Interaction of the GTS1 gene product with glyceraldehyde- 3-phosphate dehydrogenase 1 required for the maintenance of the metabolic oscillations of the yeast Saccharomyces cerevisiae Weidong Liu, Jinqing Wang, Kazuhiro Mitsui, Hua Shen and Kunio Tsurugi Department of Biochemistry, Yamanashi Medical University, Japan We previously reported that GTS1 is involved in regulating ultradian oscillations of the glycolytic pathway induced by cyanide in cell suspensions as well as oscillations of energy metabolism in aerobic continuous cultures. Here, we screened a yeast cDNA library for proteins that bind to Gts1p using the yeast two-hybrid system and cloned multiple TDH cDNAs encoding the glycolytic enzyme glyceralde- hyde-3-phosphate dehydrogenase (GAPDH). We found that the zinc-finger and dimerization sites of Gts1p were required for full ability to bind GAPDH, and Gts1ps mutated at these sites lost the ability to regulate both aerobic and unaerobic ultradian oscillations of energy metabolism. Of the three TDH genes, only TDH1 fluctuated at the mRNA level in continuous culture and its deletion resulted in the disappearance of the oscillation without any affect on growth rate. This loss of biological rhythms in the TDH1- deleted mutant was rescued by the expression of TDH1 but not of TDH2 or TDH3 under the control of the TDH1 promoter. Thus, we hypothesized that Gts1p plays a role in the regulation of metabolic oscillation by interacting with the TDH1 product, GAPDH1, in yeast. Keywords: continuous culture; glyceraldehyde-3-phosphate dehydrogenase; Gts1p; metabolic oscillation; yeast. Ultradian (cycles with a period shorter than 24 h) oscilla- tions of the glycolytic pathway were induced after addition of glucose by inhibiting mitochondrial respiration with cyanide in cell suspensions [1–3] or cell extracts [4] of yeast with a periodicity of 1–2 min as monitored by measuring the level of NAD(P)H (reviewed in [5]). The glycolytic pathway has been shown to be an autogenous oscillator under extreme nonequilibrium conditions of energy in dissipative structures, which theoretically include all living organisms [5–7]. The pathway oscillates under the primary control of phosphofructokinase [8,9], transferring energy from glucose to NADH, which acts as the feed-forward activator, and then from NADH to ATP, which acts as the feedback inhibitor. After ATP as an inhibitor has been consumed, glucose again begins to enter the glycolytic pathway. Yeast cells also exhibit sustained ultradian oscil- lations of energy metabolism, with a periodicity of  4hin continuous (chemostat) culture under aerobic conditions in an open system using a bioreactor [10–13]. (Hereafter, aerobic oscillation will be referred to as energy metabolism or metabolic oscillation in distinction from cyanide-induced glycolytic oscillation.) Energy-metabolism oscillations, which arise spontaneously under conditions of high cell density ( 5 · 10 8 cellsÆmL )1 ) [14], are detectable as a periodic change in the factors involved in energy metabo- lism such as dissolved oxygen (DO) levels, CO 2 production, glucose and ethanol concentrations, and amounts of storage carbohydrates [10–13]. DO oscillation is caused by the periodic change between respiratory and respiratory-fer- mentative phases, in which oxygen demands are relatively high and low, respectively. Although the mechanism of energy-metabolism oscillation has not been elucidated, we assume that it is similar to that of glycolytic oscillation except for the involvement of mitochondria in ATP production as the NAD(P)H level is increased during the respiratory-fermentative phase [14] and the ATP level is increased in the early respiratory phase (J. Wang & K. Tsurugi, unpublished data). The energy-metabolism oscillation is coupled to oscillations of cell division [12,13,15] and cellular responses to various stress conditions, such as heat, oxidative agents and cytotoxic compounds [14,16]. (In this report, the term ÔcouplingÕ is used to refer to a state in which multiple oscillators fluctuate with the same periodic- ity irrespective of phase.) It should be added that a cell- cycle-independent oscillation of energy metabolism with a short periodicity (20 min to 1 h) was observed under particular conditions in continuous cultures [17]. The gene GTS1 was originally isolated from a yeast cDNA library with oligonucleotides encoding three Gly-Thr repeats which had been found in the clock-related gene period [18] and was thus named GTS1 [19]. We subsequently found that the repeat was translated as an Ala-Gln repeat in the GTS1 product Gts1p [20], which is similar to the Gln-rich domain found in the clock-related protein Clock [21]. Although the structural basis of Gts1p as a clock- related protein is obscure, mutations of GTS1 showed pleiotropic effects on yeast in a gene-dosage-dependent manner; these effects included the timing of budding and sporulation and the capacity for heat tolerance [19,22], all of which are known to be clock-related in other organisms Correspondence to K. Tsurugi, Department of Biochemistry 2, Yamanashi Medical University, 1110 Shimokato, Tamaho, Nakakoma, Yamanashi 409-3898, Japan. Fax: + 55 273 6784, Tel.: + 55 273 6784, E-mail: ktsurugi@res.yamanashi-med.ac.jp Abbreviations: DO, dissolved oxygen; GAPDH, glyceraldehyde- 3-phosphate dehydrogenase; ABC, ATP-binding cassette. (Received 4 April 2002, revised 4 June 2002, accepted 12 June 2002) Eur. J. Biochem. 