In Saccharomyces cerevisiae the protein kinase Cbk1p is a member of the regulation of Ace2p and cellular morphogenesis (RAM) network that is involved in cell separation after cytokinesis, cell integrity, and cell polarity. In cell separation, the RAM network promotes the daughter cell-specific localization of the transcription factor Ace2p, resulting in the asymmetric transcription of genes whose products are necessary to digest the septum joining the mother and the daughter cell. RAM and SSD1 play a role in the maintenance of cell integrity. In the presence of a wild-type SSD1 gene, deletion of any RAM component causes cell lysis. We show here that some mutations of CBK1 also lead to a reduced fertility and a reduced expression of some of the mating type-specific genes. As polarized growth is an integral part of the mating process, we have isolated suppressors of the fertility defect. Among these, mutations in BRR1 or MPT5 lead to a restoration of fertility and a more-or-less pronounced restoration of polarity; they also show genetic interactions with SSD1. Our experiments reveal a multilayered system controlling aspects of cell separation, cell integrity, mating, and polarized growth.
CELLULAR polarity is important in both multicellular and unicellular organisms and allows certain functions to be restricted to particular parts of the cell. In the yeast Saccharomyces cerevisiae this can be seen at different levels such as the positioning of the bud site and the ellipsoidal shape of the cells, which is due to a phase of polarized growth during G2 of the cell cycle. Polarized growth is also particularly visible during filamentous growth and mating. The mating pheromones, secreted by haploid cells (MATa and MATα), block cells of the opposite mating type in G1 and provoke polarized growth leading to the formation of a long projection, or shmoo. The shmoos of opposite mating types fuse to form a diploid zygote. Setting up and maintaining cellular polarity involve the actin cytoskeleton, the septins, and the polarizome and are controlled by mitogen-activated protein kinase (MAPK) signal transduction pathways, such as the filamentation (f)-MAPK pathway and the mating (m)-MAPK pathway (for reviews see Pruyne et al. 2004; Park and Bi 2007).
In S. cerevisiae there is a network of six proteins that is necessary for a wild-type cellular polarity, the regulation of Ace2p and cellular morphogenesis (RAM) network: Cbk1p, Hym1p, Kic1p, Mob2p, Sog2p, and Tao3p (Nelson et al. 2003). The absence of any RAM component results in the formation of large aggregates of cells and a loss of polarity, leading to round rather than ellipsoidal cells (Dorland et al. 2000; Racki et al. 2000; Bidlingmaier et al. 2001; Colman-Lerner et al. 2001; Du and Novick 2002; Weiss et al. 2002; Nelson et al. 2003). This network is a two-kinase regulatory module that appears to be conserved among all eukaryotes and is involved in morphogenesis and cell cycle progression. Other well-defined examples are the mitotic exit network (MEN) in S. cerevisiae and the septation initiation network (SIN) in Schizosaccharomyces pombe (see Tamaskovic et al. 2003 and Hergovich et al. 2006 for reviews). The focal point of these networks is the regulation of a protein kinase of the nuclear Dbf2p related (NDR) family, Cbk1p in the case of the RAM network.
The NDR kinases are a subgroup of the protein kinase A (PKA)/PKG/PKC-like (AGC) serine threonine protein kinases (Hergovich et al. 2006). These kinases are regulated by the phosphorylation of at least two conserved sites (Millward et al. 1999). In Cbk1p, these correspond to serine 570, which undergoes an activating autophosphorylation, and threonine 743, which plays an essential but undefined role in the function of the protein (Jansen et al. 2006).
The role of the RAM network and the protein kinase Cbk1p in cell separation is reasonably well worked out and involves the transcription factor Ace2p (Racki et al. 2000). Ace2p controls the transcription of CTS1 (which encodes chitinase) and other genes whose products are needed to degrade the chitin-rich septum that holds mother and daughter cells together after cytokinesis (Dohrmann et al. 1992; Doolin et al. 2001; Voth et al. 2007). The localization of Ace2p is controlled throughout the cell cycle, and for most of the cycle the protein is excluded from the nucleus because of phosphorylation by Cdc28p (O'Conallain et al. 1999); in late M, the protein accumulates specifically in the nucleus of the daughter cell, the Ace2p targets are transcribed, and cell separation occurs. Cbk1p is also localized transiently to the nucleus of the daughter cells and this localization is concomitant with and dependent upon that of Ace2p (Colman-Lerner et al. 2001; Weiss et al. 2002; Nelson et al. 2003; Jansen et al. 2006; Bourens et al. 2008). A combination of genetic and biochemical studies have shown that Cbk1p phosphorylates the Ace2p nuclear export sequence (NES), disrupting its interaction with the nuclear exportin Crm1p and resulting in the accumulation of Ace2p in the nucleus (Bourens et al. 2008; Mazanka et al. 2008; Sbia et al. 2008; reviewed in Parnell and Stillman 2008).
