(Received for publication, December 9, 1996, and in revised form, March 20, 1997)
From the Department of Biochemistry and Biophysics, Iowa State University, Ames, Iowa 50011
Crystal structures of adenylosuccinate synthetase
from Escherichia coli complexed with Mg2+,
6-thiophosphoryl-IMP, GDP, and hadacidin at 298 and 100 K have been
refined to R-factors of 0.171 and 0.206 against data to 2.8 and 2.5 Å resolution, respectively. Interactions of GDP,
Mg2+ and hadacidin are similar to those observed for the
same ligands in the complex of IMP, GDP,
NO3, Mg2+ and
hadacidin (Poland, B. W., Fromm, H. J. & Honzatko, R. B. (1996).
J. Mol. Biol. 264, 1013-1027). Although crystals were grown from solutions containing 6-mercapto-IMP and GTP, the electron density at the active site is consistent with 6-thiophosphoryl-IMP and
GDP. Asp-13 and Gln-224 probably work in concert to stabilize the
6-thioanion of 6-mercapto-IMP, which in turn is the nucleophile in the
displacement of GDP from the
-phosphate of GTP. Once formed, 6-thiophosphoryl-IMP is stable in the active site of the enzyme under
the conditions of the structural investigation. The direct observation
of 6-thiophosphoryl-IMP in the active site is consistent with the
putative generation of 6-phosphoryl-IMP along the reaction pathway of
the synthetase.
6-Mercaptopurine is used in the treatment of pediatric leukemia and other cancers (1). The drug is transformed enzymically to 6-mercapto-IMP and then into other derivatives such as 6-thio-GTP, 6-thio-ITP and 6-thiomethyl-IMP. 6-Thio-GTP can be incorporated into DNA, increasing the susceptibility of DNA to damage. 6-Thiomethyl-IMP is a potent inhibitor of phosphoribosylpyrophosphate amidotransferase, suppressing de novo purine biosynthesis, and 6-mercapto-ITP is a potent inhibitor of RNA polymerase. Depletion of the adenine nucleotide pool (specifically ATP) by the action of 6-thiomethyl-IMP reduces levels of S-adenosyl-L-methionine, which in turn impedes methylation of DNA and RNA. Although mechanisms for the cytotoxic effects of 6-mercapto-IMP are known, the metabolic pathways by which 6-mercapto-IMP is inactivated as a drug are unclear.
Adenylosuccinate synthetase (IMP:L-aspartate ligase
(GDP-forming), EC 6.3.4.4) is likely to play a role in the metabolic effects of 6-mercapto-IMP. The synthetase governs the first committed step in the de novo biosynthesis of AMP (2), GTP + IMP + L-aspartate GDP + Pi + adenylosuccinate. 6-Mercapto-IMP is a competitive inhibitor with
respect to IMP (3-5) and likely inhibits the enzyme in
vivo. The combination of adenylosuccinate synthetase with
adenylosuccinate, GDP, and thiophosphate (reverse reaction) leads to
the slow generation of 6-mercapto-IMP, GTP, and aspartate (6). However,
6-mercapto-IMP, GTP, and aspartate in the presence of enzyme (forward
reaction) gives no detectable level of adenylosuccinate. This apparent
violation of microscopic reversibility has been explained by an
equilibrium constant that significantly favors GTP, 6-mercapto-IMP, and
aspartate over GDP, adenylosuccinate, and thiophosphate (6). The
interaction of 6-mercapto-IMP with adenylosuccinate synthetase then may
contribute to the suppression of de novo purine biosynthesis
but only if in vivo levels of IMP are low relative to those
of 6-mercapto-IMP.
We report here refined crystal structures at 298 and 100 K of
adenylosuccinate synthetase from Escherichia coli, grown in the presence of 6-mercapto-IMP, GTP, Mg2+, and hadacidin.
Hadacidin (see Fig. 1), a fermentation product of Penicillium
frequentans (7), is a competitive inhibitor (Ki ~ 106 M) with respect to aspartate (4-5);
the only known function of hadacidin is the inhibition of
adenylosuccinate synthetase (8). In the mechanism proposed for the
synthetase by Lieberman (9), the
-phosphate of GTP is transferred to
the 6-oxygen of IMP, whereafter aspartate displaces phosphate from the
6-phosphoryl intermediate to form adenylosuccinate. Although the
Lieberman (9) mechanism enjoys substantial support from the literature (10-12), no study has provided direct evidence for the formation of a
6-phosphoryl intermediate. Furthermore, 6-phosphoryl-IMP has never been
isolated or synthesized. Regardless of temperature, however, a
6-thiophosphoryl intermediate sits at the IMP binding site in the
crystal structures, hence providing the first direct observation of
6-thiophosphoryl-IMP and strong evidence in support of the catalytic
mechanism proposed 40 years ago by Lieberman (9).
