Fus3p and Kss1p Control G1 Arrest in Saccharomyces cerevisiae Through a Balance of Distinct Arrest and Proliferative Functions That Operate in Parallel With Far1p

Corresponding author: Elaine A. Elion, Harvard Medical School, Department of Biological Chemistry and Molecular Pharmacology, 240 Longwood Ave., Boston, MA 02115., elion{at}bcmp.med.harvard.edu (E-mail)

Communicating editor: A. P. MITCHELL

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In Saccharomyces cerevisiae, mating pheromones activate two MAP kinases (MAPKs), Fus3p and Kss1p, to induce G1 arrest prior to mating. Fus3p is known to promote G1 arrest by activating Far1p, which inhibits three Clnp/Cdc28p kinases. To analyze the contribution of Fus3p and Kss1p to G1 arrest that is independent of Far1p, we constructed far1 CLN strains that undergo G1 arrest from increased activation of the mating MAP kinase pathway. We find that Fus3p and Kss1p both control G1 arrest through multiple functions that operate in parallel with Far1p. Fus3p and Kss1p together promote G1 arrest by repressing transcription of G1/S cyclin genes (CLN1, CLN2, CLB5) by a mechanism that blocks their activation by Cln3p/Cdc28p kinase. In addition, Fus3p and Kss1p counteract G1 arrest through overlapping and distinct functions. Fus3p and Kss1p together increase the expression of CLN3 and PCL2 genes that promote budding, and Kss1p inhibits the MAP kinase cascade. Strikingly, Fus3p promotes proliferation by a novel function that is not linked to reduced Ste12p activity or increased levels of Cln2p/Cdc28p kinase. Genetic analysis suggests that Fus3p promotes proliferation through activation of Mcm1p transcription factor that upregulates numerous genes in G1 phase. Thus, Fus3p and Kss1p control G1 arrest through a balance of arrest functions that inhibit the Cdc28p machinery and proliferative functions that bypass this inhibition.

IN the presence of mating pheromones, dividing haploid a and cells of Saccharomyces cerevisiae stimulate each other to arrest at Start in G1 phase, the commitment point of the cell cycle (SPRAGUE and THORNER 1993 ). Cell synchronization at Start is required for efficient morphogenesis and fusion between mating cells. However, in the absence of mating, an arrested cell will recover from G1 arrest and resume mitotic growth, even in the presence of mating pheromone. Thus, the decision to arrest in G1 phase could be controlled by a balance between opposing forces that promote arrest and promote proliferation.

The passage through Start is determined by the activity of Cdc28p kinase, which is regulated by three G1 cyclins, Cln1p, Cln2p, and Cln3p, and two G1/S-phase cyclins, Clb5p and Clb6p (KOCH and NASMYTH 1994 ; CROSS 1995 ; NASMYTH 1996 ). The G1/S-phase cyclins provide both overlapping and unique functions for Start. Cln3p is expressed throughout the cell cycle and appears to play a more important role as a transcriptional regulator of the other G1/S-phase cyclin genes. Cln1p, Cln2p, Clb5p, and Clb6p are expressed periodically in G1 phase, with Cln1p and Cln2p having more important roles in bud emergence, and Clb5p and Clb6p having more important roles in the initiation of DNA replication and spindle pole body duplication. Additional protein kinases may also regulate the G1 to S phase transition in parallel with Cdc28p kinase. Pho85p kinase, in combination with either Pcl1p or Pcl2p cyclin partners, provides functions that overlap with Cln1p/Cdc28p and Cln2p/Cdc28p kinases (OGAS et al. 1991 ; ESPINOZA et al. 1994 ; MEASDAY et al. 1994 ). Functional overlap also exists for Cla4p kinase (CVRCKOVA et al. 1995 ) and the Pkc1p pathway (IGUAL et al. 1996 ; GRAY et al. 1997 ).

Many lines of evidence argue that G1 arrest during mating occurs primarily through inhibition of the G1 cyclins. The arrest points of both a cdc28-4 mutant and a cln1 cln2 cln3 triple mutant are similar to the mating pheromone arrest point (REED 1992 ), and dominant hyperactive alleles of G1 cyclins shorten G1 phase and prevent G1 arrest (CROSS 1988 ; NASH et al. 1988 ). In addition, mutations in individual G1 cyclins restore significant G1 arrest to fus3 and far1 mutants that do not properly arrest in the presence of -factor (CHANG and HERSKOWITZ 1990 ; ELION et al. 1990 , ELION et al. 1991A ). Two levels of G1 cyclin inhibition have been proposed: repression of transcription of the CLN1 and CLN2 genes (WITTENBERG et al. 1990 ; ELION et al. 1991A ; VALDIVIESO et al. 1993 ) and direct inhibition of the three different Clnp/Cdc28p kinases by Far1p (PETER et al. 1993 ; TYERS and FUTCHER 1993 ; PETER and HERSKOWITZ 1994 ). G1 arrest may also involve the regulation of targets other than the G1 cyclins. First, while deletion of SIC1 restores cell division to a cln1 cln2 cln3 triple mutant, presumably by activation of Clb5p/Cdc28p and Clb6p/Cdc28p kinases, -factor still inhibits the growth of this strain (SCHNEIDER et al. 1996 ; TYERS 1996 ). Second, a far3 mutant is resistant to -factor, although the G1 cyclins appear to be properly regulated (HORECKA and SPRAGUE 1996 ).

G1 arrest in response to mating pheromone is controlled by the mating MAP kinase cascade (HERSKOWITZ 1995 ). After binding of mating pheromone, the receptor activates a heterotrimeric G-protein which, together with the Ste5p scaffolding protein (ELION 1995 ), causes sequential activation of a p21-activated kinase homolog (Ste20p), a MAPKKK (Ste11p), a MAPKK (Ste7p), and two MAP kinases (Fus3p and Kss1p). Activation of Fus3p and Kss1p by -factor is coupled to G1 arrest as well as to other responses required for mating and recovery (e.g., activation of the Ste12p transcription factor, shmoo formation, fusion, and signal attenuation (ELION et al. 1990 , ELION et al. 1991A , ELION et al. 1993 ; GARTNER et al. 1992 ; MA et al. 1995 ; FARLEY et al. 1999 ).

The relative contribution of Fus3p and Kss1p to the control of G1 arrest is not known. Much of the available data support a model in which Fus3p is the major MAP kinase regulator of G1 arrest, with most of the regulation through the control of Far1p. Fus3p phosphorylates Far1p (ELION et al. 1993 ; PETER et al. 1993 ), and this phosphorylation is required to stabilize Far1p (HENCHOZ et al. 1997 ) and allow its association with the three different Clnp/Cdc28p complexes (PETER et al. 1993 ; TYERS and FUTCHER 1993 ; JEOUNG et al. 1998 ). Far1p subsequently inhibits the Clnp/Cdc28p kinases through an as-yet-undefined mechanism (GARTNER et al. 1998 ). To date, there is no clear evidence that Kss1p plays a direct role in the regulation of Far1p or G1 arrest. While null mutations in FUS3 cause a G1 arrest defect (ELION et al. 1991A ) and block pheromone-induced phosphorylation of Far1p and formation of Far1p/Clnp/Cdc28p complexes (PETER et al. 1993 ; TYERS and FUTCHER 1993 ), a null mutation in KSS1 has no obvious effect on G1 arrest (ELION et al. 1991A ). Kss1p could play an indirect role in regulating G1 arrest, because it is able to activate Ste12p (ELION et al. 1991A ), which positively regulates the FAR1 gene (CHANG and HERSKOWITZ 1990 ). This possibility is consistent with the greater -factor sensitivity of a fus3 null mutant compared to a far1 null mutant (SATTERBERG 1993 ; TYERS and FUTCHER 1993 ) and the ability of overexpressed Ste5p to restore -factor sensitivity to fus3 and far1 null mutants, but not to fus3 kss1 double mutants (ELION et al. 1991B ; LEBERER et al. 1993 ; SATTERBERG 1993 ). However, catalytically inactive Fus3p nearly completely blocks Kss1p from activating the Ste12p-dependent gene STE2 in the presence of -factor, arguing that Kss1p normally does not function in the mating pathway in the presence of Fus3p (MADHANI et al. 1997 ).