In certain genetic backgrounds the RAM network is also necessary for cell integrity. Laboratory yeast strains show a polymorphism for the SSD1 gene: some, such as CEN.PK2, carry a full-length wild-type version of the gene (SSD1 or SSD1-v); others, such as W303, carry a truncated version (ssd1-d), which is thought to be inactive (Uesono et al. 1997). In the presence of the ssd1-d allele the deletion of any of the RAM genes is viable, but in the presence of the wild-type SSD1 gene Δram alleles are lethal and lead to cell lysis (Du and Novick 2002; Jorgensen et al. 2002; Kurischko et al. 2005). Ssd1p is thought to play a role in cell wall integrity (Kaeberlein and Guarente 2002; Wheeler et al. 2003) and has been implicated in a large variety of other cellular functions, either via genetic interactions or as a high/low-copy suppressor. The protein has been shown to bind RNA (Uesono et al. 1997) and to interact directly with Cbk1p (Racki et al. 2000; Ho et al. 2002). However, despite a large amount of published data, the exact role of SSD1 is unclear.
In recent years, much progress has been made in understanding the role of the RAM network in cell separation and cell integrity, but its role in cell polarity remains unknown. Studies of the polarity phenotype involving the Δram strains are hampered by the aggregation phe-notype. We have isolated an allele of CBK1, cbk1∷3HA (identical to the allele isolated by Bidlingmaier et al. 2001), which has a loss of polarity with no significant accumulation of cell aggregates. Here we show that strains carrying the cbk1∷3HA allele have a reduced fertility. As polarized growth is an integral part of the mating process, we reasoned that the fertility defect could be linked to the polarity defect. We have isolated a series of extragenic suppressors of the fertility defect of the cbk1∷3HA allele, some of which show a concomitant restoration of cell polarity. Two of these suppressors have been cloned and are loss-of-function alleles of BRR1 and MPT5 (PUF5). We have determined the effect of the deletions of these genes on the other Δram phenotypes and the genetic interactions between CBK1, BRR1, MPT5, and SSD1.
MATERIALS AND METHODS
Strains and media:
For standard DNA propagation the Escherichia coli strain XL2-Blue (Stratagene, La Jolla, CA) was used. All the S. cerevisiae strains used are derived from W303 (Thomas and Rothstein 1989) and CEN.PK2 (a gift from K. Entian) and are listed in supporting information, Table S1. All media for bacteria were prepared as described in Sambrook et al. (1989). Yeast media were prepared according to Dujardin et al. (1980).
Basic genetic techniques and nucleic acid manipulation:
The genetic techniques were used as described in Adams et al. (1997). Gene deletions were made using the PCR procedure of Wach et al. (1994) and Petracek and Longtine (2002); GFP fusions were constructed as described by Wach et al. (1997), using the plasmid pFA6-5GA super bright (Knop et al. 1999). The CBK1 gene was mutagenized using the transposon mutagenesis system of Ross-MacDonald et al. (1997), which creates an in-phase tandem insertion of three copies of the HA epitope. One of these mutant alleles (cbk1∷3HA), an insertion of the 3HA epitope at residue 567, is identical to Cbk1-mTn3F1 isolated by Bidlingmaier et al. (2001). The suppressors were cloned using a genomic bank of the strain FL100 in the plasmid URA3 multicopy plasmid pFL44L (Bonneaud et al. 1991), which was provided by François Lacroute and Nathalie Bonnefoy. The strain SupM1 rapidly accumulated fast growing suppressors (the supM1 allele carries a nonsense mutation), and to circumvent this problem the corresponding gene was cloned by transforming several independent freshly dissected spores.
Transformation of E. coli, plasmid isolation, small-scale genomic DNA isolation, and gel electrophoresis were performed as described by Sambrook et al. (1989). Yeast were transformed as described by Gietz et al. (1992).
Isolation of extragenic suppressors of the mating deficiency of the cbk1∷3HA mutant:
Ten independent subclones of WR580-13b (MATα cbk1∷3HA ade2 LEU2) were grown to confluence on rich glucose plates for 2 days at 28°, and they were then stored in the dark at 4° for 1 week. Small patches of these cells were plated to obtain a unicellular lawn of cells and UV mutagenized to give 1–5% survival. After 2 days' growth in the dark, the plates were washed and the cell suspensions corresponding to each mutagenesis were spread on rich glucose medium to give ∼200 colonies per plate. These were replica mated to a lawn of WR580-4d (Mat a cbk1∷3HA ADE2 leu2) on a medium selecting for diploids (minus adenine, minus leucine) and were examined regularly over a 48-hr period to identify colonies that were able to cross more efficiently. A strain carrying the cbk1∷3HA allele was used as the tester strain because the mating deficiency is more pronounced in cbk1∷3HA × cbk1∷3HA crosses and this facilitated the identification of the suppressors. In total, 20,000 colonies were tested and after verification and retesting 13 suppressors with an improved mating efficiency were selected.
Quantitative mating assays were performed using a protocol adapted from Sprague (1991). The experimental and tester strains carry auxotrophic markers that allow the selection of diploids. Strains were grown in rich glucose medium to an OD600 between 1 and 2, the experimental strain was mixed with a fivefold excess of the tester strain, and 300 μl were plated on a 3-cm petri dish of rich glucose medium and incubated at 28° for 4 hr. After this the cells were resuspended, diluted, and plated on a series of selective media that allowed us to count the haploid parents and the resulting diploids. The parental cultures were also plated to check for reversion of the auxotrophic markers. The mating efficiency was calculated as the number of diploids divided by the sum of the diploids and the experimental haploid.
