Identification of Mn2+-binding Aspartates from α, β, and γ Subunits of Human NAD-dependent Isocitrate Dehydrogenase*

The human NAD-dependent isocitrate dehydrogenase (IDH), with three types of subunits present in the ratio of 2α:1β:1γ, requires a divalent metal ion to catalyze the oxidative decarboxylation of isocitrate. With the aim of identifying ligands of the enzyme-bound Mn2+, we mutated aspartates on the α, β, or γ subunits. Mutagenesis target sites were based on crystal structures of metal-isocitrate complexes of Escherichia coli and pig mitochondrial NADP-IDH and sequence alignments. Aspartates replaced by asparagine or cysteine were 206, 230, and 234 of the α subunit and those corresponding to α-Asp-206: 217 of the β subunit and 215 of the γ subunit. Each expressed, purified mutant enzyme has two wild-type subunits and one subunit with a single mutation. Specific activities of WT, α-D206N, α-D230C, α-D234C, β-D217N, and γ-D215N enzymes are 22, 29, 1.4, 0.2, 7.3 and 3.7 μmol of NADH/min/mg, respectively, whereas α-D230N and α-D234N enzymes showed no activity. The Km,Mn2+ for α-D230C and γ-D215N are increased 32- and 100-fold, respectively, along with elevations in Km,isocitrate. The Km,NAD of α-D230C is increased 16-fold, whereas that of β-D217N is elevated 10-fold. For all the mutants Km,isocitrate is decreased by ADP, indicating that these aspartates are not needed for normal ADP activation. This study demonstrates that α-Asp-230 and α-Asp-234 are critical for catalytic activity, but α-Asp-206 is not needed; α-Asp-230 and γ-Asp-215 may interact directly with the Mn2+; and α-Asp-230 and β-Asp-217 contribute to the affinity of the enzyme for NAD. These results suggest that the active sites of the human NAD-IDH are shared between α and γ subunits and between α and β subunits.

