Each Conserved Active Site Tyr in the Three Subunits of Human Isocitrate Dehydrogenase Has a Different Function*

The human NAD-dependent isocitrate dehydrogenase (IDH) is a heterotetrameric mitochondrial enzyme with 2α:1β:1γ subunit ratio. The three subunits share 40–52% identity in amino acid sequence and each includes a tyrosine in a comparable position: αY126, βY137, and γY135. To study the role of the corresponding tyrosines of each of the subunits of human NAD-IDH, the tyrosines were mutated (one subunit at a time) to Ser, Phe, or Glu. Enzymes were expressed with one mutant and two wild-type subunits. The results of characterization of the mutant enzymes suggest that βY137 is involved in NAD binding and allosteric activation by ADP. The αY126 is required for catalytic activity and likely acts as a general acid in the reaction. The γY135 is also required for catalytic activity and may be involved in proper folding of the enzyme. The corresponding tyrosines in the three dissimilar subunits of NAD-IDH thus have distinctive functions.

The human NAD-dependent isocitrate dehydrogenase (IDH) is a heterotetrameric mitochondrial enzyme with 2␣:1␤:1␥ subunit ratio. The three subunits share 40 -52% identity in amino acid sequence and each includes a tyrosine in a comparable position: ␣Y126, ␤Y137, and ␥Y135. To study the role of the corresponding tyrosines of each of the subunits of human NAD-IDH, the tyrosines were mutated (one subunit at a time) to Ser, Phe, or Glu. Enzymes were expressed with one mutant and two wildtype subunits. The results of characterization of the mutant enzymes suggest that ␤Y137 is involved in NAD binding and allosteric activation by ADP. The ␣Y126 is required for catalytic activity and likely acts as a general acid in the reaction. The ␥Y135 is also required for catalytic activity and may be involved in proper folding of the enzyme. The corresponding tyrosines in the three dissimilar subunits of NAD-IDH thus have distinctive functions.
Mammalian NAD-dependent isocitrate dehydrogenase (NAD-IDH) 2 is a mitochondrial tricarboxylic acid cycle enzyme that catalyzes the oxidative decarboxylation of isocitrate to ␣-ketoglutarate, while reducing NAD to NADH. It is a heterotetramer with three subunits in the ratio 2␣:1␤:1␥ (1), with molecular masses of ϳ37,000, 39,000, and 39,000 Da, respectively (2,3). The amino acid sequences of the ␤ and ␥ subunits of the human are 52.4% identical, whereas ␣ and ␤ subunits are only 40.4% and ␣ and ␥ subunits are only 41.6% identical. This enzyme is allosterically regulated by ADP, which decreases the K m for isocitrate by ϳ38-fold (4). NAD-IDH has two binding sites per tetramer for each ligand: isocitrate, NAD, Mn 2ϩ , and ADP (2,5), raising the question of the function of each of the subunits.
We have recently shown that homozygous mutations exclusively of the ␤ subunit of human NAD-IDH are a cause of Retinitis Pigmentosa, a hereditary degeneration of the retina that leads to blindness in patients (6). Characterization of these two types of mutant enzymes in lymphoblast cell extracts revealed a ϳ300-fold increase in K m,NAD and partial or complete loss of allosteric activation by ADP. The involvement of this enzyme in causing Retinitis Pigmentosa increases the importance of studying NAD-IDH in more detail.
