Mechanism for the degradation of origin recognition complex containing Orc5p with a defective Walker A motif and its suppression by over-production of Orc4p in yeast cells

Abstract

Orc5p is one of six subunits constituting the ORC (origin recognition complex), a possible initiator of chromosomal DNA replication in eukaryotes. Orc5p contains a Walker A motif. We recently reported that a strain of Saccharomyces cerevisiae having a mutation in Orc5p's Walker A motif (orc5-A), showed cell-cycle arrest at G₂/M and degradation of ORC at high temperatures (37 °C). Over-production of Orc4p, another subunit of ORC, specifically suppressed these phenotypes [Takahashi, Yamaguchi, Yamairi, Makise, Takenaka, Tsuchiya and Mizushima (2004) J. Biol. Chem. 279, 8469–8477]. In the present study, we examined the mechanisms of ORC degradation and of its suppression by Orc4p over-production. In orc5-A, at high temperatures, ORC is degraded by proteasomes; either addition of a proteasome inhibitor, or introduction of a mutation of either tan1-1 or nob1-4 that inhibits proteasomes, prevented ORC degradation. Introduction of the tan1-1 mutation restored cell cycle progression, suggesting that the defect was due to ORC degradation by proteasomes. Yeast two-hybrid and co-immunoprecipitation analyses suggested that Orc5p interacts preferentially with Orc4p and that the orc5-A mutation diminishes this interaction. We suggest that this interaction is mediated by the C-terminal region of Orc4p, and the N-terminal region of Orc5p. Based on these observations, we consider that ATP binding to Orc5p is required for efficient interaction with Orc4p and that, in orc5-A, loss of this interaction at higher temperatures allows proteasomes to degrade ORC, causing growth defects. This model could also explain why over-production of Orc4p suppresses the orc5-A strain's phenotype.

INTRODUCTION

The initiation of chromosomal DNA replication must be tightly regulated, and co-ordinated with cell division, to replicate the genome just once per cell cycle. In Escherichia coli, adenine nucleotides bound to the DnaA protein, the initiator of chromosomal DNA replication, play a major role in this regulation. DnaA has a high affinity for both ATP and ADP. The ATP–DnaA complex is active for DNA replication both in vivo and in vitro, but the ADP–DnaA complex is inactive [1–3]. DnaA has intrinsic ATPase activity and thus inactivates itself, suppressing over-initiation [4,5]. Acidic phospholipids interact with DnaA via conserved basic amino acid residues, and this re-activates the ADP–DnaA complex, and suppresses over-initiation of DNA replication [6–10].

In eukaryotes, ORC (origin recognition complex), a possible initiator of chromosomal DNA replication, binds to ATP and ADP [11–14]. ORC was originally identified as a six-protein complex that specifically binds to Saccharomyces cerevisiae origins of chromosomal DNA replication [11]. Homologues have been found in various eukaryotes, including humans [15]. In S. cerevisiae, ORC's ATP-binding activity has been well characterized in vitro. ORC has two subunits (Orc1p and Orc5p), which bind ATP. Orc1p, but not Orc5p, has ATPase activity [12]. Orc5p, but not Orc1p, can bind to ADP [13]. ATP binding to Orc1p, but not to Orc5p, is essential for specific binding of ORC to origin DNA [12,14]. ATP binding to Orc5p increases the affinity of Orc1p for ATP [14]. In contrast with these biochemical studies, ATP binding to ORC has not been examined extensively by genetic techniques, and therefore its role in vivo remains unknown.

We previously reported that a strain of S. cerevisiae having a mutation in the Walker A motif (consensus sequences for ATP-binding protein) of Orc5p (orc5-A) showed, at high temperatures, decreased levels of ORC, cell cycle arrest at G₂/M, and a growth defect [16]. In the present study, experiments using a proteasome inhibitor, and a mutation (tan1-1 or nob1-4) that inhibits proteasomes, suggested that the decreased ORC level is mediated by degradation by proteasomes. Since the tan1-1 mutation also suppressed the defects in cell cycle progression and cell growth in orc5-A, we presume these are also due to increased ORC degradation.

