(Received for publication, April 18, 1996, and in revised form, October 16, 1996)
From the Department of Biochemistry and Biophysics, Iowa State University, Ames, Iowa 50011
The state of aggregation of adenylosuccinate synthetase from Escherichia coli is a point of controversy, with crystal structures indicating a dimer and some solution studies indicating a monomer. Crystal structures implicate Arg143 and Asp231 in stabilizing the dimer, with Arg143 interacting directly with bound IMP of the 2-fold related subunit. Residue Arg143 was changed to Lys and Leu, and residue Asp231 was changed to Ala. Matrix-assisted laser desorption ionization mass spectroscopy and analytical ultracentrifugation of the wild-type and the mutant enzymes indicate a mixture of monomers and dimers, with a majority of the enzyme in the monomeric state. In the presence of active site ligands, the wild-type enzyme exists almost exclusively as a dimer, whereas the mutant enzymes show only slightly decreased dissociation constants for the dimerization. Initial rate kinetic studies of the wild-type and mutant enzymes show similar kcat and Km values for aspartate. However, increases in the Km values of GTP and IMP are observed for the mutant. Changes in dissociation constants for IMP are comparable with changes in Km values. Our results suggest that IMP binding induces enzyme dimerization and that two residues in the interface region, Arg143 and Asp231, play significant roles in IMP and GTP binding.
Adenylosuccinate synthetase (AMPSase)1 (IMP:L-aspartate ligase (GDP forming), EC 6.3.4.4.) catalyzes the first committed step in the conversion of IMP to AMP in the de novo synthetic pathway for purine nucleotides:
![]() |
![]() |
Three different mechanisms have been proposed for the catalytic action
of AMPSase (1). The most widely accepted mechanism involves a
6-phosphoryl-IMP intermediate formed by the nucleophilic attack of the
6-oxo group of IMP on the -phosphorus atom of GTP (5, 6). A second
nucleophilic attack by the amino group of aspartate on C-6 of
6-phosphoryl IMP displaces the phosphate and forms
adenylosuccinate.
AMPSase was first purified to homogeneity in 1976 from Escherichia coli (7) and has since been purified and characterized from many sources (1). The purA gene, which in E. coli codes for AMPSase, was cloned in 1986 (8) and used in the construction of an overexpression system (9). The crystal structure of AMPSase was determined to 2.8 Å (10) and later refined to 2.5 and 2.0 Å (11). In crystal structures, the enzyme exists as a homodimer. Two nearly independent regions contribute to the interface between the polypeptide chains of the synthetase dimer. Residues putatively involved in the binding of IMP lie at or near the interface between polypeptide chains in the dimeric form of the enzyme. One of the residues, Asp231, may play an important role in holding the subunits in close contact by hydrogen bonding to Arg147 and Lys140 of the 2-fold related subunit. Indeed, when Lys140 and Arg147 were replaced by isoleucine and leucine, respectively, Km(GTP) and Km(IMP) showed significant increases (12, 13).
Crystal structures imply that Arg143 is involved in IMP
binding to the active site of the symmetry-related subunit of the
dimer. Arg143 may also stabilize the interface through at
least one hydrogen bond. Slow phase inactivation of AMPSase was
observed when guanosine-5-O-[S-(4-bromo-2,3-dioxobutyl)] thiophosphate modifies the enzyme at Arg143. Modification
of Arg143 is prevented by adenylosuccinate alone or by a
mixture of GTP, MgCl2, and IMP (14).
The cited findings suggest a direct relationship between dimer formation and AMPSase activity. AMPSase exists as a monomer in solution (15), yet the enzyme is a dimer in crystal structures (10, 11). The discrepancy between the level of aggregation of the enzyme in solution and the crystal is the basis of an important question: Does AMPSase function as a monomer or as a dimer? MALDI mass spectroscopy, analytical ultracentrifugation, and initial rate kinetics were used to determine the state of aggregation and activity of wild-type AMPSase and several interface mutants. Arg143 was replaced with leucine to remove the guanidinium group and yet retain some of the hydrophobic attributes of the original side chain. Mutation of Arg143 to lysine retains the positive charge but limits hydrogen bonding opportunities of the side chain at position 143. Asp231 was replaced with alanine so as to disrupt the Lys140-Asp231 salt link observed in the crystalline dimer. Our findings indicate that the E. coli enzyme dimerizes in response to active site ligands and that on the basis of sequence homology, dimerization may be a property common to all known adenylosuccinate synthetases, regardless of source.