The mating time course was determined according to Gammie and Rose (2002). Equal numbers of mutant cells of both mating types grown in rich glucose medium were mixed and incubated on 2.5-cm filter discs placed on rich glucose medium at 28° for 2 or 4 hr. The cells were then washed off and fixed, and the different mating stages were counted; the results presented are the mean and standard deviation from three independent experiments (n > 100).
Live cells were grown to early logarithmic phase in rich glucose medium and observed with a Zeiss Axioplan 2 microscope linked to a Cool Snap camera (Princeton Instruments).
Total RNA was extracted with hot acidic phenol (Racki et al. 2000). cDNA was generated using SuperScript II reverse transcriptase (Invitrogen, Carlsbad, CA), using p(dN)6 random hexamers and anchored oligo(dT)23 primers (Sigma, St. Louis). Individual cDNAs were quantified by real-time PCR, using the LightCycler instrument and the FastStart DNA Master SYBR Green I reaction mixture (Roche, Indianapolis). ACT1 was used as a reference gene. Data were analyzed with the LightCycler data analysis software. All the oligonucleotides used in this study are listed in Table S2.
Total RNA of haploid strains was isolated by extraction with hot acidic phenol (Racki et al. 2000) and labeled using the LILRAK PLUS kit; internal standards came from the Two-Color RNA Spike-in kit (Agilent Technologies). The labeled targets were purified using the RNeasy Mini kit (QIAGEN, Valencia, CA). All microarrays are phosphoramidite arrays manufactured by Agilent Technologies (ref G4140B) and correspond to whole-genome coverage long oligonucleotides (60 mers). The labeling efficiency and product integrity were checked according to criteria defined by Graudens et al. (2006). Identical amounts (0.25 μg) of Cy3- and Cy5-labeled targets were mixed according to the manufacturer's instructions and incubated on the microarray slides for 17 hr at 65°, in a rotating oven at 5–6 rpm, using an Agilent hybridization system. The slides were washed as described by the manufacturer and any traces of water were removed by centrifugation at 800 rpm for 1 min. For each comparison, four microarrays were hybridized, including two biological dye swaps.
The slides were scanned with an Axon 4000B scanner (Molecular Devices, Sunnyvale, CA) equipped with 532- and 635-nm excitation lasers for Cy3 and Cy5. Each slide was scanned at 100% laser power at 5-μm resolution. Photomultiplier tube voltages were automatically adjusted to balance the distributions of the red and green intensities and to optimize the dynamics of image quantification. The level of saturated pixels was limited to 0.05%. The resulting 16-bit images were analyzed using the GenePix Pro 6.0 software, the segmentation being computed with the “adaptive circle” method.
Data were processed with the MAnGO software (Marisa et al. 2007). Backgrounds were estimated with the “morphological closing followed by opening” method and were subtracted. Not found, saturated, and bad spots were discarded from the subsequent analysis. Raw data were normalized using the print-tip loess method (Smyth and Speed 2003), which is a local regression normalization within artificial print-tip blocks. Once the biases were corrected, the differential expression of genes between two compared conditions was assessed from the biological replications. A moderated t-test with adjustment of P-values (Benjamini and Hochberg 1995) was computed to measure the significance associated with each expression difference. Genes were selected as significantly differentially expressed when their adjusted P-value was <5%, their mean fold change (average of the fold changes of the four repeats for each comparison) was >1.5 in absolute value, and their mean intensity level [log2(Cy5 × Cy3)/2] was >7.
Mutations in Cbk1p lead to a fertility defect and a reduction in the expression of the mating type-specific genes:
The cbk1∷3HA allele contains an insertion of three copies of the HA epitope at residue 567 and, as noted by Bidlingmaier et al. (2001), it presents a loss of polarity without significant aggregate formation (see Figure S1). We also noted that in plate crosses strains carrying the cbk1∷3HA allele showed a reduced mating efficiency that was more pronounced in cbk1∷3HA × cbk1∷3HA crosses; in quantitative tests they yielded 40% diploids when crossed to the wild type, compared to 70% in wild-type (wt) × wt crosses (Figure 1). Like other protein kinases in the NDR family, Cbk1p is regulated by phosphorylation (Stegert et al. 2005; Hergovich et al. 2006), the autophosphorylation of serine 570 in the activation loop, and phosphorylation of threonine 743, in a hydrophobic C-terminal motif (Jansen et al. 2006). In Cbk1∷3HAp, the three copies of the HA epitope are inserted close to the serine that is autophosphorylated. We reasoned that the phenotype of this mutation might be due in part to a perturbation of the autophosphorylation, so we created derivatives of Cbk1p that cannot be phosphorylated, Cbk1S570Ap and Cbk1T743Ap. Cbk1S570Ap shows a mating defect similar to that of Cbk1∷3HAp (Figure 1). Like Jansen et al. (2006), we found that Cbk1S570Ap has a polarity defect (see the axial ratios in Figure 5) with no accumulation of aggregates, whereas Cbk1T743Ap has a phenotype close to the deleted strain with large aggregates of round cells and is therefore not suitable for quantitative mating assays.
