Crystallographic Studies on Ascaris suum NAD-Malic Enzyme Bound to Reduced Cofactor and Identification of an Effector Site*

The crystal structure of the mitochondrial NAD-malic enzyme from Ascaris suum, in a quaternary complex with NADH, tartronate, and magnesium has been determined to 2.0-Å resolution. The structure closely resembles the previously determined structure of the same enzyme in binary complex with NAD. However, a significant difference is observed within the coenzyme-binding pocket of the active site with the nicotinamide ring of NADH molecule rotating by 198° over the C-1–N-1 bond into the active site without causing significant movement of the other catalytic residues. The implications of this conformational change in the nicotinamide ring to the catalytic mechanism are discussed. The structure also reveals a binding pocket for the divalent metal ion in the active site and a binding site for tartronate located in a highly positively charged environment within the subunit interface that is distinct from the active site. The tartronate binding site, presumably an allosteric site for the activator fumarate, shows striking similarities and differences with the activator site of the human NAD-malic enzyme that has been reported recently. Thus, the structure provides additional insights into the catalytic as well as the allosteric mechanisms of the enzyme.

Malic enzyme (ME) 1 is an oxidative decarboxylase that catalyzes the conversion of L-malate to pyruvate and carbon dioxide, using a divalent metal ion (Mg 2ϩ or Mn 2ϩ ) and NAD ϩ or NAD(P) ϩ as cofactors (1)(2)(3). The enzyme is found in prokaryotes and eukaryotes and participates in diverse metabolic pathways such as photosynthesis, lipogenesis, and energy metabolism. Mitochondrial and cytosolic isoforms of the enzyme have been identified (1,4). A sequence comparison of malic enzymes from different sources shows significant homology within the family but no homology to other proteins with the exception of the dinucleotide binding signature motif (5). Be-cause of its functional importance, the enzyme has been isolated and characterized from several sources (3,6). The mitochondrial NAD-malic enzyme (m-NAD⅐ME) from the parasitic nematode, Ascaris suum, plays a pivotal role in carbohydrate metabolism in parasitic worms (7). In the anaerobic metabolism of A. suum, malate, an intermediate in the worm's glycolytic pathway, is transported into the mitochondria where it undergoes a dismutation and is converted to pyruvate and NADH via the malic enzyme reaction and to fumarate via the fumarase reaction. Fumarate is then converted to short, branched-chain fatty acids via succinate mediated by the NADH produced in the malic enzyme reaction. The succinate dehydrogenase reaction is also involved in a site 1 oxidative phosphorylation, the main source of mitochondrial ATP (8). Since malic enzyme generates reducing equivalents (NADH) for the conversion of fumarate to succinate, it is not surprising that fumarate regulates its own utilization by activating the malic enzyme reaction (9,10).
The ascarid malic enzyme has been extensively studied in our laboratories from the standpoint of its kinetic, regulatory, chemical mechanisms and also physicochemical properties (3,6). Recently, the crystal structure of the enzyme complexed with NAD has been determined to 2.3-Å resolution and compared with the structure of the human mitochondrial enzyme, also complexed with NAD (11)(12)(13). Whereas the two enzymes have similar tertiary and quaternary structures and exhibit similarities in domain structure, there are significant differences between the structures of the two enzymes. The ascarid enzyme has 30 additional residues at its amino terminus relative to the human enzyme, and this extra sequence leads to increased interactions at the tetramer interface. Although the active site residues of the ascarid enzyme are similar to those of the human enzyme, residues interacting with NAD differ between the two (11). The two enzymes also differ in the organization of the tetramer. The ascarid malic enzyme tetramer is more flattened compared with the human enzyme as a consequence of the difference in the positioning of the C domain in the two structures relative to the A and B domains and also a difference in the positioning of the two dimers within each tetramer. The most notable difference between the two enzymes is that the human enzyme has a second NAD-binding site (exo site), distinct from the active site, at the tetramer interface, which was originally thought to be an ATP-inhibitory site but later on was suggested to be of unknown function (13). The ascarid enzyme does not have this exo site for NAD (11) and is not inhibited by ATP at physiological concentrations. 2 In the previous structure of the ascarid malic enzyme (11), the active site was located by the bound NAD and by comparison with the closed and open forms of the human enzyme (12,14). The NAD cofactor binds with the nicotinamide ring in the anti conformation with the re face directed toward the solvent. In the bound conformation, the nicotinamide ring closely approaches the pyrophosphate moiety, probably resulting in an ionic interaction between N-1 of the nicotinamide ring and one of the oxygens of the pyrophosphate backbone. In order to determine the specific effects of reduction of the nicotinamide ring on the coenzyme binding in the active site and the overall catalytic mechanism, a structure of the Ascaris malic enzyme has been solved with NADH bound at the active site. This is the first report of the structure of a malic enzyme complexed with the reduced coenzyme and reveals a huge movement of the nicotinamide ring closer to the active site residues as a consequence of reduction. Although tartronate was included in our previous crystallization trials, we were not able to clearly identify bound tartronate in the electron density map of ME⅐NAD crystals. However, the higher resolution data obtained in the present study helped to locate tartronate bound to an allosteric site that is shown to bind fumarate in the human enzyme.