Our previous work on pig heart NAD-dependent isocitrate dehydrogenase showed that this enzyme has two binding sites per tetramer for each of its ligands: isocitrate, Mn 2ϩ , NAD, ADP, NADH, and NADPH (4,5). These binding studies either indicate that these distinctive subunits have specialized functions for particular ligand sites or that the binding site for each ligand is shared between two subunits (4,5). The three types of subunits of the pig heart enzyme can be separated by chromatofocusing in the presence of urea (6). Isolated ␣, ␤, and ␥ subunits are either inactive or exhibit very low activity, but recombination of isolated ␣ with either ␤ or ␥ results in formation of either ␣␤ or ␣␥, which have substantial catalytic activity (6). These observations suggest that dimers may be the minimal functional subunits. The active site may be either within the ␣-subunits or shared between ␣ and ␤ or between ␣ and ␥ subunits.
No structure of an NAD-dependent isocitrate dehydrogenase has yet been determined. However, the crystal structure of NADP-specific isocitrate dehydrogenase of Escherichia coli (7,8), Bacillus subtilis (9), and pig mitochondria (10) are known. In the crystal structure of the manganese-isocitrate complex of pig mitochondrial NADP-specific IDH, 2 Asp-252, -275, and -279 are either direct ligands of the Mn 2ϩ or interact through a water molecule (10), whereas in the E. coli enzyme, Asp-283, -307, and -311 are similarly located (7,8). Amino acid sequence comparison of the three subunits of the human NAD-IDH enzyme with the subunits of the pig mitochondrial and E. coli NADP enzymes using ClustalW indicates relatively low sequence identity. However, certain amino acids in the E. coli and pig mitochondrial NADP-isocitrate dehydrogenases that are known to interact with Mn 2ϩ and isocitrate are conserved among the three subunits of the human NAD-IDH. We have previously reported on the roles of three conserved arginines of the NAD-dependent isocitrate dehydrogenase (11). Fig. 1 shows the amino acid sequence alignment in the region of the metal ligands of the pig mitochondrial and E. coli NADP-IDHs. Although there is relatively low homology among these enzymes, the ␣-subunit Asp-206, -230, and -234 of the human NAD-IDH can be aligned with Asp-252, -275, and -279 of the pig mitochondrial NADP-IDH as well as with Asp-283, -307, and -311 of the E. coli NADP-IDH. In addition, the ␤ subunit Asp-217 and ␥ subunit Asp-215 are comparable with Asp-252 of the pig NADP-IDH and Asp-283 of the E. coli enzyme. * This work was supported by National Institutes of Health Grant R01 HL67774. A preliminary version of some of this work has been presented (12). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 To whom correspondence should be addressed. Tel.: 302-831-2973; Fax: 302-831-6335; E-mail: rfcolman@udel.edu.
In the present study, in order to evaluate their roles in the human NAD-dependent isocitrate dehydrogenase in the binding of metal ion, in catalytic activity, and in ADP activation, we have mutated Asp-206, -230, and -234 of the ␣ subunit, Asp-217 of the ␤ subunit, and Asp-215 of the ␥ subunit. Each aspartate was separately replaced by neutral asparagine and/or cysteine. Enzyme containing one mutant subunit and two wildtype subunits was expressed in E. coli to yield complete enzyme with subunits in the ratio 2␣:1␤:1␥. Here, we describe the characteristics of these complete enzymes that harbor a single mutation in one subunit.
Site-directed Mutagenesis-The complete recombinant human NAD-dependent IDH was expressed in E. coli XL Gold ultracompetent cells (Stratagene) using the plasmid pHIDH␣␤ 2 ␥ (7.0 kbp) (11,13). Using a PCR-based protocol in accordance with the XL QuikChange site-directed mutagenesis kit (Stratagene), point mutations were introduced into a single subunit, while maintaining the other two subunits as wild type in the plasmid (pHIDH␣␤ 2 ␥). Appropriate changes in time and temperature for denaturation, annealing, and polymerization were made to the standard PCR protocol for low melting point primers in order to maximize the product yield.
The oligonucleotide primers (and their complements) used to generate mutant enzymes were as follows: 5Ј-GACAATGATCATAAAC-AACTGCTGCATGCAGC), and ␥D215N-F (5Ј-CGAGAACAT-GATTGTGAATAACACCACCATGC). The underlined codons are those mutated from aspartate to asparagine or cysteine. The primer, ␤T4S-F (5Ј-GCATCGCGGAGCCAGGCCGAGGACG), and its complement were used to change the amino acid threonine to serine at the 4th position from the N terminus in the ␤ subunit of the plasmid, pHIDH␣␤ 2 ␥. This change was made to correct the amino acid sequence of the wild-type enzyme, since threonine had mistakenly been substituted in that position in the ␤ subunit during plasmid construction (13). In each case, the occurrence of the desired mutation or change in the subunit sequence was confirmed by DNA sequencing in both directions (DNA Sequencing and Genotyping Center, University of Delaware). DNA sequencing of the other two subunits in the plasmid was also conducted to ensure that undesired mutations were not introduced into those subunits intended to be wild type. This precaution was followed because of the similarity in the ␣, ␤, and ␥ subunit amino acid sequences.
Transformation and Expression of Recombinant Human IDH Proteins in E. coli-Wild type or mutant plasmid DNA from the PCR was used to transform E. coli XL 10 Gold ultracompetent cells (Stratagene). The colonies with positive inserts were subcultured by growing overnight at 37°C in LB medium (250 ml) supplemented with ampicillin (0.1 mg/ml). These E. coli XL 10 Gold cell transformants harboring pHIDH␣␤ 2 ␥ plasmid were used to express human NAD-dependent isocitrate dehydrogenase instead of the IDH-deficient E. coli EB 106 (DE 3) cells used in our previous work (11), because the yield of the pure recombinant IDH was greater. (In our earlier study, each point mutation was introduced into the cDNA encoding a single subunit, and the plasmid encoding the complete enzyme was subsequently and laboriously assembled from the separate DNA plasmids encoding one mutant subunit and two wild-type subunits (11).) To express the enzyme, five 6-liter flasks, each containing 2 liters of LB medium with 0.1 mg/ml ampicillin, were inoculated with freshly grown E. coli XL cells (2% v/v) and were grown at 37°C while being shaken at 200 -220 rpm for 4 h or until the growth reached a midlog phase or cell density of A 600 nm ϭ 0.4 -0.6. The flasks were then placed temporarily in chilled water to lower the culture temperature to room temperature. Protein expression was induced in cells by the addition of isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 0.4 mM, and the flasks were shaken at the lower speed of 100 -140 rpm with minimal aeration at 23-25°C for 20 -22 h. (Incubation at the lower temperatures of 15 and 20°C did not improve the yield of expressed proteins.) The 10 liters of cell culture was then centrifuged at 5000 ϫ g for 10 min to separate the cells from media and suspended in a total volume of 250 ml of cold 11 mM citrate-Tris buffer, pH 7.2, containing 10% glycerol, 0.2 mM MnSO 4 , and 0.1 mM DTT. These suspensions were stored at Ϫ80°C. Frozen cells, ϳ80 -100 ml at a time, were FIGURE 1. Amino acid sequence alignment of a selected segment of ␣, ␤, and ␥ subunits of human NAD-dependent isocitrate dehydrogenase, NADP-dependent isocitrate dehydrogenases from pig (heart) mitochondria, and E. coli. The star indicates that the amino acids at the position are identical, and the dot indicates that the amino acids are similar, using Clust-alW. The Asp residues of the human enzyme subunits shown in boldface letters were mutated to Asn, and those indicated by boldface letters and underlined were replaced by Asn and Cys. The Asp residues of the porcine and E. coli enzymes numbered in superscript are aligned with the boldface Asp of the human enzyme subunits.
