Biochem. J. (2009) 424, 211–220 (Printed in Great Britain)
211
doi:10.1042/BJ20090888
Structural insight on the control of urea synthesis: identification of the
binding site for N -acetyl-L-glutamate, the essential allosteric activator
of mitochondrial carbamoyl phosphate synthetase1
Satu PEKKALA*, Ana I. MARTÍNEZ*, Belén BARCELONA*†, José GALLEGO*, Elena BENDALA‡, Igor YEFIMENKO*†‡,
Vicente RUBIO†‡2,3 and Javier CERVERA*†2,3
*Centro de Investigación Prı́ncipe Felipe (CIPF), Avda. Autopista del Saler 16, Valencia 46012, Spain, †Centro de Investigación Biomédica en Red para Enfermedades Raras
(CIBERER-ISCIII), C/Álvaro de Bazán, 10 Bajo, 46010 Valencia, Spain, and ‡Instituto de Biomedicina de Valencia (IBV-CSIC), Jaime Roig 11, Valencia 46010, Spain
NAG (N-acetyl-L-glutamate), the essential allosteric activator
of the first urea cycle enzyme, CPSI (carbamoyl phosphate
synthetase I), is a key regulator of this crucial cycle for ammonia
detoxification in animals (including humans). Automated cavity
searching and flexible docking have allowed identification of the
NAG site in the crystal structure of human CPSI C-terminal
domain. The site, a pocket lined by invariant residues and located
between the central β-sheet and two α-helices, opens at the
β-sheet C-edge and is roofed by a three-residue lid. It can tightly
accommodate one extended NAG molecule having the δ-COO−
at the pocket entry, the α-COO− and acetamido groups tightly
hydrogen bonded to the pocket, and the terminal methyl of
the acetamido substituent surrounded by hydrophobic residues.
This binding mode is supported by the observation of reduced
NAG affinity upon mutation of NAG-interacting residues of CPSI
(recombinantly expressed using baculovirus/insect cells); by the
INTRODUCTION
CPSs [CP (carbamoyl phosphate) synthetases] are complex
multidomain enzymes found in virtually all organisms [1]. These
allosterically regulated controlling enzymes synthesize CP, a
crucial compound for life, in the first committed step of the routes
of synthesis of pyrimidines, arginine and, in ureotelic vertebrates
(mainly amphibians and mammals), urea. In these latter animals
(including humans), CPSI, the enzyme that catalyses the first
step of the urea cycle, has an absolute requirement for the
allosteric activator NAG (N-acetyl-L-glutamate) [2,3]. NAG plays
a key role in controlling the flow of nitrogen through the urea
cycle [4–6], as exemplified by the deficiency of the enzyme
making NAG, NAG synthase, a deficiency that generally causes
severe neonatal hyperammonaemia [7] due to secondary CPSI
deficiency. Interestingly, there is a highly effective substitutive
therapy for NAG synthase deficiency, by using the deacylaseresistant NAG analogue NCG (N-carbamoyl-L-glutamate, also
called carglumic acid; proprietary drug name, Carbaglu® ) [7,8].
Although CPSI activation by NAG has been known for some
time [2], and despite the use of NCG in the treatment of NAG
fine-mapping of the N-chloroacetyl-L-glutamate photoaffinity
labelling site of CPSI; and by previously established structure–
activity relationships for NAG analogues. The location of the NAG
site is identical to that of the weak bacterial CPS activator IMP
(inosine monophosphate) in Escherichia coli CPS, indicating a
common origin for these sites and excluding any relatedness to
the binding site of the other bacterial CPS activator, ornithine. Our
findings open the way to the identification of CPSI deficiency
patients carrying NAG site mutations, and to the possibility
of tailoring the activator to fit a given NAG site mutation, as
exemplified here with N-acetyl-L(+
−)-β-phenylglutamate for the
W1410K CPSI mutation.
Key words: acetylglutamate, allosteric site, carbamoyl phosphate
synthetase deficiency, carglumic acid, N-carbamoyl-L-glutamate,
urea cycle defect.
synthase deficiency for nearly 30 years [7,8], the CPSI NAG site
has remained uncharacterized. Such characterization is a
requisite for understanding the mechanism of the extreme
allosteric activation of CPSI by NAG, and would have practical
interest for identifying CPSI deficiency resulting from NAG
site mutations. The patients carrying these mutations might be
amenable to treatment with very high doses of NCG or with NCG
analogues tailored to bind better to the mutant form of the site.
Until now, the only existing physical information about the
NAG site was its gross localization at the CPSI C-terminal domain,
deduced from the observation that this domain of ∼ 16 kDa
is the only part of CPSI (a polypeptide of ∼ 160 kDa) that is
photoaffinity labelled by the photoactivatable NAG analogue
[14 C]ClNAG (N-chloroacetyl-L-[14 C]glutamate) [9]. In fact, this
C-terminal domain is called the allosteric or regulatory domain in
all types of CPS because it is believed to be involved exclusively
in effector binding, relaying the regulatory information to the
catalytic domains (two homologous 40 kDa domains that catalyse
the phosphorylations of bicarbonate and carbamate) [10]. The
evidence for this conclusion was based on photoaffinity labelling
studies, first performed with CPSI [9], and then with eCPS
Abbreviations used: CAD, a trifunctional protein composed of carbamoyl phosphate synthetase II, aspartate transcarbamylase and dihydro-orotase;
ClNAG, N -chloroacetyl-L-glutamate; CP, carbamoyl phosphate; CPS, carbamoyl phosphate synthetase; eCPS, Escherichia coli carbamoyl phosphate
synthetase; hCPS, human carbamoyl phosphate synthetase; IMP, inosine monophosphate; MALDI–TOF, matrix-assisted laser-desorption ionization–
time-of-flight; MS/MS, tandem MS; NAG, N -acetyl-L-glutamate; NCG, N -carbamoyl-L-glutamate; Phe-NAG, N -acetyl-L(+
−)-β-phenylglutamate; PRPP,
phosphoribosyl pyrophosphate; rCPSI, rat carbamoyl phosphate synthetase I; rmsd, root mean square deviation.
1
Dedicated to Professor Santiago Grisola, who discovered CPSI and its activation by carbamylglutamate, and who, with V.R., prepared the first
carbamylglutamate given to a patient with NAGS deficiency.
2
These authors contributed equally to this work.
3
Correspondence may be addressed to either Javier Cervera (email [email protected]) or Vicente Rubio (email [email protected]).
c The Authors Journal compilation c 2009 Biochemical Society
212
S. Pekkala and others
(Escherichia coli CPS) [11–15]; on domain-deletion and domainexchange experiments carried out with eCPS and with CPSII
from hamster CAD [16,17] (a trifunctional protein composed of
carbamoyl phosphate synthetase II, aspartate transcarbamylase
and dihydro-orotase, which catalyses the three initial steps of
pyrimidine biosynthesis); on the observation that phosphorylation
of this domain of hamster CAD decreased UTP inhibition of
CPSII [18]; and on the determination of the crystal structures
of eCPS in complex with the inhibitor UMP [19] or with the
activators ornithine [20] or IMP (inosine monophosphate) [21].
However, the structural information gathered on eCPS has not
allowed study of where and how NAG binds to CPSI, given the
poor sequence identity (∼ 20 %) between the allosteric domains of
CPSI and eCPS [10], and because of the uncertainties on whether
the activations of both enzymes occur with similar mechanisms,
since eCPS is active in the absence of its effectors [22], whereas
CPSI is essentially inactive when NAG is not present [3]. In
addition, the two activators of eCPS, ornithine and IMP, bind to
different sites [20,21], raising the question of which one of these
sites, if any, would be analogous to the NAG site if the allosteric
domains of eCPS and CPSI were similar.
A unique opportunity for identifying the NAG site in CPSI was
provided recently when, as a part of a high-throughput structural
genomics/proteomics effort, the crystal structure of the C-terminal
domain of hCPSI (human CPSI), generated in isolated form in a
cell-free protein expression system, was deposited in the PDB
[23]. Using cavity searching and flexible docking we can now
identify the binding site for NAG in this structure. From the shape
of the binding pocket and from the chemical characteristics of
the residues lining this site, we define unambiguously the way
in which NAG binds to CPSI. This binding mode fits closely the
expectations derived from previous studies on structure–activity
relationships for NAG analogues [2,24,25], and is supported by
experimental evidence reported in the present work: the finemapping of the [14 C]ClNAG photoaffinity-labelling site within
CPSI, and the results of site-directed mutagenesis experiments
involving residues predicted to interact with the bound activator.