GTP, IMP, L-aspartate,
phenylmethylsulfonyl fluoride, and bovine serum albumin were obtained
from Sigma. A site-directed mutagenesis kit was
obtained from Amersham Corp. Restriction enzymes were obtained from
Promega. E. coli strain XL-1 blue was obtained from Stratagene. E. coli strain purA
H1238 was a gift from Dr. D. Bachman (Genetic Center, Yale University). Other reagents and chemicals used in the experiments were obtained from
Sigma if not specified.
Recombinant DNA manipulation was
performed by using standard procedures (16). The plasmid containing a
1.8-kilobase BamHI-HindIII fragment from PMS204
ligated into PUC118 was used in the mutagenesis step. All mutagenic
oligonucleotide primers used in the experiments were synthesized on a
Bioresearch 8570EX automated DNA synthesizer at the DNA Facility at
Iowa State University. Mutagenesis was performed according to the
protocol provided by Amersham Corp. The mutations were confirmed by DNA
sequencing using the chain termination method (17) at the Iowa State
University DNA Facility. The 1.8-kilobase Bam HI-HindIII
fragment with the desired mutation was ligated back into PMS204 and
transformed into XL-1 blue cells. The plasmids isolated from that
strain were used to transform E. coli strain
purA H1238, the strain from which AMPSase was
purified.
Protein concentration was determined by the Bradford (18) method using bovine serum albumin as the standard. Concentrations reported here refer to monomers.
Purification of Wild-type and Mutant Adenylosuccinate SynthetaseThe wild-type and mutant enzymes were purified as described elsewhere (9). The purity of the enzyme was checked by SDS-polyacrylamide gel electrophoresis according to Laemmli (19). AMPSase activity was determined as described earlier (20).
Kinetic Studies of Wild-type and Mutant AMPSaseThe concentrations of stock solutions of nucleotides were based on their molar extinction coefficients at 253 nm for GTP and 248 nm for IMP. For each enzyme assay, the increase at 290 nm was recorded at 25 °C. The enzyme assay solution contained 20 mM Hepes (pH 7.7) and 5 mM MgCl2. When GTP was the variable substrate, the concentration of aspartate was fixed at 5 mM, and the IMP concentration was fixed at 450 µM for wild-type AMPSase and at 12 mM for the mutants. When IMP was the variable substrate, the concentration of GTP was fixed at 300 µM, and aspartate was fixed at 5 mM. When aspartate was the variable substrate, the concentration of GTP was fixed at 300 µM, and the concentration of IMP at 450 µM for the wild-type AMPSase and 12 mM for the mutants.
To obtain the values of Kia and Kib, dissociation constants for GTP and IMP, respectively, a 5 × 5 matrix of substrates were used with each enzyme. The enzyme assay solution contained 20 mM Hepes (pH 7.7), 5 mM MgCl2, and 5 mM aspartate and varying GTP and IMP concentrations. The kinetic data were fit to Equation 1 using a program written in Minitab in place of the Omnitab program described by Siano et al. (21):
![]() |
(Eq. 1) |
Circular dichroisim
spectra for the wild-type and mutant enzymes were acquired on a JASCO
J700 spectropolarimeter equipped with a data processor. Samples (100 µg/ml of enzyme) dialyzed against 5 mM potassium
phosphate (pH 7.0), 1 mM EDTA, 1 mM
-mercaptoethanol, were placed in a 1-cm cuvette, and data points
were collected in 0.1-nm increments. Each spectrum was corrected for
background contributions of the buffer and smoothed using the
spectropolarimeter program. The data were analyzed by JASCO analysis
software or by PSIPLOT.