Adenylosuccinate synthetase was prepared as described previously from a genetically engineered strain of E. coli (13-14). The protein migrates as a single band on SDS-polyacrylamide gel electrophoresis with an apparent relative molecular weight of 48,000. Hadacidin was provided by Drs. Fred Rudolph and Bruce Cooper (Dept. of Biochemistry and Cell Biology, Rice University). All other reagents came from Sigma.
Conditions for the growth of the 6-thiophosphoryl-IMP complex were adapted from Poland et al. (15), using the method of hanging drops. Droplets contained 2 µl of enzyme solution (50 mM imidazole, 75 mM succinate, 4 mM GTP, 4 mM 6-mercapto-IMP, 5 mM hadacidin, and 20 mg/ml protein (pH 6.5)) and 2 µl of a crystallization buffer (13% polyethylene glycol 8000 (w/w), 100 mM cacodylic acid/cacodylate (pH 6.5), and 200 mM magnesium acetate). The final pH of the crystallization buffer was 6.5. Wells contained 500 µl of the crystallization buffer. Crystals of approximately 0.5 mm in all dimensions and belonging to the space group P3221 (a = b = 81.63 and c = 159.2 at 298 K, and a = b = 80.54 and c = 158.2 at 100 K) grew in about 1 week. The asymmetric unit consists of a monomer.
Data from the 6-thiophosphoryl-IMP complexes were collected on a Siemens area detector at Iowa State University and were reduced by using XENGEN (16). Data sets were 98% complete (see Table I).
|
Starting phases for the 6-thiophosphoryl-IMP complex were calculated
from the IMP/NO3 complex (15),
omitting coordinates for water molecules,
NO3
, and IMP. Refinement of the
structures involved manual fitting of models to the electron density,
using a Silicon Graphics 4D-25 and the program TOM (17), followed by a
cycle of refinement using XPLOR (18) on a Silicon Graphics 4D-35.
Constants of force and geometry for the protein came from Engh & Huber
(19). The geometry of hadacidin was based on a related structural
fragment (20), retrieved by a search of the Cambridge Data Base. Models of hadacidin with a planar nitrogen and with tetrahedral nitrogens in
L- and D-configurations were refined
individually. In early rounds of refinement, both crystal structures
were heated to 2000 K and then cooled in steps of 25-300 K. In later
rounds of refinement, the systems were heated to 1000 or 1500 K but
then cooled in steps of 10 K. After the slow cooling protocol was
completed (at 300 K), the models were subjected to 120 steps of
conjugate gradient minimization, followed by 20 steps of individual
B-parameter refinement. Individual B-parameters were subject to the
following restraints: nearest neighbor, main chain atoms, 1.5 Å2; next to nearest neighbor, main chain atoms, 2.0 Å2; nearest neighbor, side chain atoms, 2.0 Å2; and next to nearest neighbor, side chain atoms, 2.5 Å2.
Water molecules were added if (i) electron density at a level of 2.5
was present in maps based on Fourier coefficients
(Fobs
Fcalc)ei
calc
and (2Fobs
Fcalc)ei
calc,
and (ii) acceptable hydrogen bonds could be made to an existing atom of
the model. If after refinement a site for a water molecule fell beyond
3.3 Å from its nearest neighbor, that site was omitted from the model.
In addition, water molecules were deleted from the model if their
thermal parameters exceeded 80 Å2. Harmonic restraints (50 kcal/mol) were placed on the positions of oxygen atoms of water
molecules to allow new water molecules to relax by adjustments in
orientation. Occupancies of water molecules were not refined because of
the high correlation between occupancy and thermal parameters for data
of 2.6 Å nominal resolution. Thus solvent sites with B values between
50 and 80 Å2 probably represent water molecules with
occupancy parameters below 1.0 and thermal parameters substantially
lower than those reported from the refinement.
The model reported here has been deposited with the Protein Data bank, Brookhaven National Laboratory (code 1NHT). The method of Luzzati (21) indicates an uncertainty in coordinates of 0.30 Å. The amino acid sequence used in refinement is identical to that reported by Silva et al. (14). Results of the refinement are in Table I.