partially thawed and lysed, using sonication conditions described previously, while keeping the preparation in ice (11). The lysate was centrifuged at 16,000 ϫ g for 30 min, and the clear supernatant (crude extract) was separated. The protein concentration and isocitrate dehydrogenase activity were determined in this crude extract.
Purification of Recombinant IDHs from the Crude Extract-The isolation and purification of enzymes from the crude extract was modified from the previous procedure (11). Unlike IDH-deficient E. coli EB 106 cells, the E. coli XL cells used in this study possess constitutive NADP-dependent IDH enzyme in addition to the recombinant human NAD-dependent IDH enzyme. Since the human NAD-IDH and the E. coli NADP-IDH exhibit very different isoelectric points and have different molecular sizes under native conditions, they can be differentially precipitated from the crude extract by ammonium sulfate fractionation; in addition, they are bound and eluted from ion exchange resins under different conditions. The human NAD-IDH was further purified to homogeneity by gel filtration.
Ammonium sulfate precipitation was followed as described previously (11) to isolate NAD-IDH from the crude extract. The NAD-IDH precipitates predominantly between 30 and 50% saturated ammonium sulfate; this fraction was pooled and further purified. The protein precipitate was dissolved in 12 mM citrate-Tris buffer, pH 7.7, containing 20% glycerol, 0.2 mM MnSO 4 , and 0.1 mM DTT (Buffer A) and dialyzed against 6 liters of Buffer A with three changes, for 6 h each. The enzyme was applied to a DE-52 column (3 ϫ 13 cm), which had previously been equilibrated with Buffer A. The column was eluted with the same buffer until the A 280 nm reached the base line. A linear gradient was then started from 200 ml of buffer A to 200 ml of 50 mM citrate-Tris buffer, pH 7.7, containing 20% glycerol, 0.2 mM MnSO 4 , and 0.1 mM DTT. The NAD-dependent isocitrate dehydrogenase typically elutes between 150 and 230 ml of the gradient. In contrast, the E. coli NADP-dependent IDH elutes later, from about 230 ml to the end of the gradient. Fractions (5 ml each) exhibiting specific activities more than 5-6.0 units/mg (wild type) were pooled and concentrated to ϳ10 ml by ultrafiltration.
The pool was dialyzed with three changes, against 6 liters of 12 mM sodium citrate buffer, pH 5.7, containing 20% glycerol and 0.1 mM DTT (Buffer B). To facilitate NAD-IDH binding to the cation exchange resin, MnSO 4 was not added to the dialysis or to the elution buffers. Some loss of IDH activity occurred at this step. The dialyzed enzyme was applied to a cellulose phosphate cation exchange column (2.5 ϫ 11.0 cm). The column was washed with ϳ125 ml of Buffer B until the A 280 nm reached the base line. The traces of NADP-IDH carried over from the DE-52 column eluted in the wash, since E. coli IDH does not bind to cellulose phosphate under these conditions. A linear gradient, consisting of 100 ml of Buffer B and 100 ml of 0.15 M sodium citrate buffer, pH 5.7, containing 20% glycerol and 0.1 mM DTT, was used to elute the NAD-IDH. The enzyme elutes between 75 and 150 ml of the gradient. Fractions of 5.0 ml were collected in tubes containing MnSO 4 (final concentration, 2 mM) to maintain enzyme stability. Aliquots of fractions showing high specific activity were screened by SDS-PAGE. Generally, the fractions containing the NAD-IDH exhibit the pres-ence of two minor impurities with subunit molecular sizes of 90 and 45 kDa. These were pooled and concentrated to a 1-1.5-ml volume.
After dialyzing this pool against 4 liters of 50 mM citrate-Tris, pH 7.2, buffer containing 20% glycerol, 0.2 mM MnSO 4 , and 0.1 mM DTT (Buffer C) for 6 h, the enzyme was separated from the other proteins by elution through an Ultragel ACA 34 (molecular mass cut-off 350 kDa) gel filtration column (1.0 ϫ 124 cm). The proteins were eluted using Buffer C at the rate of 7.0 -7.5 ml/h, and ϳ1.0-ml volume fractions were collected. Three distinct proteins were observed, with the NAD-dependent isocitrate dehydrogenase being eluted as the middle peak. Protein fractions that were homogenous on SDS-PAGE and exhibited high specific activity for NAD-IDH were pooled and used for characterization and kinetic studies. Pure mammalian NADdependent isocitrate dehydrogenase was recognized by the appearance of two close bands in approximately equal intensity, with an upper band (␤ and ␥ subunits) of 39,000 Da and a lower band of 37,000 Da (␣ subunits) (3,14).
Protein Determination and Assay for NAD and NADP-dependent IDH Activity-In the early stages of protein purification from the crude extract, the protein concentration was determined from the A 280 nm after correcting for the ratio of A 280 /A 260 (15). In the pure preparations, the protein concentrations in mg/ml were calculated using E 280 nm 1% ϭ 6.45 (4). Enzyme activity was determined by monitoring at 340 nm the time-dependent increase in UV absorbance of NADH produced by reduction of NAD at 25°C in a 1-ml standard assay mixture containing 33 mM Tris acetate buffer, pH 7.2, 20 mM DL-isocitrate, 1 mM MnSO 4 , and 1 mM NAD as the final concentrations. The specific activity was expressed as mol/min/mg of protein. To minimize interference from NADH oxidase while assaying for IDH activity in crude preparations or ammonium sulfate fractions containing lower amounts of expressed mutant enzymes, the NADH oxidase inhibitor, 10 l of rotenone (dissolved in 100% ethanol), was added to the 1 ml of standard assay solution to give a final concentration of 2.5 M rotenone. Since the normal E. coli cells have NADP-dependent IDH at the early steps of purification (ammonium sulfate fractionation and DEAE-cellulose chromatography), enzyme samples were also assayed for NADP-dependent IDH. The NADP-IDH activity was determined at 25°C by monitoring the increase in UV absorption of NADPH at A 340 nm in a 1-ml assay mixture containing 30 mM triethanolamine chloride, pH 7.4, buffer, 0.1 mM NADP ϩ , 4 mM DL-isocitrate, and 2 mM MnSO 4 .
SDS-PAGE-Aliquots of fractions containing 10 g of protein were analyzed in 15% polyacrylamide gels containing 0.1% SDS in a discontinuous pH electrophoresis system (16) to evaluate the purity of the protein samples during purification. The solutions for preparing the stacking and resolving gel, protein staining and destaining, and other electrophoresis conditions were described previously (17).