Here we present evidence that Fus3p and Kss1p both control G1 arrest in parallel with Far1p through a combination of functions that both inhibit and promote cell division. Fus3p and Kss1p together promote G1 arrest by repressing transcription of G1/S-phase cyclin genes (e.g., CLN1, CLN2, CLB5) at a step distinct from Cln3p/Cdc28p-mediated activation of Swi4p/Swi6p. This inhibition constitutes a major portion of the cyclin regulation that occurs in the presence of -factor. Surprisingly, Fus3p and Kss1p also counteract G1 arrest through distinct mechanisms. Kss1p promotes recovery from G1 arrest by inhibiting the MAP kinase cascade at or below Ste11p. By contrast, Fus3p promotes proliferation through a novel function that is not shared by Kss1p and does not involve increasing the level of the G1 cyclins or decreasing Ste12p activity. Genetic suppression tests suggest that this function involves the activation of Mcm1p or genes under its control.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Media, strains, and yeast strain construction:
Yeast strains are described in Table 1. All strains are isogenic derivatives of EY957, which is a bar1 derivative of W303a. Yeast media were prepared as described (SHERMAN et al. 1986 ). Plasmids were integrated as described previously for fus3-6::LEU2, fus3-7::HIS3 (pYEE98, pJB225; ELION et al. 1990 , ELION et al. 1991A ), CLN2-HA LEU2 (MT104; TYERS et al. 1993 ), and CLN3-1 URA3 (pFC101-1; CROSS 1988 ). All integrations were confirmed by Southern analysis (SAMBROOK et al. 1989 ).

Halo assays:
-factor sensitivity was measured by a halo assay as described previously (ELION et al. 1990 ), using 50 µl of an overnight culture of yeast cells. -Factor peptide (synthesized by C. Dahl, Harvard Medical School) was dissolved in 90% methanol and stored at -20°. All halo assays were done at least in duplicate, using 3 µl of 50 µM synthetic -factor for bar1 strains and 8 µl of 2 mM -factor for BAR1 strains.

Growth conditions:
Strains were grown at 30° in selective SC media with 2% dextrose to an A₆₀₀ of 0.5, split in half and incubated in the presence or absence of 100 nM -factor for 2 hr (unless indicated otherwise), and then harvested. For the BAR1 strains described in Figure 2, Figure 5 mM -factor was added for 30 min. The STE11-4 far1, STE11-4 far1 fus3, and STE11-4 far1 kss1 strains each contained a FUS1-HIS3 reporter gene and were grown in media lacking histidine to avoid the propagation of sterile pseudorevertants. The growth rate of these three strains is slower than that of STE11 strains, with doubling times of 3.2, 3.5, and 2.5 hr for the STE11-4 far1, STE11-4 far1 fus3, and STE11-4 far1 kss1, respectively, compared to 1.5 hr for wild type and far1 strains. Therefore, a 4-hr -factor induction point was also done in addition to a 2-hr time point for these slower-growing strains.

View larger version (50K): In this window In a new window Download PPT slide   Figure 1. Suppression of the -factor-resistant phenotype of a far1 null. (A) Growth of far1 strains streaked onto YPD plates with the indicated amount of -factor. Strains: FAR1 CLN2 - EY1118; far1 CLN2 - EY1262; far1 cln2 - EY1290. (B) Hyperactivation of the MAPK cascade by multiple copies of the STE5 gene or the STE11-4 mutation restores -factor sensitivity to a far1 null. Pheromone sensitivity was tested in halo assays using 3 µl of 50 µM -factor (MATERIALS AND METHODS). Strains: WT - EY1118; far1 - EY1262; far1 cln2 - EY1290; far1 + STE5 2µ - EY1262 + pJB223 (STE5 URA3 2µ); STE11-4 far1 - EY1298; STE11-4 far1 fus3 kss1 - EY1346. Plates were photographed after 24 hr at 30°. (C) Catalytic activity of both Fus3p and Kss1p is required for pheromone-dependent growth arrest. Halo assays were done as in B except that 8 µl of 2 mM -factor was used for the strains that are BAR1 (see Table 1). Strains: WT - EY1118 + Ycp50 (URA3 CEN); fus3 - EY940 + Ycp50 (URA3 CEN); fus3 + fus3-K42R - EY940 + pRT5 (fus3K42R URA3 CEN); WT - EY699 + B1817 (HIS3 CEN); kss1 - EY725 + B1817; kss1 + kss1K42R - EY725 + B3697 (kss1K42R HIS3 CEN).

View larger version (25K): In this window In a new window Download PPT slide   Figure 2. Level of expression of a FUS1-lacZ reporter gene in various strains. ß-Galactosidase activity was quantitated as described in ELION et al. 1995 . The bar1 strains in A harbor FUS1-lacZ on pYBS45 and were induced for 2 hr with either 0, 1, 10, 100, or 500 nM -factor (shown from left to right). Similar results were found after a 30-min induction. The BAR1 strains in B harbor FUS1-lacZ on pJB207 and were induced with 5 µM -factor for 30 min. The units shown are the average of at least two independent experiments ± SE. Strain numbers are listed in Figure 1 except for EY1321 (STE11-4 far1 fus3) and EY1335 (STE11-4 far1 kss1).

