NAD(+)-dependent isocitrate dehydrogenase. Cloning, nucleotide sequence, and disruption of the IDH2 gene from Saccharomyces cerevisiae.

NAD(+)-dependent isocitrate dehydrogenase from Saccharomyces cerevisiae is composed of two nonidentical subunits, designated IDH1 (Mr approximately 40,000) and IDH2 (Mr approximately 39,000). We have isolated and characterized a yeast genomic clone containing the IDH2 gene. The amino acid sequence deduced from the gene indicates that IDH2 is synthesized as a precursor of 369 amino acids (Mr 39,694) and is processed upon mitochondrial import to yield a mature protein of 354 amino acids (Mr 37,755). Amino acid sequence comparison between S. cerevisiae IDH2 and S. cerevisiae NADP(+)-dependent isocitrate dehydrogenase shows no significant sequence identity, whereas comparison of IDH2 and Escherichia coli NADP(+)-dependent isocitrate dehydrogenase reveals a 33% sequence identity. To confirm the identity of the IDH2 gene and examine the relationship between IDH1 and IDH2, the IDH2 gene was disrupted by genomic replacement in a haploid yeast strain. The disruption strain expressed no detectable IDH2, as determined by Western blot analysis, and was found to lack NAD(+)-dependent isocitrate dehydrogenase activity, indicating that IDH2 is essential for a functional enzyme. Overexpression of IDH2, however, did not result in increased NAD(+)-dependent isocitrate dehydrogenase activity, suggesting that both IDH1 and IDH2 subunits are required for catalytic activity. The disruption strain was unable to utilize acetate as a carbon source and exhibited a 2-fold slower growth rate than wild type strains on glycerol or lactate. This growth phenotype is consistent with NAD(+)-dependent isocitrate dehydrogenase performing an essential role in the oxidative function of the citric acid cycle.

NAD+-dependent isocitrate dehydrogenase from Saccharomyces cerevisiae is composed of two nonidentical subunits, designated IDHl (Mr -40,000) and IDHP (Mr -39,000). We have isolated and characterized a yeast genomic clone containing the IDH2 gene. The amino acid sequence deduced from the gene indicates that IDHP is synthesized as a precursor of 369 amino acids (MI 39,694) and is processed upon mitochondrial import to yield a mature protein of 354 amino acids (Mr 37,755). Amino acid sequence comparison between S. cerevisiae IDHP and S. cerevisiae NADP+-dependent isocitrate dehydrogenase shows no significant sequence identity, whereas comparison of IDHP and Escherichia coli NADP+-dependent isocitrate dehydrogenase reveals a 33% sequence identity.
To confirm the identity of the IDH2 gene and examine the relationship between IDHl and IDHB, the IDH2 gene was disrupted by genomic replacement in a haploid yeast strain. The disruption strain expressed no detectable IDHP, as determined by Western blot analysis, and was found to lack NAD+-dependent isocitrate dehydrogenase activity, indicating that IDHP is essential for a functional enzyme. Overexpression of IDHP, however, did not result in increased NAD+dependent isocitrate dehydrogenase activity, suggesting that both IDHl and IDHP subunits are required for catalytic activity. The disruption strain was unable to utilize acetate as a carbon source and exhibited a 2fold slower growth rate than wild type strains on glycerol or lactate. This growth phenotype is consistent with NAD+-dependent isocitrate dehydrogenase performing an essential role in the oxidative function of the citric acid cycle.
Isocitrate dehydrogenase catalyzes a rate-limiting step of the citric acid cycle, the conversion of isocitrate to a-ketoglutarate coupled to the production of NADH. Mitochondrial NAD+-dependent isocitrate dehydrogenase (NAD+-IDH)' is present in all eukaryotic cells and has been purified from a variety of sources, including Saccharomyces cereuisiae (I), * This work was supported by National Institutes of Health Grant GM39404 and American Cancer Society Grant BE-116. 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.
The nucleotide sequence(s) reported in thispaper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) M74131.
$ To whom correspondence should be addressed Dept. of Biological Chemistry, University of California, Irvine, CA 92717.
' The abbreviations used are: NAD+-IDH, NAD+-dependent isocitrate dehydrogenase; kb, kilobase(s). Neurospora crmsa (2), and pig (3) and bovine (4) heart. Characterization of purified NAD+-IDH has shown it to be a complex oligomeric enzyme that is subject to extensive allosteric regulation. The enzyme from S. cereuisiae functions as an octamer composed of two nonidentical subunits, designated IDHl (MI -40,000) and IDH2 (M, -39,000) (5), and is responsive to cellular energy levels as a result of binding and activation by AMP and NAD+ (6).