The Ramachandran plot (22) for the structures reported here are
comparable with those of Poland et al. (15). As in other crystal forms of the synthetase (14-15, 23-25), the most apparent outlier is Gln-10, which exists in the same conformation in each of the
five independent polypeptide chains of three crystal forms. The program
PROCHECK (26) indicates better stereochemistry for both the low and
room temperature models than is typical for a structure of 2.5 Å resolution. The low temperature data set is superior in nominal
resolution (I
= 2
(I) at 2.6 Å relative to 3.1 Å for the room temperature data). A much improved electron density map is the most significant consequence of the superior data
acquired at low temperature. We observed no significant divergence between the room temperature and low temperature structures.
The average thermal parameter in the low temperature model varies from
5-33 Å2 for atoms of the main chain and 5-37
Å2 for atoms of side chains. The corresponding variations
for the room-temperature structure are 14-43 and 14-44
Å2, respectively, for main and side chain atoms. The
variation in average thermal parameter as a function of residue number
for the structures reported here are comparable with those of the IMP/NO3 complex (15).
The complex of 6-thiophosphoryl-IMP,
GDP, Mg2+, and hadacidin (Fig. 1) is
isomorphous to that of IMP, GDP, NO3,
Mg2+, and hadacidin (15) and to that of hydantocidin
5
-phosphate, GDP, HPO42
,
Mg2+, and hadacidin (25). A view of one monomer of the
synthetase dimer with associated ligands appears in Fig.
2. Conformational changes in the synthetase upon ligation are described
by Poland et al. (15) as are the interactions of GDP,
Mg2+, and hadacidin, which are the same here within
experimental error. Fig. 3 is a pictorial summary of the
important interactions of ligands in the active site of the
synthetase.
We focus, then, on the one feature that sets the present work apart
from the former studies. The IMP and
NO3 sites are connected by continuous
electron density, which we can interpret only as a molecule of
6-thiophosphoryl-IMP (Fig. 4). Omit maps are in complete
agreement with electron density maps generated from coefficients
2Fobs
Fcalc. Thermal parameters of atoms of
6-thiophosphoryl-IMP are comparable with those of the surrounding
protein, suggesting full occupancy of the ligand. Thus, the electron
density between the 6-sulfur and the phosphoryl group cannot arise as a
consequence of the mutually exclusive binding of phosphate and
6-mercapto-IMP, with each ligand in partial occupancy. Finally, a model
for 6-thiophosphoryl-IMP fits the electron density well, leaving behind
no significant density in difference maps. Indeed, a difference map,
based on observed data from the
IMP/NO3
and 6-thiophosphoryl-IMP
complexes and calculated phase angles from the
IMP/NO3
complex, reveals a strong and
well defined peak of electron density in the area between the 6-oxygen
of IMP and the nitrogen of NO3
(Fig.
4).
Interactions involving the 5-phosphate, the ribose, and the base
(exclusive of the 6-thiophosphoryl group) of 6-thiophosphoryl-IMP are
the same as those of IMP in an earlier study (15). The 5
-phosphate hydrogen bonds with OG1 of Thr-129, backbone amide 129, ND2 of Asn-38,
OG1 of Thr-239, and the guanidinium group of Arg-143 of the
symmetry-related monomer (Table II). In addition, three
water molecules, 554, 636, and 732, mediate interactions between the 5
-phosphate and the protein. The 5
-phosphate of 6-thiophosphoryl-IMP along with the side chain of Asp-114 (conserved in all known sequences of the synthetase) lie at the N terminus of helix H4. No direct hydrogen bonds, however, exist between helix H4 and either Asp-114 or
the 5
-phosphate.
|
The 2-OH group of the ribose of 6-thiophosphoryl-IMP hydrogen bonds
with the guanidinium of Arg-303 and the 3
-OH hydrogen bonds with
waters 524 and 699, which in turn interact with backbone carbonyls 126 and 273, respectively. Although the data are not of sufficient
resolution to unambiguously determine the state of puckering of the
ribose, the best fit to the electron density occurs with a low energy
2
-endo conformation (pseudorotation phase angle, 166°).
The torsion angles O5
-C5
-C4
-C3
(
by convention) and
O4
-C1
-N9-C4 (
by convention) are 44° (+synclinal)
and
135° (anti), respectively.