269, 3560–3569 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03047.x (reviewed in [23,24]). Further, we found that the amplitudes and durations of the cyanide-induced ultradian oscillation changed significantly as a function of the GTS1 gene dosage, whereas the frequencies of oscillation did not vary very much among the stains tested [25]. We then reported that ultradian oscillation of energy metabolism and coup- ling of oscillations of cell division and stress responses in continuous culture were disrupted by inactivation of the GTS1 gene [16]. We recently presented evidence that the metabolic oscillator drives the heat resistance oscillator composed of machinery involved in the synthesis of trehalose [26]. We hypothesized that the synthesis of trehalose parallels activation of the glycolytic pathway, and trehalose is degraded by trehalase activated by cAMP coupled to the metabolic oscillation. Deletion of GTS1 resulted in the loss of the fluctuations in the synthesis of trehalose and cAMP [26], leading to the disappearance of the oscillations. These results suggested that Gts1p plays some role in the coupling of these oscillators. Furthermore, we suggested that the rhythmic expression of Gts1p is more important than the protein level for maintenance of ultradian rhythms, as the constitutive expression of GTS1 under the control of the TPI promoter resulted in the disappearance of ultradian rhythms [16]. More recently, we reported that, when GTS1 was expressed under the control of a short (183-bp) promoter in the GTS1-disrupted mutant, the amplitude of Gts1p fluctuations was restricted, leading to attenuation of the metabolic oscillation and the uncoup- ling of stress-resistance oscillations [27]. Thus, we suggested that, for stress-resistance oscillations to occur, full fluctu- ation in the level of Gts1p is required. We are now studying the molecular mechanism by which Gts1p functions in the coupling of ultradian oscillations. This study of ultradian oscillation in yeast should contribute to our understanding of the biological rhythms in other organisms, as the energy metabolism pathway is an autogenous oscillator in all living organisms [5–7], although there may be various modifica- tions. Gts1p contains a zinc-finger motif similar to that of GATA-transcription factors [28] in the N-terminal region and a glutamine-rich strand in the C-terminal region and thus Gts1p has been conventionally classified as a tran- scription factor in the yeast genome database [29]. However, it is unlikely that Gts1p is a DNA-binding protein because it binds to neither oligomers containing the GATA motifs nor Sau3AI fragments derived from the yeast genome in gel mobility-shift assays ( 1 S. Yaguchi & K. Tsurugi, unpublished data). Now, the zinc finger is shown to be similar to that contained in GTPase-activating proteins of ADP-ribosyla- tion factors, which are considered to play a role in protein interaction rather than DNA binding [30,31]. In addition, Gts1p has a dimerization site in an 18-amino-acid region of the C-terminal portion, which plays a role in the formation of a homodimer and heterodimer with the homologous cytoplasmic domain of some ATP-binding cassette (ABC) transporters [32]. The sequence was characterized by a few acidic amino-acid residues preceded by hydrophobic amino acids. The presence of these sequences suggested that Gts1p interacts with various proteins to show pleiotropic effects on yeast. In this study of how Gts1p functions, we searched for proteins that interact with Gts1p using the yeast two-hybrid system. We found that GAPDH binds to Gts1p and that both the zinc-finger and the dimerization sites of Gts1p were involved. We showed that mutations of Gts1p that removed the zinc finger or dimerization region affected the mainten- ance of ultradian oscillations of energy metabolism in yeast. Further, we present evidence that, of the three GAPDH species in yeast, GAPDH1 is involved in the appearance and maintenance of metabolic oscillations via interaction with Gts1p. MATERIALS AND METHODS Yeast strains and culture conditions A haploid strain of the yeast S. cerevisiae,W303,wasused for cyanide-induced glycolytic oscillation in cell suspensions [25]. Cells were cultured in medium containing 10 gÆL )1 glucose, 6.7 gÆL )1 yeast nitrogen base without amino acids (Difco Laboratory, Detroit, MI, USA), and 100 m M potassium phosphate at pH 5.0 supplemented with 40 lgÆmL )1 adenine sulfate and essential amino acids. The cells harvested 1–2 h after the glucose in the medium had been exhausted [25] were washed and resuspended in 100 m M potassium phosphate, pH 6.8, to a concentration of 4 mg proteinÆmL )1 before being starved for 3 h at 30 °C. Oscillations were induced by adding 20 m M glucose to the starvedcellsandthenafter4 min,5m M potassium cyanide. The oscillations were monitored by NADH fluorescence using a spectrofluorimeter (Hitachi F-4500) in a stirred and thermostatically regulated cuvette. Another haploid strain of the yeast S. cerevisiae, S288C, was used for the continuous cultures [14]. The cells were cultured at 30 °C in a synthetic medium containing 1% glucose as defined elsewhere [10] using a modified bench-top fermenter, MDL-6C (Marubishi Bioengineering, Tokyo, Japan) with a constant volume of 500 mL. The batch and successive continuous cultures were performed as described previously [14,16], and the periodic change in respiratory- fermentative metabolism was monitored by measuring the level of DO with an oxygen electrode. Construction of GTS1 mutants for the two-hybrid assay N-Terminal and C-terminal truncated mutants of GTS1 were constructed from the wild-type GTS1 as described previously [32]. The GTS1 mutant with the KpnI–NcoI fragment deleted (for a physical map of GTS1, see Fig. 1A), named GTS1[KN], which corresponds to the dimerization site covering 18 amino-acid residues from 296 to 313, was constructed as described previously [32] and named GTS1[DKN]. To replace the cysteine residue at position 53 in the putative zinc finger with tyrosine, site-directed mutagenesis was performed according to the protocol of the kit In vitro Mutagenesis Primers (Takara, Tokyo, Japan) using primers 1 and 2 (Table 1) as 5¢ primers for the first and second PCR, respectively. The 2.4-kbp SphI–SpeIfragment thus obtained containing nucleotides )1572 to +1513, with respect to the first residue A of GTS1,namedGTS1[C53Y], was inserted into pAUR112. The mutation was con- firmed by determining the nucleotide sequence of the PCR-amplified EcoRI–SalI fragment directed on the recombinant plasmid. The EcoRI–SalI fragment was inser- ted into the plasmid (pGBT9) of the two-hybrid system. GTS1[C53Y] with the dimerization site deleted [32], named Ó FEBS 2002 Gts1p–GAPDH1 interaction in metabolic oscillations (Eur. J. Biochem. 