MATERIALS AND METHODS
Crystallization and X-ray Data Collection-Malic enzyme was purified as described previously (11). Crystals of the ME⅐NADH complex were obtained by the hanging drop-vapor diffusion method from a solution containing 100 mM Tris⅐SO 4 , pH 7.3, 100 mM sodium acetate, 15% polyethylene glycol 4000, 5 mM NADH, 10 mM tartronate, 20 mM MgSO 4 , 10 mM 2-mercaptoethanol, and 0.02% sodium azide. The crystals are typically 0.3 ϫ 0.2 ϫ 0.2 mm and are isomorphous with the crystals of ME⅐NAD complex and belong to the space group P3 1 21 with an asymmetric unit of a ϭ b ϭ 130.95 Å, c ϭ 149.13 Å, containing two monomers. For cryoprotection, crystals were first soaked in a stabilizing solution containing 25% (w/v) polyethylene glycol 4000, 15 mM NAD, and 10 mM 2-mercaptoethanol in 100 mM Tris-SO 4 , pH 7.5, for 2 h and then soaked for 24 h in a cryoprotectant solution containing the above components plus 20% ethylene glycol (v/v). Crystals were flash frozen on rayon loops by plunging them into liquid propane and stored in liquid nitrogen prior to data collection. X-ray diffraction data to 2.0 Å were collected at the Cornell synchrotron source (CHESS) F1 line ( ϭ 0.947 Å) equipped with a CCD detector. Data were processed using the DENZO/SCALEPACK programming package (15).
Model Building and Refinement-The starting model for refinement was a dimer of the Ascaris ME⅐NAD structure solved to 2.3 Å in our laboratory (Protein Data Bank code 1LLQ) (11). NAD molecules were omitted from the model for the first round of refinement. This initial model was refined with the reflection data for the ME⅐NADH crystals by simulated annealing with no symmetry constraints using the CNS (version 1.1) programming package (16). Initial 2F o Ϫ F c and F o Ϫ F c maps indicated no major differences between the two structures with the exception of the nicotinamide portion of NAD molecule and in the carboxyl-terminal region. The model was manually rebuilt using the interactive graphics model building program O (17). Since differences between the two crystallographically independent subunits (␣ a and ␣ b ) were observed, as in the structure of the enzyme-NAD complex, independent models were used for each monomer after the first two rounds of refinement. A bulk solvent correction was applied to all refinements, and the free-R factor method was used to monitor the refinement (16). Sine electron density for the NADH cofactor and tartronate molecule was clearly observed in the initial model, NADH and tartronate were added during the first round of model building. The asymmetric unit for the current model contains residues 2-603 (␣ a ) and residues 2-593 (␣ b ) of the two crystallographically independent subunits, two NADH molecules and two Mg 2ϩ bound at the active sites, two tartronate molecules bound in the dimer interface, and 410 water molecules. The subunits superimpose with a root mean square (r.m.s.) deviation of 0.34 Å between 593 corresponding ␣ carbons. Analysis of stereochemistry using the program PROCHECK (18) showed Ͼ90% of the residues in the most favored region of the Ramachandran plot. Final refinement statistics calculated using all data are given in Table I. Comparison between subunits and domains and with the human ME structures was carried out using O (17).