The base of 6-thiophosphoryl-IMP interacts with the amide side chain of
Gln-224 through its N7 and S6 atoms and with the side chain of Asp-13
through its N1 position. Interactions of 6-thiophosphoryl-IMP are
dominated by the 6-thiophosphoryl group. Backbone amides 13, 40, and
224, NZ of Lys-16, NE2 of His-41, and NE2 of Gln-224 hydrogen bond with
oxygen and sulfur atoms of the thiophosphoryl group (Fig. 5). In addition, one of the oxygens of the
thiophosphoryl group coordinates to the Mg2+. The oxygen
atoms of the thiophosphoryl group correspond in position to the oxygens
of NO3 in the
IMP/NO3
complex (15) and to three of
four oxygens of HPO42
in the
hydantocidin 5
-phosphate complex (25). His 41 interacts with the
-phosphate of GDP when NO3
resides
in the phosphoryl site, as observed in the
IMP/NO3
complex (15). In the
hydantocidin 5
-phosphate complex (25), which has
HPO42
at the phosphoryl site, His-41
interacts with the HPO42
molecule,
similar to the interaction observed for His-41 here. Presumably the
increase in formal electrostatic charge from
1 to
2 is responsible
for the interaction of His-41 with the anion located in the phosphoryl
site.
Markham and Reed (7) deduced a G of +5 kcal/mol at
pH 7 for the reaction: GTP + 6-thiophosphoryl-IMP + L-aspartate
GDP + thiophosphate + adenylosuccinate.
Even in the absence of hadacidin, which leads to the dead end kinetic
complex observed here, the combination of GTP, 6-mercapto-IMP,
aspartate, Mg2+, and enzyme should lead to undetectable
levels of product. A plausible outcome of the crystallization of GTP,
6-mercapto-IMP, Mg2+, and hadacidin, then, would be a
6-mercapto-IMP-GTP or 6-mercapto-IMP-GDP complex. The ratio of GDP to
GTP in the active site would depend largely on the rate of hydrolysis
of GTP in the bulk solvent and the rate of exchange of guanine
nucleotides between the active site and the bulk solvent. Instead, the
active site contains 6-thiophosphoryl-IMP and GDP. The presence of
6-thiophosphoryl-IMP in the active site indicates a lower free energy
for the 6-thiophosphoryl-IMP/enzyme complex than for the
substrate/enzyme complex. Thus, the enzyme must be a thermodynamic trap
for the 6-thiophosphoryl intermediate.
Several factors favor the formation of 6-thiophosphoryl-IMP in the
active site of the synthetase. First, the pKa of N1
of 6-mercapto-IMP is approximately 1-1.5 pH units lower than that of
N1 of IMP. Thus, Asp-13, which is a putative catalytic base in the
abstraction of the proton from N1 of IMP (15), may be more effective in
stabilizing the ionized form of the 6-mercapto nucleotide relative to
the 6-oxo nucleotide (Fig. 6). In addition, the S-P
thioester bond (approximately 2.1 Å in length) is 0.5 Å longer than
the O-P ester bond. The increased bond length allows the three oxygens
of the 6-thiophosphoryl group to occupy the positions of oxygen atoms
of NO3 in the
IMP/NO3
complex (15). Markham and Reed
(5) have shown synergy in the binding of IMP, GDP, Mg2+,
and NO3
to the synthetase. The
distance between N of NO3
and the
6-oxygen of IMP in the IMP/NO3
complex
is 2.7 Å, implying a strong electrostatic interaction between an
electron deficient N of NO3
and an
electron-rich 6-oxygen of IMP (15). In the present study, the 6-sulfur
(presumably stabilized initially as a thioanion) and the phosphorus
atom of PO3
(putatively derived from
the
-phosphate of GTP and stabilized in the
NO3
site) form a covalent bond as a
consequence of electrostatic interactions and the increased atomic
radii of sulfur and phosphorus relative to oxygen and nitrogen.
The proposed mechanism for the formation of 6-thiophosphoryl-IMP in the active site appears in Fig. 6. The interaction involving Asp-13 identifies it as a possible catalytic base in the abstraction of the proton from N1 of 6-mercapto-IMP and generation of a 6-thioanion. In fact, the natural substrate IMP may be in the 6-oxyanion state, when bound to the active site (15). Gln-224 may stabilize the 6-thioanion further by a hydrogen bond interaction. Mutation of Asp-13 to alanine inactivates the synthetase completely (27), and mutation of Gln-224 to glutamate reduces kcat by 1000-fold at pH 7.7.1 The interaction involving Gln-224 is relevant to reported differences in the recognition of IMP analogs by the synthetase from mammals and leishmanial parasites (28-29). Allopurinol ribonucleotide, which differs from IMP at position 7 of the purine ring system, is not recognized as a substrate by mammalian adenylosuccinate synthetase but is a poor substrate for the synthetase from leishmanial parasites. As Gln-224 alone recognizes N7 of IMP, the equivalent position in the synthetases from other organisms may be an important determinant of substrate specificity. Unfortunately, the sequence of the synthetase from a leishmanial parasite is not available.