To refine our understanding of the fertility defect in strains carrying the cbk1∷3HA allele, we determined a time course of the mating process as described in materials and methods. Mating mixtures of wild-type or cbk1∷3HA strains were examined after 2 and 4 hr, and the percentages of the different intermediates [unaffected cells, cells with a shmoo, cells that have fused (dumbbell), and zygotes producing the first diploid cell (cloverleaf)] were calculated. The results in Figure 2 show that at a given time point the level of the different mating intermediates is considerably reduced in cbk1∷3HA × cbk1∷3HA crosses compared to the wild type, suggesting that in the presence of the cbk1∷3HA allele the mating process is slowed down, beginning with shmoo formation.
To identify the genes or networks that are regulated by CBK1 in the mating process and the establishment of cellular polarity we decided to undertake a microarray study of the different cbk1 mutants. The most striking result was a reduction in the level of the expression of many of the mating type-specific genes (Table 1). Thus the different cbk1 mutations affect the expression of the mating type-specific genes and this effect is more pronounced in MATα strains.
To validate these results, we decided to perform quantitative qPCR experiments and selected three genes for each mating type. For the MATa strains we chose MFA2, STE2, and AGA2, but no significant and reproducible effects were seen on the expression of these genes (data not shown). For the MATα strains we chose MFα1, STE3, and SAG1. The results in Figure 3A show that in the cbk1 mutations, all three genes have a similar profile of expression, with SAG1 being the most strongly affected. In general, the cbk1 mutations cause a significant reduction in the expression of all three genes tested in a way that essentially parallels the severity of the cbk1 mutation. Because Cbk1p is known to regulate the activity of the transcription factor Ace2p via its localization, we also determined the effect of deleting ACE2 on the transcription of the mating type-specific genes. Our results show that deletion of ACE2 did not reduce the transcript level of our test genes.
Taken together, our results show that certain cbk1 mutant alleles lead to a reduction in fertility and a slowing down of the mating process. There is also a clear reduction in the transcription of many of the MATα mating type-specific genes in a way that is independent of the transcription factor Ace2p.
Isolation of extragenic suppressors of the fertility defect of the cbk1∷3HA mutation:
As polarized growth is an integral part of the mating process, we reasoned that extragenic mutations that increased the mating efficiency of cbk1∷3HA might show a concomitant restoration of cellular polarity and thus help us understand the role of Cbk1p in cellular polarity. To this end we isolated a series of 13 suppressors that showed an increased mating efficiency; of these, some showed no obvious effect on polarity, while others showed a partial restoration of polarity or even a hyperpolarized growth phenotype. Here we present the analysis of 2 suppressors that increase the mating efficiency and show a concomitant restoration of polarized growth: SupM1 and a complementation group of three alleles, SupM2, SupM3, and SupM4.
In plate crosses the mating efficiency phenotype is weak and not suitable as the basis of a cloning strategy; therefore we looked for other more tractable phenotypes associated with the suppressors. In a wild-type CBK1 background, supM1 conferred a recessive slow growth phenotype (Figure 4A). After transformation of a supM1 CBK1 strain by a wild-type library and the selection of transformants with a normal growth phenotype, we showed that supM1 is an allele of BRR1 (bad response to refrigeration) (Noble and Guthrie 1996a), with a nonsense mutation in codon 121 (Brr1p contains 341 aa). We have deleted BRR1 and found that this mutation confers a cryosensitivity on complete medium as reported by Noble and Guthrie (1996a,b); however, we also detected an additional growth defect at 28° in two different genetic backgrounds, W303 (ssd1-d) and CEN.PK2 (SSD1) (Figures 4A and 7A).
Strains carrying suppressors of the complementation group SupM2 all show a recessive thermosensitivity (Figure 4B). After transformation of a strain carrying the supM2 allele by a wild-type multicopy library we recovered 12 plasmids that allowed growth at 36° (File S1 and Table S3). The sequencing of candidate genes showed that the SupM2 group of suppressors are alleles of MPT5. The protein Mpt5p/Puf5p (859 aa) is a member of the RNA-binding Pumilio FBF (PUF) family (Wharton and Aggarwal 2006) and contains a PUF domain of eight repetitions between residues 223 and 530 (Barker et al. 1992; Macdonald et al. 1992). The different suppressor alleles correspond to (1) a nonsense mutation in codon 222, (2) a frameshift in codon 407 that leads to a premature stop, and (3) a missense mutation Y520H (Figure 4C). These mutations lead to the loss of all or part of the PUF domain or the modification of one repetition. We have constructed the Δmpt5 allele in a W303 background and, as previously reported, this allele, like our suppressors, is thermosensitive (Kikuchi et al. 1994; Kaeberlein and Guarente 2002). Our genomic clone gave only a partial complementation of the thermosensitive phenotype of the mpt5 mutants; initially we assumed that this was because overexpression of Mpt5p is toxic (Chen and Kurjan 1997; Tadauchi et al. 2001). In fact the MPT5 clone isolated from our library carries a frameshift due to the insertion of an adenine in a series of eight adenine nucleotides. This results in the formation of a truncated protein of 236 aa missing almost all the PUF domain. It is unlikely that the overexpression of this protein is able to complement the mpt5 mutations; more probably, there is some ribosome slippage, allowing the production of some full-length protein. Subsequently, we have shown that this mutation is present in the genome of the strain used to make our library (FL100) (Lacroute 1968). Interestingly, this mutation has also been described in another “wild-type” strain (Kennedy et al. 1997). Thus it appears that MPT5 presents a polymorphic variation in some laboratory strains.
