Regulation of the Gts1p level by the ubiquitination system to maintain metabolic oscillations in the continuous culture of yeast.

Yeast cells exhibit sustained ultradian oscillations of energy metabolism in coupling with cell cycle and stress resistance oscillations in continuous culture. We have reported that the rhythmic expression of Gts1p is important for the maintenance of ultradian rhythms. Structurally, Gts1p contains sequence motifs similar to N-degron and the ubiquitin association domain, raising the possibility that the Gts1p level is regulated by degradation via ubiquitination. When the lysine residue at the putative ubiquitination site of the N-degron was substituted with arginine, both the protein level and half-life of mutant Gts1p increased. During continuous culture, the protein level of the mutant Gts1p was elevated and did not fluctuate, leading to the disappearance of metabolic oscillation within a day. Furthermore, using three Gts1ps containing mutations in the ubiquitin association domain, we showed that the lower the binding activity of the mutant Gts1ps for polyubiquitin in vitro, the higher the protein level in vivo. Expression of the mutant Gts1ps in the continuous culture resulted in an increase in Gts1p and early loss of the oscillation. Therefore, Gts1p is degraded through conjugation with ubiquitin, and the UBA domain promoted the degradation of ubiquitinated Gts1p, causing a fluctuation in protein level, which is required for the maintenance of metabolic oscillations.

In an open system using a bioreactor, yeast cells exhibit sustained ultradian oscillations of energy metabolism in a continuous (chemostat) culture under aerobic and glucose-limited conditions (1)(2)(3)(4). The energy metabolism pathway has been proven to be an autogenous oscillator under extreme nonequilibrium energy conditions according to the theory of "dissipative structures" established by Prigogine and others (5)(6)(7). In brief, the metabolic pathway oscillates autonomously under the primary control of phosphofructokinase, 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 is consumed as fuel for various ATPases, glucose again begins to enter the glycolytic pathway. The oscillations are detectable as a periodic change in the factors involved in energy metabolism such as dissolved oxygen (DO) 1 levels, CO 2 production, glucose and ethanol concentrations, and amounts of storage carbohydrates. DO oscil-lation is due to the periodic change between respiratory and respiro-fermentative phases in which oxygen demands are relatively high and low, respectively. DO oscillations arise spontaneously in concert with cell division and are dependent on a high cell density (ϳ5 ϫ 10 8 cells/ml), regulating the cell density throughout the oscillation (8). We have reported that cellular responses to various stress conditions, such as heat, oxidative agents, and cytotoxic compounds, are regulated in concert with metabolic oscillation (8) and that the coupling 2 was disrupted by inactivation of the GTS1 gene (9). Furthermore, we suggested that the rhythmic expression of Gts1p is more important than protein level for the maintenance of ultradian rhythms, since the constitutive expression of GTS1 under the control of the TPI promoter resulted in the disappearance of ultradian rhythms (9). 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 the attenuation of the metabolic oscillation and to the uncoupling of stress-resistance oscillations (10). Thus, we suggested that, in order for stress resistance oscillations to occur, a full fluctuation in the level of Gts1p is required. Furthermore, we found that the mRNA level of GTS1 fluctuated with a broad peak at the respiro-fermentative phase, which is the opposite of the Gts1p fluctuation (10), suggesting that GTS1 expression was regulated in an oscillatory manner at the transcriptional level. However, it is likely that the oscillatory expression of GTS1 is not sufficient to cause a fluctuation in the protein level, since the protein needs to be degraded for any subsequent increase to occur and/or for regulation of the protein level.
In this paper, we examined whether the fluctuation in the level of Gts1p is controlled by the ubiquitin-proteasome system, since we found that Gts1p contains a putative N-degron (11) and ubiquitin association (UBA) domain (12). Using site-directed mutagenesis at the sequence motifs, we examined the effects of mutations on the fluctuation in Gts1p expression and the oscillations in energy metabolism during continuous culture. The results suggested that the fluctuation in the level of Gts1p is regulated by the ubiquitination system in order to maintain the oscillations of energy metabolism, in combination with the oscillatory GTS1 expression at the transcriptional level.

