Identification by Mutagenesis of Arginines in the Substrate Binding Site of the Porcine NADP-dependent Isocitrate Dehydrogenase*

Pig heart mitochondrial NADP-dependent isocitrate dehydrogenase is the most extensively studied among the mammalian isocitrate dehydrogenases. The crystal structure of Escherichia coli isocitrate dehydrogenase and sequence alignment of porcine with E. coli isocitrate dehydrogenase suggests that the porcine Arg 101 , Arg 110 , Arg 120 , and Arg 133 are candidates for roles in substrate binding. The four arginines were separately mutated to glutamine using a polymerase chain reaction method. Wild type and mutant enzymes were each expressed in E. coli , isolated as maltose binding fusion proteins, then cleaved with thrombin, and purified to yield homogeneous porcine isocitrate dehydrogenase. The R120Q mutant has a specific activity, as well as K m values for isocitrate, Mn 2 1 , and NADP 1 similar to wild type enzyme, indicating that Arg 120 is not needed for function. The specific activities of R101Q, R110Q, and R133Q are 1.73, 1.30, and 19.7 m mols/min/mg, respec-tively, as compared with 39.6 units/mg for wild type enzyme. The R110Q and R133Q enzymes exhibit K m val- ues for isocitrate that are increased more than 400- and 165-fold, respectively, as compared with wild type. The K m values for Mn 2 1 , but not for NADP 1 , are also elevated indicating that binding of the metal-isocitrate complex is impaired in these mutants. It is proposed that the positive charges of Arg 110 and Arg 133 normally strengthen the binding of the negatively charged isocitrate by electrostatic attraction. The R101Q mutant Wild Type and Mutant Enzymes— The pH de- pendence of V max was measured using the following buffers: pH 5.0–5.8 (30 m M sodium acetate), pH 5.8–7.4 (30 m M imidazole chloride), and pH 6.8–8.0 (30 m M triethanolamine chloride) buffers containing 2 m M MnSO 4 and 0.1 m M NADP 1 at 25 °C. The buffers were all 30 m M in anion concentration. To ascertain the concentration of isocitrate re-quired to saturate the R110Q and R133Q enzymes, the V max and K m values for isocitrate were determined at the lowest pH values using isocitrate solutions adjusted to the corresponding pH. The pH rate profiles were measured using 4 m M isocitrate for the wild type and R120Q enzymes, 8 m M isocitrate for R101Q enzyme, and 20 m M isoci- trate for R110Q and R133Q enzymes. Under these conditions, all the enzymes were saturated with every substrate. A stock solution of 0.25 M DL -isocitrate in water was prepared, and its pH was adjusted to 5.0 before including it in the assay mixture. The final pH of the assayed solution was measured at each pH. Spatial restraints, expressed as probability density functions from the reference protein, were then optimized. The model with the lowest value of objective function ( F , molecular probability density functions violation) fits best to the reference structure. Out of the models generated from different alignments, a model from the best alignment was chosen comparing the lowest F values.

The mitochondrial NADP-specific pig heart isocitrate dehydrogenase (EC 1.1.1.42) catalyzes the divalent metal ion-dependent oxidative decarboxylation of isocitrate to ␣-ketoglutarate, and it is considered that the metal-tribasic isocitrate complex is the preferred substrate (1). The enzyme is a homodimer (2,3), with a subunit mass of 46,600 Da consisting of 413 amino acids of determined sequence (4). A 13 C-NMR study using specifically enriched isocitrate demonstrated that all three carboxyls of the substrate remain fully ionized from pH 5.5 to 7.5 when bound to the enzyme, although the carboxylates of free isocitrate become protonated over this pH range (5). This result could be due to the presence of positively charged groups in the region of the substrate binding site. The first evidence of the importance of arginines in the function of NADP-dependent isocitrate dehydrogenase came from the inactivation of the pig heart enzyme by 2,3-butanedione (6). A maximum of four arginines were implicated in catalytic activity and, because isocitrate markedly decreased the inactivation rate, it was suggested that at least some of these residues were at or near the isocitrate binding site.
Whereas no crystal structure of a mammalian isocitrate dehydrogenase has yet been determined, the structure of the Escherichia coli NADP-dependent isocitrate dehydrogenase is known (7)(8)(9). In this bacterial enzyme, Arg 119 , Arg 129 , and Arg 153 interact with the carboxylates of isocitrate, and Ser 113 is also close to the substrate (8). Alignment of the amino acid sequence of E. coli isocitrate dehydrogenase with that of the porcine enzyme reveals only about 12% identity (plus about 34% similarity). However, there are conserved amino acids among those known to interact with the substrate in the E. coli enzyme. Fig. 1 shows the amino acid sequence alignment in this region between three eukaryotic enzymes (pig, rat, and yeast) and the E. coli enzyme. In the porcine enzyme, Arg 110 and Arg 133 line up, respectively, with the E. coli Arg 119 and Arg 153 . Arg 101 of the pig isocitrate dehydrogenase is aligned with E. coli's Arg 112 , which is close to Ser 113 . There is some uncertainty in the alignment in the middle of the region illustrated; however, Arg 120 is the only such residue in this portion of the porcine enzyme and may be comparable to Arg 132 or Arg 129 of the E. coli enzyme.
