ns
*
***
kcat
(s-1)
ATP
0.893 0.340 0.831 0.820
± 0.039 ± 0.004 ± 0.003 ± 0.019 ± 0.009 ± 0.008
0.846
0.872
0.724 0.598
± 0.022 ± 0.028 ± 0.018
0.733
kcat /Km(app)
(s-1 M-1)
7496
5112
7099
5553
5226
7161
7547
5855
1465
Inactive
7.1
± 1
20.6
± 1
6.3
± 0.4
12.7
± 0.6
13.1
± 1
12.1
± 0.6
9.5
± 1.9
20
± 2.9
14.4
± 1.5
K m(app)
(µM)
*
ns
*
*
*
ns
ns
ns
kcat
(s-1)
0.741 0.327 0.751 0.936
± 0.058 ± 0.005 ± 0.070 ± 0.035 ± 0.025 ± 0.029
0.891
0.956
0.633 0.532
± 0.058 ± 0.011 ± 0.059
0.744
XMP
kcat /Km(app)
(× 103 s-1
M-1)
104
15.9
73.8
68.2
79.2
66.9
26.7
51.7
119
25.6
± 2.8
20.3
± 2.1
23.6
± 0.6
12.3
± 0.4
36.2
± 2.2
38.9
± 0.2
12.2
± 1.4
17
± 0.3
13.1
± 1.1
K m(app)
(mM)
ns
ns
*
ns
*
*
ns
ns
kcat
(s-1)
1.081 0.465 0.939 1.240
± 0.107 ± 0.011 ± 0.004 ± 0.027 ± 0.002 ± 0.073
0.708
0.792
1.054 0.711
± 0.030 ± 0.024 ± 0.008
0.591
NH4Cl
*
ns
ns
ns
ns
ns
ns
*
kcat /Km(app)
(s-1 M-1)
42
23
72
53
57
22
27
58
35
The values are expressed as mean ± SEM of two independent experiments. The assay involved
monitoring the decrease in absorbance at 290 nm arising from the conversion of XMP to GMP at
25 °C. The results of unpaired Student’s t test are indicated as follows, ns, not significant; *, p <
0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001.
16
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
Residues on the C-terminal loop
The C-termini of the ATPPase domain in GMP synthetases comprises a 13-residue long loop (C-
terminal loop) which bears the highly conserved signature motif KPPXTXE(F/W)X (residues
547-555 in PfGMPS) (Figures 1A, S2 and S3). The ζ-amino group of Lys547 interacts with the
phosphate of XMP (Figure 5A). The peptide bonds Lys547-Pro548-Pro549 are in the cis
configuration, which, induces a turn in the polypeptide chain and enables the interaction of
Glu553 side chain carboxyl group with ribose 2’ hydroxy group of XMP (Figure 5A). In
addition, the α-carboxy of the terminal Glu555 residue forms a salt bridge with the guanidino
group of Arg336 that in turn interacts with C6 oxygen and N7 of XMP (Figure 5A). Hence,
residues of the C-terminal loop make direct and indirect contacts with XMP. The conformation
of the loop is restrained by dimer interactions involving H-bonds between the Nη2 of Arg539’
(from the neighbouring chain) and backbone CO of both Thr551 and Glu553 (Figure 5B).
17
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
Figure 5. Interactions of the C-terminal loop in PfGMPS (PDB ID 3UOW). (A) Interactions with
XMP. (B) Interdimer interactions of Arg539 from a neighbouring chain with residues on C-
terminal loop. In both images, the protein backbone is in ribbon representation with chain A
coloured green and chain B orange. The C-terminal loop of chain A is coloured magenta. XMP is
shown as ball and stick and highlighted residues are shown as sticks.
We probed the role of C-terminal loop residues Lys547, Glu553, Glu555 as well as the
importance of the interdimer interactions of Arg539 by mutating them to Leu. The K547L,
E553L and R539L mutants of PfGMPS were inactive for Gln and NH4Cl dependent GMP
18
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
formation on account of impaired adenyl-XMP formation (Figure 3B). GATase activation was
reduced by >70% suggesting that these mutants are defective in binding substrates (Figure 3A).
In contrast, the Gln and NH4Cl dependent GMP formation by the PfGMPS_E555L mutant was
only marginally lower than WT (Table 1). Also, the changes in GATase activation were minor,
and the k’ for adenyl-XMP formation was lower by only 39% (Figure 3). These results prove that
the direct interactions of Lys547 and Glu553 and the indirect interactions of Arg539 are
necessary for substrate binding and corroborate structural observations.
