Genetic and Biochemical Interactions Involving Tricarboxylic Acid Cycle (TCA) Function Using a Collection of Mutants Defective in All TCA Cycle Genes
Beata Przybyla-Zawislak, Devi M. Gadde, Kurt Ducharme, Mark T. McCammon


The eight enzymes of the tricarboxylic acid (TCA) cycle are encoded by at least 15 different nuclear genes in Saccharomyces cerevisiae. We have constructed a set of yeast strains defective in these genes as part of a comprehensive analysis of the interactions among the TCA cycle proteins. The 15 major TCA cycle genes can be sorted into five phenotypic categories on the basis of their growth on nonfermentable carbon sources. We have previously reported a novel phenotype associated with mutants defective in the IDH2 gene encoding the Idh2p subunit of the NAD+-dependent isocitrate dehydrogenase (NAD-IDH). Null and nonsense idh2 mutants grow poorly on glycerol, but growth can be enhanced by extragenic mutations, termed glycerol suppressors, in the CIT1 gene encoding the TCA cycle citrate synthase and in other genes of oxidative metabolism. The TCA cycle mutant collection was utilized to search for other genes that can suppress idh2 mutants and to identify TCA cycle genes that display a similar suppressible growth phenotype on glycerol. Mutations in 7 TCA cycle genes were capable of functioning as suppressors for growth of idh2 mutants on glycerol. The only other TCA cycle gene to display the glycerol-suppressor-accumulation phenotype was IDH1, which encodes the companion Idh1p subunit of NAD-IDH. These results provide genetic evidence that NAD-IDH plays a unique role in TCA cycle function.

THE tricarboxylic acid (TCA) cycle is a central pathway of oxidative metabolism. This pathway is critical for the oxidation of acetyl-CoA and for the production of reducing equivalents that are used by the respiratory complexes to produce ATP (Gancedo and Serrano 1989; Zeng and Deckwer 1994). Enzymes of the TCA cycle are also important for biosynthetic processes, including gluconeogenesis and amino acid and heme biosynthesis. The TCA cycle appears to be organized into a supramolecular metabolic complex that allows for interdependent function of the enzymes (Srereet al. 1997; Vélotet al. 1997). However, this metabolic complex must be flexible so that subsets of enzymes may serve other specialized needs, such as biosynthetic functions. Despite a broad understanding of TCA cycle function, TCA cycle organization and its regulation remain poorly understood.

The yeast Saccharomyces cerevisiae is an excellent model organism with which to study TCA cycle function and regulation. Many of the TCA cycle enzymes are not required for growth on fermentable carbon sources, such as glucose. However, TCA cycle enzymes become more important for cell growth when nonfermentable carbon sources, such as ethanol, acetate, glycerol, lactate, or pyruvate must be metabolized. The isolation of mutants defective in the growth on nonfermentable carbon sources has allowed for the genetic analysis of the TCA cycle and many other oxidative functions (Tzagoloff and Dieckmann 1990; McCammon 1996; McAlister-Henn and Small 1997). The eight enzymes of the TCA cycle are encoded by at least 15 different nuclear genes in S. cerevisiae (Figure 1). Four of the enzymes, citrate synthase, aconitase, fumarase, and malate dehydrogenase are encoded by the single genes, CIT1, ACO1, FUM1, and MDH1, respectively (Suissaet al. 1984; McAlister-Henn and Thompson 1987; Wu and Tzagoloff 1987; Gangloffet al. 1990). The other four enzymes are composed of subunits encoded by more than one gene. The NAD+-dependent isocitrate dehydrogenase (NAD-IDH) is an octamer composed of four Idh1p and four Idh2p subunits encoded by the IDH1 and IDH2 genes, respectively (Cupp and McAlister-Henn 1991, 1992). The α-ketoglutarate (2-oxoglutarate) dehydrogenase complex (KGDC) contains three different subunits encoded by the KGD1, KGD2, and LPD1 genes (Rosset al. 1988; Repetto and Tzagoloff 1989, 1990). Interestingly, the lipoamide dehydrogenase protein that is encoded by the LPD1 gene is also a component of the pyruvate dehydrogenase and branched chain amino acid dehydrogenase complexes. Succinyl-CoA ligase is a heterodimer composed of an α-subunit encoded by the LSC1 gene and a β-subunit encoded by the LSC2 gene (Przybyla-Zawislaket al. 1998). Ubiquinone:succinate dehydrogenase is an integral membrane protein composed of four subunits encoded by the SDH1 through SDH4 genes (Lombardoet al. 1990; Chapmanet al. 1992; Bullis and Lemire 1994; Daignan-Fornieret al. 1994). Additional genes encoding a mitochondrial citrate synthase (CIT3), aconitase (ACO2), and 4 genes that may encode alternative subunits of succinate dehydrogenase (SDH1b, YMR118c, YLR164c, and YOR297c) have been identified in the Saccharomyces genome but have not been extensively characterized (Vélotet al. 1996; Jiaet al. 1997; Colbyet al. 1998). Isozymes of malate dehydrogenase, citrate synthase, and isocitrate dehydrogenase (NADP+-dependent) have also been characterized in the cytosol, peroxisomes, and mitochondria (reviewed in McAlister-Henn and Small 1997; van Roermundet al. 1998); however, they do not appear to function directly in the TCA cycle.

Figure 1.

—The yeast TCA cycle. The eight enzymes of the TCA cycle are encoded by at least 15 genes in S. cerevisiae. The gene(s) encoding a TCA cycle enzyme are shown in the center. The number of different subunits is indicated by multiple genes; however, the oligomeric state of an enzyme is not indicated (e.g., α4β4 for NAD-IDH). Enzyme (abbreviation), gene (S. cerevisiae locus): citrate synthase (CS-1), CIT1 (YNR001c); aconitase, ACO1 (YLR304c); NAD+-dependent isocitrate dehydrogenase (NAD-IDH), IDH1 (YNL037c), and IDH2 (YOR136w); α-ketoglutarate (2-oxoglutarate) dehydrogenase complex (KGDC), KGD1 (YIL125w), KGD2 (YDR148c), and LPD1 (YFL018c); succinyl-CoA ligase (synthetase), LSC1 (YOR142w), and LSC2 (YGR244c); succinate dehydrogenase (SDH), SDH1 (YKL148c), SDH2 (YLL041c), SDH3 (YKL141w), and SDH4 (YDR178w); fumarase, FUM1 (YPL262w); malate dehydrogenase (MDH-1), MDH1 (YKL085c). Genes shown by enzyme name encode open reading frames with significant homology to an inner gene. The contributions of these outer genes to TCA cycle function are not well understood. CIT3 (YPR001w); ACO2 (YJL200c); SDH1B (YJL045c), SDH3B (YMR118c), SDH4B (YLR164w), and SDH4C (YOR297c). The gene assignments for SDH3B, SDH4B, and SDH4C are tentative pending characterization. See text for references or consult the Saccharomyces Genome Database (http//