The implications of our findings for the diagnosis and treatment of
CPSI deficiency are discussed, highlighting with one experimental
example the possibility of using modified NAG analogues to
improve activation of defective CPSI forms carrying specific
mutations in the NAG site. Finally, the proposed NAG-binding
mode is compared with those of the bacterial activators IMP and
ornithine, discussing possible evolutionary origins of the NAG
site.
EXPERIMENTAL
Identification of the NAG site and docking of the NAG molecule
The structure deposited in the PDB as file 2YVQ [23],
corresponding to the C-terminal domain of hCPS (residues 1343–
1478), was analysed for the presence of cavities with VOIDOO
(version 3.3.4) [26]. NAG was docked into the pocket identified
with VOIDOO using the GOLD (version 4.0.1) package [27], a
genetic algorithm for docking flexible ligands into protein-binding
sites that has been optimized for the prediction of ligandbinding positions. Docking calculations were unrestrained and
used the GoldScore fitness function. The preference of NAG for
binding in this pocket was confirmed by GOLD docking runs
involving the entire hCPS C-terminal domain or by docking
NAG into alternative hCPS C-terminal domain cavities. GOLD
calculations indicated the need for movement of the Trp1410
residue forming the pocket lid (see below) for full NAG entry.
Subsequent docking runs in which the side chains of Trp1410 and
c The Authors Journal compilation c 2009 Biochemical Society
the overlying Phe1445 and Gln1413 residues were allowed to rotate
freely yielded a highly stable binding pose in which the Trp1410
ring had flipped 180◦ while maintaining the stacking interaction
with the neighbouring Phe1445 and keeping a commonly observed
side chain conformation [28]. Subsequent GOLD calculations
using this alternative protein conformation led to very good NAG
docking scores, with almost total convergence (95 %) of the
NAG-binding poses [19 of 20 solutions with an rmsd (root
mean square deviation) conformational deviation of 1.35 Å (1
Å = 0.1 nm) or less; see below]. The final hCPS–NAG model was
generated by refining further the best GOLD solution by restrained
energy minimization using the ff03 force field [29] of AMBER 8.0
[30], and a generalized Born model for solvent simulation. This
last minimization used as restraints the protein–NAG hydrogen
bonds detected in the converged GOLD binding poses (see below).
Structural analysis and preparation of Figures
Structural superimpositions were performed with the LSQKAB
option of the CCP4 suite [31]. Solvent accessibility (probe radius,
1.4 Å) of residues within the eCPS (bound to IMP; PDB number
1CE8 [21]) and hCPS (PDB number 2YVQ [23]) structures was
calculated with the AREAIMol program of CCP4 [31]. Surface
potential was calculated with GRASP [32]. Multiple sequence
alignments were prepared with CLUSTALW [33], using default
parameters. Figures depicting protein structures were prepared
with PyMOL (DeLano Scientific; http://www.pymol.org).
Fine-mapping of the rCPSI (rat CPSI) photoaffinity-labelling site
Purification of rat liver CPSI, synthesis of ClNAG (unlabelled
or labelled with [14 C] in the glutamate moiety) and photoaffinity labelling of the enzyme have been reported previously
[9]. Limited proteolysis with V8 staphylococcal protease, HPLC
isolation of the resulting 18.9 kDa C-terminal labelled fragment,
and electrophoresis, fluorography and Edman sequencing were as
previously reported for [14 C]UMP-labelled eCPS [11]. Digestion
of the 18.9 kDa labelled fragment with either CNBr [35] or
trypsin (alone or followed by the V8 protease under conditions at
which the V8 protease cleaves only at glutamate residues [12,36])
have also been previously reported for [14 C]UMP-labelled eCPS
[12]. o-Iodosobenzoic acid cleavage of the 18.9 kDa labelled
fragment was carried out in the dark at 21◦C for 24 h according
to [37], in a volume of 50 μl. The reaction was terminated
with 1 μl of 1 mM dithioerythritol and the mixture was desalted
by centrifugal gel filtration [38] through a 1 ml Sephadex G25 column equilibrated with 50 % (v/v) acetonitrile/0.1 % (v/v)
trifluoroacetic acid. After removal of the solvent by freeze-drying,
the residue was subjected to SDS/urea phosphate PAGE using a
formulation [12] suitable for small peptides. For determination
of the mass of tryptic or tryptic/V8-protease fragments, the
Sephacryl S200-SDS gel-filtration technique [12] was used. The
elution position of each component is given as the distribution
coefficient, K d = (V e −V 0 )/(V i −V 0 ), taking V 0 , V i and V e as the
volumes of elution of BSA, mercaptoethanol and the component
of interest respectively.
For MS analysis, the enzyme was subjected as described
above to UV irradiation in the presence of either ClNAG
(experimental irradiation) or NAG (control irradiation) followed
by limited proteolysis with the V8 protease. After SDS/PAGE,
the Coomassie-stained 18.9 kDa band corresponding to the Cterminal domain of the enzyme from the experimental and control
samples was excised from each gel and was subjected to peptide
fingerprinting using trypsin (Promega) [39] and MALDI–TOF
CPSI site for acetylglutamate
(matrix-assisted laser-desorption ionization–time-of-flight) MS
analysis in a 4700 Proteomics Analyzer (Applied Biosystems)
used in linear mode, with concomitant MS/MS (tandem MS)
analysis of relevant peaks to acquire sequence information. The
MS and MS/MS information were analysed with MASCOT [40].
The MS analysis was carried out by the Proteomics Core Facility
of the Centro de Investigación Prı́ncipe Felipe (CIPF Valencia)Proteored (Genoma España).
Recombinant rCPSI production and mutagenesis
Recombinant rCPSI was produced using a commercial kit (Bacto-Bac® Baculovirus Expression System; Invitrogen). The rCPSI
cDNA carried by pHN3491 [41] was subcloned into the pFastBacHTA vector provided in the kit, using standard techniques for
DNA manipulation and for engineering the cDNA to replace
the CPSI mitochondrial targeting sequence by the 6 × His,
28-residue, N-terminal tag MSYYHHHHHHDYDIPTTENLYFQGAMDP, provided by the vector. After transformation of E.
coli Max Efficiency DH10Bac cells (Invitrogen), the recombinant
bacmid was recovered and transfected into Sf9 insect cells using
Cellfectin. The resulting virus was then used to infect Sf9 cells.
The best expression (monitored by SDS/PAGE) was obtained at a
virus-to-cell ratio of 2, by incubation at 27 ◦C for 65 h. Subsequent
steps were at 4 ◦C. After centrifugation of a 50 ml culture, the
cells were suspended in 3 ml of 50 mM glycyl-glycine, pH 7,
20 mM KCl, 20 % glycerol, 0.1 % Triton X-100, 5 mM imidazole,
0.1 % histidine-tagged specific protease inhibitor mixture and
5 μM cysteine protease irreversible inhibitor E-64 (the last two
components were from Sigma). The cell suspension was frozen
(in liquid nitrogen) and thawed twice, then centrifuged, and
the supernatant was mixed with 0.5 ml packed TALON resin
(Clontech) equilibrated in the same solution without Triton X100 or protease inhibitors. After 30 min gentle end-to-end mixing
and two washes with 5 ml of equilibration solution containing
10 mM imidazole, the resin was packed in a column and was
eluted with equilibration solution containing 0.15 M imidazole.
Dithiothreitol (1 mM) was added to the column effluent, and
the fractions containing CPSI (monitored by SDS/PAGE) were
pooled and concentrated to > 1 mg of protein/ml by centrifugal
ultrafiltration (100 kDa cut-off Amicon Ultra; Millipore).
