Recombinant Mouse Muscle Adenylosuccinate Synthetase OVEREXPRESSION, KINETICS, AND CRYSTAL STRUCTURE*

Vertebrates possess two isozymes of adenylosuccinate synthetase. The acidic isozyme is similar to the synthetase from bacteria and plants, being involved in the de novo biosynthesis of AMP, whereas the basic isozyme participates in the purine nucleotide cycle. Reported here is the first instance of overexpression and crystal structure determination of a basic isozyme of adenylosuccinate synthetase. The recombinant mouse muscle enzyme purified to homogeneity in milligram quantities exhibits a specific activity comparable with that of the rat muscle enzyme isolated from tissue and K m parame- ters for GTP, IMP, and L -aspartate (12, 45, and 140 (cid:1) M , respectively) similar to those of the enzyme from Escherichia coli . The mouse muscle and E. coli enzymes have similar polypeptide folds, differing primarily in the conformation of loops, involved in substrate recognition and stabilization of the transition state. Residues 65–68 of the muscle isozyme adopt a conformation not observed in any previous synthetase structure. In its new conformation, segment 65–68 forms intramolecular hydrogen bonds with residues essential for the recognition of IMP and, in fact, sterically excludes IMP from the active site. Observed differences in ligand recognition among adenylosuccinate synthetases may be due in part to conformational variations in the IMP pocket of the ligand-free enzymes.

Vertebrates possess two isozymes of adenylosuccinate synthetase. The acidic isozyme is similar to the synthetase from bacteria and plants, being involved in the de novo biosynthesis of AMP, whereas the basic isozyme participates in the purine nucleotide cycle. Reported here is the first instance of overexpression and crystal structure determination of a basic isozyme of adenylosuccinate synthetase. The recombinant mouse muscle enzyme purified to homogeneity in milligram quantities exhibits a specific activity comparable with that of the rat muscle enzyme isolated from tissue and K m parameters for GTP, IMP, and L-aspartate (12,45, and 140 M, respectively) similar to those of the enzyme from Escherichia coli. The mouse muscle and E. coli enzymes have similar polypeptide folds, differing primarily in the conformation of loops, involved in substrate recognition and stabilization of the transition state. Residues 65-68 of the muscle isozyme adopt a conformation not observed in any previous synthetase structure. In its new conformation, segment 65-68 forms intramolecular hydrogen bonds with residues essential for the recognition of IMP and, in fact, sterically excludes IMP from the active site. Observed differences in ligand recognition among adenylosuccinate synthetases may be due in part to conformational variations in the IMP pocket of the ligand-free enzymes.
Adenylosuccinate synthetase (IMP: L-aspartate ligase (GDPforming), EC 6.3.4.4) catalyzes the first committed step in the de novo biosynthesis of AMP. In vertebrates the synthetase is also a component of the purine nucleotide cycle, which interconverts IMP and AMP (1)(2)(3). Adenylosuccinate synthetases from various sources exhibit significant sequence conservation (ϳ40% sequence identity between eubacteria and mammals, for instance (4)). Nonetheless, synthetases from nearly every eukaryotic organism have ϳ30-residues added to their N termini and truncations at their C termini relative to the bacterial synthetases. Moreover, vertebrates express two distinct forms of adenylosuccinate synthetase (5). The basic isozyme (often called the muscle enzyme or AdSS1) is associated with the purine nucleotide cycle (1), whereas the acidic isozyme (often called the liver enzyme or AdSS2) is associated with de novo biosynthesis of AMP (5). The two isozymes differ not only in their isoelectric points but also in their kinetic properties, regulation, and tissue distribution (2,3).
Adenylosuccinate synthetase from Escherichia coli is the most studied of all synthetases. Its kinetic mechanism is rapid equilibrium random (6). 6-Phosphoryl-IMP is an intermediate in a two-step process that begins with the transfer of the ␥-phosphoryl group of GTP to the 6-keto group of IMP, followed by the displacement of P i from the intermediate by L-aspartate to form adenylosuccinate (7)(8)(9)(10)(11). In the absence of ligands the synthetase from E. coli is an equilibrium mixture of monomers and dimers (7,12). Active site ligands and/or high subunit concentrations favor dimer formation. Hence, crystallization experiments have thus far resulted only in dimer assemblies (9 -11, 13-19). IMP alone may stabilize the synthetase dimer and by itself triggers conformational changes in the active site of the E. coli synthetase. 1 However, recent structures of the synthetase from plants, Arabidopsis thaliana and Triticum aestivum, reveal organized active sites in the presence of GDP alone (19).