The roles of the individual subunits of NAD'-IDH in catalysis and regulation are not well understood. The number of binding sites within NAD'-IDH from S. cereuisiae for the allosteric regulator AMP, the substrate isocitrate, and the cofactor NAD', was determined using equilibrium dialysis to be fewer than the number of subunits for each compound (7). Similar results have been reported for the pig (8) and bovine (9) NAD+-IDH enzymes. Two models of subunit function and interaction can account for the results of these studies. The individual IDHl and IDHS subunits may have a specialized function of catalysis or regulation, with each containing complete, independent binding sites; alternatively, the subunits may contain half-binding sites with interaction between IDHl and IDHB required to form a complete binding pocket. Determination of the contribution of each subunit to the function and regulation of NAD'-IDH will provide insight into the control of substrate flow through the citric acid cycle.
In addition to mitochondrial NAD'-IDH, S. cereuisiae contains cytoplasmic and mitochondrial isocitrate dehydrogenase enzymes that utilize NADP+ as a cofactor. The two mitochondrial isocitrate dehydrogenase isozymes are differentially expressed in response to carbon source and oxygen levels (lo), suggesting that they function in separate pathways. The metabolic function of the NADP+-dependent isocitrate dehydrogenase (NADP'-IDH) isozymes has not been determined, but they are presumed to play a role in biosynthetic reactions that require NADP(H) and/or a-ketoglutarate. The gene encoding the mitochondrial NADP+-IDH has recently been cloned, and genomic disruption of the gene (11) did not result in the acetate-growth phenotype characteristic of yeast citric acid cycle mutants (12,13). These results suggest that the mitochondrial NADP'-IDH is not the primary isozyme of isocitrate dehydrogenase that functions in the citric acid cycle. The NADP+-IDH isozymes may, however, under certain conditions have compensatory roles in the citric acid cycle. Further analysis of the metabolic function of each isozyme will require the development of yeast strains lacking one or more of the isocitrate dehydrogenase isozymes.
We present in this paper the cloning and nucleotide sequence of the gene encoding the 1DH2 subunit of S. cereuisiae NAD'-IDH. A strain lacking IDHB was constructed by onestep genomic replacement. The disruption strain was characterized by lack of NAD+-IDH activity in uitro and by altered growth rates on a variety of carbon sources. Additionally, NAD'-dependent Isocitrate Dehydrogenase amino acid sequence comparison of the yeast IDH2 subunit and Escherichia coli NADP+-IDH revealed similarities between the eukaryotic NAD+-IDH enzyme and the prokaryotic NADP+-IDH enzyme.

MATERIALS AND METHODS
Yeast Strains and Growth Conditions-Yeast strains lacking a functional NAD+-IDH were previously identified (5) from screens of mutants that failed to grow on acetate as a carbon source (provided by Dr. M. McCammon, University of Texas Southwestern Medical Center, Dallas). Complementation experiments were perfomed using the NAD+-IDH mutants designated idhl-1 and idh2-I. These alleles were previously characterized and shown to lack the 40,000 and 39,000 molecular weight subunits, respectively (5). The AZDHZ strain was constructed using the wild type haploid yeast strain S173-6B (MATa leu2-3, 112 his3-I ura3-52 trpI-289;12).
Recombinant DNA Techniques and Nucleotide Sequencing-Plasmid purification from yeast was performed using the phenol-chloroform glass bead method (14); plasmids amplified in the E. coli strain DH5aF' were purified by the alkaline lysis method (15). Separation of plasmids and restriction fragments was achieved using 0.8% agarose gels. DNA fragments were purified by electroelution using an IBI apparatus. Purified restriction fragments were cloned into specified plasmids using T4 DNA ligase. The shuttle vectors YCp50 (16) and YEp352 (17) were utilized for expression of the IDH2 gene; a 2.8-kb EcoRI-HindIII fragment containing the IDHZ gene was inserted into the respective shuttle vector at corresponding EcoRI and Hind111 sites. Plasmids were transformed into yeast using the lithium chloride protocol (18).