Domain crosstalk.
Interdomain interface in GMP synthetases
The available GMPS structures were examined to identify residues at the interdomain interface
that could mediate domain crosstalk. The GATase domain is oriented with respect to the
ATPPase domain in the same manner in all GMPS structures except in the PfGMPS_C89A/Gln
structure where this domain is rotated by 85° (Figure 6A, C). We refer to the state of the domain
rotation in PfGMPS_C89A/Gln complex as the 85° and the others as 0°. In structures where the
GATase is in 0° rotated state, the interdomain interface is predominantly composed of the N-
terminal helix, α1 of the GATase domain and helices α11 and α12 of the ATPPase domain
(Figure 6A). The three interface helices harbour the conserved motifs (K/R)(K/R)XRE (α1
motif), DXXES (α11 motif) and KD(D/E)V(K/R) (α12 motif) (Figures 1A, S2 and S3). The
charged residues in the 4th and 5th positions of the α1 motif are replaced with Asn in sequences
from Plasmodium species. Further, the conserved D/E and K/R substitutions in the α12 motif are
limited
to
sequences
from
Plasmodium.
19
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
Figure 6. Interdomain interactions in PfGMPS. (A) Helices α1, α11 and α12 in the interdomain
interface in the 0° GATase rotated structure of PfGMPS (PDB ID 3UOW). (B) Interdomain salt
bridges in the 0° structures. The chain A in structure 3UOW is superposed on chain B of 4WIM
(apo PfGMPS), which is shown in a lighter shade of colour. The two chains superpose with an
RMSD of 1.71 Å over 457 Cα atoms. (C) Helices α11 and α12 in the interdomain interface in
85° GATase rotated PfGMPS structure (PDB ID 4WIO). (D) Interdomain salt bridges in the 85°
rotated structure. In all the images, the GATase and ATPPase domains are coloured green and
purple, respectively. The backbone is shown as line (A, C, D) and as ribbon (B). The disordered
lid-loop is shown as a dashed line (A, C). Structural superposition was performed using PyMOL.
20
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
Interdomain interactions of the 0° interface
The α1-α12 interdomain interactions are largely conserved in all GMPS structures where the
GATase is in the 0° rotated state. The first and the fourth residues of α1 motif ((K/R)(K/R)XRE)
form two interdomain salt bridges with the second and third residues of the α12 motif
(KD(D/E)V(K/R) (Table S2), respectively, though in Neisseria gonorrhoeae GMPS, the distance
between the interacting pairs is above 4 Å. In the structures of PfGMPS (PDB ID 3UOW,
4WIM), the side chains of some of these residues are disordered. However, through modelling of
missing side chains or structural superposition of protomers from two structures (3UOW and
4WIM), the atoms involved in salt bridge formation between Lys24 (α1) and Asp412/Asp413
(α12) come within a distance of ≤ 4 Å (Figure 6B and Table S2). The second salt bridge is absent
in PfGMPS as the fourth residue of the α1 motif is an asparagine.
Unlike α12, the residues of α11 across the GMPS structures, show varying levels of
disorder indicating conformational flexibility of this segment. Also, an interdomain salt bridge
involving His20 preceding the α1 motif and Glu374 of α11 motif was observed by superposing
two chains of PfGMPS (Figure 6B).[16] His20 is not conserved across GMPS sequences (Figures
1A, S2 and S3) and a similar salt bridge is observed only in N. gonorrhoeae GMPS where the
histidyl residue is replaced by Arg (Table S2). Other interdomain salt bridges are enzyme-
specific and the interacting residues are not conserved.
The significance of α1-α12 interdomain interactions was investigated in PfGMPS by
mutating Lys24 of the α1 motif and Asp412 as well as Asp413 of the α12 motif. Compared to
WT, the mutants K24L, D412A and D413A of PfGMPS displayed an increase in Km(app) value
for the substrate glutamine with PfGMPS_D412A showing the highest (7.9-fold) increase (Table
21
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
2). The kcat for Gln dependent GMP formation was unaltered in the D412A and D413A mutants
of PfGMPS, however, this parameter was lower by 60% in the K24L mutant compared to WT
(Table 2). This reduction in kcat value for PfGMPS_K24L was deemed significant by statistical
test (p < 0.001). The reduction in kcat/Km(app) for Gln dependent GMP formation in PfGMPS
mutant K24L was the highest (13.9-fold), followed by D412A (9.3-fold) and D413A (4.8-fold).