High-resolution crystal structures are available for the homodimeric NADP-dependent IDH of pig and Escherichia coli (7)(8)(9)(10)(11). The individual subunits of human NAD-IDH are only about 25-34% identical in amino acid sequence to the E. coli NADP-IDH (Fig. 1A) and about 12-18% identical to the pig NADP-IDH. Although there is a low % identity among these enzymes, the isocitrate-binding site is well conserved, including E. coli Arg-119, Arg-153, Tyr-160, and Lys-230. A previous study of pig NADP-IDH showed that Tyr-140 interacts with the ␤-carboxylate of isocitrate and acts as a general acid in the reaction (8). The Tyr-160 mutants of E. coli NADP-IDH showed a decreased k cat (12,13) and the crystal structure revealed that the tyrosine is appropriately positioned to donate a proton to the carbanion following decarboxylation (11). Partial alignment of the amino acid sequences of the human NAD-IDH and pig and E. coli NAPD-IDH, shown in Fig. 1B, indicates that Tyr-160 of the E. coli and Y140 of the pig NADP-IDH align with Tyr-126 of the ␣ subunit, Tyr-137 of the ␤-subunit and Tyr-135 of the ␥ subunit of the human NAD-IDH. In this study, to distinguish the role of each subunit, we engineered and characterized recombinant human NAD-IDH mutants of the conserved Tyr, each of which has one mutated subunit and two wild-type subunits.
Site-directed Mutagenesis-The pHIDH␣␤ 2 ␥ vector carrying all three subunits of human NAD-IDH was used as a template for the mutagenesis (14, 15). The following forward prim- Only one subunit was mutated at a time using the QuikChange kit. For each mutant plasmid, the mutation was confirmed by DNA sequencing (Allen Laboratory, University of Delaware). Plasmids containing the desired mutation were transformed into E. coli XL-10 Gold cells.
Protein Expression and Purification-E. coli XL-10 Gold cells harboring wild-type or mutant NAD-IDH plasmids were grown overnight in 250 ml of LB medium with 0.1% ampicillin at 37°C and 220 rpm. Overnight culture (30 ml) was used to inoculate five 6-liter flasks each containing 2 liters fresh LB-ampicillin medium. The culture was allowed to grow until A 600 nm of 1.0 -1.5 was reached. NAD-IDH expression was then induced with 1 mM IPTG. The cells were grown overnight at 25°C and 150 rpm and collected by centrifuging the culture at 5000 ϫ g for 10 min. The collected cells were resuspended in 12 mM citrate-Tris, pH 7.2 containing 10% glycerol, 0.2 mM MnSO 4 , and 0.1 mM DTT, and frozen at Ϫ80°C.
NAD-IDH was purified as described by Soundar et al. (15,16). The frozen pellet from 10 liters of cell culture was thawed and lysed under conditions described previously. The cell lysate was centrifuged, and the supernatant was subjected to precipitation by 30 -50% ammonium sulfate at 4°C. The enzyme was then passed through a series of DE-52 (equilibrated at pH 7.7) and cellulose phosphate (at pH 5.7) columns. Purity was assessed by SDS-PAGE (17). If required, the enzyme was concentrated to ϳ1 ml in Amicon Ultra tubes (YM-10) and passed through the Ultrogel gel filtration column as described before (15,16). In all cases, fractions exhibiting NAD-IDH activity or two bands at ϳ40 kDa by SDS-PAGE were pooled and concentrated in Amicon Ultra (Ultracel 10K) tubes.
The ␣Y126E mutant was purified similarly except that the cellulose phosphate column was omitted from the purification procedure. Instead, the fractions pooled after the DE-52 column were concentrated to ϳ1 ml and loaded on the Ultrogel gel filtration column. Purity was assessed by SDS-PAGE (17).
Protein Expression and Western Blot for the ␥Y135F Mutant-Cells harvested from 2 liters of ␥Y135F mutant and wild-type NAD-IDH cultures were resuspended in 35 ml of 12 mM citrate-Tris, pH 7.2, 10% glycerol, 0.2 mM MnSO 4 , and 0.1 mM DTT. The cells were lysed and centrifuged to remove the pellet as described before. Various amounts of the supernatant were loaded on a SDS-PAGE gel and electroblotted on a nitrocellulose membrane, as described by Huang and Colman (18). The gel and the nitrocellulose membrane were soaked briefly in 10 mM CAPS, pH 11.0, 10% methanol (transfer buffer) and sandwiched between filter papers soaked in the same buffer. The transfer was conducted overnight in the Xcell II blot module from Invitrogen in the transfer buffer at 100 mA and 4°C.