CD of Recombinant Wild Type and Mutant
Enzymes-To evaluate the secondary structure of wild-type and mutant enzymes, the ellipticity was measured as a function of wavelength between 250 and 200 nm in a 0.1-cm path length quartz cell using a Jasco model J-710 spectropolarimeter. The wildtype and mutant proteins were dialyzed against 2 liters of 25 mM

Asp in Mn 2؉ Sites of Human NAD-isocitrate Dehydrogenase
triethanolamine chloride buffer, pH 7.4, containing 10% glycerol and 0.2 mM MnSO 4 , and the same buffer was used to dilute the enzymes to 0.1 mg/ml. The mean residue molar ellipticity [] (degrees cm 2 dmol Ϫ1 ), and molar concentrations of proteins were determined using average subunit molecular weights and number of residues per subunit, as described before (11).
Determination of Amino Acid Sequence and Subunit Composition of Enzymes-The ␣, ␤, and ␥ subunits of recombinant NAD-IDH enzymes were identified by comparing their N-terminal sequences with those of the known sequences of the enzyme subunits (13,18), as described previously (11). The N-terminal amino acid sequences of wild type and mutant enzymes were determined using an Applied Biosystems protein/peptide sequencer (model Procise) equipped with an online Microgradient Delivery System (model 140 C) and a Macintosh computer (model 610). Since the N-terminal amino acid sequence of the ␣, ␤, and ␥ subunits are different, the subunit composition of wild-type and mutant enzymes was determined from the molar ratios of amino acids at positions 1, 6, and 7 upon sequencing the whole protein. Alternatively, to confirm the mutant subunit ratios of the enzymes, they were also determined from the SDS denatured subunits separated by a reverse phase column in HPLC and compared with those of wild-type enzyme, as described before (11). Effect of Various Metal Ions on the pK a of the Wild-type Human NAD-IDH-For determination of the pH-V max profile, the enzyme activity of the wild type was measured at various pH values using the following buffers: sodium acetate, pH 4.4 -5.8; MES, pH 5.4 -6.6; PIPES, pH 6.2-7.4; and triethanolamine hydrochloride, pH 6.8 -7.6. The final buffer concentration was 30 mM in 1 ml of standard assay solution. The metal ions were present at the following concentrations: MnSO 4 (1 or 2 mM), Co(NO 3 ) 2 (5 or 10 mM), and MgSO 4 (5 and 10 mM) in the 1-ml assay solutions in order to saturate the enzyme with metal ion over the entire pH range. At the lower metal ion concentra-tions, 20 mM isocitrate and 1 mM NAD were used in the assay, whereas 40 mM isocitrate and 2 mM NAD were used with the higher metal ion concentration. The results were the same with the two sets of substrate concentrations, indicating that the enzyme was saturated with respect to substrates. The dependence of the observed V max on pH was analyzed in accordance with the equation,

Michaelis-Menten Constants for Isocitrate, Metal Ions, and Coenzyme of Wild Type and Mutant Enzymes-To
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 aes is the dissociation constant of an ionizable group of the enzyme-substrate complex.

Site-directed Mutagenesis, Expression, and Purification of
Recombinant Human NAD-IDHs-Here, for the first time, sitedirected mutagenesis was carried out on the complete plasmid harboring human NAD-dependent isocitrate dehydrogenase using the PCR-based QuikChange mutagenesis method. We have successfully replaced aspartates with asparagine or cysteine at positions 206, 230, and 234 in the ␣ subunit, 217 in ␤ subunit, or 215 in ␥ subunit. Each point mutation was accomplished in the target subunit without affecting the other subunits, as demonstrated by DNA sequencing of all subunits. The wild type or mutant plasmid DNA resulting from PCR was transformed into E. coli XL Gold competent cells (Stratagene), and cells with positive inserts were used to express the recombinant enzymes. The recombinant human NAD-IDHs were purified to homogeneity as described under "Experimental Procedures," giving a yield of enzyme higher than that obtained from the IDH-deficient E. coli EB 106 cells that we used previously (11). Furthermore, as shown in Table 1, the mutagenesis and expression system described in this paper resulted in wildtype human NAD-dependent isocitrate dehydrogenase with the same specific activity as reported previously (4,11).
The purity of the enzyme preparations was assessed by SDS-PAGE, as shown in Fig. 2. The wild type and all seven purified human mutant enzymes exhibit the two close bands with equal

Specific activities of recombinant wild-type and mutant human NAD-dependent isocitrate dehydrogenases
As described under "Experimental Procedures," the specific activities for wild-type and mutant purified human NAD-IDHs were generally determined using standard assay solutions.

Enzyme
Specific activity intensity that are characteristic of mammalian NAD-dependent isocitrate dehydrogenase (3,11,14), with an upper band of 39 kDa (␤ and ␥ subunits) and a lower band of 37 kDa (two ␣ subunits). To confirm the identity of ␣, ␤, and ␥ subunits of human NAD-IDH, the purified enzymes were subjected to Edman degradation. The sequencing results show that the N-terminal sequences of the ␣, ␤, and ␥ subunits of the purified enzymes are the same as those of the distinct sequences of the three subunits of human NAD-dependent isocitrate dehydrogenase (11,13,18).

