The nonessential RGD1 gene encodes a Rho-GTPase activating protein for the Rho3 and Rho4 proteins in Saccharomyces cerevisiae. Previous studies have revealed genetic interactions between RGD1 and the SLG1 and MID2 genes, encoding two putative sensors for cell integrity signaling, and VRP1 encoding an actin and myosin interacting protein involved in polarized growth. To better understand the role of Rgd1p, we isolated multicopy suppressor genes of the cell lethality of the double mutant rgd1Δ mid2Δ. RHO1 and RHO2 encoding two small GTPases, MKK1 encoding one of the MAP-kinase kinases in the protein kinase C (PKC) pathway, and MTL1, a MID2-homolog, were shown to suppress the rgd1Δ defects strengthening the functional links between RGD1 and the cell integrity pathway. Study of the transcriptional activity of Rlm1p, which is under the control of Mpk1p, the last kinase of the PKC pathway, and follow-up of the PST1 transcription, which is positively regulated by Rlm1p, indicate that the lack of RGD1 function diminishes the PKC pathway activity. We hypothesize that the rgd1Δ inactivation, at least through the hyperactivation of the small GTPases Rho3p and Rho4p, alters the secretory pathway and/or the actin cytoskeleton and decreases activity of the PKC pathway.
IN Saccharomyces cerevisiae, the Rho family of GTPases is thought to have a central role in the polarized growth process (Drubin and Nelson 1996; Pruyne and Bretscher 2000). The main functions assigned to these GTPases involve bud formation and cell surface growth, which might occur through the involvement of the actin cytoskeleton and the secretory pathway (Imaiet al. 1996; Tanaka and Takai 1998). Although six open reading frames (ORFs) could encode Rho-GTPases in yeast (Garcia-Ranea and Valencia 1998), genetic and functional analyses have allowed the identification of five Rho members: Cdc42 and Rho1 to Rho4. These small GTPases function as binary switches, which are turned on and off by binding to GTP or GDP, respectively. The GTP-bound form interacts with its specific target and performs its cell functions (Tanaka and Takai 1998). Small GTPases are regulated by GAPs (GTPase-activating proteins), GEFs (GDP-GTP exchange factors), and a GDP dissociation inhibitor.
During the sequencing of the genome of S. cerevisiae, we identified a new gene encoding a protein with a RhoGAP homology domain (Doignonet al. 1993). This protein, called Rgd1p (for related GAP domain), was shown in vitro to be a GTPase activating protein for the Rho3 and Rho4 proteins (Doignonet al. 1999). Thus, in activating the hydrolysis of GTP, Rgd1p negatively regulates the action of these two Rho proteins. Rho3p and Rho4p play a role in bud formation and have some partially overlapping functions (Matsui and Toh-e 1992a). Deletion of RHO4 did not affect cell growth, whereas deletion of RHO3 caused a severe growth delay and a decrease in cell viability. Overexpression of RHO4 suppressed the growth defect in rho3 cells. Depletion of both RHO3 and RHO4 gene products resulted in lysis of cells with a small bud, which could be prevented by the presence of osmotic stabilizer in the medium (Matsui and Toh-e 1992b). In this latter condition, Rho3p- and Rho4p-depleted cells lose cell polarity as revealed by chitin delocalization and by random distribution of actin patches.
Analysis of rho3 suppressors has revealed genetic links with some regulatory elements of the actin cytoskeleton such as CDC42 and BEM1 (Matsui and Toh-e 1992b). We have previously shown relationships between RGD1 and the actin cytoskeleton-linked genes such as VRP1, LAS17, and MYO1; the combinations rgd1Δ vrp1Δ, rgd1Δ las17Δ, and rgd1Δ myo1Δ are synthetic lethal (Roumanieet al. 2000). Moreover, when the VRP1 product is absent, the production of GTP-constitutive forms of Rho3p and Rho4p is detrimental to yeast cells in agreement with an in vivo GAP activity of Rgd1p on both GTPases (Roumanieet al. 2000). Nevertheless, RGD1 inactivation does not lead in itself to a defect in actin organization or in budding pattern, at least under standard growth conditions. RHO3 also displays genetic interactions with the SEC4 gene whose product is involved in exocytosis (Imaiet al. 1996). Moreover, a physical interaction of Rho3p with Exo70p, a component of the exocyst complex, and with Myo2p, the myosin responsible for secretory vesicle movement, has been reported (Robinsonet al. 1999). Thus, Rho3p regulates cell polarity by simultaneously directing the rearrangements of the actin cytoskeleton and the polarized delivery and fusion of secretory vesicles to specific sites on the cell surface (Adamoet al. 1999).
We also reported genetic interactions between RGD1 and the SLG1 and MID2 genes (de Bettignieset al. 1999). SLG1 has also been designated HCS77 (Grayet al. 1997) and WSC1 (Vernaet al. 1997), but for simplicity this gene is referred to here as SLG1. Slg1p and Mid2p are both plasma membrane proteins with partial overlapping functions (Ketelaet al. 1999). They act upstream of the protein kinase C (PKC) pathway and are thought to monitor the state of the cell surface and relay the information to Pkc1p (Grayet al. 1997; Vernaet al. 1997; Jacobyet al. 1998; de Bettignieset al. 1999). The protein kinase C is mostly regulated by the small GTPase Rho1p in vivo (Nonakaet al. 1995; Kamadaet al. 1996). The Pkc1p activates a mitogen-activated protein (MAP) kinase cascade, named the PKC pathway, consisting of Bck1p (Costiganet al. 1992; Lee and Levin 1992), Mkk1p/Mkk2p (Irieet al. 1993), and the MAP kinase Mpk1p (Torreset al. 1991; Leeet al. 1993). Activation of this pathway is particularly important in response to various external stresses, including high temperature, low osmolarity, and cell wall disruption, as well as being important during mating (Heinischet al. 1999). The protein Slg1 is linked to the PKC pathway by the finding that this MAP kinase cascade is activated by heat stress via Slg1p (Grayet al. 1997). A direct interaction of Slg1p with Rom2p, one of the Rho1p-GEFs, has been recently reported and this interaction is responsible for the activation of the PKC pathway through Rho1p (Philip and Levin 2001).