We have taken a genetic approach to the study of yeast TCA cycle function. From a collection of respiratory competent mutants that were unable to utilize acetate as a carbon source (Acn- mutants), mutations in eight TCA cycle genes were identified (Denniset al. 1999). These and other mutant strains have been utilized to study many aspects of TCA cycle function. The TCA cycle is required for full induction of peroxisomes. Proliferation of this organelle requires metabolism of oleic acid (Veenhuiset al. 1987), and TCA cycle mutants cannot properly oxidize the acetyl-CoA generated from the β-oxidation of oleate in the developing peroxisomes (McCammonet al. 1990). TCA cycle function is severely compromised in mutants lacking TCA cycle components, suggesting that the entire cycle is required for function in situ (Sumegiet al. 1992). Mutants defective in succinate dehydrogenase subunits are defective in gluconeogenesis, and they are impaired in their ability to assimilate ethanol into glucose-6-phosphate (Denniset al. 1999). Glucose-6-phosphate, or other phosphorylated glucose metabolites, are required to signal repression of gluconeogenic gene expression. This signaling is defective in glyoxylate cycle, gluconeogenic, and succinate dehydrogenase mutants, but it is not a common feature associated with all TCA cycle mutants (Denniset al. 1999). The LSC1 and LSC2 genes encoding the α- and β-subunits of the ATP-dependent succinyl-CoA ligase have also recently been characterized (Przybyla-Zawislaket al. 1998). Carbon-source dependent expression of these genes is representative of other TCA cycle genes (de Risiet al. 1997). However, unlike other TCA cycle mutants, Δlsc mutants are not impaired for growth on acetate.

We have been particularly interested in NAD-IDH. Both subunits of this octameric enzyme are required for catalytic activity and for allosteric regulation (Cupp and McAlister-Henn 1993; Zhao and McAlister-Henn 1997). NAD-IDH has also been reported to bind to the 5′-nontranslated region of mitochondrially encoded mRNAs (Elzingaet al. 1993) and hence may have a regulatory function outside of its catalytic role in the TCA cycle. Mutants lacking either subunit still maintain the other subunit within the mitochondrion (Zhao and McAlister-Henn 1996; Gaddeet al. 1998). While the unassembled Idh2p subunit is soluble, the unassembled Idh1p subunit is aggregated and highly insoluble. In addition to unassembled monomeric Idh1p, a covalently modified form of this protein is also observed as a high-molecular-mass complex that accumulates as a consequence of aberrant assembly, presumably via protein oxidation (Gaddeet al. 1998).

Two distinct sets of phenotypes have been observed with idh2 mutants. Strains harboring missense alleles, including active site mutations, grow well on a medium containing glycerol (YPG), while idh2 nonsense and null mutants grow very poorly on YPG (Gadde and McCammon 1997). In addition, strains harboring these latter alleles spontaneously accumulate extragenic mutations that enhance growth on YPG. This has been termed the glycerol-suppressor-accumulation phenotype, and the extragenic mutations have been defined as glycerol suppressors. To understand the glycerol suppression phenomenon associated with idh2 mutants, Gadde and McCammon (1997) previously isolated a collection of ∼200 glycerol suppressor mutants. Mutations at the CIT1 locus were the most common suppressor identified, representing ∼20% of the entire collection. The identification of idh2 cit1 double mutants in the Acn- mutant collection was the initial observation that led to these studies (McCammon 1996). The cit1 suppressors were analyzed in some detail, including sequence analysis of several alleles. This analysis revealed that missense, nonsense, and null mutations at CIT1 were capable of enhancing the glycerol growth of idh2 strains. Biochemical analysis began to narrow the possible causes of this phenomenon. Citrate and isocitrate do not appear to accumulate to toxic levels in idh2 mutants (Gadde and McCammon 1997). The suppression phenotype seems to have no obvious relation to the Idh1p covalent complex observed in Δidh2 strains, because suppressor colonies are observed in Δidh1 Δidh2 double mutants (Gaddeet al. 1998). The levels of many TCA cycle proteins were altered in idh2 mutant strains, and the presence of cit1-suppressing mutations modulated these alterations, indicating a link to the glycerol suppression phenomenon (Gadde and McCammon 1997).

To better understand the glycerol-suppressor-accumulation phenotype associated with idh2 mutant stains, we have assembled an isogenic collection of strains defective in subunits of every TCA cycle enzyme. This collection allows a comprehensive analysis of many aspects of TCA cycle function. In addition to CIT1, we have identified mutations in six other TCA cycle genes that are capable of suppressing idh2 mutations. We provide genetic evidence that the glycerol suppression phenomenon is unique to the genes encoding NAD-IDH subunits. These results suggest the NAD-IDH plays a special and unique role in regulating TCA cycle function.


Strains and media: The strains of S. cerevisiae used in this study are listed in Tables 1 and 2. Strains were grown on three basic types of media with a variety of carbon sources: 2% (w/v) glucose, 1% (w/v) potassium acetate or sodium pyruvate, 2% (v/v) ethanol or lactate, and 3% (v/v) glycerol. Rich medium consisted of 1% yeast extract and 2% peptone, 30 mg/liter adenine sulfate (YP), and a carbon source, usually glucose (YPD) or glycerol (YPG). Synthetic medium consisted of 0.7% yeast nitrogen base (Difco, Detroit) and glucose or 3% potassium acetate (SAce3) supplemented with amino acids and other nutrients to meet the auxotrophic requirements of the strains to maintain selection of plasmid-containing transformants or to select for gene disruptions (Guthrie and Fink 1991). Semisynthetic (SS) medium contained 0.7% yeast nitrogen base, auxotrophic supplements, 0.05% yeast extract, and carbon source, usually acetate (SSAce). Plates contained 2% agar (Difco). Plasmids were amplified in the E. coli strains TG1 [supE, hsdΔ5 thi Δ(lac-proAB) F(traD36 proAB+ lacIq lacZΔM154)] and DH5α [supE44 ΔlacU169 (ϕ80 lacZΔM15) hsdR17 recA1 endA1 gyrA96 thi-1 relA1]. Plasmids were prepared using commercially available kits (QIAGEN, Chatsworth, CA).

