The earliest known step in yeast spindle pole body (SPB) duplication requires Cdc31p and Kar1p, two physically interacting SPB components, and Dsk2p and Rad23p, a pair of ubiquitin-like proteins. Components of the PKC1 pathway were found to interact with these SPB duplication genes in two independent genetic screens. Initially, SLG1 and PKC1 were obtained as high-copy suppressors of dsk2Δ rad23Δ and a mutation in MPK1 was synthetically lethal with kar1-Δ17. Subsequently, we demonstrated extensive genetic interactions between the PKC1 pathway and the SPB duplication mutants that affect Cdc31p function. The genetic interactions are unlikely to be related to the cell-wall integrity function of the PKC1 pathway because the SPB mutants did not exhibit cell-wall defects. Overexpression of multiple PKC1 pathway components suppressed the G2/M arrest of the SPB duplication mutants and mutations in MPK1 exacerbated the cell cycle arrest of kar1-Δ17, suggesting a role for the PKC1 pathway in SPB duplication. We also found that mutations in SPC110, which encodes a major SPB component, showed genetic interactions with both CDC31 and the PKC1 pathway. In support of the model that the PKC1 pathway regulates SPB duplication, one of the phosphorylated forms of Spc110p was absent in pkc1 and mpk1Δ mutants.
SACCHAROMYCES cerevisiae cells commit to a new round of cell division during G1. Activation of the G1-specific Cdc28p/cyclin complex promotes bud emergence, DNA replication, and duplication of the yeast microtubule organizing center, the spindle pole body (SPB; Pringle and Hartwell 1981; Reed 1992; Cross 1995). These major G1 cell cycle events occur independently of each other, but in a coordinated manner that has yet to be understood.
The PKC1 cell integrity pathway has been implicated in numerous cellular processes including promoting bud emergence at the G1/S transition (Mazzoniet al. 1993; Mariniet al. 1996; Zarzovet al. 1996; Grayet al. 1997; Igualet al. 1997). Pkc1p (Levinet al. 1990; Paraviciniet al. 1992; Yoshidaet al. 1992) activates a mitogen-activated protein (MAP) kinase cascade consisting of Bck1p (Lee and Levin 1992), Mkk1p/Mkk2p (Irieet al. 1993), and the MAP kinase, Mpk1p (Leeet al. 1993). The PKC1 pathway is most likely regulated by the small GTP-binding protein Rho1p in vivo (Nonakaet al. 1995; Kamadaet al. 1996). Slg1p/Wsc1p, a plasma membrane protein, acts through Rho1p to activate the PKC1 pathway (Grayet al. 1997; Vernaet al. 1997; Jacobyet al. 1998). The MAP kinase Mpk1p phosphorylates downstream targets including Rlm1p, a MADS-box transcription factor (Watanabeet al. 1995; Dodou and Treisman 1997; Watanabeet al. 1997) that regulates transcription of a number of cell-wall components (Jung and Levin 1999). In addition, Mpk1p may also target the transcription factor complex Swi4p/Swi6p (Maddenet al. 1997).
Although duplication of the SPB in G1 is also Cdc28p/Cln dependent, the specific signals that activate SPB duplication are unknown. The yeast SPB is a trilaminar disc-like structure embedded in the nuclear envelope from which nuclear and cytoplasmic microtubules are organized (Byers and Goetsch 1974, 1975). The earliest step in SPB duplication involves the formation of an electron-dense material on the cytoplasmic face of the SPB, called the satellite. The satellite is the precursor of the nascent SPB, or the spindle plaque, which forms in the cytoplasm and is then embedded into the nuclear envelope (Adams and Kilmartin 1999). The early stages of SPB duplication are independent of Cdc28p activity since mutations in CDC28 result in cell cycle arrest with a single SPB already containing a satellite (Byers and Goetsch 1974). This observation suggests that the signal to initiate SPB duplication is active before the Cdc28p/START point, but for SPB duplication to continue, a CDC28-dependent signal is required.
Analysis of the early SPB duplication genes may give insight into how cell cycle control is exerted on SPB duplication because the G1 cell cycle machinery would be expected to target proteins required in the early stages of SPB duplication. Mutations in CDC31, the yeast homolog of centrin, block the earliest stage of SPB duplication, the formation of the satellite precursor (Byers 1981; Schildet al. 1981; Baumet al. 1986). In addition to SPB duplication, Cdc31p also plays a role in morphogenesis and cell integrity via interaction with Kic1p (Sullivanet al. 1998). Mutations in the kinase domain of Kic1p and certain cdc31 alleles result in abnormal bud morphology and lysis. However, kic1 and cdc31 mutants exhibit cell-wall defects that appear to be different than those observed in pkc1 pathway mutants (Sullivanet al. 1998).
Another SPB component, Kar1p, is required at the same step of SPB assembly as Cdc31p (Rose and Fink 1987). Kar1p genetically and physically interacts with Cdc31p and is required to localize Cdc31p to the SPB (Biggins and Rose 1994; Vallenet al. 1994; Spanget al. 1995). The temperature-sensitive mutation kar1-Δ17 mislocalizes Cdc31p at high temperature and can be rescued by high dosage of CDC31 and the dominant allele CDC31-16. Formally, therefore, Kar1p functions upstream of Cdc31p. DSK2 is a nonessential ubiquitinlike protein identified as a dominant suppressor that also relocalizes Cdc31p in kar1-Δ17 mutants (Biggins and Rose 1994; Vallenet al. 1994; Bigginset al. 1996). Deletion of DSK2 in combination with deletion of another ubiquitin-like gene, RAD23, results in a temperature-sensitive block in SPB duplication prior to satellite formation (Bigginset al. 1996). Unlike mutations in KAR1, the dsk2Δ rad23Δ strain is not defective for Cdc31p localization at the SPB. However, like kar1-Δ17 mutants, the SPB duplication defect of dsk2Δ rad23Δ can be suppressed by high-copy CDC31 and by CDC31-16. These results suggest that Dsk2p/Rad23p also function upstream of Cdc31p to mediate an important function of Cdc31p at the SPB (Bigginset al. 1996).