The loss of RGD1 function amplifies the phenotype due to the SLG1 deletion and the small-budded double-mutant cells die because of defects in cell wall structure and lysis upon bud growth. In parallel, the inactivation of MID2, the other putative sensor for cell integrity signaling in S. cerevisiae (Rajavelet al. 1999), exacerbates the specific phenotype of the rgd1Δ mutant with an increase in dead cells at late exponential phase in minimal medium (de Bettignieset al. 1999). Taken together, our results suggest that Rgd1p has a regulatory role in connection with both the PKC pathway and the actin cytoskeleton organization in S. cerevisiae.
To further elucidate the function of RGD1, we isolated multicopy suppressors of the viability defect of the rgd1Δ mutation in minimal medium. Phenotypic and genetic analysis has allowed the identification of several multicopy as well as monocopy suppressors of rgd1Δ: the RHO1 and RHO2 genes encoding two GTPases (Madauleet al. 1987) involved in actin cytoskeleton organization (Yamochiet al. 1994; Kohnoet al. 1996), the MID2-homolog MTL1 (Ketelaet al. 1999; Rajavelet al. 1999), and the MKK1 gene coding for one of the MAP-kinase kinases of the PKC pathway, (Irieet al. 1993). Considering the suppressor effect of additional PKC pathway components, we show that activation of the PKC pathway prevents lethality of rgd1Δ cells. Analysis of the transcriptional activity of Rlm1p, one of the targets of the last kinase in the PKC pathway, and study of the PST1 transcription, which is positively regulated by Rlm1p, showed that the rgd1Δ mutation decreases the activity of this MAP-kinase pathway in minimal medium at late exponential phase. This decrease in PKC pathway activity is at least partly responsible for the rgd1Δ cell viability loss under particular growth or physiological conditions.
MATERIALS AND METHODS
Strains, media, and transformations: The Escherichia coli XL1-Blue (Stratagene, La Jolla, CA) was used for cloning and propagation of all plasmids. E. coli cells were cultured in DYT medium (Bacto tryptone, Bacto yeast extract, NaCl) and transformed by standard CaCl2 method (Sambrooket al. 1989). The S. cerevisiae experiments were performed mainly in X2180 genetic background and the yeast strains used are listed in Table 1. Yeast cells were grown under standard conditions either in YPD (1% Bacto yeast extract, 2% Bacto Peptone, 2% glucose) or in synthetic minimal YNB (0.67% YNB without amino acid, 2% glucose) supplemented with the appropriate nutrients. Solid media contained an additional 2% agar. Caffeine growth inhibition was assayed at 3 mg/ml; when indicated, sorbitol was added to a final molarity of 1 m. The YNB inositol-3X medium contains a threefold inositol concentration (6 mg/liter instead of 2 mg/liter). Unless otherwise stated, the growth temperature used was 30°. Yeast transformations were carried out by the lithium acetate method (Agatepet al. 1998).
DNA manipulations, plasmids, and yeast genomic DNA library: Standard procedures were used for DNA manipulations (Sambrooket al. 1989). DNA sequencing was performed using ALF DNA sequencer (Amersham Pharmacia Biotech).
All the plasmids used in this study are listed in Table 2. The high-copy vector pRS425 and low-copy vector pRS415 carrying RGD1 and MID2 were already described (de Bettignieset al. 1999). The RHO1 and RHO2 genes were subcloned from the library YEp13 plasmids as 2.7-kb and 4.9-kb HindIII-PvuII DNA fragments, respectively, and inserted between the HindIII and SmaI sites of the centromeric plasmid pRS315. In a similar way, the 2.9-kb SalI-HindIII fragment carrying the MTL1 gene was inserted into the corresponding sites of pRS315. The MKK1 gene was amplified by PCR and cloned between the XhoI and PstI sites in pRS315. The high-copy plasmid YEp352 and low-copy plasmid YCp50 containing the URA3 marker and bearing PKC1 were kindly provided by D. Levin as was the BCK1 gene and its hyperactive allele BCK1-20 in pRS314. DNA fragments bearing the alleles of BCK1 were removed from the plasmid using SacI and XhoI restriction enzymes and cloned in the corresponding sites of pRS426 and pRS316 plasmids, which contain the URA3 marker. The low-copy plasmid pRS316 containing the MPK1 gene was a gift from C. Mann. The high-dosage MPK1 was obtained by removing a KpnI-SpeI DNA fragment from this plasmid and inserting it into the corresponding sites of pRS426.
The genomic library was constructed by B. Daignan-Fornier (Daignan-Fornieret al. 1994) by digesting yeast DNA partially with Sau3A and cloning fragments ranging from 2 to 8 kb into the BamHI site of the YEp13 shuttle vector. This vector, which contains the S. cerevisiae LEU2 gene as a selection marker, is present at a high copy number in transformed cells (Broachet al. 1979).
