Variations in the Response of Mouse Isozymes of Adenylosuccinate Synthetase to Inhibitors of Physiological Relevance*

Vertebrates have acidic and basic isozymes of adenylosuccinate synthetase, which participate in the first committed step of de novo AMP biosynthesis and/or the purine nucleotide cycle. These isozymes differ in their kinetic properties and N-leader sequences, and their regulation may vary with tissue type. Recombinant acidic and basic synthetases from mouse, in the presence of active site ligands, behave in analytical ultracentrifugation as dimers. Active site ligands enhance thermal stability of both isozymes. Truncated forms of both isozymes retain the kinetic parameters and the oligomerization status of the full-length proteins. AMP potently inhibits the acidic isozyme competitively with respect to IMP. In contrast, AMP weakly inhibits the basic isozyme noncompetitively with respect to all substrates. IMP inhibition of the acidic isozyme is competitive, and that of the basic isozyme noncompetitive, with respect to GTP. Fructose 1,6-bisphosphate potently inhibits both isozymes competitively with respect to IMP but becomes noncompetitive at saturating substrate concentrations. The above, coupled with structural information, suggests antagonistic interactions between the active sites of the basic isozyme, whereas active sites of the acidic isozyme seem functionally independent. Fructose 1,6-bisphosphate and IMP together may be dynamic regulators of the basic isozyme in muscle, causing potent inhibition of the synthetase under conditions of high AMP deaminase activity.

The impact of the PNC on the metabolism of various tissues is unsettled (18 -21), as is the assignment of the two isozymes to mutually exclusive metabolic roles (18). Moreover, in muscle, where exercise can increase IMP concentration up to 50-fold, the PNC may work asynchronously; AMP deaminase works only when AdSS1 and/or adenylosuccinate lyase is quiescent (3,(22)(23)(24)(25). During recovery (restoration of basal IMP levels) AMP deaminase is inactive (24,26). Indeed, interactions with myosin activate AMP deaminase, a process regulated by the decrease in the ATP concentration during exercise (24,26). No study has demonstrated regulation of AdSS1 in the context of the PNC, although slight inhibition of the basic isozyme occurs at high concentrations of IMP (8,13,27).
Mouse recombinant AdSS1 has a significantly lower K m for IMP than that reported for the basic isozyme isolated from either rat or rabbit (28). Reported K m values for adenylosuccinate synthetases vary considerably (2,6,28) due, in part, to variations in assay protocols and conditions of assay, as well as intrinsic differences in the synthetases themselves (2,6). The low natural abundance of AdSS2 has been an impediment to its purification and rigorous evaluation (7,9,18). Preparations of AdSS2 from malignant cells, such as Novikoff ascites tumor cells (29) and Yoshida sarcoma tumor cells (30) save in one instance (30), specific activities significantly lower than that of AdSS1 (2,6). Human and mouse AdSS2 have been cloned (3,4) and the latter overexpressed in COS (African green monkey kidney) cells, but no kinetic characterization was reported (3). Native states of oligomerization for each isozyme remain ambiguous, as reports of monomeric and dimeric AdSS1 and AdSS2 are in the literature (8, 10, 29 -31). In contrast, the synthetase from Escherichia coli is active as a dimer (32,33) and exists in a monomer-dimer equilibrium (34).
Reported here are first instances of heterologous overproduction and kinetic characterization of mouse AdSS2. The k cat values for recombinant AdSS2 and AdSS1 are almost identical. AdSS2, relative to AdSS1, has a slightly lower K m for IMP and GTP and a significantly higher K m for L-aspartate. High (but physiologically relevant) concentrations of IMP inhibit AdSS1 but not AdSS2. Adenylosuccinate, GDP, and GMP are strong inhibitors of both isozymes, but AMP, which potently inhibits AdSS2, is a weak inhibitor of AdSS1. Furthermore, the kinetic mechanism of AMP inhibition differs for the two isozymes. Fru-1,6-P 2 might be a physiologically significant inhibitor of AdSS1 but not of AdSS2. Truncated isozymes (N-terminal leader sequence removed) retain the kinetic properties and state of oligomerization of their full-length counterparts. Mouse isozymes exhibit a monomer-dimer equilibrium, and both GTP and IMP stabilize the dimer. Differences in mouse synthetases reported here do not support mutually exclusive metabolic roles for the two isozymes. Moreover, our findings support the asynchronous operation of the PNC in muscle.