In this paper, to test the roles of the positively charged Arg 101 , Arg 110 , Arg 120 , and Arg 133 as determinants of isocitrate affinity and catalysis by porcine dehydrogenase, each of these arginines was mutated to the neutral glutamine. Here, we report the procedures for mutagenesis, expression, and purification of the mutant enzymes, as well as the characterization of the physical and kinetic properties of these enzymes. A preliminary version of this work has been presented (10).

EXPERIMENTAL PROCEDURES
Materials-Oligonucleotides were synthesized by the Operon Technologies, Inc. (Alameda, CA). Plasmid pMal-c2, E. coli strain TB1, T4 DNA ligase, and amylose resin were obtained from New England Biolabs (Beverly, MA). The restriction enzymes, BamHI and Bpu1102I, as well as proteinase K were purchased from Life Technologies, Inc. Pfu DNA polymerase was obtained from Stratagene (La Jolla, CA). Dyeconjugated DNA primers were synthesized by LI-COR, Inc. ( Site-directed Mutagenesis-A 1.2-kbp 1 cDNA-encoding pig heart mitochondrial NADP-specific isocitrate dehydrogenase was previously cloned into a bacterial expression vector pMal-c2 (New England Biolabs), which yields the enzyme as a fusion protein with maltose-binding protein (11). This vector was later reconstructed with a thrombin cleavage site, as described previously, and was used to express the wild type enzyme (12). This vector was used in the present study for oligonucleotide-directed mutagenesis using a megaprimer PCR method (13). The mutagenesis was carried out in two PCR steps: the first step amplifies a segment of template DNA using a mutagenic oligonucleotide and a reverse end primer encoded with a BamHI restriction site. The PCR product was then incorporated in the second PCR step to produce a 1.66-kbp DNA in conjunction with a third forward end primer encoded with a Bpu1102I restriction site.
The Bpu1102I site is located in the malE gene 379 base pairs upstream of IDP1 DNA, and the BamHI site is encoded at the terminal of the IDP1 DNA. The oligonucleotides used to generate mutant enzymes were: R101Q (5Ј-CCAATGGAACCATCCAGAACATCCTCGGGG), R110Q (5Ј-GGGACTGTTTTCCAGGAGCCCATCATCTGC), R120Q (5Ј-GCAAGAACATCCCACAGCTTGTGCCTGGC), and R133Q (5Ј-GCCCA-TCACCATCGGCCAGCACGCTCACGGCGACC). The underlined codons are those mutated to glutamine. The end primers, included to flank the DNA region for amplification, were BpuF (5Ј-GGCGTGCTGAGCG-CAGGTATTAACGCCGCCAGTCCG) and BamHI R (5Ј-CTAGAGGAT-CCTTACTACTGCCGGCCCAGAGCTCTGTC). PCR was carried out in a Robocycler Gradient 96 (Stratagene, La Jolla, CA) in a 50-l reaction mixture containing 50 -100 ng of 7.88-kbp template DNA, 500 ng of each oligonucleotide, 0.25 mM dNTPs, and 2.5 units of cloned Pfu DNA polymerase. Denaturation, annealing, and polymerization were performed at 95°C for 1 min, 55°C for 1 min, and 72°C for 2.5 min, respectively, for 30 cycles. Each reaction was initiated with an additional denaturation at 95°C for 1 min and terminated at 72°C for 10 min after the above 30 cycles of reaction were completed.
The first PCR step DNA product was purified by agarose gel electrophoresis and extracted using the Ultraclean 15 DNA purification system. Before the overhangs are removed from the 1.66-kbp second step PCR product, 200 l of the PCR pool was treated with 4 l of 10% SDS and 6 l containing 6 g of proteinase K enzyme at 55°C for 30 min. DNA was then extracted twice using phenol:chloroform:isoamyl alcohol (25:24:1 v/v) and washed in chloroform once. DNA was precipitated in cold ethanol with 3 M sodium acetate buffer, pH 5.2, at Ϫ80°C for 15 min. The DNA was separated by centrifugation, and the pellet was washed in 70% ethanol and then dissolved in 10 mM Tris chloride, pH 7.5, buffer. To remove the overhangs, the DNA in 25 l was treated with 20 units of each of BamHI and Bpu1102I at 37°C for 5 h. The DNA was then gel-purified and ligated. The ligation mixture containing a molar ratio of 2:1 of insert DNA to 6.22-kbp vector DNA was pretreated at 65°C for 5 min and then incubated at 16°C for 15 h with 400 units of T4 DNA ligase. Vector DNA was previously gel-purified after treating 7.88 kbp of DNA with BamHI/Bpu1102I and calf intestinal alkaline phosphatase enzymes. The ligation mixture was transformed into E. coli TB1 competent cells and grown on LB plates containing ampicillin (0.1 mg/ml). Colonies were subcultured, and the plasmid DNA was isolated using the Promega DNA purification system. To ensure proper ligation of insert and vector, the plasmid was digested with BamHI/ Bpu1102I enzymes, and the release of 1.66 kbp of DNA was observed. The desired mutation and fidelity of PCR amplification was confirmed by sequence analysis using the LI-COR sequencing protocol (LI-COR 4200 Long Readir Sequencer) at the University of Delaware's Cell Biology Core Facility.