By employing the SPB duplication mutants kar1-Δ17 and dsk2Δ rad23Δ in different genetic screens, we identified multiple genetic interactions between the PKC1 and the SPB duplication pathways. Members of the PKC1 pathway, in high copy, could suppress the temperature sensitivity of kar1-Δ17, dsk2Δ rad23Δ, and cdc31-2. Over-expression of different members of the PKC1 pathway specifically suppressed the SPB duplication mutants but not other G2/M-arrested cells. In addition, we analyzed many combinations of double mutants and found extensive synthetic lethality and synthetic growth defects. To understand the functional basis of the synthetic interactions, we analyzed the phenotypes of the double mutants with respect to cell integrity and G2/M arrest. In a mpk1Δ kar1-Δ17 double mutant, the G2/M arrest defect was clearly exacerbated. We also found that phosphorylation of Spc110p, an essential SPB component, was defective in pkc1ts and mpk1Δ mutants, suggesting that the PKC1 pathway may directly or indirectly regulate Spc110p phosphorylation. Our results demonstrate an important functional link between PKC1 pathway activity and SPB duplication. We propose that PKC1 MAP kinase signaling positively regulates an SPB component, possibly to coordinate SPB duplication and bud emergence during G1.
MATERIALS AND METHODS
Microbial techniques and yeast strain construction: Yeast media and microbial techniques were essentially as previously described (Roseet al. 1990). Bacterial media were as described (Sambrooket al. 1989), and bacterial strain XL1-Blue was used for all bacterial manipulations. All bacterial strains are listed in Table 1.
All yeast strains used are listed in Table 2. Yeast strains were constructed using standard genetic techniques. All yeast strains were isogenic to S288c. To generate double mutants for synthetic lethality studies, strains with mutations in cdc31-1, kar1-Δ17, spc10-220, mps2-1, and dsk2Δ rad23Δ were kept covered with a URA3 plasmid bearing the appropriate wild-type gene to prevent increased ploidy. PKC1 pathway components were then disrupted in the covered mutant strains. The resulting double mutant strains containing wild-type (WT) URA3 plasmids were streaked on 5-fluoroorotic acid (5-FOA) at 23° to select for plasmid loss (Boekeet al. 1984). When double or triple mutant combinations were viable on 5-FOA at 23°, growth at higher temperatures was assayed by spotting 10-fold serial dilutions.
In most cases, double mutants between PKC1 pathway components and SPB duplication genes were generated either by PCR-mediated gene disruption (Baudinet al. 1993) or with disruption plasmids listed in Table 1. In a few cases, the double/triple mutants were generated as segregants of genetic crosses. Specifically, the slg1Δ kar1-Δ17 strain (MY4646) was constructed by mating a kar1-Δ17 [KAR1 URA3] (MS2373) strain to a slg1Δ (MY4455), a cdc31-1 slg1Δ double mutant (MY6862) by crossing MY4455 to MY3885, the dsk2Δ rad23Δ slg1Δ (MY4916) strain by mating MY4449 to MY4455, the pkc1ts kar1-Δ17 strain (MY6805) by crossing DL523 (Levin and Bartlett-Heubusch 1992) to MS2374, and a dsk2Δ rad23Δ pkc1ts strain (MY6296) by mating DL523 (Levin and Bartlett-Heubusch 1992) to MY4197.
kar1-Δ17 synthetic lethality screen and identification of mpk1: To identify mutations that result in synthetic lethality with kar1-Δ17, a “shuffle-mutagenesis” screening strategy was employed. MS2373 and MS2376, ura3 kar1-Δ17 strains of both mating types, were constructed, each carrying a KAR1 URA3-based centromeric plasmid to suppress the mutant defect. The strains were mutagenized with ethyl methanesulfonate (Roseet al. 1990) to 40% viability and plated on synthetic complete medium lacking uracil (SC-ura) to select for the covering plasmid. Mutations that cause slow growth or lethality after loss of the covering plasmid were assayed by their sensitivity to 5-FOA media, which selects against the URA3 plasmid. Out of 60,000 colonies screened, we identified four 5-FOA-sensitive mutants whose 5-FOA sensitivity was recessive and segregated as single genes. The 5-FOA sensitivity of the four mutants was rescued by a second plasmid bearing KAR1 on a LEU2-based vector, as expected for mutations that affect KAR1 function. Complementation analysis revealed that two of the mutations defined unique linkage groups and were subsequently identified as mutations in REG1 and NEM1 (W. Khalfan and M. D. Rose, unpublished observations). The remaining two mutations, SL6 and SL18, belonged to the same linkage group and caused temperature sensitivity at 37°, and the temperature sensitivity was linked to 5-FOA sensitivity because they cosegregated in crosses (<3.1 cM).
We cloned the wild-type gene corresponding to the SL6 mutation by complementation of the temperature sensitivity of kar1-Δ17 SL6 strain bearing the KAR1 LEU2 plasmid (MS5589) strain. MS5589 was transformed with a URA3 YCp50 library (Roseet al. 1987). Of 18,000 transformants examined, 7 were Ts+. All seven plasmids rescued the temperature sensitivity of SL6 when reintroduced into yeast. The insert junctions of three plasmids were sequenced and contained two open reading frames in common, YHR029C and MPK1/SLT2. A plasmid containing only the MPK1/SLT2 gene (p666, kindly provided by Dr. Levin), rescued the temperature sensitivity of MS5589 and the 5-FOA sensitivity of MS5584. To confirm that the MPK1 gene and SL6 locus were allelic, an mpk1Δ kar1-Δ17 double mutant (MS5886) was generated, shown to be 5-FOA sensitive, and the 5-FOA sensitivity could be suppressed with a LEU2 MPK1 plasmid. To establish linkage, MS5831, a strain containing the SL6 mutation alone, was crossed to a mpk1Δ strain (MS5933). The resulting diploid strain was Ts−, indicating that mpk1Δ and SL6 fail to complement. Because a homozygous mpk1Δ diploid cannot sporulate, an MPK1 LEU2 plasmid was introduced into the diploid. Upon sporulation, all spores that did not contain the MPK1 LEU2 plasmid (43/43) were temperature sensitive, indicating that there were no wild-type recombinants. We concluded that MPK1 and SL6 are allelic because they are closely linked (<4.7 cM).
dsk2Δ rad23Δ high-copy suppressor screen: Strain MY3592 was transformed with a YEp24 genomic library (Carlson and Botstein 1982). A total of 21,300 Ura+ transformants (17 genome equivalents) were selected on SC-ura medium at 30° and subsequently replica printed to SC-ura at 37°. Thirty putative Ts+ colonies were picked and retested. Plasmids were recovered from yeast as previously described (Roseet al. 1990) and plasmid linkage of the Ts+ phenotype was tested after retransformation into MY3592. The major class of plasmids (11) contained the CDC31 gene. RAD23 was isolated twice. DSK2 was not isolated, presumably because of its toxicity when overexpressed. Two independent secondary screens were used to classify the remaining 17 suppressors. Because rad23Δ, but not dsk2Δ, causes UV sensitivity, two plasmids that conferred UV resistance were deemed specific to rad23Δ and not studied further. To identify suppressors that were specific for SPB duplication, the plasmids were tested for their ability to suppress KAR1 and CDC31 mutations. Nine plasmids partially suppressed CDC31 mutations in an allele-specific manner. To identify the suppressing genes, we sequenced the ends of the library inserts using primers from the YEp24 vector. Two of them contained the SLG1 gene and were strong suppressors of both kar1-Δ17 and cdc31. One plasmid contained the PKC1 gene.