269) 3561 GTS1{[DKN] + [C53Y]}, was obtained by removing the KpnI–NcoI fragment from GTS1[C53Y]. GTS1[DC] was constructed by deleting the ClaI-ClaI fragment from GTS1. GTS1{[DC] + [DNS]} and GTS1{[DC] + [DKS]} were obtained by deleting the NcoI–SalIandKpnI–SalI frag- ments, respectively, from GTS1[DC]. The yeast two-hybrid system The two-hybrid assay was performed using the Matchma- ker Two-hybrid System (Clontech) as described previously [32]. To screen for Gts1p-interacting proteins, a yeast cDNA library prepared using a cDNA synthesis kit (Amersham Pharmacia Biotech) was inserted into the plasmid pGAD424 (LEU2 Amp r ) downstream of the activation domain of GAL4. The recombinant plasmids were transformed together with the recombinant plasmid pGBT9 (TRP1 Amp r ) carrying GTS1 in-frame downstream of the DNA-binding domain of GAL4. To determine the binding site of Gts1p and GAPDH, the wild-type and mutant GTS1 genes were inserted into pGBT9, and the cloned cDNA encoding the C-terminal 97 amino-acid residues of GAPDH3 (GAPDH3-C97) was inserted into pGAD424. The interactions between the prey and bait hybrid proteins were determined as activation of the lacZ reporter gene by measuring the b-galactosidase activity of the cells using either or both the colony lift assay and liquid culture assay for b-galactosidase as described previously [32]. The protein levels of the Gts1p mutants were examined by Western blotting using antibody to Gts1p as described previously [10]. Transformation with a chimeric plasmid harboring GTS1 The GTS1-deleted mutant gts1D(W303) was produced using the strain W303 as described previously [22,25]. To transform gts1D(W303) with GTS1, GTS1[DKN] and GTS1[C53Y], the constructs were inserted into the vector pYX222 at the multicloning site. These transformants were named pYXGTS1/gts1D(W303), pYXGTS1[DKN]/gts1D (W303) and pYXGTS1[C53Y]/gts1D(W303), respectively. The GTS1-deleted mutant gts1D was produced using strain S288C as described previously [14,16]. GTS1[DKN] and GTS1[C53Y] inserted into pAUR112 were transformed into gts1D and the transformants were named pACGTS1 [DKN]/gts1D and pACGTS1[C53Y]/gts1D, respectively. Preparation of TDH -deleted mutants TDH1, TDH2 and TDH3 encoding GAPDH1, GAPDH2 and GAPDH3, respectively [33,34], were cloned by PCR directed against yeast genomic DNA from S288C. Synthetic oligonucleotides for 5¢ and 3¢ primers, 3 and 4 for TDH1, Table 1. List of synthetic oligonucleotides used for primers (1–13) or probes (14, 15) in this experiment. No. Sequence (5¢-3¢) 15¢-CCCGAAGCATGCTGTGCCTAC-3¢ 25¢-CCCGAAGCATGCTGTGCTCAC-3¢ 35¢-GGAGAATTCGTTGGGCTGAGCTTCTGATCC-3¢ 45¢-TTGGGATCCTTAAGCCTTGGCAACATATTC-3¢ 55¢-CTTTGAATTCTGCTGTAACCCGTACATGCC-3¢ 65¢-ATTAGAATTCGCGGCTAAAGTTAAGCATGC-3¢ 75¢-GAAAACTGGATCCGACTTGTATGCTAAAGG-3¢ 85¢-TTGGGATCCTTAAGCCTTGGCAACATATTC-3¢ 95¢-GAAAACTGGATCCGACTTGTATGCTAAAGG-3¢ 10 5¢-TAATAGGAATTCTGATCATTTTGTTTTGTG-3¢ 11 5¢-AACAAGAATTCATGGTTAGAGTTGC-3¢ 12 5¢-AGGAGCTCGAGTTAAGCCTTGGCAAC-3¢ 13 5¢-AACTAACTAGTACTTGTATGCTAAAGG-3¢ 14 5¢-TTGTGTGTGTTGGTGAAATATCAAACC AAGTTCTTGATGAATTTC-3¢ 15 5¢-GTGTATTTTTCTTCGTTAACACCCATGACGAA CATTGGGGCGGTG-3¢ Fig. 1. Determination of the binding site of Gts1p in the 97 C-terminal residues of GAPDH3 (GAPDH3-C97) using N-terminal and C-terminal deletion mutants (A), and mutants with modifications in the zinc finger and/or dimerization site of GTS1 (B) in the two-hybrid system. Binding activity was determined by measuring b-galactosidase activity in colony lifts and in liquid cultures. The activity in liquid cultures is indicated by A 420 due to decomposition of o-nitrophenyl b- D -galacto- pyranoside (NpGal) as a substrate after incubation for 1 h at 30 °Cin a cell suspension containing 10 6 cells [32]. Bait genes indicate DNAs inserted in-frame downstream of the DNA-binding domain of GAL4. For prey, a cDNA encoding GAPDH3-C97 inserted downstream of the activation domain of GAL4 was used. Open boxes and solid lines indicate open reading frames and deleted regions, respectively. Lateral- striped and vertical-striped boxes indicate approximate positions of a zinc finger and the dimerization site, respectively. In (B), black boxes indicate approximate positions of the disrupted zinc finger. 3562 W. Liu et al.(Eur. J. Biochem. 269) Ó FEBS 2002 5and4forTDH2, and 6 and 4 (Table 1) for TDH3, respectively, were used. The PCR products were inserted into pUC19 vector, and the resulting plasmids, named pUCTDH1,pUCTDH2 and pUCTDH3,weredigestedto remove the promoters and 5¢ halves of the TDH genes. The remaining fragments containing the 3¢ halves of the TDH genes were ligated with fragments from the plasmid ASAJ2682 containing the kan gene as a selective marker [35]. The fragments containing mutated TDH genes (TDH1::kan, TDH2::kan,andTDH3::kan) were purified from the plasmids and transformed into S288C. The TDH- deleted mutants were identified by determining physical maps of products of PCR directed against genomic DNA fragments from each mutant. In addition, the TDH1- deleted mutant (tdh1D) was confirmed by analyzing the length of TDH1 mRNA by Northern blotting, and TDH2- deleted and TDH3-deleted mutants (tdh2D and tdh3D, respectively) by analyzing the length of the gene loci by Southern blotting. For the experiment to rescue tdh1D, the 1811-bp BamHI– BamHI fragment carrying the TDH1 ORF and its 804-bp 5¢ upstream sequence was PCR-amplified on genomic DNA using oligonucleotides 7 and 8 (Table 1) as 3¢ and 5¢ primers, respectively. The fragment was inserted into the SmaIsiteof pAUR112 and transformed into tdh1D. To construct TDH3 under the control of the promoter of TDH1, the 804-bp BamHI–EcoRI fragment was PCR-amplified on genomic DNA using oligonucleotides 9 and 10 (Table 1) for 3¢ and 5¢ primers, respectively. The BamHI–EcoRI fragment ligated to the EcoRI–BamHI fragment from pUCTDH3 was inserted into the SmaI site of pAUR112. To construct TDH2 under the control of the promoter of TDH1,the TDH2 ORF was amplified by PCR using oligonucleotides 11 and 12 (Table 1) for 5¢ and 3¢ primers, respectively, and, after digestion with EcoRI, the PCR product was ligated with the 804-bp BamHI–EcoRI fragment. The ligated fragment was amplified by PCR using oligonucleotides 13 and 12 (Table 1) as 5¢ and 3¢ primers, respectively. The PCR product was digested with SpeIandXhoI and inserted into the cognate sites of pRSA103. The recombinant plasmid carrying TDH2 and TSH3 under the control of the TDH1 promoter was transformed into tdh1D. Northern and Southern blotting For Southern-blot analysis of TDH2 and TDH3,the EcoRI–SnaBI fragment from pUCTDH2 encoding the 5¢-upstream sequence between )649 and )204, with respect to the first nucleotide A of TDH2,andtheEcoRI–NspV fragment from pUCTDH3 encoding the 5¢-upstream