The Role of Cdh1p in Maintaining Genomic Stability in Budding Yeast

Corresponding author: Orna Cohen-Fix, 8 Center Dr., Bldg. 8, Rm. 319, Bethesda, MD 20892-0840., ornacf{at}helix.nih.gov (E-mail)

Communicating editor: B. ANDREWS

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Cdh1p, a substrate specificity factor for the cell cycle-regulated ubiquitin ligase, the anaphase-promoting complex/cyclosome (APC/C), promotes exit from mitosis by directing the degradation of a number of proteins, including the mitotic cyclins. Here we present evidence that Cdh1p activity at the M/G₁ transition is important not only for mitotic exit but also for high-fidelity chromosome segregation in the subsequent cell cycle. CDH1 showed genetic interactions with MAD2 and PDS1, genes encoding components of the mitotic spindle assembly checkpoint that acts at metaphase to prevent premature chromosome segregation. Unlike cdh1 and mad2 single mutants, the mad2 cdh1 double mutant grew slowly and exhibited high rates of chromosome and plasmid loss. Simultaneous deletion of PDS1 and CDH1 caused extensive chromosome missegregation and cell death. Our data suggest that at least part of the chromosome loss can be attributed to kinetochore/spindle problems. Our data further suggest that Cdh1p and Sic1p, a Cdc28p/Clb inhibitor, have overlapping as well as nonoverlapping roles in ensuring proper chromosome segregation. The severe growth defects of both mad2 cdh1 and pds1 cdh1 strains were rescued by overexpressing Swe1p, a G₂/M inhibitor of the cyclin-dependent kinase, Cdc28p/Clb. We propose that the failure to degrade cyclins at the end of mitosis leaves cdh1 mutant strains with abnormal Cdc28p/Clb activity that interferes with proper chromosome segregation.

CELL cycle progression must be carefully regulated to preserve genome integrity. In addition to the many proteins that carry out the structural and mechanical aspects of duplicating and segregating chromosomes, an extensive network of regulatory proteins oversees these events. The fidelity of chromosome segregation is ensured, in part, by the spindle assembly checkpoint that regulates the metaphase-to-anaphase transition (reviewed in GARDNER and BURKE 2000 ). Kinetochores that lack bipolar attachments to the mitotic spindle send out a signal that is recognized and transmitted by the checkpoint proteins Mad1p, Mad2p, Mad3p, Bub1p, Bub3p, and Mps1p. The signal itself is not well understood, but it probably stems from a kinetochore not occupied by microtubules and/or absence of tension on an unattached or mono-attached kinetochore (ZHOU et al. 2002 ). In budding yeast the end result of the checkpoint signaling pathway is stabilization of the anaphase inhibitor, Pds1p. This is achieved by inhibiting a ubiquitin ligase, the anaphase-promoting complex/cyclosome (APC/C), which in conjunction with Cdc20p is required to promote Pds1p's ubiquitination and subsequent degradation (VISINTIN et al. 1997 ). Pds1p blocks anaphase initiation by binding to and inhibiting the separase Esp1p, so named because it promotes separation of sister chromatids through cleavage of a cohesin subunit, Scc1p (CIOSK et al. 1998 ; UHLMANN et al. 1999 ). Once all chromosomes have established bipolar attachments to the spindle, the checkpoint signal ceases and Pds1p is ubiquitinated and degraded, releasing Esp1p to promote anaphase initiation. Under normal conditions, neither the checkpoint pathway nor Pds1p is essential, but if spindle assembly or kinetochore function is compromised both the checkpoint and Pds1p become essential (HOYT et al. 1991 ; LI and MURRAY 1991 ; YAMAMOTO et al. 1996A ). Pds1p plays a similar role in response to DNA damage, where the activation of the DNA damage checkpoint pathway results in Pds1p's stabilization (YAMAMOTO et al. 1996B ; COHEN-FIX and KOSHLAND 1997 ; SANCHEZ et al. 1999 ; WANG et al. 2001 ). In addition to its role as an anaphase inhibitor, Pds1p also acts as an activator of Esp1p by promoting its nuclear localization (JENSEN et al. 2001 ; AGARWAL and COHEN-FIX 2002 ). Esp1p must have a Pds1p-independent mechanism for entering the nucleus that functions adequately at room temperature, but cells lacking Pds1p are temperature sensitive because they lack sufficient Esp1p in the nucleus to initiate anaphase (YAMAMOTO et al. 1996A ; JENSEN et al. 2001 ). Because Pds1p both promotes Esp1p's nuclear localization and inhibits Esp1p's activity, the cell ensures that any Esp1p that colocalizes with its nuclear substrate remains inactive until Pds1p is degraded.

The cyclin-dependent kinase (CDK), Cdc28p, is a critical regulator of cell cycle progression in budding yeast. Cdc28p pairs with at least nine different cyclins during the cell cycle (six Clbs and three Clns; reviewed in MENDENHALL and HODGE 1998 ). The type of cyclin associated with Cdc28p, and perhaps the absolute level of kinase activity, determine which Cdc28p substrates get phosphorylated and to what extent. Among Cdc28p's targets are other regulatory proteins that are responsible for cell cycle processes such as the metaphase-to-anaphase transition or the exit from mitosis. Because the mechanisms controlling different aspects of cell cycle progression are highly interconnected, disruption of one part of the system may have adverse effects on later events.

Not surprisingly, Cdc28p/cyclin activity is highly regulated. Cdc28p/cyclin activity is inhibited by a number of mechanisms including phosphorylation, cyclin degradation, and binding of inhibitory proteins. Inhibitory phosphorylation of Cdc28p is carried out by Swe1p. Swe1p is the budding yeast homolog of Wee1, a kinase found in fission yeast and higher eukaryotes, which phosphorylates and inhibits mitotic Cdk/cyclin at the G₂/M transition. Unlike in other organisms, the timing of mitotic entry during normal cell cycles in budding yeast is not regulated by inhibitory phosphorylation of the mitotic CDK (AMON et al. 1992 ; SORGER and MURRAY 1992 ); nonetheless, the biochemical functions of Wee1 and Swe1p appear to be conserved. Swe1p is specific for Cdc28p/Clb complexes and probably phosphorylates Cdc28p/Clb during normal growth because mutants that cannot remove the Swe1p-dependent phosphorylation from Cdc28p show a G₂/M delay that is eliminated when SWE1 is deleted (BOOHER et al. 1993 ). The degradation of the mitotic cyclins (e.g., Clb2) is mediated primarily by the APC/C that is associated with Cdh1p, a Cdc20p homolog (SCHWAB et al. 2001 ). Cdh1p, like Cdc20p, functions as a substrate specificity factor for the APC/C (SCHWAB et al. 1997 ; VISINTIN et al. 1997 ). In addition to the mitotic Clbs, APC/C^Cdh1p directs the ubiquitination of a number of substrates late in mitosis and in G₁, including the septin-associated kinase Hsl1p (BARRAL et al. 1999 ; BURTON and SOLOMON 2000 ), the spindle motor Cin8p (HILDEBRANDT and HOYT 2001 ), and the spindle-associated protein Ase1p (JUANG et al. 1997 ; VISINTIN et al. 1997 ). By targeting mitotic cyclins for degradation, APC/C^Cdh1p inactivates Cdc28p/Clbs, a prerequisite for mitotic exit. However, Cdh1p is not essential because mitotic Cdc28p/Clb activity in G₁ is also squelched by the binding of Sic1p, a CDK inhibitor. The requirement for either Cdh1p or Sic1p is underscored by the fact that although neither protein is essential, the cdh1 sic1 double mutant is inviable.

