Location of the Coenzyme Binding Site in the Porcine Mitochondrial NADP-dependent Isocitrate Dehydrogenase*

The structure of crystalline porcine mitochondrial NADP-dependent isocitrate dehydrogenase (IDH) has been determined in complex with Mn2+-isocitrate. Based on structural alignment between this porcine enzyme and seven determined crystal structures of complexes of NADP with bacterial IDHs, Arg83, Thr311, and Asn328 were chosen as targets for site-directed mutagenesis of porcine IDH. The circular dichroism spectra of purified wild-type and mutant enzymes are similar. The mutant enzymes exhibit little change in Km for isocitrate or Mn2+, showing that these residues are not involved in substrate binding. In contrast, the Arg83 mutants, Asn328 mutants, and T311A exhibit 3-20-fold increase in the \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(K_{m}^{NADP}\) \end{document}. We propose that Arg83 enhances NADP affinity by hydrogen bonding with the 3′-OH of the nicotinamide ribose, whereas Asn328 hydrogen bonds with N1 of adenine. The pH dependence of Vmax for Arg83 and Asn328 mutants is similar to that of wild-type enzyme, but for all the Thr311 mutants, pKes is increased from 5.2 in the wild type to ∼6.0. We have previously attributed the pH dependence of Vmax to the deprotonation of the metal-bound hydroxyl of isocitrate in the enzyme-substrate complex, prior to the transfer of a hydride from isocitrate to NADP's nicotinamide moiety. Thr311 interacts with the nicotinamide ribose and is the closest of the target amino acids to the nicotinamide ring. Distortion of the nicotinamide by Thr311 mutation will likely be transmitted to Mn2+-isocitrate resulting in an altered pKes. Because porcine and human mitochondrial NADP-IDH have 95% sequence identity, these results should be applicable to the human enzyme.

The porcine mitochondrial NADP-dependent isocitrate dehydrogenase (EC 1.1.1.42) (NADP-IDH) 1 is a divalent metal-dependent enzyme that catalyzes the oxidative decarboxylation of isocitrate to ␣-ketoglutarate using NADP as a cofactor. The enzyme is a homodimer (1,2), with a molecular mass of 46.6 kDa and 413 amino acids per subunit (3), which binds 1 mol of NADPH or NADP ϩ per mol of enzyme subunit (4). NADP-dependent isocitrate dehydrogenase is not only involved in the citric acid cycle, but also has an important role in preventing oxidative damage in mitochondria through NADPH regeneration (5,6).
We have described the crystal structure of porcine NADP-IDH complexed with Mn 2ϩ and isocitrate (7). The enzymebound Mn 2ϩ is hexacoordinate, with Asp 252 , Asp 275 , two water molecules, and two oxygens of isocitrate acting as ligands (7)(8)(9). Mutagenesis experiments have supported the importance of Asp 252 and Asp 275 in the Mn 2ϩ -isocitrate site and have implicated Asp 279 (8,9), Arg 101 , Arg 110 , and Arg 133 (10), Lys 212 , and Tyr 140 (11), and Ser 95 , Asn 97 , and Thr 78 (12) as participants in the substrate site. Our recent findings suggested that the pH dependence of the enzyme's V max is caused by the deprotonation of the metal-bound hydroxyl of isocitrate in the enzymesubstrate complex prior to the transfer of a hydride to NADP (9).