EXPERIMENTAL PROCEDURES
Materials-E. coli strain BL21 (DE3), plasmid pET28b, nickel-nitrilotriacetic acid-agarose, and the thrombin cleavage capture kit were from Novagen, Inc. Restriction enzymes, DNA ligase, and Vent Polymerase were from New England Biolabs. All other reagents were from Sigma unless noted otherwise.
Construction of Full-length and Truncated Synthetases-The cloning of full-length AdSS1 into the expression plasmid pET28b was described previously (28). cDNA for mouse acidic adenylosuccinate synthetase (AdSS2) was kindly provided by Dr. F. B. Rudolph (Department of Biochemistry and Cell Biology, Rice University, Houston, TX) as a pSPORT1clone (11). A fragment of 1371 bp was amplified using the following primers: forward, 5Ј-CCCTTGTCATATGTCGATCTCCGAG-AGCAGC-3Ј (NdeI restriction site underlined), and reverse, 5Ј-CCGC-TCGAGTTAGAAGAGCTGAATCATGGACTC-3Ј (XhoI restriction site underlined). Insertion of the amplified fragment into corresponding sites of the pET28b expression vector resulted in the plasmid pAdSS2a. An NdeI restriction site located into the AdSS2 open reading frame was removed by a silent mutation. Truncated AdSS1 (AdSS1-Tr) and AdSS2 (AdSS2-Tr) were generated using the forward primers 5Ј-CCCTTGTC-ATATGACTGGCTCTCGCGTGACCGTG-3Ј and 5Ј-CCCTTGTCATAT-GGGGAACCGGGTGACTGTGGTG-3Ј, respectively (NdeI restriction sites underlined). All constructs were checked by sequencing (Iowa State University DNA sequencing facility).
Expression and Purification of Recombinant Isozymes-AdSS1 was produced and purified as described previously (28). The same procedure was used for AdSS1-Tr. AdSS2 was expressed in E. coli BL21 (DE3) at 37°C in LB media containing 30 g/ml kanamycin. Cells were collected by centrifugation (4,000 ϫ g for 10 min at 4°C), resuspended in lysis buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, 10 mM imidazole, pH 8, and 1 mM phenylmethanesulfonyl fluoride), and then disrupted by sonication. After centrifugation (24,000 ϫ g for 30 min), the supernatant 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 10 volumes of each of the three buffers, differing from the above only in the concentration of imidazole (20,30, and then 40 mM). Bound protein eluted with 300 mM imidazole. After dialysis in 50 mM HEPES, 50 mM NaCl, 1 mM dithiothreitol, and 0.5 mM EDTA, pH 7.0, the enzyme was loaded at 0.5 ml/min onto a DEAE-Sepharose column, equilibrated with dialysis buffer. AdSS2 was eluted by a linear gradient (0 -200 mM NaCl). AdSS2-Tr was overproduced and purified as above with one modification; cell cultures were maintained at 25°C after induction. At 37°C the truncated protein appears in inclusion bodies. Removal of the polyhistidyl tag employed the thrombin cleavage capture kit. Recombinant E. coli adenylosuccinate synthetase was overproduced and purified as described elsewhere (34).
Enzyme Assay-Protein concentration was determined by the method of Bradford (36), using bovine serum albumin as a standard. Enzyme activity was determined at an absorbance of 280 nm and at 22°C as described previously (37). A standard assay buffer for AdSS1 contained 40 mM HEPES, pH 6.7, 8 mM magnesium acetate, 150 M GTP, 250 M IMP, and 2 mM aspartate. For AdSS2, the assay buffer contained 40 mM HEPES, pH 6.7, 8 mM magnesium acetate, 150 M GTP, 200 M IMP, and 8 mM L-aspartate. The reaction was started by the addition of up to 1 g/ml enzyme. Under these conditions the reaction was linear for 1 min. The Hill coefficient for Mg 2ϩ was determined by varying the concentration of magnesium acetate from 0.2 to 4 mM and from 0.05 to 2 mM for AdSS1 and AdSS2, respectively. K i values for Fru-1,6-P 2 , AMP, GDP, and GMP were determined by holding two substrates at saturating levels and varying the concentration of the third substrate over 1Ϫ8 K m , at different fixed concentrations of inhibitors ranging over 0.5Ϫ2 K i . In experiments to determine the K i of IMP inhibition, concentrations of GTP varied from 15 to 200 M, those of IMP ranged from 40 to 4,000 M, and concentrations of L-aspartate and Mg 2ϩ were 2 and 8 mM, respectively. Kinetic data were analyzed with the computer program GraFit (38).