Structure of the ME⅐NADH Complex-
The tertiary and quaternary structure of the Ascaris NAD-malic enzyme in complex with the cofactor NADH is essentially identical to that of the binary complex with NAD (11). Overall, the structure is a tetramer, which is organized as a dimer of dimers ( Fig. 1) (the r.m.s. deviation between the ␣ carbon backbones of the corresponding monomers, dimers, and tetramers of the two complexes are 0.25, 0.32, and 0.34 Å, respectively). Two types of subunit interfaces have been defined: a dimer interface between two subunits composing two monomers and a tetramer interface formed between two dimers (Fig. 1). The monomer is organized into four domains (A, B, C, and D) as previously described for Ascaris ME (11) and other malic enzymes (14). The active site is a cleft, formed between the C and B domains, and one NADH cofactor is bound to each of the four active sites in the tetramer. The B and C domains in the ME⅐NADH complex superimpose on the corresponding domains in the ME⅐NAD complex with r.m.s. deviations of 0.18 and 0.30 Å, respectively. The active site, like that of the ME⅐NAD complex (11), is in an open conformation (i.e. the residues implicated in substrate binding and catalysis (Arg 181 , Lys 199 , Asp 295 , Asp 272 , and Asn 434 ) are poorly positioned for their prospective roles.
In the NAD-bound structure, electron density for three residues, Ala 601 , Ser 602 , and Met 603 at the C terminus in the D-domain (11) was either not present or weak. In the present structure, the three residues could be unambiguously modeled but do not have a significant impact on the overall structure.
The tetramer in the present structure has four NADH molecules (one per monomer) bound in the active site and four tartronate molecules (one per monomer) bound within the dimer interface (Fig. 1).
The NADH-binding Site-The conformation of the NADH cofactor and its interactions with the enzyme are similar to those of the ME⅐NAD complex with the exception of a dramatic difference in the conformation and interactions of the nicotinamide ring (Fig. 2, A and B). The nicotinamide ring in the NADH complex rotates ϩ198°about the N-glycosylic bond relative to its orientation in the ME⅐NAD complex, with the si face exposed to the solvent-accessible region of the active site ( Fig.  2A). Binding interactions observed in the ME⅐NAD complex between the nicotinamide ring amide group and residues Gly 477 and Asn 479 are broken as a consequence of this rotation, and new hydrogen bonds between the carboxamide side chain and the enzyme are formed between active site residues Asp 295 and Arg 181 (Table II and Fig. 2B). A water molecule (wat139) occupies the site vacated by the nicotinamide amide group and forms hydrogen bonds with Gly 477 and Asn 479 .
In addition to NADH, Mg 2ϩ is also bound to the active site (Fig. 2B). The metal ion is coordinated to the carboxylate oxy- for a randomly selected 5% of the data excluded from the refinement.
FIG. 1. Ribbon diagram of the structure of Ascaris mitochondrial ME in complex with NADH, Mg 2؉ , and tartronate. A, the tetramer viewed down one 2-fold axis (indicated by the black oval) with the tetramer and dimer interfaces indicated by the arrows. The four subunits are colored blue, yellow, green, and tan. The four NADH and four tartronate ligands are shown as red ball-and-stick models, and their binding sites are indicated for one dimer. B, the Ascaris ME dimer viewed down the 2-fold axis corresponding to the dimer interface showing the locations of the two tartronate binding sites more clearly. The NH 2 and COOH termini of each monomer are indicated. For one subunit, the two helices (␣ A3 and ␣ A4 ) that contribute most of the residues involved in tartronate binding are colored purple and labeled. This figure was generated using MOLSCRIPT (31).  (12).