Deletion of BRR1 and MPT5 also suppresses the fertility defect of cbk1∷3HA:
The nature of the mutations isolated in the BRR1 and MPT5 genes suggested that the suppression of the fertility defect of the cbk1∷3HA allele was due to a loss of function. To determine if deletion of these genes also suppressed the fertility defect we measured the mating efficiency of strains carrying the deleted alleles in the presence of either the wild-type CBK1 gene or cbk1∷3HA and in both mating types. In the presence of cbk1∷3HA, the deletion of either BRR1 or MPT5 significantly increased the mating efficiency, with Δbrr1 restoring an essentially wild-type mating efficiency (Figure 1). Deletion of the genes in a wild-type CBK1 background had little or no effect. Thus Δbrr1 and Δmpt5 alleles also act as suppressors of the fertility defect of cbk1∷3HA, and these were used in preference to the original suppressors in our subsequent experiments.
When we examined the effect of the deletion of BRR1 and MPT5 on the time course of the mating process, we found that the two genes had different effects: the deletion of BRR1 leads to a very slight slowing of the mating process, while deletion of MPT5 accelerates the mating process. In terms of the suppressors, Δbrr1 has no discernible effect as cbk1∷3HA Δbrr1 strains are indistinguishable from cbk1∷3HA strains. The deletion of MPT5 has a dramatic effect, as cbk1∷3HA Δmpt5 strains show an accelerated mating process compared to the wild type, albeit less than the Δmpt5 strains (Figure 2).
As our cbk1 mutants have a defect in the expression of some of the mating type-specific genes, we examined the effect of Δbrr1 and Δmpt5 on the expression of these genes. In MATα strains deleting BRR1 or MPT5 in a CBK1 background does not significantly affect the transcription of MFα1, STE3, or SAG1. In the cbk1 mutant backgrounds, Δbrr1 seems to have only a slight effect, essentially on MFα1, but Δmpt5 produces a clear increase in the transcript levels of all three genes (Figure 3B).
From these results it is clear that the deletion of BRR1 or MPT5 suppresses the fertility defect of cbk1∷3HA in both mating types, and this suppression could be due in part to a restoration of the transcription of the mating type-specific genes. Also we have shown that the deletion of MPT5 is able to overcome the slowing down of the mating process caused by the cbk1∷3HA allele.
The brr1 and mpt5 suppressor alleles have a polarity phenotype:
The suppressors of the fertility defect of cbk1∷3HA were isolated in an attempt to clarify the role of Cbk1p and the RAM network in cellular polarity. Both the BRR1 and the MPT5 deletions affect cellular polarity. We measured the axial ratio of haploid strains carrying the deletion alleles in the presence of three forms of CBK1: wild type, cbk1∷3HA, and Δcbk1. It is clear from the images and the axial ratios that deletion of MPT5 has a much stronger effect on polarity than the deletion of BRR1 (Figure S2 and Figure 5). Deletion of BRR1 leads to a modest hyperpolarization in a CBK1 background (axial ratio 1.29 compared to 1.2 for the wild type) and a limited restoration of polarity in cbk1∷3HA or Δcbk1. In contrast, deletion of MPT5 leads to a clear hyperpolarization (axial ratio 1.43 compared to 1.2 for the wild type, Figure 5; see also Prinz et al. 2007) and restores cellular polarity in a cbk1∷3HA or Δcbk1 background, leading to a slightly hyperpolarized axial ratio (1.3 and 1.29 compared to 1.2 for the wild-type strain).
Thus to different extents, the suppression of the cbk1∷3HA fertility defect by both Δbrr1 and Δmpt5 is accompanied by a restoration of cellular polarity during vegetative growth; and at least in the case of Δmpt5 the polarity effect is independent of CBK1 as it is also seen in a Δcbk1 background.
Deletion of BRR1 or MPT5 leads to an aggregation phenotype:
Our initial observations of the fertility defect suppressors of cbk1∷3HA showed that they contained aggregates of cells, a phenotype that is associated with a loss of Cbk1p function; we decided to quantify this effect. In a wild-type CBK1 background, Δbrr1 and Δmpt5 lead to a slight reduction in the level of individual cells, but no significant increase in the level of large aggregates (containing >12 cells). In the presence of cbk1∷3HA, the deletion of either gene leads to a clear shift toward larger aggregates (>29 cells: cbk1∷3HA, 1.3%; cbk1∷3HA Δbrr1, 11%; and cbk1∷3HA Δmpt5, 16.5%) (Figure 6A). Thus the deletion of BRR1 or MPT5 appears to accentuate the aggregation phenotype of the cbk1∷3HA allele.