Screening for multicopy suppressors: The rgd1Δ mid2Δ S. cerevisiae strain LBD-A7 was transformed with the YEp13 genomic DNA library and transformants were grown on YNB solid medium for 7 days. Two milliliters of a liquid methylene blue staining solution (methylene blue 0.1 g/liter; trisodium citrate dihydrate 20 g/liter) was poured gently onto the plate surface. After an overnight storage at 4°, the transformants forming white or paler colonies than rgd1Δ mid2Δ were selected, and cells picked up from the top of the colony were streaked on YNB to get rid of cross-contamination of the colonies. After this primary screening, the selected transformants were cultured in liquid YNB inositol-3X medium, which gives a rgd1Δ mid2Δ cell lethality near 100% as determined using methylene blue staining (see Figure 1). The growth and the rate of dead cells were monitored after 45 and 65 hr of culture and transformants displaying a reduced lethality rate were further analyzed. Library plasmid DNA was extracted from these transformants using the Robzyk and Kassir (1992) procedure and transferred into XL1-Blue. Plasmid DNA was then submitted to restriction analysis. To confirm the plasmid dependency of the multicopy suppressor effect, the YEp13-based plasmids were reintroduced into the original double-mutant strain, and lethality rate was determined again by methylene blue staining after 45 and 65 hr of growth. Both ends of the inserts carried by the plasmids that still led to a reduced lethality of the double-mutant strain were sequenced.
Determination of cell lethality: Two different methods were used to determine lethality rate. First we used microscopic examination after staining with methylene blue as described previously (Rose 1975; de Bettignieset al. 1999). The lethality rate was then calculated from the counting of at least 400 cells. We also used staining with propidium iodide and analysis by flow cytometry (Deereet al. 1998). The FACScalibur flow cytometer (Becton Dickinson, San Jose, CA) and the Cellquest Software were used to determine the lethality rate from counting 10,000 events. As a gate was used to discard cell aggregates from analysis, the values determined were slightly lower using flow cytometry than methylene blue staining. The method used was indicated in each experiment.
Detection of the MID phenotype: The mating pheromone induced death (MID) phenotype was revealed, as previously described (Iidaet al. 1994; de Bettignieset al. 1999). Briefly, the lethality rate of shmoos of MATa cells was measured 5 and 7 hr after exposure to 6 μm α-factor.
Rlm1p transcriptional activity: The plasmids YS116, pBTM116, and pYW71 were given by K. Matsumuto (Watanabeet al. 1997). The plasmid YS116 is a YEp-based URA3 plasmid harboring the lacZ reporter gene containing LexA DNA-binding sites in its promoter. The yeast shuttle vector pBTM116 produces the LexA DNA-binding domain alone and the plasmid pYW71 the fusion protein LexA-Rlm1ΔN in which the MADS box DNA-binding domain of Rlm1p has been replaced with the DNA-binding domain of LexA. For convenience, the TRP1 marker harbored by these plasmids was removed using the XbaI and NaeI unique sites and replaced by the SmaI-NheI fragment of the YDp-L plasmid carrying LEU2 (Berbenet al. 1991) to give pBLM116 and pYL71 plasmids, respectively (Table 2). The transactivation activity of LexA-Rlm1ΔN was measured by using the lacZ reporter gene carried by the plasmid pYS116. β-Galactosidase assays were performed as described previously (Kaiseret al. 1994). The same cell amount corresponding to an OD600 equivalent to a 0.3 unit was used for each assay, which was performed in triplicate. Activities were calculated according to the adapted Miller formula.
Northern blot analysis: Cells cultivated in YNB were collected and washed in 0.9% NaCl before freezing in dry ice. Total RNAs were extracted from 2 × 108 cells as described previously (Aveset al. 1985). Five-microgram samples of total RNAs, denatured with glyoxal, were separated by agarose gel electrophoresis and transferred to a GeneScreen nylon membrane (Dupont, Wilmington, DE; New England Nuclear, Boston) as described previously (Whiteet al. 1986). PST1 and RPB4 DNA fragments were obtained by PCR using the primers 5′-TGTTGAATGATTGGGCTGGG-3′ and 5′-AAGAAGCAACAACAAGGAGG-3′ for PST1 and 5′-GAATGTTTCTACATCAACC-3′ and 5′-GAGTGTTTCTAGGTTTGAC-3′ for RPB4. Probes were labeled with [α-32P]dCTP (Amersham, Buckinghamshire, UK) using a random priming kit (Promega, Madison, WI). After hybridization, blots were washed according to the GeneScreen recommendations and quantification was achieved using a Phosphor-Imager (Storm 860, Molecular Dynamics, Sunnyvale, CA). The RBP4 gene of which the product is present during all growth phases (Rosenheck and Choder 1998) was used as an internal standard; PST1 mRNA levels were normalized to RPB4, using the first point as an arbitrary unit of 1.