Microscopy: To examine the G2/M arrest phenotype of the SPB duplication mutants, strains were grown at 23° to early logarithmic phase and one-half of the cultures were shifted to the indicated temperatures for 4-8 hr. To examine the nuclear morphology, 1 ml of each culture was collected by centrifugation and stained with 4′,6-diamidino-2-phenylindole (DAPI) essentially as described (Roseet al. 1990). DAPI was obtained from Accurate Biochemicals and Scientific Corp. (Westbury, NY).
For cell cycle analysis of mpk1Δ kar1-Δ17, strain MS5886 was grown on 5-FOA at 23° for several days until small, slow-growing colonies appeared (incubation at lower temperatures failed to give any colonies). As a control, strain MS2373 was also grown on 5-FOA at 23°. The colonies arising on the 5-FOA plate were propagated in YPD liquid where the double mutant strain continued to grow very slowly compared to the control strain.
For live/dead analysis, strains were grown as described above. FUN-1 dye (Millardet al. 1997), was added to 1 ml of culture to a final concentration of 10 μm (Molecular Probes, Eugene, OR). The cultures were incubated at room temperature, in the dark, for 0.5 hr. Cells were examined by differential interference and fluorescence microscopy (Axiophot, Carl Zeiss, Inc., Thornwood, NY).
Spc110 Western blot analysis: Affinity-purified Spc110 antibodies were obtained from T. Davis (University of Washington, Seattle, WA) and crude Spc110 antibodies were obtained from Dr. Stirling (University of Dundee, Dundee, United Kingdom). Crude Spc110 antibodies were purified against a GST-Spc110p fusion (Stirlinget al. 1994). Protein extracts were prepared by trichloroacetic acid (TCA) precipitation (Wrightet al. 1989). Protein extracts were separated on a 6.5% SDS-PAGE as described (Laemmli 1970). A 1:1000 dilution of affinity-purified Spc110p antibodies was used for Western blotting.
Overexpression of members of the PKC1/MPK1 pathway suppresses SPB duplication mutants: Mutations in CDC31, KAR1, DSK2, and RAD23 result in defects at an early step during SPB duplication. Specifically, cdc31, kar1-Δ17, and dsk2Δ rad23Δ mutants arrest in G2/M as large-budded cells with a monopolar spindle and a single SPB (Byers 1981; Vallenet al. 1992; Bigginset al. 1996). All combinations of dsk2Δ, rad23Δ, and kar1Δ17 mutants can be suppressed by high-copy CDC31, suggesting that Cdc31p is downstream in this pathway. To identify other genes that interact with the CDC31-related network of SPB duplication genes, we screened for highcopy suppressors of the temperature sensitivity of dsk2Δ rad23Δ (see materials and methods). As expected, the most frequent high-copy suppressor was CDC31. In addition, we identified SLG1/WSC1 and PKC1, as partial suppressors of the growth defect of dsk2Δ rad23Δ at restrictive temperatures (Figure 1A). SLG1/WSC1 encodes a plasma membrane protein that signals to Pkc1p via Rho1p (Vernaet al. 1997; Jacobyet al. 1998). Pkc1p activates the MAP kinase module consisting of Bck1p, Mkk1p/Mkk2p, and Mpk1p (Lee and Levin 1992; Irieet al. 1993; Leeet al. 1993).
To determine whether the suppression by high-copy SLG1 and PKC1 was relevant to SPB duplication, we examined their ability to suppress mutations in SPB components that show genetic interactions with dsk2Δ rad23Δ. As shown in Figure 1, B and C, high-copy SLG1 and PKC1 also partially suppressed kar1-Δ17 and cdc31-2. Two other alleles of CDC31, cdc31-1 and cd31-5, were not suppressed by any of the plasmids at any temperature (data not shown). We also tested whether the downstream components of the PKC1 pathway, BCK1-20 and MPK1 2μ, could suppress dsk2Δ rad23Δ, kar1-Δ17, and cdc31-2. As shown in Figure 1, BCK1-20 was a suppressor of kar1-Δ17 but not of the other mutations. High-copy MPK1 was a very weak suppressor of kar1-Δ17. All highcopy suppression results are summarized in Figure 1D. These results show that overexpression of multiple components of the PKC1 pathway can suppress mutations in all of the genes that are known to affect the first step in SPB duplication. Strikingly, the upstream components of the PKC1 pathway, PKC1 and SLG1, were consistently the strongest suppressors.
Slg1p suppresses the SPB duplication defects through Pkc1p: High-copy SLG1 was a better suppressor of kar1-Δ17 than high-copy PKC1. On the other hand both were equally good suppressors of cdc31-2. We therefore wanted to determine whether SLG1 suppresses the SPB defects through PKC1 or via an independent pathway. We tested whether PKC1 lies downstream of SLG1 for suppression of SPB duplication mutants by asking whether high-copy PKC1 can suppress a dsk2Δ rad23Δ slg1Δ triple mutant. The triple mutant showed a stronger temperature-sensitive growth defect than the dsk2Δ rad23Δ double mutant. Nevertheless, high-copy PKC1 suppressed both the dsk2Δ rad23Δ slg1Δ triple mutant and the dsk2Δ rad23Δ double mutant well (Figure 2A). Therefore, Pkc1p does not require Slg1p to suppress dsk2Δ rad23Δ consistent with PKC1 being downstream of SLG1. We next tested whether Slg1p required Pkc1p for suppression. This experiment is complicated by the fact that PKC1 is an essential gene. We therefore determined whether high-copy SLG1 could suppress a dsk2Δ rad23Δ pkc1ts triple mutant. Once again, the temperature sensitivity of the dsk2Δ rad23Δ pkc1ts mutant was more severe than that of either dsk2Δ rad23Δ or pkc1ts. We found that high-copy SLG1 did not suppress the dsk2Δ rad23Δ pkc1ts triple mutant at 37° (Figure 2B), consistent with Slg1p being upstream of Pkc1p's essential function. However, we found that, at an intermediate temperature of 32°, high-copy SLG1 did suppress the dsk2Δ rad23Δ pkc1ts triple mutant as well as DSK2 (Figure 2B). This result can be interpreted in two ways. First, the pkc1ts allele may be partially active at 32°, and high-copy SLG1 may be suppressing by enhancing the partial activity. Alternatively, suppression at 32° may be due to a function of Slg1p that is partially independent of Pkc1p.