sequence between )572 and )25, with respect to the first nucleotide A of TDH3, were used for probes of TDH2 and TDH3, respectively. The probes were labeled using the kit PCR DIG-Labeling Mix (Roche Diagnostics, Mannheim, Germany). The HindIII and HpaI fragments of genomic DNAs of tdh2D and tdh3D, respectively, were Southern- blotted and visualized using these probes. For Northern blotting, TDH1 mRNA was detected using the synthetic oligonucleotide 14 (Table 1), which is comple- mentary to the 5¢ end extension specific for TDH1 mRNA spanning nucleotide positions )91 to )46 from the first nucleotide A of TDH1 [33]. TDH2 and TDH3 mRNAs were detected together using oligonucleotide 15 (Table 1), which is identical with the complementary sequence between + 371 and + 435, with respect to the first nucleotide A of each gene, having 30% dissimilarity from the corresponding region of TDH1 mRNA. The probes were labeled with the DIG Oligonucleotide Tailing kit (Roche Diagnostics) and found to have almost the same specific activities. Then, 20 lg(forTDH1)and10 lg(forTDH2 and TDH3)oftotal RNA extracted from cells using Isogen (Wako, Tokyo, Japan) were electrophoresed in a 1% agarose gel containing 2.2 M formaldehyde, and fluorescence was measured using a fluorimetric image analyzer Fuji LAS-1000 (Fuji-Film Co., Tokyo, Japan) after Northern blotting. The relative mRNA level was corrected by total RNA amount based on ethidium bromide staining after agarose gel electrophoresis. Western blotting and immunoprecipitation Western-blot analysis was performed as described previ- ously [16], and immunoprecipitation was performed using anti-(mouse GAPDH) IgG obtained from Funakoshi Co. (Tokyo, Japan). Determination of GAPDH activity GAPDH activity was determined as described previously [34]. RESULTS Interaction between Gts1p and GAPDH in the yeast two-hybrid system The yeast cDNA library was screened for proteins interact- ing with Gts1p using the yeast two-hybrid system. Of eight positive clones, four contained cDNA fragments of TDH3 encoding GAPDH3 and the others contained all different genes. As GAPDH is a key enzyme in the energy- metabolism pathway, we decided to study it in detail as a candidate for a Gts1p-binding protein. The four fragments of TDH3 all differed in length, encoding the C-terminal 120, 100, 97, and 76 amino-acid residues of GAPDH3. However, as the shortest fragment, named GAPDH3-C76, had only weak binding activity in the two-hybrid assay compared with the others (data not shown), the one with the C-terminal 97 residues (GAPDH3-C97) was used to search for binding site(s) of Gts1p. The screening of a series of deletion mutants of GTS1 [32] for activity to bind GAP- DH3-C97 in the two-hybrid system revealed that the binding sites were located in both the N-terminal and C-terminal portions of Gts1p, as truncation of either portion resulted in a significant loss of binding activity (Fig. 1A). Keeping in mind that Gts1p was characterized structurally by a putative zinc finger and a dimerization site located in the N-terminal and C-terminal portions, respect- ively, we examined the binding sites using a new series of GTS1 mutants with modifications at these sites (Fig. 1B). The binding activity of Gts1p-DC encoded by GTS1[DC] lacking the ClaI–ClaI region in the middle of Gts1p was even higher than that of wild-type Gts1p. Comparing the binding activity of Gts1p-(DC+DKS)withthatofGts1p- (DC+DNS), deletion of the dimerization site from Gts1p- DC resulted in almost complete loss of the activity (Fig. 1B), suggesting that the dimerization site plays an important role Ó FEBS 2002 Gts1p–GAPDH1 interaction in metabolic oscillations (Eur. J. Biochem. 269) 3563 in the binding to GAPDH. However, the finding that Gts1p lacking only the dimerization region (Gts1p-DKN) still had some binding activity suggested the presence of other sites. Consistently, the binding activity of Gts1p-C53Y with a disrupted zinc finger was much weaker than that of Gts1p- DKN, and Gts1p-(DKN + C53Y) lacking the zinc finger together with the dimerization site lost all activity (Fig. 1B). Thus, for full activity of Gts1p to bind GAPDH-C97, both the zinc finger and dimerization site were required. Yeast contains three GAPDH proteins, named GAP- DH1, GAPDH2, and GAPDH3, which are encoded by TDH1, TDH2,andTDH3, respectively [33,34]. The amino- acid sequence of GAPDH3 has  90% identity with that of GAPDH1 and 98% identity with that of GAPDH2 in either the C-terminal 97 residues or whole sequences. As these proteins are so similar, we examined whether the binding activity to Gtsp1 is specific for GAPDH3 using whole lengths of GAPDH proteins as prey in the two-hybrid system. The results showed that there were no differences among them in terms of their ability to bind Gts1p as bait (data not shown). Therefore, it is likely that all GAPDH proteins possess similar binding activity to Gts1p, although cDNA fragments encoding GAPDH3 were obtained in the screening assay for Gts1p-binding proteins. Binding of Gts1p to GAPDH in cell lysate The binding activities of Gts1p-DKN and Gts1p-C53Y to GAPDH were further tested by the coimmunoprecipitation assay (Fig. 2). Cell lysates expressing the wild-type and mutant Gts1p proteins were immunoprecipitated with antibodies to GAPDH, and Western blots of immunopre- cipitates were stained with a labeled anti-Gts1p IgG. Consistent with the result from the two-hybrid assay, neither Gts1p-DKN nor Gts1p-C53Y could bind to GAP- DH-C97 (Fig. 2). Why Gts1p-C53Y migrated slower than Gts1p on SDS/PAGE is not known. Effect of Gts1p mutants on cyanide-induced glycolytic oscillation To investigate whether Gts1p interacts with GAPDH to regulate the ultradian oscillations in yeast, GTS1[C53Y] and GTS1[DKN] were transformed into gts1D(W303), and the transformants were subjected to experiments on cyanide- induced glycolytic oscillation. We previously reported that the duration of cyanide-induced glycolytic oscillations was lengthened by 27% in the GTS1-deleted mutant and shortened by 20% in the GTS1-overexpressing strain compared with the wild-type cell [25]. Similarly the duration of the oscillations was severely shortened when the wild-type GTS1 was overexpressed in gts1D (pYXGTS1/gts1D) (Fig. 3). When gts1D cells overexpressed Gts1p-C53Y or Gts1p-DKN, the duration of the oscillations