The blot was fixed with 5% milk in 0.1% Tween in Tris-buffered saline, pH 7.6 (TBST), pH 7.6 for 1 h. The blot was then soaked in 1:200 dilution of the 1 o antibody (goat polyclonal anti-IDH3G IgG) in 5% milk for 1 h at room temperature. The blot was then washed with TBST and soaked in 1:2000 dilution of the horseradish peroxidase-conjugated 2 o antibody (antigoat IgG) for 20 min. The blot was washed with TBST and 1:1 solution of the ECL substrate (Pierce) was added to the blot. The blot was soaked in the solution for 5 min before viewing under the Fluorochem 8800 (Alpha Innotech). The area of each band was compared using the software ImageJ 1.40g.
N-terminal Sequencing-The N-terminal amino acid sequence was 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). The molar ratio was calculated as described by Bzymek and Colman (19).
Determination of Enzyme Activity-The standard conditions for assaying the enzyme activity were at 25°C in 1 ml of 33 mM Tris acetate, pH 7.2, 20 mM DL-isocitrate, 1 mM NAD, and 1 mM Mn 2ϩ . The specific activity is defined as mol of NADH produced/min/mg of enzyme, when assayed under standard conditions, with the formation of NADH being monitored by A 340 nm (15). The ␣Y126E mutant was assayed in 30 mM MES at pH 6.1 and compared with the wild type at the same pH. The enzyme concentrations were determined from the A 280 nm (E 280 nm 1% ϭ 6.45) (2). The kinetic parameters were determined by fitting the data in the Michaelis-Menten equation using the program SigmaPlot.
pH Profile-For determination of the pH-profile, enzyme activity of the wild-type and mutants was measured in 1 ml at various pH values using the following buffers: sodium acetate (pH 4.4 -5.8), MES (pH 5.4 -6.6), and PIPES (pH 6.2-7.4) and triethanolamine hydrochloride (pH 6.8 -7.6). All the buffers had a final concentration of 30 mM. The pH dependence of V max was determined for the wild-type and each mutant with 20 mM isocitrate, 1 mM NAD, and 1 mM Mn 2ϩ , unless otherwise stated.
Circular Dichroism-Secondary structure of the wild-type and mutant enzymes was determined using Aviv circular dichroism spectrometer (Model 400) in a 0.1 cm pathlength quartz cuvette. The enzymes were first dialyzed against 2 liters of 25 mM triethanolamine hydrochloride, pH 7.4, containing 10% glycerol and 0.2 mM MnSO 4 . The molar ellipticity was determined using average molecular weights and average number of residues per subunit, as described before (15).

RESULTS
Site-directed Mutagenesis, Protein Expression, and Purification-Plasmids containing the wild-type and mutant NAD-IDH were successfully expressed in E. coli XL-10 Gold cells. All the mutants of IDH, except the ␥Y135F and ␣Y126E, were purified using the established protocol described by Soundar et al. (15,16). As assessed by SDS-PAGE (Fig. 2), the wild-type and mutant NAD-IDH enzymes were pure, exhibiting two protein bands, with the upper band reflecting the ␤ and ␥ subunits (39 kDa) and the lower band representing the ␣ subunit (37 kDa) (1,15,20).
Under standard assay conditions (20 mM DL-isocitrate, 1 mM NAD, and 1 mM MnSO 4 , at pH 7.2) the enzymes showed the specific activities recorded in Table 1. The wild-type enzyme exhibited a specific activity of 26.3 mol of NADH produced/ min/mg of enzyme. The ␣Y126S, ␣Y126F, and ␣Y126E mutants had no activity under the standard conditions, suggesting that the Tyr in the ␣ subunit is essential for the catalytic function. In contrast, the ␤Y137S, ␤Y137F, and ␤Y137E showed lower specific activities of 3.1, 12.2, and 7.9 mol/min/mg, respectively, indicating that this Tyr in the ␤ subunit may contribute to, but is not required for, activity.