Circular Dichroism Spectra of Wild Type and Mutant
Enzymes-To test whether the mutations affect the conformation of NAD-IDH, the circular dichroism spectra of the wildtype and mutant enzymes were determined. The CD spectra of all of the mutant enzymes are superimposable on that of the wild-type enzyme, reflecting appreciable amounts of ␣-helical content. The results indicate that these mutations do not cause detectable change in the secondary structure of the NAD-IDH.
Determination of Subunit Composition of Recombinant Enzymes-Using the N-terminal sequences of ␣, ␤, and ␥ subunits obtained from Edman degradation of the whole purified enzymes, the subunit compositions were determined, as described under "Experimental Procedures." The subunit compositions of the wild-type and mutant human NAD-dependent isocitrate dehydrogenases, as shown in Table 2, indicate that the ratio of ␣/␤/␥ subunits is ϳ2:1:1, which is characteristic for NAD-dependent isocitrate dehydrogenase isolated from pig heart (2,3) and is also similar to that previously obtained for human NAD-IDH (11). These results also indicate that the mutagenesis, expression, and purification procedures described in this paper yield complete recombinant human IDH enzymes with subunits assembled in the correct composition.
Determination of Specific Activities of Wild Type and Mutant Recombinant Enzymes-The specific activities of wild-type and mutant enzymes were determined, in most cases, under the standard assay conditions, which include 20 mM isocitrate and 1 mM MnSO 4 . The concentrations of isocitrate and Mn 2ϩ were increased to assay those mutant enzymes that did not show activity with the standard assay, with the results reported in Table 1. At each position, aspartic acid was replaced by asparagine. Only in the case of D206N is the specific activity the same or higher than that of wild-type enzyme, suggesting that ␣-Asp-206 is not involved in the catalytic activity of the enzyme. In contrast, the D230N and D234N mutants exhibited almost no activity even when assayed at 70 mM isocitrate and 60 mM MnSO 4 , indicating that ␣-Asp-230 and ␣-Asp-234 are essential for catalytic activity.
In the case of the pig heart mitochondrial NADP-specific isocitrate dehydrogenase, it has been found that cysteine is an acceptable (albeit less effective) substitute for critical aspartates in the manganese-isocitrate binding site (19). Therefore, for human NAD-IDH, we replaced ␣-Asp-230 and ␣-Asp-234 with cysteine. The results shown in Table 1 indicate a specific activity of ␣-D230C as high as 6.6% that of wild type when assayed at unusually high concentrations of Mn 2ϩ , whereas ␣-D234C has the measurable specific activity of 0.9%, as compared with that of WT, under standard assay conditions. Clearly, cysteine is a better substitute than asparagine for the naturally occurring aspartate at these positions.
Because the NADP-specific isocitrate dehydrogenases have been shown to have three aspartates in the region of the metalisocitrate site that are important for their function (7,8,19), we anticipated that the NAD-dependent isocitrate dehydrogenase might also have three critical aspartates. Since ␣-D206N is highly active, ␣-Asp-206 was not a candidate for a functionally important aspartate; therefore, we evaluated by mutagenesis the corresponding aspartates in the ␤ and ␥ subunits. Table 1 shows that under standard assay conditions, the specific activity of ␤-D217N is reduced to one-third that of wild type, implying that ␤-Asp-217 has some influence on catalytic function. More notable, however, is the effect of replacing ␥-Asp-215: ␥-D215N exhibits only 17% of the specific activity of the wildtype enzyme, and that activity is observed only when the Mn 2ϩ concentration is elevated 20-fold over that in the standard assay. The ␥-Asp-215 appears to contribute to binding of metal ion by the enzyme. Table 3 shows that among the ␣ subunit mutants, D206N exhibits Michaelis constants for Mn 2ϩ and NAD, as well as a V max value similar to those of wild type, implying a lack of involvement of Asp-206 in the metal or coenzyme binding functions of the enzyme. (Although the V max of ␣-D206N is a little higher than that of wild type, the values of k cat /K m are slightly lower for the mutant enzyme (Tables 3 and

TABLE 2 Subunit composition of recombinant wild type and mutant human NAD-dependent isocitrate dehydrogenases
As described under "Experimental Procedures," the subunit composition was determined from the molar ratios of amino acid yields (cycles 1, 6, and 7) obtained by Edman degradation of whole protein and by estimating the relative distribution of peak areas in HPLC separation of subunits obtained from the whole enzyme (11).