was lengthened Fig. 2. Determination of binding activity of the dimerization-site-deleted Gts1p-DKN and zinc-finger-disrupted Gts1p-C53Y for GAPDH in vivo by coimmunoprecipitation. The protein levels of Gts1p in cell lysates from the GTS1-deleted mutants [gts1D(W303)] expressing the wild- type (pYXGTS1/gts1D) (lane 1), zinc-finger-disrupted (pYXGTS1 [C53Y]/gts1D) (lane 2) or dimerization-site-deleted (pYXGTS1[DKN]/ gts1D) (lane 3) were determined by Western blotting using antibodies to Gts1p (upper panel). The cell lysates were immunoprecipitated with antibodies to GAPDH, and coprecipitated Gts1p was detected by Western blotting using antibodies to Gts1p (lower panel). Fig. 3. Representative patterns of ultradian oscillation of the NADH fluorescence at 20 °C in cell suspensions of gts1D(W303) (A), gts1D(W303) overexpressing the wild-type (pYXGTS1/gts1D)(B), gts1D(W303) overexpressing the zinc-finger-disrupted Gts1p (pYXGTS1 [C53Y]/gts1D) (C), and gts1D(W303) overexpressing the dimerization site-deleted Gts1p (pYXGTS1[DKN]/gts1D)(D).Cells were harvested about 1.5 h after the diauxic shift. Glucose (20 m M final concentra- tion) was added at )5minandKCN(5m M )attimezero. 3564 W. Liu et al.(Eur. J. Biochem. 269) Ó FEBS 2002 severalfold compared with that of the GTS1-overexpressing strain (Fig. 3). The fact that the oscillation was lengthened by inactivation of GTS1 [25] suggested that mutant Gts1ps without the ability to bind GAPDH could not function like the wild-type Gts1p to regulate cyanide-induced glycolytic oscillation. Effect of Gts1p mutants on the coupling of biological rhythms in continuous cultures As we previously reported [16], the wild-type Gts1p, expressed with the centromeric recombinant plasmid carry- ing GTS1 with its upstream region of about 1.0 kbp (pACGTS1[N-C]) rescued the metabolic oscillations lost in the GTS1-deleted mutant (gts1D) in continuous culture (Fig. 4A). The energy-metabolism oscillation continued for about 4 days, although the amplitude of DO oscillation was about 30% lower than it was in the wild-type cell [16] (also summarized in Table 2). To investigate whether Gts1p mutants that had lost the ability to bind GAPDH can rescue the metabolic oscillations in gts1D, the recombinant plas- mids carrying GTS1[C53Y] and GTS1[DKN] in centro- mere-based plasmids, named pACGTS1[DKN]/gts1D and pACGTS1[DKN]/gts1D were transformed into gts1D,and the transformants subjected to continuous culture. Although the cells grew normally reaching a critical density at the beginning of the culture, oscillations of energy metabolism did not appear in either transformant (Fig. 4 and Table 2). This suggested that Gts1p–GAPDH interac- tion was required for metabolic oscillation. Changes in the TDH mRNA levels during the energy-metabolism oscillation in continuous cultures GAPDH is generally considered to be a constitutively expressed protein. However, the finding suggesting that GAPDH is involved in the regulation of metabolic oscilla- tion raised the possibility that at least one component of GAPDH fluctuates at the protein level to function in the metabolic oscillation. We examined the activity and protein concentration of GAPDH as a whole during metabolic oscillation in continuous culture, but found no apparent fluctuations (data not shown). Thus it is possible that only a particular GAPDH species was oscillating. It has been reported that TDH2 and TDH3 were expressed predominantly in exponentially growing cells, whereas the expression of TDH1 was increased in the stationary phase and under conditions of stress [36,37]. Consistent with such reports, we found that the mRNA level of TDH2/TDH3 [determined together, as their nuc- leotide sequences are too similar (93% identity) to discrimi- nate between them by Northern-blot analysis] in the stationary-phase cells was decreased by about 20% and that the TDH1 mRNA was increased by 50%, occupying about 20% of total TDH mRNA, compared with the case in exponentially growing cells (data not shown). The mRNA levels of TDH1 and TDH2/TDH3 were determined during continuous culture in wild-type cells (Fig. 5). The TDH1 mRNA level fluctuated with the same periodicity as energy metabolism, whereas the level of TDH2/TDH3 mRNAs did not fluctuate. The result suggested that the expression of TDH1 apparently fluctuated in concert with metabolic oscillation, although it remains possible that the expression of TDH2 and TDH3 fluctuated, with opposite phases canceling out the oscillations of each other. Effect of TDH gene disruption on energy-metabolism oscillation in continuous cultures 2 To further investigate which TDH genes participate in the regulation of the ultradian oscillations in yeast, we constructed deletion mutants of each gene and subjected them to continuous culture. As the deletion of TDH genes reportedly inhibits the growth of yeast to varying degrees Fig. 4. Representative patterns of the DO oscillation in a continuous culture of gts1D expressing the wild-type Gts1p (pACGTS1[N-C]/gts1D) as a control (A), gts1D expressing the dimerization-site-deleted Gts1p (pACGTS1[DKN]/gts1D)(B),andgts1D expressing the zinc-finger-dis- rupted Gts1p (pACGTS1[C53Y]/gts1D)(C).Continuous culture was started at time zero and continued at a dilution rate of 0.1 h )1 with a synthetic medium containing 1% glucose at 30 °C. The energy meta- bolism oscillation was monitored by measuring the level of dissolved oxygen (DO). Ó FEBS 2002 Gts1p–GAPDH1 interaction in metabolic oscillations (Eur. J. Biochem. 