Analytical Ultracentrifugation-Sedimentation equilibrium experiments were performed in a Beckman Optima XL-A analytical ultracentrifuge using an An-60 Ti rotor, rotor speeds of 9,000 and 15,000 rpm, a temperature of 4°C, and protein concentrations of 0.1Ϫ0.5 mg/ml in 20 mM HEPES, pH 7.2, 20 mM NaCl, and 1 mM dithiothreitol. Centrifugation in the presence of ligands (25 M IMP, 25 M GTP, 1 M hadacidin, 2 mM magnesium acetate) employed only one concentration of protein (0.3 mg/ml). Samples were centrifuged for 12 h. Equilibration was verified then by 3 scans recorded at 4-h intervals. Stepwise radial scans were performed at 280 nm, using a step-size of 0.001 cm, with each datum being the average of 30 measurements. Data were analyzed using the "Ideal" model on the Optima XL-A Analysis software (version 2.0). Partial specific volumes of 0.741, 0.743, 0.740, and 0.745 ml/g for AdSS1, AdSS1-Tr, AdSS2, and AdSS2-Tr, respectively, were determined from the amino acid composition and published tables (39). Samples were centrifuged at 40,000 rpm for 15 h to sediment all protein, and then radial scans were recorded to obtain a base-line correction for each cell. Inclusion of the second virial coefficient did not improve fits, indicating that nonideality is not present in the system.
Partial Purification of Mouse Isozymes from Tissue-Mouse liver and skeletal muscle were disrupted using a Polytron homogenizer (Brinkmann Instruments) in a buffer containing 50 mM HEPES, pH 7.0, 50 mM NaCl, 1 mM dithiothreitol, 0.5 mM EDTA, 1 mM phenylmethanesulfonyl fluoride, and 15 g/ml leupeptin. The homogenate was centrifuged at 25,000 ϫ g for 30 min. The recovered supernatant fluid was heated (60°C, 1 min) and then centrifuged (25,000 ϫ g, 45 min). The supernatant was subjected to ammonium sulfate fractionation. Proteins that precipitated in the range of 40Ϫ60% (w/v) ammonium sulfate were retained. At this point adenylosuccinate synthetase activity was readily detected. Precipitated protein was dialyzed against 25 mM HEPES, 25 mM NaCl, 1 mM dithiothreitol, and 0.5 mM EDTA, pH 7.5, and then loaded onto a DEAE-Sepharose column equilibrated with the same buffer. The column retained the acidic but not the basic isozyme. 1 M NaCl washed the acidic isozyme from the column. Identities of the acidic and basic isozymes were confirmed by Western blots and their kinetic properties.
Antibodies and Western Blots-Polyclonal antibodies against recombinant AdSS1 were raised in rabbits at the Iowa State University protein facility and were then purified by affinity chromatography, using an Econo-Pac serum IgG purification kit (Bio-Rad). A Sepharose 4B column containing immobilized AdSS2 eliminated cross-reacting antibodies. Protein transfer to a nitrocellulose membrane was performed as recommended by the manufacturer (Bio-Rad) in a buffer containing 25 mM Tris, pH 8.3, 192 mM glycine, 20% methanol, and 0.05% SDS. Detection of antigen-antibody complexation employed the Opti-4CN kit (Bio-Rad).