Overexpression of Pkc1p pathway components does not suppress G2/M-arrested mutants in general: The SPB duplication mutations that showed genetic interactions with the PKC1 pathway cause a G2/M arrest at the restrictive temperature. Therefore, we considered the possibility that the PKC1 pathway may simply be required for maintaining the integrity of large-budded cells at G2/M. If so, activation of the pathway may simply suppress the growth defect of SPB mutants by suppressing the loss of viability of G2/M-arrested cells. To address this possibility, we examined whether high-dosage SLG1 or PKC1 could rescue the temperature sensitivity of a variety of mutants that arrest at G2/M through different mechanisms. We examined cdc13-1, defective for telomere metabolism (Wood and Hartwell 1982; Hartwell and Smith 1985; Garviket al. 1995), cdc15-2, defective in APC activation in M phase (Jaspersenet al. 1998), and cdc7-1, defective for DNA replication (Njagi and Kilbey 1982; Sclafaniet al. 1988). Moreover, to address whether SPB mutants may generally interact with the PKC1 pathway, we tested two other SPB duplication mutants not known to be linked to Cdc31p, spc110-220 (Sundberget al. 1996) and mps2-1 (Wineyet al. 1991). We found that cdc13-1, mps2-1, cdc15-2, or cdc7-1 strains were not suppressed by 2μ SLG1 or 2μ PKC1 (Figure 3). Therefore, we conclude that overexpression of PKC1 pathway genes do not generally suppress mutants that arrest in G2/M.
Another trivial possibility is that overexpression of PKC1 pathway components suppresses SPB duplication mutants by extending the length of G1 and thereby allowing extra time for SPB duplication to occur. To test this possibility, we monitored cell cycle progression in strains either overexpressing or deleted for PKC1 pathway components after release from various synchronizations. We found that in all cases the cells proceeded through the cell cycle with equivalent kinetics (Ivanovska and Rose 2000, and data not shown). Therefore, the PKC1 pathway does not suppress the SPB duplication mutants by altering progression through G1.
SPC110 is genetically linked to the CDC31 network of SPB duplication genes: Strikingly, the spc110-220 mutant strain was strongly suppressed by high-copy PKC1 and SLG1 (Figure 3). This observation raised the possibility that Spc110p may have functions related to those of Kar1p, Cdc31p, or Dsk2p and Rad23p. Spc110p is a component of the inner SPB plaque that binds calmodulin (Geiseret al. 1993; Kilmartinet al. 1993). The spc110-220 mutation leads to defective SPB assembly at the restrictive temperature ostensibly because of reduced interaction between Spc110p and calmodulin (Sundberget al. 1996). Although binding of Cdc31p to Spc110p has been observed in vitro (Geieret al. 1996), no evidence for in vivo interaction has been reported. To test whether spc110-220 shares characteristics with kar1-Δ17 and dsk2Δ rad23Δ, we asked whether spc110-220 could be suppressed by high-copy CDC31 or CDC31-16, known suppressors of dsk2Δ rad23Δ and kar1-Δ17 (Vallenet al. 1994; Bigginset al. 1996); see Figure 4. We found that high-copy CDC31, but not CDC31-16, partially rescued spc110-220 (Figure 4). In contrast, mps2-1, an SPB mutant that is not suppressed by overexpression of the Pkc1p pathway, was also not suppressed by any of the plasmids. Therefore, we concluded that high-copy CDC31 suppressed spc110-220 specifically. Because CDC31-16 did not suppress spc110-220, suppression of spc110-220 must occur by a mechanism different from that of kar1-Δ17. These results suggest that Cdc31p also interacts with Spc110p in vivo, consistent with the in vitro data. In summary, increased dosage of Pkc1p and Slg1p suppressed all four mutations that can be suppressed by high-copy Cdc31p. These findings suggest that the PKC1 pathway interacts with one or several proteins in the Cdc31p SPB assembly network.
Mechanism of high-dosage suppression: We next wanted to investigate how high-dosage SLG1 and PKC1 suppressed the growth defect of the SPB mutants. The simplest model is that the PKC1 pathway contributes to SPB function. However, the PKC1 pathway has a well-established role in cell-wall integrity. Therefore, there is a formal possibility that genes in the CDC31 network play an additional role in cell-wall integrity and that these SPB mutants suffer cell-wall defects that can be suppressed by SLG1 and PKC1. The compromised cell walls in PKC1 pathway mutants cause cell lysis and death. We therefore tested whether the kar1-Δ17 and dsk2Δ rad23Δ mutants exhibited cell lysis, using FUN-1, a fluorescent viability probe (Millardet al. 1997). The majority of pkc1ts mutant cells (98%) were dead after 6 hr at the nonpermissive temperature of 37° (Table 3). In contrast, under similar conditions, we found only a small percentage of dead cells in the dsk2Δ rad23Δ and kar1-Δ17 strains at the restrictive temperature (8 and 16% respectively) and these were not affected by high-copy SLG1 or high-copy PKC1 (Table 3). These results indicate that decreased lysis/cell death is not likely to be the mechanism by which SLG1 and PKC1 suppress the SPB duplication mutants.