Previous studies from our laboratory suggest that Arg 314 , His 315 , and Tyr 316 interact with the 2Ј-phosphate of NADP and are determinants of the coenzyme specificity of isocitrate dehydrogenase (13,14). His 309 also contributes to the coenzyme binding site (13). Although the porcine mitochondrial NADPspecific isocitrate dehydrogenase has not been crystallized in complex with coenzyme, NADP has been modeled within the crystal structure of this mammalian enzyme using a structural alignment between the porcine enzyme-Mn 2ϩ -isocitrate complex and seven determined crystal structures of bacterial IDH-NADP complexes (7). Fig. 1 shows a view of the model of the NADP-Mn 2ϩ -isocitrate complex of pig mitochondrial NADP-IDH. To evaluate the location of NADP in this model, we have selected as targets for site-directed mutagenesis three amino acids distributed throughout the predicted coenzyme site: Arg 83 , Thr 311 , and Asn 328 . Arg 83 and Thr 311 are positioned close to the nicotinamide ribose and 5Ј-phosphate, whereas Asn 328 is near the adenine ring ( Fig. 1). A structure-based alignment of two regions of the amino acid sequences of NADP-isocitrate dehydrogenases from six species is shown in Fig. 2. Arg 83 is conserved among mammals but not in lower organisms, while Thr 311 and Asn 328 are conserved in all species. The results of this report demonstrate that all three of these amino acids contribute to and delineate the extent of the coenzyme binding site of the mitochondrial porcine NADP-dependent isocitrate dehydrogenase. expressed as a maltose-binding fusion protein with a thrombin cleavage site located between the maltose-binding protein and the porcine isocitrate dehydrogenase (15). Site-directed mutagenesis of pMALcIDP1 was performed using the QuikChange XL Kit to produce mutant DNA (11,12). The oligonucleotides (forward and reverse primers) used to generate the mutant enzymes by the QuikChange method are listed in Table I. The underlined codons are those nucleotides coding for the mutations. R83K, R83Q, T311A, T311N, T311S, N328D, and N328S mutant enzymes were generated for the present studies. All mutations were confirmed by the BigDye terminator cycle sequencing method performed at the Agricultural DNA Sequencing Facility, University of Delaware.

Materials
Expression and Purification of Wild-type and Mutant Enzymes-The plasmids were expressed in E. coli strain TB1, and the pure fusion proteins from wild-type and mutant enzymes were isolated as maltosebinding proteins using an amylose column as previously described (10,15). The fusion proteins were cleaved by human thrombin (10,15) and purified on a DE-52 ion exchange column (14). The purity of the proteins was assessed by SDS-polyacrylamide gel electrophoresis (10,(13)(14)(15) and by N-terminal amino acid sequencing on an Applied Biosystem gas-phase sequencer (Model Procise) equipped with an on-line microgradient Delivery System (Model 140C) and a computer (Model 610 Macintosh). The protein concentration for the purified isocitrate dehydrogenase was determined from E 280 nm 1% ϭ 10.8 (16). A subunit molecular mass of 46.6 kDa (15) was used to calculate the concentration of enzyme subunits.

Circular Dichroism Spectra of the Wild-type and Mutant
Enzymes-CD spectra were measured at room temperature using a Jasco model J-710 spectropolarimeter (Jasco, Inc., Easton, MD). Measurements of ellipticity as a function of wavelength were made as described previously (8,10), using 413 as the number of amino acids per subunit of NADP-dependent isocitrate dehydrogenase. The concentrations of wild-type and mutant enzymes were determined using the A 280 nm and the Bio-Rad dye-binding assay based on the method of Bradford (17).
For K m determinations, the concentration of NADP, isocitrate, or Mn 2ϩ was varied, whereas the other substrates were maintained at saturating concentrations. The K m values for NADP were determined by varying the concentrations of NADP while maintaining the isocitrate and Mn 2ϩ at the saturating concentrations of 4 mM and 2 mM, respectively. The assay solution for measuring K m at pH 7.4 contained 30 mM triethanolamine hydrochloride, and the assay solution for measuring K m at pH 5.5 contained 30 mM sodium acetate buffer. The K m values (with standard errors) were determined from direct plots of velocity versus substrate concentration using Sigma Plot software. The pH dependence of V max for the reaction catalyzed by wild-type and mutant isocitrate dehydrogenases were determined over the range pH 5-8, using the buffers previously described (9). The reaction rates were determined using 4 mM isocitrate, 0.1-10 mM NADP, and 2-6 mM Mn 2ϩ , as indicated for each enzyme.

Expression and Purification of Wild-type and Mutant Enzymes-
The pig heart NADP-dependent isocitrate dehydrogenase mutant enzymes were generated using expression vector pMALcIDP1, and the QuikChange-XL kit method with oligonucleotides encoding substituted amino acids. At position 83, the positively charged arginine was substituted with glutamine (to eliminate the charge but maintain the hydrogen bonding potential) or lysine (to maintain the positive charge, but change the shape). At position 311, threonine was replaced by alanine (which is only slightly smaller, but lacks the hydrogen bonding potential), by serine (which is also slightly smaller, but retains the possibility of hydrogen bonding) or by asparagine (which is much larger, yet maintains the hydrogen bonding capability). At position 328, asparagine was changed to the negatively charged aspartate or to the neutral, but smaller, serine. The wild-type and all mutant enzymes expressed in E. coli as maltose fusion proteins, were cleaved by thrombin and purified, as described under "Experimental Procedures." The wild-type and mutant enzymes each exhibit a single protein band on polyacrylamide gels containing SDS and each one has the same subunit molecular mass (about 46.6 kDa); representative samples are shown in Fig. 3. The N-terminal amino acid sequences of these purified porcine isocitrate dehydrogenases also demonstrate that these recombinant porcine enzymes are not contaminated with the E. coli isocitrate dehydrogenase, since the N-terminal sequences of the E. coli and porcine enzymes differ in 9 of the first 10 amino acids.