The Tartronate Binding Site-Tartronate, a competitive inhibitor of the ascarid malic enzyme with respect to the substrate malate, was included in all Ascaris ME crystallizations in order to characterize the malate-binding site. Although electron density is observed in the active site corresponding to the tartronate binding site as identified in the human ME⅐NAD (12), the density in the ascarid malic enzyme structure is weak and does not conclusively demonstrate that tartronate is bound to the active site (Fig. 3A). This may be due to partial occupancy of the malate-binding site. However, density for tartronate is observed at a different site, within the dimer interface near a noncrystallographic 2-fold axis (Fig. 3B). Here the tartronate molecule is bound to a highly positively charged pocket (Figs. 4, A and B) and is tightly anchored by strong hydrogen bond/ionic interactions with residues from two adjacent subunits. The majority of the residues contributing to the binding site are in a cleft formed between the ␣ A3 (residues 76 -87) and ␣ A4 (residues 93-104) helices of a subunit (Figs. 3B and 5A). Oxygen atoms O-1 and O-2 from one tartronate carboxylate group are within hydrogen-bonding distance (Table III) to the N-⑀ atom of the side chain amide group of Gln 78 (Gln 64 in human ME) and a salt bridge with the guanidinium group of Arg 105 (Arg 91 in human ME), respectively (Fig. 5, A and B). Oxygen atom O-4 of the other carboxylate forms a salt bridge with the guanidinium group of Arg 105 (Arg 91 in human ME), whereas its O-5 atom forms a salt bridge with the NH 2 of the guanidinium group of Arg 81 (Arg 67 in human ME). The hydroxyl oxygen atom (O-3) of tartronate is within hydrogenbonding distance to the N-⑀ of Arg 81 . The remainder of the tartronate binding site is formed by residues 140 -142 from the adjacent subunit. These residues are part of a coil, which links the A and B domains of this subunit. The hydroxyl group of Tyr 141 is within hydrogen bonding distance to a carboxylate oxygen of tartronate. Binding of tartronate to this site contributes to the subunit interaction across the dimer interface. It is interesting that the dimer interface in the A. suum malic enzyme contains several charged residues and provides a continuous, positively charged pocket (Fig. 4, A and B) for the binding of negatively charged molecules such as tartronate and malate. The tartronate site also contains a water molecule (wat74; Fig. 3B), which is bound to a tartronate carboxylate oxygen and the amide NH of Met 64 . Its location corresponds to a water molecule bound to the fumarate-binding site in the human ME⅐NAD.
Identification of bound tartronate in the Ascaris ME⅐NADH complex also definitively indicated that tartronate is bound at the same site in the Ascaris ME⅐NAD complex. In the ME⅐NAD complex, although density was observed at the tartronate binding site, the resolution of the data did not permit identification of this density as tartronate. Comparison of the ME⅐NADH and  TTN) and active site residues shown as ball-and stick models. B, ribbon diagram of the tartronate-binding site within the dimer interface of Ascaris ME with tartronate and key residues indicated as ball-and-stick models and a water molecule bound to tartronate shown as a sphere. Electron density (2F o Ϫ F c , contoured at 1) for tartronate is also shown. Generated using MOLSCRIPT (31), BOBSCRIPT (32), and RASTER-3D (33).
ME⅐NAD structures revealed that the unidentified density in ME⅐NAD was also tartronate and that the binding interactions are identical to those observed in ME⅐NADH.

DISCUSSION
Overall Structure-The present structure, the first of a malic enzyme with reduced dinucleotide bound to the active site, provides new insights into substrate binding modes and helps resolve some of the questions associated with the catalytic mechanism of the enzyme. In addition, the structure reveals the identity of the allosteric site, that binds the activator fumarate. The backbone conformation and the quaternary structure observed in the ME⅐NADH complex are very similar to those of the ME⅐NAD complex (the r.m.s. deviation between the main chain atoms of the two structures is 0.25 Å, monomer; 0.32 Å, dimer; and 0.34 Å, tetramer). There are no noticeable differences in the dimer and tetramer interactions. As stated above, the final three residues at the C terminus in the Ddomain (Ala 601 , Ser 602 , and Met 603 ) can be visualized but have no significant impact on the overall structure of the ME⅐NADH complex.