Expression and Purification of Wild Type and Mutant Enzymes-The procedure previously described for expression and purification of wild type enzyme (12) was adopted here for mutant enzyme preparation, with a few modifications. E. coli cells transformed with mutant pMal-cIDP1 were grown at 37°C in 8 liters of LB broth (4 ϫ 2 liter) containing ampicillin (0.1 mg/ml) to A 600 nm ϭ 0.4 cell density. Isopropyl thio-␤-Dgalactopyranoside was added to 0.3 mM concentration to induce expression, and the incubation was continued for 20 -24 h at 25°C. Cells were collected by centrifugation at 5000 ϫ g for 5 min and suspended in 800 ml of cold 0.02 M triethanolamine chloride buffer, pH 7.4, containing 10% glycerol, 0.2 M Na 2 SO 4 , and 2 mM MnSO 4 (Buffer A). Cells were frozen in 2 ϫ 400 ml volume and lysed by sonication at 20 KHz and 475 W for 10 min with a large probe in a semithawed condition in cold water. The lysate was centrifuged at 16,000 ϫ g for 10 min, and the clear supernatant was collected after removing the debris. The fusion protein was separated from the 800 ml of crude extract by passing it through 75 ml of amylose resin in 2 h in a sintered glass funnel at 4°C. The resin had previously been equilibrated in Buffer A. This step removes the endogenous E. coli isocitrate dehydrogenase, which is not bound to the amylose resin (12). After washing with 2 liters of Buffer A, the resin was transferred into a column, and the fusion protein was eluted with 10 mM maltose as described (12).
The enzyme was cleaved from the maltose-binding protein by incubating the fusion protein with human or bovine plasma thrombin (41 units/mg fusion protein) at 25°C for 48 h in the presence of 4 mM isocitrate and 2 mM MnSO 4 to stabilize the cleaved enzyme. The digest was dialyzed against 0.018 M triethanolamine chloride buffer, pH 7.1, containing 10% glycerol, 0.06 M Na 2 SO 4 , and 1 mM MnSO 4 (Buffer B) and applied to a Matrex gel Red-A resin equilibrated with Buffer B to separate pig heart isocitrate dehydrogenase from the maltose-binding protein and thrombin. The enzyme was eluted by 0.018 M triethanolamine chloride buffer, pH 8.0, containing 10% glycerol, and 0.4 M Na 2 SO 4 (12). The fractions that contained high specific activity were pooled and reapplied to the amylose resin as described (12) to remove any uncleaved fusion protein. The final gel filtration step described previously (12) has been eliminated, as this modified procedure yields more than 95% pure enzyme after the second step of purification.
SDS-Polyacrylamide Gel Electrophoresis-Proteins from uninduced and isopropyl thio-␤-D-galactopyranoside-induced E. coli TB1 cells and protein aliquots at different steps of purification were analyzed in a 15% polyacrylamide slab gel (Laemmli system) containing 0.1% SDS in a discontinuous pH electrophoresis system (14). Preparation of stacking and resolving gel, electrophoresis running conditions, protein fixing, staining, and destaining solutions were followed as described (15).
Determination of Amino Acid Sequence-The N-terminal amino acid sequences of wild type and mutant enzymes (1 nmol) were determined using an Applied Biosystem gas-phase protein sequenator (Model 470A) equipped with an on-line phenylthiohydantoin analyzer (Model 120) and computer (Model 900A).
Fast Performance Liquid Chromatography of the Wild Type and Mutant Enzymes-Gel filtration was carried out using a Amersham Pharmacia Biotech fast performance liquid chromatography system equipped with a Superose-12 column (1 ϫ 30 cm, Amersham Pharmacia Biotech). Wild type and mutant enzymes (0.1-1.0 mg) in 0.4 ml were applied to the column, which was previously equilibrated with 0.1 M triethanolamine chloride buffer, pH 7.7, containing 10% glycerol and 0.3 M Na 2 SO 4 . The proteins were eluted at a flow rate of 0.3 ml/min, and 0.6-ml fractions were collected. Standard proteins, applied and eluted under the same conditions used for isocitrate dehydrogenase, were catalase (232 kDa), aldolase (158 kDa), albumin (67 kDa), ovalbumin (43 kDa), and chymotrypsinogen A (25 kDa). The column had a total volume (V t ) of 20.1 ml and a void volume (V o ) of 6.9 ml, as determined by L-tyrosine and blue dextran 2000, respectively. The protein concentrations in the fractions were measured at A 280 nm from which the elution volume (V e ) for each protein was determined.