Site-directed mutagenesis was performed by the overlapping
extension method using a commercial kit (QuikChange;
Stratagene), the pFastBac-HTA vector containing the CPSI cDNA
as template, a high-fidelity thermostable DNA polymerase (Pfu
Turbo from Stratagene) and the appropriate mutagenic forward
and reverse oligonucleotides (see Supplementary Table S1 at
http://www.BiochemJ.org/bj/424/bj4240211add.htm). Mutations
were either to alanine (T1394A) or to residues expected
to cause the smallest possible structural change (N1437D,
N1440D and T1391V). The W1410K mutation introduces the
residue found in bacterial CPSs (see Supplementary Figure S1
at http://www.BiochemJ.org/bj/424/bj4240211add.htm). Mutant
plasmids were sequenced to assure the respective presence and
absence of the desired and unwanted mutations. Production of the
mutant proteins by the insect cells and mutant protein purification
were as for the wild-type enzyme.
Enzyme activity assay
CPSI activity was assayed at 37 ◦C in a solution containing 50 mM
glycyl-glycine, pH 7.4, 0.1 M KCl, 0.1 M NH4 Cl, 5 mM ATP,
7 mM MgSO4 , 20 mM KHCO3 , 1 mM dithiothreitol, 5 mM Lornithine, 10 units/ml ornithine transcarbamylase (Sigma) and
213
NAG as indicated or, when indicated, Phe-NAG [N-acetyl-L(+
−)β-phenylglutamate] (prepared as described in [25]), determining
citrulline after 10 min, and, when indicated, ADP [3,42]. Results
of at least duplicate assays were fitted to hyperbolic kinetics with
GraphPad Prism (GraphPad Software).
Other assays
CD spectra in the far-UV region (195–250 nm) were obtained
at 4 ◦C with a Jasco 810 spectropolarimeter in 0.2 ml of
20 mM Tris/HCl, pH 7.4, 20 mM KCl, 20 % glycerol, 0.5 mM
dithiothreitol and 1–2 μM protein in a 0.1 cm-pathlength cell.
Each spectrum was the average of ten scans. Protein was
determined by the Bradford assay [43] using BSA as standard.
RESULTS AND DISCUSSION
Proposed site and mode of NAG binding in CPSI
Since the structure of the C-terminal domain of hCPS (PDB
file 2YVQ [23]) was obtained using the acellularly produced
isolated domain without the natural context provided by the
remaining domains of the protein, there could be doubts on
whether the structure is the genuine one or an artefact. The
latter possibility appears to be excluded by the observation that,
despite the poor sequence identity with the allosteric domain of
eCPS (20 % identity; see Supplementary Figure S1), the structure
of the two domains is virtually identical (rmsd of 1.65 Å for
superimposition of 108 Cα atoms; Supplementary Figure S2
at http://www.BiochemJ.org/bj/424/bj4240211add.htm): an αβα
open sandwich in which the central parallel sheet of five elements
(strand order, 32145) is covered on one side by helices 2, 1 and 5
and on the other by helices 3 and 4 (Figure 1A).
A previous study [44] demonstrated the binding of one NAG
molecule per subunit of rat liver CPSI. Using the hCPS Cterminal domain structure [23], and utilizing automated cavitysearching [26] and ligand-docking [27] algorithms, we only
detected one pocket containing the approximate dimensions
required to accommodate the NAG molecule (Figures 1A and 1B).
The pocket, opening at the sheet C-edge, is located between the β
sheet and the helical layer containing helices 3 and 4 (Figure 1A).
In this pocket, the indolic ring of Trp1410 , along with Phe1445 and
Lys1444 , forms a lid separating the cavity from the surrounding
solution (Figure 1B). Initial docking experiments indicated that,
in the rotamer observed in the crystal structure of the empty
pocket, the Trp1410 side chain would hamper optimal fitting of
NAG inside the cavity. Thus, when flexibility of the side chains of
Trp1410 and the overlying Phe1445 and Gln1413 residues was allowed,
a highly stable binding pose was detected (Figure 1D). In this
conformation, the Trp1410 aromatic ring has flipped 180o around the
Cβ–Cγ bond relative to the conformation observed in the crystal
structure of the hCPS C-terminal domain, while maintaining the
stack with Phe1445 (Figure 1D). This alternative Trp1410 side-chain
conformation is commonly observed in proteins [28]. Subsequent
calculations using this alternative protein conformation led to
much better NAG docking scores and to almost total convergence
(95 %) of the NAG-binding poses (Figure 1D).
In the converged solutions (Figure 1E), the NAG molecule
adopts a largely extended conformation and places its α-COO−
and acetamido groups oriented towards one wall and the floor
of the pocket respectively, whereas the glutamate side chain
is oriented towards the pocket entry. This NAG-binding mode
results in appropriate interactions with the residues lining the
cavity: the α-COO− is encircled by the Asn1440 and Asn1449
c The Authors Journal compilation c 2009 Biochemical Society
214
Figure 1
S. Pekkala and others
The C-terminal domain of CPSI and the NAG site
(A) Representation of the hCPS allosteric domain (PDB file 2YVQ), coloured to enhance visibility of the structure. Secondary structure elements are labelled. The pocket identified with VOIDOO
is shown in semi-transparent surface representation. An NAG molecule bound to this pocket according to the results of docking with GOLD (see the text) is shown in a ‘sticks’ representation.
(B) and (C) Surface representations of the NAG site as seen in the crystal structure of hCPS C-terminal domain (PDB file 2YVQ) (B), and the IMP site of eCPS (PDB file 1CE8 [21]) (C), in identical
orientations. In (B), the side chains of the NAG pocket roof residues Lys1444 , Phe1445 and Trp1410 are shown in a sticks representation to allow visualization of the pocket interior. In (C), the IMP
activator is shown in a sticks representation. (D) NAG docking to hCPSI. The side chains proposed to surround the NAG molecule are shown in a thin stick representation, in the conformations
observed in the crystal structure of the empty NAG site (PDB file 2YVQ, black), superimposed with the conformations resulting from GOLD flexible docking of NAG (in blue). The converged NAG
binding poses are shown by thicker sticks. (E) Stereoview of the proposed hCPS site bound to NAG. Secondary structure elements are labelled. (F) Schematic plot (drawn with LIGPLOT [51]) of the
interactions between the protein and NAG. Distances are in Å. (G) Stereoview of the superimposed side chains (thinner sticks) of the residues making the IMP site (green carbons) and the proposed
NAG site (black carbons) in eCPS and hCPSI respectively, with the bound IMP and NAG molecules shown with green and black carbons respectively.
c The Authors Journal compilation c 2009 Biochemical Society
CPSI site for acetylglutamate
side-chain amide groups on one side and by the NH group of the
flipped Trp1410 ring on the other, forming with them a network of
hydrogen bonds, while the α-NH group forms a hydrogen bond
with the Thr1391 side-chain (Figure 1F). The acetamido group
has its methyl end in a hydrophobic nest formed by Leu1363 ,
the γ -methyl of Thr1391 , Trp1410 , Ile1423 and Ile1452 (Figure 1E),
whereas its carbonyl group points towards the Asn1437 side-chain
amide, with which it forms a hydrogen bond (Figure 1F). The
-OH groups of Thr1391 and Thr1394 are close to the outward-facing
NAG δ-COO− , donating hydrogen bonds to it (Figure 1F). The
positive surface potential at the pocket entry (estimated with
GRASP [32]; not illustrated) fits the binding of the negatively
charged δ-COO− . Finally, the Cβ atom of the glutamate side
chain also faces outwards, making contacts with the indolic ring
of Trp1410 (Figure 1E). These proposed NAG-interacting residues
are strictly conserved in CPSs, with some of them (Trp1410 , Asn1440
and Asn1449 , see Supplementary Figure S1) being conserved
exclusively in NAG-sensitive CPSs, supporting the functional
importance of this pocket and its implication in NAG binding.
The proposed mode of NAG binding agrees with prior
structure–activity studies with NAG analogues
A structure–activity study [25] concluded that NAG binds, with
its glutamate moiety extended in a low-energy conformation, to
a CPSI site containing the subsites for the α-COO− and for the
acetamido group of NAG fixed relative to each other (deduced
from the abolition of binding by adding a methylene group
between the asymmetric C atom and the α-COO− of NAG [25]),
whereas the subsite for the δ-COO− appeared less constrained.