The levels of GATase activation were lower by 45%, 34% and 42% in the K24L, D412A and
D413A mutants, respectively, while the k’ for adenyl-XMP formation did not vary significantly
(Figure 3). Though PfGMPS_K24L is a GATase mutant, the kcat for NH4Cl dependent GMP
formation, a step that is independent of GATase activity, is reduced by half (Table 2). The
second residue of the α1 motif, Arg25, interacts with the residues on the loop containing His208
and Glu210, which together with Cys89 constitute the GATase catalytic triad (Figure S4).
Mutation of Arg25 to Leu however did not result in any significant change in kinetic properties,
the level of GATase activation or the k’ for adenyl-XMP formation (Table 2 and Figure 3).
Unlike the mutation of lid-loop residues which results in complete inactivation, interface
residues are distal to the GATase and ATPPase active site pockets and hence the observed
changes are not drastic. However, the significant reduction in kcat/Km(app) upon mutating Lys24,
Asp412 and Asp413 suggests an essential role for these residues in mediating domain crosstalk.
This is supported by the results of our previous study where we had probed the α1-α11
interactions. In a manner identical to Lys24, mutation of His20 on α1 results in a two-fold
reduction in kcat for both Gln and NH4Cl dependent GMP formation while the results of mutating
the interacting partner, Glu374 on α11, suggested that this residue plays multiple roles including
domain crosstalk and adenyl-XMP formation.[16]
22
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
Interdomain interactions of the 85° interface
In the PfGMPS_C89A/Gln structure (PDB ID 4WIO), helices α11 and α12 retain the
conformation that is observed in the 0° structure, however, they interact with a different set of
GATase residues on account of GATase rotation (Figure 6C). Although the side chains of many
residues constituting this 85° interface are disordered, an examination of the structure shows that
the side chains of the residue pairs Glu213/Lys376 and Lys160/Asp412 could be involved in
forming interdomain salt bridges (Figure 6D). The GATase residues Glu213, Lys160 and the
ATPPase residue Lys376 located following the α11 motif, are not conserved while Asp412 is a
member of the invariant α12 motif (Figures 1A, S2 and S3).
Mutation of Glu213 to Ala as well as Lys376 to Leu results in an increase in the Km(app)
value for Gln by 3.9 and 4.4-fold, respectively (Table 2) whereas the changes in kcat are minor. In
PfGMPS_E213A, the levels of GATase activation and the k’ for adenyl-XMP formation are
reduced by 36% and 41%, respectively, while the changes for the K376L mutant are not
significant (Figure 3). Mutation of Lys160 to Leu did not affect the kinetic parameters (Table 2),
the levels of GATase activation or the k’ for adenyl-XMP formation (Figure 4). However, as
discussed above, mutation of Asp412, which is also involved in interdomain salt bridges in the
0° interface, impairs domain crosstalk. The increase in the Km(app) value for Gln upon mutating
Glu213 and Lys376, suggests a role for even the 85° interface residues in domain crosstalk.
Discussion
The incorporation of nitrogen into XMP catalyzed by GMP synthetases requires the production
of a reactive nitrogen atom (with its lone pair of electrons) in the GATase domain, activation of
the substrate XMP in the ATPPase domain and the interdomain transfer of nitrogen (ammonia).
In addition, GMP synthetases are characterized by allostery as events in the ATPPase domain
23
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
upregulate Gln hydrolysis in the GATase domain. Also, the catalytic events in the two domains
are synchronized. To identify the residues involved in these processes, we resorted to structural
analysis, site-directed mutagenesis and enzyme assays. These investigations have helped us map
the ATPPase active site as well as the residues involved in domain crosstalk.
Mapping functional residues in the ATPPase catalytic pocket
Our studies have revealed that residues that are critical for the functioning of the ATPPase
domain are located on the interface helices α11 and α12, lid-loop and C-terminal loop. Whereas
Asp371 on α11 is essential for adenyl-XMP formation[16], Lys411 and Lys415 as shown in this
study, are involved in the binding of ATP. Interestingly, helices α11 and α12 also harbour
residues whose side chains contact the GATase domain implicating a dual function of catalysis
in the ATPPase domain and interdomain crosstalk (see below).