Following mitosis, both APC/C^Cdh1p and Sic1p continue to function as CDK inhibitors throughout G₁ of the next cell cycle (HUANG et al. 2001 ). Sic1p activity during G₁ is important for DNA replication (LENGRONNE and SCHWOB 2002 ). Assembly of prereplication complexes on replication origins, which normally occurs during G₁, is inhibited by Cdc28p/Clb activity (DAHMANN et al. 1995 ). In sic1 mutants, Cdc28p/Clb5,6p activity during G₁ is too high, and some origins are never primed for replication (LENGRONNE and SCHWOB 2002 ). Consequently, DNA replication is not completed in a timely manner, and some cells attempt to segregate incompletely replicated chromosomes, resulting in extensive chromosome loss. In this report we present evidence that, like Sic1p, Cdh1p is also important for more than mitotic exit. However, Cdh1p activity at the end of mitosis is distinct from that of Sic1p and is required to ensure high-fidelity chromosome segregation during mitosis of the next cell cycle.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Yeast strains:
The genotypes of the strains used in this work are listed in Table 1. Strains are derived from the W303 background except where indicated. To create strain KR3011, a PDS1 disruption cassette was cut out of pAY55 (YAMAMOTO et al. 1996A ) with ApaI and KspI and transformed into a bar1 version of ymw2 (gift of M. Solomon, Yale University; originally from M. Walberg and R. Davis, Stanford University). Sources of other mutant alleles used were: bub2::URA3 (from strain KH128, gift of S. Biggins), mad2::URA3 (from strain KH141, gift of S. Biggins), cdh1::HIS3 [from strain 1120, gift of A. Amon, Massachusetts Institute of Technology (MIT)], sic1::HIS3 (from strain 708, gift of A. Amon, MIT), rad9::HIS3 (SE1, gift of S. Elledge, Baylor College of Medicine), and mrc1::S.p. his5+ (SE2, gift of S. Elledge, Baylor College of Medicine). The sources of the chromsome III fragment (SPENCER et al. 1990 ) were JCY149 and JCY150, both of which were gifts of J. Campbell (National Institutes of Health) and were made on the basis of the strategy described in SPENCER et al. 1990 . The source of the tetO::URA3 and GFP-tetR::LEU2 constructs was strain 6752, a gift of K. Nasmyth, Vienna.

The cdh1::kan (used for all cdh1::kan strains except 3124), pds1::kan, and ade3::kan alleles were created by PCR-based one-step gene disruption (LONGTINE et al. 1998 ). PCR was done with plasmid pFA6a-kanMX6 as template, forward primers that consisted of 50 bases 5' of the coding region of the gene to be disrupted followed by F₁, and reverse primers that consisted of 50 bases 3' of the coding region of the gene to be disrupted followed by a modified version of R1 (5'-CGATGAATTCGAGCTCGTTT-3'). The sequences of the primers were as follows: KRO162 (cdh1-forward), 5'-CTCCGATTTTTGTCACCCTTCCTTCTAGTCTTCATCCTAAATTTAGTTGCCGGATCCCCGGGTTAATTAA-3'; KRO163 (cdh1-reverse), 5'-TTTTTTTTACAGAATTTTTGAGATGATATTACTACTATGAAAACCCTTTACGATGAATTCGAGCTCGTTT-3'; KRO171 (pds1-forward), 5'-TTACACTTCTGCGGTACCAAGCTAGATTAAGTGCTAGATAATAAACCTTTCGGATCCCCGGGTTAATTAA-3'; KRO172 (pds1-reverse), 5'-TATCTGTATATACGTGTATATATGTTGTGTGTATGTGAATGAGCAGTGGATCGATGAATTCGAGCTCGTTT-3'; KRO197 (ade3-forward), 5'-TGAGACCAGGTAACGAGACGAACACAACTTTACAAGTCAAATAAGAAATCCGGATCCCCGGGTTAATTAA-3'; KRO198 (ade3-reverse), 5'-AAAAAAACTTTTGCATTTGTCTTTATTAAATTCTATATAATTAAGTTGTCCGATGAATTCGAGCTCGTTT-3'.

swe1::kan with 400 bases of flanking sequence 5' and 3' was amplified by PCR from a swe1::kan strain (S228c background; American Type Culture Collection, Manassas, VA) using primers KRO140 (swe1-forward) 5'-GTGGGAGATAGGGGGCTATTCG-3' and KRO141 (swe1-reverse) 5'-GAACTTTTGGTGGTCCAGCGTGG-3' and transformed into W303.

The mad2::ura3::HIS3 allele was created using the marker swap plasmid method as described in CROSS 1997 . Plasmid pUH7 was digested with SmaI to release HIS3 flanked by URA3 sequences. The fragment was transformed into a mad2::URA3 strain and his⁺ ura^- transformants were isolated. The above strains were crossed with each other and with other appropriately marked W303 strains to create the strains in Table 1 numbered KR3038–KR3099-3B.

hxt13::URA3 was constucted by PCR using pRS306 (SIKORSKI and HIETER 1989 ) as a template and primers consisting of 50 bases 5' and 3' to HXT13 followed by sequences that anneal to the regions flanking the URA3 gene in pRS306 [KRO230 (hxt13-forward) 5'-CACGTAAGGCATAACAATCAAAAAAAGAAAAAAGAAACAAAAGTTAAACCGCATCAGAGCAGATTGTACTG-3' and KRO231 (hxt13-reverse) 5'-AACTATAATATACAATGTTGCCTATCAAGACAAACATATGCACTCTATGACTCCTTACGCATCTGTGCGG-3']. The PCR product was tranformed into BY4742 (MAT his31 leu20 lys20 ura30; S288c background; Yeast Consortium, ResGen, Invitrogen, Huntsville, AL) to create strain 3118. sic1::kan and cdh1::kan (strain 3124) with 400 bases of flanking sequence 5' and 3' were amplified from the sic1::kan and cdh1::kan strains from the Saccharomyces Genome Deletion Project MATa Collection (Yeast Consortium; ResGen, Invitrogen) using the following primers: KRO235 (sic1-forward), 5'-GGCCAACTCTTGTTGTAGTTG-3'; KRO195 (sic1-reverse), 5'-GTCACTTCTAGCAAATTTGG-3'; KRO236 (cdh1-forward 2), 5'-GTCTCCACCATAACCATAGAAG-3'; and KRO164 (cdh1-reverse 2), 5'-GACGCCTGTAATATGTCATG-3'. The PCR products were transformed into strain 3118 to create strains 3123 and 3124.