As the aggregation phenotype is known to be due to a mislocalization of Ace2p and that Cbk1p is necessary for the correct localization of Ace2p, we decided to determine if the accentuated aggregation phenotype of the suppressors was caused by a modification of the localization of Ace2p or Cbk1p. To do this we looked at the localization of GFP-tagged proteins in strains (CBK1 and cbk1∷3HA) deleted for BRR1 or MPT5. Because the addition of a C-terminal GFP to Cbk1∷3HAp aggravates the aggregation phenotype of this mutant (data not shown), we determined its localization using Mob2-GFP, a RAM component that both colocalizes with Cbk1p and is dependent on Cbk1p for its localization (Nelson et al. 2003). These results show no significant change in the localization of Ace2p, suggesting that the aggregation phenotype of the Δbrr1 and Δmpt5 strains is independent of the localization of Ace2p (Figure S2). The situation is less clear for Cbk1p/Mob2p, especially in the Δbrr1 cbk1∷3HA context, as these proteins have weak nuclear signals that are difficult to see in cell aggregates or cells that have perturbed morphologies. Thus, it is not possible to rule out a subtle variation in the localization of these proteins.
To see if Ace2p function is affected by the deletion of BRR1 or MPT5, we examined the level of transcription of two genes that are directly controlled by this factor, CTS1 and DSE2 (Dohrmann et al. 1992; Colman-Lerner et al. 2001). The results of the qPCR experiments presented in Figure 6B show essentially identical effects on the transcript levels of both CTS1 and DSE2. In a wild-type context (CBK1), Δbrr1 has no effect, while Δmpt5 results in a 40–50% reduction in the level of CTS1 and DSE2 transcripts. In the presence of cbk1∷3HA, transcription of the genes is already reduced by ∼60% and deletion of either BRR1 or MPT5 causes a further reduction, which is more severe in the case of MPT5. As one would expect, the severity of the effect of Δbrr1 and Δmpt5 on the transcription of CTS1 and DSE2 follows the severity of the aggregation phenotypes (Figure 6, A and B). We have also shown that Δbrr1 and Δmpt5 do not affect the transcription of ACE2 (data not shown). These results show that Δbrr1 and Δmpt5 reduce the levels of at least two Ace2p targets in a way that appears to be independent of both the transcription of the ACE2 gene and the localization of the protein.
Genetic interactions between CBK1, BRR1, MPT5, and SSD1:
In the presence of a wild-type SSD1 gene, CBK1 and all the RAM genes are essential (Du and Novick 2002; Jorgensen et al. 2002; Kurischko et al. 2005). Kikuchi et al. (1994) and Kaeberlein and Guarente (2002) have shown that Δmpt5 strains are thermosensitive at 36° in a Δssd1 context and that this thermosensitivity is suppressed by SSD1; we have confirmed these results in our strains (data not shown). We have also deleted BRR1 in CEN.PK2, which carries the wild-type SSD1 gene. The deletion of BRR1 leads to a significant slow growth phenotype at 28°, which is clearly aggravated by the deletion of SSD1 (Figure 7A); this effect can also be seen at 18° and 36° (data not shown). Thus SSD1 acts as a suppressor of the growth defects of both Δbrr1 and Δmpt5 strains. This led us to ask if there is also an interaction between BRR1 and MPT5. To address this we analyzed the segregation of a Δbrr1/BRR1 Δmpt5/MPT5 diploid in a W303 (ssd1-d) and CEN.PK2 (SSD1) background. Figure 7B shows that in the absence of SSD1 (W303 ssd1-d), the double deletion, Δbrr1 Δmpt5, is inviable at both 28° and 36°, whereas in the presence of SSD1, the double mutant is able to grow slowly at 36°. Thus Δbrr1 and Δmpt5 show a synthetic lethal phenotype that is partially suppressed by SSD1 at 36°.
Taken together, these results show that BRR1, MPT5, and SSD1 are intricately related and we have previously shown that Cbk1p interacts physically with Ssd1p (Racki et al. 2000). To determine how MPT5 affects the genetic interaction between SSD1 and CBK1, we analyzed the segregation of a Δcbk1/CBK1 Δmpt5/MPT5 diploid in a CEN.PK2 (SSD1) background (Figure 7C). As expected, the deletion of CBK1 led to a lethal phenotype, but this was partially suppressed when MPT5 was also deleted. The double mutant, Δcbk1 Δmpt5, is thermosensitive: the strain showed a uniform slow growth at 28° and formed aggregates similar to a Δcbk1 Δssd1 strain; individual cells had a very heterogeneous shape and tended to lyse easily. Thus Δmpt5 is a weak suppressor of the lethality of Δcbk1 in a wild-type SSD1 background.
These experiments have uncovered new interactions between CBK1, BRR1, MPT5, and SSD1 and reveal a complex network that controls aspects of cell separation, cell integrity, mating, and polarized growth.
The protein kinase Cbk1p and the RAM network are critical for maintaining cell polarity and efficient cell separation. We have shown here that cbk1∷3HA and cbk1S570A have a reduced mating efficiency in both MATa and MATα strains and that all our cbk1 mutants show a reduction in the level of the transcripts of some mating type-specific genes in MATα strains. It is known that Δram strains have difficulty forming robust shmoos when treated with α-factor (Bidlingmaier et al. 2001; Nelson et al. 2003) and we have observed that cbk1∷3HA and cbk1S570A strains are slow to form shmoos and form small shmoos (data not shown). Furthermore, our analysis of the time course of the mating process shows that in cbk1∷3HA strains, the onset of shmoo formation is delayed, and it appears that other critical steps, such as cell fusion (to form the dumbbells) are also delayed (see Figure 2). Taken together, these data suggest that the fertility defect of the cbk1∷3HA is not simply the result of a reduction in the transcription of some of the mating type-specific genes.