269) 3565 [34], we determined the cell densities at the beginning of the continuous culture and thereafter. The tdh1D mutant grew as fast as the wild-type cell, reaching a critical cell density (Table 2), and DO oscillations disappeared within a day (Fig. 6A) similar to the case of gts1D [16]. The growth of tdh2D was disturbed, but the cell density reached 90% of that of the wild-type at the beginning of the continuous culture (Table 2). The culture showed a DO oscillation with wavelength and amplitude similar to those of the wild-type (Fig. 7A and Table 2), but it disappeared in 2 days because eof disturbance of cell growth, as the cell density decreased to 4.14 · 10 8 mL )1 at the end of the oscillation. During the transient oscillation, the TDH1 mRNA level oscillated, whereas that of TDH3 did not (Fig. 7B). The result confirmed the fluctuation of the TDH1 mRNA level and eliminated the possibility that the expressions of TDH2 and TDH3 fluctuated with opposite Table 2. Cell densities and durations and amplitudes of the DO oscillations in continuous cultures of the wild-type and GTS1 and TDH mutants. Strain Cell density (· 10 )8 ÆmL )1 ) a Duration (h) Amplitude (%) b Wild-type 5.03 ± 0.08 (8) c 132 ± 4.4 40.4 ± 2.21 gts1D 5.12 ± 0.11 (5) (No or short oscillations) pACGTS1[N-C]/gts1D 5.10 ± 0.18 (3) 117 ± 7.8 28.4 ± 2.89 pACGTS1[DKN]/gts1D 5.08 ± 0.23 (3) (No oscillations) pACGTS1[C53Y]/gts1D 5.03 ± 0.15 (3) (No or short oscillations (<12 h)) tdh1D 5.01 ± 0.13 (3) (No or short oscillations (<15 h)) tdh2D 4.58 ± 0.20 (3) 30 ± 4.5 31.8 ± 1.80 tdh3D 3.32 ± 0.27 (2) (No continuous cultures) d pACTDH1/tdh1D 4.95 ± 0.26 (3) 97 ± 8.5 28.2 ± 2.80 pACTDH1pr.TDH2/tdh1D 3.40 ± 0.40 (2) (No continuous cultures) pACTDH1pr.TDH3/tdh1D 2.45 ± 0.35 (2) (No continuous cultures) a Cell density at the beginning of continuous culture. b The amplitude was estimated by subtracting the DO concentration (%) of the valley from that of the peak of the next wave. c The number of experiments used for statistical analysis. d Continuous cultures could not be started because of the disturbance of cell growth. Fig. 5. Changes in the TDH mRNA levels during the DO oscillation in a continuous culture of the wild-type S288C. Cell samples were harvested on the second day of the DO oscillation. TDH1 (s)andTDH2/TDH3 mRNA levels (d) were determined by Northern-blotting analysis. Total RNA stained with ethidium bromide was conventionally used for the quantitative control for applied RNA amounts. Fig. 6. Representative patterns of the DO oscillation in continuous cul- tures of TDH1-deleted mutant (tdh1D) (A) and cells transformed with TDH1 under the control of its own promoter (B). Continuous culture was started at time zero and continued at a dilution rate of 0.1 h )1 with a synthetic medium containing 1% glucose at 30 °C. 3566 W. Liu et al.(Eur. J. Biochem. 269) Ó FEBS 2002 phases (Fig. 6). The growth of tdh3D was most disturbed among the three TDH mutants as reported in the previous report [34] without reaching the cell density or showing the appearance of DO oscillations (Table 2). These results suggested that the TDH1 gene is the most likely one required for metabolic oscillation, while the others are predominantly engaged in the growth of cells. Effect of TDH gene expression on the rescue of the metabolic oscillations in tdh1 D To examine the possibility that TDH1 is involved in the regulation of metabolic oscillations, we examined whether the expression of the TDH1 gene in tdh1D can rescue the disappearance of biological rhythms by transforming the centromeric recombinant plasmid carrying TDH1 with its upstream region of about 0.8 kbp. The energy-metabolism oscillation appeared and continued for about 4 days (Fig. 6B), although the wavelength of the oscillation was about 35% shorter and the amplitude about 30% lower than in the wild-type cell (Table 2). In contrast, when TDH2 or TDH3 was expressed in tdh1D under the control of the promoter of TDH1, the cell growth of either transformant was even more disturbed than that of the parental tdh1D, and thus continuous cultures could not be started (Table 2). Thus, GAPDH2 and GAPDH3 could not replace GAP- DH1 to rescue the DO oscillation in tdh1D. DISCUSSION Using the two-hybrid system to search for Gts1p-binding proteins with a yeast cDNA library, only TDH3 cDNAs encoding GAPDH3 were cloned. However, three other results in this report suggested that GAPDH1 interacts with Gts1p to regulate ultradian oscillations in yeast. First, the mRNA level of GAPDH1 fluctuated in concert with energy metabolism whereas those of GAPDH2 and GAPDH3 were constant. Secondly, inactivation of TDH1 caused the disappearance of the metabolic oscillation without affecting growth of the yeast, whereas inactivation of either TDH2 or TDH3 was considered to primarily cause disturbance of cell growth. Thirdly, disappearance of metabolic oscillation in tdh1D was rescued by transformation of TDH1 but not by transformation of the other genes. This discrepancy is probably explained by the difference in mRNA abundance in the cDNA library, with TDH3 mRNA being the most and TDH1 mRNA the least abundant (< 10%) in vegetatively growing cells [36]. Although there are several substitutions of amino acids in the C-terminal 97 residues between GAPDH1 and the others, they are all very homologous. Furthermore, GAPDH1 and GAPDH3 did not show any differences in terms of ability to bind Gts1p in the two-hybrid assay. This lack of difference suggested that the C-terminal region of GAPDH1 possesses the same binding activity as that of GAPDH3. However, the result cannot rule out the possibility that GAPDH1 has a preference for Gts1p, as all GAPDH proteins exist as homotetramers in vivo, whereas they are monomers in the two-hybrid system. Alternatively, the subcellular localiza- tion of GAPDH1 may be different from others so that it preferentially associates with Gts1p in vivo. In this report, we suggested that Gts1p interacted with GAPDH via the zinc finger in the N-terminal region and the dimerization site in the C-terminal region. However, whether or not these two regions are located adjoining each other is not known because of lack of knowledge of the 3D structure. In addition, we cannot explain why deletion of the ClaI–ClaI region spanning these structures increased the ability to bind GAPDH in the two-hybrid analysis. However, it is pointed out that the binding sites are related, if not directly, to nucleotide binding, as the zinc finger is homologous to that of GTPase-activating proteins of ADP-ribosylation factors [28,30,31] and the dimerization domain is similar to the sequence downstream of Walker’s motif A of some ABC transporters [32,38]. The deletion of the ClaI–ClaIregion may facilitate interaction. Alternatively, as the ClaI–ClaI region contains a ubiquitin-association domain (residues 193–234) [39], which was found to be involved in a rapid degradation of Gts1p itself ( 3 T. Saito, K. Mitsui, Y. Hamada & K. Tsurugi, unpublished results), deletion of the region may stabilize the protein (Gts1p-DC) causing elevation of the protein level. In fact, the Gts1p-DC level was elevated in most transformed cells in the experiments of the two-hybrid assay. Fig. 7. Representative patterns of the DO oscillation in continuous cul- tures of TDH2-deleted mutant (tdh2D)(A)andtheTDH1 (s) and TDH3 mRNA levels (d) during the DO oscillation (B). Continuous culture was started at time zero and continued at a dilution rate of 0.1 h )1 with a synthetic medium containing 1% glucose at 30 °C. In (A), the horizontal bar indicates the time when cell samples for (B) were collected. Ó FEBS 2002 Gts1p–GAPDH1 interaction in metabolic oscillations (Eur. J. Biochem. 