In orc5-A, over-production of Orc4p, another subunit of ORC, can also suppress the defects seen at high temperature [16]. This suppression was specific to orc5-A: Orc4p did not restore function in orc5-1, another temperature-sensitive strain that has a mutation outside the ATP-binding domain [16]. Furthermore, the suppression was specific to Orc4p; over-production of each of the other subunits (Orc1p, Orc2p, Orc3p and Orc6p) did not restore function [16]. These results revealed that Orc4p and ATP binding to Orc5p are closely linked. One possibility is that Orc5p can preferentially interact with Orc4p via its Walker A motif domain. In the present study, we used yeast two-hybrid analysis to suggest that Orc5p's interaction with Orc4p depends on the Walker A motif of Orc5p. We therefore consider that ATP binding to Orc5p is important for Orc5p to interact with Orc4p, which in turn is important for the stability of ORC in cells, and hence DNA replication and cell growth. We also suggested that the Orc4p–Orc5p interaction is mediated by the C-terminal region of Orc4p and the N-terminal region of Orc5p that contains the Walker A motif.

EXPERIMENTAL

Strains, plasmids and medium

S. cerevisiae strains are listed in Table 1 [16–20]. Cells were cultured in SC (synthetic complete) medium.

Table 1. Yeast strains.

Strains	Genotypes	Reference
EGY48	MATα trp1 ura3 his3 6lexAop-LEU2	[18]
FY24	MATα trp1δ63 ura3-52 leu2δ1	[17]
tan1-1	MATα ura3-52 leu2δ1 tan1-1	[17]
W303-1A	Mataade2-1 can1-100 his3-11,15 leu2-3,112 trp1-1 ura3-1	[19]
YY411	Mataade2-1 can1-100 his3-11,15 leu2-3,112 trp1-1 ura3-1 orc5-A	[16]
YY412	Mataade2-1 can1-100 his3-11,15 leu2-3,112 trp1-1 ura3-1 orc5-A orc2-3HA:	The present study
Y138	Mataade2-1 can1-100 his3-11,15 leu2-3,112 trp1-1 ura3-1 nob1-4::LEU2	[20]
NT411	MATα trp1δ63 ura3-52 leu2δ1 orc5-A	The present study
NT412	MATα trp1δ63 ura3-52 leu2δ1 orc5-A tan1-1	The present study
NT422	Mataade2-1 can1-100 his3-11,15 leu2-3,112 trp1-1 ura3-1 orc5-A nob1-4::LEU2	The present study

The orc5-A gene (orc5K43E) was introduced into pRS406 to create pRS406-orc5-A [16]. This plasmid was transformed into FY24 and tan1-1 strains and transformed cells were selected on SC agar plates lacking uracil. The resultant strains were transferred to plates containing 5-fluoro-orotic acid (the two-step gene replacement method) [21]. After checking by PCR, the resultant strains were named NT411 (FY24, orc5-A) and NT412 (FY24, tan1-1, orc5-A) respectively. We introduced the nob1-4 mutation into the orc5-A mutant to construct NT422 strain (W303-1A, orc5-A, nob1-4).

Modifications of ORC2 and ORC6 genes with 3×HA (haemagglutinin) were performed as described previously [22]. PCR was performed using pFA6a-3HA-TRP1 or pFA6a-3HA-His3MX6 plasmid as a template and primers containing DNA sequences for the C-terminal region of ORC6 or ORC2 gene respectively. The amplified DNA was transformed into YY411. The construct of the resultant strain (YY412) was confirmed by the colony PCR method.

FACS analysis of cellular DNA content

Samples were prepared as previously described [23,24], with the following modifications. Cells were pelleted by centrifugation, washed with sterilized water, and fixed in 70% (v/v) ethanol for 12 h. Cells were again pelleted, resuspended in 50 mM sodium citrate, sonicated for 1 min, treated with 0.25 mg/ml RNaseA for 1 h at 50 °C, and then with 1 mg/ml proteinase K for 1 h at 50 °C. DNA was stained with 50 μg/ml of propidium iodide, and 20000 cells from each sample were scanned with a FACSCalibur instrument (Becton Dickinson).