We next tested whether high-dosage SLG1 and PKC1 suppressed the G2/M arrest of kar1-Δ17, dsk2Δ rad23Δ, and spc110-220 mutants, caused by the failure in SPB duplication. The G2/M arrest is demonstrated by large-budded cells with a single nucleus, indicating failure of the duplicated DNA to segregate. If high-dosage SLG1 and PKC1 suppressed the SPB duplication defects of the mutants, we would expect a decrease in the percentage of G2/M-arrested cells. Indeed, SLG1 and PKC1 overexpression partially suppressed the G2/M arrest defect of kar1-Δ17 at 35° (from 44 to 30%) and SLG1 overexpression partially suppressed at 37° (from 40 to 23%) (Table 4), consistent with the partial suppression of the temperature sensitivity of kar1-Δ17. Similarly, the G2/M defect of dsk2Δ rad23Δ was partially suppressed by high-copy SLG1 at 37° (from 71 to 44%; Table 4). Moreover, high-copy PKC1 suppressed the G2/M arrest phenotype of spc110-220 (from 76 to 36%). Therefore, SLG1 and PKC1 partially suppressed the G2/M arrest of kar1-Δ17, dsk2Δ rad23Δ, and spc110-220 mutants, implying that they suppress the SPB duplication defects. Our findings support the idea that SLG1 and PKC1 provide a positive function for SPB duplication.
Synthetic lethal interactions between SPB duplication mutants and multiple members of the PKC1 pathway: If the Pkc1p pathway does regulate SPB duplication, then PKC1 pathway mutations should be partially compromised for SPB duplication. If so, their function should be revealed by synthetic lethal interactions with the relevant SPB mutants. In an independent genetic screen, we identified mutations in MPK1, the PKC1 pathway MAP kinase, as being synthetically lethal with kar1-Δ17 (see materials and methods). In Figure 5A, we demonstrate synthetic lethality by the inability of a strain to segregate a URA3-based plasmid containing a wild-type copy of one of the genes mutated on the chromosome. kar1-Δ17 and mpk1Δ all grew on 5-FOA, a wild-type control, at 23°. In contrast, the mpk1Δ kar1-Δ17 double mutant strain was extremely sensitive to 5-FOA because of a requirement for the KAR1 URA3 plasmid for survival. Thus, the mpk1Δ kar1-Δ17 double mutant combination is synthetically lethal.
We next asked whether the synthetic lethality observed between mpk1 and kar1-Δ17 could be extended to other SPB components and members of the PKC1 pathway. First, to test whether other SPB mutants showed synthetic lethal interactions with mpk1Δ, we deleted MPK1 in the dsk2Δ rad23Δ, cdc31-1, and spc110-220 mutant strains containing the appropriate wild-type gene on URA3 plasmids. The mpk1Δ was synthetically lethal with dsk2Δ rad23Δ and spc110-220 mutations (Figure 5A). The cdc31-1 mpk1Δ double mutant showed a greatly exacerbated growth defect at temperatures permissive for either single mutant (Figure 5B). In contrast, the mps2-1 mpk1Δ double mutant did not show a synthetic lethal phenotype (Figure 5A) confirming that the synthetic lethal interactions are restricted to the CDC31 network of SPB duplication genes. In summary, all the CDC31-related SPB mutants analyzed showed either synthetic lethality or synthetic growth defects when MPK1 was deleted.
Next, we tested whether mutations in other components of the PKC1 pathway showed synthetic lethal interactions with the SPB duplication mutants. All the double mutant combinations tested showed synthetic lethality or severely exacerbated growth defects (Figure 5, A and B). In particular, the bck1Δ kar1-Δ17 double mutant was synthetically lethal, while the other combinations had restrictive temperatures lower than either parent. Figure 5C summarizes all of the exacerbated growth defects and synthetic lethal interactions observed. On the basis of both high-dosage suppression and synthetic lethal genetic interactions, we conclude that the PKC1 pathway provides a function that is crucial for survival of the SPB duplication mutants.
Characterization of the double mutants: The genetic interactions presented thus far do not distinguish whether the synthetic defects affect SPB duplication per se. Although this seems most likely, an alternative is that mutations in SPB components could have secondary defects in cell integrity that are exacerbated in the absence of the PKC1 pathway. To assess the nature of the synthetic defects directly, we analyzed the phenotypes of the double mutant strains. This analysis is technically compromised by the fact that the terminal phenotype associated with a defect in SPB duplication (arrest at G2/M with a large bud and an unduplicated SPB) requires an extended period of bud growth. Bud growth is extremely sensitive to defects in the PKC1 pathway causing cells to lyse before the large bud stage. Thus, at the restrictive temperature, most double mutants arrested as a mixture of small-budded and large-budded cells and so were not informative as to whether there was an increase in the SPB duplication defect (data not shown). Synchronizing the cultures in G1 with α-factor, followed by release from arrest, did not alleviate the problem (data not shown). However, mpk1Δ exhibited a strong synthetic lethality with kar1-Δ17 at a temperature where the single mutant did not lyse (Table 5). A slow-growing mpk1Δ kar1-Δ17 mutant propagated at 23° showed an increase in G2/M-arrested cells (53%) relative to the kar1-Δ17 single mutant (33%) without showing an elevated frequency of dead cells (Table 5). Thus, this double mutant showed an increase in G2/M arrest, consistent with the hypothesis that the PKC1 pathway plays a positive role in SPB duplication.