All of the enzymes were dialyzed
against 5 mM potassium phosphate buffer (pH 7.0), 1 mM EDTA, and 1 mM -mercaptoethanol. Protein
concentration was adjusted to 1 mg/ml. Samples of 0.5-1.0 µl were
loaded with 0.5 µl of freshly made 3,5-dimethoxy-4-hydroxy-cinnamic acid matrix. Bovine serum albumin was used as the internal calibration standard. Data were collected on a Finnigan LASERMAT 2000 MALDI-time of
flight mass analyzer in the Protein Facility of Iowa State University
and were analyzed by the LASERMAT 2000 data processing software.
Analytical ultracentrifugation experiments were performed using a Beckman Optima XL-A ultracentrifuge. The temperature of the rotor (AN-60 Ti) was set at 4 °C. Rotor speeds were set at 10,000, 14,000, and 18,000. Wild-type and mutant AMPSase samples were prepared in 5 mM potassium phosphate buffer (pH 7.0) at concentrations of 2.9-11.6 µM corresponding to absorbences of 0.2-0.8 (280 nm, 1-cm cuvette). Concentration-dependent equilibrium sedimentation was performed with concentrations of 4.3 and 58 µM, corresponding to A280 nm readings of 0.3 and 4.0. Samples of AMPSase in the presence of ligands were prepared by dialysis overnight against 5 mM potassium phosphate (pH 7.0), 5 mM succinate, 5 mM MgCl2, 20 µM IMP, and 20 µM GTP. Protein samples were centrifuged at least 10 h before the collection of data. Stepwise radial scans were performed at 280 nm for the wild-type and mutant enzymes with and without ligands, and at 280, 295, and 300 nm for the concentration-dependent equilibrium sedimentation. Each reading is the average of 30 points with nominal spacing of 0.001 cm between radial positions. Absorbance readings were measured at 1-h intervals to ensure that equilibrium had been reached. Three scans were averaged, and the data were analyzed by the method of Van Hold and Weischet (22) using the "IDEAL" model on the Optima XL-A Analysis Software (version 2.0) to get the apparent molecular mass. The partial specific volume of AMPSase, 0.737 cm3/g, was calculated by the method of Cohn and Edsall (23).
Equilibrium sedimentation data for the wild-type and mutant enzymes obtained at three different rotor speeds were analyzed using the "SELF" model in multiple data set analysis program Optima XLA. The association constants (Ka) for the monomer-dimer equilibrium were obtained using the following equation:
![]() |
(Eq. 2) |
![]() |
![]() |
(Eq. 3) |
X-ray crystallographic studies suggest that interface residues
play an important role in maintaining the quaternary structures of
AMPSase (10, 11). Two very important residues in this context are
Arg143 and Asp231. Arg143 in one
subunit is ligated to the 5-phosphoryl of IMP in the active site of
the juxtaposed subunit (11). Asp231 forms a salt link with
Lys140, a residue putatively essential for AMPSase activity
as a means of stabilizing subunit-subunit association. Experiments
involving mutation of residues Arg143 and
Asp231 were undertaken to gain insight into the role of
these interface residues.
Arg143 and Asp231 are conserved in the nine known AMPSase sequences (8, 24-29, 31)2 (Table I), as are Arg147 and Lys140. In crystal structures, the side chain of Asp231 forms a salt bridge with Lys140, and the carbonyl of Asp231 hydrogen bonds to Arg147 (10, 11). Chemical modification or mutation of Lys140 or Arg147 inactivates the synthetase (12, 13). Arg143 from each subunit projects into the IMP binding site of the symmetry-related subunit of crystallographic dimers (11).
|
The oligonucleotide primers used in the mutagenesis experiments are shown in Table II together with the sequencing primers for confirming the mutants. Arg143 was changed to leucine and lysine. Asp231 was mutated to alanine. These mutations altered the charge states of residues and/or hydrogen bonding interactions observed in crystallographic structures while keeping the size of the side chain comparable with that of the original residue. All the mutants were purified by using procedures similar to those for wild-type AMPSase with some modifications. The D231A mutant bound to the phenyl sepharose CL-4B so tightly that it could be eluted only by water. All the enzymes exhibited greater than 95% purity on the basis of SDS-polyacrylamide gel electrophoresis and a molecular mass of 48 kDa (data not shown).
|
Circular dichroisim spectra of the mutant and wild-type enzymes were superimposable (data not shown) from 200 to 260 nm, indicating no global alterations in the secondary structures of the mutants relative to wild-type AMPSase.