Assessment of ORC subunits in yeast chromatin

Yeast spheroplasts were lysed with Triton X-100, and samples were processed into soluble (supernatant) and chromatin (insoluble precipitate) fractions by centrifugation, as previously described [24]. Equivalent amounts (total protein) of chromatin fractions were subjected to electrophoresis on 7.5 or 10% polyacrylamide gels containing SDS, transferred to a PVDF membrane, and probed with monoclonal antibodies against Orc3p (SB3), Orc5p (SB5) and HA (12CA5) [24,25].

Yeast two-hybrid analysis

Plasmids pSH18-34 (a reporter plasmid in which the lacZ gene is located downstream of a 6-fold repeated operator of the lexA gene), pEG202 [a plasmid to express BD fusions (binding domain fusions) under the control of the ADH promoter] and pJG4-5 [a plasmid to express AD fusions (activation domain fusions) under the control of a GAL1 promoter] were purchased from OriGene Technologies. In pJG4-5, the ampicillin resistance gene was replaced by a chloramphenicol resistance gene, to construct pJG4-5Cm^R. DNA fragments containing intact or partial ORF (open reading frame) for each ORC subunit were prepared by PCR and inserted into pEG202 or pJG4-5Cm^R.

EGY48 cells were transformed with pSH18-34, a pEG202 derivative and a pJG4-5Cm^R derivative.

β-Galactosidase assay: cells were collected by centrifugation, suspended with 1 ml of Z buffer (100 mM phosphate buffer, pH 7.0, 10 mM KCl, 1 mM MgSO₄ and 50 mM 2-mercaptoethanol) to which were added three drops of CHCl₃ and one drop of 0.1% SDS. After mixing and incubation for 5 min at 37 °C, the β-galactosidase reaction was started by addition of 0.2 ml of ONPG (o-nitrophenyl β-D-galactopyranoside) and stopped by addition of 0.5 ml of 1 M Na₂CO₃. After centrifugation, the absorbance at 420 nm of supernatant was determined. One unit of β-galactosidase hydrolyses 1 μmol of ONPG per minute.

Co-immunoprecipitation assay

Yeast whole cell extract was prepared by the standard glass beads method in the presence of protease inhibitors. Whole cell extract was incubated with Dynabeads Protein G (Dynal Biotech) conjugated with an antibody against HA (12CA5) for 1 h at 4 °C with rotation. After centrifugation, precipitates were washed and finally suspended with SDS sample buffer (150 mM Tris/HCl, pH 6.8, 50% glycerol, 1.4 M 2-mercaptoethanol, 5% SDS and 0.0025% bromophenol blue). Samples were analysed by immunoblotting with an antibody against HA or LexA.

RESULTS

In orc5-A, ORC degradation depends on the proteasome system

Most intracellular proteins are degraded, in a controlled way, by the ubiquitin–proteasome system. Polyubiquitinated proteins are degraded by 26 S proteasomes, which are composed of a 20 S proteasome core and a 19 S regulatory subunit [26–29]. We here examined whether the proteasome system degrades ORC in orc5-A (YY411) at high temperatures. After temperature shift-up (from 24 to 37 °C), both Orc3p and Orc5p disappeared from orc5-A cells (Figure 1A), as described previously [16]. Pre-incubating cells with a proteasome inhibitor [MG-132 (the proteasome inhibitor carbobenzoxy-L-leucyl-L-leucyl-leucinal)] [30,31] completely suppressed this disappearance (Figure 1A). We also examined the state of ORC subunits other than Orc3p and Orc5p under these conditions. As shown in Figure 1(B), Orc2p and Orc6p were decreased in orc5-A cells at 37 °C and this phenotype was suppressed by pre-incubation of cells with MG-132. Although we could not detect Orc1p and Orc4p by immunoblotting even in the wild-type strain, we presume that all ORC subunits disappear from orc5-A cells at 37 °C.

Figure 1. Effect of a proteasome inhibitor on ORC degradation in orc5-A cells.