More than two decades ago, the basic isozymes from rabbit (20) and rat (21) were purified and characterized. Subsequent studies confirmed a random sequential kinetic mechanism for these muscle synthetases (6,8). Several groups reported crystals of the basic isozyme (20,21), as well as preliminary x-ray diffraction data (22), but no structure of the basic isozyme has appeared in the literature. Only the cDNA encoding the mouse muscle isozyme has been cloned (4). Reported here are first instances of heterologous overexpression and structure determination of the basic isozyme of adenylosuccinate synthetase. The enzyme is expressed in E. coli using an N-terminal, polyhistidyl tag to facilitate purification. The purified enzyme crystallizes readily in the absence of ligands. The resulting crystal structure of the ligand-free protein at 2.5 Å resolution reveals a polypeptide fold similar to those of synthetases from E. coli and plants. The active sites of the basic isozyme and the E. coli synthetase, however, differ significantly in the conformation of loops, which define, in part, the IMP and GTP binding sites. Indeed, the basic isozyme, as seen here, cannot bind IMP as does the E. coli synthetase without a conformational change involving residues that up to now have been conformationally invariant among all known synthetase structures.
(DE3), and the thrombin cleavage capture kit were from Novagen, Inc. Nickel-nitrilotriacetic acid-agarose was from Qiagen. All other reagents, including GTP, IMP, L-aspartate, bovine serum albumin, and DEAE-Sepharose were from Sigma unless noted otherwise.
Construction of pAdSS1a Expression Plasmid-cDNA for mouse muscle adenylosuccinate synthetase (AdSS I) was kindly provided by Dr. F. B. Rudolph (Department of Biochemistry and Cell Biology, Rice University, Houston, TX) as a pBluescript SK clone (4). A fragment of 1351 base pairs was amplified using the following primers: forward, 5Ј-CCCTTGTCATATGTCCGGGACCCGAGCCTC-3Ј (NdeI restriction site underlined) and reverse, 5Ј-CCGCTCGAGAAAAAGCTGGATCAT-GGACTCTC-3Ј (XhoI restriction site underlined). Insertion of the amplified fragment into corresponding sites of the pET28b expression vector resulted in the plasmid pAdSS1a.
Expression and Purification of the Recombinant Muscle Isozyme-Protein was expressed in E. coli BL21 (DE3) and grown at 37°C in LB medium containing 50 g/ml kanamycin. After the cell culture reached an A 600 of 1.4, isopropyl-1-thio-␤-D-galactopyranoside was added to a final concentration of 0.3 mM, and the culture was maintained at 15°C overnight. The cells were collected at 4000 ϫ g for 10 min at 4°C. Harvested cells were disrupted by sonication in lysis buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, 10 mM imidazole, 10% glycerol, pH 8). After centrifugation at 20,000 ϫ g for 30 min, SDS-polyacrylamide gel electrophoresis of samples from both the supernatant and pellet revealed the production of a protein (ϳ52 kDa), which accounted for ϳ10% of the total protein. Most of the protein, however, was in inclusion bodies.
The supernatant solution was loaded onto a nickel-nitrilotriacetic acid-agarose column previously equilibrated with 50 mM NaH 2 PO 4 , 300 mM NaCl, 10 mM imidazole, pH 8. The column was washed sequentially with two buffers (10 volumes of each), differing from that above only in the concentrations of imidazole (first 20 mM and then 30 mM). An imidazole concentration of 300 mM eluted the target protein. After dialysis in 50 mM Hepes, 50 mM NaCl, 1 mM dithiothreitol, and 0.5 mM EDTA, pH 7.5, the enzyme was loaded at 0.5 ml/min onto a DEAE-Sepharose column, equilibrated with the dialysis buffer. At pH 7.5 all contaminating proteins bind to the resin, whereas the recombinant muscle isozyme elutes with the void volume. SDS-polyacrylamide gel electrophoresis of 30 g of the His-tagged protein revealed no additional bands of protein. The recombinant protein can be frozen at Ϫ80°C and thawed in the dialysis buffer without any significant loss of activity. The His tag was removed as described in the protocol of the thrombin cleavage capture kit. The recombinant protein, with and without the His tag, was used in subsequent kinetics and crystallization experiments.