Media:
Liquid yeast culture media was prepared as described in AUSUBEL et al. 1995 . Amino acid dropout powders for synthetic media were purchased from Bio101 (Carlsbad, CA). YP + galactose solid media were prepared like YP + dextrose (AUSUBEL et al. 1995 ) except that 2% galactose was substituted for dextrose. 5-Fluoroanthranilic acid (FAA) solid media was prepared as described in TOYN et al. 2000 with FAA purchased from Aldrich Chemical (Milwaukee). Other solid yeast media and Luria broth (LB) + ampicillin plates were purchased from KD Medical (Columbia, MD). Escherichia coli minimal medium + leucine and tryptophan contained: 1.5% agar, 1x M9 salts (6 µg/ml Na₂HPO₄, 3 µg/ml KH₂PO₄, 0.5 µg/ml NaCl, and 1 µg/ml NH₄Cl), 1 mM magnesium sulfate, 1 µM FeCl₃, 1 µg/ml thiamine, 0.5% dextrose, and 26 µg/ml each tryptophan and leucine.

Plasmids:
PDS1 was cut from pOC20 [CEN/URA3/PDS1, with the PDS1 gene inserted between the EcoRI and BamHI sites of plasmid pRS316 (SIKORSKI and HIETER 1989 )] with BamHI and EcoRI and ligated into the corresponding sites in pRS314 (CEN/TRP1; SIKORSKI and HIETER 1989 ) to create pKR201. ADE3 (coding region plus 500 bases on each side) was amplified by PCR from pDK255 (gift of D. Koshland) with primers that introduced a 5' SacI site (KRO105) and a 3' NotI site (KRO106). Primer sequences were: KRO105, 5'-GGGTATGAGCTCTACGTGAGCTAAAGCACAGATTG-3'; and KRO106, 5'-GGATAAGCGGCCGCGTAGTCCAATACCGTTTTTG-3'. The PCR product was digested with SacI and NotI and ligated into the corresponding sites of pKR201 to create pKR204. Plamids 189-26 (CDH1) and 189-20 (SWE1) were isolated from a CEN/URA3 genomic library (ROSE et al. 1987 ). The genomic insert in plasmid 189-26 corresponds to chromosome VII bases 485,079–496,384. The genomic insert in plasmid 189-20 corresponds to chromosome X bases 74,516–80,656. To create pKR252 (CEN/TRP1/CDH1), plasmid 189-26 was cut with EcoRI, and the resulting 6.7-kb fragment including CDH1 was ligated into pRS314 (CEN/TRP1; SIKORSKI and HIETER 1989 ). pJM1091 (gift of J. Harrison, Duke University) consists of the SWE1 open reading frame with 900 bases of promoter sequence and 10–12 C-terminal MYC tags in pRS316 (CEN/URA3; SIKORSKI and HIETER 1989 ). AD10 [SIC1 in pRS425 (2µ/LEU2; SIKORSKI and HIETER 1989 )] was a gift of F. Cross, Rockefeller University. Plasmids pDK243, pDK368-1, and pDK368-7 (gifts of E. Hogan and D. Koshland, Carnegie Institute) were described in HOGAN and KOSHLAND 1992 .

Synthetic lethal screening:
The synthetic lethal screen was based on the sectoring strategy of BENDER and PRINGLE 1991 . Strain KR3011 (pds1 ade2 ade3 trp1) with the nonessential plasmid pKR204 (CEN/TRP1/ADE3/PDS1) was grown overnight in 25 ml synthetic complete media lacking tryptophan. Cells were washed with water and resuspended in 25 ml 0.1 M sodium phosphate buffer, pH 7.0. Three-milliliter aliquots of cells were incubated in either 100 µl (27 mM) or 150 µl (40 mM) of methanesulfonic acid ethyl ester (EMS; Sigma, St. Louis) for 90 min. Viability after this treatment ranged from 10 to 30%. The EMS was inactivated by washing twice with 3 ml 5% sodium thiosulfate. Cells were resuspended in media, plated on YP + dextrose after the appropriate dilution to get 300 colonies per plate, and incubated at 23°. A total of 18,000 colonies were screened. Colonies that failed to sector (i.e., completely red colonies) were restreaked and then tested for sensitivity to FAA that selects against the TRP1 plasmid, pKR204. Nonsectoring, FAA-sensitive strains were kept for further analysis.

Cloning of SL189:
Synthetic lethal candidate strain SL189 with pKR204 was transformed with a CEN/URA3 yeast genomic library (ROSE et al. 1987 ). Transformants were selected on synthetic complete media lacking uracil and then replica plated onto FAA to select against pKR204. Thirty-eight FAA-resistant colonies were isolated out of 14,000 transformants. Library plasmids were recovered from yeast by the yeast-boiling DNA miniprep procedure (ROBZYK and KASSIR 1992 ) and transformed into E. coli. Plasmids were sequenced using primers RO91 5'-GCTTTGGCCGCCGCCCAGTCCTGCTGCC and RO92 5'-CATCGGTGATGTCGGCGATATAGGCGCC that flank the genomic DNA insert.

Gap repair:
The CDH1-containing library plasmid, 189-37 (genomic insert equals chromosome VII bases 483,481–496,380), was digested with PvuII to remove 7.9 kb of the genomic insert including CDH1. The backbone was gel purified and religated to create pKR217. The SWE1-containing library plasmid, 189-10 (genomic insert equals chromosome X bases 68,067–80,656), was digested with AflII and SnaBI to remove 10.1 kb of the genomic insert including SWE1. (SnaBI cut the insert into two pieces so it would not comigrate with the backbone on a gel.) The backbone was gel purified and religated to create pKR216. Strain SL189 (carrying pKR204) was transformed with pKR217 that had been linearized with PvuII or pKR216 that had been linearized with AflII. Plasmids (a mixture of pKR204 and pKR216 or pKR217 derivatives) were isolated from ura⁺ transformants by the yeast boiling DNA miniprep procedure (ROBZYK and KASSIR 1992 ) and transformed into E. coli strain MC1066 [galU galK strAr hsdR^- (lacIPOZYA)X74 trpC9830 leuB6 pyrF74::Tn5(km^r)] in which it is possible to select for the yeast URA3 gene. Transformants were plated on LB + ampicillin and then replica plated onto minimal media + tryptophan and leucine to select for URA3-containing plasmids. Restriction digests were performed to determine whether the gap repair was successful.