All these traits are reflected in our model in the present study for
NAG binding: the NAG molecule is extended, the subsites for the
α-COO− and acetamido groups involve the internal regions of
the pocket and the δ-COO− binds at the pocket opening
(Figure 1E), which should offer an sterically less constrained
context. Prior observations that thiocarbamoylglutamate does not
activate CPSI [2], but that N-acetoxyglutarate is an activator
[24], also agree with our present proposal (Figures 1E and 1F),
since in our model the O atom of the NAG acetamido group is
appropriately placed for forming one hydrogen bond with the
enzyme. The inactive N-acetyl, N-methyl-L-glutamate analogue
[25] would be unable to form the proposed hydrogen bond
between the αN and Thr1391 (Figures 1E and 1F), in addition
to causing steric clash with the site. The inability of N-butyrylL-glutamate [2] to activate CPSI parallels the observation of lack
of space within the binding pocket (Figure 1E) to accommodate
the N-butyryl instead of the N-acetyl group. In contrast, little
reshuffling would be needed for replacing the N-acetyl group
of NAG by the modestly larger N-propionyl or N-chloroacetyl
groups and, indeed, the corresponding NAG analogues are quite
good activators of CPSI [25]. The structure also accounts for
the binding of NCG, given the similar size and overall shape of the
carbamoyl and acetyl groups. However, the higher polarity and
different geometry of the NH2 in the carbamoyl group relative
to the CH3 of the acetyl substituent would be expected to impair
somewhat the interactions with the hydrophobic cavity that hosts
this methyl group of NAG, explaining the ∼ 10-fold lesser affinity
of CPSI for NCG relative to NAG [25]. The importance of the
hydrophobic interactions with the methyl group are illustrated by
the observation that the affinity of CPSI for N-formyl-L-glutamate,
which lacks this methyl group, is two orders of magnitude lower
than for NAG [25]. Finally, in agreement with the prediction from
our model that the glutamate side chain of NAG faces towards
the pocket entry, and, in particular, that the NAG Cβ atom is
partially exposed and contacts the pocket roof residue Trp1410 , the
215
NAG analogue with a phenyl substituent at the Cβ , Phe-NAG,
was a good CPSI activator [25]. Thus the extra phenyl group of
Phe-NAG can be accommodated in this outward-facing part
of the site, possibly by reshuffling the roof over the site somewhat.
Overall, the evidence derived from structure–reactivity studies of
NAG analogues supports the binding of NAG in the way proposed
in the present study.
Mapping of the site of CPSI photoaffinity labelling by
N -chloroacetyl-L-glutamate supports the proposed mode of NAG
binding
rCPSI, an enzyme that is functionally virtually identical to hCPSI
[45] and which has 94.4 % sequence identity with it [46], was
photoaffinity labelled in its ∼ 16 kDa C-terminal domain with the
radioactive and photosensitive NAG analogue [14 C]ClNAG [9].
Given its closeness to the methyl group of NAG (Figure 1E), the
-OH group of Thr1391 appears a good candidate for forming the covalent bond with the NAG radical generated upon irradiationtriggered Cl atom loss [9]. Because of the low efficiency of the
labelling (∼ 5 % of enzyme molecules [9]), direct identification
of the labelled residue has not been possible thus far. However,
the results of specific protein cleavage techniques agree with the
cross-linking of [14 C]NAG to Thr1391 . These studies have been
carried out with the 18.9 kDa labelled C-terminal fragment
produced by limited proteolysis with the V8 staphylococcal
protease (Figures 2A and 2B), a fragment which was proven
by N-terminal sequencing to begin at Met1329 (Figure 2B). Upon
CNBr cleavage at methionine residues [35] (Figures 2A and 2C),
the radioactive fragment exhibited an electrophoretic mass of
∼ 17 kDa, as expected (Figure 2A, second bar) for the fragment
that is C-terminal to the most distal methionine, Met1350 . Upon
o-iodosobenzoic treatment to cleave at tryptophan residues [37]
(Figures 2A and 2C), the radioactivity migrated according to a
molecular mass, estimated by SDS-urea/PAGE [12] (a suitable
technique for small fragments), of 7.3 kDa. This value fits only the
mass (7.72 kDa) of the most N-terminal (residues 1329–Trp1397 )
of the three fragments that should be generated upon cleavage
at tryptophan residues (additional two fragments, 1398–Trp1410
and 1411–1500, with respective masses of 1.4 and 10 kDa). Thus
the enzyme is photolabelled upstream of the o-iodosobenzoic
cleavage point at Trp1397 and downstream of the CNBr cleavage
point at Met1350 , in the region between residues 1351 and
1397.
We showed [9] that exhaustive tryptic digestion of [14 C]ClNAGlabelled reCPSI yields a single labelled fragment, as indicated
by HPLC peptide mapping (Figure 2D). The same minimal
labelled tryptic peptide was obtained (given the identical elution
position of the radioactivity from reversed-phase HPLC, results
not shown) by digestion of the whole enzyme and of the 18.9 kDa
C-terminal V8 fragment, indicating that this labelled minimal
tryptic peptide is contained entirely within the 18.9 kDa Cterminal V8 fragment (as expected, since the enzyme is labelled
at some point between residues 1351 and 1397, far from the Nend of the 18.9-kDa V8 fragment). This labelled tryptic peptide
behaves in gel filtration [12] as expected for the mass (4.16 kDa)
of the minimal tryptic peptide (residues 1388–1424) that contains
Thr1391 (Figures 2A and 2F). All other minimal tryptic peptides
covering the 1351–1397 region (Figure 2A, fourth bar) are much
smaller. The minimal tryptic peptide that contains Thr1391 includes
two glutamate residues and thus should be cleaved by the V8
staphylococcal protease [36]. If the [14 C]NAG is cross-linked to
Thr1391 , the radioactive peptide produced by sequential tryptic V8
digestion should have a mass of 0.77 kDa. Both predictions are
c The Authors Journal compilation c 2009 Biochemical Society
216
Figure 2
S. Pekkala and others
Mapping of the site of photoaffinity labelling of CPSI with the NAG analogue ClNAG
Except in panel (G), in which the analogue was not radioactive, all other panels reflect the results obtained after photoaffinity labelling with [14 C] ClNAG. (A) Cleavage techniques used to identify
the region where NAG is bound. Bars represent enzyme polypeptides, cleavage points are indicated with scissors and are localized with black vertical lines, with the cleavage site residue number
below the bar. The radioactive fragment obtained in each step is coloured grey. Numbers within the bars are sequence-deduced fragment masses (in kDa), including the expected mass increase (188
Da) if NAG was cross-linked to the fragment. The shadowed bar regions are those limited by the CNBr and o -iodosobenzoic cleavage points (see the text). Letters on the right of the horizontal bars
give the panels illustrating the results of each cleavage technique. (B) HPLC isolation of the C-terminal labelled fragment obtained after limited digestion with the V8 staphylococcal protease of the
photoaffinity-labelled enzyme. The insets show the results, revealed by Coomassie Blue staining or by fluorography, of SDS/PAGE of selected peaks. (C) SDS/PAGE and fluorography of the cleavage
c The Authors Journal compilation c 2009 Biochemical Society
CPSI site for acetylglutamate
satisfied by the present experimental evidence (Figures 2E and
2F), which reveals: (1) a change in the elution of the radioactive
peak from the reversed-phase HPLC column upon V8 digestion
of the tryptic fragment; and (2) the elution of this radioactivity
from the gel filtration column at the expected position for
0.77 kDa. The only other peptide (residues 1393 to 1414) falling
in the 1351–1397 region (the region encompassing the label) that
is generated by the V8 protease from the minimal tryptic peptide,
has a much larger mass (2.56 kDa) and thus does not fit the gel
filtration results. Taken together, all these observations add up to
render it highly probable that Thr1391 is the site of photoaffinity
labelling by [14 C]ClNAG.