In earlier studies on the structure of the ME⅐NAD complex, it was speculated that the residues implicated in metal ion binding (Glu 271 , Glu 272 , and Asp 295 ) were not properly positioned for binding the metal ion. In the present structure, the binding site for magnesium is clearly identified (Fig. 2B). Whereas the role of Asp 295 in metal ion binding was further confirmed by earlier site-directed mutagenesis studies (19), a role for Glu 271 and Asp 272 was not postulated. These two residues are part of the malic enzyme sequence QFEDFA (positions 269 -274), which agrees well with the consensus metal ion binding sequence XXDDXX, where X is an uncharged or hydrophobic residue (20). All of the residues suggested as metal ion ligands are homologous to residues in the human enzyme (12). The fact that Mg 2ϩ was not observed bound at the corresponding site in the ME⅐NAD complex indicates that the metal ion affinity may be sensitive to the oxidation state of the cofactor or that the low resolution data for ME⅐NAD crystals does not reveal the metal ion-binding site. The binding of the metal ion is consistent with the required ordered addition of metal ion prior to malate. FIG. 4. A, electrostatic potential surface of a monomer of Ascaris mME as viewed in roughly the same orientation as in Fig. 3B. B, electrostatic potential surface of two monomers showing the dimer interface along the vertical axis. Two tartronates are indicated as red CPK models. This figure was generated using GRASP (28). Although the active sites of the ME⅐NADH and ME⅐NAD complexes differ in their cofactor conformations and the presence of Mg 2ϩ bound in ME⅐NADH complex, the active site main chain and side chain conformations are very similar to one another.
Conformational Change in the Coenzyme-A comparison of the coenzyme binding pockets in the present structure and the previously reported ME⅐NAD structure shows interesting changes in the conformation of the nicotinamide ring as a consequence of reduction (Fig. 2, A and B). In the ME⅐NAD complex, as NAD is bound, the re face of the nicotinamide ring is directed toward the solvent such that its carboxamide side chain is directed away from the catalytic pocket into the protein structure. In the present structure, however, the reduced cofactor has its nicotinamide ring moved into the active site through a rotation about the N-glycosylic bond by ϩ198°, and its si face is now directed toward the solvent. The carboxamide side chain, which was originally hydrogen-bonded to a backbone carbonyl and the amide side chain of Asn 479 , now interacts with Arg 181 , which in turn interacts with the ␣-carboxylate of L-malate, and Asp 295 , the putative general base, which in turn interacts with the 2-hydroxyl of malate in the Michaelis complex (Table II). The proS proton of the reduced nicotinamide ring is now directed toward the active site. It is interesting to note that whereas the nicotinamide ring rotates by ϩ198°into the active site, other catalytic residues are unaffected. As shown in Fig. 2A, the density around the nicotinamide ring in a difference map calculated from ME⅐NADH complex with and without NADH is very strong, and there are strong hydrogen-bonding contacts between the carboxamide group of the nicotinamide ring with Arg 181 and Asp 295 , suggesting that the nicotinamide ring is indeed locked in this conformation and that the observed conformational change is not simply due to the conformational flexibility allowed by the Rossman fold as in the case of lactate dehydrogenase (21,22). The mobility of the nicotinamide ring reported in this study is also observed in other enzymes, and it has been proposed to play a crucial role in the catalytic mechanism of at least two other enzymes, aldehyde dehydrogenase (23) and 6-phosphogluconate dehydrogenase (24).
Implications to Catalytic Mechanism-A three-step acidbase catalytic mechanism is proposed for ascarid malic enzyme based on pH and isotope partitioning studies (25)(26)(27). The assignment of the general acid, general base, and binding groups shown in Scheme 1 is based on the x-ray structure of the ME⅐NAD complex (11) and site-directed mutagenesis studies (19). Generally, L-malate is converted to oxaloacetate facilitated by Asp 295 , which acts as a general base to accept a proton from the 2-hydroxyl of L-malate (19). The hydride at C-2 of malate is transferred to the 4-position of the nicotinamide of NAD, with the positively charged pyridinium nitrogen providing the driving force for reduction. The guanidinium of Arg 181 helps to orient malate for reaction by forming an ion pair/ hydrogen bond with the ␣-carboxyl of malate, and Lys 199 acts in a similar role by interacting with the ␤-carboxyl. The metal ion (Mg 2ϩ ) aids in the orientation of malate for catalysis and facilitates proton transfer to Asp 295 by binding to the ␣-hydroxyl of malate. In the second step, oxaloacetate is decarboxylated to enolpyruvate with the metal ion acting as a Lewis acid and protonated Asp 295 acting as a general acid to protonate the enolate oxygen. Finally, enolpyruvate is tautomerized to pyruvate, with Asp 295 acting as a general base and the ⑀-amino group of Lys 199 acting as a general acid to protonate C-3 of enolpyruvate (29). Thus, the catalytic mechanism requires a protonated form of the general acid (Lys 199 ) and an unprotonated form of the general base (Asp 295 ). The optimum protonation state for Asp 295 (unprotonated) and Lys 199 (protonated) is observed in the V/K malate pH-rate profile (26), although the catalytic role of Lys 199 is not realized until CO 2 is released. The observation of the Lys 199 pK in the V/K profile indicates the importance of the protonated form of Lys 199 in binding malate. Whereas the existence of the protonated form of Lys 199 is important for optimum binding of the ␤-carboxylate of L-malate, it poses a problem in the decarboxylation step, where the ionic/hydrogen-bonding interaction between Lys 199 and the carboxylate would be anticatalytic. Thus, during the subsequent hydride transfer step, the interaction between the carboxylate and Lys 199 must be eliminated.