Screening and characterization of rgd1Δ mid2Δ multicopy suppressor genes: Loss of the Rho-GAP encoding RGD1 function results only in a slight cell mortality in YNB medium, making genetic screens difficult. The rgd1Δ mutant presents an ~15% lethality rate beginning at the late exponential phase, the dead cells being mainly small budded (Bartheet al. 1998). We noted an enhancement of the rgd1Δ-specific viability loss at the late exponential phase when MID2 was inactivated. The percentage of dead small-budded cells is ~60% in rgd1Δ mid2Δ (de Bettignieset al. 1999). Thus, to easily screen multicopy suppressors of the rgd1Δ mutation, the rgd1Δ mid2Δ double mutant was used. This strain was transformed with the yeast YEp13 genomic library and screened for restoration of cell viability, selecting the whiter or clear blue-colored colonies after addition of a methylene blue solution onto the plate surface. From ~20,000 yeast transformants, 82 positive clones were retained. To confirm and to better discriminate the suppressor effects, the 82 transformants were cultured in liquid YNB inositol-3X medium, a medium containing threefold the inositol concentration of standard YNB. Indeed, whereas the wild-type and mid2Δ strains grew normally under this growth condition, the cell lethality of the rgd1Δ and rgd1Δ mid2Δ strains was higher and that of the double mutant close to 100% at late exponential phase (Figure 1). The cell growth and viability were monitored after 45 and 65 hr of culture when the mutant response was observable. Of the 82 isolated transformants, 22 presented a cell lethality lower than that of the double mutant. To verify that the suppression was plasmid dependent, the plasmids extracted from the 22 yeast transformants were reintroduced in rgd1Δ mid2Δ. Cell viability was again examined from the new transformants after 45 and 65 hr of culture in liquid YNB inositol-3X. After this step, only 12 plasmids partially suppressing the cell mortality of the double mutant at the late exponential phase and in stationary phase were retained. The insert junctions of plasmids were sequenced and different genes were identified comparing the sequences with the Saccharomyces cerevisiae Genome Database. The RGD1 and MID2 genes were isolated two and four times, respectively, giving an internal screening control. Because the inserts of the remaining plasmids contained just one complete ORF, we were able to directly assign the specific multicopy suppressor effects to the RHO1, RHO2, MKK1, and MTL1 genes. RHO1 and RHO2 genes were found two times each, and the MKK1 and MTL1 genes were found only one time each.
Phenotypic study of the multicopy suppressor genes in rgd1Δ mid2Δ: To specify the effect of the different suppressor genes, a phenotypic study was first undertaken in the double-mutant background. Each double-mutant strain transformed by one of the previously identified suppressor genes was cultivated in liquid YNB inositol-3X medium and the cell viability was monitored during growth by flow cytometry (Figure 2). As expected, the strain containing the RGD1 gene on YEp13 presented the lowest lethality. The viability value was consistent with the absence of lethality of mid2Δ, and in that way this transformant behaved like a mid2Δ strain (de Bettignieset al. 1999). In the same way, the strain containing the YEp13 plasmid-borne MID2 gene displayed the same behavior as a rgd1Δ strain. Indeed, after 30 hr of growth this strain exhibited ~25% viability loss. Overexpression of RHO1, RHO2, MKK1, and MTL1 led to a suppressor effect with an intermediate percentage of cell lethality compared to 85% obtained with the plasmid YEp13. The strains carrying the high-copy plasmid-borne RHO1 or RHO2 gave similar results after 30 hr of growth with ~40% of mainly small-budded dead cells; it was ~50% for MKK1 and MTL1. The suppressor effect was also observed in YNB standard medium with the same gradation. The percentages of cell lethality were 60, 5, and 18% for the double-mutant strain containing the YEp13 without insert and carrying RGD1 or MID2, respectively. Overexpression of RHO1 and RHO2 gave a value of 20%, whereas it was slightly higher (25%) for the strain overexpressing MKK1 and MTL1 (data not shown).
Introduction of the rgd1Δ mutation into the mid2Δ strain accentuates the mid2Δ caffeine sensitivity and the rgd1Δ mid2Δ double mutant is hypersensitive to caffeine, which can be remedied by addition of 1 m sorbitol (de Bettignieset al. 1999). We wanted to determine whether these suppressor genes could also suppress the caffeine hypersensitivity of the double mutant. Growth of the double mutants transformed with YEp13 and with YEp13 carrying the different genes were examined on YPD plates containing 3 mg/ml caffeine (Figure 3). The different plasmids introduced in the wild-type strain did not modify caffeine sensitivity (Figure 3) and no growth difference was observed when rgd1Δ mid2Δ carrying these plasmids was cultivated in the presence of caffeine and sorbitol (data not shown). In the presence of caffeine only, introduction of MID2 in rgd1Δ mid2Δ resulted in the rgd1 phenotype, as expected. Since the rgd1Δ alone is not sensitive to the drug, MID2 overexpression completely suppressed the double-mutant hypersensitivity. In the same way, RGD1 overexpression only partially suppressed this sensitivity, consistent with what we obtained in a mid2Δ strain (de Bettignieset al. 1999). In agreement with their suppressor effects on cell viability, overexpression of MKK1, MTL1, and RHO2 in the double mutant also decreased the caffeine sensitivity. Surprisingly, RHO1 overexpression did not show any effect under these test conditions. The lack of response of the RHO1 gene with respect to caffeine seems to indicate a cellular mechanism for RHO1 action that is distinct from that occurring in the other suppressor genes.