Circular Dichroism Spectra of Wild-type and Mutant Enzymes-CD spectra of wild-type and mutant isocitrate dehydro-genases were measured to evaluate whether there are changes in secondary structure in these mutant enzymes. The CD spectra of R83Q, R83K, T311A, T311S, T311N, N328S, and N328D mutant enzymes are very similar to that of the wild-type enzyme (data not shown). These results suggest that the mutations do not cause any appreciable conformational changes.
Kinetic Characteristics of Wild-type and Mutant Enzymes-The kinetic parameters of wild-type and mutant porcine NADPdependent isocitrate dehydrogenases are summarized in Table  II The specific activity of wild-type enzyme, measured under the standard assay conditions at pH 7.4, containing 4 mM isocitrate, 2 mM Mn 2ϩ , and 0.1 mM NADP, is 37.8 mol min Ϫ1 mg Ϫ1 . Table II also shows the V max values of wild-type and the mutant isocitrate dehydrogenases extrapolated to saturating concentrations of NADP. The V max value of wild-type enzyme is 42.9 mol min Ϫ1 mg Ϫ1 and the corresponding values of R83Q, R83K, T311A, and N328S mutant enzymes are only a little lower, whereas the V max of N328D enzyme is reduced to 36% of that of wild-type enzyme. Notably, the T311N mutant enzyme has a markedly decreased V max value (less than 1% of the wild-type value); the relatively large size of the asparagine which replaces the threonine may so distort the binding of the coenzyme that the maximum catalytic rate is affected. In contrast, substitution of threonine at position 311 by serine not only does not decrease activity but actually gives a large increase in V max (82.4 mol min Ϫ1 mg Ϫ1 ). However, it is notable that k cat /K m NADP for the T311S mutant is similar to that of wild type (Table II).  pH Dependence of V max -The pH-V max profiles of wild-type and mutant isocitrate dehydrogenases were measured using saturating concentrations of substrate and coenzyme (4 mM isocitrate, 2-6 mM Mn 2ϩ , and 1-10 mM NADP). The V max of the recombinant wild-type enzyme depends on the basic form of an ionizable group of the enzyme-substrate complex (9). The dependence of V max on pH was analyzed by Equation 1, where V max,obs is the maximum velocity measured at each pH, V max,int is the intrinsic maximum velocity, which is independent of pH, and K es is the dissociation constant of an ionizable group of the enzyme-substrate complex. Over the pH range 5-8, the pH-V max profiles for all Arg 83 and Asn 328 mutant enzymes are similar to that of wild type, whereas those for Thr 311 mutant enzymes are shifted toward higher pH, yielding higher pK es values. Table III summarizes the pK es values for wild type, as well as the Arg 83 , Thr 311 , and Asn 328 mutant enzymes. For all the enzymes, substrates saturated the active site under the conditions used. Increasing the already high concentrations of NADP to 10 mM and Mn 2ϩ to 6 mM did not change the pH-V max profiles. The pK es values of the Arg 83 mutant enzymes and N328D are similar to that of the wild-type enzyme (about 5.2 to 5.4). The V max values of these mutant enzymes probably depend on the ionizable form of the same group as in the wildtype enzyme. The pK es of the N328S mutant enzyme is slightly higher, 5.63, which may reflect a relatively small perturbation of the pK es of the same ionizable group seen in wild type. In contrast, the pK es values of all Thr 311 mutant enzymes are significantly higher than that of the wild-type enzyme. Although the V max,int of the Thr 311 mutant enzymes ranges from 0.33 to 82 mol min Ϫ1 mg Ϫ1 , all three Thr 311 mutant enzymes have similar pK values, 6.0 -6.2.