Circular Dichroism of the Wild Type and Mutant Enzymes-Circular dichroism was conducted using a Jasco model J-710 spectropolarimeter. The asterisks indicate that the amino acids at that position are identical, and the "." indicates that the amino acids are similar. The Arg residues of the porcine enzyme (numbered above the sequences) were mutated to Gln residues. The corresponding Arg residues of the E. coli enzyme are numbered below the sequences. Clustal W was used for sequence alignment.
Measurement of the ellipticity as a function of wavelength between 250 and 200 nm were made using a 0.1-cm path length quartz cell. Wild type and mutant proteins (0.15 mg/ml) in 0.025 M triethanolamine chloride buffer, pH 7.7, containing 10% glycerol and 0.075 M Na 2 SO 4 were used for circular dichroism measurements. The mean residue molar ellipticity [] (deg cm 2 mol Ϫ1 ) was calculated from the equation [] ϭ /(10nCl), where is the measured ellipticity in millidegrees, C is the molar concentration of protein, l is the path length of the cell in cm and n is the number of residues/subunit of enzyme (n ϭ 413 for pig heart NADP-dependent isocitrate dehydrogenase).
Kinetic Studies of Wild Type and Mutant Enzymes-Enzyme assays were performed at 25°C by monitoring the time-dependent reduction of NADP ϩ from the UV absorbance of NADPH at 340 nm. The standard assay solution (1 ml) was 30 mM triethanolamine chloride buffer (pH 7.4), 0.1 mM NADP ϩ , 4 mM DL-isocitrate, and 2 mM MnSO 4 ; and the specific activity is defined as the mol of NADPH produced/min/mg protein under these conditions. A subunit M r of 46,600 (4) was used to calculate the concentration of enzyme subunits and the protein concentration was determined from E 280 nm 1% ϭ 10.8 (16). For K m determinations, the concentration of either coenzyme, isocitrate or Mn 2ϩ , was varied, and the other substrates were maintained at the standard concentrations. For the isocitrate K m determination, concentrations of isocitrate varied between 0.005 and 20.0 mM; for the Mn 2ϩ K m measurement, 0.1 M to 10 mM MnSO 4 ; and for the NADP ϩ K m determination, 5 M to 2.5 mM NADP ϩ were used. The K m and V max values were calculated from the Lineweaver-Burk plots.
pH-Rate Profiles of Wild Type and Mutant Enzymes-The pH dependence of V max was measured using the following buffers: pH 5.0 -5.8 (30 mM sodium acetate), pH 5.8 -7.4 (30 mM imidazole chloride), and pH 6.8 -8.0 (30 mM triethanolamine chloride) buffers containing 2 mM MnSO 4 and 0.1 mM NADP ϩ at 25°C. The buffers were all 30 mM in anion concentration. To ascertain the concentration of isocitrate required to saturate the R110Q and R133Q enzymes, the V max and K m values for isocitrate were determined at the lowest pH values using isocitrate solutions adjusted to the corresponding pH. The pH rate profiles were measured using 4 mM isocitrate for the wild type and R120Q enzymes, 8 mM isocitrate for R101Q enzyme, and 20 mM isocitrate for R110Q and R133Q enzymes. Under these conditions, all the enzymes were saturated with every substrate. A stock solution of 0.25 M DL-isocitrate in water was prepared, and its pH was adjusted to 5.0 before including it in the assay mixture. The final pH of the assayed solution was measured at each pH.
Molecular Modeling of Pig Heart NADP-dependent Isocitrate Dehydrogenase-As the crystal structure of the mammalian enzyme has not yet been determined, modeling studies were carried out on an SGI workstation using Insight II (Biosym/MSI, San Diego, CA). The sequence alignment was performed using Clustal W for sequences of NADP-dependent isocitrate dehydrogenase of pig heart (mitochondrial), E. coli, yeast (mitochondrial), and rat (cytosolic), which is almost the same as we published earlier in Fig. 3 of Ref. 4. The crystal structure of E. coli NADP-dependent isocitrate dehydrogenase was used as the reference structure (8) for the Modeler program (a module under homology), which calculated energy-minimized models for the monomeric structure of the pig heart NADP-dependent isocitrate dehydrogenase. Spatial restraints, expressed as probability density functions from the reference protein, were then optimized. The model with the lowest value of objective function (F, molecular probability density functions violation) fits best to the reference structure. Out of the models generated from different alignments, a model from the best alignment was chosen by comparing the lowest F values.