We also obtained independent evidence using MS for the
labelling by ClNAG of the minimal tryptic peptide encompassing
residues 1388–1424. MS fingerprinting with trypsin (see the
Experimental section) confirmed the assignment of the 18.9 kDa
fragment generated by the V8 protease to the C-terminal domain
of the enzyme, and identified the tryptic fragment encompassed
between residues 1388 and 1424 by both its mass (calculated
mass, 3967.93 Da; experimental mass in two determinations,
3968.24 and 3968.49 Da) and by sequence information obtained
by MS/MS. This tryptic peptide was detected (Figure 2G) in
similar amount (judged from the peak size) both in the fingerprint
of the 18.9 kDa fragment generated after control irradiation in
the presence of NAG (NAG lacks the Cl substituent and thus it
is not photosensitive and does not photolabel the enzyme [9]) or
in the experimental irradiation with ClNAG, in line with the fact
that a very small fraction of the protein is labelled with ClNAG
[9]. Nevertheless, in the tryptic digest of the sample that was
photoaffinity labelled with ClNAG, but not in that in which the
ClNAG was replaced by NAG, a small peak was observed with
a mass of 4155.94 Da (Figure 2G), which corresponds, within
experimental error, to the expected mass (4155.09 Da) of the
1388–1424 tryptic fragment modified by cross-linking to NAG.
The small amount of this peptide, which agrees with the wellknown low-labelling efficiency by ClNAG [9], prevented further
analysis. Nevertheless, the mass of the peak and its exclusive
finding in the experimental sample agrees with the conclusion that
the 1388–1424 tryptic peptide is the one where ClNAG labelling
takes place.
Site-directed mutagenesis studies confirm NAG site identification
Using a baculovirus-based insect cell expression system, we
were successful in producing mature rCPSI (that is, the rat
liver enzyme lacking the N-terminal mitochondrial targeting
sequence) in active form, engineered to incorporate a His6 tag
at its N-terminus for rapid purification. Under optimal expression
conditions (see the Experimental section), ∼ 0.5 mg of largely
homogeneous (Figure 3A) active (specific activity, 1.3 +
− 0.08
μmol of CP · min−1 · mg−1 ) rCPSI was obtained from 50 ml
of culture after a single step of Co2+ chelation chromatography.
The enzyme exhibited normal substrate kinetics (K m values for
217
ATP, HCO3 − and NH4 + , 1.06+
−0.11 mM, 6.43 +
− 0.60 mM and
NAG
1.07 +
− 0.1 mM respectively) and NAG activation kinetics (K a
0.11 +
0.01
mM,
Figure
3A).
−
We generated enzyme forms carrying single amino acid
mutations affecting NAG site residues predicted to be important
for NAG binding. The mutant proteins were produced in the same
way and with similar yield and purity as in the case of wild-type
CPSI (Figure 3B), and appeared well folded (judged from the
similarity of their far-UV CD spectra relative to the spectrum of
wild-type CPS; results not shown). In agreement with the fact
that the catalytic machinery is not provided by the C-terminal
domain [10,13,20], all the mutants were active in the presence
of NAG, producing CP and ADP with the normal ADP/CP
product ratio of 2 (results not shown). However, all mutations
significantly increased the concentrations of NAG required for
activation (Figure 3C), as expected for residues involved in NAG
binding. Of particular note was the increase of ∼ 400-fold and
∼ 900-fold in the K a NAG triggered respectively by the T1391V
and T1394A mutations, supporting the role of the hydroxy groups
of these threonine residues in NAG binding (Figures 1E and
1F). The increase of ∼ 70–110-fold in the K a NAG caused by the
N1437D and N1440D mutations also agrees with the role of these
asparagine residues as hydrogen bond donors to the NAG αCOO− and acetamido carbonyl (Figures 1E and 1F). Finally, the
W1410K mutation, chosen because it replaces a key pocket roof
residue by the highly conserved lysine found at this position in
bacterial CPSs, increased the K a NAG ∼ 60-fold (Figure 3C). This
result is consistent with the observed partial hampering of NAG
entry by the Trp1410 ring (see above).
The rate of the enzyme reaction for the T1391V and T1394A
mutants, at saturation of NAG, was a third relative to wild-type
CPSI (Figure 3C), a decrease that appears not to be due to enzyme
inactivation, since these mutant rCPSI forms were as stable as
wild-type CPS at 4 ◦C, 25 ◦C and 37 ◦C (results not shown).
Thus the enzyme–NAG complex appears genuinely less active
for these mutants than for the wild-type enzyme, suggesting the
involvement of Thr1391 and Thr1394 in allosteric signal transmission
to the catalytic domains. Since these residues bind to the δ-COO−
of NAG (Figures 1E and 1F), their involvement in allosteric signal
transmission agrees with the previous conclusion [25] that the
subsite for δ-COO− is involved in the conformational changes
associated with NAG activation.
An unexpected finding of the present study is the observation
that the W1410K mutation resulted in a modified specificity for the
activator. Thus with this mutant, the affinity was increased when
Phe-NAG was used instead of NAG (Figure 3D). Phe-NAG was
reported [25] and is confirmed in the present study (Figure 3D)
to activate rCPSI with a K a that is ∼ 2-fold higher than for NAG,
but the W1410K mutant binds Phe-NAG considerably better than
NAG (respective K a values, 0.55 and 6.3 mM; Figure 3D, insert).
As discussed below, this finding, which supports the interaction
of the NAG Cβ atom with Trp1410 , may have practical implications
in the treatment of CPSI deficiency.
products obtained by treating the 18.9-kDa labelled fragment isolated in (B) with o -iodosobenzoic acid (left) or with CNBr (right). The minus sign indicates no treatment. The masses of the radioactive
products obtained indicated with dashed lines are in italics; St denotes mass standards with values to the left and right of the respective panels. (D and E) HPLC separation of the radioactive
fragments obtained by (D) exhaustive tryptic digestion of the 18.9-kDa labelled fragment isolated in (B) and (E) by subsequent digestion with the staphylococcal V8 protease. (F) Gel-filtration
determination (Sephacryl S200-SDS technique) of the mass of the labelled minimal tryptic peptide (䊏) or of the radioactive product generated by subsequent digestion with V8 protease (䊉). The
samples applied to the column are the radioactive peaks of panels (D) and (E). Radioactivity was determined in the column effluent (see the Experimental section for details). 䊊, mass standards.
(G) Linear MALDI–TOF spectra of trypsin-digested C-terminal domain of the enzyme. The numbers given correspond to the experimental masses (in Da) for the peak corresponding to the minimal
tryptic peptide that contains Thr1391 , and for the peak proposed here to correspond to this same peptide labelled with NAG (see the text). Upper panel: spectrum of ClNAG-irradiated sample. Lower
panel: spectrum of control sample.
c The Authors Journal compilation c 2009 Biochemical Society
218
S. Pekkala and others
Implications of the present studies for clinical CPSI deficiency
The diagnosis of genetic diseases ideally aims at establishing
the disease-causing potential of each amino acid substitution
found in a patient presenting the disease, thus excluding that
the mutations found could merely be trivial low-frequency
polymorphisms having little impact on enzyme functionality or
stability. The robust baculovirus expression/purification system
developed here provides a well-suited system for carrying out
these pathogenicity-testing studies in the case of human CPSI
deficiency, and, therefore, can be valuable in the diagnosis and
evaluation of this genetic condition. In addition, the present
characterization of the NAG site also has practical potential for
identifying mutations exclusively affecting NAG activation. A
number of amino acid substitutions have been reported already
in patients of CPSI deficiency that are candidates to affect NAG
regulation, since they fall in the C-terminal domain of the enzyme
[47–49]. We have begun a study aiming at applying the present
knowledge to predict the potential effects of these C-terminal
domain CPSI mutations, carrying out in parallel expression
studies to test the quality of our structure-based predictions.
Hopefully, the knowledge of the C-terminal domain structure
and on the NAG site will provide in the near future the answer
to whether a mutation is likely to impair NAG binding. In
such mutations the CPSI catalytic machinery should be intact
and it would appear possible, at least in principle, to restore
enzyme activity by the administration of large enough doses of
NCG [7,8]. Moreover, our observation that the modified NAG
analogue Phe-NAG is a better activator than NAG for W1410K
rCPSI suggests that N-carbamoyl-L-β-phenylglutamate would be
a better treatment than NCG in the case that a CPSI deficiency
patient were found to carry the W1410K mutation. Clearly, our
findings raise the novel possibility of synthesizing modified NCG
analogues specially tailored to fit individual NAG site mutations
for the treatment of CPSI deficiency.