EXPERIMENTAL PROCEDURES
Yeast Strains and Media-The haploid strain S288C (MAT␣ mal gal2 SUC2) of Saccharomyces cerevisiae was used. FL599 (ise1 Ϫ ) (13) was a gift from Dr. J. Nittis (Harvard University). In the batch cultures, cells were incubated at 30°C in a synthetic medium con-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
sisting of a yeast nitrogen base without amino acids (Difco) containing 2% glucose with the required amino acids and bases.
Conditions for Continuous Culture-The cells were cultured at 30°C, pH 5.5, in a synthetic medium containing 1% glucose as defined elsewhere (2) using an improved bench top fermenter, MDL-6C (Marubishi Bioengineering, Tokyo), at a constant volume of 500 ml (8). Continuous cultures were performed at a dilution rate of 0.1 h Ϫ1 with aeration at 1 liter/h and stirring at 420 rpm. The periodic change in respiro-fermentative metabolism was monitored by measuring the level of DO with an oxygen electrode.
Construction of Mutant GTS1-To obtain the wild type GTS1, a 2.30-kilobase pair XhoI-SpeI fragment containing both GTS1 and a 1.0-kilobase pair upstream region was obtained by PCR directed against genomic DNA using oligonucleotides GTS-Xho and GTS-3ЈDS (Table I and Fig. 1A) as 5Ј-and 3Ј-primers, respectively, and inserted into the multicloning site of the centromere-based vector pRSA103, which is derived from pRS413 (Stratagene, La Jolla, CA) containing AUR1 from pAUR112 (Takara, Tokyo) in place of HIS3 (9). The XbaI site in the multicloning site of the recombinant plasmid was removed by deletion of the SacI-SpeI fragment to obtain pRSA-GTS1.
For construction of GTS1[K17R] encoding Gts1p(K17R) containing a substitution at position 17 (arginine for lysine), a PCR-amplified fragment obtained using GTS-Nhe and K17R-2 as 5Ј-and 3Ј-primer (Table  I), respectively, was ligated with a fragment obtained using K17R-1 and GTS-Spe (Table I) as primers, after digestion with AflIII. The PCR product amplified on the ligated fragment using GTS-Xho and GTS-Spe (Table I) as 5Ј-and 3Ј-primers, respectively, was digested with XhoI and XbaI and inserted into the cognate sites of pRSA-GTS1 to obtain pRSA-GTS1 [K17R]. The mutation in pRSA-GTS1[K17R] was confirmed by sequencing the GTS1 site in the recombinant plasmid and incorporated into the GTS1-disrupted strain gts1⌬ (8,9).
To construct GTS1[TGF] encoding a substitution in MGF between amino acids 205 and 207 of Gts1p, a fragment amplified using TGF and GTS-3ЈSal as 5Ј-and 3Ј-primer (Table I), respectively, was ligated with a fragment amplified using GTS-5ЈRI and GTS-3ЈSal as primers, after digestion with BsrI and SalI, respectively. The EcoRI-SalI segment from the ligated fragment was inserted into the cognate sites of pYX222 (Takara), and the BglII-XbaI fragment obtained from the recombinant plasmid was inserted into the cognate sites of pRSA-GTS1, to obtain pRSA-GTS1 [TGF]. Similarly, constructs GTS1[MAQ] and GTS1 [MGQ] were obtained using oligonucleotides MAQ and MGQ as 5Ј-primers (Table I) Treatment with a Proteasome Inhibitor and Western Blotting-FL599 (ise1 Ϫ ) was cultured until the A 600 value reached 1.0 and mixed with MG132 at a final concentration of 50 M. After incubation for a specific period, cells were harvested, and the protein levels of Gts1p and actin in the cell lysates were determined by Western blotting as described previously (9). The relative protein level of Gts1p to actin was determined with a lumino-image analyzer (LAS1000; Fuji Film, Tokyo).