269) 3567 We previously reported that two acidic amino-acid residues which are preceded by hydrophobic amino acids were important for the interaction between the dimeriza- tion domains of Gts1p and some ABC transporters [32]. In this report, of the four positive clones of GAPDH genes, the shortest one encoding 76 C-terminal amino-acid residues had only weak binding activity compared with the one encoding 97 residues. Thus, it is likely that a core binding site of GAPDH exists 97–76 residues from the C-terminus. The sequence of the region (237- VDVSVVDLTVKLDKETTYDEI-257) is rich in acidic amino acids preceded by hydrophobic amino acids, although the overall homology to the Gts1p dimerization region is low. Possibly, the dimerization domain plays a role in the interaction with various proteins via as yet unidentified mechanisms, and so Gts1p can have pleio- tropic effects on yeast phenotypes. The glycolytic enzyme GAPDH is one of the most abundant proteins and is generally considered to be constitutively expressed. In higher organisms, one copy of the gene encoding GAPDH is expressed (reviewed [40]) and has been reported to have a variety of activities including apoptosis, nuclear RNA transport, microtubule assembly, protein kinase reactions, and DNA replication [40]. Fur- thermore, it should be pointed out that the protein level of GAPDH in Neurospora crassa [41] and that of chloroplast GAPDH in the dinoflagellate Gonyaulax polyhedra [42] have been reported to undergo circadian oscillation. On the other hand, yeast contains three members of the GAPDH family, and GAPDH1 has been suggested to have a unique function among the GAPDH proteins, as it is predomin- antly synthesized in stationary phase or stressed cells [36,37] and alone it cannot support growth [34]. Structurally, while there are no nonhomologous substitutions between GAPDH2 and GAPDH3, there are seven such substitutions between GAPDH1 and the others. In particular, although GAPDH1 does not have N-degron 4 , the N-terminal amino acid of GAPDH1 is isoleucine, which is known as a destabilizing residue according to the N-end rule [43], and thus differs from the others, which have the stabilizing residue valine. Thus, from the structural viewpoint, it is possible that GAPDH1 plays a unique role among the three GAPDHs. Here, we showed that the GAPDH1 level fluctuated in concert with metabolic oscillation, and that deletion of the gene resulted in disappearance of the oscillation. Thus, we hypothesized that, in yeast, the metabolic oscillator drives other oscillators via the interac- tion of Gts1p and GAPDH1, but the precise molecular mechanism remains to be clarified. In addition, as the mutations in GAPDH affected cyanide-induced glycolytic oscillation, which supposedly is not coupled to other oscillators, it is possible that Gts1p plays some direct role in the oscillation of the glycolytic pathway. Under- standing the biological rhythms in yeast will contribute to understanding them in other organisms, as it has been shown that the energy metabolism pathway is an autogen- ous oscillation in dissipative structures including all living organisms [5–7]. REFERENCES 1. Aldridge, J. & Pye, E.K. (1976) Cell density dependence of oscil- latory metabolism. Science 259, 670–671. 2. Aon, M.A., Cortassa, S., Westerhoff, H.V. & van Dam, K. (1992) Synchrony and mutual stimulation of yeast cells during fast gly- colytic oscillations. J. Gen. Microbiol. 138, 2219–2227. 3. Richard, P., Teusink, B., Westerfhoff, H.V. & van Dam, K. (1993) Around the growth phase transition S. cerevisiae’s make-up favours sustained oscillations of intracellular metabolites. FEBS Lett. 318, 80–82. 4. Chance, B., Hess, B. & Betz, A. (1964) DPNH oscillation in a cell- free extract of S. Carlsbergensis. Biochem. Biophys. Res. Commun. 16, 182–187. 5. Goldbeter, A. (1996) Biochemical Oscillations and Cellular Rhythms. Cambridge University Press, Cambridge. 6. Prigogine, I., Lefever, R., Goldbeter, A. & Herschkowitz-Kauf- man, M. (1969) Symmetry-breaking instabilities in biological systems. Nature 223, 913–916. 7. Nicolis, G. & Prigogine, I. (1977) Self-organization in non- equilibrium systems. From Dissipative Structures to Order Through Fluctuations. Wiley, New York. 8. Higgins, J. (1964) A chemical mechanism for oscillation of gly- colytic intermediates in yeast cells. Proc. Natl Acad. Sci. USA 51, 989–994. 9. Hess, B. & Boiteux, A. (1971) Oscillatory phenomena in bio- chemistry. Annu. Rev. Biochem. 40, 237–258. 10. Parulekar, S.J., Semones, G.B., Rolf, M.J., Lievense, J.C. & Lim, H.C. (1986) Induction and elimination of oscillations in continuous cultures of Saccharomyces cerevisiae. Biotechnol. Bioeng. 28, 700–710. 11. Stra ¨ ssle, C., Sonnleitner, B. & Fiechter, A. (1988) A predictive model for spontaneous synchronization of Saccharomyces cere- visiae grown in continuous culture. I. Concept. J. Biotechnol. 7, 299–318. 12. Porro, D., Martegani, E., Ranzi, B.M. & Alberghina, L. (1988) Oscillations in continuous cultures of budding phase: a segregated parameter analysis. Biotechnol. Bioeng. 32, 411–417. 13. Chen, C.I. & McDonald, K.A. (1990) Oscillatory behavior of Saccharomyces cerevisiae in continuous culture. II. Analysis of cell synchronization and metabolism. Biotechnol. Bioeng. 36, 28–38. 14.Wang,J.,Liu,W.,Uno,T.,Tonozuka,H.,Mitsui,K.& Tsurugi, K. (2000) Cellular stress responses oscillate in syn- chronization with the ultradian oscillation of energy metabolism in the yeast Saccharomyces cerevisiae. FEMS Microbiol. Lett. 189, 9–13. 15. Mu ¨ nich, T., Sonnleitner, B. & Fiechter, A. (1992) The decisive role of the Saccharomyces cerevisiae cell cycle behavior for dynamic growth characterization. J. Biotechnol. 22, 329–352. 