Kinetic Analysis of AMPSase MutantsThe kinetic parameters for GTP, IMP, and aspartate with various forms of AMPSase are summarized in Table III. The Km values for aspartate illustrate that the mutants differed little from that of the wild-type enzyme, suggesting that Arg143 and Asp231 are not involved in the binding of aspartate. In addition, the mutant and wild-type AMPSases had comparable kcat values; however, Km values for GTP and IMP exhibited significant increases for the mutant enzymes relative to wild-type AMPSase. The Km(GTP) values for R143K, R143L, and D231A showed 2-, 10-, and 20-fold increases, respectively, compared with that of the wild-type enzyme. On the other hand, even more dramatic changes were observed for the Km values of IMP. Increases in Km(IMP) of 100-, 100-, and 60-fold for the mutants R143K, R143L, and D231A, respectively, were observed relative to that of wild-type AMPSase. In addition, increases in Kia(IMP) of 20-30-fold were also observed for the mutant enzymes, relative to that of the wild-type enzyme. These findings suggest that the residues in question contribute significantly to the binding of both IMP and GTP by stabilizing the dimer through hydrogen bonding, by direct interaction with IMP, or by both. The similar kinetic properties of R143K and R143L mutants with respect to IMP binding suggest that the positive charge of lysine alone did not restore wild-type properties, suggesting a precise hydrogen bonding and stereochemical role for Arg143. The models for R143K and R143L are presented and discussed below (Fig. 1).
|
MALDI Mass Spectrometry Analysis of AMPSase
AMPSase in
solution is a monomer (15), whereas it is a dimer in crystals (10, 11).
We assumed that subunit association might involve interface residues
such as Arg143; therefore, experiments were undertaken to
evaluate this possibility. The mutant enzymes and wild-type AMPSase
were analyzed by MALDI mass spectroscopy. A typical spectrum is shown
in Fig. 2. Given the estimated molecular mass of the
monomer as 46 kDa, three ionized species were observed from 20 to 110 kDa, corresponding to mass values of 23.4 kDa (M = 46.6 kDa, Z = 2), 46.6 kDa (M = 46.6 kDa, Z = 1), 92.7 kDa (2M = 92.7 kDa,
Z = 1). The existence of a 92.7-kDa peak indicates the
presence of dimers.
Equilibrium Sedimentation of Wild-type and Mutant AMPSases
Subunit association of wild-type AMPSase was evaluated
at three different concentrations (32). Considering the molecular mass
of the monomeric and dimeric enzymes, three different centrifugation speeds were utilized (10,000, 14,000, and 18,000). The molecular mass
of AMPSase does not depend on either the centrifugation speed or the
protein concentration under the conditions tested. Typical equilibrium
sedimentation data are shown in Fig. 3a. Fig.
3b shows the relationship between wild-type protein
concentration and apparent molecular mass. At low concentrations
(A280 nm = 0.1), the apparent molecular mass
was 45 kDa, which matches the result determined by MALDI mass
spectroscopy. At higher concentrations (A280 nm 0.65), the apparent molecular mass approaches 60 kDa, which is
between the molecular mass for monomeric and dimeric AMPSase. This
observation strongly indicates the existence of a monomer-dimer
equilibrium. Dissociation constants for dimer-monomer equilibria of
wild-type and mutant AMPSases are shown in Table IV. At
micromolar concentrations of enzyme in the absence of ligands, the
Kd values are approximately 10 µM for
all the enzymes, indicating low concentrations of dimers under these
conditions. In the presence of active site ligands, the mutant enzymes
exhibited small decreases in Kd values,
suggesting a slight increase of dimers, whereas the wild-type enzyme
exhibited a large decrease in Kd, implying that
virtually all the protein is present as dimer. Active site ligands
apparently are less effective in the stabilization of the mutant
dimers. Our findings support the important roles played by the
interface residues in forming AMPSase dimers.
|
Conflicting reports have characterized E. coli AMPSase as a monomer (15) and a dimer (9) in solution. Crystal structures of AMPSase clearly show the enzyme as a dimer in both the unligated (10, 11) and substrate-ligated states (33). AMPSase is putatively regulated by such feedback inhibitors as GDP, adenylosuccinate, and adenine nucleotides (1). The concentration of the latter class of compounds may not vary in the cell because of the adenylate kinase equilibrium. An alternative mode of regulation may be linked to the state of association of the enzyme. We therefore undertook experiments to determine the enzyme's state of association under different experimental conditions.