Determination of mRNA Levels by Northern Blotting-Northern blotting of GTS1 and GTS1[K17R] mRNAs was performed using as probes digoxygenin-labeled PCR products amplified with the DIG labeling kit (Roche Molecular Biochemicals) using GTS-5ЈRI and GTS-3ЈKpn (Table  I) as 5Ј-and 3Ј-primers, respectively. Actin mRNA was detected using the digoxygenin-labeled PCR product as probe, which was amplified using Actin-N and Actin-C (Table I) for 5Ј-and 3Ј-primers, respectively. The relative mRNA level of GTS1 to actin was determined with the lumino-image analyzer Detection of the Gts1p-Ubiquitin Conjugate-The EcoRI-HindIII fragment containing the open reading frame of GTS1 without the stop codon was amplified on genomic DNA using GTS-5ЈRI and GTS-3ЈHind as 5Ј-and 3Ј-primer, respectively, and inserted into the cognate sites of pYX222 containing the sequence of hemagglutinin (HA) epitope in a multicloning site (R&D Systems, Minneapolis, MN). pYXHA-GTS1 carrying a fusion gene construct encoding Gts1p tagged with HA epitope at the C terminus was incorporated into gts1⌬. The transformed cells were harvested when the A 600 of the culture reached 1.0 and cell lysate was prepared by shaking with glass beads in radioimmune precipitation buffer (9). The cell lysate was mixed with Pansorbin (Calbiochem) and anti-HA antibody (Cell Signaling Technology, Berkeley, CA), and the immunoprecipitate was subjected to Western blotting using anti-ubiquitin antibody (Biogenesis, Poole, UK).   Fig. 1A.
The Yeast Two-hybrid System-The two-hybrid analysis was performed using a Matchmaker two-hybrid system (CLONTECH, Palo Alto, CA) as described previously (14). The open reading frame of UBI4 encoding five ubiquitin proteins in a head-to-tail arrangement (15) was amplified on genomic DNA using UBI-5ЈRI and UBI-3ЈSal (Table I) as 5Ј-and 3Ј-primer, respectively, and the EcoRI-SalI fragment from the PCR product was inserted into the cognate sites of pGAD424 in-frame with the gene encoding the activation domain of GAL4. On the other hand, the open reading frames of wild type and mutant GTS1 genes were amplified in the recombinant plasmid pRSA103 using GTS-5ЈRI and GTS-3ЈSal (Table I) as 5Ј-and 3Ј-primer, respectively. The EcoRI-SalI fragments obtained from the PCR products were inserted into the cognate sites of pGTB9 in-frame with the gene encoding the DNAbinding domain of GAL4. The interactions between the prey and bait hybrid proteins were determined for the activation of the lacZ reporter gene by measuring the ␤-galactosidase activity of the cells using the liquid culture assay (14).

Ubiquitin-related Amino Acid Sequences in Gts1p and the Effect of a Proteasome Inhibitor on the Gts1p Level-Gts1p
contains two sequences homologous to ubiquitin-related motifs, N-degron and the UBA domain (Fig. 1). If the first methionine residue is removed from Gts1p in vivo, arginine, one of the destabilizing residues, is located at the N terminus and the ubiquitin-binding residue lysine occupies position 17. In addition, basic residues are present at positions 3 and 10, which have been reported to increase the efficiency of N-degron (16). Thus, the N-terminal sequence of Gts1p almost completely fits the consensus sequence of N-degron if the first methionine residue is removed. On the other hand, the UBA domain con- sists of about 40 amino acid residues comprising three ␣-helices linked by two loops (12,17). The putative UBA domain in Gts1p has 80% homology to the consensus sequence including three well conserved amino acid residues (Gly-206, Leu-217, and Ala-227) (Fig. 1B). The presence of these sequences would suggest that Gts1p is degraded by the ubiquitin-proteasome system. To test this possibility, the effect of the proteasomespecific inhibitor MG132 on the degradation of Gts1p was examined using FL599 (ise1 Ϫ ) cells having enhanced permeability to the drug (13). Western blotting analysis after the treatment with MG132 in vivo showed that the Gts1p level had increased by more than 4-fold 4 h after the treatment with the drug (Fig. 2), suggesting the involvement of proteasomes in the degradation of Gts1p.
Association of Gts1p with Ubiquitin-To investigate whether Gts1p binds ubiquitin in vivo, we first tried to detect ubiquitin-Gts1p conjugates in the wild type cells but could not detect any signals by Western blotting using anti-ubiquitin antibody in the immunoprecipitate with anti-Gts1p antibody (data not shown). Then we constructed a mutant overexpressing HAtagged Gts1p (HA-Gts1p), named TMpHA-GTS1, since we thought that Gts1p-ubiquitin conjugates might be minor components. Western blot analysis of the cell lysate from TMpHA-GTS1 using anti-Gts1p showed that there are several signals in addition to the one corresponding to HA-Gts1p at 45 kDa (Fig.  3A). Although the additional signals could not be found in the Western blot of cell lysate from the wild type cell, we thought that they are possibly some modified forms of Gts1p, since none of them was found in the cell lysate from gts1⌬ (Fig. 3A).