Western Blot on ␥Y135F Mutant-The ␥Y135F mutant could not be purified using the conventional methods followed for wild-type and the other IDH mutants. There was also no detect- able activity in its cell lysate (Table 1). To evaluate whether the ␥Y135F mutant enzyme was expressed and to compare the expression with that of the wild-type enzyme, both the wildtype and ␥Y135F mutant (treated in the same manner from growth to cell lysis) were blotted on a nitrocellulose paper, using similar protein concentrations for the wild-type and mutant enzymes. The Western blot was performed using a 1 o antibody against the C terminus of the ␥ subunit of human NAD-IDH. Fig. 3 shows that the areas and intensity of the bands seen on the blots for wild-type and ␥Y135F mutant are approximately the same (areas: 4172 and 4411; intensities: 1050600 and 1124805 respectively), indicating that the ␥Y135F mutant is expressed well, although it is inactive.
N-terminal Sequencing-The N-terminal amino acid sequence of the mature human NAD-IDH subunits are distinguishable. The first 10 amino acids of the ␣ subunit are TGGVQTVTLI; those of the ␤ subunit are ASRSQAEDVR; while those of the ␥ subunit are FSEQTIPPSA. Thus, a molar ratio of amino acids was used to calculate the ratio of the subunits using the first 10 amino acids of the sequence, as described in Bzymek and Colman (19). Table 2 shows that the molar ratio for the purified wild-type and mutant enzymes was ϳ2:1:1 for ␣, ␤, and ␥ subunit, respectively.

Determination of Secondary Structure by Circular Dichroism-
CD spectra of the mutant and wild-type enzymes were used to assess any change in the secondary structure caused by the mutations. Fig. 4 shows the CD spectra of the wild-type and mutant enzymes. The CD spectra are similar, reflecting mostly ␣-helical structure.
Kinetic Parameters for the Wild-type and ␤Y137 Mutant Enzymes at pH 7.2-To evaluate whether the mutation affected the enzyme affinity for its ligands, kinetic parameters at pH 7.2 were determined for all the ␤ subunit mutants for comparison with wild-type enzyme; the results are recorded in Table 3. The K m,Mn 2ϩ is similar in all the ␤Y137 mutants and the wild-type (Table 3, column 2). The K m,isocitrate for all the ␤ subunit mutants is similar to that of the wild-type when measured in the absence of ADP (Table 3, column 3).
In contrast, the mutants exhibit marked changes in the K m for isocitrate when measured in the presence of 1 mM ADP (Table 3, column 4). Whereas in wild-type enzyme the K m,isocitrate decreases 40-fold when ADP is present, it is almost unchanged in the ␤Y137S mutant and decreases only 6-fold in the ␤Y137F mutant. The effect of ADP in the ␤Y137E mutant is most similar to wild type, with the K m for isocitrate decreasing 38-fold when ADP is present. The mutants also exhibit a marked increase in K m,NAD (Table 3, column 5). K m,NAD for the ␤Y137S mutant is 10-fold higher, while that for the ␤Y137F mutant is only 2-fold higher than that for the wild-type. The ␤Y137E mutant is similar to the wild-type. The changes in the V max for the mutant enzymes are not remarkable (Table 3, column 6). For the ␤Y137S, ␤Y137F, and ␤Y137E mutants the V max values are 32, 47, and 30%, respectively, of that of the wild-type. These results suggest that Tyr-137 of the ␤ subunit is not critical for enzyme activity.