Pkc1p pathway activity is required for proper Cdc31p function in vivo: The genetic interactions suggest that Pkc1p pathway activity is required for optimal function of SPB duplication genes. All of the genes suppressed by PKC1 overexpression can also be suppressed by high-copy CDC31. Therefore, we next asked whether the PKC1 pathway is required for Cdc31p to suppress the SPB mutations. Specifically, we tested whether high-copy CDC31 and the dominant alleles CDC31-16 and DSK2-1 could suppress kar1-Δ17 in the absence of the Pkc1p pathway component Slg1p. The dominant alleles CDC31-16 and DSK2-1, but not high-copy CDC31, can suppress a complete deletion of KAR1 (Vallenet al. 1994) and suppress kar1-Δ17 as well as KAR1 (Figure 6, top). Previous analysis has failed to uncover any mutations that compromise the ability of CDC31-16 and DSK2-1 to suppress kar1-Δ17 (W. Khalfan and M. Rose, unpublished observations; Bigginset al. 1996). As shown in Figure 6, CDC31-16 or DSK2-1 plasmids did not suppress the slg1Δ kar1-Δ17 as well as KAR1 at 37°. Moreover, high-copy CDC31 completely failed to suppress the double deletion even at 30° (Figure 6, bottom). These results can be interpreted in two ways. First, Slg1p may be required for optimal function of Cdc31p. By this scheme the PKC1 pathway would lie upstream or in a parallel pathway leading to activation of Cdc31p. Alternatively, Cdc31p itself may be required to activate the PKC1 pathway. In this scenario, the PKC1 pathway would lie downstream of Cdc31p, and mutations in cdc31 and the other SPB duplication genes would be predicted to alter PKC1 pathway activity. We tested this second idea by measuring PKC1 pathway activity in the SPB duplication mutants in a number of ways. First, as a measure of MKK1 and MKK2 activity, we tested whether MPK1 was phosphorylated in a kar1-Δ17 strain upon mild heat shock treatment (Kamadaet al. 1995; Vernaet al. 1997). We found that Mpk1p's ability to be phosphorylated was not compromised in the kar1-Δ17 strain (data not shown). We also examined the in vitro kinase activity of Pkc1p immunoprecipitated from a kar1-Δ17 strain, using myelin basic protein (MBP) as the substrate (Watanabeet al. 1994). Again, we found no evidence that Pkc1p kinase activity was affected, at least in vitro (data not shown). Finally, using a LacZ-reporter assay system, we examined the transcriptional activity of Rlm1p, a MADS-box transcription factor whose activity depends on Mpk1p phosphorylation (Watanabeet al. 1997). Although Rlm1p transcriptional activity was indeed lower in the kar1-Δ17 strain, this was a general effect of G2/M arrest, since another G2/M mutant, cdc13, also showed reduced levels of Rlm1p transcriptional activity. Taken together, it seems unlikely that the CDC31 pathway activates the PKC1 pathway. We therefore interpret the inability of high-copy CDC31 to suppress the kar1-Δ17 slg1Δ double mutant (Figure 6) as being due to reduced activity or levels of Cdc31p in the slg1Δ mutant. The observation that CDC31-16 and DSK2-1, but not high-copy CDC31, could partially suppress the kar1-Δ17 slg1Δ double mutant at 37° suggests that the dominant suppressors may act by a different mechanism that is partially independent of the PKC1 pathway. In support of the idea that they act differently, CDC31 2μ, but not CDC31-16, suppressed the temperature-sensitive growth defect of spc110-220 (Figure 4).
Phosphorylation of Spc110p is defective in pkc1ts and mpk1Δ mutants: We next sought to explore the molecular mechanism by which the PKC1 pathway affects SPB duplication. One possibility is that the PKC1 pathway phosphorylates an SPB component. On the basis of the genetic interactions, possible candidates include Kar1p, Cdc31p, and Spc110p. We first examined Kar1p in the different PKC1 pathway mutants and found no change in the mobility of various Kar1p forms (data not shown). We next tested Spc110p, which exists as at least two distinct phosphorylated forms of molecular weights 112 kD and 120 kD. The 120-kD form (p120) arises from additional serine/threonine phosphorylation of the 112-kD form and is cell cycle regulated, predominating in small-budded cells with duplicated DNA and short spindles (Friedmanet al. 1996; Stirling and Stark 1996). The SPB duplication gene Mps1p is one of the kinases responsible for Spc110p phosphorylation (M. Winey and T. Davis, unpublished communication). On the basis of the time of the appearance of the p120 form and analysis of cdc mutants, the phosphorylation that produces the p120 form does not appear to be required for SPB duplication (Friedmanet al. 1996). We reasoned that the phosphorylation state of Spc110p could serve as a useful marker for earlier event(s) in SPB duplication. We therefore examined the phosphorylation of Spc110p in asynchronous wild-type, pkc1ts, and mpk1Δ cultures grown in YPD at 30° and shifted to 37° for 3 hr. An asynchronous wild-type culture grown at 37° contained both p120 and p112 forms, α-factor-arrested wild-type strain contained the faster migrating p112 form, and nocodazole-arrested wild-type cells contained the p120 form (Figure 7 and Friedmanet al. 1996; Stirling and Stark 1996). In contrast to the wild-type strain, the p120 form was absent in the pkc1ts mutant at 37°. Similarly, an mpk1Δ culture showed reduced levels of p120 at permissive temperature (30°) and drastically reduced levels of p120 at 37°.
The disappearance of the p120 form in pkc1ts and mpk1Δ strains is not likely to be due to the accumulation of cells at a stage in the cell cycle prior to Spc110p phosphorylation. First, although the pkc1ts mutant accumulates small-budded cells, the small-budded cells have already initiated DNA replication and formed short spindles (Levin and Bartlett-Heubusch 1992; Ivanovska and Rose 2000). Therefore, on the basis of the pkc1ts mutant phenotype, we would not expect a block in the production of the p120 form due to cell cycle arrest. Second, in the case of the mpk1Δ mutant, the strain did not arrest at any particular stage of the cell cycle at the restrictive temperatures. We directly examined the cell cycle distribution of pkc1ts and mpk1Δ strains from aliquots taken at the time of protein extract preparation. As shown in Table 6, while there were differences in their cell cycle distribution compared to wild type, the distribution was not correlated with the presence or absence of the p120 form. These results suggest that the lack of phosphorylation is not due to a block in cell cycle progression. Taken together, our results suggest that Pkc1p and/or Mpk1p may directly or indirectly phosphorylate Spc110p. These results may provide a mechanistic basis for the genetic interactions between thePKC1 pathway and SPB duplication genes.