Enzyme Assay-Protein concentration was determined by the method of Bradford (23), using bovine serum albumin as a standard. Enzyme activity was determined at an absorbance of 280 nm and 22°C as described previously (24). The assay buffer contained 20 mM Hepes, pH 7.2, and 8 mM magnesium acetate. Using up to 1 g/ml enzyme, the reaction was linear for 1 min. K m and V max values for each substrate were obtained by holding the other two substrates at saturating levels (100 M for GTP, 300 M for IMP, and 2000 M for L-aspartate) and varying the third substrate. GTP was varied from 5 to 100 M, IMP varied from 25 to 500 M, and L-aspartate varied from 100 to 1000 M. All of the kinetic data were analyzed with the computer program ENZFITTER (25).
Crystallization-Crystals were grown by the method of hanging drops. Equal parts of a protein solution (10 mg/ml of protein in 50 mM Hepes, pH 7.5, 50 mM NaCl, 1 mM dithiothreitol, 0.5 mM EDTA) and precipitant solution (200 mM calcium acetate, 100 mM cacodylate, pH 6.5, 18% polyethylene glycol 8000) were combined in 4-l droplets. The wells contained 500 l of the precipitant solution. Equal dimensional prisms of 100 microns appeared within 3 days at 22°C. The crystals were transferred in three steps (10-min intervals) to solutions containing 100 mM calcium acetate, 50 mM cacodylate, pH 6.5, 19% (w/v) polyethylene glycol 8000, and glycerol at 7, 14, and 21% (v/v) and then frozen in liquid nitrogen.
Data Collection, Model Building, and Refinement-The data were collected at Beamline 9-2 of Stanford Synchrotron Radiation Laboratory, using the CCD detector, ADSC Quantum 4. The wavelength of x-radiation was 0.979 Å, and the temperature of data collection was 120 K. Data reduction was done with Denzo/Scalepack package (26). The structure was solved by molecular replacement with the program AmoRe (27). The initial model was the E. coli synthetase (Protein Data Bank identifier 1SON) (18). Model building and refinement employed the programs XTALVIEW (28) and CNS (29), respectively. Force constants and parameters of stereochemistry came from Engh and Huber (30). Individual thermal parameters were refined subject to the follow-ing restraints: nearest neighbor, main chain atoms, 1.5 Å 2 ; next-tonearest neighbor, main chain atoms, 2.0 Å 2 ; nearest neighbor, side chain atoms, 2.0 Å 2 ; and next-to-nearest neighbor, side chain atoms, 2.5 Å 2 . Minimum and maximum allowable thermal parameters were 5 and 100 Å 2 , respectively. Criteria for the addition of water molecules were identical to those of previous studies (9 -11). Estimates of coordinate error used the method of Luzzati (31). Evaluation of stereochemistry of the refined model employed PROCHECK (32). Superposition of structures employed software from the CCP4 package (33). Calculations of accessible surface area employed CNS (29), using the approach of Lee and Richards (34) and a probe radius of 1.4 Å.