YY411 and YY412 (in which Orc2p and Orc6p are tagged with HA) cells were cultured in SC medium to exponential phase at 24 °C and further incubated in the presence or absence of MG-132 (10 μg/ml) for 1 h. Then the incubation temperature was changed to 37 °C, and the culture was sampled as shown. Chromatin fractions were prepared and analysed by immunoblotting using monoclonal antibodies specific for Orc3p, Orc5p or HA. As loading controls, samples were stained with silver (A, B). Control samples (shown as ‘C’) were taken before the MG-132 treatment (A).

The intensity of the Orc5p band increased after addition of MG-132 (Figure 1A). This increase was diminished by pre-incubation of cells with cycloheximide (an inhibitor of protein synthesis; results not shown), suggesting that this increase is due to the stimulation by MG-132 of the expression of Orc5p.

To test further whether the proteasome system was responsible for the disappearance of ORC, we examined the effect of a mutation which disrupts the proteasome system. The tan1-1 mutation is located in the PRE8 gene, which encodes one of the subunits of the 20 S proteasome core [17]. We introduced the orc5-A mutation into the tan1-1 mutant, and into its corresponding wild-type strain (FY24), to construct strains NT412 and NT411 respectively (Table 1). Similar to experiments using the W303-1A genetic background, in NT411, the orc5-A mutation caused ORC disappearance at high temperatures (Figure 2A). However, ORC did not disappear from NT412 (FY24, orc5-A, tan1-1) cells, indicating that tan1-1 stabilized ORC5-A (ORC containing the protein product of the orc5-A gene). Results in Figures 1(A) and 2(A) strongly suggest that in the orc5-A strain, ORC disappearance at high temperatures is mediated by the proteasome system. We confirmed that 37 °C is a semi-permissive temperature for the tan1-1 mutant; the mutant could grow at 37 °C (Figure 4) but not at 38.5 °C (results not shown) and results in Figure 2(A) suggested that the proteasome system is partially inactivated at 37 °C in the mutant, this being consistent with results in a previous paper [17].

Figure 2. Effect of the tan1-1 or nob1-4 mutation on ORC degradation caused by the orc5-A mutation.

FY24 (wild-type), NT411 (orc5-A) and NT412 (orc5-A, tan1-1) (A) or W303-1A (wild-type), YY411 (orc5-A) and NT422 (orc5-A, nob1-4) (B) cells were cultured in SC medium and sampled, and chromatin fractions were prepared and analysed by immunoblotting, as described for Figure 1.

Figure 4. The effect of the tan1-1 mutation on temperature-sensitive growth caused by the orc5-A mutation.

NT412 (orc5-A, tan1-1), NT411 (orc5-A), FY24 (wild-type) and the tan1-1 strain were incubated on SC agar plates at 24 °C or 37 °C for 2 days.

The nob1-4 is a mutation in the gene encoding a protein that binds to the 19 S regulatory subunit and is essential for the ubiquitin–proteasome system. Furthermore, this mutation causes a temperature-sensitive growth phenotype [20]. We introduced the orc5-A mutation into the nob1-4 mutant to construct strain NT422. As shown in Figure 2(B), both Orc3p and Orc5p were decreased in YY411 (W303-1A, orc5-A) but not in NT422 (W303-1A, orc5-A, nob1-4), confirming that the ORC disappearance at high temperatures in the orc5-A strain is mediated by the ubiquitin–proteasome system.

We next examined the effect of tan1-1 on other characteristics of orc5-A. In a W303-1A genetic background, orc5-A caused cell cycle arrest at G₂/M phase at high temperatures [16]. FACS analysis of NT411 showed that cells with 2C DNA content (G₂/M phase) accumulated after temperature shift-up (Figure 3), which is consistent with results from YY411 (W303-1A, orc5-A) cells [16]. On the other hand, in NT412 (FY24, orc5-A, tan1-1) cells, populations of both G1 and G₂/M cells (1C and 2C respectively) were observed even 10 h after temperature shift-up (Figure 3). In the tan1-1 single mutant, cell cycle progression was not so affected, when compared with the wild-type strain, FY24 (Figure 3). Thus, in NT412, the tan1-1 mutation suppressed cell cycle arrest at G₂/M phase. We therefore conclude that in orc5-A, cell cycle arrest is probably due to the proteasome-dependent degradation of ORC.

Figure 3. FACS analysis, showing the effect of the tan1-1 mutation on cell cycle arrest caused by the orc5-A mutation.