Swi4p but not Rlm1p function is required in SPB duplication mutants: Having established that Pkc1p and the MAP kinase module affect SPB duplication, we wanted to identify the downstream effectors of the PKC1 pathway that transduce the signal to the SPB. One well-established downstream target of Mpk1p is Rlm1p, a MADS-box family transcription factor whose activity is MPK1 dependent (Watanabe et al. 1995, 1997; Dodou and Treisman 1997). However, unlike mpk1Δ, deletion of RLM1 does not result in cell morphogenesis and/or lysis defects, indicating that there must be additional downstream effectors of MPK1 activity. The transcription factor complex consisting of Swi4p/Swi6p (SBF) has also been suggested to be a second downstream effector of MPK1 (Maddenet al. 1997). Swi4p and Swi6p physically associate with Mpk1p, and phosphorylation of Swi4p and Swi6p is dependent on Mpk1p (Maddenet al. 1997). These results suggest that Swi4/Swi6 complex may be a target of PKC1 pathway activity. However, SBF activity is also thought to be required to turn on the PKC1 pathway via Cdc28p/Clns (Mazzoniet al. 1993; Mariniet al. 1996; Zarzovet al. 1996; Grayet al. 1997). To determine whether Rlm1p and/or Swi4p are important for the survival of the SPB mutants, we generated kar1-Δ17 rlm1Δ and kar1-Δ17 swi4Δ double mutants containing KAR1 URA3 plasmids and examined their ability to lose the plasmid on 5-FOA at a range of temperatures. As shown in Figure 8A, we found that the kar1-Δ17 swi4Δ strain was synthetically lethal, similar to the kar1-Δ17 mpk1Δ double mutant. In contrast, the kar1-Δ17 rlm1Δ double mutant did not grow any more poorly than either single mutant alone at a range of temperatures (Figure 8B). Since the cdc31-1 mpk1Δ double mutant also has a severe growth defect, we examined whether cdc31-1 mutants were sensitive to mutations in either SWI4 or RLM1. Similar to the results obtained with the kar1-Δ17 swi4Δ double mutant, a cdc31-1 swi4Δ double mutant was synthetically lethal, whereas a cdc31-1 rlm1Δ double mutant did not show any synthetic growth defects (Figure 8, C and D). These results argue that Swi4p's function is crucial for survival of the SPB duplication mutants, kar1-Δ17 and cdc31-1, in activating the PKC1 pathway and/or transmitting a signal downstream of Mpk1p. Clearly however, Rlm1p is not the target of Mpk1p activity that is relevant for SPB duplication.
Because Swi4p might mediate its effects through transcriptional regulation, and because Cdc31p function is compromised in the kar1-Δ17 slg1Δ mutant (Figure 6), one possible explanation for the genetic interactions would be if the PKC1 pathway regulated CDC31 transcription. To address this question, we examined the levels of CDC31 mRNA in asynchronously growing wild-type, pkc1ts, mpk1Δ, and swi4Δ mutants either at 23° or 37°. We found that the levels of CDC31 mRNA were similar in the mutant and wild-type strains (data not shown). We also found the levels of Cdc31p protein in pkc1ts mutant to be comparable to that of wild type at 37°. In addition, the levels of Kar1p and Spc110p were wild type in pkc1ts and mpk1Δ strains (data not shown). Therefore, it is unlikely that the PKC1 pathway regulates the synthesis of Cdc31p, Kar1p, or Spc110p. It remains possible, however, that the expression of an unidentified SPB component is regulated by the PKC1 pathway.
Genetic interactions with SPB duplication genes suggest a role for the PKC1 pathway in SPB duplication: We identified numerous genetic interactions between SPB duplication genes in the CDC31 pathway and the PKC1 signal transduction cascade. Overexpression of SLG1 or PKC1 partially rescued the temperature sensitivity of dsk2Δ rad23Δ, kar1-Δ17, and cdc31-2. We assume that overexpression leads to activation of the PKC1 pathway. The suppression was specific to SPB duplication defects and was not due to general suppression of G2/M arrest. High-dosage SLG1 and PKC1 did not simply rescue cell-wall-related defects of the SPB mutants, because we did not find significant cell lysis in dsk2Δ rad23Δ and kar1-Δ17. It is interesting that the most upstream members of the PKC1 pathway, SLG1 and PKC1, were the strongest suppressors. We present two interpretations of this observation. First, the PKC1 pathway may branch at one or more points and the different components of the PKC1 pathway may suppress the SPB duplication mutants by different mechanisms. Alternatively, consistent with the observations of other researchers, overexpression of upstream components may activate the PKC1 pathway more strongly than overexpression of downstream components, resulting in different levels of suppression of the SPB duplication mutants (Grayet al. 1997).
We also observed somewhat different behaviors among the different PKC1 pathway components in our synthetic lethal analyses. The severity of the interactions was the reverse of the suppression, with the most downstream components showing the most severe synthetic phenotypes in combination with SPB duplication mutations. The SPB mutants were particularly sensitive to mutations in MPK1. With the exception of cdc31, all the SPB mutants showed synthetic lethality with mpk1Δ, while mutations in the other PKC1 pathway genes resulted in viable strains with more severe growth defects. In addition, the mpk1Δ kar1-Δ17 double mutant had an exacerbated G2/M arrest phenotype, suggesting a requirement for the MPK1 gene in SPB duplication. It is possible that Mpk1p is crucial for the survival of SPB mutants because it is targeted and activated by other MAP kinase signaling pathways, especially in the absence of upstream components of the PKC1 pathway. The presence of such cross-talk may help to explain why mutations in upstream components do not exacerbate the growth defect of SPB duplication mutants as severely as mutations in MPK1.
The SPB mutants cdc31 and kar1 showed synthetic lethality with swi4Δ arguing that, like Mpk1p, Swi4p is also crucial for the survival of the SPB duplication mutants. The Swi4p/Swi6p transcription factor complex (SBF) is a proposed target of Mpk1p (Maddenet al. 1997). However, SBF is also an activator of the PKC1 pathway via the Cdc28p/Clns (Grayet al. 1997). Therefore, while the synthetic lethal results argue that Swi4p is as crucial as Mpk1p, they do not discriminate whether the upstream or downstream activity of Swi4p is important. The lack of genetic interactions between the SPB mutants and rlm1Δ clearly rule out the possibility that Rlm1p is the important target of Mpk1p in the SPB duplication mutants. Recently, a genome-wide analysis has identified numerous genes that are transcriptionally upregulated or downregulated in response to MPK1 activity (Jung and Levin 1999). The transcriptional regulation of most of the genes identified in the genome analysis was mediated by Rlm1p, and all genes code for components of the cell-wall biosynthesis machinery (Jung and Levin 1999). Since loss of Mpk1p function results in a much more severe growth defect than the loss of Rlm1p, Mpk1p must have other targets as well. It is possible that other genes that are transcriptionally regulated by Mpk1p were missed in the genome analysis. However, the results of Jung and Levin (1999) raise the interesting possibility that Mpk1p activates its targets post-transcriptionally. Indeed, the transcript levels of CDC31 and protein levels of Cdc31p, Kar1p, and Spc110p are normal in the PKC1 pathway mutants, suggesting that the pathway may regulate SPB component(s) post-translationally.
Formally, the PKC1 pathway may function upstream, in parallel, or downstream of the CDC31 pathway. If the PKC1 pathway is downstream, then mutations in the SPB duplication genes may affect the level of PKC1 pathway activity. We tested this hypothesis using several different markers for PKC1 pathway activity and in all cases we found that mutations in SPB duplication genes did not alter the activity of the PKC1 pathway. Therefore, it is unlikely that the CDC31 pathway is upstream of the PKC1 pathway.