Measurement of chromosome fragment, plasmid, and chromosome loss:
ade2 mutant cells were transformed with a nonessential fragment of chromosome III that carries the ADE2 and URA3 genes (SPENCER et al. 1990 ) (Table 2). Loss of the fragment gave rise to red sectors in an otherwise white colony. Strains without the SIC1 plasmid were grown overnight in synthetic complete medium lacking uracil to select for the chromosome fragment. Strains carrying the SIC1 plasmid were grown overnight in synthetic complete medium lacking uracil and leucine to select for the plasmid as well. Cultures were diluted to a density corresponding to 300 colonies per plate and plated on YP + dextrose. Cells that lost the chromosome fragment during the first division on the plate gave rise to half-sectored colonies whereas those that lost the fragment before plating formed completely red colonies. The percentage loss rate per cell division was calculated using the formula: 100 x (half-sectored colonies)/(total colonies - red colonies). Between 3000 and 6000 colonies without the SIC1 plasmid and between 1000 and 2000 colonies with the SIC1 plasmid were scored for each genotype.

pDK243 and pDK368-7 plasmid loss rates (Table 6) were determined similarly except that the strains used were ade2ade3 double mutants and the plasmid carried the ADE3 gene so that cells that lost the plasmid gave rise to white sectors in an otherwise red colony (KOSHLAND et al. 1985 ). The wild-type (WT; A364a) and cdc6-1 strain cells were grown to log phase at 23°, incubated at 36° for 3 hr, and then plated immediately. All other strains were grown at 23° and then plated. For all strains, plates were incubated at 23°. Between 400 and 1000 colonies were scored for each genotype/plasmid combination.

Missegregation of chromosome V (Table 3 and Table 4) was monitored using strains that expressed a tet repressor-green fluorescent protein (GFP) fusion and carried an array of tet operators integrated at the URA3 locus on chromosome V (MICHAELIS et al. 1997 ). Cells were grown to midlog phase in YP + dextrose, fixed, stained with 4',6-diamidino-2-phenylindole (DAPI) to visualize nuclei, and observed under the microscope. Segregation was classified as normal if a single GFP dot was observed in unbudded, small-budded, or large-budded cells with a single nucleus, or if two dots were observed in postanaphase cells, one dot in each nucleus. Other patterns were scored as missegregation events. Approximately 200–400 cells were scored for each genotype.

Gross chromosomal rearrangement assay:
Gross chromosomal rearrangement (GCR) assays were done as described in MYUNG et al. 2001 . URA3 was integrated into the HXT13 locus, which is located 7.5 kb telomeric to the CAN1 locus on the left arm of chromosome V. Between 10⁹ and 5 x 10⁹ hxt13::URA3 cells were grown to midlog phase in YPD and plated on canavanine/fluoroorotic acid (FOA) plates (10⁹ cells per 150-mm plate) to select for cells that had lost both URA3 and CAN1. Dilutions of the same cultures were plated on synthetic complete medium to determine the total number of viable cells.

Microscopy:
Cells were fixed for microscopy in media with 4% paraformaldehyde (Electron Microscopy Services, Fort Washington, PA) for 1 hr at 23°, washed with 1x phosphate-buffered saline (PBS), and stored at 4°. Immediately before observation, the fixed cells were sonicated gently to break up clumps, incubated in 1% Triton-X-100 for 5 min, mixed at a 1:1 ratio with Vectashield with DAPI mounting medium (Vector Laboratories, Burlingame, CA), and placed on a slide. Observations were done with a Nikon Eclipse E800 microscope with a Nikon 100x Plan Apo phase objective and filter sets for DAPI and GFP.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

A screen for mutations that are lethal in combination with pds1:
To search for novel proteins important for cell cycle progression, we conducted a synthetic lethal screen to identify mutants that are dependent on Pds1p for viability at 23°, a temperature at which, under normal growth conditions, Pds1p is not required. We anticipated that this approach will reveal several different classes of proteins: those acting in parallel to Pds1p in promoting nuclear localization of Esp1p or factors involved in processes such as spindle assembly or DNA metabolism whose absence would render cells dependent on Pds1p's checkpoint function. Following the strategy of BENDER and PRINGLE 1991 , we deleted the chromosomal copy of PDS1 in an ade2 ade3 trp1 mutant strain and provided PDS1 on a CEN plasmid that also carried TRP1 and ADE3 (pKR204). Cells that carried the plasmid were red due to the ade2 mutation whereas cells without the plasmid were white due to their ade2 ade3 genotype. Under nonselective conditions, cells occasionally lost the plasmid, giving rise to white sectors in an otherwise red colony. If, however, cells sustained a mutation that rendered PDS1 essential, viability then depended on the pKR204 plasmid, resulting in solid red colonies. The pds1 ade2 ade3 trp1/pKR204 strain was mutagenized with EMS and strains that formed red, nonsectoring colonies on YP + dextrose at 23° were isolated. The nonsectoring strains were then tested for FAA sensitivity. FAA is toxic to cells that have a wild-type copy of TRP1 (TOYN et al. 2000 ), thus allowing growth only of cells that had lost the Trp1p-encoding plasmid, pKR204. Strains that required PDS1 for viability were unable to grow without pKR204 and, consequently, were FAA sensitive. Finally, we eliminated those strains that remained FAA sensitive after transformation with a CEN/URA3/PDS1 plasmid (for example, those that integrated pKR204 into the yeast genome). Out of 18,000 mutagenized cells, we obtained 19 synthetic lethal mutants that fell into 15 complementation groups. In this work, we describe the characterization of one of these strains, SL189, which has a mutation in the gene encoding for the APC/C activator, CDH1 (SCHWAB et al. 1997 ; VISINTIN et al. 1997 ).

Identification of CDH1 as the mutated gene in SL189:
Strain SL189 + pKR204 (CEN/TRP1/PDS1) was transformed with a CEN/URA3 yeast genomic library. Cells that obtained a library plasmid that allowed them to grow without pKR204 were selected on the basis of FAA resistance. Library plasmids that conferred FAA resistance had genomic inserts that included PDS1, CDH1 (plasmid 189-26), and the CDK inhibitor SWE1 (plasmid 189-20; Fig 1A). To determine whether the SWE1 or CDH1 genes were mutated in SL189, we isolated the genomic alleles of these genes from strain SL189 by gap repair of the corresponding library plasmids from which the coding regions of these genes were removed. The genomic copy of SWE1 in SL189 appeared to be functional because the SWE1 gap-repaired plasmid allowed SL189 to grow without the PDS1 plasmid (data not shown). We also confirmed that a centromeric plasmid carrying the wild-type SWE1 gene alone (pJM1091; Fig 1A) was able to suppress the synthetic lethality of SL189. Thus, SWE1 is probably a suppressor of the pds1 synthetic lethality in SL189. The CDH1 gap-repaired plasmid, on the other hand, could not suppress the synthetic lethality of SL189, suggesting that the mutation responsible for the synthetic lethality had been transferred to the CDH1 plasmid. We sequenced the gap-repaired CDH1 allele and found that it had a C-to-T mutation at base 460 (out of 1701 bases) that changed the codon for Arg 154 to a STOP codon (this allele was named cdh1-189). To test directly whether cdh1 is synthetically lethal with pds1, the meiotic products of a pds1/PDS1 cdh1/CDH1 heterozygous diploid were analyzed, and no viable double-mutant spores were identified. When the heterozygous diploid was transformed with a PDS1 plasmid before sporulation, viable double-mutant spores were obtained, but they all required the PDS1 plasmid for viability. pds1 cdh1 cells transformed with plasmids expressing either CDH1 (plasmid 189-26) or SWE1 (pJM1091) were able to grow in the absence of the PDS1 plasmid (Fig 1B). On the basis of this evidence, we conclude that pds1 and cdh1 are synthetically lethal and that the nonsense mutation in cdh1-189 is likely to be responsible for the synthetic lethality with pds1 in SL189.