Effect of pH on the Observed V max -Previous studies with the homodimeric pig NADP-IDH demonstrated that, in contrast to the wild-type enzyme, the Y140E enzyme exhibits an increase in activity as the pH decreases (8). To determine whether the corresponding tyrosines in the heterodimeric human NAD-IDH behave similarly, the pH dependence of V max was measured for the wild-type and mutant enzymes. As shown in     Fig. 5A, the pH profiles of the wild-type and ␤Y137 NAD-IDH mutants are similar with apparent pK values (Ϯ S.E.) of 6.37 Ϯ 0.05, 6.33 Ϯ 0.10, 6.25 Ϯ 0.04 and 6.45 Ϯ 0.07 for the wild-type, ␤Y137S mutant, ␤Y137F mutant and ␤Y137E mutant, respectively. The ␣Y126S and ␣Y126F mutants are both inactive throughout the pH range. In contrast, the ␣Y126E mutant shows a rise in the observed V max as the pH decreases (Fig. 5B) with a pK of ϳ5.8 Ϯ 0.25. This result suggests that the glutamate replacing ␣Y126 (and presumably the natural Y126) must be protonated for the enzyme to be active and therefore the ␣Y126E mutant of human NAD-IDH is similar to the Y140E mutant of pig NADP-IDH (8).
Kinetic Parameters for the Wild-type and ␣Y126E Mutant at pH 6.1-Although the ␣Y126 mutants are inactive under standard conditions at pH 7.2, the ␣Y126E has some activity at pH 7.2 (0.06 mol/min/mg) in the presence of the higher 5 mM NAD. Because the pH profile of ␣Y126E shows a higher V max at pH 6.1, the kinetic parameters for the ␣Y126E mutant were determined at pH 6.1 and compared with those of the wild-type enzyme at the same pH. Table 4 shows that the K m,Mn 2ϩ is 30-fold higher in the ␣Y126E mutant as compared with the wild-type. The K m,isocitrate is 25-fold lower than that of the wildtype enzyme and, in the presence of 1 mM ADP, the K m,isocitrate is reduced almost equally in the wild-type (3.6-fold) and the ␣Y126E mutant (4.3-fold). K m,NAD for the ␣Y126E mutant is 29-fold higher than that of the wild-type. The V max of the wild-type at pH 6.1 is 14.4 mol/min/mg whereas that for the ␣Y126E mutant is only 1.03 mol/min/mg, suggesting a critical role for the residue in enzyme activity.

DISCUSSION
The ␣, ␤, and ␥ subunits of human NAD-IDH are clearly related, but each has a distinct amino acid sequence (Fig. 1A). However, the amino acid residues in the isocitrate-binding site of the pig and E. coli NADP-IDH, one of which is a conserved tyrosine, can be identified in all 3 subunits of the human NAD-IDH (Fig. 1B). This observation raises questions about the contribution of these residues, and hence of the subunits to enzyme function. Do they perform a similar or a different role, and is the active site shared between the subunits? The studies reported in this report, in which the tyrosine of one subunit at a time is replaced, provide insights into the role of the tyrosines in these three types of subunits.
In the case of the ␤Y137 mutants, their kinetic properties revealed similar affinity toward Mn 2ϩ and toward isocitrate, as that of wild type; these results suggest that ␤Y137 is not involved in binding either Mn 2ϩ or isocitrate. The relatively small changes in the V max also indicate that this residue is not essential for enzyme catalysis. The most striking changes in the ␤Y137 mutants are their substantial increase in the K m for NAD, and their marked decrease in the ability of ADP to lower the K m for isocitrate. The ␤Y137S and ␤Y137F mutants exhibit a higher K m for NAD as compared with that of the wild type, with ␤Y137S featuring the highest K m for NAD. The replacement of Tyr with Glu, however, causes almost no difference in the enzyme affinity toward NAD. Serine is considerably smaller than tyrosine and even though it has an aliphatic -OH, it is not surprising that Ser cannot interact with NAD as does Tyr at the same position. Phenylalanine is similar in size to tyrosine but lacks the phenolic -OH. Glutamate is similar in length to Tyr  and has the polar -COO instead of the -OH but lacks the aromatic group. The observation that Glu is the best substitute for Tyr at ␤Y137, indicates that the phenolic -OH or the -COO of Glu contributes to the affinity of the enzyme for NAD. The allosteric effect of ADP is completely lost in the ␤Y137S mutant enzyme and is partially lost in the ␤Y137F mutant, whereas the ␤Y137E enzyme exhibits about the same allosteric effect as that of the wild-type enzyme. The same order of the effectiveness of the replacement amino acids (i.e. Ser Ͻ Phe Ͻ Glu) suggests that the same properties determine the allosteric effect of ADP on the enzyme. Previous studies have shown that mutations at other positions in the ␤ subunit resulted in a higher K m for NAD and complete or partial loss of allosteric activation by ADP (6,15,16,19). None of these studies showed a major effect on the V max of the enzyme. Hence, we conclude that the ␤ subunit is important for NAD binding and for the allosteric effect of ADP, but it is not critical for enzyme activity.