Enzyme
Subunit ratio ␣ ␤ ␥  (20 -22).) With an increase in K m,NAD of ϳ16-fold for D230C as compared with wild type, Asp-230 may also participate in coenzyme binding. Furthermore, D230C exhibits an extrapolated V max of only ϳ8% that of wild type, indicating that ␣-Asp-230 is a determinant of the maximum catalytic rate. For the ␣-D234C mutant, whereas the K m values for Mn 2ϩ and NAD are similar to those of wild type, the K m for Cd 2ϩ is 10-fold lower than that of wild type, suggesting that the cysteine at position 234 interacts strongly with Cd 2ϩ . However, the extrapolated V max with either Mn 2ϩ or Cd 2ϩ as divalent metal ion is only 1% that of wild type, implying that ␣-Asp-234 is needed for catalytic function.
For the ␤-D217N mutant, whereas the K m for Mn 2ϩ is unchanged and V max is ϳ65% that of wild type, the K m for NAD is elevated ϳ10-fold, indicating that Asp-217 may contribute to the binding of the coenzyme (Table 3). Surprisingly, D217N exhibits much stronger affinity for Cd 2ϩ , as indicated by a 20-fold lower K m value as compared with wild type. However, in this case, the V max (with Cd 2ϩ ) is only 4% of that of the same enzyme when Mn 2ϩ is the cation, suggesting an unusual and detrimental interaction of Cd 2ϩ with this mutant enzyme.
Notably, the ␥ subunit mutant, D215N, exhibits a 100-fold elevation in its K m,Mn 2ϩ and a significant increase in the K m for Cd 2ϩ . These results indicate that ␥-Asp-215 is likely to be another ligand for the Mn 2ϩ .
Kinetic Constants for Isocitrate and Activation of IDH Enzymes by ADP- Table 4 records in column 1 the K m values for the substrate isocitrate. Among the mutant enzymes, only ␣-D230C and ␥-D215N exhibit large increases (10 -20-fold) in the K m for isocitrate. These are the same enzymes with elevated K m values for Mn 2ϩ . These results are consistent with involvement of ␣-Asp-230 and ␥-Asp-215 in the binding of metalisocitrate at the active site.
All mammalian NAD-dependent isocitrate dehydrogenases are activated by ADP, which decreases the K m for isocitrate while leaving V max unchanged (1). Table 4 shows that the recombinant wild-type human NAD-IDH exhibits a 7-fold decrease in K m,isocitrate in the presence of 1 mM ADP; this result is consistent with previous reports for NAD-IDH (1,11,23). All of the mutants exhibit a substantial decrease in the K m for isocitrate when ADP is added, implying that none of these aspartates is directly involved in the allosteric effect of ADP.
Effect of Various Metal Ions on pK a of Recombinant Wild-type Human NAD-IDH-The pH dependence of V max of the porcine NAD-dependent isocitrate dehydrogenase was determined previously with Mn 2ϩ as the cation, yielding a value of 6.5-6.6 as the pK a for the enzyme-substrate complex (24,25). This result was interpreted as indicating a requirement in the catalytic reaction for the basic form of an enzymic histidine or acidic amino acid (i.e. Glu or Asp). Another possible ionizable group is the ␣-hydroxyl of enzyme-bound isocitrate, which may be a ligand for Mn 2ϩ ion, as reported in the crystal structure for

Michaelis-Menten constants for Mn 2؉ , Cd 2؉ , and NAD of recombinant wild-type and mutant NAD-IDH
The assays were conducted as described under "Experimental Procedures" using standard assay solutions containing various concentrations of metal or coenzyme. The Michaelis-Menten constants were determined from Lineweaver-Burk plots. ND, not determined. a To calculate the mol of dimeric enzyme/mg of protein, a molecular weight of 80,000 was used for NAD-IDH (1.25 ϫ 10 Ϫ8 mol of dimeric enzyme/mg of protein). b The constant was determined using varying concentrations of NAD under standard conditions except that the isocitrate concentration was 60 mM and the MnSO 4 concentration was 20 mM. c The activities were determined under standard assay conditions except for 10 mM NAD. d This mutant enzyme was assayed using standard assay solution, which includes 70 mM isocitrate. e The kinetic constants of this mutant for NAD were determined except for 70 mM isocitrate plus 25 mM Mn 2ϩ under otherwise standard conditions.

TABLE 4 Kinetic constants for isocitrate and ADP activation of wild type and mutant NAD-IDHs
As described under "Experimental Procedures," the activities were measured under standard assay conditions, which includes various concentrations of isocitrate and to determine the effect of ADP, 1 mM as final concentration of ADP was maintained unless otherwise indicated. the pig heart mitochondrial NADP-dependent isocitrate dehydrogenase (10,19). The pK of a water molecule bound to a metal ion depends on the metal ion to which it is bound, with the pK of Mg 2ϩ Ͼ Mn 2ϩ Ͼ Co 2ϩ (26,27). To determine whether the pH dependence of V max for the NAD-IDH could be due to the ionization of the metal-bound ␣-OH of isocitrate in the enzyme-substrate complex, pK aes values were determined for wild-type NAD-IDH using Co 2ϩ , Mn 2ϩ , or Mg 2ϩ as the essential divalent metal ions. In each case, the V max versus pH curve was the same when measured at two different high concentrations of metal ion, isocitrate, and NAD (as described under "Experimental Procedures"), indicating that the enzyme was saturated with substrate over the entire range of pH. Table  5 shows the values of pK aes for human wild-type NAD-dependent isocitrate dehydrogenase in the presence of Mg 2ϩ , Mn 2ϩ , or Co 2ϩ . The pK aes changes with the metal ion used in the order Mg 2ϩ Ͼ Mn 2ϩ Ͼ Co 2ϩ . These results are consistent with the pK aes , reflecting the deprotonation of the metal-bound ␣-hydroxyl of isocitrate bound to the enzyme.