Fragments of the IDH2 gene were subcloned into Bluescribe plasmids (Strategene) and used for nucleotide sequencing. Doublestranded dideoxynucleotide sequencing as described by Wang (19) was performed using Sequenase version 2.0 (US. Biochemical Corp.). Commercially available primers corresponding to plasmid sequences (T3 and T7) were used for most sequencing reactions; one reaction was primed using an internal synthetic oligonucleotide (Operon, San Pablo, CA).
The AZDH2 strain was constructed by deletion-insertion. The 2.8kb EcoRI-Hind111 fragment in Bluescribe was digested with XhoI and HincII to remove the IDH2 coding region. A yeast HIS3 gene with compatible ends was ligated into the plasmid, and recombinant plasmids were selected by screening with a 3ZP-labeled DNA fragment containing the HIS3 gene. Selected plasmids were amplified and digested with EcoRI and EcoRV to yield a linear 3.0-kb fragment. The digested DNA was transformed into the yeast strain S173-6B. Replacement of the wild type gene in His+ transformants by integration of the disrupted gene at the IDHZ locus was confirmed by Southern blot analysis.
DNA probes used for screening and Southern blot analysis were labeled by the random primer technique (20) using [a-32P]CTP (Amersham Corp.). Oligonucleotide probes were 5' end-labeled (15) with [Y-~'P]ATP (Amersham Corp.) using polynucleotide kinase. The degenerate oligonucleotide (ACNGTNAAA/GCAA/GCC) corresponding to potential coding sequences for the amino-terminal region of IDH2, amino acids 2-6, was synthesized by the Protein and Nucleic Acid Laboratory (University of California, Irvine).
Cell Fractionation and Activity Assays-Yeast mitochondria were prepared as described by Daum et al. (21) from strains cultivated overnight in YP medium with 2% glycerol/2% lactate as carbon sources. Briefly, pelleted cells were washed and resuspended in a buffer containing zymolase (ICN Pharmaceuticals) to produce spheroplasts. Cells were broken using a Dounce homogenizer, and mitochondria were isolated by differential centrifugation at 10,000 rpm. Mitochondrial extracts for use in enzyme assays and Western blot analyses were obtained by vortexing with glass beads followed by centrifugation in a microcentifuge to remove membranes. Protein concentrations were determined using the Bradford dye-binding method (22). NAD+-IDH activity was measured spectrophotometrically as previously described (5). Assays contained 40 mM Tris-HC1 (pH 7.5), 4 mM MgCIz, 0.25 mM NAD', and 5-40 pg of mitochondrial protein from extracts. All asssys were performed at 24 "C and initiated by the addition of 2.5 mM DL-isocitrate.
Western Blot Analysis-Samples of mitochondrial protein extracts (20 pg each) for Western blot analysis were electrophoresed on 10% polyacrylamide-sodium dodecyl sulfate gels (23). Separated proteins were transferred to polyvinylidene membranes (Immobilon, Millipore Corp.) using a graphite dry blotting apparatus (24). The membrane was stained with Coomassie Blue, blocked with 5% bovine serum albumin, and incubated in a 1:300 dilution of anti-IDH antiserum (5). Detection was performed using '2SI-labeled protein-A followed by autoradiography.

RESULTS
Cloning and Nucleotide Sequencing-As a strategy to isolate the ZDHl and IDH2 genes, yeast strains lacking functional NAD+-IDH, idhl-1 and idh2-1 (5), were used for complementation. The IDH-strains were transformed with a yeast genomic DNA library containing 5-15-kb fragments of yeast genomic DNA cloned into the shuttle vector YCp50. This vector has a single copy Cen origin of replication and a yeast URA3 gene for selection. Transformants were initially selected on minimal (YNB) medium plus glucose plates that lacked uracil to determine transformation efficiency. Transformants were collected and replated on rich (YP) acetate for selection of plasmids that would complement the IDH-phenotype, a failure to grow with acetate as a carbon source. A single complementing plasmid was isolated from one of more than 10,000 independent transformants and was found to contain an 8-kb yeast genomic DNA insert. The gene complementing the IDH-defect was localized to a 2.8-kb EcoRI-H i d 1 1 fragment by Southern blot analysis using a degenerate oligonucleotide probe encoding the amino terminus of IDH2, as described under "Materials and Methods." A partial restriction map of the 2.8-kb EcoRI-Hind111 fragment and the strategy used for nucleotide sequencing of this region are shown in Fig. 1.