In the case of the ␣ subunit mutants, the ␣Y126S and ␣Y126F are inactive, and the ␣Y126E is inactive at pH 7.2 implying that Tyr-126 is required for enzyme activity. It is striking that the ␣Y126E mutant is active at low pH indicating that, when protonated, Glu can partially substitute for Tyr. The ␣Y126 could be acting as a general acid that protonates the enolate after decarboxylation to form ␣-ketoglutarate; this role has been proposed for Tyr-140 of the pig NADP-IDH (8). Previous studies have shown that replacement of different amino acids in the ␣ subunit of human NAD-IDH (Arg-88, Asp-230, and Asp-234) yield inactive enzyme (15,16). These reports, together with the results presented in this study, indicate that alpha is the catalytic subunit.
In the case of the ␥ subunit mutant, although there is no activity in the cell lysate of the mutant, it is clear that the protein is expressed. The lack of activity can be interpreted to indicate either that ␥Y135 is essential for activity or that this tyrosine is important for the correct folding of the enzyme. Since the enzyme did not bind normally to the ion exchange resins generally used for the purification of NAD-IDH, it appears that there is a major change in the exposed ionic groups of the enzyme. Therefore it seems likely that ␥Y135 plays a structural role (or a role in the correct folding of the enzyme). Other amino acids of the ␥ subunit have been implicated in isocitrate, Mn 2ϩ , and NAD binding, as well as the allosteric effect of ADP (15,19).
Earlier studies on the human and pig NAD-IDH demonstrated that there are only two binding sites per tetramer for each ligand (2,5), suggesting that the subunits may have specialized functions. The ␣, ␤, and ␥ subunits are clearly related to each other but, as the subunits deviated in amino acid sequence, their roles in the enzyme became differentiated. The ␣ subunit is required for catalytic reaction. The ␤ subunit is important for optimal binding of the coenzyme and for the allosteric effect of ADP. The ␥ subunit may contribute to optimal affinity for isocitrate, Mn 2ϩ and NAD and to the allosteric effect of ADP, as well as promoting proper folding of the enzyme. This study shows that the NAD and Mn 2ϩ sites are shared between the ␣ and ␤ subunits. Earlier studies have indicated that the active sites are shared between ␣ and ␤ subunits and between ␣ and ␥ subunits (16,19). Thus, although catalysis by the simple NADP-dependent isocitrate dehydrogenase is carried out by a single subunit type, the more complex allosteric NAD-specific isocitrate dehydrogenase requires three distinct subunits. a Experiments were carried out in 1 ml of 30 mM MES at pH 6.1, 40 mM isocitrate, 5 mM NAD, 1 mM MnSO 4 resulting in a higher V max at pH 6.1 compared to that seen in the pH profile for the wild-type enzyme. b Experiments were carried out in 1 ml of 30 mM MES at pH 6.1, 20 mM isocitrate, 5 mM NAD, 1 mM MnSO 4 .