Protein aggregation is influenced by protein concentration, ligands, pH, temperature, and ionic strength (34-36). Discrepancies in reported molecular masses of AMPSase from different sources may stem from the precise conditions under which mass determinations were made (1). We used two widely differing approaches, MALDI mass spectroscopy and equilibrium sedimentation, to determine molecular mass. Data from MALDI mass spectroscopy clearly revealed the presence of a 92-kDa species (Fig. 2), a finding fully consistent with the existence of AMPSase as a dimer in solution. The MALDI technique, however, cannot provide a measure of the relative amounts of monomer and dimer. Equilibrium sedimentation was performed to determine whether a dynamic equilibrium exists between the monomer and dimer forms. Our results suggest that without ligands, AMPSase is a mixture of both monomers and dimers (50% monomer and 50% dimer at 11.4 µM wild-type enzyme) and the equilibrium shifting to favor the dimer at higher concentrations (91% dimer and 9% monomer at 58.2 µM wild-type enzyme). These observations are in harmony with x-ray diffraction studies in which only dimers were observed (10, 11, 33).
We failed to observe two distinct species in centrifugation experiments, suggesting that a rapid equilibrium exists relative to the sedimentation rate between the monomers and dimers in the absence of ligands. Only dimers were detected, however, when substrates were added to the wild-type enzyme. Ligands clearly shift the equilibrium toward the dimer and the kinetic barrier in dimer formation, and dissociation is relatively low. On the other hand, mutant enzymes in the presence of ligands exhibited only slightly lower Kd values than in the absence of ligands, indicating a much weaker ligand-induced dimerization of mutant enzymes relative to the wild-type enzyme. Considering that Arg143 and Asp231 are well conserved in all the AMPSase sequences, the results here also suggest that AMPSase from E. coli may require both subunits for catalytic activity at physiological ligand concentrations and that this may be a general property of all AMPSases, regardless of source. AMPSase, however, is not the only protein that changes its state of subunit association upon ligand binding. Briehl demonstrated that lamprey hemoglobin exists predominantly as monomers when oxygenated and as oligomers when deoxygenated (37).
This study focused on two residues of AMPSase, Asp231 and
Arg143, that are involved in putative subunit-subunit
interactions. Arg143 from one subunit hydrogen bonds to the
5-phosphoryl group of IMP in the active site of the symmetry-related
subunit (33). In addition, Arg143 hydrogen bonds to a
backbone carbonyl of the juxtaposed subunit. Because Arg143
is conserved in all nine known sequences of AMPSase (8, 24-29, 31),2 we suggest that its role in the E. coli
enzyme is also conserved in all other known AMPSases. The mutants R143L
and R143K exhibit approximately the same kcat
and Km values for aspartate as does wild-type
AMPSase with small changes in the Km values for GTP
(2- and 10-fold increases, respectively, for R143K and R143L). However,
100-fold increases in Km for IMP were observed for
both mutant enzymes. Despite great differences in the hydrophobicity
and electrostatic charge of lysine and leucine residues at position
143, they exhibit very similar kinetic properties in terms of IMP
binding.
The spatial relationship of the side chain of position 143 to the
active site of the monomer related by the 2-fold symmetry is shown in
Fig. 1. The guanidinum group of Arg143 bonds to the
backbone carbonyl of the symmetry related subunit, as well as the
5-phosphate of IMP (Ref. 33 and Fig. 1b). By model
building, atom NZ of lysine 143 can occupy the same position as the
guanidinum nitrogen responsible for the intersubunit hydrogen bonds
(Fig. 1c). Thus, the increase in Km for
the R143K mutant is not immediately obvious from the modeling study.
However, the substitution of a lysyl for an arginyl side chain at
position 143 leaves a packing void at the interface between monomers.