The lid-loop subsequent to helix α11 contains the invariant IK(T/S)HHN motif. The
residue Lys386 participates in ATP binding, whereas His388 and His389 are involved in
catalyzing the formation of the adenyl-XMP intermediate. The increase in electrophilicity of C2
of XMP enabled through the formation of the unstable adenyl-XMP intermediate facilitates
attack by ammonia and thereby the replacement of the C2 oxygen with nitrogen. Transfer of
AMP from ATP to XMP is an adenylation reaction that is facilitated by the ATP α-phosphorous
undergoing geometric as well as electrostatic changes involving the development of strong
negative charges on the non-bridging oxygen atoms in the transition state.[27,28] Since
stabilization of transition state underlies enzymatic rate enhancements, all phosphoryl transfer
enzymes, including adenylating enzymes, employ positively charged groups to interact with and
neutralize the negative charges on the non-bridging oxygen atoms.[29] Our results suggest a
dominant role for His389 and also a possible function for His388 in the stabilization of the
24
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
transition state as both H388/389A mutants of PfGMPS bind ATP.Mg2+ and XMP, but are highly
or completely impaired in adenyl-XMP formation. Crystal structures also lend support to the
catalytic role of His388 and His389 as these two residues along with Lys386 are the only ones
with positively charged side chains lying within an interacting distance of ATP α-phosphate.
The lid-loop is disordered in most structures of GMPS suggesting that it is
conformationally dynamic. This is also seen in the adenylating enzyme tyrosyl-tRNA synthetase,
wherein the loop containing the conserved KSMKS motif changes its conformation from an open
to a closed-form upon binding ATP. This closure results in a movement of the ζ-amino group of
Lys235 (KSMKS motif) close to the ATP α-phosphate thereby resulting in catalysis.[30,31] The
results with the PfGMPS lid-loop mutants suggest that even in GMP synthetases, binding of
ATP.Mg2+ and XMP induces conformational changes in the lid-loop culminating in the
repositioning of the IK(T/S)HHN motif residues in proximity to ATP α-phosphate, facilitating
the synthesis of adenyl-XMP intermediate through residues His388 and His389 as well as the
formation of GMP through interactions of Asn390. This design of placing catalytic residues on a
conformationally dynamic lid-loop has enormous functional implications as it ensures that the
labile adenyl-XMP intermediate is synthesized only after a closed catalytic pocket is established.
The sealed pocket protects the adenyl-XMP intermediate from attack by water and also enables
ammonia channelling.
The C-terminus of the ATPPase domain mediates dimerization in PfGMPS and
contributes many residues that interact with the substrate XMP. Our experiments suggest that
residues Lys547 and Glu553 on the C-terminal loop are critical for binding XMP. Moreover, the
interactions of Arg539’ from the neighbouring protomer with the C-terminal loop facilitates
substrate binding by maintaining the correct conformation of the C-terminal loop.[16] The
25
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
contribution of residues from the neighbouring protomer in substrate binding sheds light on the
essentiality of dimerization in PfGMPS and GMP synthetases from prokaryotes and parasitic
protozoa.[6,13,15] Human GMPS and GMP synthetases from higher eukaryotes possess a 150-
residue insertion that enables them to function as monomers.[32] These results highlight a
significant difference between the active sites of the ATPPase domain of GMPS from pathogens
and the human host, which could be exploited in the design of specific inhibitors.
Mechanism of allostery and domain crosstalk in GMP synthetases
The GATase domain is connected to the ATPPase domain through a flexible linker which
permits an 85° rotation. In structures of two-domain type GMPS where the GATase is in 0° state,
the interdomain interactions mediated by GATase helix α1 and ATPPase helices α11 and α12 are
conserved. Even upon 85° rotation, helices α11 and α12 remain part of the interface, suggesting
their central role in domain crosstalk. A total of nine residues in the 0° and 85° interfaces were
mutated and the enzymatic assays implicate the interface residues in domain crosstalk.