Expression and Purification of Wild Type and Mutant Enzymes-
The porcine NADP-dependent isocitrate dehydrogenase, with glutamine substituted for arginine at each of the positions 101, 110, 120, and 133, were generated using expression vector pMALcIDP1 by a megaprimer PCR method (3). The first and second PCR runs yielded approximately 0.86 and 1.66 kbp DNAs, respectively. Although the 1.66-kbp DNA was incubated with the BamHI and Bpu1102I restriction enzymes to remove the overhang and was gel purified, it did not initially ligate with the 6.22-kbp vector DNA. However, the ligation was successful if the 1.66 kbp DNA had been treated with proteinase K and SDS prior to incubation with the restriction enzymes. A small number of colonies resulted from self-ligated vector DNA on transformation, even though the vector DNA had been treated with alkaline phosphatase. Such colonies were avoided, and colonies that yielded plasmid DNA, which produced a 1.66-kbp insert DNA upon BamHI/Bpu1102I digestion, were selected for expression. Nucleotide sequence analysis showed that the mutations had been introduced in all of the mutant plasmid DNAs and that the sequences of the PCRamplified DNA were correct.
The cells were grown in 8 liters of culture medium, and the isopropyl thio-␤-D-galactopyranoside-induced expression time was increased from 4 h (12) to 20 -24 h to improve the level of expression of the fusion protein. The sonication of cells in one pulse for 10 min yielded more soluble fusion protein than obtained previously (12). The time for incubation of the fusion protein with thrombin was also increased from 18 to 48 h. Under these conditions only trace amounts of uncleaved fusion protein remained. This procedure yielded approximately 15 mg of pure protein as compared with the 1.2 mg obtained by the procedure reported previously (12). Fig. 2 shows that the wild type and mutant enzymes eluted after the second amylose column were homogeneous, as evaluated by polyacrylamide gels containing sodium dodecyl sulfate. The subunit molecular size of the four mutant proteins is similar to that of wild type enzyme. 2 The sequences of the porcine and E. coli isocitrate dehydrogenases differ in nine of the first ten amino acids. The N-terminal sequencing of these purified wild type and mutant porcine enzymes indicate that they are not contaminated with the E. coli enzyme.
Gel Filtration of the Wild Type and Mutant Enzymes-To test the molecular size of the native wild type and mutant enzymes, they were subjected to gel filtration on a Superose-12 column, equilibrated with 0.1 M triethanolamine chloride, pH 7.7, containing 10% glycerol and 0.3 M Na 2 SO 4 . The wild type and mutant enzymes all elute between aldolase and bovine serum albumin with an average elution volume of 10.8 ml. The results indicate that wild type and mutant isocitrate dehydrogenases exist as dimers in solution. The molecular weight of the isocitrate dehydrogenases was determined using a plot of 2 Although it appears that the mobility of the R120Q mutant is slightly retarded relative to the other samples, this difference is not considered significant because it was observed only in this gel and is probably due to uneven loading of this sample. Circular Dichroism Spectra of the Wild Type and Mutant Enzymes-Circular dichroism spectra of the expressed enzymes were measured to ascertain whether these mutations cause conformational change in the enzyme. The spectra of all mutants are similar to that of the wild type enzyme; all exhibit minima at 208 and 225 nm. These results indicate that the mutations do not cause appreciable change in the secondary structure of these isocitrate dehydrogenases. Table I shows the specific activities of wild type and mutant enzymes under standard conditions, as described under "Experimental Procedures." The purified wild type enzyme expressed in E. coli cells exhibits a specific activity, which is similar to that of the enzyme isolated previously from pig heart mitochondria in our laboratory (15). The R101Q and R110Q mutants show marked decreases in specific activity, whereas the R133Q mutant is 50% as active as wild type, indicating that the functions of these three mutant enzymes are impaired. The mutant R120Q has a specific activity similar to that of wild type.

Kinetic Parameters of the Wild Type and Mutant Enzymes-
All four mutants exhibit values of K m for the coenzyme, NADP ϩ , similar to that of wild type enzyme, as shown in Table  I. These results indicate that these arginines are not involved in coenzyme binding. The R120Q mutant also has a K m for Mn 2ϩ similar to that of wild type indicating that Arg 120 is not involved in Mn 2ϩ binding. In contrast, the R101Q, R110Q, and R133Q mutants show K m values for Mn 2ϩ that are increased 300-, 100-, and 1200-fold, respectively, consistent with a role for these arginines in isocitrate-Mn 2ϩ binding. Table II shows the kinetic constants for the substrate, isocitrate, wild type, and mutant enzymes. The R120Q mutant has V max and K m values similar to those of wild type enzyme, indicating that Arg 120 is not involved in the function of the enzyme. The R101Q mutant exhibits a k cat value that is only 4% that of wild type, and it has a K m for isocitrate that is increased 18-fold. The R110Q and R133Q mutants exhibit k cat values that are, respectively, 10 and 53% that of wild type. However, they have K m values for isocitrate that are increased more than 400-and 165-fold, respectively, as compared with wild type. These data indicate that the affinity of the enzyme for isocitrate is greatly impaired by replacing Arg 110 and Arg 133 . The last column in Table II shows the catalytic efficiency (k cat /K m ) of mutant enzymes, compared with that of wild type enzyme. Whereas mutant R120Q is similar to that of wild type, R101Q and R133Q mutants are 1000-fold poorer in effi-ciency compared with that of wild type. The R110Q mutant exhibits a 10,000-fold decrease in efficiency compared with wild type, because it has both a high value for K m for isocitrate and a lower V max .