RESULTS AND DISCUSSION
Enzyme Characterization-Previous studies of the rat and rabbit muscle isozymes isolated from tissue reported a high K m for IMP and a low K m for L-aspartate relative to the acidic isozymes (5,20,21,35). Largely because of these findings, the basic and acidic isozymes were cast as integral components of the purine nucleotide cycle and de novo AMP biosynthesis, respectively (5,35). The K m values for GTP, IMP, and L-aspartate of the recombinant mouse muscle isozyme are 12 Ϯ 2, 45 Ϯ 7, and 140 Ϯ 15 M, respectively, values that are more similar to those of the E. coli synthetase (20 -50, 20 -60, and 200 -350 M, respectively) (36 -38), than values reported for other muscle isozymes isolated from tissue (10 -120, 107-700, and 250 -330 M, respectively) (5,20,21,35,39). The relatively large range in reported K m values for the synthetase is probably due in part to technical difficulties in establishing linear progress curves. GDP is a potent product inhibitor of all known adenylosuccinate synthetases (2) and greatly limits the linear range of progress curves in the determination of K m for GTP. Furthermore, the synthetase may exist as a mixture of monomers and dimers governed by an equilibrium constant sensitive to substrate concentration (12). Hence, even the oligomeric state of the synthetase may change with substrate concentration at a fixed enzyme concentration. The highest values for K m come from single point, stopped assays (5,21), which in addition to the complications noted above, provide no information by which to assess the linearity of the progress curve. Excluding determinations based on single point assays, the K m values for enzymes from rabbit and rat muscle are 10 -24, 107-320, and 290 -300, respectively (20,35,39,40), for GTP, IMP, and Laspartate, values approximately 2-fold higher than those reported here for the recombinant enzyme from mouse muscle.
The recombinant mouse muscle isozyme has almost the same specific activity as the rat muscle enzyme (5.74 versus 6.24 mol/min/mg) (21) and is approximately five times more active than the protein isolated from rabbit muscle (1.0 mol/min/mg) (41). We observe no evidence of substrate inhibition for IMP up to a maximum concentration of 500 M. Furthermore, thrombin cleavage of the His-tagged synthetase and subsequent purification do not alter the kinetic parameters of the recombinant enzyme. In fact, the thrombin-cleaved protein forms crystals isomorphous to those described below, with a structure identical to within experimental uncertainty.
Structural Features of the Muscle Isozyme-Crystals of ligand-free, mouse muscle enzyme diffract to a resolution (2.5 Å) comparable with crystalline synthetases from other sources. The space group and unit cell dimensions are virtually identical to those reported for the rat muscle isozyme (22). Statistics of data collection and refinement are in Table I. The stereochemistry of the refined model, as analyzed by the program PROCHECK (32), exceeds that typically observed for a structure of 2.5 Å resolution. No residues lie in disallowed regions of the Ramachandran plot, and 87% of all the residues are in the most favored regions. The average uncertainty in coordinates is ϳ0.3 Å. Thermal parameters vary from 16 to 100 Å 2 , the latter being an upper limit imposed by the software.
We observe reliable electron density beginning with residue 26. Hence, the first 25 residues plus the His tag construct are disordered and not included as part of the model. The His tag/leader element (ϳ50 residues) is presumably intact, given the successful isolation of the recombinant protein from nickelnitrilotriacetic acid-agarose and the lack of any evidence for heterogeneity (by gel electrophoresis) in the purified protein after 1 week of storage at room temperature. The observed starting position in the electron density precedes the beginning of the E. coli sequence and structure by five amino acid residues. The trace through these initial five residues follows a path different from that of the plant synthetases (Fig. 1). Electron density is weak or missing for residues 151-164, which correspond to residues 120 -130 of the E. coli synthetase. Corresponding residues from plant synthetases are disordered as well in the absence of bound IMP. Loop 120 -131 of the E. coli synthetase becomes ordered, however, upon ligation of the active site (9). As expected from the high level of sequence identity (Fig. 2), the overall fold of the mouse muscle synthetase is almost identical to that of other synthetases (Fig. 1). The structure consists of a ␤-sheet core made up of nine parallel strands (B14, B10, B7, B1, B6, B2, B3, B4, and B5) and one antiparallel strand (B15). The core ␤-sheet is flanked by subdomains that contain 11 ␣-helices, seven antiparallel ␤-strands, and five 3 10 helices.
Secondary structures of the mouse muscle, E. coli, and plant synthetases are in agreement, with only a few differences. Residues 393-398 of the ligand-free, mouse muscle isozyme form a single-turn, 3 10 helix, similar in conformation to the corresponding residues of the ligated synthetase from E. coli and the plant synthetases but different from that of the ligandfree enzyme from E. coli. Loop 330 -336 of the mouse muscle enzyme is in an ordered conformation, whereas the corresponding segment (loop 298 -304) of ligand-free E. coli enzyme is disordered. Hydrogen bonds, involving backbone carbonyls 333 and 334 with the side chains of Ser 428 , Arg 431 , and Asn 435 of a lattice neighbor, may stabilize the conformation of the loop 330 -336 of the muscle synthetase. In crystal structures of the plant synthetases, the corresponding loops are also involved in lattice contacts, despite differences in space groups (P4 1 2 1 2 for T. aestivum and I23 for A. thaliana).