Nucleotide sequence analysis revealed an open reading frame ( Fig. 2) that encodes the amino terminus of the mature IDH2 polypeptide (nucleotides 46-89), which was previously determined by amino acid sequence analysis (5). An ATG codon located 45 nucleotides upstream from the sequence encoding the mature form of IDHZ is presumed to be the initiator methionine codon, because it is the first methionine codon in the identified open reading frame and is preceded by a potential TATA box at relative nucleotide positions -13 to -17. Assuming that translation begins at this methionine codon, the deduced amino acid sequence predicts a precursor protein of 369 amino acids (Mr 39,694); cleavage of a 15residue amino-terminal peptide upon mitochondrial import would yield the mature protein of 354 amino acids (M, 37,755). The presumptive mitochondrial targeting peptide is similar in size and properties to other previously characterized yeast mitochondrial import peptides (25,26).
IDH2 Gene Disruption-To confirm the identity of the IDH2 gene and to examine the relationship between IDH2 and IDH1, the majority of the coding region of the cloned gene was replaced with the yeast HZS3 gene and used for onestep genomic replacement (27). Fig. 3A illustrates the deletion-insertion construct in which a 1.8-kb fragment containing the coding region and adjacent 3"noncoding region was replaced by a 1.7-kb fragment containing the yeast HZS3 gene. Genomic recombination in a His-haploid yeast strain was achieved by transformation with a linear 3.0-kb EcoRI-EcoRV fragment (cf. Fig. 1) containing the HZS3 disruption. TO confirm integration at the ZDH2 locus, Southern blot analysis was performed on genomic DNA isolated from the wild type and disruption strains that had been digested with EcoRI plus EcoRV. As shown in Fig. 3B for DNA from the wild type strain, two bands with expected sizes of approximately 1.1 and 1.9 kb hybridized with a 32P-labeled IDH2 probe (the 2.8kb EcoRI-Hid11 DNA fragment). In DNA from the disruption strain, however, a single 3.0-kb fragment hybridized to the same probe, indicating the expected loss of the internal EcoRV site with replacement of the IDH2 locus by the HIS3 gene. To confirm that the observed alteration was due to recombination of the HIS3 disruption construct at the IDH2 locus, Southern blots of the same DNA digests were hybrid-ized with a 32P-labeled HIS3 probe. The endogenous HIS3 gene was detected as a 5.0-kb fragment in DNA from both the wild type and the disruption strains. An additional 3.0-kb fragment, corresponding to the 3.0-kb fragment also detected by the 32P-labeled IDH2 probe, was present in DNA from the disruption strain, indicating integration of the HIS3 gene at the IDH2 locus. The absence of expression of the IDH2 subunit in the disruption (MDH2) strain was confirmed by Western blot analysis (Fig. 4). Using an anti-IDH antiserum prepared against the purified yeast holoenzyme (5)

FIG. 3. Disruption construct and Southern blot analysis of
the IDH2 genomic disruption. A, the IDH2-coding region and adjacent 3' region between the indicated XhoI and HincII sites were replaced by the yeast HIS3 gene, as described under "Materials and Methods." Genomic integration was achieved by transformation using a linear 3.0-kb EcoRI-EcoRV fragment containing the IDHZ disruption. B, Southern blot analysis of DNA from the wild type strain and the AIDH2 strain. Aliquots (10 pg) of total genomic DNA isolated from wild type (lanes I ) and AIDH2 (lanes 2) were digested with EcoRI and EcoRV and subjected to electrophoresis. Southern blot analysis was performed using as '"P-labeled probes the 2.8-kb EcoRI-Hind111 DNA fragment from IDH2 (left panel) or a 1.7-kb BamHI fragment containing the HIS3 gene (right panel) as described in the text. subunit proteins were detected in mitochondrial extracts from the wild type strain (lane 1 ), whereas mitochondrial extracts from the AIDH2 strain contained a single cross-reacting protein of M , -40,000 (IDH1) (lane 2). This result provides further evidence that the gene cloned encodes IDHB and also establishes that the IDHl subunit is encoded at a separate genomic locus.