View larger version (49K): In this window In a new window Download PPT slide   Figure 1. The effect of CDH1 and SWE1 plasmids on the growth of pds1 cdh1 double mutants. (A) Strain SL189 (pds1 cdh1-189) carrying plasmid pKR204 (CEN/TRP1/ADE3/PDS1) was transformed with pRS316 (CEN/URA3, vector), CEN/URA3 genomic library plasmids containing CDH1 (189-26), SWE1 (189-20), or a CEN/URA3 plasmid containing the SWE1 gene alone (pJM1091). Transformants were grown at 23° on YP + dextrose (YPD; left) or FAA (right) to select for cells that have lost pKR204. (B) pds1 cdh1 cells (strain KR3044-3A) carrying plasmid pKR204 were transformed with pRS316, 189-26, or pJM1091 and grown at 23° on FAA.

Mutations in the spindle assembly checkpoint are deleterious to cdh1 mutants:
We next investigated which function(s) of Pds1p was required for viability in cdh1 mutants. Overexpression of ESP1 from a galactose-inducible promoter did not rescue the pds1 cdh1 mutant, suggesting that Pds1p's role in Esp1p activation was unlikely to be relevant to the pds1 cdh1 synthetic lethality (data not shown). Pds1p is an essential part of the spindle assembly checkpoint pathway (GARDNER and BURKE 2000 ) and also assists in arresting the cell cycle after DNA damage (YAMAMOTO et al. 1996B ; COHEN-FIX and KOSHLAND 1997 ; GARDNER et al. 1999 ). If Pds1p's checkpoint function(s) was important for the viability of cdh1 mutants, we would expect to see genetic interactions between CDH1 and other checkpoint proteins. Thus, we created mutants that lacked Cdh1p and a component of each of four cellular checkpoint pathways: (1) the spindle assembly checkpoint (Mad2p; Fig 2A and Fig C), (2) the DNA damage checkpoint (Rad9p; Fig 2B and Fig C), (3) the DNA replication checkpoint (Mrc1p; ALCASABAS et al. 2001 ; TANAKA and RUSSELL 2001 ; Fig 2B and Fig C), and (4) the spindle-positioning checkpoint (Bub2p; HOYT 2000 ; Fig 2A and Fig C), although Pds1p has so far not been shown to be involved in this checkpoint. We found that cdh1 single mutants grew significantly more slowly than wild type or any of the other single mutants (Fig 2A and Fig B). mad2 cdh1 double mutants, while viable, were much slower growing than the cdh1 single mutant (Fig 2A). This interaction was specific to the spindle assembly checkpoint as rad9 cdh1 and mrc1 cdh1 double mutants grew as well as cdh1 single mutants (Fig 2B), and bub2 cdh1 double mutants were, at most, slightly slower growing (Fig 2A).

View larger version (79K): In this window In a new window Download PPT slide   Figure 2. Genetic interactions between cdh1 and checkpoint mutants. (A) Wild-type (strain KR3074-1B, WT), mad2 (strain KH141), bub2 (strain KH128), cdh1 (strain KR3038-6D), mad2 cdh1 (strain KR3072-3D), and bub2 cdh1 (strain KR3071-2B) cells were grown on YP + dextrose at 23°. (B) Wild-type (strain KR3074-1B, WT), rad9 (strain SE1), mrc1 (strain SE2), cdh1 (strain KR3044-1B), rad9 cdh1 (strain KR3091-3D), and mrc1 cdh1 (strain KR3093-10A) cells were grown on YP + dextrose at 23°. (C) The double-mutant combinations rad9 cdh1 (strain KR3091-3D), mrc1 cdh1 (strain KR3093-10A), mad2 cdh1 (strain KR3072-3D), and bub2 cdh1 (strain KR3071-2B) were grown on YP + dextrose at 23°. (D) mad2 cdh1 cells (strain KR3074-3B) were transformed with pRS316 (CEN/URA3, vector), a CEN/URA3 genomic library plasmid containing CDH1 (189-26), or a CEN/URA3 plasmid expressing SWE1 alone (pJM1091). Transformants were grown on synthetic medium lacking uracil at 23°. (E) mad2 swe1 (strain KR3098-6B), cdh1 swe1 (strain KR3098-12B), mad2 cdh1 (strain KR3098-2A), and mad2 cdh1 swe1 (strain KR3099-3B) cells all carrying a CEN/TRP1/CDH1 plasmid (pKR252) were spotted onto YP + dextrose plates (left) or FAA plates (right) at 23° to determine whether they could grow in the absence of pKR252.

The spindle assembly checkpoint pathway delays the metaphase-to-anaphase transition until all chromosomes are properly attached to the spindle. Therefore, mutants with compromised spindle or kinetochore function would be expected to exhibit a checkpoint-dependent G₂/M delay, which would be manifested by an abnormally high proportion of G₂/M cells in an asynchronously growing culture. The fraction of G₂/M cells in an asynchronously growing culture of the cdh1 single mutant was similar to wild type (Fig 3). We also saw no significant differences among wild-type, mad2, cdh1, and mad2 cdh1 strains when we timed the interval between bud emergence and nuclear division in single cells, although we would not have been able to detect delays of <15 min (data not shown). It is possible that cdh1 cells undergo a G₂/M delay that is too brief to dramatically affect the overall cell cycle distribution of cells in an asynchronous culture but is nonetheless important for cell survival (see below).

View larger version (47K): In this window In a new window Download PPT slide   Figure 3. The cell cycle distribution of asynchronous populations of wild-type, mad2, cdh1, and mad2 cdh1 cells. Wild-type (WT; strains KR3066-8C and KR3066-3D), mad2 (strains KR3067-4B and KR3068-3C), cdh1 (strains KR3066-6A and KR3066-7B), and mad2 cdh1 (strains KR3066-3B, KR3066-6B, and KR3066-7D) cells were grown to midlog phase in YP + dextrose, fixed, stained with DAPI to visualize nuclei, and examined under the microscope. Cells were classified as unbudded (G1), small budded (S), large budded with a single nucleus (G2/M), or large budded with nuclei in mother and bud (late M). Results are averages of two independent experiments for each strain.