DISCUSSION
In this study, the characterization of mutant enzymes with replacements at Arg 83 , Thr 311 , and Asn 328 have implicated these three amino acids in the coenzyme site of the porcine mitochondrial NADP-dependent isocitrate dehydrogenase. The Arg 83 and Asn 328 mutant enzymes, as well as the T311A enzyme have strikingly elevated K m values for NADP, while the K m values for the other substrates are similar to those of wild-type enzyme. In the structure shown in Fig. 1, the guanidino group of Arg 83 is close to the 3Ј-OH of the nicotinamide ribose, and to the 5Ј-O of that ribose, which is linked to a phosphorus of the pyrophosphate moiety. The distance between the 3Ј-OH and the closest nitrogen of Arg 83 is 4.6Å; however, there is an enzyme-bound water only 2.83Å from that N of Arg 83 , which may mediate a hydrogen bond between the 3Ј-OH of the ribose and Arg 83 . The closest distance is 5.7 Å between Arg 83 and the 5Ј-O of the pyrophosphate. It is apparent that the positive charge of the amino acid at position 83 contributes to the affinity between the enzyme and NADP because the K m NADP for R83K is one-fourth that of R83Q. The structure of the NADP complex of the human cytosolic NADP-dependent isocitrate dehydrogenase has recently been reported as the second mammalian IDH to be crystallized (18); however, the functions of the amino acids near the coenzyme have not been evaluated by mutagenesis. The porcine mitochondrial NADP-IDH is 77% identical plus similar in amino acid sequence to human cytosolic IDH and an overlay of the structures of the porcine mitochondrial IDH (PDB 1LWD) and the human cytosolic IDH (PDB 1T09) shows the strong structural resemblance of the two enzymes. In the structure of the human cytosolic enzyme, the corresponding Arg 82 is also considered to form a hydrogen bond with the 3Ј-OH of the nicotinamide ribose of NADP (18).
In the porcine mitochondrial IDH (Fig. 1), Thr 311 is close to the oxygen in the ring of the furanose form of the nicotinamide ribose of NADP, as well as to the 5Ј-O of the ribose. The ability to form a hydrogen bond and the size of the amino acid at position 311 are important in NADP binding and in the catalytic reaction. Replacement of Thr 311 by alanine, which cannot form hydrogen bonds, results in more than a 10-fold increase in K m NADP . In contrast, when Thr 311 is replaced by serine, which can form hydrogen bonds, the K m NADP is only 1.7ϫ that of wild-type enzyme. Substitution of the larger asparagine for Thr 311 allows hydrogen bonding, but the enzyme-NADP complex must be distorted in accommodating the larger side chain. Thus, for the T311N enzyme, the K m NADP is similar to that of the wild-type enzyme, but V max is less than 1% that of wild type. Thr 311 is the closest, of the three amino acids we have here mutated, to the enzyme-bound Mn 2ϩ -isocitrate. In the catalytic dehydrogenation of isocitrate, a hydride is transferred from isocitrate to the N-4 position of the nicotinamide of NADP. It is reasonable that any perturbation of the enzyme-bound nicotinamide of NADP, which likely occurs when a larger amino acid is substituted at position 311, results in a decreased V max . Indeed substitution of any amino acid at position 311 causes a change in the pH dependence of V max , as indicated by the elevated pK es for all the Thr 311 mutant enzymes (from 5.2 in wild type to about 6.0 in all the Thr 311 mutant enzymes). We have previously attributed the pH dependence of V max to the deprotonation of the metal-bound hydroxyl of isocitrate in the enzyme-substrate complex (9). Although Thr 311 is about 12Å from the enzyme-bound Mn 2ϩ (Fig. 1), any distortion of the nicotinamide could readily be transmitted to the nearby Mn 2ϩisocitrate with a concomitant change in the pK es . It is notable that the corresponding Thr 311 in the human cytosolic IDH (18) occupies a similar location to that of Thr 311 in the mitochondrial enzyme, and it may have the same role in interacting with NADP.