16. Wang, J., Liu, W., Mitsui, K. & Tsurugi, K. (2001) Evidence of the involvement of the GTS1 gene product in the regulation of bio- logical rhythms in the continuous culture of the yeast Saccharo- myces cerevisiae. FEBS Lett. 489, 81–86. 17. Sohn, H., Murray, D.B. & Kuriyama, H. (2000) Ultradian oscil- lation of Saccharomyces cerevisiae during aerobic continuous culture: hydrogen sulphide mediates population synchrony. Yeast 16, 1185–1190. 18. Jackson, F.R., Bargiello, T.A., Yun, S.H. & Young, M.W. (1986) Product of per locus of Drosophila shares homology with pro- teoglycans. Nature 320, 185–188. 19. Mitsui, K., Yaguchi, S. & Tsurugi, K. (1994) The GTS1 gene, which contains a Gly-Thr repeat, affects the timing of budding and cell size of the yeast Saccharomyces cerevisiae. Mol. Cell Biol. 14, 5569–5578. 20. Yaguchi, S., Mitsui, K. & Tsurugi, K. (1997) Reexamination of the nucleotide sequence of GTS1, a candidate clock-related gene of Saccharomyces cerevisiae. Biochem. Arch. 13, 97–105. 21. Steeves, T.D., King, D.P., Zhao, Y., Du Sangoram, A.M.F., Bowcock, A.M., Moore, R.Y. & Takahashi, J.S. (1999) Molecular cloning and characterization of the human CLOCK gene: expression in the suprachiasmatic nuclei. Genomics 57, 189–200. 3568 W. Liu et al.(Eur. J. Biochem. 269) Ó FEBS 2002 22. Yaguchi,S.,Mitsui,K.,Kawabata,K.,Xu,Z.&Tsurugi,K. (1996) The pleiotropic effect of the GTS1 gene product on heat tolerance, sporulation and the life span of Saccharomyces cerevi- siae. Biochem. Biophys. Res. Commun. 218, 234–237. 23. Dunlap, J.C. (1993) Genetic analysis of circadian clocks. Ann. Rev. Physiol. 55, 683–728. 24. Hall, J.C. (1990) Genetics of circadian rhythms. Ann. Rev. Genet. 24, 659–697. 25. Wang, J., Mitsui, K. & Tsurugi, K. (1998) The GTS1 gene product influences the ultradian oscillation of glycolysis in cell suspension of the yeast Saccharomyces cerevisiae. Biochem. Biophys. Res. Commun. 244, 239–242. 26.Uno,T.,Wang,J.,Mitsui,K.,Umetani,K.,Tamura,K.& Tsurugi, K. (2002) Ultradian rhythm of trehalose levels coupled to heat resistance in continuous cultures of the yeast Saccharomyces cerevisiae. Chronobiol. Int. 19, 361–375. 5 27. Tonozuka,H.,Wang,J.,Mitsui,K.,Saito,T.,Hamada,Y.& Tsurugi, K. (2001) Analysis of the upstream regulatory region of the GTS1 gene required for its oscillatory expression. J. Biochem. 130, 589–595. 28. Ireland, L.S., Johnston, G.C., Drebot, M.A., Dhillon, N., DeMaggio, A.J., Hoekstra, M.F. & Singer, R.A. (1994) A member of a novel family of yeast ÔZn-fingerÕ proteins mediates the trans- ition from stationary phase to cell proliferation. EMBO J. 13, 3812–3821. 29. Mewes, H.W., Albermann, K., Bahr, M., Fishman, D., Gleissner, A., Hani, J., Heumann, K., Kleine, K., Maierl, A., Oliver, S.G., Pfeiffer, F. & Zollner, A. (1997) Overview of the yeast genome. Nature 387, 7–65. 30. Poon, P.P., Wang, X., Rotman, M., Huber, I., Cukierman, E., Cassel, D., Singre, R.A. & Johnston, G.C. (1996) Saccharomyces cerevisiae Gcs1 is an ADP-ribosylation factor GTPase-activating protein. Proc. Natl Acad. Sci. USA 93, 10074–10077. 31. Blader, I.J., Jamie, M., Cope, T.V., Jackson, T.R., Profit, A.A., Greenwood, A.F., Drubin, D.G., Prestwich, G.D. & Theibert, A.B. (1999) GCS1, an Arf guanosine triphosphatase-activating protein in Saccharomyces cerevisiae, is required for normal actin cytoskeletal organization in vivo and stimulates actin poly- merization in vitro. Mol. Biol. Cell 10, 581–596. 32. Kawabata,K.,Mitsui,K.,Uno,T.,Tamura,K.&Tsurugi,K. (1999) Protein interactions of Gts1p of Saccharomyces cerevisiae throughout a region similar to a cytoplasmic portion of some ATP-binding cassette transporters. Eur. J. Biochem. 259, 112–119. 33. Holland, J.P., Labienice, L., Swimmer, C. & Holland, M.J. (1983) Homologous nucleotide sequences at the 5¢ termini of messenger RNAs synthesized from the yeast enolase and glyceraldehyde- 3-phosphate dehydrogenase gene families. J. Biol. Chem. 258, 5291–5299. 34. McAlister, L. & Holland, M.J. (1985) Isolation and character- ization of yeast strains carrying mutations in the glyceraldehyde- 3-phosphate dehydrogenase genes. J. Biol. Chem. 260, 15013– 15018. 35. Naumovski, L. & Friedberg, E.C. (1983) Construction of plasmid vectors that facilitate subcloning and recovery of yeast and Escherichia coli DNA fragments. Gene 22, 203–209. 36. McAlister, L. & Holland, M.J. (1985) Differential expression of the three yeast glyceraldehyde-3-phosphate dehydrogenase genes. J. Biol. Chem. 260, 15019–15027. 37. Boucherie,H.,Bataille,N.,Fitch,I.T.,Perrot,M.&Tuite,M.F. (1995) Differential synthesis of glyceraldehyde-3-phosphate dehydrogenase polypeptides in stressed yeast cells. FEMS Microbiol. Lett. 125, 127–133. 38. Decottignies, A. & Goffeau, A. (1997) Complete inventory of the yeast ABC proteins. Nat. Genet. 15, 137–145. 39. Hofmann, K. & Bucher, P. (1996) The UBA domain: a sequence motif present in multiple enzyme classes of the ubiquitination pathway. Trends Biochem. Sci. 21, 172–173. 40. Sirover, M.A. (1999) New insight into an old protein: the func- tional diversity of mammalian glyceraldehyde-3-phosphate dehy- drogenase. Biochim. Biophys. Acta 1432, 159–184. 41. Shinohara, M.L., Loros, J.J. & Dunlap, J.C. (1998) Glycer- aldehyde-3-phosphate dehydrogenase is regulated on a daily basis by the circadian clock. J. Biol. Chem. 273, 446–452. 42. Fagan, T., Morse, D. & Hastings, J.W. (1999) Circadian synthesis of a nuclear-encoded chloroplast glyceraldehyde- 3-phosphate dehydrogenase in the dinoflagellate Gonyaulax polyhedra is translationally controlled. Biochemistry 38, 7689– 7695. 43. Varshavsky, A. (1996) The N-end rule: functions, mysteries, uses. Proc. Natl Acad. Sci. USA 93, 12142–12149. Ó FEBS 2002 Gts1p–GAPDH1 interaction in metabolic oscillations (Eur. J. Biochem. 269) 3569 . Interaction of the GTS1 gene product with glyceraldehyde- 3-phosphate dehydrogenase 1 required for the maintenance of the metabolic oscillations of the yeast Saccharomyces cerevisiae Weidong. previously [14 ,16 ]. GTS1[ DKN] and GTS1[ C5 3Y] inserted into pAUR 112 were transformed into gts1D and the transformants were named pACGTS1 [DKN]/gts1D and pACGTS1[C5 3Y] /gts1D, respectively. Preparation of TDH -deleted. severely shortened when the wild-type GTS1 was overexpressed in gts1D (pYXGTS1/gts1D) (Fig. 3). When gts1D cells overexpressed Gts1p-C5 3Y or Gts1p-DKN, the duration of the oscillations was lengthened Fig.