RESULTS
Sequence Comparison and Validation-Nucleotide sequences of mouse AdSS2, reported here and by Guicherit et al. (11), differ at positions 499 (C in Ref. 11 is now G) and 595 (A becomes G). As a consequence, Arg 167 and Thr 199 become glycine and alanine, respectively, identical to corresponding residues in human AdSS2 (4). Moreover, Gly 167 is invariant among synthetases from 43 organisms, representing 30 major phylogenetic lineages. The published sequence of human AdSS2 (4) also differs with respect to other sequence information; Ala 24 should be arginine and an additional residue (proline) comes after position 24. The N-terminal leader sequence in mouse and human AdSS2 then each have 26 amino acid residues (Fig. 1). Sequence identity between AdSS1 and AdSS2 from the same source (mouse or human) is ϳ75% but exceeds 95% between like isozymes from different sources. N-terminal leader sequences of AdSS1 and AdSS2 from the same source are only 25% identical. In contrast, N-terminal leader sequences between mouse and human AdSS1 are 97% identical, and those between mouse and human AdSS2 are 73% identical.
Purity and Yield-AdSS1-Tr and AdSS2-Tr are shorter by 28 and 26 amino acid residues, respectively, than their full-length counterparts (Fig. 1). The yield of purified AdSS1-Tr was comparable with that of full-length AdSS1 (5-10 mg/liter of cell culture), but substantial quantities of AdSS1 and AdSS1-Tr did appear in inclusion bodies. In contrast, yields of recombinant AdSS2 and AdSS2-Tr were ϳ 25 mg/liter of cell culture, ϳ20% of the total soluble protein. SDS-PAGE of samples revealed a single band of ϳ50 kDa. Full-length and truncated AdSS2 are stable for several days at 4°C. AdSS2 is stable with respect to freeze/thaw cycles in buffers supplemented with 30% glycerol. Full-length AdSS1, with or without their N-terminal polyhistidyl tags, have identical kinetic parameters and crystallize under the same conditions (28). Similarly, truncation of the N-terminal polyhistidyl tag from AdSS2 does not change its kinetic properties (data not shown).
Thermal Stability-In the absence of ligands, the E. coli synthetase is in a monomer-dimer equilibrium (K d ϳ10 M) (34). The presence of active site ligands, such as IMP and GTP, significantly increases thermal stability of E. coli synthetase. Moreover, the thermal stability of the E. coli synthetase increases with protein concentration. 2 In the absence of ligands, AdSS1 and AdSS2 are more stable than the E. coli synthetase, but IMP and GTP do not greatly enhance the thermal stability of the mammalian isozymes (Fig. 2). AdSS1-Tr and AdSS2-Tr have the same thermal stability as their full-length counterparts, ruling out an effect due to the N-terminal leader sequence.
Native Molecular Weight of AdSS1 and AdSS2-Equilibrium sedimentation ultracentrifugation indicates single species of molecular mass 86.0 Ϯ 3.9 and 90.7 Ϯ 2.5 kDa for AdSS1 and AdSS2, respectively. Predicted masses are 101.1 and 100.7 kDa, derived from amino acid sequences of recombinant AdSS1 and AdSS2, respectively. Equilibrium ultracentrifugation of samples in the presence of IMP, GTP, Mg(acetate) 2 , and hadacidin (N-hydroxy-N-formylglycine, a potent competitive inhibitor with respect to L-aspartate) showed increased molecular masses of 103.6 Ϯ 5.4 and 101.5 Ϯ 4.3 kDa for AdSS1 and AdSS2, respectively. The behavior of AdSS1-Tr and AdSS2-Tr was similar to their full-length counterparts. In the absence of active site ligands, molecular masses were 86.1 Ϯ 5.   nantly dimers, but a small shift from monomer to dimer occurs in the presence of ligands. The K d value for the mammalian isozymes then, in the absence of ligands, is significantly lower than that of E. coli synthetase (34). The molecular weight (78 kDa) of a yeast synthetase, determined by gel filtration, lies between that of a monomer and dimer (40), whereas plant synthetases are either monomers or dimers, depending on the methodology of mass determination (41).
K m , k cat , pH, and Buffer Effects-K m values for AdSS1-Tr are comparable with those of AdSS1 (28), but k cat is slightly lower, perhaps due to a small component of misfolded AdSS1-Tr. Relative to AdSS1, AdSS2 has a lower K m for IMP, a similar K m for GTP, and significantly higher K m for L-aspartate (Table I).
AdSS2-Tr has a slightly higher K m value for L-aspartate than that of AdSS2. K m values for IMP and L-aspartate of tissuederived AdSS1 and AdSS2 were similar to those of the recombinant isozymes (data not shown).