View larger version (93K): In this window In a new window Download PPT slide   Figure 3. Cln2p/Cdc28p kinase levels in far1 and fus3 strains. (A) STE11-4 restores -factor-dependent inhibition of Cln2-HAp/Cdc28p kinase in the absence of Far1p. Lanes 1, 2—WT (CY326); lanes 3, 4—far1 (CY327); lanes 5, 6—STE11-4 far1 (CY329); lanes 7, 8—STE11-4 far1 fus3 (CY378); lanes 9, 10—STE11-4 far1 kss1 (CY330); lane 11—no Tag. Exponentially growing strains harboring the CLN2-HA gene were treated (+) or not treated (-) with -factor for 2 hr, and then extracts were prepared and assayed for Cln2-HA kinase activity and Cln2-HA protein. Fold inhibition is the average of two to three experiments. (B) Quantitation of specific activity of Cln2-HAp/Cdc28p kinase in far1 and fus3 strains. Strains were treated for 1 hr with -factor, except for the wild-type strain, which was treated for 15 min. Based on prior normalization, equal amounts of Cln2-HA protein were immunoprecipitated from the different extracts (10–40 mg total protein). After immunoprecipitation and washes, the protein A beads were resuspended in 2 ml of lysis buffer. Of this, 200 µl was used for the Cln2-HA kinase assay, and the remainder was used for the Cln2-HA immunoblot. Strains are in Table 1. Lanes 1, 2—WT; lanes 3, 4—far1; lanes 5, 6—fus3; lanes 7, 8—STE11-4 far1; lane 9—no Tag. (C) Cln2-HAp/Cdc28p kinase activity and Cln2-HA protein in far1 and fus3 strains. Lanes 1, 2—far1 (CY327); lanes 3, 4—fus3 (CY328); lanes 5, 6—far1 fus3 (CY358); lanes 7, 8—WT (CY326); lane 9—no Tag. (D) STE5OP restores of -factor-dependent inhibition of Cln2-HAp/Cdc28p kinase in the absence of Far1p. Fold inhibition is on the basis of one experiment. Lanes 1, 2—WT + 2µ (CY326 + Yep24); lanes 3, 4—WT + STE5-2µ (CY326 + pJB223); lanes 5, 6—far1 + 2µ (CY327 + Yep24); lanes 7, 8—far1 + STE5-2µ (CY327 + pJB223); lane 9—no Tag. All strains were induced with 100 nM of -factor for 2 hours except for as described in B.

View larger version (57K): In this window In a new window Download PPT slide   Figure 4. Fus3p and Kss1p regulate the level of G1/S-phase cyclin mRNAs. (A) Fus3p and Kss1p repress CLN1, CLN2, CLB5 and activate CLN3, PCL2. Lanes 1, 2—WT, lanes 3, 4—far1; lanes 5, 6—STE11-4 far1, lanes 7, 8—STE11-4 far1 fus3; lanes 9, 10—STE11-4 far1 kss1; lanes 11, 12—STE11-4 far1 fus3 kss1. (B) CLN3-1 does not bypass Far1p-independent repression of CLN1, CLN2. Lanes 1, 2—CLN3 (EY-1118); lanes 3, 4—CLN3-1 (CY385); lanes 5, 6—STE11-4 far1 (EY1298); lanes 7, 8—STE11-4 far1 CLN3-1 (CY384). In a parallel experiment, the percentage of unbudded cells in the STE11-4 far1 and STE11-4 far1 CLN3-1 strains after a 2-hr exposure to 100 nM of -factor were found to be: STE11-4 far1, 40% - -factor, 69% + -factor. STE11-4 far1 CLN3-1, 35%; -factor, 46% + -factor. For A and B, logarithmically growing strains were induced with 100 nM of -factor for 2 hr and Northern analysis was performed as described in MATERIALS AND METHODS. + indicates -factor induction. Note that the histone probe detects both HTA1 and HTB1 mRNAs. Strain numbers are listed in the legends to Figure 1 and Figure 2.

View larger version (85K): In this window In a new window Download PPT slide   Figure 5. Dominant CLN2 and CLB5 cyclin genes suppress Far1p-independent arrest. Halo assays used 3 µl of 50 µM -factor. Strains: CLN3 - EY1118; CLN3-1 - CY385; STE11-4 far1 CLN3 - EY1298; STE11-4 far1 CLN3-1 - CY384; WT + CEN and STE11-4 far1 + CEN are EY1118 and EY1298 bearing Ycp50; WT + pGAL-CLN2 and STE11-4 far1 + pGAL-CLN2 are EY1118 and EY1298 bearing pGAL-CLN2 URA3 CEN. YIpGAL1-CLB5 strains contain a functional CLB5-HA gene (ES2669; Table 1) integrated at the URA3 locus. Plates were photographed after 24 hr at 30°.

Preparation of yeast extracts:
Cells were collected at 4°, washed once with cold sterile water, and frozen in dry ice/ethanol. Whole cell extracts were prepared by lysis with glass beads, as described in SURANA et al. 1991 . Protein concentrations were determined using the Bio-Rad (Hercules, CA) protein assay.

ß-Galactosidase assays:
ß-Galactosidase activity was measured as described (ELION et al. 1995 ) using yeast extracts prepared as described above.

Immunoprecipitation, immunoblot analysis, and kinase assays:
Immunoprecipitations were performed as described (ELION et al. 1993 ) using 12CA5 mouse monoclonal antibody to detect the HA1 epitope (FIELD et al. 1988 ). Depending on the abundance of the protein in question, 2 mg (for Cln2-HAp) to 40 mg (for detection of Cln2-HAp in a wild-type strain after -factor treatment) of protein was used for immunoprecipitation. For quantitation of immunoprecipitated kinase activity, the protein A beads were washed twice more with kinase reaction buffer (20 mM Tris-HCl, pH 7.5, 7.5 mM MgCl₂, 0.1 mM EGTA (MENDENHALL 1993 and reactions were carried out in 20 µl containing 1.4 µl [-³²P]ATP (ICN, Costa Mesa, CA; 5000 Ci/mmol),1 µl 10 mM ATP, and 1 µg H1 histone for 15 min at 25° (SURANA et al. 1991 ). Reactions were terminated by addition of 2x PAGE sample buffer, boiled for 5 min and loaded onto a 12% SDS-PAGE gel. Phosphorylated H1 was visualized by autoradiography. Quantitations were done with a Molecular Dynamics (Sunnyvale, CA) Phosphorimager. For immunoblot analysis of immunoprecipitated proteins, protein A beads were washed, pelleted, and boiled in sampler buffer immediately before SDS-PAGE (8%, 10%, or 12% depending on the protein) as described in KRANZ et al. 1994 . Blots were developed with an Amersham (Buckinghamshire, UK) ECL kit according to manufacturer's instructions using Fuji RX X-ray film.

Northern analysis:
Total RNA was isolated by extraction with hot and acidic phenol as described (COLLART and OLIVIERO 1992 ), transferred to nitrocellulose, and probed according to SAMBROOK et al. 1989 . The following probes were used: 1.3-kb EcoRI-NcoI CLN1 from 419, 0.7-kb XhoI-HindIII CLN2 from 810 (HADWIGER et al. 1989 ), 1.4-kb EcoRI-XhoI CLN3 from pFC101-1 (CROSS 1988 ), 1.5-kb SalI-BamHI from Ycp-GAL1-CLB5, 1.3-kb BglII-NsiI PCL2 from pBA623, 5.5-kb NheI-SalI H2A1 + H2B1 from pCC68, and 2.0-kb XhoI-HindIII ACT1 from pYEE15 (ELION et al. 1991A ). Blots were probed sequentially. Northern blots were reprobed with ACT1 as a control for loading and retention of RNA on the membrane.