Asn 328 of the porcine mitochondrial IDH is located at the other end of the coenzyme binding site near the adenine ring TABLE III Kinetic parameters for the pH-V max profiles for wild-type and mutant enzymes The pK es values were measured in different pH buffers containing various concentrations of NADP, Mn 2ϩ , and isocitrate. The pH-V max curves were superimposable when 0.5 or 1 mM NADP, 2 or 4 mM Mn 2ϩ and 4 or 8 mM isocitrate were used for wild type; when 1, 7, or 10 mM NADP, 2, 4, 5, or 6 mM Mn 2ϩ and 4 mM isocitrate were used for R83Q; when 1, 5, 7, or 10 mM NADP, 2 or 6 mM Mn 2ϩ and 4 mM isocitrate were used for R83K; when 1, 2, 5, 7, or 10 mM NADP, 2, 4, 5, or 6 mM Mn 2ϩ and 4 mM isocitrate were used for T311A; and when 1 or 2 mM NADP, 2 mM Mn 2ϩ and 4 mM isocitrate were used for N328S. For T311N and T311S, 0.1 mM NADP, 2 mM Mn 2ϩ and 4 mM isocitrate were used; and for N328D, 1 mM NADP, 2 mM Mn 2ϩ and 4 mM isocitrate were used. These results indicate that these enzymes were saturated with substrates over the full pH range and that the measured values were actually V max values as a function of pH. The substrate concentrations are all high relative to the K m values. For example, for wild-type, R83Q, R83K, and T311A, at pH 5.5 the K m values for NADP were 17-261 M, and for Mn 2ϩ 1-6 M.  (Fig. 1). The nitrogen of the carbamido side chain of the asparagine is within hydrogen bonding distance (2.70Å) of the N1 of the adenine, and thus can facilitate the binding of NADP. Replacement of Asn 328 by the much smaller Ser results in a 20-fold increase in the K m NADP , likely reflecting the increased distance between Ser 328 and the N1 of the purine ring. Substitution of the similarly sized but negatively charged Asp for Asn 328 is also detrimental to the affinity between enzyme and NADP (K m NADP is 4.4ϫ that of wild-type enzyme), but less so than in the case of N328S. In the human cytosolic NADPspecific isocitrate dehydrogenase, Asn 328 is also positioned to hydrogen bond to the N1 of the NADP adenine (18). Indeed, most of the amino acid side chains close to the NADP bound to the porcine mitochondrial IDH occupy similar locations in the crystal structure of the human cytoplasmic IDH (18).
In previous studies from our laboratory, mutagenesis experiments indicated that Arg 314 and Tyr 316 (14), as well as His 315 (13) of the porcine mitochondrial NADP-specific IDH dehydrogenase interact with the 2Ј-phosphate of NADP and contribute to the coenzyme specificity of the enzyme. His 309 not only can affect coenzyme binding, but also influences the enzyme interaction with metal ion in the presence of isocitrate (13). Although sequence alignment indicates there is only 16% identity between the porcine and E. coli IDHs (7), amino acids corresponding to His 309 , Arg 314 , and Tyr 316 have been observed in the NADP site of the crystalline isocitrate dehydrogenase of E. coli (19,20). In the hyperthermophile Aeropyrum pernix, the adenine ring of NADP lies between the main chain of Asn 356 and the side chain of His 343 (21). The Asn 356 and His 343 of A. pernix IDH are equivalent to Asn 328 and His 309 , respectively, of porcine mitochondrial NADP-IDH (Fig. 2). It has been pointed out that the use of NADP by prokaryotic IDHs is an ancient adaptation to growth on acetate (22), so it is not surprising that most of the amino acids which interact with the coenzyme are conserved. However, only in the other crystalline mammalian enzyme, human cytosolic NADP-IDH (18), are all of the amino acids identified in the coenzyme site of the porcine mitochondrial NADP-IDH found in approximately the same location.
The present study of site-directed mutagenesis in the coenzyme site of porcine mitochondrial NADP-dependent isocitrate dehydrogenase is consistent with the crystal structure of the Mn 2ϩ -isocitrate complex of the same enzyme with NADP positioned in silico (7), and with the recently reported crystal structure of the human cytosolic IDH complexed with NADP (18). Arg 83 , Thr 311 , and Asn 328 are all located close to enzymebound NADP where they are determinants of the enzyme affinity for coenzyme. In addition, Thr 311 is in the vicinity of the nicotinamide ring of NADP, which must be close to the enzymebound metal-isocitrate, and therefore the amino acid at position 311 can indirectly influence the ionization of the metalliganded hydroxyl of isocitrate in the active site. The porcine mitochondrial NADP-IDH has 95% amino acid sequence identity with human mitochondrial NADP-dependent isocitrate dehydrogenase. Therefore it is likely that the results of this study will be applicable to the human enzyme.