To detect Gts1p-ubiquitin conjugates in TMpHA-GTS1, the cell lysate was immunoprecipitated with anti-HA antibody, and Gts1p-ubiquitin conjugates were detected in the immunoprecipitate by Western blotting using anti-ubiquitin antibody (Fig. 3B). A major signal suggestive of Gts1p-ubiquitin conjugates had a molecular mass of 55 kDa, which corresponds approximately to that of HA-tagged Gts1p (45.6 kDa) conjugated with one molecule of ubiquitin (8.5 kDa) (Fig. 3B, right  lane). Since there is no band at the position corresponding to the 55-kDa signal in the Western blot of the original lysate obtained using anti-Gts1p antibody (Fig. 3B, left lane), the 55-kDa conjugate is thought to be a minor component in the Gts1p population. In addition, none of the signals found in the Western blot of the lysate (Fig. 3, left lane) turned out to be a ubiquitin conjugate. At present, we do not know the nature of these signals. It should be pointed out that there are some minor bands stained like a smear in the high molecular weight region above the major band. Thus, the results suggested that Gts1p is conjugated with ubiquitin in a predominantly monoubiquitinated form.
Effect of Mutation at the Putative Ubiquitin Ligation Site of N-degron of Gts1p on the Metabolic Oscillation-To examine whether the N-degron of Gts1p is involved in its degradation, a mutant Gts1p(K17R) containing arginine at position 17 instead of lysine was expressed in the GTS1-disrupted mutant, gts1⌬, and the level of Gts1p(K17R) was determined (Fig. 4). The level was 2.4-fold that of the wild type protein, whereas the mRNA level was slightly decreased (0.8-fold), suggesting that the mutant Gts1p accumulated due to some deficiency at the step after transcription (Fig. 4). The half-lives of Gts1p and Gts1p(K17R) were then compared after the inhibition of protein synthesis with cycloheximide (Fig. 5). The results showed that the halflife of Gts1p(K17R) was about 12 h, which is ϳ4 times longer than that of the wild type Gts1p (3.5 h), suggesting that Gts1p(K17R) was accumulated due to a defect in degradation. Thus, the N-degron of Gts1p is probably involved in the degradation of Gts1p.
We examined the effect of the K17R mutation in Gts1p on the maintenance of the energy-metabolism oscillation monitored by measuring the concentration of DO in continuous culture (Fig. 6). As we previously reported (9), the wild type Gts1p, expressed with the centromeric recombinant plasmid carrying GTS1[N-C] with its upstream region of about 1.0 kilobase pair rescued the metabolic oscillations lost in gts1⌬ (Fig. 6A). The energy metabolism oscillation continued for about 4 days, al-though the amplitude of DO oscillation was about 30% lower than it was in the wild type cell (Table II). The Gts1p level oscillated at a level nearly identical to that of the wild type cell (data not shown) as reported previously (9). In contrast, the oscillation in gts1⌬ cells expressing Gts1p(K17R) ceased within a day (Fig. 6B). The Gts1p(K17R) level was twice the peak value of Gts1p in the wild type cells during the transient DO oscillation (Fig. 6B). The fluctuation in the level of Gts1p(K17R) was severely restricted (Fig. 6C) compared with that of the wild type Gts1p expressed in either the wild type or gts1⌬ cells transformed with GTS1 (9) (Table II). In addition, the amplitude of the cell cycle oscillation also decreased. These results suggest that the N-degron of Gts1p is involved in the degradation of protein necessary for the fluctuation at the protein level and support the notion that periodic change in the Gts1p level is necessary for the persistent DO oscillation during the continuous culture.