Like pds1 cdh1 mutants, mad2 cdh1 cells were sensitive to SWE1 levels. mad2 cdh1 strains grew better when extra SWE1 was provided on a centromeric plasmid (pJM1091, Fig 2D), and conversely, mad2 cdh1 swe1 triple mutants were inviable (Fig 2E). These results suggest that the growth defects in the pds1 cdh1 and mad2 cdh1 strains may have the same underlying cause. Because pds1 cdh1 and mad2 cdh1 strains are affected by the level of Swe1p, an inhibitor of the cell cycle kinase, Cdc28p/Clb (BOOHER et al. 1993 ), and because Clb cyclins are known targets of Cdh1p (SCHWAB et al. 1997 , SCHWAB et al. 2001 ), the mutant phenotypes of these strains may be related to abnormal regulation of Cdc28p/Clb activity.

mad2 cdh1 strains exhibit extensive chromosome loss:
The spindle assembly checkpoint pathway is necessary for high-fidelity chromosome transmission in cells in which the spindle or kinetochores are compromised in some way. The poor growth of mad2 cdh1 raised the possibility that cdh1 mutant cells had spindle/kinetochore defects that, in the absence of the spindle checkpoint, led to chromosome missegregation followed by cell lethality. To test this idea, we compared the loss rate of a nonessential fragment of chromosome III (SPENCER et al. 1990 ) in wild-type, mad2, cdh1, and mad2 cdh1 strains (Table 2, left column). The fragment is transmitted during mitosis of wild-type cells nearly as well as native chromosomes and it carries the ade2 suppressor, SUP11, allowing its presence to be monitored in a colony color assay (SPENCER et al. 1990 ). We calculated the percentage of half-sectored colonies, which reflects fragment loss during the first mitotic division on the plate. Of 6000 wild-type colonies examined, none had half-sectors, implying that the loss rate was no greater than 0.02% per division. This result is in agreement with the published value of 0.017% per division (SPENCER et al. 1990 ). The mad2 strain had a loss rate of 0.4% (at least 20-fold greater than that of wild type), which is also consistent with published data (WARREN et al. 2002 ). In the cdh1 strain chromosome fragment loss was elevated at least 180-fold over wild type; in the mad2 cdh1 double mutant, fragment loss was at least 400-fold greater than that of wild type. Deletion of CDH1 alone substantially reduces the fidelity of chromosome transmission, suggesting that some of the damage caused by this mutation is not recognized by the spindle checkpoint; however, the checkpoint is exerting a protective effect because the loss rate is significantly higher in the cdh1 mad2 double mutant.

Chromosome transmission was also monitored by an assay in which chromosome V is visualized by expression of a GFP-tagged tet repressor that binds to an array of tet operators integrated into the URA3 locus of chromosome V (MICHAELIS et al. 1997 ; Table 3). In a wild-type population, preanaphase cells exhibit a single GFP spot, while postanaphase cells exhibit two GFP spots, one in the mother nucleus and one in the bud nucleus. Other patterns can arise from chromosome transmission errors. When we examined the distribution of GFP spots in wild-type, mad2, cdh1, and mad2 cdh1 strains, our results agreed well with the results of the chromosome fragment loss assay: wild-type cells showed no evidence of errors, mad2 and cdh1 cells had a moderate level (1%), and mad2 cdh1 cells had a greatly elevated level (6.1%). Importantly, all of the abnormal patterns counted were consistent with missegregation events in which both sister chromatids segregated to the same cell (e.g., unbudded cells with more than one GFP spot or postanaphase cells with two spots in the same nucleus). If the other 15 chromosomes in mad2 cdh1 are missegregated as often as chromosome V (6%), then the fraction of cells in the mad2 cdh1 culture that have the correct complement of chromosomes is only (1 - 0.06)¹⁶ = 0.37. Assuming that most missegregation events resulting in chromosome loss are lethal, this frequency of chromosome missegregation could easily account for the slow growth of mad2 cdh1 strains. Finally, as mentioned earlier, mad2 cdh1 cells do not accumulate in any particular phase of the cell cycle (Fig 3); instead, it is likely that they stop growing whenever a critical protein encoded by the chromosome(s) they have lost becomes limiting.

Chromosome segregation in pds1 cdh1:
To study the terminal phenotype of the pds1 cdh1 strain, we exploited the fact that this strain, while dead in YP + dextrose, does grow, albeit poorly, in YP + galactose (F. CROSS, personal communication; Fig 4A). We do not have an explanation for this phenomenon. We considered the possibility that the pds1 cdh1 strain benefited from progressing through the cell cycle more slowly, a consequence of using a suboptimal carbon source like galactose. However, pds1 cdh1 did not grow in dextrose synthetic complete medium, in which the doubling time of wild-type cells is comparable to that in YP + galactose (data not shown). There is accumulating evidence that a carbon source affects the expression of many genes, including some cell cycle regulatory genes (CROSS et al. 2002 ). It is plausible that the expression of factors involved in spindle assembly or cell cycle control may be altered in the presence of galactose in a way that is helpful to pds1 cdh1 cells. When pds1 cdh1 cells growing in YP + galactose were switched to YP + dextrose, they lost viability over a period of days (Fig 4B). We examined these cells for chromosome segregation defects using the tetO/GFP-tetR system described above (Table 4). All wild-type cells examined showed normal segregation of chromosome V in both dextrose- and galactose-containing media. In galactose, 70% of pds1 cdh1 cells were viable, and 5.6% had missegregated chromosome V. After 48 hr in dextrose, viability had dropped to 6% and >20% of cells had abnormal GFP patterns. In pds1 and cdh1 single-mutant cells, missegregation levels were higher in dextrose than in galactose (2.1 vs. 0% for pds1 cells and 2.1 vs. 1.0% for cdh1 cells), but in all cases, missegregation events were far less frequent in the single mutants than in the double mutant.

View larger version (65K): In this window In a new window Download PPT slide   Figure 4. Viability and cell cycle distribution of pds1 cdh1 mutants grown in galactose and dextrose. (A) Wild-type (WT; strain KR3056-8A), pds1 (strain KR3057-18A), cdh1 (strain KR3054-8D), and pds1 cdh1 (strain KR3056-12C) cells were grown at 23° on YP + galactose (YPG, left) or YP + dextrose (YPD; right). (B) Wild-type (WT; strains KR3055-10A and KR3056-8A), pds1 (strains KR3057-7C and KR3057-18A), cdh1 (strains KR3054-8D and KR3056-1C), and pds1 cdh1 (strains KR3056-1A and KR3056-12C) cells were grown at 23° to midlog phase in YP + galactose liquid medium, transferred to YP + dextrose liquid (0 hr), and grown for an additional 48 hr with dilutions as necessary to maintain the cultures in log phase. At various times, cells were counted and plated on YPG to determine the percentage of viable cells in the culture. (C and D) Samples from the cultures described in B were collected at 0 hr (immediately before the switch from YP + galactose to YP + dextrose (C) and after 48 hr in YP + dextrose (D). Cells were fixed, stained with DAPI, and observed under the microscope. Cell cycle phase was scored as in Fig 2.

Finally, the cell cycle distributions of pds1 cdh1 strains in both galactose and dextrose were similar overall to those of wild type and the two single mutants (Fig 4C and Fig D). Like the mad2 cdh1 strain, pds1 cdh1 cells probably died for a variety of different reasons related to which particular chromosomes were lost.

cdh1 mutants and DNA replication:
Recently, it was shown that the Cdc28p/Clb inhibitor Sic1p is important not only for regulating mitotic exit but also for keeping Cdc28p/Clb activity low during G₁, which allows prereplication complexes (pre-RCs) to assemble on DNA replication origins (LENGRONNE and SCHWOB 2002 ). sic1 mutant cells have a high rate of plasmid and chromosome loss not because of segregation defects but because excessive Cdc28p/Clb activity interferes with origin firing, slowing DNA replication. The delay escapes checkpoint surveillance and cells attempt to separate their sister chromatids while replication intermediates are still present on the DNA. Because Cdh1p cooperates with Sic1p to inhibit Cdc28p/Clb activity during G₁, we considered the possibility that chromosome loss in cdh1 mutants might also be due to replication defects.