Subunit Interface of the Dimer-The molecular 2-fold axis of the mouse muscle synthetase coincides with a crystallographic 2-fold axis of symmetry. The contact area of the subunit interface (excluding residues with thermal parameters in excess of 80 Å 2 ) is about the same in the mouse muscle (5150 Å 2 ), ligand-free E. coli (5270 Å 2 ) (15), A. thaliana (4830 Å 2 ) (19), and T. aestivum (5250 Å 2 ) (19) enzymes. The subunit interface of the mouse muscle dimer has fewer salt links than those of the E. coli or plant synthetases, a comparable number of hydrogen bonds, and more hydrophobic contacts. Lys 174 makes a salt link with Asp 263 (donor-acceptor distance of 2.6 Å), which is an interaction conserved in all known synthetase structures (Fig.  2). Hydrogen bonds between subunits of the mouse muscle synthetase replace other salt links of E. coli and plant synthetases. Hydrogen bonds number 28, 32, 24, and 23 at the subunit interfaces of mouse muscle, ligand-free E. coli, A. thaliana, and T. aestivum synthetases, respectively, using a distance cutoff of 3.3 Å and a thermal parameter cutoff of 80 Å 2 . Unique to the mouse muscle synthetase are hydrogen bonds, Asp 184 -Asn 392# , Ser 187 -Asn 392# , Asp 234 -backbone carbonyl 389#, and Gln 289 -backbone carbonyl 355#. (The # symbol designates a residue from the subunit related by molecular symmetry). Hydrophobic contacts involve two clusters of nonpolar side chains: Val 236 , Phe 264# , Met 352# , and Val 353# , and close to the molecular 2-fold axis Met 206 , Phe 207 , Met 206# , and Phe 207# (Fig. 3). Plant synthetases have nonpolar residues corresponding to Phe 207 , Val 236 , Phe 264 , and Val 353 , but Met 206 and Met 352 of the mouse muscle enzyme correspond to polar residues in the plant synthetases.
Catalytic function of adenylosuccinate synthetase from E. coli is sensitive to the integrity of the subunit interface. Mutations of Lys 140 and Asp 231 , which correspond to Lys 174 and Asp 263 , respectively, of the mouse muscle enzyme, inactivate the synthetase (42) or elevate the K m for IMP and GTP by 40and 20-fold, respectively (38). At subunit concentrations below 10 M, and in the absence of active site ligands, the E. coli synthetase is largely monomeric (12) and presumably inactive, although the latter has yet to be demonstrated. Synthetases from plants reportedly behave as monomers in gel filtration, but light scattering data are consistent with a dimer (19). The basic isozyme is reportedly a stable dimer (2, 3, 21, 39). Conformational variations among synthetases in regions corresponding to residues 184 -193, 204 -213, and 385-395 of mouse muscle enzyme (Figs. 1 and 3) are at or close to the subunit interface of the dimer. Known crystal structures of synthetases, however, reveal no certain basis for differences in dimer stability, assuming that such differences exist.
The present structure provides no insight as to the role (if any) of the 30-amino acid extension to the N terminus of the mouse muscle enzyme (hereafter called the leader sequence). Leader sequences among members of the synthetase family are dissimilar. The fusion of lacZ to the N terminus of the acidic human isozyme, however, results evidently in an inactive construct (43). As significant evidence stands against the direct participation of the leader sequence in catalysis (the E. coli synthetase, for instance, has no leader sequence), the fusion protein may be incapable of forming an active dimer. No evidence suggests, however, any change in catalytic properties or structure because of the presence of the N-terminal His tag.