The crystallographically observed conformational change in the nicotinamide ring may provide an explanation as to how the hydrogen bond between Lys 199 and the ␤-carboxyl of malate may be eliminated. A mechanism incorporating the structural data obtained in the present study is shown in Scheme 1.
Malate is bound such that its ␣-carboxylate is hydrogen-bonded to the guanidinium of Arg 181 and coordinated to Mg 2ϩ . The ␤-carboxylate of malate is proposed to hydrogen-bond to Lys 199 , FIG. 5. Ribbon diagrams for the comparison between the tartronate and fumarate binding sites of Ascaris and human mME. A, tartronate binding site in Ascaris ME; B, fumarate binding site in human mME. The ␣ a and ␣ b subunits in both of the enzymes are colored blue and tan, respectively. Tartronate (TTN) in Ascaris ME and fumarate in human ME (FM) are shown as red and cyan CPK models. Key residues involved in binding ligands are shown as ball-and-stick models. The NH 2 and COOH termini of each monomer are indicated. This figure was generated using MOLSCRIPT (31) and RASTER-3D (33). and the ␤-hydroxyl of malate is hydrogen-bonded to Asp 295 and coordinated to Mg 2ϩ . The nicotinamide ring of NAD ϩ is bound with its carboxamide side chain hydrogen-bonded to a backbone NH and carbonyl and to the amide side chain of Asn 479 . The distance between the pyridinium nitrogen of NAD ϩ and the pyrophosphate moiety of the cofactor is about 5.7 Å. Hydride transfer from C-2 of malate to C-4 of the nicotinamide ring occurs and is partly driven by the positively charged pyridinium ring and general base catalysis by Asp 295 . The reduction of the nicotinamide ring results in a rotation by 198°about the N-glycosidic bond, placing the carboxamide side chain within hydrogen-bonding distance to Asp 295 and Arg 181 . The rotation is thought to result, at least partly, from loss of the ionic interaction between the pyridinium nitrogen of NAD ϩ and the pyrophosphate moiety of the cofactor. The movement of the cofactor is proposed to be reversible but with the equilibrium position favoring that shown in the structure (Fig. 2). The movement of the nicotinamide ring along with the sp 3 to sp 2 hybridization change at C-2 that occurs as oxalacetate is formed is proposed to shift the position of the bound oxalacetate such that 1) the ␤-carboxylate moves away from Lys 199 and 2) the C-3-C-4 bond to the ␤-carboxylate is now orthogonal to the C-2-C-3 plane of oxalacetate, favoring decarboxylation. Decarboxylation then occurs facilitated by the Lewis acidity of the Mg 2ϩ and with Asp 295 acting as a general acid. Finally, tautomerization occurs aided by general base-general acid catalysis via Asp 295 -Lys 199 pair. The products pyruvate and NADH are then released. Final confirmation of the above hypothesis should come from the structure of a ternary complex of the enzyme with NADH and an analogue of malate or oxaloacetate.
Of interest is the change in mechanism observed for the malic enzyme, from "stepwise" to "concerted," when the dinu-cleotide substrate is changed to more oxidizing dinucleotides acetylpyridine adenine dinucleotide, thio-NAD ϩ , or pyridine aldehyde adenine dinucleotide (30). The change in mechanism probably reflects a difference in the binding of the pyridine ring of the alternative substrates, which results in a difference in the conformation of the bound malate. A proposed link between the conformation of the bound nicotinamide ring and malate/ oxalacetate was made above. It is not difficult to imagine a difference in the bound conformation of the pyridine ring, given the interaction of the carboxamide side chain of NAD ϩ with Asn 479 . In all cases, the side chains of the alternative dinucleotide substrates differ with respect to the hydrogen-bonding capability, size, and hydrophobicity in comparison with the carboxamide side chain of NAD ϩ .