Cell morphology and flow cytometry:
Cells were fixed and stained with 4',6-diamidino-2-phenylindole (DAPI) using a protocol provided by S. Dutcher (TRUEHEART et al. 1987 ) and then counted after brief sonication to determine the percentage of unbudded cells. For flow cytometric DNA quantitation (FACS) analysis, cells were fixed with 70% ethanol and processed essentially as described (HUTTER and EIPEL 1979 ), and then briefly sonicated immediately before DNA quantitation. FACS was done using the Flow Cytometry Facility at the Dana-Farber Cancer Institute.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Hyperactivation of Fus3p and Kss1p by Ste11-4p or excess Ste5p restores G1 arrest to a far1 mutant:
The phenotype of a far1 cln2 double mutant first suggested that mating pheromone promotes G1 arrest through additional pathways that operate in parallel with Far1p (CHANG and HERSKOWITZ 1990 ). A far1 cln2 double mutant is completely resistant to -factor at a concentration that causes a wild-type strain to arrest growth (Figure 1A, left). Unless indicated otherwise, all strains are deleted for the -factor protease gene, BAR1, to avoid complications from recovery as a result of -factor degradation (CIEJEK and THORNER 1979 ). This -factor resistance is presumably due to the high levels of Cln3p/Cdc28p and Cln1p/Cdc28p kinase that result from the loss of Far1p inhibition (PETER et al. 1993 ; TYERS and FUTCHER 1993 ). However, when the concentration of -factor is raised to higher levels, a far1 cln2 double mutant arrests in G1 phase (Figure 1A, panels 2 and 3, and Figure 1B; CHANG and HERSKOWITZ 1990 ). The fact that an increase in -factor concentration can restore G1 arrest to a far1 cln2 strain raises the possibility that the MAP kinase cascade activates parallel pathways to regulate G1 arrest.

We directly tested the possibility that Fus3p and Kss1p promote G1 arrest independently of Far1p, by determining whether an increase in the level of signaling through the MAPK cascade could restore G1 arrest to a far1 null mutant. Pathway activity was increased in two independent ways, either by increasing the concentration of Ste5p, a limiting component required for MAP kinase activation (KRANZ et al. 1994 ), or by a gain-of-function allele of the MAPKKK Ste11p (STE11-4) that causes constitutive hyperactivation of the MAPKK Ste7p that activates Fus3p and Kss1p (STEVENSON et al. 1992 ). Overproduction of Ste5p (STE5^OP) or introduction of STE11-4 in a far1 strain increases -factor-induced FUS1 expression across a wide range of -factor concentrations, although STE11-4 is significantly more potent than STE5^OP in the absence of -factor (Figure 2). Halos assays show that STE5^OP and STE11-4 restore nearly as much -factor sensitivity to a far1 null strain as does a deletion in the CLN2 gene (Figure 1B). STE11-4 restores more -factor sensitivity to the far1 null than does STE5^OP, consistent with the greater pathway activity in this strain (Figure 2).

The restoration of pheromone sensitivity to a far1 strain by STE11-4 or STE5^OP requires the components of the mating signal transduction cascade including Fus3p and Kss1p. Deletion of positive regulators of the pathway (i.e., STE4, STE5, STE7, STE12) or overexpression of negative regulators (i.e., GPA1, SST2) blocks the pheromone sensitivity of the far1 STE11-4 strain. Overexpression of positive regulators that either enhance Ste11p activity (i.e., STE5, STE50) or the amount of active Ste7p (i.e., STE7) further enhances the -factor sensitivity of the STE11-4 far1 strain (data not shown). Thus, it is possible to arrest growth in far1 cells by simply increasing the level of activity of the mating MAP kinase cascade during -factor induction.

Null mutations in both FUS3 and KSS1 completely block the -factor sensitivity of the STE11-4 far1 strain, demonstrating that the arrest is completely dependent upon the two mating MAP kinases (Figure 1B). Substitution of catalytically inactive fus3K42R for FUS3 in STE11 FAR1 and STE11-4 far1 strains completely blocks -factor-induced arrest, indicating that Fus3p kinase activity is essential for the arrest (Figure 1C and data not shown). Substitution of catalytically inactive kss1K42R for KSS1 in a STE11 FAR1 strain causes partial resistance to -factor and reduced levels of FUS1 expression (Figure 1C and Figure 2), indicating that Kss1p kinase activity is also required for efficient arrest. STE11-4 is unlikely to promote G1 arrest through inappropriate activation of either Mpk1 or Hog1, the other two MAPKs expressed in haploid cells. Mutation of Mpk1, the MAPK in the Pkc1 pathway, does not reduce the ability of STE11-4 to restore arrest to a far1 strain (data not shown). Hog1 is an attenuator of the pathway and inhibits Fus3p tyrosine phosphorylation (HALL et al. 1996 ). Thus, Fus3p and Kss1p are specifically required to mediate this arrest.

The increased -factor sensitivity in the presence of STE5^OP or STE11-4 could arise from effects on cell division at any point in the cell cycle. Overexpression of a stable form of Far1p outside of G1 phase leads to G2 arrest, possibly from inappropriate inhibition of Clbp/Cdc28p kinases (MCKINNEY and CROSS 1995 ), and STE11-4 can cause a G2/M delay during pseudohyphal growth (KRON et al. 1994 ). To determine whether the effects of STE5^OP and STE11-4 were G1 phase-specific, we quantitated their effects on budding and DNA synthesis after a short-term exposure to -factor. For STE11-4 strains, a 4-hr -factor induction time point was done in addition to a 2-hr time point, because STE11-4 causes cells to divide more slowly (3.2-hr doubling time for STE11-4 far1 compared to 1.5 hr for wild type and far1 strains; see MATERIALS AND METHODS for details). STE5^OP and STE11-4 both cause a far1 strain to arrest in G1 phase in the presence of -factor, as shown by a greater accumulation of unbudded cells (Table 2). STE11-4 somewhat increases the amount of inhibition of DNA synthesis [31% wild-type inhibition for far1 vs. 42% wild-type inhibition for STE11-4 far1; Table 2, shown as "1C DNA (%)"], while STE5^OP has no obvious effect, indicating that most of the arrest is due to a block in budding for both strains. As predicted from the halo assay, a fus3 kss1 double null mutation completely blocks -factor-induced inhibition of budding and DNA synthesis by STE11-4 (i.e., STE11-4 far1 fus3 kss1 behaves like ste2 or fus3 kss1 strains; Table 2) and by STE5^OP (data not shown), demonstrating that Fus3p and Kss1p inhibit budding and DNA synthesis in the absence of Far1p.

View this table: In this window In a new window   Table 2. Effect of -factor treatment on budding arrest and inhibition of DNA synthesis

Elevated signaling is not required for the MAP kinases to inhibit DNA synthesis in a far1 null:
To further confirm that Fus3p and Kss1p promote G1 arrest through mechanisms distinct from Far1p, we compared the arrest behavior of a far1 single mutant to fus3 far1, kss1 far1, and fus3 kss1 double mutants that do not have enhanced levels of signaling. As previously observed (TYERS and FUTCHER 1993 ), a far1 mutant undergoes partial inhibition of DNA synthesis in the presence of -factor (31% wild-type inhibition; Table 2), although the strain is resistant to -factor, as measured by a halo assay (Figure 1B) and the accumulation of unbudded cells (Table 2). This partial inhibition is not detected in a ste2 mutant and therefore requires signal transduction through the -factor receptor (Table 2). Moreover, deletion of either FUS3 or KSS1 reduces the amount of inhibition of DNA synthesis that occurs in a far1 strain (12% wild type for fus3 far1 and 8% wild type for kss1 far1), and DNA synthesis is not inhibited at all in a fus3 kss1 double mutant (Table 2). Analysis of the levels of total Cdc28p kinase in these strains shows that these effects on DNA synthesis are mirrored at the level of Cdc28p kinase activity (data not shown), substantiating the results with the STE11-4 far1 strain.