Analysis of the multicopy suppressor effect in rgd1Δ and mid2Δ mutants: Considering the results obtained in the double-mutant background, we investigated whether the identified genes were multicopy suppressors of rgd1Δ and/or mid2Δ mutations. We therefore undertook similar phenotypic studies in the single mutants and we first tested the effect of gene overexpression on cell viability (Figure 4). As mid2Δ does not show any cell mortality in minimal medium, this test was applied only to the rgd1Δ strain. To exacerbate its defect and thus to better assess the suppressor effect, the rgd1Δ strain was cultivated in liquid YNB inositol-3X medium; in this growth condition the cell mortality reached >40% instead of 15% in the standard YNB. First of all, we verified that the YEp13-borne RGD1 in the rgd1Δ strain gave restoration of cell viability as previously obtained with other high-copy plasmids (de Bettignieset al. 1999). For the RHO1, RHO2, MTL1, and MKK1 genes, we observed again a partial rescue of rgd1Δ cell viability with percentages ranging from 15 to 30%. This test also revealed a suppressor effect due to MID2 overexpression with ~25% cell mortality. These results, based on viability rescue, show that all the genes isolated in our screen, as well as MID2, are rgd1Δ multicopy suppressors. The caffeine response was not tested in the single mutants because rgd1Δ does not exhibit any sensitivity toward caffeine compared to wild type and the mid2Δ sensitivity was not easily usable to examine the suppression effects. To discriminate the effects of RHO1, RHO2, MTL1, and MKK1 in the two single mutants, we next examined the MID phenotype shown by both single mutants. As for the mid2Δ strain (Onoet al. 1994), rgd1Δ shmoos die when exposed to the mating pheromone (de Bettignieset al. 1999). Therefore, MATa rgd1Δ and mid2Δ cells containing the different high-copy plasmids were treated with a 6 μm α-factor in YNB and the viability of shmoos was examined to determine the suppressor effects of RHO1, RHO2, MTL1, and MKK1 (Figure 5). As expected, RGD1 in the rgd1Δ background complemented the shmoos lethality in the presence of the pheromone. Unlike the result in the cell viability test, no effect was detected when MID2 was used to suppress the MID phenotype of rgd1Δ. Conversely, the introduction of RGD1 in high copy did not suppress the MID phenotype of the mid2Δ mutant; however, the complementation of MID phenotype was not complete even when the MID2 gene itself was used. In parallel, it was verified that YEp13-borne RGD1 and MID2 did not lead to shmoos lethality in a wild-type strain (data not shown). When we addressed the suppressor effects of RHO1, RHO2, MTL1, and MKK1 in mid2Δ, we observed a partial suppression from these genes. The RHO genes were more efficient than MKK1 and MTL1, the homolog of MID2. In rgd1Δ, the RHO1, RHO2, and MKK1 genes clearly suppressed the MID phenotype, and RHO1 and RHO2 showed the strongest response with shmoos lethality of 8 and 14%, respectively. On the contrary, MTL1 overexpression did not suppress the rgd1Δ mutation, but rather was detrimental by increasing the shmoos lethality. Except for the MTL1 overexpression effect on the rgd1 MID phenotype, these results are consistent with a multicopy suppressor effect of the RHO1, RHO2, MTL1, and MKK1 genes that is specific to the rgd1Δ mutation.
Suppressor effect of the genes carried by low-copy plasmid in rgd1Δ: The suppressor effect was also investigated from these genes carried by the low-copy plasmids pRS315 or pRS415. These plasmids were introduced in the rgd1Δ strain and cell viability was followed by flow cytometry during growth in YNB inositol-3X medium (Figure 6). For unclear reasons, the rgd1Δ strain containing the low-copy plasmid showed a slightly reduced lethality (25%) with respect to what we observed with the high-copy plasmid YEp13. Introduction of RGD1 complemented its mutation and MID2 partially suppressed the cell viability loss of rgd1Δ strain. Concerning RHO1, RHO2, MTL1, and MKK1, we again found a suppressor effect; the RHO1 gene presented the strongest suppressor with <10% cell lethality after 30 hr of culture. Interestingly, MTL1 in low copy gave a better suppressor effect than in high copy, with ~15% lethality at 30 hr, similar to RHO2 and MKK1 ones. To determine if the suppression by low-copy genes was also relevant to the other phenotype caused by the RGD1 inactivation, we examined the MID phenotype (Figure 7). As with the high-copy plasmid, RHO1, RHO2, and MKK1 partially suppressed the MID phenotype of rgd1Δ. The more pronounced suppression was obtained with the RHO genes. For MTL1, in contrast to what we observed with the high-copy plasmid, its introduction into the low-copy plasmid allowed the detection of a partial suppressor effect in agreement with the results observed from the cell viability test. As before, MID2 did not modify the rgd1Δ shmoos lethality. Thus, even if both inactivations of RGD1 and MID2 led to the MID phenotype, they affect different mechanisms. Taken together, the results show the involvement of these four genes in suppression of both rgd1Δ phenotypes even when expressed from low-copy plasmids.
Suppression by PKC pathway components in rgd1Δ: The identification of RHO1 and MKK1, two genes involved in the PKC pathway in S. cerevisiae, as rgd1Δ suppressors, led us to examine the suppression effects of other components belonging to this pathway. Thus, high-copy and low-copy plasmids carrying the PKC1, BCK1, and the SLT2/MPK1 genes acting upstream and downstream of MKK1 were transformed into the wild-type and rgd1Δ strains. Cell growth and viability of these transformed strains grown in YNB inositol-3X medium were observed (Figure 8). No deleterious effect on cell proliferation was observed as a result of PKC1, BCK1, and MPK1 overexpression. When carried by high-copy plasmids, only MPK1 overexpression allowed a net restoration of the cell viability of the single mutant. A similar response was obtained with MPK1 carried on a centromeric plasmid. Unlike with the high-copy plasmid, PKC1 carried on YCp50 presented little suppressor effect. In addition, the activated mutant allele BCK1-20 isolated as a suppressor of a pkc1 deletion, whose product probably mimics the phosphorylated active form of Bck1p (Lee and Levin 1992), was introduced on high-copy pRS426 or low-copy pRS316 vectors in the LBG44-6B strain. In both cases, a strong suppression effect was observed with <10% of dead cells after 40 hr of culture (Figure 8). Our results show that an increase in the PKC pathway activity suppresses the cell mortality of the rgd1Δ mutant.