If chromosome loss in cdh1 and sic1 mutants has a common cause, then overexpression of SIC1 might compensate for the lack of CDH1 and improve chromosome transmission in cdh1 mutants. Expression of SIC1 from a high-copy plasmid does in fact reduce the chromosome loss rate of cdh1 single mutants more than fivefold from 3.6 to 0.7% (Table 2, right column). SIC1 overexpression can only partially rescue the chromosome loss in mad2 cdh1 double mutants, reducing the rate from 8.3 to 4.6%. Intriguingly, these results suggest that there may be two separate defects contributing to chromosome loss in cdh1 cells. One defect, responsible for the 3.6% loss rate in cdh1 single mutants, is not recognized by the spindle checkpoint, which is functional in these cells, but is ameliorated by overexpressing SIC1. These phenotypes suggest a defect in replication initiation like that seen in sic1 mutants. Consistent with this possibility, overexpression of SIC1 eliminates nearly half of the chromosome loss events in mad2 cdh1 double mutants. The second defect, which is likely to stem from a spindle or kinetochore malfunction, accounts for the remaining half of the chromosome loss in mad2 cdh1 cells. Chromosome loss due to this defect is not affected by SIC1 overexpression but is suppressed by an intact spindle checkpoint.

We next performed two assays to look for evidence of replication defects in cdh1 cells. First, we measured the rate of GCR in cdh1 cells. GCR, which is characterized by large deletions and nonreciprocal translocations, occurs when cells are unable to repair double-strand breaks by homologous recombination. sic1 mutants have extremely high rates of GCR (575-fold elevated relative to wild type) because they frequently incur double-strand breaks while attempting to segregate chromosomes that are still undergoing replication (LENGRONNE and SCHWOB 2002 ). To measure GCR, we integrated URA3 into the left arm of chromosome V near the CAN1 locus. Because no essential genes are distal to CAN1 and URA3, cells can survive GCR events that result in the deletion of both marker genes. Such cells can be identified because they will be canavanine and 5-FOA resistant. Consistent with previously reported results (LENGRONNE and SCHWOB 2002 ), we found that GCR was very high in sic1 mutants (833-fold elevated relative to wild type; Table 5). In contrast, GCR in cdh1 cells was nearly the same as in wild-type cells (1.6-fold elevated relative to wild type). Although both cdh1 and sic1 cells undergo chromosome loss, the results of this experiment suggest that the loss occurs for different reasons in the two mutant strains.

Second, we tested whether plasmid loss is suppressed in cdh1 mutants by increasing the number of origins of replication (ARSs) on the plasmid. Extra ARSs improve plasmid transmission in mutants with defects in replication initiation by increasing the chances that the plasmid will get a competent pre-RC under conditions where pre-RC assembly is difficult (HOGAN and KOSHLAND 1992 ). We found that a plasmid with a single origin of replication (pDK243) was lost approximately twofold more frequently than a plasmid with eight ARSs (pDK368-7) from WT (303), mad2, cdh1, and mad2 cdh1 cells (Table 6; HOGAN and KOSHLAND 1992 ). Because we observed a twofold difference between the loss rates of pDK243 and pDK368-7 in wild-type as well as mutant cells, it is unlikely that this difference reflects a replication defect in the mutants. As a control, we transformed the same pDK243 and pDK368-7 plasmids into a cdc6-1 strain in which plasmid loss has previously been shown to be suppressed by extra ARSs (HOGAN and KOSHLAND 1992 ) and a congenic WT strain (A364a) and measured plasmid loss after incubating the strains at 36° for 3 hr, the restrictive temperature for the cdc6-1 mutation. In cdc6-1 cells, there was a 3.3-fold decrease in the loss rate of pDK368-7 as compared to that of pDK243 while there was virtually no difference between the loss rates of the two plasmids in the wild-type cells (Table 6). We also examined sic1 mutants because LENGRONNE and SCHWOB 2002 reported that the loss rate of pDK368-7 is sixfold lower than that of pDK243 in this strain (HOGAN and KOSHLAND 1992 ; LENGRONNE and SCHWOB 2002 ); however, in our hands, there was no significant difference between the loss rates of pDK243 and pDK368-7 in sic1 cells (loss of pDK368-7 was only 1.7-fold lower than loss of pDK243). We do not know the reason for this discrepancy with the published results. Nonetheless, since we were able to observe an effect of additional ARSs on plasmid loss in cdc6-1 cells, and because we did not see any effect in cdh1 or cdh1 mad2 cells, we conclude that these experiments do not support a role for Cdh1p in replication initiation.

Finally, sic1 and cdh1 mutants differ in their sensitivities to deletion of PDS1. While the pds1 sic1 strain grew noticeably more slowly than wild type, it was still much healthier than a pds1 cdh1 strain (Fig 5). This is in agreement with LENGRONNE and SCHWOB 2002 , who found that mutation of PDS1 and other spindle and DNA replication checkpoint components had little, if any, effect on sic1 strains. Taken together, our results suggest that although there is some overlap in the mutant phenotypes of cdh1 and sic1 and overexpression of Sic1p can compensate for some of the cdh1 defects, there are also important differences. Thus, Sic1p and Cdh1p may play distinct roles in ensuring genomic integrity.

View larger version (106K): In this window In a new window Download PPT slide   Figure 5. Genetic interaction of pds1 and sic1. Wild-type (strain KR3089-1D, WT), pds1 (strain KR3089-6C), sic1 (strain KR3089-1C), cdh1 (strain KR3055-17B), pds1 sic1 (strain KR3089-4D), and pds1 cdh1 (strain KR3056-12D; maintained on YP + galactose before streaking onto YP + dextrose) cells were grown on YP + dextrose at 23°.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Our fundamental finding in this study is that cdh1 mutant cells rely on the spindle assembly checkpoint pathway to prevent rampant chromosome loss. The simplest interpretation of our results is that chromosome segregation is disrupted by accumulation of a protein that normally is degraded by APC/C^Cdh1p. APC/C^Cdh1p has many substrates, including several that are involved in mitotic progression and spindle function, such as Clb2p (SCHWAB et al. 1997 ), the polo-like kinase Cdc5p (CHARLES et al. 1998 ; SHIRAYAMA et al. 1998 ), the septin-associated kinase Hsl1p (BURTON and SOLOMON 2000 ), the microtubule motor Cin8p (HILDEBRANDT and HOYT 2001 ), and the microtubule-binding protein Ase1p (JUANG et al. 1997 ; VISINTIN et al. 1997 ), raising the question of which substrate(s) is critical. It is quite possible that to ensure genomic integrity more than one APC/C^Cdh1p substrate must be degraded. JUANG et al. 1997 showed that overexpression of nondegradable Ase1p when combined with a mad1 mutation resulted in a spindle assembly checkpoint-dependent mitotic delay and a high rate of mortality. Likewise, overexpression of nondegradable Cin8p caused a metaphase-like arrest with an abnormal spindle morphology, but dependence on the spindle assembly checkpoint was not determined (HILDEBRANDT and HOYT 2001 ). Finally, overexpression of Cdc5p resulted in growth arrest that was not restricted to a particular cell cycle stage (CHARLES et al. 1998 ). It is not known whether nondegradable Ase1p, Cin8p, or Cdc5p expressed from their endogenous promoters, equivalent to the situation in cdh1 mutants, would have the same effect, but it is conceivable that failure to degrade one or more of these proteins contributes to chromosome segregation defects.