Active Site-Loops 38 -53, 120 -131, 298 -304, and 417-421 of the E. coli synthetase undergo significant conformational where i runs over multiple observations of the same intensity and j runs over crystallographically unique intensities.
b All unique data in the resolution range 39.6 -2.5 Å. c R factor ϭ ⌺ʈF obs ͉ Ϫ ͉F calc ʈ/⌺͉F obs ͉, ͉F obs ͉ Ͼ 0. d R factor based upon 10% of the data randomly culled and not used in the refinement. change in response to ligation of the active site (9,38). The corresponding loops in the mouse muscle enzyme (residues 68 -83, 151-165, 330 -336, and 448 -452, respectively) are probably sensitive to active site ligands as well. For the most part, residues in direct contact with active site ligands in the E. coli synthetase are identical to corresponding residues of the mouse muscle synthetase. By analogy to the E. coli synthetase, loops 151-165, 330 -336, and 448 -452 are part of the IMP-, the L-aspartate-, and the GTP-binding pockets, respectively, and specific residues of loop 68 -83 probably interact with IMP, GTP, and Mg 2ϩ .
Because the aforementioned loops seem important to the function of all known synthetases and yet have different sequence numbers, we suggest a generalized nomenclature. Hereafter, we suggest that the names Switch loop, IMP loop, Asp loop, and GTP loop represent segments corresponding to residues 38 -53, 120 -131, 298 -304, and 417-421, respectively, of the E. coli synthetase (Fig. 1).
As noted above, the IMP loop of the mouse muscle enzyme is disordered. In structures of the E. coli and plant synthetases without IMP, the corresponding residues are also disordered.
In the presence of IMP the conserved threonines at positions 162 and 163 of the mouse muscle enzyme should hydrogen bond with the 5Ј-phosphoryl group of IMP. The IMP loop of the mouse muscle enzyme has three more residues than those from E. coli and plant synthetases (Fig. 2). Glu 118 of E. coli synthetase, however, hydrogen bonds with the IMP loop of the ligated enzyme (9). If Glu 149 (the corresponding residue of the mouse muscle synthetase) plays a similar role, then the additional residues may influence the conformation of the mouse muscle IMP loop only as it emerges from helix H3. Conformational differences in this area of the IMP loop, however, could still influence interactions between subunits of a synthetase dimer (Fig. 3).
In contrast to the disordered Asp loop of the E. coli synthetase, the corresponding loop of the mouse muscle enzyme adopts a well defined conformation, probably because of the stabilizing effect of lattice contacts (see above). The conformation of the Asp loop in the basic isozyme is similar to that of the E. coli synthetase in its complex with hadacidin (an analog of L-aspartate) (9). On the basis of model building and energy minimization, the Asp loop of the mouse muscle synthetase will retain its observed conformation but will move 5 Å toward the active site to hydrogen bond with hadacidin.
Superposition of the mouse muscle and E. coli synthetases reveals a 1.6 Å root mean square difference between their respective GTP loops. The observed difference may be related to a much larger conformational difference in the Switch loop, described below. Residues of the GTP loop define in part the binding pocket for the base of the guanine nucleotide. In the E. coli synthetase, Ser 414 , the side chain of which hydrogen bonds with the O-6 atom of the guanine nucleotide, and Pro 417 , which stacks against the base of the guanine nucleotide, are Gly 420 and Lys 423 , respectively, in the mouse muscle isozyme. Neither changes in residue type nor conformation, however, greatly influence the K m for GTP (see above). In the plant enzymes Ser 414 is also replaced by glycine. Evidently, the backbone amide of the glycine recognizes the O-6 atom of the guanine nucleotide as well as the OG atom of Ser 414 .
The Switch loop of the E. coli synthetase adopts a ligand-free conformation or a ligated conformation, both of which differ significantly from that of the mouse muscle synthetase (Fig. 1). The root mean square difference between corresponding ␣-carbons of the mouse muscle loop relative to the ligand-free and ligated E. coli loops is 3.0 and 2.7 Å, respectively. Hydrogen bonds involving Asn 116 , Lys 119 , and Gly 120 with Asp 81 , Glu 79 , and Tyr 80 , respectively, stabilize the observed conformation of mouse muscle Switch loop. The above residues, with the exception of Gly 120 , are different in E. coli and plant synthetases (Fig. 2). The ligand-free position of the corresponding loop in plant synthetases is unknown. Variations in the sequence of the plant synthetases relative to E. coli and mouse muscle synthetases may give rise to yet another loop conformation in the absence of ligands.