The 6-thioanion then displaces GDP from the -phosphate of bound GTP
(Fig. 6). Both His-41 and Mg2+ probably play important
roles in stabilizing charge development on the
- and
-phosphates
of GTP in the transition state (15). A His-41 to asparagine mutant is
inactive (27) and Mg2+ is required for activity of the
synthetase (9).
The second reaction governed by the synthetase involves the
nucleophilic displacement of thiophosphate from the 6-phosphoryl intermediate by the amino group of L-aspartate
(Fig. 7). The conformation and interactions of aspartate
are inferred from enzyme-bound hadacidin (15). The putative
conformation of bound aspartate favors a hydrogen bond between its
-carboxylate and its
-amino group, suggesting a catalytic
function for the
-carboxylate in abstracting a proton from that
amino group. No direct evidence is available regarding the catalytic
role of the
-carboxylate of aspartate. However, an
-amino acid is
a substrate of the synthetase only if it bears a negatively charged
substituent with carboxylate-like geometry at the
-carbon (30).
Hydroxylamine is the only known substrate for the synthetase that
departs significantly from aspartate in structure and charge
distribution. Hydroxylamine may not require the catalytic support of
the
-carboxylate because the pKa of its amine is
significantly lower than that of the
-amino group of aspartate.
Assuming a catalytic function for the -carboxylate of aspartate
(there are no other candidates provided by the enzyme (15)), a total of
three hydrogen bonds, then, may be involved in promoting the second
reaction: (i) the putative interaction between the
-amino and
-carboxylate groups of aspartate, (ii) the hydrogen bond between
Asp-13 and N1 of the phosphoryl intermediate, and (iii) the hydrogen
bond between His-41 and the 6-phosphoryl group. The first interaction
enhances the nucleophilicity of the amino group, the second hydrogen
bond enhances the electrophilicity of C6 of the intermediate, and the
third hydrogen bond stabilizes the developing charge on the leaving
group (thiophosphate in this case). The extent to which these three
hydrogen bonds promote catalysis in the second reaction cannot be
answered on the basis of the present study.
Poland et al. (15) suggest a more significant interaction between Asp-13 and Mg2+ after formation of the phosphoryl intermediate, hence transforming Asp-13 from a catalytic base in the phosphotransfer step into a catalytic acid for the second reaction. In the present complex, we observe no significant movement of Asp-13 toward the Mg2+. Coordinate uncertainty (0.3 Å), however, could obscure small but catalytically significant changes in the distance of Asp-13 to Mg2+.
Lieberman (9) was the first to propose the formation of
6-phosphoryl-IMP as an intermediate on the reaction pathway governed by
the synthetase, but the results of his study are consistent with two
other proposed mechanisms (2). Miller and Buchannan (31) suggest an
attack of aspartate on C6 of IMP in concert with an attack by the
6-oxygen of IMP on the -phosphate of GTP. Markham and Reed (6) were
unable to find spectral evidence for the formation of a 6-phosphoryl
intermediate and, as a consequence, suggested that the addition of
aspartate to C6 of IMP preceded the phosphorylation of O6. Isotope
exchange studies at equilibrium by Cooper et al. (12),
however, support the formation of 6-phosphoryl-IMP prior to the
nucleophilic attack of aspartate. Furthermore, Bass et al.
(11) demonstrated the chemical exchange of 18O from the
,
-bridging position of GTP to the oxygens of the
-phosphate of
GTP in the presence of enzyme and IMP or in the presence of enzyme,
IMP, and succinate (as an analog of aspartate). Isotope exchange could
occur only if the
-phosphate dissociates from GTP for an interval
long enough to allow the rotational isomerization of the
-phosphate
group. The absence of labeled phosphate in solution demonstrated that
the dissociated
-phosphate remained bound to the enzyme. As IMP was
required for isotope exchange, Bass et al. (11) suggested
that the dissociated
-phosphate existed as 6-phosphoryl-IMP. The
observation here of 6-thiophosphoryl-IMP in the active site of the
synthetase is entirely consistent with the results and conclusions of
the isotope exchange studies.
The atomic coordinates and structure factors (code 1NHT) have been deposited in the Protein Data Bank, Brookhaven National Laboratory, Upton, NY.
We thank Drs. Fred Rudolph and Bruce Cooper for their generous gift of hadacidin.