Dependence of Velocity on pH for Wild Type and Mutant Enzymes-The pH dependence of V max was determined for the recombinant wild type and mutant enzymes from about pH 5.2 to 7.8, for comparison with the pH rate profile previously reported for the enzyme isolated from pig hearts (17)(18)(19). The enzyme has been shown to be stable over this pH range. Although the K m values for isocitrate are relatively high for some of the mutant enzymes, we have ascertained that each enzyme is saturated with respect to isocitrate over the entire pH range at the concentration of the substrate chosen for that enzyme (4 mM isocitrate for the wild type and R120Q enzymes, 8 mM isocitrate for R101Q enzyme, and 20 mM isocitrate for R110Q and R133Q enzymes).
The dependence of observed V max obs on pH was analyzed in accordance with the equation, where V max obs is the maximum velocity observed at a given pH, V max int is the intrinsic pH independent maximum velocity, and K aes is the dissociation constant for the enzyme-substrate complex. The pK aes values for wild type and mutant enzymes are summarized in Table III. To allow comparisons of the shapes of the curves for the various enzymes, the activity of each mutant enzyme at each pH was expressed as a fraction of its own intrinsic maximum velocity (V max obs /V max int ). Fig. 3 shows the plots of (V max obs /V max int ) for both wild type and mutant enzymes. The points in the figure are experimental data, whereas the lines represent fits to Equation 1, for V max obs /V max int using the pK aes values in Table III. The pH-V max profile for the mutant R120Q is similar to that of wild type enzyme. For mutant R101Q, the pK aes is slightly lower than that of wild type enzyme. In contrast, both mutants R110Q and R133Q exhibit striking increases in pK aes to 6.43 and 7.40, respectively, as compared with 5.54 for wild type enzyme. These results suggest that the positive charges of Arg 110 and Arg 133 normally lower the pK of the nearby catalytic base to facilitate its ionization.

DISCUSSION
The dimeric bacterial and eukaryotic NADP-dependent isocitrate dehydrogenases carry out the same catalytic reaction, with similar kinetic parameters (12,20). They also have about the same number of amino acids; yet the level of amino acid conservation is low (about 12% identity plus 34% similarity overall). Based on the sequence alignment of the mammalian and E. coli isocitrate dehydrogenases, consideration of the relatively few amino acids, which are conserved in all species, and analysis of the crystal structures of the E. coli enzyme, Arg 101 , Arg 110 , Arg 120 , and Arg 133 were chosen for testing by mutagenesis as candidates for interaction with the carboxylates for isocitrate. Neutral glutamine was selected to replace arginine because it lacks the positive charge of the wild type amino acid but is polar and is similar in size to arginine. It is clearly desirable to change a single variable at a time in designing mutants to minimize generalized effects on the overall structure of the enzyme. In contrast, Jennings et al. (21), in constructing mutants of the rat cytoplasmic NADP-dependent isocitrate dehydrogenase, substituted the negatively charged glutamate for arginine and did not evaluate the effect of this substitution on the structure of the enzyme. If the arginine normally participates in electrostatic attraction of a carboxylate of isocitrate, then replacement of Arg by glutamate could introduce a local, intense electrostatic repulsion between the enzyme and substrate, which would make it difficult to measure the altered kinetic parameters of the mutant enzymes. In fact, half of the Arg to Glu mutant rat cytoplasmic isocitrate dehydrogenases exhibited such low activity, that they were not characterized kinetically (21). In contrast, all of our arginine to glutamine mutants retained sufficient activity to allow measurement of a variety of kinetic parameters. These studies show that only the R120Q mutant is similar to the wild type enzyme in function, whereas the other mutants show impaired function either in isocitrate binding or in catalysis. Wild type and mutant pig mitochondrial isocitrate dehydrogenases were found, in the present study, to elute at comparable positions on gel filtration fast protein liquid chromatography, indicating that they all have the normal dimeric structure. Furthermore, the circular dichroism spectra of the wild type and mutant enzymes are nearly superimposable, showing that the mutations do not cause appreciable change in the secondary structure of the enzymes.