Rlm1p transcriptional activity in the rgd1Δ mutant: The previous results suggest that the RGD1 inactivation might decrease the signaling activity of the PKC pathway. To test whether the rgd1Δ mutation could lower the PKC pathway activity, we monitored the transcriptional activation of Rlm1p in the mutant background. The RLM1 gene encodes a member of the MADS-box family of transcription factors, which was identified initially as a loss-of-function mutant that suppresses the lethality associated with a high level of Mkk1p-S386P (Watanabeet al. 1995). RLM1 has a key role in transmitting the cell integrity signal. Recently, it was reported from a genome-wide analysis that the majority of genes whose expression changed following kinase-cascade activation was regulated through the transcription factor Rlm1p (Jung and Levin 1999). To follow the ability of Rlm1p to activate transcription, we quantified expression of the lacZ reporter gene directed by the LexA-Rlm1ΔN fusion protein in wild-type and rgd1Δ mutant cells. The LexA-Rlm1p chimera is phosphorylated by the MAP kinase Mpk1p and activates, in turn, the transcription of the LexA-operator-controlled lacZ reporter gene (Watanabeet al. 1997). Thus, to determine whether the rgd1Δ mutation could decrease the PKC pathway activity and, subsequently, whether a defect in PKC signaling triggered the cell viability loss in rgd1Δ, the β-galactosidase activity and the cell lethality were monitored during growth in YNB (Figure 9). The β-galactosidase activity measured in wild-type cells increased with cultivation time up to the entry into stationary phase. As the activity increases exponentially during the exponential growing phase and since β-galactosidase is relatively stable in S. cerevisiae, we suppose that the RLM1 activation progressively increases during this phase up to the stationary phase. We then observed a net decrease of the reporter activity in stationary phase, suggesting that the RLM1 activity was then weakened in the wild-type resting cells. In the rgd1Δ mutant, the β-galactosidase activity was in the same range as the wild type at the beginning of the culture. However, it was lower when rgd1Δ cells died (Figure 9A). To better compare the RLM1 transcriptional activity with the cell mortality appearance, the data of β-galactosidase activity and lethality rates were analyzed as functions of growth (Figure 9B). In this presentation, we observed that for cell densities up to one OD unit, the β-galactosidase activity increased and was identical in wild-type and rgd1Δ cells, indicating that the Rlm1p was similarly activated in both strains at the beginning of the growth. As shown before, from the time when rgd1Δ cell mortality appeared, the reporter activity increased less than in wild-type cells. Such a result could be explained by the mortality of part of the yeast culture, but it is difficult to correlate the activity variation with the viability loss. To determine whether a PKC-activation defect could trigger the rgd1Δ mortality or, alternatively, whether the lethality might cause the activity change, we tried to examine the timing of both events during growth in the mutant with respect to wild type. Our results showed that the two parameters varied at the same time in rgd1Δ and did not permit us to conclude that RLM1 is involved in activating change in cell mortality appearance.
A similar study was initiated with the double-mutant rgd1Δ mid2Δ; for that, the β-galactosidase activity was measured during growth in YNB medium from the double mutant, the respective single mutants, and the wild-type strain containing the two plasmids each expressing the reporter and transactivator genes. Four time points were chosen and again the reporter activity was compared with the lethality rate (Figure 10). First, a similar β-galactosidase activity defect was observed in mid2Δ and rgd1Δ, although the mid2 cells displayed the same viability as wild-type cells. The defect of the RLM1 transcriptional activity was consistent with the known activator role of Mid2p on the PKC pathway and on Mpk1p phosphorylation under stress conditions (Ketelaet al. 1999; Rajavelet al. 1999). Our results also revealed the role of Mid2p on Rlm1p activity through the PKC pathway during vegetative growth at 30°. For rgd1Δ mid2Δ, while the cell lethality was <5% at 19 hr, the β-galactosidase was very low compared to other strains. In the same way, at 25 and 37 hr, whereas the lethality was not complete, no significant β-galactosidase activity was detected. The lack of reporter activity in the double-mutant background is due to the combined effects of both mutations. It may only partly be explained by mid2Δ inactivation and indicates some role for Rgd1p in activating the PKC pathway. Taken together, these results suggest that the decrease of the PKC pathway activity following rgd1Δ inactivation is in part responsible for cell lethality, but is not enough in itself to trigger lethality and that another defect due to RGD1 inactivation and cumulative to the PKC pathway activation defect is involved in lethality appearance.
Transcriptional regulation of the PST1 gene: To confirm that the changes in β-galactosidase activity in the wild-type and rgd1Δ strains reflected changes in PKC pathway activation, we examined the expression of the PST1 gene that was demonstrated to be dependent on the PKC pathway activity through the transcription factor Rlm1p (Jung and Levin 1999). The PST1 transcriptional level was determined by Northern analysis from the wild-type and rgd1Δ strains grown in YNB; cell lethality of both strains was also measured. Quantification of the PST1 mRNA levels was achieved through Phosphor-Imager measurements and normalized to the expression of RPB4 whose transcript is present at similar levels during all growth phases (Rosenheck and Choder 1998). The expression profile of PST1 along the growth (Figure 11) is strikingly similar to the profile of the β-galactosidase activity described in Figure 9A and allows us to estimate the evolution of the PKC pathway activity. The result confirms that in a wild-type strain the PKC activity rises suddenly at the end of the exponential phase to reach a peak when cells enter the stationary phase and then diminishes in resting cells. This activation of the PKC pathway seems dramatically reduced in rgd1Δ and lethality appears in this mutant precisely at the time when induction of the activity should occur. Given that the expression of PST1 is normalized to that of RPB4, differences in expression cannot be explained by the lethality of 15% of the cells. Thus, inactivation of the RGD1 gene leads to a clear defect in PKC pathway activation, at least under the growth conditions used.