The origin of the NAG site of CPS
Comparison of the structures of the allosteric domains of hCPSI
and eCPS reveals that the NAG site of CPSI is topographically
equivalent to the site for the relatively poor activator of eCPS, IMP
(Figures 1B, 1C and 1G and Supplementary Figures S1 and S2).
The IMP site of eCPS is an angle-shaped open furrow formed by
the same structural elements that also form the NAG site in CPSI
(Figure 1C). In the bacterial enzyme, the distance between the
N-ends of α-helix 4 and the β3–α3 connector is larger than
in hCPSI, leaving a space where the purine base of IMP sits. In
hCPSI, this space is covered by a roof formed by the mutually
stacked bulky side chains of (from outwards to inwards) Lys1444 ,
Phe1445 and Trp1410 (Figure 1B), of which Trp1410 is invariant in
NAG-sensitive CPSs (Supplementary Figure S1). Because of this
Figure 3
studies
Production of recombinant CPSI, and site-directed mutagenesis
(A) Recombinant (reCPSI) and liver-purified (liCPSI) rCPSI exhibit identical NAG activation
kinetics. Velocity is expressed as a fraction of the velocity extrapolated at an infinite NAG
concentration. The insets illustrate SDS/PAGE analysis (Coomassie Blue staining) of CPSI
expression. Left panel, crude centrifuged extracts of the insect cells infected with the baculovirus
c The Authors Journal compilation c 2009 Biochemical Society
lacking or carrying the CPSI coding sequence (lanes labelled Control and reCPSI respectively;
St denotes mass standards). Right panel, purified recombinant enzyme compared with the
enzyme purified from rat liver. (B) SDS/PAGE (Coomassie Blue staining) of the purified mutant
CPSI forms prepared here. (C) Effects of the mutations on CPSI activation by NAG. Velocities
are given relative to those for the same concentration of the wild-type enzyme. The numbers
to the right correspond to estimates from each curve for the K a NAG (given in mM units) and
for the velocity at infinite NAG concentration (given relative to that of wild-type). (D) Influence
of the W1410K mutation (insert) on activation by NAG and by the NAG analogue Phe-NAG.The
larger plot illustrates the activation kinetics for NAG and its analogue on the wild-type enzyme.
The inset reveals that the analogue is a better activator than NAG for the W1410K mutant CPS.
Velocities are expressed as for panel (C).
CPSI site for acetylglutamate
roof, the part of the site that in eCPS binds the purine base of IMP
is obliterated and the NAG site becomes a pocket.
The similarity between the NAG and IMP sites also extends to
the interactions made by IMP and NAG (Figure 1G). The δ-COO−
of NAG can be equated with the IMP phosphate, a phosphate that,
in eCPS, makes hydrogen bonds with the side chains of Thr974 and
Thr977 , corresponding to Thr1391 and Thr1394 of CPSI. The two IMP
ribose hydroxy groups appear to be antecedents of both O atoms
of the NAG α-COO− , since they are encircled by Asn1015 , Thr1017
and Ser1026 , corresponding to the three hCPS asparagine residnes,
Asn1437 , Asn1440 and Asn1449 , that encircle in CPSI the NAG αCOO− (Figure 1G). Even the indolic ring of Trp1410 is reminiscent
(Figures 1B, 1C and 1G) of the purine base of bound IMP, as it
is in an approximately equivalent place, although in a different
orientation, than the purine base of the nucleotide effector of
bacterial CPS. The similarity of the binding of IMP to eCPS and
of NAG to CPSI is also highlighted by the overall coincidence, in
the alignment of the sequences of both enzymes, of the residues
that change their exposition with the binding of the corresponding
effector (Supplementary Figure S1, black dots).
The structure of eCPS revealed the presence of different sites
for its two activators ornithine and IMP [20,21]. Our results, by
proving the equivalence of the NAG site of CPSI and the IMP
site of bacterial CPS, exclude a relationship between the NAG
and the ornithine sites and indicate that the NAG and IMP sites
derive from the same ancestral site. Since PRPP (phosphoribosyl
pyrophosphate) is an IMP analogue, and shares with IMP a
phosphoribosyl moiety, the PRPP site of the pyrimidine-specific
CPSII of CAD [50] most likely derives from the same ancestral
site. In fact, CAD contains a tryptophan residue that aligns with
Trp1410 of CPSI (Supplementary Figure S1), and thus the PRPP
site of CAD might resemble the NAG site of CPSI by being a
pocket rather than a furrow and by having the indolic ring of this
tryptophan in the site in which in eCPS is occupied by the adenine
ring of IMP, a ring that is missing in PRPP.
AUTHOR CONTRIBUTION
All the authors performed the research. Javier Cervera and Vicente Rubio designed the
research, analysed the data and wrote the paper.
ACKNOWLEDGEMENTS
We thank José Luis Llácer (IBV-CSIC) for help with Figures and structural calculations, the
Proteomics Core Facility of the Centro de Investigación Prı́ncipe Felipe (CIPF Valencia)Proteored (Genoma España) for diligent mass spectrometric service, and Professor Santos
Fustero and his group (Laboratorio de Moléculas Orgánicas, CIPF Valencia, Spain) for
making one lot of non-labelled ClNAG.
FUNDING
This work was supported by the Spanish Ministry for Science (MEC and MICINN) [grant
numbers BFU2007-66781 and BFU2008-05021]; and by the Valencian Government
[grant number AP-035/08].
REFERENCES
1 Rubio, V. and Cervera, J. (1995) The carbamoyl-phosphate synthase family and
carbamate kinase: structure-function studies. Biochem. Soc. Trans. 23, 879–883
2 Grisolı́a, S. and Cohen, P. P. (1953) Catalytic role of glutamate derivatives in citrulline
biosynthesis. J. Biol. Chem. 204, 753–757
3 Rubio, V., Britton, H. G. and Grisolı́a, S. (1983) Mitochondrial carbamoyl phosphate
synthetase activity in the absence of N-acetyl-L-glutamate. Mechanism of activation by
this cofactor. Eur. J. Biochem. 134, 337–343
219
4 Meijer, A. J., Lamers, W. H. and Chamuleau, R. A. (1990) Nitrogen metabolism and
ornithine cycle function. Physiol. Rev. 70, 701–748
5 Shigesada, K., Aoyagi, K. and Tatibana, M. (1978) Variations in acetylglutamate level and
its possible significance in control of urea synthesis in mammalian liver. Eur. J. Biochem.
85, 385–391
6 Tatibana, M. and and Shigesada, K. (1976) Regulation of urea biosynthesis by the
acetylglutamate-arginine system. In The Urea Cycle (Grisolı́a, S., Báguena, R. and Mayor,
F., eds), pp. 301–313, John Wiley and Sons, New York
7 Bachmann, C., Krähenbühl, S., Colombo, J. P., Schubiger, G., Jaggi, K. H. and Tönz, O.
(1981) N-acetylglutamate synthetase deficiency: a disorder of ammonia detoxication. N.
Engl. J. Med. 304, 543
8 Rubio, V. and Grisolı́a, S. (1981) Treating urea cycle defects. Nature 292, 496
9 Rodrı́guez-Aparicio, L. B., Guadalajara, A. M. and Rubio, V. (1989) Physical location of
the site for N-acetyl-L-glutamate, the allosteric activator of carbamoyl phosphate
synthetase, in the 20-kilodalton COOH-terminal domain. Biochemistry 28, 3070–
3074
10 Rubio, V. (1994) Structure-activity correlations in carbamoyl phosphate synthetases. In
Carbon Dioxide Fixation and Reduction in Biological and Model Systems (Brändén, C.-I.
and Schneider, G., eds), pp. 249–264, Oxford University Press, Oxford
11 Rubio, V., Cervera, J., Lusty, C. J., Bendala, E. and Britton, H. G. (1991) Domain structure
of the large subunit of Escherichia coli carbamoyl phosphate synthetase. Location of the
binding site for the allosteric inhibitor UMP in the COOH-terminal domain. Biochemistry
30, 1068–1075
12 Cervera, J., Bendala, E., Britton, H. G., Bueso, J., Nassif, Z., Lusty, C. J. and Rubio, V.
(1996) Photoaffinity labelling with UMP of lysine 992 of carbamyl phosphate synthetase
from Escherichia coli allows identification of the binding site for the pyrimidine inhibitor.