Fus3p and Kss1p inhibit Cln2p/Cdc28p kinase independently of Far1p:
We next determined whether the level of Cln2p/Cdc28p kinase was reduced in the STE11-4 far1 strain as an explanation for the increased pheromone sensitivity and budding arrest. An epitope-tagged CLN2 gene under the control of its own promoter (TYERS et al. 1993 ) was integrated in single copy into the strains to be tested. In a wild-type strain, the steady state levels of Cln2-HAp protein and active Cln2-HAp/Cdc28p kinase are greatly reduced by -factor, causing an overall 26-fold reduction in the level of Cln2-HAp/Cdc28p kinase (Figure 3A, lanes 1 and 2). As previously reported (VALDIVIESO et al. 1993 ; PETER and HERSKOWITZ 1994 ), a far1 mutant has high levels of Cln2 protein and active Cln2p/Cdc28p kinase (Figure 3A, lanes 3 and 4). STE11-4 restores a dramatic 20-fold inhibition of Cln2p/Cdc28p kinase to the far1 strain (Figure 3A, lanes 5 and 6). Similar low levels of Cln2p/Cdc28p kinase are also found in the STE5^OP far1 strain (Figure 3D). Thus, -factor can inhibit Cln2p/Cdc28p kinase independently of Far1p.

The Far1p-independent inhibition of Cln2p kinase is blocked by a null mutation in FUS3 (Figure 3A, lanes 7 and 8; STE11-4 far1 fus3), demonstrating a clear role for Fus3p in negatively regulating Cln2p/Cdc28p kinase that is distinct from Far1p. Deletion of Kss1p also blocks, to a lesser extent, the inhibition of Cln2p/Cdc28p kinase (Figure 3A, lanes 9 and 10; STE11-4 far1 kss1). The level of Cln2p/Cdc28p kinase activity in a STE11-4 far1 kss1 triple mutant is reproducibly two-fold greater than that of a STE11-4 far1 double mutant, despite equal levels of Cln2p protein (Figure 3A; compare lanes 5 and 6 with lanes 9 and 10), suggesting that Kss1p modestly inhibits Cln2p/Cdc28p kinase. Thus, both MAP kinases regulate Cln2p/Cdc28p kinase independently of Far1p, possibly at several levels. However, Fus3p plays a much greater role.

Fus3p and Kss1p do not inhibit the specific activity of Cln2p/Cdc28p:
We examined the specific activity of Cln2p/Cdc28p kinase in far1 and STE11-4 far1 strains to determine whether the MAP kinases inhibit the activity of Cln2p/Cdc28p kinase independently of Far1p (Figure 3B). Large-scale preparations of whole cell extracts were made from wild-type, far1, and STE11-4 far1 strains grown in the absence or presence of -factor to be able to immunoprecipitate equal amounts of Cln2p under both conditions. A 15-min -factor induction was done for the wild-type strain, because of the rapid loss of Cln2p, while 1-hr inductions were done for the other strains. Samples were then preequalized so that equal amounts of Cln2p would be immunoprecipitated from each of the extracts (10–40 mg protein; MATERIALS AND METHODS). As shown in Figure 3B, a ~3-fold reduction in Cln2p/Cdc28p kinase-specific activity in a wild-type strain is detected after 15 min in -factor, presumably because of inhibition by Far1p. This level of inhibition may be an underestimate per responding cell, because only a small percentage of cells are at the Start arrest point. Cln2p/Cdc28p kinase has equally high specific activity in the far1 and STE11-4 far1 strains after a 1-hr exposure to -factor (~0.8-fold inhibition for both strains). Thus, the enhanced sensitivity and G1 arrest of the STE11-4 far1 strain is unlikely to be due to an effect on the specific activity of Cln2p/Cdc28p kinase.

Fus3p and Kss1p repress transcription of CLN1, CLN2, and CLB5:
The reduction in Cln2p/Cdc28p kinase by hyperactivation of Fus3p and Kss1p by STE11-4 could be the result of enhanced post-transcriptional inhibition of Cln2p. We therefore determined whether the -factor-dependent reduction in Cln2p protein detected in the STE11-4 far1 strain involves more rapid turnover of Cln2 mRNA or protein. On the basis of shut-off experiments using a GAL1-CLN2-HA gene, we find no evidence for enhanced post-transcriptional inhibition of Cln2p in the STE11-4 far1 strain either in the absence or presence of -factor (data not shown).

We next determined whether transcriptional repression of the G1 cyclin genes is the primary cause of the decreased levels of Cln2p. As previously shown (NASH et al. 1988 ; WITTENBERG et al. 1990 ), transcription of the CLN1 and CLN2 genes decreases in the presence of -factor in a wild-type strain (Figure 4A, lanes 1 and 2), while the expression of CLN3 is slightly increased (Figure 4A). In a far1 null strain, the addition of -factor for 2 hr does not reduce expression of either CLN1 or CLN2, nor does it increase the expression of CLN3 (Figure 4A, lanes 3 and 4). Strikingly, STE11-4 restores nearly wild-type inhibition of transcription of the CLN1 and CLN2 genes to the far1 null (Figure 4A, lanes 5 and 6), largely accounting for the 26-fold reduction in Cln2p/Cdc28p kinase (Figure 3A). This inhibition contrasts with transcriptional activation of two other cyclin genes that promote budding and are implicated in recovery, CLN3 and PCL2 (Figure 4A; NASH et al. 1988 ; MEASDAY et al. 1994 ).

The repression of the CLN1 and CLN2 genes is mediated by the combined action of Fus3p and Kss1p. Null mutations in either FUS3 or KSS1 partially block the inhibition of transcription in the STE11-4 far1 strain to similar extents (Figure 4A: lanes 7 and 8, lanes 9 and 10; STE11-4 far1 fus3 and STE11-4 far1 kss1), while null mutations in both FUS3 and KSS1 fully block the inhibition (Figure 4A, lanes 11 and 12; STE11-4 far1 fus3 kss1). Fus3p and Kss1p also equivalently regulate transcriptional activation of the CLN3 and PCL2 genes. This pattern of control contrasts the opposing effects of Fus3p and Kss1p on the transcription of the FUS1 gene (ELION et al. 1991A ; Figure 2).

Fus3p and Kss1p also repress transcription of the CLB5 gene (Figure 4A), with tighter repression than that observed for CLN2. In contrast to the pattern of control of CLN1 and CLN2, only a double deletion of FUS3 and KSS1 blocks repression of the CLB5 gene. The transcriptional repression of the CLB5 gene is unlikely to be an indirect consequence of inhibition of the G1 cyclins, because repression occurs efficiently in the STE11-4 far1 fus3 strain that has high levels of Cln2p/Cdc28p kinase (Figure 3A). Additional experiments suggest that -factor does not significantly alter the levels of Clb5p or Clb5p/Cdc28p kinase activity in wild-type or far1 cells (based on shut-off experiments using CLB5-HA expressed from the GAL1 promoter; data not shown).