Previous reports have shown that expression of NAD'-IDH is highly regulated and that approximately equivalent levels of the IDHl and IDHB subunits are maintained in wild type cells (5). As shown in Fig. 4, lane 2, disruption of IDH2, however, does not appear to alter levels of IDH1. This suggests that expression of the two subunits may not be coordinately controlled. To further examine expression of IDHl and IDH2, single or multicopy shuttle vectors containing the IDHZ gene were transformed into the AIDH2 strain, and levels of the subunits were examined by Western blot analyses. As shown in Fig. 4, lune 3, expression of IDH2 using the single copy vector YCp50 resulted in wild type levels of IDHP and a ratio of IDH1:IDHP aproximating 1:1. Overexpression of IDHZ using the multicopy vector YEp352, on the other hand, resulted in approximately a 30-fold increase, as determined by densitometry, in the levels of IDH2 as compared with IDHl (Fig. 4, lune 4 ) . These results suggest that the IDHl and IDH2 genes are independently transcribed and that subunit protein levels do not regulate expression.
Activity and Growth Phenotype of the Disruption Strain-To determine if IDHl or IDH2 can independently catalyze the conversion of isocitrate to a-ketoglutarate, mitochondrial extracts from strains expressing different levels of IDH2 were assayed for NAD'-IDH activity ( Table I). The AIDHZ strain contained no detectable NAD+-IDH activity; this lack of activity was not due to an inhibitory factor in the AIDH2 extract, because assays containing a mixture of wild type extract plus AIDHZ extract exhibited full activity. These results indicate that IDHl cannot independently catalyze the conversion of isocitrate to a-ketoglutarate and that IDHP is essential for NAD'-IDH activity. Overexpression of IDH2, on the other hand, using the multicopy vector YEp352 did not result in elevated NAD'-IDH activity, even though levels of IDHB are approximately 30-fold higher in these extracts. This result suggests that, like IDH1, IDH2 alone does not exhibit NAD'-IDH activity. Together, these data indicate that both subunits are required for a functional enzyme. An attempt to reconstitute the NAD'-IDH activity in uitro by combining IDHl in AIDH2 extracts with IDHB in AlDH2(YEp352-IDH2) extracts was unsuccessful. Although this was not extensively studied, formation of the active oligomeric form of the enzyme may require conditions or other cellular components that were absent in these assays.
The metabolic function of NAD+-IDH was examined by growth phenotype analysis of the AIDH2 strain. Because the original selection used for yeast strains lacking NAD'-IDH activity was an inability to grow on semisynthetic medium with 2% sodium acetate as the carbon source, we characterized " IDH2 deletion strain containing the plasmid indicated in paren-Mixed assay in which the two extracts were combined and allowed theses.
to equilibrate at room temperature for 5 min prior to assay. the growth phenotypes of the AIDH2 strain using acetate and other nonfermentable carbon sources. As shown in Table 11, the AIDHZ strain grew at reduced rates on all nonfermentable carbon sources and exhibited wild type growth rates only on glucose (rich or semisynthetic medium). Growth rates for the AIDHZ strain on glycerol or lactate were reduced 2.5-fold in rich media and 1.5-fold in semisynthetic media as compared with the wild type strain. The difference between growth on nonfermentable carbon sources in rich medium uersus semisynthetic medium may be due to increased expression of other enzymes under conditions of limiting nitrogen that provide compensation for the disruption defect. The AIDHZ strain doubled no more than once within a 48-h period following dilution from a rich (YP) glucose starter culture into rich or semisynthetic acetate medium. Expression of IDHZ using shuttle vectors in the AIDH2 strain was found to restore near wild type growth rates on all carbon sources, confirming that the metabolic defect is due to a loss of NAD+-IDH activity. A similar acetate-phenotype has been reported for yeast strains containing defects in the citric acid cycle isozymes of citrate synthase (12) or mitochondrial malate dehydrogenase (13). Thus, the observed growth phenotype for the AIDH2 strain is consistent with earlier suggestions (5) that the isocitrate dehydrogenase that utilizes NAD' is the isozyme that functions in the citric acid cycle.