Biochemistry 35, 7247–7255
13 Cervera, J., Conejero-Lara, F., Ruiz-Sanz, J., Galisteo, M. L., Mateo, P. L., Lusty, C. J. and
Rubio, V. (1993) The influence of effectors and subunit interactions on Escherichia coli
carbamoyl-phosphate synthetase studied by differential scanning calorimetry. J. Biol.
Chem. 268, 12504–12511
14 Bueso, J., Lusty, C. J. and Rubio, V. (1994) Location of the binding site for the allosteric
activator IMP in the COOH-terminal domain of Escherichia coli carbamyl phosphate
synthetase. Biochem. Biophys. Res. Commun. 203, 1083–1089
15 Bueso, J., Cervera, J., Fresquet, V., Marina, A., Lusty, C. J. and Rubio, V. (1999)
Photoaffinity labeling with the activator IMP and site-directed mutagenesis of histidine
995 of carbamoyl phosphate synthetase from Escherichia coli demonstrate that the
binding site for IMP overlaps with that for the inhibitor UMP. Biochemistry 38,
3910–3917
16 Czerwinski, R. M., Mareya, S. M. and Raushel, F. M. (1995) Regulatory changes in the
control of carbamoyl phosphate synthetase induced by truncation and mutagenesis of the
allosteric binding domain. Biochemistry 34, 13920–13927
17 Liu, X., Guy, H. I. and Evans, D. R. (1994) Identification of the regulatory domain of the
mammalian multifunctional protein CAD by the construction of an Escherichia coli
hamster hybrid carbamyl-phosphate synthetase. J. Biol. Chem. 269, 27747–
27755
18 Carrey, E. A. and Hardie, D. G. (1988) Mapping of catalytic domains and phosphorylation
sites in the multifunctional pyrimidine-biosynthetic protein CAD. Eur. J. Biochem. 171,
583–588
19 Thoden, J. B., Huang, X., Kim, J., Raushel, F. M. and Holden, H. M. (2004) Long-range
allosteric transitions in carbamoyl phosphate synthetase. Protein Sci. 13, 2398–2405
20 Thoden, J. B., Raushel, F. M., Benning, M. M., Rayment, I. and Holden, H. M. (1999) The
structure of carbamoyl phosphate synthetase determined to 2.1 Å resolution. Acta
Crystallogr. D Biol. Crystallogr. 55, 8–24
21 Thoden, J. B., Raushel, F. M., Wesenberg, G. and Holden, H. M. (1999) The binding of
inosine monophosphate to Escherichia coli carbamoyl phosphate synthetase. J. Biol.
Chem. 274, 22502–22507
22 Meister, A. and Powers, S. G. (1977) Glutamine-dependent carbamyl phosphate
synthetase: catalysis and regulation. Adv. Enzyme Regul. 16, 289–315
23 Xie, Y., Ihsanawati, Kishishita, S., Murayama, K., Takemoto, C., Shirozu, M. and
Yokoyama, S. (2008) Crystal structure of MGS domain of carbamoyl-phosphate
synthetase from Homo sapiens , Protein DataBank doi:10.2210/pdb2yvq/pdb.
24 Schooler, J. M., Fahien, L. A. and Cohen, P. P. (1963) 2-Acetoxyglutarate as an activator
of carbamoyl phosphate synthetase. J. Biol. Chem. 238, 1909–1910
25 Britton, H. G., Garcia-España, A., Goya, P., Rozas, I. and Rubio, V. (1990) A
structure-reactivity study of the binding of acetylglutamate to carbamoyl phosphate
synthetase I. Eur. J. Biochem. 188, 47–53
26 Kleywegt, G. J. and Jones, T. A. (1994) Detection, delineation, measurement and display
of cavities in macromolecular structures. Acta Crystallogr. D Biol. Crystallogr. 50,
178–185
c The Authors Journal compilation c 2009 Biochemical Society
220
S. Pekkala and others
27 Verdonk, M. L., Cole, J. C., Hartshorn, M. J., Murray, C. W. and Taylor, R. D. (2003)
Improved protein-ligand docking using GOLD. Proteins 52, 609–623
28 Lovell, S. C., Word, J. M., Richardson, J. S. and Richardson, D. C. (2000) The
penultimate rotamer library. Proteins 40, 389–408
29 Duan, Y., Wu, C., Chowdhury, S., Lee, M. C., Xiong, G., Zhang, W., Yang, R., Cieplak, P.,
Luo, R., Lee, T. et al. (2003) A point-charge force field for molecular mechanics
simulations of proteins based on condensed-phase quantum mechanical calculations.
J. Comput. Chem. 24, 1999–2012
30 Case, D. A., Cheatham, III, T. E., Darden, T., Gohlke, H., Luo, R., Merz, Jr, K. M., Onufriev,
A., Simmerling, C., Wang, B. and Woods, R. (2005) The Amber biomolecular simulation
programs. J. Comput. Chem. 26, 1668–1688
31 CCP4 (Collaborative Computational Project 4) (1994) The CCP4 suite: programs for
protein crystallography. Acta Crystallogr. D. Biol. Crystallogr. 50, 760–763
32 Honig, B. and Nicholls, A. (1995) Classical electrostatics in biology and chemistry.
Science 268, 1144–1149
33 Thompson, J. D., Higgins, D. G. and Gibson, T. J. (1994) CLUSTAL W: improving the
sensitivity of progressive multiple sequence alignments through sequence weighting,
position specific gap penalties and weight matrix choice. Nucleic Acids Res. 22,
4673–4680
34 Reference deleted
35 Gross, E. (1967) The cyanogen bromide reaction. Methods Enzymol. 11, 238–255
36 Drapeau, G. R. (1986) Cleavage at glutamic acid with staphylococcal protease. Methods
Enzymol. 47, 189–191
37 Mahoney, W. C. and Hermodson, M. A. (1979) High-yield cleavage of tryptophanyl
peptide bonds by o-iodosobenzoic acid. Biochemistry 18, 3810–3814
38 Penefsky, H. S. (1977) Reversible binding of Pi by beef heart mitochondrial adenosine
triphosphatase. J. Biol. Chem. 252, 2891–2899
39 Shevchenko, A., Jensen, O. N., Podtelejnikov, A. V., Sagliocco, F., Wilm, M., Vorm, O.,
Mortensen, P., Boucherie, H. and Mann, M. (1996) Linking genome and proteome by
mass spectrometry: large-scale identification of yeast proteins from two dimensional gels.
Proc. Natl. Acad. Sci. U.S.A. 93, 14440–14445
40 Perkins, D. N., Pappin, D. J. C., Creasy, D. M. and Cottrell, J. S. (1999) Probability-based
protein identification by searching sequence databases using mass spectrometry data.
Electrophoresis 20, 3551–3567
Received 17 June 2009/11 September 2009; accepted 15 September 2009
Published as BJ Immediate Publication 15 September 2009, doi:10.1042/BJ20090888
c The Authors Journal compilation c 2009 Biochemical Society
41 Nyunoya, H., Broglie, K. E., Widgren, E. E. and Lusty, C. J. (1985) Characterization and
derivation of the gene coding for mitochondrial carbamyl phosphate synthetase I of rat.