FY24 (wild-type), NT411 (orc5-A), the tan1-1 strain and NT412 (orc5-A, tan1-1) cells were cultured in SC medium to exponential phase at 24 °C, and then the temperature was increased to 37 °C. The culture was sampled as indicated, and cellular DNA content was analysed by FACS.

We also examined the effect of tan1-1 on temperature-sensitive growth in orc5-A cells. As shown in Figure 4, cells containing the orc5-A mutation (NT411) could not grow at 37 °C, as described previously [16]. Interestingly, the tan1-1 mutation completely restored ability to form colonies at 37 °C (Figure 4), and NT412 cells showed a growth curve similar to that of the wild-type strain (results not shown). Cell growth of the tan1-1 single mutant was similar to that of the wild-type strain (results not shown). All these results strongly suggest that in the orc5-A strain, the growth defect is due to proteasome-dependent degradation of ORC.

Yeast two-hybrid analysis of the interaction between Orc5p and Orc4p in vivo

Our previous studies showed that Orc5p interacts with Orc4p via the Walker A motif region [16]. We have used a yeast two-hybrid system to test this idea. The ORF encoding wild-type Orc5p, or Orc5-Ap, was fused to a lexA DNA-binding domain to construct BD fusions. The ORF encoding each of the ORC subunits (Orc1p–Orc6p) was fused to a B42-HA transcriptional activation domain, to construct AD fusions. Plasmids containing the AD fusion genes and the BD fusion genes are introduced into cells; if the fusion proteins interact, they activate expression of reporter genes, β-galactosidase and the LEU2 gene product, which allows growth on SC agar plates without leucine [32].

First, we studied wild-type Orc5p. Co-expression of a BD fusion of ORC5 (BD-ORC5) and an AD fusion of ORC4 (AD-ORC4) caused a much higher level of the β-galactosidase activity than the vector control, suggesting that Orc5p and Orc4p interact with each other (Figure 5A). Co-expression of BD-ORC5 and AD-ORC2 also caused high level of β-galactosidase activity, showing that Orc5p and Orc2p interact with each other (Figure 5A). Co-expression of BD-ORC5 and AD fused to ORC6 or other ORC genes (ORC1, ORC3) caused little or negligible respectively, stimulation for β-galactosidase activity, suggesting that Orc5p does not interact strongly with these subunits (Figure 5A). We confirmed, by immunoblotting analysis, that all AD and BD fusion proteins were expressed approximately equally (results not shown). Therefore, in cells, Orc5p seems to have the high affinity for Orc4p. It also interacts with itself (Figure 5A).

Figure 5. Interaction between Orc5p and Orc4p, and its dependency on the ATP-binding domain of Orc5p.

(A) EGY48 cells were co-transfected with pSH18-34, a pEG202 derivative (pEG202 alone, or pEG202-ORC5 or pEG202-orc5-A; BD fusions), and a pJG4-5Cm^R derivative (pJG4-5Cm^R alone, or pJG4-5Cm^R-ORC1, -ORC2, -ORC3, -ORC4, -ORC5 or -ORC6; AD fusions). The activity of β-galactosidase in cells was determined and expressed as total units in 1 ml of yeast cell culture with D₆₀₀ (attenuance) value as 1.0. (B) Whole cell extract was prepared from EGY48 cells having pSH18-34, pEG202 (vector) or its derivative (pEG202-ORC5 or pEG202-orc5-A) and pJG4-5Cm^R-ORC4 and precipitated with or without antibody against HA (Orc4p is tagged with HA in pJG4-5Cm^R-ORC4). Each sample [7.5% input of whole cell extract and precipitates with (+) or without (−) antibody against HA] was analysed on the same gel by immunoblotting with an antibody against HA or LexA.

We next studied the orc-5A mutant. Co-expression of BD-orc5-A and AD-ORC4 caused much the same level of β-galactosidase activity as the vector control (Figure 5A), suggesting that the mutation destroys Orc5p's ability to interact with Orc4p. We confirmed that BD-ORC5 and BD-orc5-A were expressed approximately equally (results not shown), and conclude that the interaction between Orc4p and Orc5p requires the ATP-binding domain of Orc5p. Results in Figure 5(A) also suggest that the interaction between Orc5p with itself, and with Orc2p, also requires the ATP-binding domain of Orc5p.