The R120Q mutant enzyme has a specific activity, K m values for isocitrate, Mn 2ϩ , and NADP ϩ and pH-V max profile similar to those of wild type enzyme. These characteristics indicate that Arg 120 is not involved in either catalysis or substrate binding. The resemblance between the sequences of the mammalian and the E. coli isocitrate dehydrogenase is weak between Asp 123 and Met 149 of the E. coli enzyme (Fig. 1). Apparently, the porcine enzyme does not have an arginine equivalent to Arg 129 of the E. coli enzyme that forms hydrogen bonds and/or salt bridges to the ␤-carboxylate of isocitrate (8).
Both the R110Q and R133Q mutant enzymes feature striking increases in K m values for isocitrate (400-and 165-fold, respectively) and elevated K m values for Mn 2ϩ (100-and 1200fold, respectively). These results indicate that positive charges of Arg 110 and Arg 133 normally stabilize the binding of Mnisocitrate by electrostatic interaction with the negatively charged isocitrate. Conversely, elimination of the positive charge by substitution of the neutral glutamine for arginine weakens the affinity of isocitrate for the enzyme. In addition, the R110Q and R133Q mutant enzymes exhibit upward shifts of their pH-V max profiles to yield pK aes values of 6.4 and 7.4, respectively, as compared with 5.5 for the wild type enzyme. Previous studies (17)(18)(19) have reported on the dependence of V max on pH for the pig heart NADP-dependent isocitrate dehydrogenase. This titration curve could best be described by a requirement for catalysis of the unprotonated form of an enzymatic functional group of pK about 5.6. Because the pK increased when the catalytic activity was measured in 20% ethanol, a solvent of lower dielectric constant (17), but did not change when the temperature was varied from 10 to 30°C (19), the pK was attributed to ionization of a carboxyl group in the enzyme-substrate complex. Because 13 C-NMR evidence indicated that enzyme-bound isocitrate remained fully ionized over this pH range (5), the pK of 5.6 can be attributed to an enzymatic carboxyl group. The isocitrate dehydrogenase reaction is thought to be initiated by removal of a proton from the -OH of isocitrate prior to the transfer of a hydride to NADP ϩ (1, 8); a general base on the enzyme has been postulated to catalyze this proton transfer reaction. A recent mutagenesis study in which aspartic acids of the pig heart enzyme were replaced by asparagines suggests that either Asp 275 or Asp 279 functions as the catalytic base because the D275N and D279N mutants exhibited large changes in pK aes (22). Our present results on the arginine to glutamine substitutions at positions 110 and 133 thus imply that the positive charges of Arg 110 and Arg 133 normally act to lower the pK of Asp 275 /Asp 279 ; when the positively charged arginines are substituted by neutral glutamines, the nearby Asp 275 /Asp 279 becomes a weaker acid, as indicated by the increase in pK aes of the enzymatic carboxyl group in the a These values are higher than the specific activities given in Table I because the V max values given here were obtained by extrapolation to saturating concentrations of isocitrate for each mutant, whereas the "specific activity" given in Table 1 was measured under the "standard conditions," which includes an isocitrate concentration of 4 mM. For wild type enzyme, 4 mM isocitrate is a saturating concentration, but that is not the case for the R110Q and R133Q mutants because of their high K m values for isocitrate.  3. pH-V max profiles for the wild type and mutant isocitrate dehydrogenases. For each enzyme, every pH is divided by its own intrinsic pH-independent maximum velocity so that the shapes of the curves can readily be compared. The profiles for the wild type (Ⅺ), R101Q (OE), R110Q (E), R120Q (q), and R133Q (ࡗ) are shown.
enzyme-substrate complex. The increase in pK aes is greater in the R133Q than in the R110Q mutant enzyme, leading to the prediction that Arg 133 is closer than Arg 110 to the catalytic base of the enzyme.
The R101Q mutant enzyme exhibits a modest decrease in k cat with a modest increase in K m for isocitrate and an elevated K m for Mn 2ϩ . The pH-V max profile is only slightly changed as compared with wild type enzyme. The results indicate that Arg 101 may have a role in catalysis. As shown in Fig. 1, porcine Arg 101 is aligned with E. coli Arg 112 , which is next to Ser 113 . The activity of E. coli isocitrate dehydrogenase is regulated by phosphorylation, with almost complete inactivation resulting from phosphorylation of Ser 113 (20,(23)(24)(25). Because replacement of Ser 113 by the negatively charged glutamate or aspartate also causes inactivation of the E. coli enzyme due to loss of the ability to bind isocitrate, it has been proposed that electrostatic repulsion and steric hindrance between the phosphoryl group and the ␥-carboxylate group of isocitrate are the major causes of inactivation in the phosphorylated E. coli enzyme (24). The pig heart enzyme is not regulated by phosphorylation; indeed in the eukaryotic enzymes, asparagine is located at the position equivalent to Ser 113 in the E. coli enzyme (Fig. 1), so these enzymes cannot be phosphorylated at that site. Nevertheless, it is not surprising that mutation of the nearby Arg 100 of rat cytosolic isocitrate dehydrogenase to the negatively charged glutamate causes inactivation (21), presumably from electrostatic repulsion of isocitrate. In our study of the porcine enzyme, we here show that even changing Arg 101 to the neutral glutamine results in a marked decrease in k cat (Table II), indicating that the positively charged arginine itself must be important in the overall catalytic reaction of isocitrate dehydrogenase.