Values of k cat and K m of the E. coli synthetase are sensitive to pH in the range of 7.0Ϫ7.7 (42). As the pH in exercising muscle can decrease from 6.9 to 6.4, we determined whether changes in pH influence kinetic parameters of AdSS1 or AdSS2; however, K m values remained constant from pH 6.5 to 7.2. The pH optima for AdSS1 and AdSS2 lie between 6.6 and 6.9. As the pH optima are nearly out of the buffering range of HEPES (pK a 7.5), other more suitable buffers were tested. MES, MOPS, and PIPES buffers support similar turnovers for AdSS2 and the E. coli synthetase. In contrast, AdSS1 in 40 mM MES and PIPES has only 40 and 10%, respectively, of the specific activity of AdSS1 in HEPES.
Metal Requirement-AdSS2 and the E. coli synthetase reach maximum activities in 2 mM Mg(acetate) 2 , but AdSS1 requires 8 mM Mg(acetate) 2 . Concentrations above 10 mM Mg(acetate) 2 are inhibitory for all systems. Inhibition is more pronounced when Cl Ϫ is a counterion to Mg 2ϩ instead of acetate. Hill coefficients for Mg 2ϩ are 1.1 Ϯ 0.1 for AdSS1 and 1.0 Ϯ 0.2 for AdSS2. The Hill coefficient for Mg 2ϩ is 2 for the E. coli synthetase (43). Crystal structures of AdSS1 (44,45), AdSS2, 3 and the E. coli synthetase (46,47), however, reveal only one Mg 2ϩ per subunit. Evidently, studies of AdSS1 and AdSS2 will not clarify the role of the "second" Mg 2ϩ inferred by the kinetics of the E. coli synthetase.
Metabolite and Substrate Inhibition-Mouse synthetases are subject to potent inhibition by adenylosuccinate, GDP, AMP, and GMP (Table II). In contrast to product and GMP inhibition, AMP inhibits AdSS1 and AdSS2 with marked differences. The K i value for AMP is 12-fold lower for AdSS2 than AdSS1. Moreover, AMP is a competitive inhibitor with respect to IMP for AdSS2 but noncompetitive with respect to IMP for AdSS1.
Fru-1,6-P 2 is the only intermediate of glycolysis that inhibits mammalian adenylosuccinate synthetases at concentrations below 1 mM (1, 2). Fru-1,6-P 2 inhibits AdSS1 more potently than AdSS2 in plots of reciprocal velocity against 1/GTP and 1/L-aspartate (6-and 8-fold, respectively), but inhibition of the two isozymes is nearly the same in plots of reciprocal velocity against 1/IMP. Fru-1,6-P 2 is a noncompetitive inhibitor with respect to all substrates.
High concentrations of IMP (Ͼ1 mM) at a saturating concentration of GTP (Ͼ150 M) inhibit AdSS1 but not AdSS2; however, IMP inhibits both isozymes at concentrations of GTP below its K m value (Fig. 3). Scheme I represents IMP inhibition of AdSS2 and AdSS1. In the case of AdSS2, K ii Ͼ Ͼ K i , whereas for AdSS1 K ii must assume a value comparable with that of K i in order to observe IMP inhibition in the presence of saturating concentrations of GTP. In the context of a saturating concentration of L-aspartate and a rapid equilibrium Random Bi Bi kinetic mechanism for IMP and GTP (37,48), the expression for initial velocity associated with Scheme I is shown in where A and B are IMP and GTP concentrations, respectively; K a and K b are their corresponding Michaelis constants; K ia is the dissociation constant of IMP from the free enzyme; K i is the inhibition constant for IMP in the absence of GTP; and K ii is the inhibition constant for IMP in the presence of GTP. When only IMP concentrations vary at saturating concentrations of L-aspartate and GTP, Equation 1 simplifies to Equation 2, The  Fig. 4. High absorbances at 280 nm due to high concentrations of IMP and GTP thwarted the determination of K i and K ii for IMP inhibition of AdSS2; however, K i for AdSS2 must exceed 8 mM (data not shown).