For biochemical confirmation of the interaction between Orc5p and Orc4p, we performed co-immunoprecipitation experiments by use of crude whole cell extracts. Whole cell extracts were prepared from yeast expressing both AD-ORC4 (HA–Orc4p) and BD-ORC5 (LexA–Orc5p) or BD-orc5-A (LexA–Orc5-Ap), immunoprecipitated with an antibody against HA and detected by immunoblotting with an antibody against HA or LexA. As shown in Figure 5(B), approximately equal amounts of HA–Orc4p and LexA–Orc5p or LexA–Orc5-Ap were expressed in each strain and approximately equal amounts of HA–Orc4p were precipitated using the antibody against HA, suggesting that co-immunoprecipitation experiments worked well under the conditions. A significant amount of LexA–Orc5p was co-precipitated with HA–Orc4p, showing that Orc5p physically interacts with Orc4p under the conditions. A smaller amount of LexA–Orc5-Ap than of LexA–Orc5p was co-precipitated with HA–Orc4p, showing that the interaction of Orc4p with Orc5-Ap is weaker than that with LexA–Orc5p. These results suggest that the physical interaction between Orc4p and Orc5p requires the ATP-binding domain of Orc5p.

For further confirmation that Orc5p's interaction with Orc4p depends on the ATP-binding domain of Orc5p, we examined colony formation on SC agar plates without leucine. Co-expression of BD-ORC5 and AD-ORC4 permitted colony formation, confirming that Orc4p and Orc5p interact (results not shown). Cells co-expressing BD-orc5-A and AD-ORC4 fusions had only a weak ability to form colonies, especially at higher temperatures, suggesting that Orc5-Ap and Orc4p might interact weakly (results not shown). There was no clear difference between BD-ORC5 and BD-orc5-A for colony formation on SC agar plates with leucine at any temperatures (results not shown). These results confirm that Orc4p interacts weakly with Orc5-Ap, compared with its interaction with Orc5p, especially at high temperatures. Since in ORC5-A (ORC containing the protein product of the orc5-A gene), Orc5p loses its ability to bind to ATP [12], ATP binding to Orc5p is presumably important for Orc5p and Orc4p to interact.

We then tried to determine the domain of each subunit (Orc5p and Orc4p) involved in the interaction. As shown in Figure 6(A), we created three fragments (N-terminal, middle and C-terminal) of Orc5p and Orc4p, constructed fusions, and examined the interaction between these fragments and an intact subunit. Co-expression of BD-ORC5 and AD-ORC4 caused a high level of β-galactosidase activity (Figure 6B). Co-expression of BD-N-terminal ORC5 (termed ORC51) and AD-ORC4 gave an even higher level of β-galactosidase activity. The other fragments of Orc5p seemed not to interact with Orc4p (Figure 6B). Co-expression of BD-ORC4 and AD-ORC5 caused the high level of β-galactosidase activity (Figure 6C). Co-expression of BD-C-terminal ORC4 (termed ORC43) and AD-ORC5 gave the lower but significant level of β-galactosidase activity (Figure 6C). The other fragments of Orc4p seemed not to interact with Orc5p. We confirmed that all fusion proteins were expressed approximately equally (results not shown). These results suggest that the N-terminal region of Orc5p interacts with the C-terminal region of Orc4p.

Figure 6. Mapping domains involved in the interaction between Orc5p and Orc4p, by using a yeast two-hybrid assay.

Schematic diagram of Orc5p, Orc4p and their partial fragments (A). Amino acid residues of full-length Orc5p and Orc4p and their partial fragments were numbered (A). EGY48 cells co-transfected with pSH18-34, pJG4-5Cm^R derivatives and pEG202 derivatives. The activity of β-galactosidase in cells was expressed as described in the legend of Figure 5 (B–D).