We have identified Δbrr1 and Δmpt5 as suppressors of the mating defect of cbk1∷3HA strains, and subsequent experiments have shown that to a greater or a lesser extent they also affect the polarity, aggregation, and cell integrity phenotypes of these strains.
Brr1p is a nuclear protein (Huh et al. 2003) and this localization is not modified in Δcbk1 strains (data not shown). Originally BRR1 was identified as being required for the stability of the snRNAs and shown to interact with the snRNPs (Noble and Guthrie 1996a,b). The deletion was cryosensitive, and we have shown that it has a severe growth defect at 28° in two different strain backgrounds. The cryosensitivity is partially suppressed by the overexpression of the spliceosomal protein SmD1. The BRR1 deletion is also synthetic lethal with ccr4 and pop2 (which encode components of the Ccr4–Not complex that mediates 5′–3′ mRNA deadenylation) and with snu114 (GTPase component of the U5 snRNP) (Brenner and Guthrie 2005; Pan et al. 2006). It has been found in association with the spliceosomal proteins Smd1p, Smd2p, Smx2p, Smx3p, and Sto1p, which are part of the nuclear cap-binding complex (Gavin et al. 2002, 2006; Krogan et al. 2006), and recently Δbrr1 was shown to have a synthetic phenotype with the deletion of TGS1, the trimethylguanosine synthase (Hausmann et al. 2008). Thus many interactions link Brr1p with RNA metabolism, particularly splicing and degradation, but the exact molecular role of the protein is not known.
Although it is clear that Δbrr1 suppresses the fertility defect of cbk1∷3HA in a way that is independent of mating type, the mechanism by which this happens is unclear. Δbrr1 has very little effect on the expression of the mating type-specific genes and the mating time courses of cbk1∷3HA and cbk1∷3HA Δbrr1 strains are essentially identical.
Mpt5p/Puf5p is a member of the PUF family of RNA-binding proteins and it recruits the Ccr4–Not complex to the 3′-UTR of some mRNAs, stimulating deadenylation and reducing translation (Barker et al. 1992; Macdonald et al. 1992; Goldstrohm et al. 2006, 2007; Wharton and Aggarwal 2006). Systematic studies have suggested that Mpt5p can bind to >200 mRNAs, indicating that Mpt5p could regulate many cellular processes (Gerber et al. 2004; Seay et al. 2006). Several Mpt5p mRNA targets are well characterized such as HO, involved in mating type switching, and STE7 and TEC1, two components of the (f)-MAPK pathway. Mpt5p has also been implicated in the regulation of many other pathways, such as silencing and replicative life span, although the mechanisms involved remain unclear (Kennedy et al. 1997; Cockell et al. 1998; Tadauchi et al. 2001; Irie et al. 2002; Ohkuni et al. 2006; Prinz et al. 2007). Mpt5p also regulates, at least in part, the cell wall integrity (CWI) pathway. The GTPase Rho1p, which is in turn activated by the sensors Mid2p and Slg1p, activates Pkc1p, the central regulatory kinase in the pathway; Lrg1p inhibits this pathway. Mpt5p binds to the LRG1 mRNA, inhibiting its expression, thus activating the CWI pathway. The growth defect of Δmpt5 ssd1 strains is suppressed by the absence of Lrg1p or the overexpression of MID2, SLG1, and PKC1 (Takeuchi et al. 1995; Hata et al. 1998; Kaeberlein and Guarente 2002; Levin 2005; Ohkuni et al. 2006; Stewart et al. 2007). We also isolated SLG1 and PKC1 as multicopy suppressors of the mpt5 mutation during the cloning of the gene (see Table S3).
Mpt5p also shows a complex series of interactions with different factors involved in mating: it interacts physically with Sst2p, the GTPase activating protein for Gpa1p that regulates desensitization to the α-pheromone; the protein also interacts with Fus3p (m-MAPK) and Kss1p (m-MAPK and f-MAPK), and it regulates STE7 mRNA, which is also part of the m-MAPK cascade. Finally, the deletion of MPT5 suppresses the mating defect of a fus3 mutant (Chen and Kurjan 1997; Bardwell 2005; Prinz et al. 2007). In general Mpt5p seems to regulate pathways, or individual proteins, by binding to the 3′-UTR of mRNAs, leading to degradation or a reduction of translation.
It is clear that the mating defect of the cbk1∷3HA strains results from a combination of causes. Recently Kurischko et al. (2008) have shown that Cbk1p and the RAM network are involved in regulating growth, bud emergence, and secretion. In particular, Cbk1p can phosphorylate Sec2p and inhibition of Cbk1p leads to a delay in the polarized localization of the exocytosis regulators Sec4p and Sec2p. Exocytosis and remodeling of the cell wall and cell membrane are crucial elements of shmoo formation and cell fusion during mating. Thus a partial disruption of these processes, together with the reduced transcription of some of the mating type-specific genes, could explain the delay in the onset of mating in the cbk1∷3HA strains.