J. Biol. Chem. 260, 9346–9356
42 Yefimenko, I., Fresquet, V., Marco-Marin, C., Rubio, V. and Cervera, J. (2005)
Understanding carbamoyl phosphate synthetase deficiency: impact of clinical mutations
on enzyme functionality. J. Mol. Biol. 349, 127–141
43 Bradford, M. M. (1976) A rapid and sensitive method for the quantification of microgram
quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72,
248–254
44 Alonso, E. and Rubio, V. (1983) Binding of N-acetyl-L-glutamate to rat liver
carbamoylphosphate synthetase (ammonia). Eur. J. Biochem. 135, 331–337
45 Rubio, V., Ramponi, G. and Grisolia, S. (1981) Carbamoyl phosphate synthetase I of
human liver. Purification, some properties and immunological cross-reactivity with the rat
liver enzyme. Biochim. Biophys. Acta 659, 150–160
46 Haraguchi, Y., Uchino, T., Takiguchi, M., Endo, F., Mori, M. and Matsuda, I. (1991)
Cloning and sequence of a cDNA encoding human carbamyl phosphate synthetase I:
molecular analysis of hyperammonemia. Gene 107, 335–340
47 Summar, M. L. (1998) Molecular genetic research into carbamoyl phosphate synthetase I:
molecular defects and linkage markers. J. Inherit. Metab. Dis. 2, 30–39
48 Summar, M. L., Hall, L. D., Eeds, A. M., Hutcheson, H. B., Kuo, A. N., Willis, A. S., Rubio,
V., Arvin, M. K., Schofield, J. P. and Dawson, E. P. (2003) Characterization of genomic
structure and polymorphisms in the human carbamyl phosphate synthetase I gene. Gene
311, 51–57
49 Eeds, A. M., Hall, L. D., Yadav, M., Willis, A., Summar, S., Putnam, A., Barr, F. and
Summar, M. L. (2006) The frequent observation of evidence for nonsense-mediated decay
in RNA from patients with carbamyl phosphate synthetase I deficiency. Mol. Genet. Metab.
89, 80–86
50 Tatibana, M. and Shigesada, K. (1972) Control of pyrimidine biosynthesis in mammalian
tissues. V. Regulation of glutamine-dependent carbamoyl phosphate synthetase:
activation by 5-phosphoribosyl 1-pyrophosphate and inhibition by uridine triphosphate.
J. Biochem. 72, 549–560
51 Wallace, A. C., Laskowski, R. A and Thornton, J. M. (1995) LIGPLOT: a program to
generate schematic diagrams of protein-ligand interactions. Protein Eng. 8,
127–134
Biochem. J. (2009) 424, 211–220 (Printed in Great Britain)
doi:10.1042/BJ20090888
SUPPLEMENTARY ONLINE DATA
Structural insight on the control of urea synthesis: identification of the
binding site for N -acetyl-L-glutamate, the essential allosteric activator of
mitochondrial carbamoyl phosphate synthetase1
Satu PEKKALA*, Ana I. MARTÍNEZ*, Belén BARCELONA*†, José GALLEGO*, Elena BENDALA‡, Igor YEFIMENKO*†‡,
Vicente RUBIO†‡2,3 and Javier CERVERA*†2,3
*Centro de Investigación Prı́ncipe Felipe (CIPF), Avda. Autopista del Saler 16, Valencia 46012, Spain, †Centro de Investigación Biomédica en Red para Enfermedades Raras
(CIBERER-ISCIII), C/Álvaro de Bazán, 10 Bajo, 46010 Valencia, Spain, and ‡Instituto de Biomedicina de Valencia (IBV-CSIC), Jaime Roig 11, Valencia 46010, Spain
Figure S1
Alignment of the relevant C-terminal domain sequences of hCPSI, hamster CPSII (a component of CAD) and the large subunit of eCPS (CarB)
Background shading in dark and light colours reveals respective invariance or conservative replacement of a given residue in either NAG-sensitive CPSI and CPSIII (dark and light blue), eukaryotic
pyrimidine-specific CPSII (red and magenta) or bacterial carB-type CPSs (green and yellow). The following CPSs have been aligned to determine sequence conservations within each class of CPS:
CPSI and CPSIII from man, rat, mouse, opossum, chicken (although not ureotelic, chicken has genes for ornithine transcarbamylase and CPSI), bullfrog, zebra fish, the shark Squalus acanthias ,
the amphioxus Branchiostoma floridae and the sea urchin Strongylocentrotus purpuratus ; CPSII from human, hamster, Squalus acanthias, Drosophila melanogaster , the slime mold Dictyostelium
discoideum and the ascomycetes Emericella nidulans , Schizosaccharomyces pombe and Saccharomyces cerevisiae ; CarB from E. coli , Salmonella typhimurium , Pseudomonas aeruginosa , Serratia
protomacularis , Aquifex aeolicus , and pyrimidine-specific CPSs from Bacillus subtilis and Bacillus stearothermophilus . Black and grey shading denotes, respectively, invariance or conservation in
at least two CPS families. The dots above and below the CPSI and CarB sequences respectively indicate residues that change their accessibility with NAG (CPSI) or IMP (CarB) binding, using a
probe radius of 1.4 Å. Letters above the CPSI sequence show the mutations introduced here. Secondary structure elements of the CPSI and CarB allosteric domains are shown as arrows (β strands)
and cylinders (helices) above and below the corresponding sequences. In the CPSI sequence, Thr1391 , the residue proposed here to be cross-linked to ClNAG upon photoaffinity labelling of the
enzyme, is circled. In CAD, a horizontal line underlines the GAGGRR sequence proposed [1] to be involved in the binding of the polyphosphate moiety of the inhibitor UTP. In addition, the serine
residue that is phosphorylated by cAMP-dependent protein kinase is indicated [2]. The vertical blue and red lines under the Lys993 and His995 of eCPS mark the points of photocross-linking of UMP
and IMP respectively to this enzyme [1,3].
1
Dedicated to Professor Santiago Grisola, who discovered CPSI and its activation by carbamylglutamate, and who, with V.R., prepared the first
carbamylglutamate given to a patient with NAGS deficiency.
2
These authors contributed equally to this work.
3
Correspondence may be addressed to either Javier Cervera (email [email protected]) or Vicente Rubio (email [email protected]).
c The Authors Journal compilation c 2009 Biochemical Society
S. Pekkala and others
Table S1
Synthetic oligonucleotides used in site-directed mutagenesis
The triplets introducing the desired mutations are indicated in bold.
Figure S2 Stereo view of the superimposition of the Cα traces of the
allosteric domains of hCPS and eCPS
hCPS (in black, residue numbers are labelled in multiples of 10) and eCPS (in green). NAG is
represented with carbon atoms in black, in the way it is proposed to be bound to hCPS. IMP
bound to eCPS is shown with carbon atoms in green.
REFERENCES
1 Cervera, J., Bendala, E., Britton, H. G., Bueso, J., Nassif, Z., Lusty, C. J. and Rubio, V.
(1996) Photoaffinity labelling with UMP of lysine 992 of carbamyl phosphate synthetase
from Escherichia coli allows identification of the binding site for the pyrimidine inhibitor.
Biochemistry 35, 7247–7255
2 Carrey, E. A. and Hardie, D. G. (1988) Mapping of catalytic domains and phosphorylation
sites in the multifunctional pyrimidine-biosynthetic protein CAD. Eur. J. Biochem. 171,
583–588
3 Bueso, J., Lusty, C. J. and Rubio, V. (1994) Location of the binding site for the allosteric
activator IMP in the COOH-terminal domain of Escherichia coli carbamyl phosphate
synthetase. Biochem. Biophys. Res. Commun. 203, 1083–1089
Received 17 June 2009/11 September 2009; accepted 15 September 2009
Published as BJ Immediate Publication 15 September 2009, doi:10.1042/BJ20090888
c The Authors Journal compilation c 2009 Biochemical Society
Mutation
Direction
Sequence
T1391V
T1391V
T1394A
T1394A
W1410K
W1410K
N1437D
N1437D
N1440D
N1440D
Forward
Reverse
Forward
Reverse
Forward
Reverse
Forward
Reverse
Forward
Reverse
5 -GCTTTTTGCCGTAGAAGCCACATCAGAC-3
5 -GTCTGATGTGGCTTCTACGGCAAAAAGC-3
5 -GCCACAGAAGCCGCATCAGACTGGCTCAACGCC-3
5 -GTTGGCGTTGAGCCAGTCTGATGCGGCTTCTGTGGC-3
5 -GCCACCCCAGTGGCTAAGCCATCTAGGGAAGGACAG-3
5 -GGGATTCTGTCCTTCCTGAGATGGCTTAGCCACTGGG-3
5 -GCATTGACCTAGTGATTGACCTCCCCAATAAC-3
5 -GGTGTTGTTATTGGGGAGGTCAATCACTAGGTC-3
5 -GTGATTAACCTCCCCGATAACAACACCAAATTTG-3
5 -GACAAATTTGGTGTTGTTATCGGGGAGGTTAATC-3