To test this idea, we examined the interaction between Orc43p and Orc51p using a yeast two-hybrid system. As shown in Figure 6(D), we could not detect the significant interaction between BD-ORC43 and AD-ORC51 or BD-ORC51 and AD-ORC43. We confirmed that all fusion proteins were expressed approximately equally (results not shown). We consider a possibility that although the N-terminal region of Orc5p interacts with the C-terminal region of Orc4p, full-length of Orc5p and Orc4p are necessary to recognize the counterpart.

DISCUSSION

In the present study, we examined the mechanism of ORC disappearance from the orc5-A strain at high temperatures. Experiments with a chemical proteasome inhibitor, or with the tan1-1 or nob1-4 mutant, strongly suggested that ORC degradation is mediated by the proteasome system. At present, it is not clear whether Orc5p is the direct target of polyubiquitination, because we could not detect the polyubiquitination of Orc5-Ap in cells under our experimental conditions (results not shown). Observation that the introduction of the tan1-1 mutation in the orc5-A mutant restored cell growth at high temperatures surprised us, because we considered that the growth defects were primarily due to a loss of function or structure of Orc5-A, and that elimination by the proteasome system was secondary to this. Precise identification of defects of the double mutant (orc5-A, tan1-1) is important in order to understand the role of ATP-binding to Orc5p in cells. Although we tried to find these, the mutant's colony formation ability at 37 °C, and cell viability after 37 °C incubation, were much the same as in wild-type cells.

FACS analysis revealed that in the orc5-A strain, at high temperatures, the cell cycle arrests at G₂/M. This arrest was also suppressed by tan1-1, suggesting that it is also due to ORC degradation by proteasomes. In human cells, one ORC subunit, Orc6p, plays an important role in chromosome segregation and cytokinesis. Orc6p localized to kinetochores during mitosis, and inhibition of Orc6p expression resulted in multi-polar spindles, aberrant mitosis and the formation of multinucleated cells [33]. Orc6p could also be required for chromosome segregation and cytokinesis in yeast, and this could explain why ORC degradation in the orc5-A mutant causes cell cycle arrest.

Analysis with a yeast two-hybrid system suggested that Orc5p preferentially interacts with Orc4p. These in vivo results are consistent with results from previous studies using other approaches in vitro. Lee and Bell reported, using protein-DNA cross-linking, that Orc4p and Orc5p are located very close to each other on origins of replication. They also showed that when ORC was purified from over-producing cells, lack of Orc5p led to a loss of Orc4p from purified ORC [34]. Therefore it seems that Orc4p and Orc5p interact both in vitro and in vivo. Kneissl et al. [35] studied the interaction between subunits of mouse ORC by use of a yeast two-hybrid system. The interaction between Orc4p and Orc5p was weak, compared with interactions between other subunits [35]. Thus the structure of ORC seems to be different in yeast and mouse. Analysis with a yeast two-hybrid system also suggested the Orc5p–Orc5p interaction and its dependence on ATP binding to Orc5p. We previously reported that ORC forms a multimer on origin DNA in vitro under the conditions of high concentrations of ORC [25]. It is an interesting idea that ORC forms a multimer in vivo under some conditions. Further in vivo analysis on the state of ORC at distinct phases of the cell cycle or under some environmental conditions is necessary to test this idea.

We suggest here, from use of the yeast two-hybrid system, that the interaction between Orc4p and Orc5p involves an ATP-binding site on Orc5p, and thus presumably depends on ATP bound to Orc5p; since purified Orc5p alone (without other subunits) does not bind ATP in vitro, we could not test the dependence of Orc4p–Orc5p interaction on ATP. In the orc5-A strain, both the defect in Orc4p–Orc5-Ap interaction, and enhanced ORC degradation, were apparent at high temperatures. Loss of the Orc4p–Orc5p interaction has probably led to the degradation of ORC. This idea is interesting because it is connected to another novel idea that ORC degradation is regulated by ATP binding to Orc5p even in wild-type cells. However, up to now, there is no reported evidence suggesting that wild-type ORC becomes unstable under some conditions. Furthermore, we previously reported that the ATP–Orc5p complex is relatively stable compared with the Orc1p–ATP complex [36]. Again, further analysis of the stability of wild-type ORC at distinct phases of the cell cycle or under some environmental conditions is necessary to test this idea.