Regardless of the different ligand-free conformations for the above loop, the hydrogen bond between His 71 and Asp 51 (His 41 and Asp 21 , respectively, in the E. coli synthetase) is present in each of the ligand-free synthetases, for which crystal structures are available. Furthermore, His 41 of the E. coli synthetase is an essential catalytic residue (9,36), and Asp 21 represents onehalf of an important switching mechanism, which defines the ligated and ligand-free conformations of the E. coli synthetase (7,9,38). In the proposed switching mechanism (7), hydrogen bond formation between Asn 38 and the 5Ј-phosphoryl group of IMP leverages the Switch loop into its ligated conformation, which is stabilized further by the salt link between Asp 21 and Arg 419 (of the GTP loop). Because these residues are invariant over the known synthetase family, the histidine-aspartate and arginine-aspartate links, defined above, are probably hall-marks of the ligand-free and ligated conformations, respectively, of all adenylosuccinate synthetases.
Conformational differences in the aforementioned loops are not entirely surprising, because these structures are flexible. The gain or loss of a single hydrogen bond can result in a significant conformational difference. Residues 35-38 of the E. coli synthetase, however, are well ordered with low thermal parameters. Hence, the significant difference in conformation for residues 65-68 of the mouse muscle enzyme relative to residues 35-38 of the E. coli enzyme is unanticipated (Fig. 4). In fact, not only is Asn 68 of the mouse muscle synthetase removed from the 5Ј-phosphoryl pocket of IMP, segment 65-68 sterically excludes IMP from binding (Fig. 4). Furthermore, a hydrogen bond between Asn 265 and backbone amide 67 may stabilize the observed conformation of residues 65-68. Asn 256 may itself be an important catalytic residue. Gln 224 of the E. coli synthetase, which corresponds to Asn 256 of the mouse muscle enzyme, hydrogen bonds with the atoms N-7 and O-6 of IMP and putatively stabilizes the 6-oxyanion form of IMP (9,37). Finally, residues 65-68 have elevated thermal parameters that suggest conformational mobility and the possibility of a ligand-induced conformational change. The conformation of this loop has been verified by the collection of a data set from a second crystal, by the use of omit electron density maps, and by the refinement of an E. coli-like model, the latter resulting in an inferior fit to the electron density. The altered conformation of residues 65-68 may be a distinct feature of the ligand-free, basic isozyme and may go hand-in-hand with the observed conformational differences in the Switch loop.
Any functional significance attached to the altered conformation of segment 65-68 is merely speculative, but the muscle isozyme is inhibited at least 10-fold more potently by fructose 1,6-bisphosphate than the E. coli synthetase. 2 Others have suggested a connection between glycolysis and the purine nucleotide cycle on the basis of the in vitro inhibition of the muscle isozyme by fructose 1,6-bisphosphate (2, 5), but whether the in vivo concentrations of fructose 1,6-bisphosphate in muscle ever rise to levels, which could cause appreciable inhibition of the synthetase, is unknown. Spector and Miller (40) report allopurinol ribonucleotide as a weak inhibitor of the synthetase from rabbit muscle, yet this IMP analog is a substrate for the synthetase from the protozoan Leishmania donovanii (44). Variations in substrate specificity between muscle and protozoan synthetases may arise from conformational differences involving residues 65-68. Atom N-7 of IMP may need to hydrogen bond with Asn 256 of the mouse muscle enzyme to trigger a conformational change in residues 65-68. If so, atom N-7 of IMP would compete with backbone amide 67 for the side chain of Asn 256 (Fig. 4). The absence of a proton acceptor at position 7 of allopurinol ribonucleotide would give backbone amide 67 a clear advantage in stabilizing a conformation, which is antagonistic toward the IMP analog. The existence of an active site conformation that excludes IMP also allows for alternative mechanisms of regulation. Conceivably, an inhibitor need not look like IMP or bind to the active site as does IMP to inhibit the mouse muscle enzyme. Any ligand that stabilizes the IMPantagonistic conformation observed here would be an inhibitor of the synthetase. The muscle isozyme, then, may be sensitive to more than just feedback inhibition by AMP. Evidence indicates, for instance, an association of the muscle synthetase with F-actin (45,46). Hence, interactions between the muscle isozyme and other muscle proteins could influence or even determine the level of synthetase catalysis in vivo.