DISCUSSION
E. coli and mouse synthetases exhibit a monomer-dimer equilibrium, in which GTP and IMP stabilize the dimer. Mouse isozymes in the absence of ligands, however, are more stable than the E. coli synthetase on the basis of thermal stability and analytical ultracentrifugation. The monomer-dimer equilibrium may be a property common to all adenylosuccinate synthetases. The dimer is the active form of the E. coli synthetase, and evidently each of its subunits independently achieves maximum velocity in the presence of saturating substrates (32). Indeed, an arginyl side chain (Arg 143 in the E. coli synthetase) critical to the recognition of IMP (34) comes to the active site from a symmetry-related subunit of the dimer and is present in all known sequences of synthetase. Hence, regulatory mechanisms that impair subunit dimerization are conceptually possible but have not been demonstrated for any synthetase in vivo. The increased stability of the mouse dimers relative to E. coli dimer could be exploited, however, in the design of drugs (antibiotics) that target the subunit interface.
N-terminal truncations of the mouse isozymes do not alter dimer stability or kinetics. Therefore, the functional differences exhibited by the two isozymes arise from their core sequences, which are ϳ75% identical. Nevertheless, the N-leader sequences of AdSS1 and AdSS2 diverge significantly, and yet each is conserved across mammalian species. The latter observation suggests possible differences in regulation, protein-pro-   (50 -51), but no hard evidence as yet implicates the N-leader sequence of AdSS1 in these interactions. Previous studies (8,13,27) have documented IMP inhibition of AdSS1, but have disagreed as to whether GTP antagonizes or reinforces the phenomenon (8,13). Nonetheless, IMP does bind to the GTP pocket of AdSS1 as evidenced by the crystal structure of an IMP complex (44). Furthermore, all residues of the GTP pocket are identical in AdSS1 and AdSS2. The EA 2 complex in Scheme I then may well represent IMP molecules bound to the IMP and GTP pockets within the same subunit of a dimer. High levels of GTP would relieve such inhibition, but AdSS1 remains sensitive to IMP even in the presence of saturating GTP. Hence, AdSS1 has an alternative mechanism that allows IMP to inhibit in the presence of GTP (EBA 2 complex of Scheme I). As will be discussed below, the alternative inhibitory site for IMP in AdSS1 may be the symmetry-related IMP pocket of the dimer.
AMP inhibits AdSS2 ϳ12-fold more strongly than AdSS1, and by a different kinetic mechanism (competitive versus noncompetitive with respect to IMP). Furthermore, in crystalline complexes of the E. coli synthetase (52), AdSS1 (45), and AdSS2 3 AMP binds only to the IMP pocket. AMP ligation of the IMP pocket accounts for competitive inhibition of AdSS2, but noncompetitive inhibition of AdSS1 suggests one of two alternative mechanisms. (i) AMP promotes the dissociation of an active AdSS1 dimer into inactive monomers. (ii) AMP inhibits one subunit of the AdSS1 dimer by binding to the IMP pocket of the other subunit. The former mechanism is unlikely, given the stabilizing effect of IMP on the dimer (34), but the latter is   plausible because of the proximity of IMP pockets in the dimer (Fig. 5). Helix 5 and the IMP loop (the latter binds the 5Јphosphoryl group of AMP) are conformationally dynamic elements in AdSS1 and could interact across the subunit interface. In contrast, helix 5 in AdSS2 is immobilized by hydrogen bond interactions and hence is less likely to transmit conformational changes between active sites. 3 IMP and AMP inhibition in AdSS1 then could stem from a common mechanism that involves the interaction of active sites of the dimer. AdSS2 on the other hand seems more like the E. coli synthetase in its properties of IMP and AMP inhibition. Kang et al. (32) have demonstrated functionally independent (non-interacting) subunits in the dimer of the E. coli synthetase.