The inactivation of the RGD1 gene encoding a Rho-GAP for the small GTPases Rho3p and Rho4p (Doignonet al. 1999) gives a viability loss at late exponential phase in minimum medium (de Bettignieset al. 1999). Dead cells present a small bud as some mutants of the PKC pathway do (Levinet al. 1990); however, adding sorbitol to the medium does not compensate for this defect. Likewise, the MID phenotype of rgd1Δ is not prevented by osmotic stabilization, contrary to the observations with mutations in MPK1/SLT2 (Erredeet al. 1995). However, as for mutants altered in the PKC-signaling pathway (Costiganet al. 1992; Martinet al. 1996), the heat-shock sensitivity of the rgd1Δ mutant is suppressed by the addition of 1 m sorbitol (de Bettignieset al. 1999); in addition, the RGD1 inactivation exacerbates the caffeine sensitivity of mid2Δ mutants, which is also rescued by sorbitol. Such phenotypic data suggest that RGD1 presents some functional links with the cell wall integrity pathway. This proposition is reinforced by the discovery of genetic interactions between RGD1 and either SLG1 or MID2, two putative sensors for cell wall integrity signaling in S. cerevisiae (Grayet al. 1997; Ketelaet al. 1999; Rajavelet al. 1999).
In this study, we isolated RHO1, RHO2, MKK1, and MTL1 as multicopy suppressors of the rgd1Δ mid2 cell lethality. All four suppressor genes are specific to the rgd1Δ mutation; they partially suppress both rgd1Δ defects, cell lethality and the MID phenotype, and they also work when carried by low-copy plasmids. RHO1 was the more efficient suppressor with RHO2, MKK1, and MTL1 giving a suppression response in the same range. We also found that MID2 was a low-copy suppressor but only of the rgd1Δ cell viability, indicating that the shmoo lethality in rgd1Δ and mid2Δ should be due to distinct altered mechanisms. RHO1, MKK1, and MTL1 were previously shown to be involved in the PKC-signaling transduction pathway. MKK1 encodes one of the MAP-kinase kinases of the PKC pathway. The Rho1p GTPase mediates bud growth by controlling polarization of the actin cytoskeleton and cell wall synthesis (Drgonovaet al. 1999). It controls the cell integrity signaling pathway through two functions. A first essential function of Rho1p is to bind and activate the protein kinase C (Kamadaet al. 1995; Nonakaet al. 1995; Martinet al. 2000). Recently, it was shown by two-hybrid experiments that the cytoplasmic domains of Slg1p and Mid2p interact with the Rho1p-exchange factor Rom2p. The function of the sensor-Rom2p interaction would be to stimulate nucleotide exchange toward the small G-protein, Rho1p (Philip and Levin 2001). Second, Rho1p serves as an integral regulatory subunit of the 1,3-β-glucan synthase complex and stimulates this activity in a GTP-dependent manner (Drgonovaet al. 1996). In addition, it is postulated that Rho1p controls the actin cytoskeleton via its interaction with Bni1p, which binds to profilin (Evangelistaet al. 1997; Imamuraet al. 1997). MTL1 encodes a polypeptide showing 50% identity with Mid2p. Overexpression of MTL1 partially suppressed the pheromone sensitivity of mid2Δ. As for mid2Δ, Mpk1p activation is diminished in the mtl1Δ mutant in response to heat shock (Rajavelet al. 1999). Concerning RHO2, it is involved, as is RHO1, in bud formation and organization of the actin cytoskeleton (Matsui and Toh-e 1992b). Rho2p was shown to be the only yeast Rho family member that can repolarize the actin cortical patches in the pfyΔ strain (Marcouxet al. 2000). It is interesting to note that MID2 overexpression as well as ROM1 and ROM2, two genes coding for exchange factors of Rho1p (Ozakiet al. 1996), also suppress the profilin-deficient phenotype of yeast cells (Marcoux et al. 1998, 2000). It was thus proposed that Mid2p, Rom2p, and Rom1p are in the same signaling pathway, acting upstream of Rho2p to correct the profilin-deficient phenotype (Marcouxet al. 2000). This is consistent with the identification of ROM2, RHO2, and MID2 as multicopy suppressors of the cik1 and kar3 mutations giving a chromosome instability and a karyogamy defect, respectively (Manninget al. 1997). From these and our data, we can suppose that the RHO2 suppressor gene is also linked by some means or other to the cell wall integrity pathway. It will be interesting to study the involvement of Rho2p in the PKC pathway activity.
In view of our results and the data from the literature, we can interpret the suppressor effect of RHO1, MKK1, and MTL1 as the consequence of the overactivation of the PKC pathway. The effect of inactivation and overexpression of MID2 on the rgd1Δ lethality is in agreement with such a hypothesis. Likewise, the synthetic lethality of rgd1Δ slg1Δ and the associated phenotypes could be explained by the diminution of PKC pathway activity caused by the action of both mutations. Nevertheless, although Mid2p and Slg1p show overlapping functions and signal through Rho1p (Ketelaet al. 1999; Rajavelet al. 1999), SLG1 is not a dosage suppressor of the rgd1Δ phenotypes (data not shown), revealing a discriminating function in signal transduction between these two proteins. For RHO2, even if no direct link were established with the cell wall integrity pathway, we can postulate that the RHO2 suppressor effect also acts by increasing the PKC pathway activity. Besides the suppressors, we present evidence indicating that activation of the PKC pathway is necessary to compensate for the defects caused by the RGD1 deletion. Indeed, among the high- and low-dosage kinases of the PKC pathway, MPK1 encoding the last kinase suppressed the rgd1Δ mortality in the same range as RHO2, MKK1, and MTL1; a slight effect was visualized with PKC1 in low copy. In addition, with the hyperactive Bck1-20p form in low copy number the rescue of cell viability is very close to that obtained when RGD1 is carried by the same plasmid, showing that the overactivation of the PKC pathway is sufficient to compensate for the rgd1Δ defect.