Genetic analysis of glycerol suppressors: Glycerol suppressor mutations are defined as extragenic mutations that enhance growth of idh2 null, nonsense, or frameshift mutations (Gadde and McCammon 1997). A collection of glycerol suppressor mutations of the idh2-4 and idh2-5 alleles was previously isolated (Gadde and McCammon 1997). The suppressor mutations were genetically separated from the idh2 alleles by standard tetrad analysis. Strains harboring the glycerol suppressor alleles were determined by their ability to enhance growth of idh2 strains on YPG and by their growth defects on nonfermentable carbon sources. Complementation groups were determined by genetic crosses to strains of opposite mating type as previously described (McCammon 1996). For alleles from several complementation groups (e.g., Table 5, part B), growth defects on nonfermentable carbon sources were very slight. Complementation analysis required replication of crosses on SAce3 plates and the successive transfer of the replicas onto two to three additional SAce3 plates. The slight growth defects relative to control crosses were eventually discerned.

Suppressor frequency: The frequency of glycerol suppressors was calculated in Δidh1, Δidh2, and Δidh1 Δidh2 strains. Overnight cultures in YPD were collected by centrifugation, washed once with water, diluted, and inoculated onto YPG and YPD plates. Glycerol suppressor colonies (i.e., large colonies among a lawn of small nonsuppressor colonies) were scored on YPG plates after 4 days at 30°. Colony-forming units were calculated on YPD plates. Mitochondrial petite mutants appear white in the ade2 background of these strains and were not counted into the colony-forming units. The glycerol suppressor frequency was calculated as the ratio of glycerol suppressors per colony-forming units and represents the average of at least three independent experiments.

Construction of deletion mutants: The TCA cycle gene disruptions listed below (Rothstein 1991) were usually constructed in strains MMYO11 or I2ΔHL. The 5′ upstream primer is listed first, followed by the 3′ downstream primer.

  • CIT1: a 1.8-kb PCR product containing the cit1::LEU2 disruption construct was amplified from strain CS1- genomic DNA (Kispalet al. 1988) with the primers 5′-CGTTGAAG GAGAGATTTGCTGA and 5′-CTGTACCACCTTCATGAT CAGAATG and used to disrupt the CIT1 gene.

  • CIT2: a 2.4-kb PCR product containing the cit2::URA3 construct was amplified from strain CS2- (Kispalet al. 1988) with the primers 5′-TTGGTGACGTTAATCTAAAGATAG and 5′-TATGCAGAGGGGTGTAAAAGTAGGC and used to disrupt the CIT2 gene.

  • CIT3: a 1.2-kb cit3::URA3 construct was amplified from a plasmid copy of URA3 with the hybrid primers (Lorenzet al. 1995) 5′-GTACAAAGGCTTCTACCGGGCGCACATATAT GCAGAAGGattttttttttattcttttttttgatttcgg and 5′-TTACAAC TTGTTAACATTGCTTGCTTTGGTAAGTGCTTCgctttttctt tccaattttttttttttcgt, where the uppercase nucleotides correspond to CIT3 and the lowercase correspond to URA3. Disruption was confirmed with the primers 5′-AATAGCA GTTTTGATGACT and 5′-CTACCTATAGACAGCATCAT, which amplified a 1.8-kb band from Δcit3 strains as compared to a 2.0-kb fragment from CIT3 strains.

  • ACO1: a 4.3-kb aco1::URA3 disruption construct was PCR-amplified from plasmid pAD2 (Gangloffet al. 1990) with the primers 5′-AAACAGGCCATTTACCTACGC and 5′-ACGACCAGTTGCTTCCAAAT.

  • ACO2: a 3.4-kb EcoRI fragment containing the aco2::HIS3 construct was excised from plasmid pRS-Δaco2 (Vélotet al. 1996) for disruption of ACO2. Disruption was confirmed through use of the primers 5′-CGTATTATTCGATTTGCGCAC and 5′-TTTGCACCGTCTATATGCAGA, which amplified a 2.8-kb band from Δaco2 strains as compared with a 2.5-kb fragment from ACO2 strains.

  • IDH1: a 1.9-kb fragment containing the idh1::LEU2 disruption construct was PCR-amplified from genomic DNA of strain ΔIDH1 (Cupp and McAlister-Henn 1992) with the primers 5′-TTCACTAAACGTTGCTTTGCG and 5′-GCCAATG TTGCCTAAGATGGT for disruption of IDH1.

  • IDH2: a 3.0-kb EcoRI-EcoRV fragment containing the idh2::HIS3 disruption construct was excised from plasmid YCpIDH2-HIS3 (Gadde and McCammon 1997) for disruption of IDH2.

  • KGD1: a 1.6-kb fragment containing the kgd1::URA3 deletion construct was PCR-amplified from strain W303ΔKGD1 (Repetto and Tzagoloff 1989) for disruption of KGD1 with the primers 5′-CATACTTCTTCCTAATTTCCCCAA and 5′-TCCCTAAAGGATCTATATGGG.

  • KDG2: a 2.1-kb PCR product containing the kdg2::HIS3 deletion construct was amplified from genomic DNA of strain W303ΔKDG2 (Repetto and Tzagoloff 1990) with the primers 5′-AGGAGTACCGTATAATGAACCAA and 5′-TCA AACAAGCTAACAAGGTCGAA. Because of the incompatibility of Δkgd2::HIS3 with Δidh2::HIS3, KGD2 was disrupted in strain 22B, which harbors the idh2-4 nonsense allele (Gadde and McCammon 1997).

  • LPD1: a 1.4-kb PCR product containing the lpd1::URA3 construct was amplified from strain MML22 (Lantermanet al. 1996) and used to disrupt the LPD1 gene with the primers 5′-GTCTTAACGTTGGATGTATC and 5′-CGCTTCTTTCAATTGCTCTTC.

  • LSC1: a 3-kb fragment containing the lsc1::LEU2 construct was excised from plasmid pGLSC1-LEU2 and used to disrupt the LSC1 gene (Przybyla-Zawislaket al. 1998).