Less clear is whether the PKC1 functions upstream or in parallel with the CDC31 pathway. We found that Slg1p is required for Cdc31p's SPB function (Figure 6). This result suggests that the PKC1 pathway affects Cdc31p, either by activating Cdc31p directly or by activating other components with which Cdc31p interacts at the SPB. Given that Spc110p phosphorylation was defective in pkc1ts and mpk1Δ mutants and that Spc110p functionally interacts with Cdc31p, Spc110p may well be one of the targets of PKC1 pathway activity. However, given the timing of the appearance of p120, it is unlikely that this particular phosphorylation is the relevant regulatory event. The p112 form has not been resolved from the unphosphorylated form of the protein (Friedmanet al. 1996). Therefore, it is possible that the p120 appears as a consequence of earlier phosphorylation(s) of Spc110p. In such a scenario, the lack of p120 could be an indication that earlier phosphorylation events relevant to SPB duplication have not occurred in the pkc1ts and mpk1Δ mutants.
While our data suggest that the PKC1 pathway is involved in SPB duplication, pkc1 mutants have not been reported to exhibit defects in SPB duplication. The slg1Δ strain does have a phenotype that can be taken as including an SPB defect; in the cold, the mutant fails to initiate SPB duplication even though it buds and eventually lyses (Ivanovska and Rose 2000). The possibility of genetic redundancy and the pleiotropic nature of the mutant phenotype might easily mask an SPB defect in pkc1 mutants.
Spc110p may interact with Cdc31p at the SPB: In the course of testing the specificity of the genetic interactions, we found that SLG1 and PKC1 also suppressed the SPB duplication defect of spc110-220, a mutation in a gene not known to be related to genes in the Cdc31p pathway. We also found that overexpression of CDC31 suppressed spc110-220. The point mutation in the spc110-220 allele disrupts binding of Spc110p to Cmd1p, and overexpression of Cmd1p rescues the spc110-220 temperature sensitivity (Geiseret al. 1993; Sundberget al. 1996). High-copy CDC31 has also been found to partially suppress the temperature sensitivity of a cmd1 allele defective for SPB duplication (Geieret al. 1996). Two hypotheses are consistent with these findings. First, CDC31 might suppress spc110-220 by substituting for its close relative, calmodulin. In this view, Cdc31p and Spc110p would not normally interact in vivo. Consistent with this, Cdc31p and Spc110p have been localized to different regions of the SPB; the major portion of Cdc31p localizes to the half-bridge (Spanget al. 1993) and Spc110p localizes to the region between the central and inner plaques (Rout and Kilmartin 1990). Because Cdc31p and Spc110p would not share functions in vivo, the PKC1 pathway would suppress Cdc31p and Spc110p by different mechanisms.
An alternative view, which we favor, is that Spc110p and Cdc31p have interrelated functions that our genetic suppression analysis has uncovered. This interpretation is supported by the observation that, in vitro, an Spc110p peptide containing the calmodulin binding site binds Cdc31p as well as it binds calmodulin (Geieret al. 1996). In this more parsimonious model, the PKC1 pathway would suppress mutations in cdc31 and spc110 by the same mechanism because Spc110p and Cdc31p have shared functions in vivo. One place where Cdc31p and Spc110p might interact is at the junction between the half-bridge and the inner plaque of the SPB.
To explain our genetic data, we propose that the PKC1 pathway may signal to the Cdc31p network or to some other target at the SPB to activate SPB duplication. Because we observed that phosphorylation of Spc110p was defective in pkc1ts and mpk1Δ strains, we suggest that Spc110p may be a direct or indirect target of PKC1 pathway regulation. Does overexpression of PKC1 pathway components suppress SPB duplication mutants by restoring the p120 form of Spc110p? We think that this is unlikely because the p120 phosphorylated form is present in the SPB duplication mutants at the nonpermissive temperature, including spc110-220, kar1Δ17, and cdc31 mutants (Friedmanet al. 1996; W. Khalfan, I. Ivanovska and M. D. Rose, unpublished observation). Therefore, the loss of p120 is not likely to be the defect in these mutants. More likely, Spc110p phosphorylation in pkc1 and mpk1 mutants may simply reflect the state of the SPB, revealing that prior event(s) important for SPB assembly under the control of the Pkc1p pathway have not been executed properly.
Timing of regulation: The kar1-Δ17, cdc31, and dsk2Δ rad23Δ mutants are defective for an early step in SPB duplication, the formation of the satellite, which occurs independently of Cdc28p/Clns activation. Hence, if the PKC1 pathway regulates SPB duplication in a positive manner, it may act, at this early stage, in a manner independent of Cdc28p/Clns. Indeed there is substantial evidence to suggest that the PKC1 pathway can function independently of Cdc28p/Clns activity. First, mutations in MPK1/SLT2 and PKC1 have been isolated as enhancers of Start-defective cdc28 mutants (Mazzoniet al. 1993; Mariniet al. 1996) suggesting that the PKC1 pathway functions in parallel with Cdc28p. Second, MPK1 activation during vegetative growth is only partially dependent on CDC28; Mpk1p tyrosine phosphorylation is reduced but not eliminated in cdc28ts mutants, again suggesting that the PKC1 pathway can act partially independently of Cdc28p (Zarzovet al. 1996). Alternatively, the PKC1 pathway may be a transducer of the START/Cdc28p signal to promote the later steps in SPB duplication. Finally, the PKC1 pathway could activate SPB duplication by acting on Cdc28p/Clns. This view is supported by the interaction between MPK1 and the Swi4p/Swi6p complex (SBF), a major transcriptional activator of the G1 cyclins (Maddenet al. 1997). Regardless of the details, the PKC1 signal transduction pathway, which also promotes bud emergence, is ideally suited to coordinate two major events at the G1/S transition, SPB duplication and bud emergence.
We thank D. Levin, T. Davis, M. Winey, M. Stark, and D. Stirling for strains, antibodies, and useful discussions. We also acknowledge Naz Erdeniz for critical reading of the manuscript. This research was supported by National Institutes of Health grant GM52526 to M.D.R. I.I. and W.K. were supported by fellowships from the New Jersey Commision on Cancer Research.
Communicating editor: E. W. Jones
- Received February 25, 2000.
- Accepted April 12, 2000.
- Copyright © 2000 by the Genetics Society of America