DISCUSSION
Subunit Interactions and Function-We present here the first reported cloning and nucleotide sequence of a gene encoding an NAD+-dependent isocitrate dehydrogenase. Prior to this work, the only gene for an isocitrate dehydrogenase known to function in the citric acid cycle that had been cloned was that encoding the NADP+-dependent isocitrate dehydrogenase from E. coli (28). The NAD+-IDH enzyme from S. cereuisiue, however, is distinctly different from the E. coli enzyme. In addition to different cofactor specificities, the yeast enzyme exists as an octamer of two nonidentical subunits (5), whereas NADP+-IDH from E. coli exists as a dimer of identical subunits (29). Moreover, the enzyme from E. coli is regulated solely by phosphorylation (30), whereas the yeast enzyme is allosterically regulated by several compounds including AMP, NAD+, and citrate (6). The functional significance of a complex oligomeric NAD+-IDH enzyme in eukaryotic cells has been the subject of much speculation. Because

TABLE I1
Growth rates for wild type and IDHZ disruption strains Yeast strains were cultivated aerobically in a 30 "C shaker on either rich (YP) or semisynthetic (YNB/YE) medium containing the indicated carbon source a t a concentration of 2%. Logarithmic cell growth was measured spectrophotometrically at Asoo,,. The average value for duplicate experiments is reported. Glc, glucose; G11, glycerol; Lac, lactate; Ace, acetate. the E. coli enzyme is not subject to allosteric regulation, it is possible that the eukaryotic NAD'-IDH enzyme in parallel with compartmentation of the citric acid cycle has evolved as a more complex oligomeric enzyme to allow for allosteric regulation.
Based on the studies that indicate four isocitrate-and two AMP-and NAD+-binding sites per octamer, it has been suggested that the individual subunits of yeast NAD'-IDH have separate functions of catalysis or regulation (6). Sequence differences among tryptic peptides from the individual subunits of pig heart NAD'-IDH have also been interpreted to suggest possible differences in functions of the subunits (31). Our findings with the disruption of the IDH2 gene, however, do not provide support for a model of independent subunit function in which one subunit is catalytically active in the absence of a regulatory subunit. Strains expressing only IDHl showed no detectable NAD+-IDH activity, indicating that IDHl in the absence of IDH2 is not catalytically active. Additionally, overexpression of IDHB in the presence of wild type levels of IDHl did not yield an increase in activity, suggesting that IDHB can not function independently of IDH1. These results imply that both IDHl and IDHB are required for catalysis and suggest two possible models of subunit interaction and function. In one model, amino acid residues required for product formation, including either the substrate-binding site or catalytic site, are shared between IDHl and IDH2; this is similar to the E. coli enzyme in which each subunit contains half-sites for isocitrate binding which interact with one another to form a complete binding site (32). Alternatively, the amino acid residues required for product formation may be localized on a single subunit, but interactions with the regulatory subunit would be required for catalysis. Determination of the correct model will likely require sequence analysis of the IDHl gene, structural studies of the active octamer and isolated subunits, and alteration of the IDH2 and IDHl genes by site-specific mutagenesis.
Although the precise function of each subunit can not be determined at this time, our results provide evidence that the two subunits of the yeast enzyme are not a result of posttranslational modification of a single gene product. Deletion of the IDHZ gene does not alter expression of IDH1, confirming that IDHl and IDH2 are encoded by separate genes. Additionally, overexpression of IDH2 using the multicopy shuttle vector YEp352 did not result in increased levels of IDH1. This result indicates that expression of one gene does not affect expression of the other.
IDH2 Sequence Comparisons-Despite the differences in cofactor specificity, subunit structure, and regulation between the yeast NAD+-IDH and E. coli NADP'-IDH enzymes, IDH2 and the prokaryotic polypeptide exhibit considerable similarity in amino acid sequence. The alignment shown in Fig. 5 yields an overall sequence identity of 33%, and two regions, residues 82-115 and 268-316 of IDH2, contain greater than 55% identity. Structural analysis of the E. coli enzyme by xray crystallography has established that residues 110-120, corresponding to residues 95-105 in IDHB, function in isocitrate binding (32). Ser-113 of the E. coli enzyme forms a hydrogen bond with isocitrate (33), and phosphorylation at this site inhibits binding of isocitrate (34). Ser-98 of yeast IDHB aligns with this residue and may participate in substrate binding in the yeast NAD+-IDH octamer. Additional residues , which are near the binding site of isocitrate in the three-dimensional structure of the E. coli enzyme (32), are also conserved in IDH2 (Arg-104, . The second region of high sequence similarity, amino acids 268-NAD+-dependent Isocitrate Dehydrogenase  Amino acid sequence comparison of yeast IDH2, E. coli IDH, and pig heart IDH. The sequences for yeast IDH2 and E. coli IDH (27) were deduced from genomic DNA clones; the peptide sequences for pig heart IDH were determined from cysteinyl-containing tryptic peptides (30). Sequence alignment was determined using the Fast A program. Numbers correspond to residue positions in the yeast IDHB protein sequence. Residues of E. coli IDH and pig heart IDH that are identical with aligned residues in yeast IDHZ are starred (*), with boxes indicating identical residues referred to in text. Gaps (-) were introduced to optimize alignment. The mitochondrial presequence processing site of yeast IDHZ is indicated by an arrow. 316, contains the histidine postulated to be the acid-base catalyst present in many dehydrogenases. His-339 of the E. coli enzyme, corresponding to His-281 of IDH2, is the residue believed to perform this function (32). The alignment of residues in the E. coli enzyme involved in isocitrate binding and acid-base catalysis with residues in IDH2 suggests that the IDH2 subunit makes important contributions to the active site of the yeast enzyme.