  • LSC2: a 1.7-kb fragment containing the lsc2::URA3 construct was excised from plasmid pGLSC2-URA3 and used to disrupt the LSC2 gene (Przybyla-Zawislaket al. 1998).

  • SDH1: a 2.0-kb EcoRI fragment containing the sdh1::ADE2 construct was purified from plasmid pSDH1-ΔADE2 (obtained from Dr. Bernard D. Lemire) and used to disrupt the SDH1 gene.

  • SDH2: a 1.2-kb sdh2::URA3 construct was amplified from a plasmid copy of URA3 with the hybrid primers 5′-TTGAAC GTTATTGAGAAGGAAGGCCTTTTGTTTGGTGADGAA Gattttttttttattcttttttttgatttcgg and 5′-GATAGTCTAGGCAA ATGCCAAAGATTTCTTAATTTCAGCAATAGCgctttttctttc caattttttttttttcgt, where the uppercase nucleotides correspond to SDH2 and the lowercase nucleotides correspond to URA3, and used to disrupt the SDH2 gene.

  • SDH3: a 2.0-kb EcoRI-SalI fragment containing the sdh3::TRP1tained from Dr. Bernard D. Lemire) and used to disrupt the SDH3 gene.

  • SDH4: a 0.8-kb fragment containing the sdh4::TRP1 construct was PCR-amplified from genomic DNA of strain sdh4w2 (Robinson and Lemire 1996) with the primers 5′-GAAGAA CACAGGCGCAATTTAG and 3′-CGCATAATAATGACGA TATTAGGA and used to disrupt SDH4.

  • FUM1: strain W303ΔFUM1 harboring the fum1::LEU2 allele (Wu and Tzagoloff 1987) was mated to strain 27D harboring the idh2-5 frameshift allele, and Δfum1 and idh2-5 Δfum1 segregants were obtained by tetrad analysis.

  • MDH1: a 1.2-kb mdh1::URA3 construct was amplified from a plasmid copy of URA3 with the hybrid primers, 5′-TTG TCAAGAGTAGCTAAACGTGCGTTTTCCTCTACACTTG CCAACCCTattttttttttattcttttttttgatttcgg and 5′-CCCTATTT TTCACTCTATTTCTGATCTTGAACAATCTATTTAgcttttt ctttccaattttttttttttcgt, where the uppercase nucleotides correspond to MDH1 and the lowercase nucleotides correspond to URA3.

View this table:

Genotypes of S. cerevisiae strains used in this study

View this table:

Genotypes of TCA cycle deletion strains

Typical PCR reactions consisted of 10 mm Tris-HCl, pH 8.3, 50 mm KCl, 2.5 mm MgCl2, 1 μm of each primer, 0.2 mm of each dNTP, 0.025 units/μl of Taq polymerase, and 0.5–1.5 ng of template DNA per microliter. For gene disruptions, the reaction volume was 100 μl and for confirmation of disruptions the volume was 25 μl. For most PCR reactions the following cycling conditions were applied to isolate DNA fragments from plasmids and yeast genomic DNA: 5 min at 95° (1 cycle); 30 sec at 94°, 30 sec at 55°, and 2 min at 72° (30 cycles); and 3 min at 72° (1 cycle). For reactions with hybrid primers, the PCR amplification protocol was slightly modified: 5 min at 95° (1 cycle); 30 sec at 94°, 30 sec at 50°, and 2 min at 72° (3 cycles); 30 sec at 94°, 30 sec at 55°, and 2 min at 72° (30 cycles); and 3 min at 72° (1 cycle).

Confirmation of constructs: Four different methods were used to confirm the proper disruption of TCA cycle genes:

  1. Complementation: Strains were confirmed by complementation to previously documented TCA cycle mutants.

  2. PCR amplification: Chromosomal DNA was isolated from disrupted strains (Roseet al. 1990), and the predicted disruption was confirmed by PCR amplification. In all cases analyzed, primers flanking the TCA gene yielded an altered product of the predicted size. In addition, primers from an internal region of disrupting markers that would yield a product from the disrupted strain only and not from the corresponding wild type were used. The following marker primers were used: 5′-AGAAAGATGTGAAATTCTTTG (ADE2), 5′-ATCCAAACCTTTTTACTCCAC (HIS3), 5′-AA GGACCAAATAGGCAATG (LEU2), 5′-ATGCAGTTGGAC GATATCAA (TRP1), and 5′-GCATGACAATTCTGCTAA CAT (URA3). In properly disrupted strains the PCR products were observed with the upper strand primer of the disrupted gene and lower strand primer of the marker gene.

  3. Complementation with plasmids: The available TCA cycle genes on plasmids were transformed into the appropriate disrupted strain (Schiestl and Geitz 1989) and assayed for their ability to complement the mutant phenotype on nonfermentable carbon sources.

  4. Protein analysis: In some cases mutation of a particular TCA cycle gene was confirmed by loss of enzyme activity and/or loss of protein by immunoblot analysis, which was performed as previously described (Sumegiet al. 1992; Gadde and McCammon 1997; Gaddeet al. 1998; Przybyla-Zawislaket al. 1998).

Protein analysis: Whole-cell lysates and mitochondrial pellets were prepared as previously described (McCammon 1996). Protein concentration was determined using Bradford assay with bovine serum albumin as a standard (Bradford 1976). Proteins were resolved on 10% PAGE, transferred to nitrocellulose membranes, probed with antisera, and detected by chemiluminescence (ECL; Amersham, Arlington Heights, IL).

Figure 2.

—Glycerol suppression phenotypes of TCA cycle mutants. Strains were grown on YPD medium. Cells were collected by centrifugation, washed once in water, and resuspended to a final concentration of 2 OD600 units/ml of water. The culture was serially diluted in 10-fold steps, and 10 μl of each dilution was spotted onto YPG plates. Plates were photographed after incubation at 30° for 4 days. The genes analyzed are summarized in Table 4.


TCA cycle mutant collection and glycerol suppression: A set of isogenic yeast strains was constructed that harbored mutations in TCA cycle genes. These mutants were initially tested for the glycerol suppressor accumulation phenotype. Only the Δidh1 and Δidh2 mutations displayed this phenotype on YPG (Figure 2). Fast-growing colonies were observed among the lawn of cells from these two deletion strains after 3–4 days of incubation at 30°. The suppressor phenotype was also observed in a Δidh1 Δidh2 double-deletion mutant and in Δidh1idh1 and Δidh2idh2 diploids. The suppressor frequency in Δidh1, Δidh2, and Δidh1 Δidh2 haploid strains was ∼2 × 10-4 relative to colony-forming units on YPD. The phenotype was also observed in idh1 nonsense alleles but not with missense alleles. These results indicate that, of the TCA cycle genes, the glycerol-suppressor-accumulation phenotype is unique to IDH1 and IDH2 alleles lacking the corresponding Idh1p or Idh2p subunit.