DISCUSSION
The ␣, ␤, and ␥ subunits of human NAD-dependent isocitrate dehydrogenase exhibit strong resemblance in amino acid sequence, yet their isoelectric points are distinct. Amino acid sequence comparison using ClustalW reveals 32% identity plus 35% similarity among the three subunits. The ␤ and ␥ subunits have a stronger resemblance in sequence (53% amino acid identity plus 17% similarity) and are 12-15 amino acids longer than ␣. It is likely that the differences among subunits arose from gene duplication followed by divergent evolution; however, whether the sequence diversity is associated with specialized functions for the subunits has not yet been established. The minimum molecular weight for a complete functional enzyme is 160,000 (2), and, with its three types of distinguishable subunits (2 ␣, ␤, and ␥), the enzyme has only two binding sites per enzyme tetramer for every ligand tested (4,5). A major question for this allosteric enzyme is whether the enzyme active site is entirely located within the two ␣ subunits per tetramer and the ␤ and ␥ subunits have regulatory functions or, alternatively, whether the two active sites per tetramer are shared between dissimilar subunits (i.e. ␣␤ and ␣␥). In this study, we sought to address the issue by identifying the subunit location of the aspartates, which are ligands of the required divalent metal ion.
The pig mitochondrial NADP-dependent isocitrate dehydrogenase, with a determined structure for the manganese-isocitrate complex (10), can serve as a guide to our study of the mammalian NAD-specific isocitrate dehydrogenase. The porcine NADP-IDH is a dimer of identical subunits, as are the NADP-IDHs of E. coli (7) and B. subtilis (9). Each manganeseisocitrate bound to crystalline porcine NADP-IDH is located close to the interface between the two subunits (10), and amino acids from both subunits contribute to binding and catalysis (19,28). For manganese-isocitrate bound at the B subunit, the direct ligands of the hexacoordinate Mn 2ϩ are the carboxylate of Asp-252 from the A subunit and of Asp-275 from the B subunit, the ␣-hydroxyl and ␣-carboxylate of isocitrate, and two water molecules (10). Asp-279 of the B subunit is located a little further from the Mn 2ϩ but is within hydrogen bonding distance of the two water molecules that are coordinated to the metal ion (10,19). The roles of Asp-252, Asp-275, and Asp-279 of the porcine NADP-specific isocitrate dehydrogenase have been demonstrated by site-directed mutagenesis (19,29).
Here, we selected the target sites for mutagenesis in the human NAD-dependent isocitrate dehydrogenase because of their sequence alignment with the three critical aspartates of the porcine NADP-IDH. Asp-279 of the NADP-IDH corresponds to ␣-Asp-234 of NAD-IDH; Asp-275 of the NADP-enzyme is equivalent to ␣-Asp-230; and Asp-252 of the opposite subunit of the NADP-IDH aligns with ␣-Asp-206, ␤-Asp-217, and ␥-Asp-215. We replaced each of these amino acids with asparagine, because it lacks the negative charge of aspartate, which is important in binding metal cations but retains the size and shape of aspartate. In addition, at the two positions at which the asparagine replacement yielded a completely inactive enzyme (␣-D230N and ␣-D234N), we substituted cysteine, because, with its -SH group, it is known to be a metal ligand in a variety of proteins.
The results for mutants of the NAD-dependent isocitrate dehydrogenase can be summarized as follows. At position ␣-230, although ␣-D230N has no detectable activity, ␣-D230C exhibits a low but measurable V max (1.9 mol of NADH/ min/mg of enzyme) as compared with WT (22 mol NADH/ min/mg enzyme). These results indicate that aspartate at this position is important for function, whereas cysteine is an acceptable, but poorer, alternative. The ␣-D230C mutant exhibits increases of about 32-fold and 10-fold in the K m values for Mn 2ϩ and isocitrate, respectively, which is consistent with ␣-Asp-230 acting as a direct ligand of the Mn 2ϩ . In addition, the D230C enzyme has a 16-fold elevation in K m for NAD, which might indicate that normally ␣-Asp-230 participates in the binding of the coenzyme. However, an enzyme ligand of the Mn 2ϩ may also be close to the isocitrate ␣-hydrogen, which is transferred as a hydride to the nicotinamide of NAD. Thus, perturbation of the manganese-isocitrate binding by changing a ligand of the metal ion may indirectly have an adverse effect on the affinity of the enzyme for NAD. These characteristics are similar to those of mutants of Asp-275 of the porcine NADP-IDH (19,29).
At position ␣-234, the asparagine mutant is inactive, and the ␣-D234C mutant has a specific activity only 1% that of the wildtype NAD-IDH. However, the K m values for Mn 2ϩ , isocitrate, and NAD are normal. This pattern is similar to the properties of mutants of Asp-279 of the porcine NADP-IDH (19,29), which