Presumably this void is filled by a water molecule, in which case more
than one conformational arrangement between that water molecule and the
lysyl side chain is possible, but only one of these arrangements is
comparable with the interaction exhibited by Arg143 in the
crystal structures.
Asp231, also conserved in known AMPSase sequences, forms an
intersubunit salt bridge with the side chain of Lys140 and
a hydrogen bond to Arg147 through its backbone carbonyl. As
for the R143K and R143L mutants, D231A showed about the same
kcat and Km for aspartate as
wild-type AMPSase. However, Km values for GTP and IMP increased 20- and 60-fold, respectively, relative to the wild-type enzyme. These results are qualitatively similar to those for the R143K
and R143L mutants and are consistent with the hypothesis that the
dimeric form of the synthetase is functionally active (11). The
increase in Km for GTP with the mutants may be due
to synergism in the binding of IMP and GTP. Markham and Reed, for
instance, observed synergism in the inhibition of AMPSase by GDP and
nitrate (38, 39), both of which are competitive inhibitors with respect
to GTP. Furthermore, on the basis of isotopic scrambling reactions
(40), the -phosphoryl group of GTP probably exchanges between the
two sites, where it is either covalently linked to the 6-oxo group of
IMP or covalently linked to the
-phosphoryl group of the guanine
nucleotide. In the absence of the dimer interface, AMPSase cannot
provide an appropriate environment for binding IMP, which probably
impairs the exchange process just described and leads to a weaker
association of GTP with the enzyme.
Combining the two lines of evidence from biophysical analysis and
initial rate kinetics on both the wild-type and mutant enzymes, we
established a clearer relationship between the association states and
enzymatic activity of E. coli AMPSase. MALDI mass
spectroscopy revealed the existence of a 92-kDa species for all the
mutant enzymes, indicating that mutations at positions 231 and 143 do not prevent dimerization. R143L, in the presence of ligands, showed an
apparent molecular mass of a dimer as determined by gel filtration analysis (data not shown). These observations suggest that neither Arg143 nor Asp231 are essential for dimer
formation. On the other hand, position 231 and 143 mutants studied here
exhibit little change in Kd values for the dimer to
monomer equilibrium in the presence or the absence of ligands and
higher Km(IMP) and Km(GTP) values
with little change in kcat values, implying that
dimerization of AMPSase is required for the precise recognition and
binding of substrates IMP and GTP. Assuming that the mutant enzymes
retain the rapid equilibrium random ter ter mechanism of wild-type
AMPSase (1), the comparable kcat values for the
mutants and wild-type AMPSase suggest that neither Arg143
nor Asp231 play a role in the rate-limiting step of
catalysis (interconversion from
AMPSase-MgGTP-2-IMP-aspartate to
AMPSase-adenylosuccinate-MgGDP1-Pi).
An interesting question concerns the significance of the AMPSase
monomer-dimer equilibrium and its physiological implications, if any.
The enzyme should exist primarily as a monomer in the micromolar
concentration range, and substrates should shift the monomer-dimer
equilibrium to dimer. Only if the monomer and dimer have dissimilar
activities at physiological ligand concentrations can the monomer-dimer
equilibrium be of significance to regulation. Interestingly, in the
enzyme concentration range of 20-150 pM, there is a
several minute lag in adenylosuccinate production when the reaction is
initiated with the enzyme (data not shown). When the enzyme is
preincubated with MgGTP2 and IMP and the reaction is
initiated with L-aspartate, however, the lag is eliminated.
These findings are consistent with either an inactive monomer or a
monomer of low activity relative to the dimer. Perhaps AMPSase is a
monomer in the absence of nucleotide substrates but is activated by
de novo purine nucleotide biosynthesis. This scenario may
represent a major control mechanism of adenylosuccinate biosynthesis.
It is also noteworthy that AMPSase from Saccharomyces cerevisiae was identified as a single stranded DNA binding protein that specifically recognizes an autonomously replicating sequence (30). Because most of the single stranded DNA-binding proteins are oligomers, the dimerization phenomenon may indicate a dual role for AMPSase as an enzyme and as a regulatory element in DNA replication.
We thank Siquan Luo for help with the analytical ultracentrifuge.