Because there is not yet a three-dimensional structure for any mammalian NADP-dependent isocitrate dehydrogenase, we have generated an energy-minimized structure for the porcine NADP-specific isocitrate dehydrogenase, using the Modeler Program of Insight II, based on the sequence alignment and the x-ray coordinates of the E. coli enzyme complexed with metal isocitrate. Fig. 4 shows the portion of the model, which includes the substrate and the four arginine residues selected for mutation in this study. Arg 120 is far away from the substrate site in this model: the closest distance between this arginine's guanido group and isocitrate is 19.5 Å to the ␣-carboxylate. This long distance is consistent with our experimental results that the R120Q enzyme is similar to wild type, with little change in the specific activity, K m values or pK aes . For the rat cytoplasmic isocitrate dehydrogenase, even replacement of the corresponding Arg 119 by the negatively charged Glu caused no decrease in V max , although there is about a 15-fold increase in the K m for isocitrate, probably due to electrostatic repulsion (21). We conclude that Arg 120 in the porcine mitochondrial isocitrate dehydrogenase is not needed for function. Fig. 4 shows both Arg 110 and Arg 133 close to the substrate isocitrate. Arg 110 is only about 2.9 Å to the ␤-carboxylate of isocitrate, suggesting possible involvement in either hydrogen bonding or electrostatic interaction, whereas it is about 4.8 Å to the ␣-carboxylate of isocitrate. Arg 133 is located about equidistant to the ␣and ␤-carboxylates (ϳ4.6 Å), suggesting that it likely participates in electrostatic attraction of isocitrate. This model is consistent with the appreciable increases in the K m for isocitrate in both the R110Q and R133Q enzymes, with the K m effect for R110Q being greater, in agreement with the more striking proximity of Arg 110 to isocitrate. In the crystal structure of the E. coli enzyme, similar roles have been shown for the corresponding Arg 119 and Arg 153 (8).
The model of the pig isocitrate dehydrogenase is also consist-ent with an effect of the positively charged Arg 110 and Arg 133 in lowering the pK of the catalytic base, postulated to be Asp 275 or Asp 279 (22). Both of these arginines are closer to Asp 275 (which may strengthen the case for this residue as the catalytic base), with Arg 133 estimated as 4.9 Å from the carboxylate of Asp 275 , whereas Arg 110 is 6.8 Å from the same carboxylate. This difference in distances is consistent with the greater increase in pK a values upon eliminating the positive charge in the R133Q as compared with the R110Q mutant. Arg 101 is located no closer than 12.9 Å from isocitrate in the model (Fig. 4) and therefore cannot be directly involved in binding substrate in the active site. This conclusion is consistent with the observation that the R101Q mutant enzyme exhibits smaller increases in the K m for isocitrate than do the R110Q or R133Q mutants. Similarly, Arg 101 is not close to the proposed enzymic general base, Asp 275 or Asp 279 (14.7 and 21.0 Å, respectively), consistent with the result that the pK aes is not increased in the R101Q mutant. The role of Arg 101 cannot be established conclusively from the model. It is possible that even at this distance the positive charge of Arg contributes to an orientation of the isocitrate that is favorable for the catalytic reaction. An alternative explanation can also be envisioned: examination of the model for pig isocitrate dehydrogenase indicates that Arg 101 is located along a channel from the exterior solvent to the active site at which isocitrate is bound in the enzyme-substrate complex. Isocitrate may traverse this channel in the "on" reaction, whereas ␣-ketoglutarate may leave the active site via this channel in the "off" rate. Thus, Arg 101 may play a role in the overall reaction distinct from that involving removal of a proton from isocitrate by the catalytic base. How- FIG. 4. Homology model of the isocitrate binding site of porcine NADP-dependent isocitrate dehydrogenase. This model was generated using the Modeler program under homology, which is part of the Insight II suite, as described under "Experimental Procedures." The four Arg residues mutated to Gln are colored in cyan, isocitrate is shown in white, and the Mn 2ϩ is in green. ever, a definitive assignment of a role for Arg 101 must await additional experiments and the experimental determination of the enzyme's structure.
The homology model of pig mitochondrial NADP-dependent isocitrate dehydrogenase has allowed us to visualize the substrate binding site of the enzyme and to summarize the results of mutagenesis experiments. The model is consistent with most of the experimental results. Whereas it cannot substitute for an actual crystal structure of the porcine enzyme, the homology model can be useful for suggesting additional sites for mutagenesis and for conceptualizing results until the protein structure is determined experimentally.