Growth of the TCA cycle mutants on several nonfermentable carbon sources was monitored. In general, the most permissive carbon source was ethanol, followed by glycerol, pyruvate, and lactate, while acetate was the least permissive carbon source. The TCA cycle genes could be grouped into five phenotypic categories. Strains lacking the ACO1 gene would not grow on any of the nonfermentable carbon sources (Table 3). Similarly, strains harboring deletion mutations in five other genes, SDH1, SDH2, SDH3, SDH4, and FUM1, grew poorly on all of the nonfermentable carbon sources. Strains deleted for one of five genes, IDH1, IDH2, KGD1, KGD2, and LPD1, were able to grow well only on ethanol, with minimal growth on glycerol or pyruvate. Mutants in two genes, CIT1 and MDH1, were able to utilize all of the carbon sources except for acetate. Finally, strains lacking either the LSC1 or LSC2 genes could essentially grow on all of the carbon sources and were the only TCA cycle mutants that were able to grow on acetate. These deletion strains did display a slow growth phenotype on glycerol or pyruvate, which was previously utilized for molecular and genetic analyses of these genes (Przybyla-Zawislaket al. 1998). Many of the TCA cycle mutations displayed slow growth phenotypes on YPG similar to the Δidh1 and Δidh2 mutations. Thus, slow growth alone does not explain why only the Δidh mutants accumulate glycerol suppressors.

The TCA cycle genes were also deleted in a Δidh2 background to test them as glycerol suppressor mutations. Deletion of the CIT1, LSC1, LSC2, and MDH1 genes enhanced glycerol growth of Δidh2 strains (Figure 2; summarized in Table 4). For the most part, there was little difference between the glycerol growth of the single Δcit1, Δlsc1, Δlsc2, and Δmdh1 deletion strains and the strains harboring these mutations in a Δidh2 background. The remaining TCA cycle deletions did not enhance growth of Δidh2 strains on YPG. In most cases the opposite phenotype was observed, and these double mutants grew worse than either of the single mutants (e.g., the Δkgd2 strains). This is the phenotype normally expected for mutations affecting two steps in the same pathway, and it stands in sharp contrast to the glycerol suppression phenotype reported here. In addition, Δkgd1 mutations were not suppressed by Δcit1 or Δmdh1 mutations, indicating that these latter mutations do not function as suppressors of defects at KGD1. These results support previous observations that glycerol suppressors only arise in idh1 and idh2 strains.

While most of the TCA cycle null mutations did not function as glycerol suppressors, this did not prevent their accumulation in an idh2 strain. Upon longer incubation (7–10 days) of double mutants on YPG plates at 30°, glycerol suppressor colonies were observed that grew slightly faster than the double mutants but slower than either of the single mutants. Meiotic analysis of the glycerol suppressor mutations isolated in idh2-5 Δkgd1 and idh2-5 Δfum1 strains revealed that the glycerol suppressor mutations enhanced growth of the idh2-5 mutation but not of the Δkgd1 or Δfum1 mutations. This indicates that the suppressor mutations arose in the double-mutant strains because of the idh2 allele. The accumulation of glycerol suppressors in the double mutants was used as another indicator that the second mutation was not capable of suppressing the Δidh2 mutation (this phenotype is more apparent in Figure 5).

View this table:

Growth of TCA cycle deletion mutants on nonfermentable carbon sources

View this table:

Summary of TCA cycle genes and glycerol suppression phenotype

View this table:

Suppressor genes of idh2 mutations

Analysis of spontaneous glycerol suppressor mutations: We have previously reported the isolation of ∼200 spontaneous glycerol suppressor mutations (Gadde and McCammon 1997). This collection was isolated in two strains of opposite mating type harboring the idh2-4 mutation (a nonsense mutation at codon 210 of the IDH2 gene) and in similar strains harboring the idh2-5 mutation (a frame-shift mutation at codon 57). Two general strategies were used to identify the suppressor genes. First, complementation analysis was performed using tester strains harboring mutations in the TCA cycle and other related genes. Second, the suppressor mutations were separated from the idh2 mutations by sporulation. Strains harboring suppressor mutations were backcrossed to the suppressor collection to identify other alleles of that suppressor gene. The genes identified by this analysis were initially given ACN designations to reflect their inability to utilize acetate (McCammon 1996; Gadde and McCammon 1997).

Glycerol suppressor mutations were identified in 15 different genes representing ∼70% of the strains in the collection (Table 5). Mutations in 5 genes constituted ∼80% of the alleles identified. As previously reported, mutations at CIT1 were the most common, representing ∼20% of the entire collection (Gadde and McCammon 1997). Alleles of three genes, KGD1, ACN48, and ACN55, each composed ∼10% of the collection. Finally, mutations in ACN57 represented ∼6% of the collection.

The suppressor genes were placed into three categories on the basis of phenotypic and genetic analysis (Table 5). The mutations in the first group (eight genes) displayed a typical Acn- phenotype: an inability to grow on acetate (SSAce) while retaining the ability to grow on glycerol (YPG; McCammon 1996). Members of the second group displayed very subtle or mild growth defects on nonfermentable carbon sources. Of the seven genes placed in this category, two TCA cycle genes were identified: KGD1 and KGD2. Very little if any growth defect was observed by conventional analysis, such as streaking or spotting cells on plates containing nonfermentable carbon sources. Temperature-sensitive phenotypes were also not observed. This imposed a major constraint on the ability to analyze these alleles in detail. Crosses of strains had to be successively replicated several times onto plates containing 3–4% acetate, which is an inhibitory concentration (McCammon 1996). A further complication was the occurrence of intragenic complementation among some of the suppressor alleles. Identification of the TCA cycle suppressor genes by complementation analysis was greatly aided by the use of deletion strains that displayed tight growth phenotypes on nonfermentable carbon sources.