Overexpression of either CLN2 or CLB5 suppresses Far1p-independent G1 arrest:
One could argue that the arrest we detect in the STE11-4 far1 strain is not tied to the observed transcriptional repression of the G1/S cyclin genes. To test the hypothesis that transcriptional repression of the G1/S cyclin genes is causal to the arrest we observe, we determined whether dominant G1/S cyclin genes that circumvent the transcriptional repression imposed by Fus3p and Kss1p are able to bypass Far1p-independent arrest. As shown in Figure 5, overexpression of either CLN2 or CLB5 using the strongly inducible GAL1 promoter confers -factor resistance to the STE11-4 far1 strain in addition to the wild-type strain (GAL-CLN2, YIpGAL-CLB5; Figure 5). Two additional observations support the view that Far1p-independent arrest involves a G1 arrest block that is a consequence of transcriptional repression of the G1/S genes. First, the pattern of expression of histone H2A and H2B mRNAs (encoded by HTA1/HTB1) mirrors that of CLN1 and CLN2 (Figure 4A), consistent with a block at Start in G1 phase. Second, we find that the STE11-4 far1 strain does not undergo an enhanced loss of viability compared to a far1 strain after long-term (18-hr) exposure to a high concentration of -factor (100 nM; data not shown), indicating that the cells are arrested by -factor rather than dying. Collectively, these findings strongly argue that transcriptional repression of the G1/S cyclin genes is a primary cause of Far1-independent arrest.

Fus3p and Kss1p block Cln3p/Cdc28p from activating the CLN1 and CLN2 promoters:
Periodic transcription of the CLN1, CLN2, and CLB5 genes in G1 phase is controlled by Swi4p/Swi6p and Mbp1p/Swi6p transcription factor complexes (KOCH and NASMYTH 1994 ; BREEDEN 1996 ), which are positively regulated by Cln3p/Cdc28p kinase (CROSS 1995 ; NASMYTH 1996 ). We tested whether Fus3p and Kss1p mediate transcriptional repression of the G1 cyclin genes through inhibition of Cln3p/Cdc28p, by determining whether a dominant CLN3-1 allele could bypass Far1p-independent inhibition of CLN1, CLN2 transcription. The dominant CLN3-1 mutation stabilizes Cln3p and allows Cln3p/Cdc28p kinase to hyperactivate transcription of the CLN1 and CLN2 genes in the presence of -factor. This activation requires Swi4p and Swi6p (CROSS and TINKELENBERG 1991 ; DIRICK and NASMYTH 1991 ). Cln3-1p efficiently bypasses transcriptional repression of the CLN1 and CLN2 genes, as shown by the high levels of CLN1 and CLN2 mRNAs in the CLN3-1 strain in the presence of -factor (Figure 4B, lanes 3 and 4) and the resulting -factor resistance of this strain in a halo assay (Figure 5). Strikingly, however, Cln3-1p does not activate CLN1 and CLN2 to an obvious degree in the STE11-4 far1CLN3-1 strain (Figure 4B, lanes 7 and 8). The absence of transcriptional activation of CLN1 and CLN2 correlates with greatly reduced -factor resistance for this strain compared to the CLN3-1 strain (Figure 5). Thus, Cln3-1p/Cdc28p is unable to activate the CLN1, CLN2 promoters in the STE11-4 far1 strain.

One interpretation of this finding is that STE11-4 inhibits CLN3-1, either transcriptionally or post-transcriptionally. Northern analysis demonstrates that the CLN3 gene is properly upregulated by -factor in STE11-4 strains (Figure 4A). Immunoblot analysis shows that the steady state levels of epitope-tagged Cln3p are the same in wild-type as in STE11-4 far1 strains (data not shown). Two additional observations suggest that the Cln3-1 protein is still functional in the STE11-4 far1 strain. First, CLN3-1 does confer some -factor resistance to the STE11-4 far1 strain (as shown by the slightly more turbid and smaller diameter halo, Figure 5), consistent with the fact that CLN3-1 can weakly substitute for CLN1 and CLN2 for passage through Start (CROSS 1995 ). Second, CLN3-1 still promotes budding in the STE11-4 far1 strain, under the same -factor conditions that prevent CLN3-1 from activating the CLN1/CLN2 genes (Figure 4B legend). Collectively, these data argue that Cln3-1p/Cdc28p complexes are selectively blocked for transcriptional activation at the G1/S cyclin promoters.

Fus3p and Kss1p may also promote proliferation in addition to G1 arrest:
The results presented thus far show that Fus3p and Kss1p play positive roles in the regulation of Far1-independent G1 arrest. Therefore, we would predict that STE11-4 far1 fus3 and STE11-4 far1 kss1 strains should be less sensitive than a STE11-4 far1 strain in a halo assay because of the elevated levels of the G1/S cyclins. Surprisingly, however, deletion of either FUS3 or KSS1 in the STE11-4 far1 strain causes enhanced -factor sensitivity in a halo assay (Figure 6), although deletion of both genes causes -factor resistance (Figure 1B). A trivial explanation of slower growth rate for the more sensitive strains does not account for the increase in -factor sensitivity, because long-term incubation of the plates does not result in smaller halos. Alternatively, the greater sensitivity might be due to hyperinduction of the Ca²⁺-dependent pathway (MOSER et al. 1996 ) that mediates -factor-induced cell death (IIDA et al. 1990 ). However, viability counts of the STE11-4 far1 fus3 and STE11-4 far1 kss1 strains exposed to 100 nM -factor for 18 hr indicate that the -factor-induced sensitivity is not due to decreased viability and that the fus3 null mutation decreases the percentage of -factor-induced cell death that occurs in the presence of -factor (data not shown). One interpretation of these observations is that the halo assay represents the net sum of both G1 arrest and proliferative functions, and that Fus3p and Kss1p also promote proliferation in addition to G1 arrest.

View larger version (43K): In this window In a new window Download PPT slide   Figure 6. MCM1 requires FUS3 but not KSS1 to promote proliferation. Strains: WT, STE11-4 far1, STE11-4 far1 fus3, STE11-4 far1 kss1 are EY1118, EY1298, EY1321, and EY1335 bearing either Yep24 or MCM1-2µ plasmids. Strains were tested for -factor sensitivity in a halo assay using 3 µl of 50 µM -factor. Plates were photographed after 36 hr at 30°.

Comparative phenotypic analysis of additional fus3 and kss1 strains (summarized in Table 3) suggests that Fus3p and Kss1p may counteract G1 arrest through distinct functions. First, a fus3 null mutant undergoes significantly more budding arrest than does a far1 mutant, despite slightly higher levels of Cln2p/Cdc28p kinase (Figure 3C) of equivalently high specific activity. Furthermore, a fus3 null mutation causes enhanced budding arrest in the background of a far1 null, as shown by the greater partial budding arrest in a fus3 far1 double mutant compared to a far1 single mutant. Again, the greater budding arrest occurs despite high levels of active Cln2p/Cdc28p kinase (Figure 3C). Assessment of pathway activity using the FUS1-lacZ reporter gene shows that fus3 and STE11-4 far1 fus3 strains have reduced Ste12p activity compared to wild-type and STE11-4 far1 control strains (Figure 2; ELION et al. 1991A ). These observations suggest that Fus3p regulates proliferation through a function that is not linked to increased G1 cyclin levels or reduced Ste12p activity.