In order to examine whether the active site of human enzyme could accommodate such a conformational change in the nicotinamide ring, NADH was modeled into the active site of the open structure of the human enzyme (14) such that the nicotinamide ring was rotated by 198°relative to its position in NAD molecule. The new conformation was indeed feasible without any steric hindrance, and the contact distances between the nicotinamide ring and the corresponding enzyme residues were similar to those observed in ascarid enzyme.
The Tartronate Binding Site and Regulatory Mechanism-Tartronate, a dicarboxylic analogue of malate and fumarate, is tightly bound in an allosteric site via hydrogen-bonding interactions with the side chains of two arginines (Arg 105 and Arg 81 ) and a glutamine (Gln 78 ). All of the residues shown are homologous to those in the fumarate allosteric site depicted in the recent structure of the human enzyme-ATP complex (Fig. 5, A  and B) (13). The human mitochondrial ME is allosterically activated by fumarate. Although many of the residues that SCHEME 1. Proposed mechanism for the NAD-malic enzyme. The mechanism is described under ''Discussion.'' Arg 181 is considered to lie above the plane that contains malate, with the nicotinamide ring in the same plane as malate but with its carboxamide pointing back and into the plane of the paper. Rotation causes the carboxamide side chain to be pointing up out of the paper plane.
bind fumarate in human ME are homologous or functionally similar to those involved in tartronate binding in Ascaris ME, the binding modes of these ligands differ significantly between the two enzymes. A multiple sequence alignment for malic enzymes from selected species (Fig. 6) shows conservation of all of the above residues. It is interesting to note that fumarate activation has not been reported in all of the species that have the conserved residues under consideration.
The Ascaris malic enzyme is also allosterically activated by fumarate with an activation constant of 40 M (9, 10). Activation is expressed as a decrease in the off-rate for bound malate (10). Although the activation by fumarate under conditions of saturating NAD ϩ and Mg 2ϩ is only 2-fold (9), with reactants maintained at estimated physiologic concentrations, the activation is 15-fold. 3 Thus, the allosteric site where tartronate is bound is most likely an activator site for fumarate. This conclusion is further supported by the observation that the R105A mutant of the Ascaris malic enzyme, which has a V/E t value identical to that of the wild type enzyme, is no longer activated by fumarate, consistent with tartronate being bound to the fumarate activator site. 3 The distance between the active site and the tartronate allosteric site is ϳ30 Å. However, there are structural relationships between the sites, which could allow for transmission of the allosteric signal. Within a subunit, Arg 105 participates in binding tartronate. The backbone oxygen of the adjacent residue, Asp 104 , is hydrogen-bonded to the side chain of Lys 143 , which is part of a coil sequence (residues 140 -142) that connects the A and B domains of a subunit (Fig. 5A). This coil is also directly involved in forming the second tartronate binding site within the dimer interface via Tyr 141 (Fig. 5A). The aminoterminal end of this coil is linked to the ␣ A6 helix (residues 37-144), which contributes active site residue Tyr 126 . Thus, the tartronate-binding site is linked to the coil connecting the A and B domains of a monomer (residues 140 -143). It is interesting to note that the conformation of this coil differs significantly from that of the corresponding coil in human ME (residues 126 -129) despite almost identical sequences (Figs. 5B and 6). In particular, there is no interaction corresponding to the Asp 104 -Lys 143 pair, and the positions of the homologous residues Arg 142 (Ascaris) and Arg 128 (human) differ. In addition, in place of Tyr 141 in the Ascaris enzyme, the human enzyme has a phenylalanine (Phe 127 ).
Overall, the evidence presented in this study identifies an allosteric site and provides new insights into the mechanism of allosteric activation of the A. suum malic enzyme. However, the structure of the Ascaris malic enzyme with fumarate bound to this allosteric site will further confirm the identity of this site and enhance our understanding of the allosteric mechanism.
FIG. 6. Sequence alignment of residues in the Ascaris mME tartronate binding site with the homologous regions in human (and other) MEs. Key residues involved in tartronate binding (Ascaris) and fumarate binding (human) are indicated.