The third class of mutants displayed enhanced growth on YPG. Mutations in two genes, ACN48 and ACN62, representing ∼13% of the entire collection, were identified with this phenotype. Colonies of strains harboring the acn48 and acn62 mutations were bigger than the wild-type strains on YPG (Figure 3). The doubling time of the acn48 strain was 4.5 hr compared with the wild type of 5.7 hr in YPG. The presence of the acn48 mutation in an idh2 strain decreased the doubling time from ∼65 hr to ∼9 hr. The enhanced glycerol growth phenotype was not useful for complementation analysis. Fortunately, these strains also displayed another growth phenotype. While growth on glycerol was enhanced, the presence of acetate in the medium prevented growth on glycerol. This phenotype was observed with almost all of the Acn- collection (McCammon 1996).

TCA cycle suppressors: Mutations in seven TCA cycle genes were identified in the spontaneous suppressor collection: CIT1, KGD1, KGD2, LPD1, LSC1, LSC2, and MDH1. Mutations in these genes represent ∼35% of the entire collection. Mutations in CIT1, LSC1, LSC2, and MDH1 were predicted from the studies of the TCA cycle deletion mutant described above. However, the identification of the mutations in the three genes encoding subunits of the KGDC was somewhat surprising. Single-mutant alleles of KGD2 and LPD1 were identified, and these mutations displayed opposite phenotypes by themselves. The kgd2 allele displayed almost no growth defect on nonfermentable carbon sources, while the lpd1 allele displayed growth defects on most of the non-fermentable carbon sources tested.

Figure 3.

—Growth phenotypes of ACN48 alleles. Growth of strains harboring ACN48 or acn48 alleles on YPG and SSAce2 plates. Strains were processed as in Figure 2.

Suppressor mutations at the KGD1 locus were the second most abundant class. The 18 kgd1 alleles fell into two phenotypic groups. The major group, represented by the kgd1-24 allele, displayed nearly normal growth on nonfermentable carbon sources (Figure 4). Complicating analysis of this group was the occurrence of intragenic complementation between members. The other group, represented by kgd1-7, displayed a typical Acn- phenotype (Figure 4). Immunoblots of mitochondrial fractions from strains harboring these latter alleles indicated that the Kgd1 protein was present in all six strains. Analysis of TCA cycle enzymes in a strain harboring the kgd1-7 allele indicated that KGDC activity was ∼10% of wild-type levels. Taken together, these data indicate that “leaky” or partial-function mutations at KGD1 can suppress the idh2 mutations but that null mutations cannot suppress (Table 4). The same probably occurs for the defects in the other KGDC subunits.

Suppressor analysis of partial function TCA cycle alleles: Suppressor mutations in eight TCA cycle genes were not observed, either among the spontaneous suppressor collection (Table 5) or when the deletion mutation was constructed in the Δidh2 background (Figure 2; Table 4). Mutations in these genes, ACO1, IDH1, IDH2, SDH1, SDH2, SDH3, SDH4, and FUM1, cause severe growth defects on nonfermentable carbon sources that were similar in several cases to a mitochondrial petite mutation (Figure 2; Table 3). Because partial-function mutations in the three genes encoding KGDC subunits were identified as suppressors, we tested whether partial-function mutations in some TCA cycle genes might also be capable of functioning as idh2 suppressors. Leaky alleles of ACO1 (10% of wild-type aconitase activity), SDH2 (30% of wild-type succinate dehydrogenase activity), and SDH4 (Denniset al. 1999) were constructed in a Δidh2 background and tested for their ability to enhance growth on YPG. None of these alleles were capable of functioning as glycerol suppressors (Figure 5; Table 4).

Two mutations that directly affected NAD-IDH activity were also tested. Strains harboring the idh1-1 allele express immunodetectable Idh1p protein (Sumegiet al. 1992), do not accumulate glycerol suppressors, and did not suppress the idh2-4 allele. The idh2-2 allele harbors a missense mutation that alters an arginine at residue 119 of the Idh2p precursor that is predicted to be involved in binding isocitrate within the active site of NAD-IDH (Gadde and McCammon 1997). Strains harboring the idh2-2 allele did not accumulate suppressors. A idh2-2idh2 diploid strain did not accumulate glycerol suppressor colonies and grew well on YPG (Figure 5). Rather than functioning as a suppressor mutation, the idh2-2 missense allele partially complements the null allele by providing a catalytically inactive Idh2p subunit. This supports the idea that the glycerol-suppressor-accumulation phenotype is associated with the absence of Idh2p. In summary, because all of the mutant aco1, idh1, sdh2, and sdh4 alleles displayed growth phenotypes similar to mutation in TCA cycle genes that were capable of suppressing idh2 alleles, the ability to grow on YPG is not sufficient for a TCA cycle allele to function as a suppressor. Therefore, these results suggest that mutations in some TCA cycle genes will not be able to suppress idh2 mutations either as null mutations or as leaky mutations. This reinforces the observations that only a certain subset of TCA cycle genes are capable of suppressing idh2 defects.

Figure 4.

—Growth phenotypes of KGD1 alleles. Growth of strains harboring KGD1 or kgd1-7, Δkgd1, and kgd1-24 alleles on YPG and SSAce plates. Strains were processed as in Figure 2.

Figure 5.

—Glycerol suppression phenotypes of TCA cycle partial-function and isozyme mutations. Growth of partial-function, missense, or isozyme mutations on YPG in IDH2 and idh2 backgrounds. Strains were processed as in Figure 2. The idh2-4 and idh2-5 alleles were used to construct the aco2 idh2-4 and kgd1-7 idh2-5 strains, respectively.

Isozyme mutations do not suppress: Mutations defective in isozymes of TCA cycle proteins were also tested for their ability to suppress idh2 mutations. Two genes encoding citrate synthases were tested. CIT2 encodes a peroxisomal citrate synthase, while CIT3 apparently encodes an additional mitochondrial matrix isozyme (Kimet al. 1986; Jiaet al. 1997). Deletion of these genes individually did not significantly reduce growth on nonfermentable carbon sources, and neither mutation was capable of suppressing idh2 (Figure 5, Table 4). ACO2 encodes a mitochondrial aconitase isozyme that may be important primarily during fermentation (Vélotet al. 1996; van Den Berget al. 1998). The Δaco2 mutation had very little effect on growth on nonfermentable carbon sources and did not suppress idh2. Finally, MDH2 encodes a cytoplasmic malate dehydrogenase isozyme that is involved in gluconeogenesis (Minard and McAlister-Henn 1991; McAlister-Henn and Small 1997). The mdh2-3 allele lacks immunodetectable MDH-2 protein and is unable to utilize acetate, ethanol, or pyruvate as carbon and energy sources (McCammon 1996). This defect at MDH2 did not enhance growth of an idh2 mutation on glycerol (Figure 5, Table 4). Two conclusions can be drawn from these results. First, the ability to suppress idh2 mutations on glycerol is not a property shared by all genes involved in oxidative metabolism. Second, the ability to suppress idh2 mutations is not a property shared by all citrate synthase or malate dehydrogenase isozymes.