In a wild-type background, deletion of MPT5 leads to a hyperpolarization of the cells and an acceleration of the mating process. In cbk1∷3HA cells, Δmpt5 stimulates the expression of some of the mating type-specific genes, restores polarity, and accelerates the mating process. Prinz et al. (2007) have shown that the hyperpolarization of Δmpt5 strains requires the presence of TEC1. In conjunction with Ste12p, Tec1p activates the filamentation pathway; Mpt5p binds the TEC1 mRNA and inhibits the expression of Tec1p. Thus in the absence of Mpt5p, Tec1p probably causes a low-level constitutive expression of the filamentation pathway, leading to a hyperpolarization of the cells. In a similar way, given the known interactions between Mpt5p and elements of the m-MAPK pathway, it is possible that deletion of MPT5 leads to a low-level activation of this pathway. Lrg1p could also constitute an element of this system: the protein is required for efficient cell fusion (Fitch et al. 2004) and its expression is inhibited by Mpt5p; thus deletion of MPT5 would lead to increased levels of Lrg1p and improved cell fusion.
Our results and those from several other laboratories have shown that a complex series of genetic interactions link CBK1 (and the RAM network in general) with SSD1, BRR1, and MPT5 and also that Ssd1p interacts physically with Cbk1p and Mpt5p (Racki et al. 2000; Tarassov et al. 2008). No clear unified picture is immediately evident from all these interactions, but there are certain parallels that draw these proteins together. Cbk1p, Ssd1p, and Mpt5p have all been linked to the maintenance of cell integrity; Ssd1p, Brr1p, and Mpt5p are all involved in RNA metabolism; and recently Tao3p, another RAM component, has been implicated in snoRNA biogenesis (Qiu et al. 2008). Brr1p and Mpt5p are known to interact with the Ccr4–Not complex that mediates 5′–3′ mRNA deadenylation and Cbk1p is also linked to this complex as it is a two-hybrid partner of one of the subunits (Not3p) (Racki et al. 2000). Cbk1p is also two-hybrid partner of Dsf2p, which is a deletion suppressor of the Δmpt5 thermosensitive phenotype (W. J. Racki, unpublished results; Ohkuni et al. 2006). All these results are suggestive of a role for these proteins in regulating gene expression.
Recently Mazanka et al. (2008) determined a consensus motif for phosphorylation by Cbk1p and noted an unusual requirement for a histidine residue at −5 to the phosphorylation site. Ssd1p has an exact match to this consensus and both Brr1p and Mpt5p have very close matches including the histidine at −5; thus it is possible that all three proteins are phosphorylated by Cbk1p. The observation that in each case the phenotype associated with the CBK1 deficiency is overcome by deletion of one of the genes (lethality for Δssd1 and reduced fertility for Δbrr1 and Δmpt5) suggests that phosphorylation by Cbk1p has an inhibitory role. In the case of Ssd1p and Mpt5p, we can speculate a little further: Cbk1p could negatively regulate Ssd1p by phosphorylation, with a constitutively active Ssd1p leading to cell lysis in the absence of RAM. There is no clear evidence to indicate if Brr1p and Ssd1p are positive or negative effectors, but in general Mpt5p seems to be a negative regulator. Thus no, or a reduced, Cbk1p inhibitory phosphorylation would lead to a constitutive repression that would be relieved by the deletion of the MPT5 gene (see the discussion of Lrg1p above). These different interactions are summarized in the model in Figure 8. Given the links between Cbk1p and Mpt5p, and Mpt5p and the m-MAPK and f-MAPK pathways, it is interesting to note that the Neurospora crassa homolog of Cbk1p, COT1, has recently been shown to interact genetically with two MAP kinases, MAK1 and MAK2, which participate in the regulation of filamentous growth, hyphal fusion, and sexual development (Maerz et al. 2008).
It is clear that the RAM network interacts with Ssd1p, Brr1p, and Mpt5p to regulate a series of parallel, and probably at least partially redundant, pathways controlling aspects of cell separation, cell integrity, mating, and polarized growth (Figure 8). Some of these functions are essential for viability but may be lethal when overexpressed in an uncoordinated way leading to cell lysis, and this could be why the cell evolved such a complicated multilayered system to control them.
We thank W. Racki for providing the original cbk1∷3HA strain, A.-M. Bécam for invaluable technical assistance, G. Dujardin and N. Bonnefoy for critical reading of the manuscript, H. Delacroix for helpful discussions concerning the microarray experiments, and L. Kuras for help with the qPCR experiments. This work was financed by the Centre National de la Recherche Scientifique (CNRS), by a “Subvention Fixe” from the Association pour la Recherche sur le Cancer, and by an Action Concertée Incitative Biologie Cellulaire, Moléculaire et Structurale grant from the French Ministry of Research. M.B. thanks the Institut de Chimie des Substances Naturelles du CNRS, Gif-sur-Yvette, France, for financial support.
We dedicate this article to the memory of Piotr Slonimski, who died in Paris on April 25th, 2009 at the age of 86.
Supporting information is available online at http://www.genetics.org/cgi/content/full/genetics.109.105130/DC1.
↵1 Present address: Institut de Génétique et Microbiologie UMR8621, Université Paris-Sud 11, F-91405, Orsay, France.
Communicating editor: M. D. Rose
- Received May 15, 2009.
- Accepted June 8, 2009.
- Copyright © 2009 by the Genetics Society of America