Kss1p appears to counteract G1 arrest by a distinct mechanism. Although kss1 and STE11-4 far1 kss1 strains also have enhanced -factor senstivity (Table 3), the enhanced sensitivity correlates with slower recovery, as shown by a delay in resumption of cell division upon -factor withdrawal in a STE11-4 far1 kss1 strain compared to the STE11-4 far1 strain (Table 4). The delay in recovery correlates with increased pathway activity, as shown by modestly enhanced expression of the FUS1-lacZ reporter gene (Table 3; also see Figure 2), increased PCL2 mRNA levels (Figure 4A), and decreased Cln2p levels (Figure 3A). (The STE11-4 far1 kss1 strain has less Cln2p compared to the STE11-4 far1 fus3 strain, despite similar levels of CLN2 mRNA.) Thus, Kss1p may promote proliferation during recovery through downregulation of the pathway.

View this table: In this window In a new window   Table 4. Part of the pheromone sensitivity of STE11-4 far1 strains is due to a recovery defect

Overexpression of MCM1 suppresses Far1p-independent arrest:
We attempted to determine whether Swi4p, Swi6p, and Mcm1p transcription factors that control the G1 to S transition might constitute direct or indirect targets of either Fus3p or Kss1p by testing their ability to confer -factor resistance to the STE11-4 far1 strain when overexpressed. Putative targets that might be expected to confer -factor resistance in this test either could be inhibited by Fus3p and Kss1p to promote G1 arrest (such as regulators of CLN1, CLN2, CLB5) or activated to promote recovery (such as regulators of CLN3, PCL2). Swi4p and Swi6p positively regulate CLN1, CLN2, CLB5, and PCL2 in addition to other genes (KOCH and NASMYTH 1994 ). Mcm1p is a MADS box regulator of CLN3, SWI4, and genes involved in DNA synthesis (MCINERNY et al. 1997 ), cell wall biosynthesis and metabolism (KUO and GRAYHACK 1994 ), the G2/M transition (ALTHOEFER et al. 1995 ), and mating (ELBLE and TYE 1991 ; OEHLEN et al. 1996 ; KUO et al. 1997 ). Overexpression of Swi4p and Swi6p, as well as a truncated form of Swi4p (which confers slightly more -factor resistance than full-length Swi4p in a wild-type strain, L. BREEDEN personal communication), does not have an effect on the -factor sensitivity of a STE11-4 far1 strain (data not shown). In addition, co-overexpression of Swi4p and Swi6p (both under the control of the GAL1 promoter), even in the presence of CLN3-1, does not bypass the arrest of a STE11-4 far1 strain (data not shown).

By contrast, overexpression of MCM1 confers significant -factor resistance to both wild-type and STE11-4 far1 strains (Figure 6). Mcm1p is unlikely to cause -factor resistance through inappropriate activation of -specific genes, because excess Mcm1p does not induce expression of -factor or inhibit mating in these strains (data not shown). Mcm1p is also unlikely to promote cell division solely through upregulation of CLN3 and SWI4, because overexpression of SWI4 in the presence of CLN3-1 has no effect in the STE11-4 far1 strain and Mcm1p still bypasses G1 arrest in a cln3 strain (data not shown). In addition, Mcm1p is not bypassing Far1p-independent arrest through overexpression of the G2 cyclins (i.e., CLB2; SIEGMUND and NASMYTH 1996 ), because a GAL1-CLB2 gene does not bypass the arrest of either wild-type or STE11-4 far1 strains (data not shown). These observations suggest that Fus3p and Kss1p may regulate G1 arrest through Mcm1p or genes, in addition to SWI4 and CLN3, that are under Mcm1p control.

Mcm1p requires Fus3p but not Kss1p to counteract G1 arrest:
We tested whether Mcm1p can suppress the arrest of STE11-4 far1 fus3 and STE11-4 far1 kss1 strains to determine whether Mcm1p strictly requires either Fus3p or Kss1p to promote cell division in the presence of -factor. Strikingly, excess Mcm1p fails to bypass the -factor arrest of the STE11-4 far1 fus3 strain (Figure 6). By contrast, excess Mcm1p efficiently suppresses the -factor arrest of the STE11-4 far1 kss1 strain (Figure 6), in addition to that of kss1, hog1, and mpk1 deletion strains (data not shown). Thus, Mcm1p specifically requires Fus3p to promote cell division. These findings argue compellingly for a physiologically relevant role for Mcm1p in regulating proliferation in the presence of -factor. In addition, they argue that Fus3p is required for Mcm1p to promote proliferation, and that this function is not shared by Kss1p.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Fus3p and Kss1p promote G1 arrest independently of Far1p:
To analyze the contribution of Fus3p and Kss1p to G1 arrest that is independent of Far1p, we devised far1 strains with intact G1/S cyclins that undergo G1 arrest as a result of hyperactivation of the mating pathway. Our analysis suggests that Fus3p and Kss1p promote pheromone-induced G1 arrest in at least two ways: through activation of Far1p-dependent inhibition of three G1-cyclin-dependent kinases (Figure 7A), and through Far1p-independent repression of transcription of at least three G1/S cyclin genes (Figure 7B). Whereas Fus3p and Kss1p contribute equally to pheromone-dependent activation of the FAR1 gene (FARLEY et al. 1999 ), the vast majority of the control of Far1p is through Fus3p. Fus3p is as essential as Far1p in inhibiting the specific activity of Cln2p/Cdc28p kinase, although Kss1p may weakly inhibit Cln2p/Cdc28p (Figure 3A). Fus3p and Kss1p together repress transcription of the G1/S cyclin genes, although Fus3p has a greater role (Figure 7B). Transcriptional repression is likely to account for the majority of the inhibition of Cln1p/Cdc28p, Cln2p/Cdc28p, and Clb5p/Cdc28p in a wild-type strain, since overexpression of CLN2 or CLB5 is sufficient to override Far1p-independent G1 arrest. This view is consistent with the rapid drop in CLN1,CLN2, and CLB5 mRNAs that is induced by -factor in a wild-type strain (Figure 4; WITTENBERG et al. 1990 ) and the greater inhibitory effect of Far1p on Cln3p/Cdc28p than on Cln2p/Cdc28p (GARTNER et al. 1998 ; JEOUNG et al. 1998 ).

View larger version (22K): In this window In a new window Download PPT slide   Figure 7. Summary of levels of control of G1 arrest and proliferation by Fus3p and Kss1p.

We demonstrate for the first time that Kss1p has a role in regulating pheromone-induced G1 arrest that is distinct from transcriptional control of the FAR1 gene. The detection of a clear effect of a kss1 null mutation on G1 arrest necessitated conditions that allowed the evaluation of Far1p-independent arrest in the presence of the G1/S cyclins, presumably because Kss1p functions redundantly with Fus3p and Far1p. The existence of Kss1p-dependent regulatory events that control G1 arrest is supported by the -factor resistance of a kss1K42R mutant (Figure 1C). Our findings argue that Kss1p is required for efficient G1 arrest in a wild-type strain, although Fus3p plays a much greater role, as in mating (MADHANI et al. 1997 ; FARLEY et al. 1999 ).