It is interesting to note that Cdc20p, the Cdh1p homolog that targets substrates to the APC/C at the metaphase-to-anaphase transition, is itself degraded in an APC/C-dependent manner during G₁ and is therefore likely to be a Cdh1p substrate (PRINZ et al. 1998 ; SHIRAYAMA et al. 1998 ). Thus, the absence of Cdh1p in cdh1 mutants may be partially compensated for if Cdc20p is stabilized and APC/C^Cdc20p activity persists into late mitosis and G₁.

The fact that the severity of the mutant phenotype in pds1 cdh1 and mad2 cdh1 strains is influenced by the level of the Cdc28p/Clb inhibitor Swe1p suggests that the critical proteins requiring degradation in an APC/C^Cdh1p-dependent manner are the Clb cyclins. Although it is formally possible that altered Cdc28p/Clb levels do not have a deleterious effect in pds1 cdh1 and mad2 cdh1 cells and that Swe1p simply delays the G₂/M transition long enough for the spindle to recover from damage caused by the accumulation of other Cdh1p substrates, we favor the idea that these mutants suffer from abnormal Cdc28p/Clb activity caused by the failure to degrade mitotic cyclins (see below). At the end of mitosis, Clb cyclins are degraded and Sic1p binds to the Cdc28p/Clb complex. These mechanisms appear to have the same end result, inactivation of Cdc28p/Clb, but they are not equivalent. Wild-type cells degrade Clb2p from anaphase onset throughout the next G₁ until APC/C^Cdh1p is inhibited by increasing Cdc28p activity at the G₁/S boundary (AMON et al. 1994 ; YEONG et al. 2001 ). From G₁/S through metaphase, CLB genes are transcribed and Cdc28p/Clb gradually accumulates (AMON et al. 1993 ). Cells that lack Cdh1p, on the other hand, are likely to maintain high levels of Cdc28p/Clb that is kept inactive by Sic1p during G₁. We hypothesize that when Sic1p is degraded at the onset of S phase, these cells immediately are subjected to high Cdc28p/Clb kinase activity (Fig 6). During much of interphase, cdh1 mutants may have higher Cdc28p/Clb activity than wild-type cells do, which could lead to inappropriate phosphorylation of certain substrates, acceleration of progression through interphase, and reduced fidelity of some cell cycle events such as kinetochore or spindle assembly. In support of this idea, cdh1 cells are significantly smaller than wild-type cells, indicating that the balance between cell division and cell growth has been altered (JORGENSEN et al. 2002 ; WASCH and CROSS 2002 ). One prediction of this model is that artificially reducing Cdc28p/Clb activity, for example, by deleting CLB2, should improve the growth of pds1 cdh1 and mad2 cdh1 cells. However, because several of the double-mutant combinations adversely affect viability (e.g., pds1 clb2 are synthetically lethal) we have not been able to create pds1 cdh1 clb2 and mad2 cdh1 clb2 strains. We have also attempted to compare Clb2p levels and Cdc28p/Clb2p activity in wild-type and cdh1 cells, but we have not been successful due to difficulties with synchronizing cdh1 cultures. At this point, we do not know which interphase processes are being derailed in the cdh1 mutant. Because an effect on cell growth and viability arises only when the spindle assembly checkpoint is compromised, and because cdh1 mutants that lack the checkpoint have a severe chromosome loss phenotype, we suspect that some aspect of kinetochore or spindle function is impaired. Although we did not notice gross defects in spindle morphology in our cdh1 mutant, WASCH and CROSS 2002 recently reported that cdh1 strains do exhibit spindle abnormalities. If the problem lies in kinetochore assembly or in microtubule attachment to the kinetochore, a plasmid with a suboptimal centromere should be especially poorly transmitted in checkpoint-defective cdh1 cells. In fact, we observed that mad2 cdh1 cells lose a plasmid that has a minimal centromere (pRS412; CEN/ADE2; SIKORSKI and HIETER 1989 ) more than three times more often than they lose a plasmid that has a centromere that more closely resembles the chromosomal one (pDK243; Table 5; HOGAN and KOSHLAND 1992 ). Wild-type, mad2, and cdh1 cells, on the other hand, lose the two plasmids at approximately the same rate (our unpublished observation). Thus, we speculate that the absence of Cdh1p leads to defects in microtuble-kinetochore attachments in the subsequent cell cycle.

View larger version (23K): In this window In a new window Download PPT slide   Figure 6. Hypothetical model of the effect of loss of Cdh1p on the profile of Cdc28/Clb activity and Clb levels throughout the cell cycle (see text).

Several explanations may account for why pds1 cdh1 mutant strains are less viable and have a more severe chromosome loss phenotype than do mad2 cdh1 mutants. First, even though the spindle assembly checkpoint is absent in mad2 cdh1 cells, Pds1p still binds and inhibits Esp1p, delaying anaphase for at least the length of time it takes to degrade Pds1p. Cells lacking Pds1p, on the other hand, do not have the protection of this brief delay and may suffer more chromosome loss. Second, Pds1p may indirectly promote the activation of Sic1p: Pds1p enhances the accumulation of Esp1p in the nucleus (JENSEN et al. 2001 ; AGARWAL and COHEN-FIX 2002 ); Esp1p, in turn, promotes mitotic exit as part of the Cdc14 early anaphase release (FEAR) network that ultimately leads to the activation of Sic1p and Cdh1p (VISINTIN et al. 1998 ; SHOU et al. 1999 ; VISINTIN et al. 1999 ). Therefore, pds1 mutants may have lower FEAR activity than wild-type cells and, consequently, are likely to activate Sic1p less well. Because of the additional effect of reduced Sic1p activity, interphase Cdc28p/Clb activity in pds1 cdh1 may be even higher than that in mad2 cdh1, resulting in a more severe chromosome loss phenotype. This study demonstrates that Sic1p and Cdh1p play roles that are overlapping in some respects but distinct in others in maintaining genomic stability. Both cdh1 and sic1 mutants lose chromosomes. Chromosome loss in cdh1 mutants is effectively suppressed by overexpressing SIC1, suggesting that a target common to both Sic1p and Cdh1p is likely to be responsible for genomic instability in cdh1 strains. In sic1 mutants chromosome loss is due to a replication initiation defect (LENGRONNE and SCHWOB 2002 ). sic1 cells fail to assemble a full complement of prereplication complexes on origins in G₁, a problem that, intriguingly, escapes the notice of checkpoint systems, and attempt anaphase before replication is complete. This defect accounts for the high rate of GCR in sic1 mutants and the fact that plasmid transmission in these cells is improved if the plasmid has extra ARSs (LENGRONNE and S