We have characterized a novel phenotype associated with the two genes encoding subunits of the TCA cycle NAD+-dependent isocitrate dehydrogenase. Mutants lacking either subunit of NAD-IDH grow poorly on glycerol medium and accumulate extragenic mutations, termed glycerol suppressors, that enhance growth on this nonfermentable carbon source. With the aim of understanding the mechanism behind the glycerol suppression phenomenon, two sets of genetic experiments were carried out. The first series of experiments was designed to determine which TCA cycle mutations displayed the glycerol suppressor accumulation phenotype. The second series of experiments was to determine which genes could suppress idh2 mutations. Results from both experiments are summarized in Tables 4 and 5. Analysis of the glycerol suppression phenomenon concentrated on TCA cycle genes, and several approaches were taken to study the role(s) of these genes in the glycerol suppression phenotype. Deletion mutations in TCA cycle genes were constructed, and these mutations were screened on YPG for a glycerol-suppressor-accumulation phenotype. Only the Δidh1 and Δidh2 strains displayed the glycerol-suppressor-accumulation phenotype (Figure 2; Table 4). The phenotype is probably not related to the ability to grow on glycerol because many of the TCA cycle mutants, like the Δidh2 and Δidh1 strains, grew poorly on YPG (Figure 2; Table 3).

Four TCA cycle genes were identified with the ability to suppress as null mutations: CIT1, LSC1, LSC2, and MDH1. Mutations in seven TCA cycle genes were identified among a spontaneous glycerol suppressor collection, including all of the genes that were capable of suppressing as nulls. Mutations in three additional TCA cycle genes were identified that encoded subunits of the KGDC complex, the enzyme downstream of NAD-IDH. Only partial function mutations in the KGD1, KGD2, and LPD1 genes could suppress while the null mutations could not. Because leaky suppressor mutations were identified in all seven of these TCA cycle genes, partial function mutations of TCA cycle genes that were not observed in the spontaneous suppressor collection (ACO1, IDH1, SDH2 and SDH4) were tested, and none were able to suppress. In sum, these results indicate that defects in only a certain subset of TCA cycle genes are capable of enhancing growth of idh2 mutants on glycerol (Table 4).

The TCA cycle gene defects therefore fall into three categories on the basis of their ability to suppress idh2 (Table 4). The first group, ACO1, IDH1, SDH1, SDH2, SDH3, SDH4, and FUM1, are apparently unable to suppress, either as null or as leaky mutants. Except for Δidh1, these mutations cause severe growth defects on all of the nonfermentable carbon sources tested. At the other extreme are the CIT1, LSC1, LSC2, and MDH1 genes that can suppress as null mutations and as partial function mutations. These four genes displayed the mildest growth defects on nonfermentable carbon sources, which may in part explain their ability to function as suppressors. Perhaps the most interesting set of suppressor mutations was in the genes encoding KGDC subunits. While null mutations in the KGD1, KGD2, and LPD1 genes did not suppress, leaky or partial-function alleles of each of these genes were capable of suppression. Mutations at KGD1 were among the most common spontaneous suppressor mutations identified. KGDC functions immediately downstream of NAD-IDH within the TCA cycle. Interestingly, mutations in the ACO1 and ACO2 genes that encode forms of aconitase, the enzyme upstream of NAD-IDH, did not suppress. While several TCA cycle enzymes were elevated in idh2 strains, KGDC activity was diminished (Gadde and McCammon 1997). Mammalian KGDC and NAD-IDH are in physical contact with one another (Srereet al. 1997), indicating a direct relationship between these two enzyme complexes. However, it is not clear whether physical interactions within the TCA cycle are necessary or even important for the glycerol-suppressor-accumulation phenotype.

Analysis of several genes related to TCA cycle function indicates that not all mutations in genes of oxidative metabolism function as glycerol suppressors. Both extra- and intramitochondrial isozymes of citrate synthase and malate dehydrogenase fail to suppress Δidh2 mutations. As stated above, partial-function alleles of several TCA cycle genes that do not suppress as null mutations also fail to function as suppressors. In conjunction with the distribution of alleles among the suppressor genes, these results suggest that there are some limits to the number and types of genes that can function as suppressors when mutated.

All of the TCA cycle genes tested displayed growth defects on nonfermentable carbon sources. However, there was a wide range of growth phenotypes displayed by these mutations. While the Δlsc1 and Δlsc2 mutants grew on every carbon source with only moderate growth defects on glycerol and pyruvate, many other mutants were essentially unable to grow on any of the five nonfermentable carbon sources used. It is somewhat puzzling why mutations in different genes of the same pathway would have such a wide range of growth phenotypes. Two main factors contribute to the growth properties of a particular oxidative mutation on nonfermentable carbon sources. First, nonfermentable carbon sources differ in how they are metabolized, and mutational defects may exacerbate these differences. Most nonfermentable carbon sources can be converted to acetyl-CoA, which is one of the central intermediates of oxidative metabolism. However, carbon sources differ in how they are converted to this intermediate (Gancedo and Serrano 1989), and one of the main differences appears to be the amount of high-energy intermediates, such as reducing equivalents, that are derived from the oxidation of a carbon source. For instance, the oxidation of ethanol generates two more NADH molecules than the oxidation of acetate. Many other factors also contribute to the differences in how these carbon sources are metabolized, including the accumulation (e.g., AMP) or depletion (e.g., CoA) of metabolites that can exert regulatory or even toxic effects.