Fus3p and Kss1p block the ability of Cln3p/Cdc28p to activate Swi4p/Swi6p:
How might Fus3p and Kss1p repress the expression of CLN1, CLN2, and CLB5? Cln3p/Cdc28p is a potent activator of transcription of CLN1 and CLN2 (STUART and WITTENBERG 1995 ), making it a potential target of negative control. The fact that hyperactive Cln3-1p/Cdc28p is unable to stimulate expression of CLN1 and CLN2 in a STE11-4 far1 strain in the presence of -factor, although it does so in a wild-type strain (Figure 4), argues strongly that Fus3p and Kss1p block the function of Cln3p/Cdc28p at the G1 cyclin promoters. However, this inhibition may be indirect, because the majority of -factor-induced inhibition of Cln3p/Cdc28p is from Far1p (JEOUNG et al. 1998 ), STE11-4 does not lower Cln3p levels (data not shown), and Cln3-1p is still active in the STE11-4 far1 strain (Figure 4B legend). Thus, Fus3p and Kss1p may inhibit another component of the transcription apparatus or activate a repressor that blocks the expression of the G1/S cyclins. Overexpression of SWI4 and SWI6, either alone or in combination, does not confer -factor resistance to wild-type or STE11-4 far1 cells, arguing against the simplest view that Swi4p or Swi6p is a rate-limiting target of the MAP kinases. To date, Dig1p/Rst1p and Dig2p/Rst2p are the only known repressors in the mating pathway (TEDFORD et al. 1997 ). However, Dig1p/Rst1p and Dig2p/Rst2p may have a function that pertains to the G1 cyclins in addition to Ste12p on the basis of two hybrid interactions with Cln1p and Cln2p (COOK et al. 1997 ; TEDFORD et al. 1997 ). Alternatively, Fus3p and Kss1p may upregulate a Ste12p-dependent repressor gene. This possibility is consistent with the observation that a GAL1-STE12 gene induces cells to accumulate in G1 phase in W303a (DOLAN 1996 ), although excess Ste12p does not produce G1 arrest in a W303a far1 strain (SATTERBERG 1993 ).

Fus3p and Kss1p may promote proliferation:
Our analysis suggests that the mating MAP kinases also counteract G1 arrest through overlapping and distinct functions. Fus3p and Kss1p together enhance expression of the CLN3 and PCL2 genes in the presence of -factor. This activation is likely to promote recovery, because cln3 and pho85 null mutations delay recovery (Table 4; NASH et al. 1988 ). The PCL2 gene has a pheromone response consensus sequence (TGAAACA) upstream of the ATG, so its activation may occur through Ste12p, as is the case for genes that downregulate the pathway (i.e., GPA1, SST2, and MSG5).

Fus3p may also promote proliferation by a mechanism that does not involve upregulation of Ste12p or the Cdc28p machinery (Figure 7B). A fus3 null undergoes significant -factor-dependent inhibition of budding and DNA synthesis, despite reduced Ste12p activity and elevated levels of CLN1, CLN2 mRNAs (Figure 4A), and Cln2p/Cdc28p kinase. These fus3 null phenotypes are also detected in far1 and STE11-4 far1 strains (Table 3), consistent with the loss of a proliferative function. Fus3p may attenuate the pathway activity through its function as a kinase (GARTNER et al. 1992 ; KRANZ et al. 1994 ; ERREDE and GE 1996 ) or through functions not dependent upon catalytic activity (FARLEY et al. 1999 ). Fus3p promotes cell division through a function that is not shared by Kss1p, on the basis of the dependency of Mcm1p for Fus3p but not Kss1p. Fus3p is required for vegetative growth (ELION et al. 1991A ), raising the possibility that these proliferative functions are related.

By contrast, Kss1p appears to promote recovery by inhibiting the activity of the pathway, possibly near the Ste11p step (Figure 4 and Figure 7B). Kss1p also appears to positively regulate the abundance of Cln2p (Figure 3A). Kss1p could inhibit the pathway and increase the level of Cln2p indirectly through its ability to act as a repressor of Ste12p when catalytically inactive (MADHANI et al. 1997 ), or because of a distinct function(s) that requires kinase activity.

Fus3p may activate Mcm1p or genes under its control:
Fus3p may promote proliferation through the MADS box transcription factor Mcm1p or gene products under its control. Support for this comes from the observation that Mcm1p specifically requires Fus3p, but not Kss1p, Hog1p, or Mpk1p to suppress Far1p-independent G1 arrest. Genetic tests argue that Mcm1p does not promote proliferation solely through upregulation of the CLN3 and SWI4 genes, implying that additional Mcm1p-dependent genes are required (such as cell wall and DNA synthesis genes (KUO and GRAYHACK 1994 ; KUO et al. 1997 ; MCINERNY et al. 1997 ). Further work is needed to determine which genes must be upregulated by Mcm1p and whether Fus3p regulates Mcm1p or gene products under its control. Mcm1p might be directly phosphorylated by Fus3p in response to pheromone. Mcm1p is known to be phosphorylated in response to another extracellular stimulus, high salt (KUO et al. 1997 ). Alternatively, Fus3p may activate Mcm1p indirectly through the regulation of a protein required for Mcm1p function.

Are the G1/S cyclins the only targets of negative control?
Our experiments reveal, quite surprisingly, that it is possible to inhibit DNA synthesis and budding under conditions of high Cln2p/Cdc28p kinase and not further inhibit DNA synthesis under conditions of reduced Cln2p/Cdc28p kinase. It is possible that conditions that uncouple the control of budding from G1 cyclin levels lead to activation of budding or DNA synthesis checkpoint apparati and cell cycle arrest (SIA et al. 1996 ; WEINERT 1998 ). On the other hand, the absence of a strict correlation among Cln2p/Cdc28p levels, budding arrest, and inhibition of DNA synthesis may hint at the existence of auxiliary mechanisms that inhibit budding and DNA synthesis in a wild-type cell and indicate that Fus3p and Kss1p have additional targets of negative control. This possibility is consistent with the existence of three recessive mutations that confer -factor resistance to a far1 cln2 strain without a clear effect on the G1 cyclins or Ste12p activity, far3 (HORECKA and SPRAGUE 1996 ), par2, and par3 (V. CHERKASOVA and E. A. ELION, unpublished data). These mutations could either block G1 arrest or promote recovery.

Proper control of Ste11p may be critical for G1 arrest:
How does a cell arrest in the presence of -factor if Fus3p and Kss1p have both cell division arrest and proliferative functions? Our analysis of Far1p-independent arrest suggests that proper regulation of Ste11p may be central to determining whether a cell arrests or divides. First, the ability to restore G1 arrest to a far1 mutant is specific to Ste5p and Ste11-4p. Overproduction of other rate-limiting components such as Ste4p (Gß), Ste20p, Bem1p, and Ste12p does not restore G1 arrest to a far1 mutant (LYONS et al. 1996 ; data not shown). Ste5p and the STE11-4 mutation both overcome the basal inhibitory state of Ste11p (STEVENSON et al. 1992 ; CHOI et al. 1994 ), suggesting that an override of negative control of Ste11p is required to induce Far1p-independent arrest. Such an override may circumvent the inhibitory effects of high levels of Cln2p/Cdc28p kinase that accumulate in a far1 null and that inhibit the activity of Fus3p at the Ste20p/Ste11p step and promote recovery (WASSMAN and AMMERER 1997 ; LEZA and ELION 1999 ). It is interesting to speculate that the relative levels of active Fus3p and Kss1p determine whether a cell arrests or divides. For example, a latent increase in the level of activity of proliferative pathways that operate parallel to Cdc28p (ESPINOZA et al. 1994 ; MEASDAY et al. 1994 ; CVRCKOVA et al. 1995 ; IGUAL