The Binding Activity of the UBA Domain of Gts1p for Ubiquitin-The UBA domain has been identified in various classes of proteins including those in the ubiquitin-proteasome system (12). The function of the domain has not been elucidated, although it was reported that the UBA domains of Rad23p and Ddi1p, DNA-damage-inducible proteins, bind to ubiquitin noncovalently in the yeast two-hybrid system (18). We investigated whether the UBA domain of Gts1p can associate with ubiquitin using the yeast two-hybrid system. When the monoubiquitin gene, which was amplified by polymerase chain reaction from UBI4 encoding five ubiquitin genes in tandem (15), was used as a bait in the system, Gts1p as prey did not bind to the protein (data not shown). However, when the whole UBI4 gene was used as bait, Gts1p did bind to the polyubiquitin protein, although the binding activity as determined using ␤-galactosidase was about half that of Gts1p proteins (14) (Table III). To further determine the binding activity of Gts1p to ubiquitin via the UBA domain, we introduced site-directed mutations in the amino acid triplet MGF between positions 205 and 207, since G-206 is the one of the most conserved residues among UBA sequences (12) (Fig. 1). We constructed four mutant Gts1ps containing TGF, MAF, MGQ, and MAQ in place of MGF, named Gts1p(TGF), and so on. However, since gts1⌬ expressing Gts1p(MAF) grew very slowly, it could not be used in further experiments. We do not know the reason why gts1⌬ expressing Gts1p(MAQ), which contains another mutation at F-207, grew fairly well, but it should be pointed out that MAF is more hydrophobic than the wild type MGF, whereas MAQ retains some hydrophilicity similar to that of MGF. Of the three mutant Gts1ps tested, Gts1p(MAQ) was reduced most in terms of the activity to bind ubiquitin followed by that of a Relative level of Gts1p to actin at the peak and trough of the fluctuations. The Gts1p level in each strain was normalized taking the peak level of the wild type cells as 100.
b Average and S.D. of the amplitude of waves (n Ͼ 20) of the DO oscillation estimated by subtracting the DO concentration (%) of the trough from that of the peak of the next wave.
c Average duration (h) of the DO oscillation. d no, no significant fluctuation in the protein level.
Gts1p(TGF) (Table III). Since the binding activity of Gts1p(MGQ) to ubiquitin did not change as much as that of the wild type, Gly-206 is the most important residue for binding in the three mutant Gts1ps, which is consistent with the fact that Gly-206 is well conserved among UBA sequences. Thus, it is likely that the UBA domain of Gts1p can associate with ubiquitin, as shown in a previous report (18). Effect of Mutations in the UBA Domain of Gts1p on Its Degradation-To examine whether the UBA domain is involved in the degradation of Gts1p itself, the protein levels of the mutant Gts1ps were measured (Fig. 7). Among the three mutant Gts1ps, the protein level of Gts1p(MAQ) was increased the most, followed by that of Gts1p(TGF) (Fig. 7). This was the reverse of the case for the binding activity for polyubiquitin (Table III), showing that the lower the binding activity for polyubiquitin, the lower the degradation capacity of the protein itself.
Effect of Mutations in the UBA Domain of Gts1p on the Metabolic Oscillation-When Gts1p(MAQ), whose activity to bind ubiquitin was most affected among the three, was expressed in gts1⌬ in continuous culture, no oscillation in the concentration of DO was seen (Fig. 8). The Gts1p level was about twice the peak value in the wild type cells and seemed to gradually increase (Fig. 8 and Table II). On the other hand,  gts1⌬ cells expressing Gts1p(TGF) whose activity to bind ubiquitin is intermediate among the three showed a transient DO oscillation, which disappeared within a day ( Fig. 9A and Table  II). The Gts1p(TGF) level during this transient oscillation did not fluctuate and remained equivalent to the peak value of Gts1p (Fig. 9B). In contrast, gts1⌬ cells expressing Gts1p(TGQ) exhibited a metabolic oscillation with a high amplitude similar to that of the wild type cell (Table II). However, the oscillation did not last as long (ϳ3 days) (data not shown), probably because the amplitude of the Gts1p(TGF) level was significantly restricted, although the protein level was similar to that of the wild type cell (Table II). These results suggested that the UBA domain of Gts1p is involved in the degradation of the protein to regulate the fluctuation in the protein level, which is in turn required to maintain the oscillation of energy metabolism.