As reported earlier, GATase activation induced by the binding of ligands to the ATPPase
domain is characterized by a reduction in the Km(app) value for Gln, arising from the formation of
a high-affinity binding pocket.[16] Disruption of the interdomain salt bridges between Lys24
(helix α1) and Asp412/Asp413 (helix α12) seen in the 0° structure, results in 4-8-fold increase in
the Km(app) value for Gln and a 5-14-fold reduction in the kcat/Km(app) value. Similarly, mutation of
Glu213 and Lys376 (helix α11), the interacting pair in the 85° rotated structure led to a 4-fold
increase in Km(app) value for Gln. The increase in Km(app) value for Gln implicates these residues,
and the interdomain interaction involving them, in allosteric GATase activation. This apart, we
had earlier shown that mutation of Glu374 (helix α11) that interacts with His20 (helix α1) in the
0° structure also impairs GATase activation. The role of α1 in domain crosstalk is supported by
26
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
NMR experiments on two-subunit type GMPS from Methanocaldococcus jannaschii where the
binding of substrate-bound ATPPase subunit induced chemical shift changes in residues on helix
α1 as well as the loops containing the catalytic triad of the GATase subunit.[9]
Interestingly, although Glu213 is a GATase residue that is far away from the ATPPase
active site, the PfGMPS_E213A mutant displays a reduction in the k’ for adenyl-XMP formation
by 41%. These results validate that rotation of the GATase domain and interactions in the 85°
rotated state are integral to allosteric activation of GATase domain as well as catalysis in the
ATPPase domain. Corroborating these observations is a study on EcGMPS, in which through
juxtaposition of the GATase active site against the ATPPase active site, His186 (analogous to
Glu213 in PfGMPS) was seen to form an interdomain salt bridge with Glu383 on α12. Mutation
of His186 to Ala resulted in an increase in the Km value for Gln by 50-fold and uncoupling of the
two reactions.[33]
Taken together, the results from the 0° and 85° interface mutants suggest that the
interface helices are conduits for information exchange pertaining to GATase activation and
domain crosstalk. In addition, these results suggest that allosteric activation of the GATase
domain is a cumulative effect of numerous interdomain interactions in both the 0° and 85°
interfaces. It is possible that apart from the two interfaces observed in the available crystal
structures, the interdomain interactions in the intermediate states of GATase rotation are also of
considerable significance. Our inability to obtain an interface mutant with a very large increase
in Km(app) value for Gln reflecting GATase activation in the wild type enzyme, suggests the
presence of a network of interactions possibly through multiple steps in the trajectory of domain
rotation.
27
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
Interface helices link catalysis in ATPPase domain with the GATase activation
The experiments described above have enabled the identification of various residues involved in
the functioning of the ATPPase domain. Many of these residues are located on the interface
helices α11 and α12 and the lid-loop that is connected to α11. The residues on these structural
elements play critical roles in substrate binding (Lys386, Lys411 and Lys415), catalyzing the
formation of adenyl-XMP intermediate (Asp371, His388 and His389) and facilitating the
reaction of ammonia with adenyl-XMP to generate GMP (Asn390). Whereas residues Asp371,
Lys411 and Lys415 on helices α11 and α12 that are essential for ATPPase catalysis have their
side chains pointed into the ATPPase pocket, residues Lys376, Glu374 and Asp412/413 have
their side chains directed at the interdomain interface where they interact with GATase residues
(Figure 7). This arrangement permits the residues on helices α11 and α12 to relay the state of
substrate occupancy and/or the formation of adenyl-XMP intermediate to the residues on
GATase helix α1. An examination of the structure shows that residues on helix α1 interact with
the GATase active site loop which harbours two residues, His208 and Glu210 of the catalytic
triad (Figure S4). This permits coupling the events in ATPPase domain (ATP/XMP binding and
adenyl-XMP intermediate formation) with GATase activation leading to the co-ordination of the
catalytic events across the two domains.
28
This article is protected by copyright. All rights reserved.

10.1002/cbic.202000158
ChemBioChem
Figure 7. Schematic illustrating the coupling of catalysis in ATPPase domain with allostery.
Residues on ATPPase helices α11 and α12 that face the ATPPase pocket along with those on the
lid-loop play roles in binding of substrates and ATPPase catalysis. The residues on α11 and α12
that face the interface form salt bridges with residues on GATase helix α1. The interactions in
the 0° GATase rotated state is shown here. Residues Lys547 and Glu553 on the C-terminal loop
are essential for substrate binding and the conformation of the loop is stabilized by interactions
of Arg539’ from the neighbouring chain. Key inter-residue and protein-ligand interactions are
shown as blue and brown dashes, respectively. The GATase catalytic triad residues Cys89,
His208 and Glu210; and Asp172 which is involved in GATase activation are also shown.
The lid-loop augments these processes since it is connected to the interface helix α11 and
thus the conformational dynamics of the lid-loop are linked to GATase rotation. Following
substrate binding, the lid-loop closes over the ATPPase active site thereby facilitating the
formation of adenyl-XMP intermediate through interactions of His388 and His389. Apart from
enabling the synthesis of adenyl-XMP intermediate, the closed conformation of the lid-loop
forms a sealed pocket for ammonia tunnelling. This pocket creates a solvent-free environment
protecting the adenyl-XMP intermediate from attack by water while also preventing the
protonation of ammonia, which would render it incapable of carrying out the nucleophilic attack.
Conclusions
By mapping various residues that play critical roles in ATPPase catalysis and domain crosstalk,
the results presented herein provide insight into the mechanism of synchronization of active sites.
29
This article is protected by copyright. All rights reserved.