Irrespective of whether subunit interactions in the dimer are the basis for noncompetitive AMP inhibition of AdSS1, an unresolved issue still remains. How does AdSS1 exclude AMP from its IMP pocket and still retain high affinity for IMP? The conformation of the pre-Switch loop of AdSS1 in its ligand-free state differs from that of other synthetases (28), including AdSS2. 3 The pre-Switch loop in AdSS1 blocks the 5Ј-phosphoryl pocket and holds the side chain of Asn 256 in an intramolecular hydrogen bond. Hence, IMP and AMP must sacrifice binding energy to overcome the antagonistic conformation of the pre-Switch loop in AdSS1. The 4 -5-fold increase in the K m of IMP and the 11-12-fold increase in K i of AMP are consistent with less favorable interactions of each nucleotide with the IMP pocket of AdSS1 relative to AdSS2. Unlike AMP, however, IMP can in principle recover free energy lost in its binding interaction by forming 6-phosphoryl-IMP; atoms N-7 and N-6 of AMP cannot both hydrogen bond with the side chain of Asn 256 , whereas atoms N-7 and O-6 of 6-phosphoryl-IMP can and do hydrogen-bond with the side chain of Asn 256 (44,45). As IMP and AMP levels in most tissues are ϳ60 and 200 M, respectively (53), fluctuations in the relative concentrations of AMP and IMP would influence the activity of AdSS2, whereas variations in the concentration of IMP alone would influence the activity of AdSS1.
The assignment of AdSS1 and AdSS2 to separate metabolic roles by Nakagawa and co-workers (7,12) rests largely on different susceptibilities to Fru-1,6-P 2 inhibition (7, 12); however, the reported K i values (0.6 and 1.6 mM for AdSS1 and AdSS2, respectively) are 30 -80-fold higher than physiological concentrations of Fru-1,6-P 2 . Nonetheless, all kinetic parameters reported by Nakagawa and co-workers (7,10,12,30) are much higher than those determined by other groups (2, 6). Stayton et al. (6), for instance, report noncompetitive Fru-1,6-P 2 inhibition of AdSS1 from rat (K i ϳ130 M) and rabbit (K i ϳ50Ϫ100 M), consistent with the findings here. Although an inhibitory site distinct from the active site accounts for the kinetic mechanism of inhibition, a preliminary crystallographic study reveals Fru-1,6-P 2 bound as an analogue of 6-phosphoryl-IMP. 3 Fru-1,6-P 2 could be a potent competitive inhibitor with respect to IMP at low concentrations of IMP (K i of 16 -20 M), but when IMP is at saturation, Fru-1,6-P 2 must bind elsewhere. The location of this alternative Fru-1,6-P 2 site is unknown. The potent mode of Fru-1,6-P 2 inhibition is nearly equal for AdSS1 and AdSS2. On the other hand, Fru-1,6-P 2 binds to the alternative site of AdSS1 with higher affinity than to that of AdSS2.
As the concentration of Fru-1,6-P 2 in most tissues is ϳ20 M, but can increase 3-fold in muscle during contraction (54,55), Fru-1,6-P 2 could be a dynamic regulator of AdSS1 in muscle. AdSS1 is the major if not sole form of the synthetase in muscle. The documented 60-and 3-fold increases in the in vivo concentrations of IMP and Fru-1,6-P 2 during vigorous exercise should severely limit AdSS1 activity, suggesting an asynchronous PNC. Declining ATP levels in exercising muscle activate AMP deaminase, whereas the concomitant increase in levels of IMP and Fru-1,6-P 2 inhibit AdSS1. Cessation of vigorous exercise leads to the restoration of ATP and Fru-1,6-P 2 levels, the inactivation of AMP deaminase, and the relief of Fru-1,6-P 2 inhibition of AdSS1. As the IMP concentration diminishes, the activity of AdSS1 actually accelerates until IMP is no longer saturating.
Continuous operation of the PNC in muscle is difficult to reconcile with the accumulation of IMP in vivo and the requirement for GTP to drive the AdSS1 reaction. The re-conversion of GDP to GTP is at best unfavorable in the face of diminished concentrations of ATP (24). Hence, the PNC is an unlikely source for the ammonia produced during muscle contraction and probably cannot provide fumarate to the Krebs cycle as originally proposed (5,13). Indeed, partial and complete deficiency in muscle AMP deaminase does not influence Krebs cycle anaplerosis, phosphocreatine hydrolysis, adenine nucleotide ratios, or exercise performance (21). Moreover, the fate of IMP accumulated during contraction varies with muscle type (56 -59); IMP primarily goes to AMP via the PNC, but some is released as inosine and hypoxanthine. These lost nucleotides must be re-synthesized de novo in order to restore adenine nucleotide pools (60). Hence, AdSS1 activity related to de novo purine biosynthesis and the PNC may be inseparable in muscle tissue.