Second, alternative mechanisms that bypass a particular defect may also distinguish the ability of certain mutants to grow on particular nonfermetable carbon sources. Some enzymes are encoded by more than one gene, while other enzymes may be able to bypass a particular block and allow for a compound to be metabolized. For instance, the relatively mild growth defects of the LSC genes were recently rationalized (Przybyla-Zawislaket al. 1998). The accumulation of succinyl-CoA because of the loss of succinyl-CoA ligase activity could be bypassed by isocitrate lyase, which can provide succinate for succinate dehydrogenase, or by acetyl-CoA hydrolase, which can hydrolyze succinyl-CoA to succinate or possibly act as a transferase to form succinate by transferring the CoA to another acyl chain, such as acetate. Both isocitrate lyase and acetyl-CoA hydrolase enzymes are highly induced on acetate, which may contribute to the ability of the Δlsc mutants to grow on acetate (Przybyla-Zawislaket al. 1998). Many of the isozymes differ in their substrate specificity and/or cofactor preference (e.g., NAD+ vs. NADP+), may be located in distinct subcellular compartments (e.g., cytosol or peroxisomes), and may be regulated in distinct manners by allosteric or transcriptional mechanisms (e.g., CIT2, MDH3, and IDP3 are highly induced primarily by oleate; van Roermundet al. 1998). Therefore the presence of isozymes does not automatically determine whether a gene defect can be effectively bypassed (McAlister-Henn and Small 1997). While carbon source metabolism and enzymatic bypass appear to be the main contributing factors to the growth phenotypes of particular TCA cycle mutations, the details for each metabolic block remain poorly defined.

Mutations in eight genes that do not encode TCA cycle enzymes were observed in the spontaneous glycerol suppressor collection. Little is known about the function of these genes. They are involved in oxidative metabolism because mutations in these genes display growth phenotypes on nonfermentable carbon sources. While these genes do not encode subunits of TCA cycle enzymes, it is possible that they may encode functions that are directly or indirectly related to TCA cycle function. The TCA cycle is organized within the mitochondrion in a highly protein-concentrated environment of the matrix and inner membrane (Srere 1987; Srere et al. 1987, 1997). Just about every TCA cycle enzyme interacts in some way with other proteins, such as inner membrane metabolite carrier proteins, transaminases, and anapleurotic, biosynthetic, and other catabolic enzymes that utilize or generate TCA cycle metabolites. While many of these proteins are located within the matrix/inner membrane compartment, others are located outside of the mitochondrion. The complex array of interacting proteins underscores the central role that the TCA cycle plays in intermediary metabolism.

There are two likely explanations for the relatively weak growth phenotypes observed with many of these undefined suppressor mutations. First, these mutations may result in proteins that retain partial function. Similar partial-function alleles were observed in all seven TCA cycle genes identified in the spontaneous suppressor collection. By analogy to the TCA cycle suppressor mutations, it is likely that null mutations in many undefined suppressor genes will display more severe growth phenotypes. Second, many genes involved in oxidative metabolism display little or no growth phenotype on nonfermentable carbon sources. Genes of this type include CIT2 and CIT3, encoding peroxisomal and mitochondrial isozymes of citrate synthase (Kimet al. 1986; Jiaet al. 1997), ACS1, encoding acetyl-CoA synthetase (de Virgilioet al. 1992), CAT1, encoding the carnitine acetyltranferase (Kispalet al. 1993), IDP1, encoding a mitochondrial NADP+-dependent isocitrate dehydrogenase (Haselbeck and McAlister-Henn 1993), CTP1, encoding the tricarboxylic acid carrier (Kaplanet al. 1996), and four of the five genes encoding subunits of the pyruvate dehydrogenase complex (Miranet al. 1993; Pronket al. 1996). All of these genes encode proteins whose function is related to the TCA cycle either directly or indirectly.

These results add new insight into the glycerol suppression phenomenon. Two functions for NAD-IDH have been proposed. First, it is an enzyme of the TCA cycle. Second, it appears to bind to mitochondrial mRNAs and has been proposed to be a regulator of mitochondrial protein expression and/or mitochondrial biogenesis (Elzingaet al. 1993). While it is clear that the glycerol-suppressor-accumulation phenotype is not associated with NAD-IDH catalytic activity (Gadde and McCammon 1997), the relationship between NAD-IDH and mtRNA binding is not well defined. RNA-binding residues of Idh1p appear to be distinct from catalytic and allosteric residues (L. A. Grivell, personal communication). In addition, alteration of these RNA-binding residues via mutagenesis imparts no detectable growth phenotype on nonfermentable carbon sources, suggesting that NAD-IDH mtRNA binding plays a subtle role in mitochondrial function. Furthermore, it is unclear how defects in another TCA cycle enzyme, such as citrate synthase, could affect this proposed extracatalytic function of NAD-IDH.

The current working model to understand the glycerol suppression phenotype is based on a structural role of the NAD-IDH oligomeric complex in oxidative metabolism. This phenotype is not observed with NAD-IDH missense mutants defective in the oxidative decarboxylation of isocitrate to α-ketoglutarate; rather it is observed only upon the loss of NAD-IDH octameric structure. The genetic and biochemical data reveal a link between glycerol suppression and specific alterations in other TCA cycle enzymes. This points to an extracatalytic role of NAD-IDH in some aspect of TCA cycle function. One possibility is that NAD-IDH plays a unique role in maintaining the structure of a TCA cycle metabolic complex. Upon the loss of NAD-IDH octamer, the TCA cycle supramolecular structure may collapse without dissociating. The loss of other TCA cycle functions, as with the glycerol suppressor mutations, may result in the further dissociation of the complex and allow the remaining enzymes to function more independently of one another. As seen in this report, certain defects may be more important than others for dissociation of the complex. Alternatively, the cell may be able to accommodate the loss of some TCA cycle enzymes over others in allowing for growth on glycerol (see previous discussion). While this model is consistent with in situ metabolic labeling, protein tethering, and crytallographic modeling studies (Sumegiet al. 1992; Lindbladh et al. 1994a,b; Vélotet al. 1997), it is not supported by recent studies on mistargeting NAD-IDH and MDH-1 to the cytosol (Zhao and McAlister-Henn 1996; Small and McAlister-Henn 1997). Further analysis will be required to dissect the alterations to TCA cycle function implicated by these studies.


The authors thank Dr. A. Tzagoloff for the Kgd1 antiserum and Drs. R. J. Dickinson, G. J-M. Laughin, B. D. Lemire, L. McAlister-Henn, I. E. Scheffler, P. A. Srere, and A. Tzagoloff for plasmids and yeast strains used in this report. The assistance of L. A. Leonard and H. J. Mroczkowski during the initial characterization of the kgd1 and acn48 suppressor mutations is acknowledged. This work was funded by the National Science Foundation grant MCB-9604225.


  • Communicating editor: M. Johnston

  • Received November 6, 1998.
  • Accepted February 12, 1999.


View Abstract