On the basis of the phenotypes of the RGD1 deleted strain and the genetic relations between RGD1 and the PKC pathway, RGD1 was thought to be required for the activation of the PKC pathway during the growth phase in which PKC pathway activity seemed optimal. Indeed, the follow-up of the RLM1 transcriptional activity and of the PST1 transcription level indicates that, in our conditions, the PKC pathway activity is not constant during growth and increases particularly in the transition phase and that the inactivation of the RGD1 gene decreases the level of PKC pathway activation when the mutant expresses the phenotype. Similar analysis revealed that MID2 inactivation led to a significant reduction in Rlm1p activation without any significant lethality and showed that a decrease in PKC pathway activation in itself is not enough to cause lethality. Moreover, introduction of the rgd1Δ mutation in the mid2Δ strain led to a dramatic reduction in Rlm1p activation, which seems to precede the very high lethality of the strain. Taken together and considering that overactivation of the PKC pathway suppresses the lethality of the rgd1Δ, these results allow us to conclude that the rgd1Δ mutation alone weakens the activity of the PKC pathway, but also causes another more subtle defect, which, cumulative with the decrease of the PKC pathway, subsequently results in cell lethality during a particular growth phase in minimal medium, where the activation of this signaling pathway seems particularly important. Hence, the small-budded death phenotype of the rgd1Δ mutant at late exponential phase, which is reminiscent of the phenotype of some PKC pathway mutants, is at least partly due to a failure to activate the PKC pathway under this growth phase.
Thus, we show that RGD1 acts somewhere upstream of the PKC pathway. Given that Rgd1p has been shown to have a Rho-GAP activity toward Rho3p and Rho4p, we have to consider that the defect of PKC pathway activation at late exponential phase observed in the rgd1Δ might be mediated by the small GTPases Rho3p and Rho4p. Indeed, a strain carrying another RGD1 deletion removing the Rho-GAP domain presents a cell lethality like rgd1Δ (Bartheet al. 1998). In addition, consistent with such a hypothesis, introduction of the constitutively active forms of either Rho3p or Rho4p (Roumanieet al. 2000) in the wild-type background and growing in YNB inositol-3X medium triggers a cell mortality at late exponential phase with dead small-budded cells, as for the rgd1Δ strain (our unpublished data). Bud growth occurs as the result of two combined events: first, an increased synthesis and assembly of cell wall components and, second, polarization of growth by rearrangement of the cytoskeleton and of the secretory machinery to specifically deliver cell wall constituents to the bud. It was recently reported that, beyond its control in actin organization, Rho3p plays a role in exocytosis (Adamoet al. 1999). Thus the RGD1 inactivation might lead to subtle changes in the cellular mechanisms achieved by Rho3p and Rho4p and to an impairment in polarization and vesicle transport that might indirectly affect cell wall assembly. Such a hypothesis is corroborated by the diminished resistance of the double mutant rgd1Δ mid2Δ, with regard to that of the single mutant mid2Δ, to Calcofluor white and Congo red, two drugs interfering with cell wall assembly (de Bettignieset al. 1999).
We can postulate, then, whether the decrease of the PKC pathway activity in the rgd1Δ mutant would be the consequence of an altered polarized growth whose effects would be observable in particular physiological states. Such a hypothesis would be consistent with the findings that Pkc1p and Slg1p are essential for the repression of rRNA and ribosomal protein genes in response to a defect in the secretory pathway (Nierras and Warner 1999; Liet al. 2000). These authors propose that in a secretory defective cell that can no longer synthesize either the plasma membrane or the cell wall, the continued synthesis of proteins leads to osmotic stress and repression of ribosome synthesis. Likewise, the mrs6-2 mutation that leads to low levels of Ypt protein prenylation, and causes vesicle polarization defects and thermosensitive growth, can be suppressed by genes involved in cell wall maintenance, such as SLG1 and MPK1 (Bialek-Wyrzykowskaet al. 2000). It was proposed that at high temperature mrs6-2 cells have a vesicle-polarization alteration that causes defects in the formation of the cell wall. In rgd1Δ we can imagine the reciprocal effect with a secretory pathway more active and/or active in the wrong conditions, due to the absence of negative regulation of Rho3p and Rho4p by its GAP, and consequently a structurally modified cell envelope leading to a signaling decrease of PKC pathway. Further experiments will shed some light on this hypothesis.
We are grateful to Drs. D. Levin (Baltimore), C. Mann (CEA, Saclay), K. Matsumuto (Nagoya, Japan) for providing strains and plasmids. We thank Marie Beneteau and Virginie Agra for technical participation in this work, and Dr. Sanne Jensen for proofreading of the manuscript. G. de Bettignies was the recipient of a MENRT fellowship. This work was supported by grants from the University Victor Segalen Bordeaux 2 and the Centre National de la Recherche Scientifique.
Communicating editor: L. Pillus
- Received January 22, 2001.
- Accepted September 13, 2001.
- Copyright © 2001 by the Genetics Society of America