T A L L L S~L T N H & C Q I Q N A V L S T I A S G P~----T G D L A G T A T T S S F T E A V I~
To obtain the alignment shown in Fig. 5, it was necessary to introduce two large gaps into the IDH2 sequence. The gap of 23 amino acids introduced between residues 146 and 147 of IDHZ spans the "clasp region" of the E. coli enzyme where the major subunit-subunit interactions occur (32). The absence in IDHZ of residues corresponding to those involved in dimerization of the E. coli enzyme indicates that the subunits of yeast NAD+-IDH interact using contacts different from those of the E. coli protein.
Amino acid sequence comparisons with pig heart NAD+-IDH are also included in Fig. 5. The pig enzyme, like the yeast enzyme, has been shown to be composed of nonidentical subunits having molecular weights of -40,000 (35). Pig NAD+-IDH, however, is a tetramer having an a2:/3:y structure (36). Several cysteine-containing peptides from each of the three types of subunits have been isolated and sequenced (31). Sequences from three of five peptides of the a-subunit, four of five peptides of the P-subunit, and all five peptides of the y-subunit could be aligned with IDH2; this resulted in 113 identical residues out of the total 265 residues aligned from all peptides, corresponding to a 43% sequence identity. Because a high degree of sequence similarity is observed with each of the three subunits, we cannot determine which subunit of the pig enzyme most closely resembles the yeast IDHZ subunit from the limited data available. One region of sequence similarity of particular significance, however, is the region encompassing IDHB residues 195-205. Asp-197 of IDHB aligns with an aspartate residue found to be chemically labeled by reaction of pig NAD+-IDH with an analogue of the allosteric modifier ADP (37). This result, in addition to the sequence similarity with the E. coli enzyme, suggests that the IDH2 subunit may have a regulatory site as well as catalytic function.
Relationship of NAD+and NADP-IDH Isozymes-In addition to the NAD+-IDH, two NADP+-IDH isozymes in S. cereuisiae catalyze the conversion of isocitrate to a-ketoglutarate. Although they function in similar reactions, the three isozymes of isocitrate dehydrogenase are structurally distinct.
The mitochondrial NADP+-IDH enzyme is similar to the E. coli enzyme in that it is a homodimer composed of subunits with a molecular weight of -45,000 and does not appear to be allosterically regulated. The gene encoding mitochondrial NADP+-IDH from S. cerevisiae has been cloned and the nucleotide sequence determined (11); however, no significant similarity (<20%) is observed when the protein sequence is compared with those for E. coli NADP+-IDH or yeast IDH2.
The lack of primary structure similarity between the mitochondrial NADP+-IDH and NAD+-IDH isozymes supports the idea that these isozymes function in separate metabolic pathways.
An important implication of the ability of the IDH2 disruption strain to grow on glycerol or lactate is that another enzyme is partially compensating for the lack of NAD+-IDH activity. We presume that this is the NADP+-IDH, with the most likely candidate being the mitochondrial isozyme. Both isocitrate and a-ketoglutarate, however, can cross the mitochondrial membrane, and thus the cytoplasmic isozyme may also contribute a-ketoglutarate to be utilized for citric acid cycle function. The growth phenotype of the AIDH2 strain suggests an essential role of NAD+-IDH in the oxidative function of the citric acid cycle and that the NADP+-IDH isozymes can function to provide a-ketoglutarate, although at a reduced rate, to the cycle in the absence of a functional NAD+-IDH isozyme. This hypothesis can be tested by disruption of the NADP+-IDH genes in the strain lacking IDH2 and analysis of the resulting growth phenotypes.