DISCUSSION
In this report, we suggested that Gts1p was degraded via the ubiquitin-proteasome system through ubiquitination at K-17, although it is not clear whether the first methionine, one of the stabilizing residues according to the N-end rule (11), is removed. In fact, we previously sequenced the N-terminal portion of Gts1p isolated from yeast overexpressing the protein under the control of the TPI promoter in a multicopy vector and found that this protein, if not all proteins, contained the first methionine. 3 However, the result does not eliminate the possibility that the overexpressed Gts1p partly evades excision of the first methionine. The results of the present study showing that Gts1p had a half-life of 3.5 h in the wild type cells after cycloheximide treatment suggested that Gts1p has a destabilizing residue at the N terminus, taking into consideration that inhibition of protein synthesis with cycloheximide generally promotes the stabilization of proteins and mRNAs. Furthermore, the K17R mutation resulted in an increase in the level of Gts1p and a longer half-life, suggesting ubiquitination at Lys-17. In addition, we detected a Gts1p-ubiquitin conjugate in cell lysate, although it was of a monoubiquitinated form. This result is in contrast to other reports (19,20), where overexpressed Myc and x-ray-induced p53 formed multiubiquitin conjugates in a ladder-like fashion. Monoubiquitination has been reported to target membrane-bound proteins for endocytosis, leading to eventual degradation in vacuoles (21,22). However, it is unlikely that Gts1p is degraded in vacuoles, since the protein is located in the supernatant of the cell lysate and the degradation is inhibited by the proteasome inhibitor MG132. It is possible that the Gts1p associated with multiubiquitin chains undergoes rapid degradation.
We showed that the UBA domain of Gts1p is involved in the interaction with ubiquitin using the two-hybrid assay and that the domain promoted the degradation of Gts1p itself. The result of the assay is consistent with the report by Bertolaet et al. (18) in which the UBA domains of Rad23p and Ddi1p bind ubiquitin noncovalently. The result showing that Gts1p did not bind to the ubiquitin monomer in the two-hybrid system is in close agreement with a recent report (23) that indicates that UBA domain-containing proteins like Rad23 and Mud1 bind more than 300 times more tightly to polyubiquitin than to monoubiquitin. The binding of the polyubiquitin of the three mutant Gts1ps at the UBA domain in vitro decreased in parallel with the capacity for degradation of Gts1p in vivo. Gts1p(MAQ) containing a Gly to Ala substitution at position 206, which most affected the association with polyubiquitin, was most deficient in the degradation of Gts1p. The Gly residue is one of the most conserved in the UBA domain and is located in the loop linking ␣-helices 1 and 2 and is known to comprise a part of the hydrophobic surface to which the ubiquitin molecule is thought to bind (18,24). The result suggested that the UBA domain is involved in the degradation of multiubiquitinated Gts1p in an autoregulated (autocatalytic) manner, since the expression of Gts1p(MAQ) which contains an intact N-degron resulted in the complete loss of the fluctuation of Gts1p level in the continuous culture. This may be a reason why multiubiquitinated Gts1p was not detected in cell lysate by Western blotting. However, since this domain is contained in many proteins without N-degron and reportedly interacts with proteins other than ubiquitin (24), the possibility remains that it has as yet unknown functions in addition to the degradation of Gts1p itself.
In this report, we presented evidence that the degradation of Gts1p by the ubiquitin-proteasome system is necessary for a rhythmic expression of the protein. It now remains to be elucidated when and how the ubiquitination system is activated during metabolic oscillation in a continuous culture. Since the ubiquitin-proteasome system requires a large amount of ATP, it is assumed that the system is activated in the respiratory phase when much more ATP is produced in mitochondria than in the respiro-fermentative phase when glycolysis is predominant. Consistently, the decrease of Gts1p begins in the middle of the respiratory phase (9,10). However, since the expression of GTS1 is regulated in an oscillatory manner at the transcription level (10), the regulation at the rhythmic expression of Gts1p may be much more complicated than anticipated here.
It would be interesting to note that the ubiquitin-proteasome system is involved in the regulation of biological rhythm in higher organisms. The three clock-related PAS proteins from Arabidopsis contain the F-box characteristic of proteins that direct ubiquitin-mediated degradation (25,26), and in Drosophila a clock-related protein called timeless (TIM) is known to be degraded by the ubiquitin-proteasome system in response to light (27). These and our results suggested the involvement of the ubiquitin-proteasome system in the regulation of biological rhythms.
In this report, we showed, for the first time, that the ubiquitin-proteasome system is involved in the regulation of the ultradian oscillation of the energy metabolism by means of degrading Gts1p in yeast. These results are very important and suggest for further study on the molecular mechanism of how ultradian rhythms in yeast are regulated. Furthermore, since the energy metabolism pathway has been proven to be an autogenous oscillator in dissipative structures including all living organisms, these data will contribute to studies on the biological rhythms in other organisms.