Asp in Mn 2؉ Sites of Human NAD-isocitrate Dehydrogenase
is a "second shell" ligand of the metal ion. If ␣-Asp-234 has a corresponding location in the human NAD-IDH, replacement of this aspartate can indirectly affect the orientation of the substrate at the active site by changing the interaction of amino acid 234 with water coordinated to the Mn 2ϩ . Indirect effects of changing second shell ligands have been reported for other enzymes (30). Mutation of the third conserved aspartate of the ␣-subunit, ␣-Asp-206, has little effect on the kinetic characteristics of NAD-IDH. The ␣-D206N mutant has nearly normal K m values for Mn 2ϩ , isocitrate, and NAD, and a V max that is not decreased compared with wild-type enzyme. Clearly, ␣-Asp-206 is not involved either in the binding of substrates or in the catalytic reaction. These results contrast with the major changes in the kinetics of the porcine NADP-IDH upon mutation of Asp-252 (19), although the sequence alignment suggested that these aspartates were equivalent in the two enzymes. The explanation must lie in the fact that in the crystal structure of the pig NADP enzyme, Asp-275 and Asp-279 come from one subunit, whereas Asp-252 is contributed to the active site from the opposite subunit (10,19). Thus, the actual amino acids of the NAD-IDH, which correspond to Asp-252 of the crystalline NADP enzyme are likely to come from the ␤ and/or ␥ subunits.
The ␥-D215N mutant exhibits a K m for Mn 2ϩ that is elevated 100-fold, with a 20-fold increase in the K m for isocitrate, and only a small perturbation in the K m for NAD. These are characteristics expected if ␥-Asp-215 is a direct ligand of Mn 2ϩ , equivalent to Asp-252 of the NADP-IDH. Once this mutant enzyme is saturated with substrates, however, its V max is only decreased to ϳ40% of the wild-type value, indicating that the major role of ␥-Asp-215 is in binding to the metal ion, which, in turn, is coordinated to isocitrate.
In contrast, the corresponding ␤-D217N mutant exhibits a normal affinity for metal ion and for isocitrate, a V max that is only decreased to ϳ65% of the normal value, but a striking 10-fold elevation in the K m for NAD. The ␤-Asp-217 functions as a determinant of the affinity between the enzyme and its coenzyme.
The NAD-dependent isocitrate dehydrogenase is regulated by ADP, which acts by lowering the K m for isocitrate (1,4). All of the mutants examined in this study exhibit a decreased K m for isocitrate in the presence of ADP. Thus, none of these aspartates is required for the allosteric response to ADP. In a previous paper from this laboratory (11), we found that mutation of ␤-Arg-99 and ␥-Arg-97 resulted in loss of the allosteric response to ADP, leading to the suggestion that the ␤ and ␥ subunits are responsible for the regulation of the enzyme by nucleotides. The present study does not contradict the earlier conclusion. However, it indicates that the ␤ and ␥ subunits have additional roles in the active site: contributing a ligand for Mn 2ϩ and a determinant of the NAD affinity.
It has been proposed that isocitrate dehydrogenase reactions are initiated by the abstraction of a proton, facilitated by an enzymic general base, from the C2-hydroxyl of isocitrate prior to the transfer of the hydride to the nicotinamide of the coenzyme (31). For the NAD-IDH wild type enzyme, the pH dependence of V max was thought to reflect the deprotonation of that enzymic group in the enzyme-substrate complex; since the pK aes was about 6.5 when determined with Mn 2ϩ , the enzymic general base was postulated to be an aspartate or glutamate. However, there is another possibility that should be considered; the pK of 6.5 could represent the ionization of the metal-coordinated C2-hydroxyl of isocitrate, despite the fact that this pK is much lower than that of the hydroxyl of isocitrate when free in solution. Such an identification of the ionizable group has been made for the complex of manganese-isocitrate with the porcine NADP-IDH and is consistent with the crystal structure of this enzyme (10,19). There are other enzymes for which an enzymemetal-water complex exhibits a pK much lower than that of the metal-water complex alone (32,33). In the case of the NAD-dependent isocitrate dehydrogenase, the positively charged arginines in the region of the isocitrate site (11) may be responsible for the decreased pK. Characteristic of the ionization of a metalcoordinated water bound to an enzyme (34), or of a metalcoordinated hydroxyl group of a substrate bound to an enzyme (19), the pK changes with the metal ion, decreasing from Mg 2ϩ to Mn 2ϩ to Co 2ϩ . In the data for the wild-type human NAD-IDH presented in this paper, pK aes ranges from 6.78 with Mg 2ϩ to 6.51 with Mn 2ϩ to 6.16 with Co 2ϩ , which is consistent with the designation of the enzyme-bound metal-coordinated ␣-hydroxyl of isocitrate as the ionizable group influencing V max .
An important issue addressed in this paper is the functional role of the three dissimilar subunits of human NAD-dependent isocitrate dehydrogenase. Previous studies from this laboratory (11,(35)(36)(37) using mutagenesis and affinity labeling have provided evidence that the ␤ and ␥ subunits contain the allosteric ADP sites. The results presented in this paper constitute the first experimental demonstration that a metal-isocitrate site of human NAD-dependent isocitrate dehydrogenase has contributions from both ␣ and ␥ subunits (␣-Asp-230 and ␥-Asp-215 provide ligands for the Mn 2ϩ site, and ␣-Asp-234 is required for high catalytic activity, probably affecting the Mn 2ϩ through its water ligands). These results explain the earlier observations that the addition of ␥ subunit to ␣ subunit results in formation of an ␣␥ dimer with substantial regain in activity (6) and that the addition of ␥ subunit to an ␣ 2 ␤ complex enhances activity about 7-fold (35,38). The enzyme has two Mn 2ϩ binding sites per tetramer (4). For one Mn 2ϩ site, we have identified three aspartates (␣-Asp-230, ␣-Asp-234, and ␥-Asp-215), whereas for the second Mn 2ϩ site, we have located only two aspartates associated with metal binding (␣-Asp-230 and ␣-Asp-234), since the corresponding aspartate on the ␤ subunit does not seem to influence Mn 2ϩ interaction. Either the two Mn 2ϩ sites of the enzyme are not identical, or we have yet to find the third aspartate (on the ␤ subunit), which interacts with the metal ion. The data of this paper also constitute an experimental demonstration that the NAD binding site includes amino acids from both the ␣ and ␤ subunits (␣-Asp-230 and ␤-Asp-217). Thus, the catalytic site is not the exclusive function of the ␣ subunit but requires interaction of all of the subunits.
Yeast NAD-dependent isocitrate dehydrogenase is another extensively studied eukaryotic allosteric enzyme that is activated by a purine nucleotide (in this case, by AMP). This enzyme is composed of two types of distinct subunits, IDH1 and IDH2, of which the IDH2 subunit has been designated as the catalytic subunit, whereas the IDH1 subunit is the regula-tory subunit; however, there are some interactions between the subunits (39 -41). The yeast enzyme has recently been crystallized (42), and it is anticipated that the subunit interactions will be clarified when the structure is determined. The mammalian NAD-dependent isocitrate dehydrogenase has evolved a further level of complexity, with three distinguishable subunits. During the oxidative decarboxylation, NAD reacts with Mn 2ϩisocitrate; therefore, the active site must involve all three subunit types for optimal catalysis, whereas the allosteric activator ADP has thus far been found to be associated with only the ␤ and ␥ subunits.