A synthetic analysis of the Saccharomyces cerevisiae stress sensor Mid2p, and identification of a Mid2p-interacting protein, Zeo1p, that modulates the PKC1-MPK1 cell integrity pathway

The cell wall of Saccharomyces cerevisiae is a complex and dynamic structure, essential for the maintenance of cell shape and integrity. Composed of 1,3-β-glucan, 1,6-β-glucan, mannoproteins and chitin, the cell wall undergoes changes in shape and composition during the vegetative budding cycle, as well as in the alternative developmental pathways of mating and sporulation. The integrity of the cell wall during these dynamic periods requires PKC1, a gene encoding the budding yeast homologue of mammalian protein kinase C, with mutants displaying a cell lysis defect (Levin & Bartlett-Heubusch, 1992). Cells lacking functional Pkc1p require an osmotic stabilizer for growth, and are sensitive to caffeine and high temperature. When shifted to a hypo-osmotic medium, pkc1Δ mutants are unable to maintain cell wall integrity, and arrest growth with a small bud that lyses at its tip. Although the role of Pkc1p appears to be the activation of the Mpk1p MAP kinase cascade, it is also responsible for a less well characterized secondary pathway co-ordinating ribosome and membrane synthesis (Nierras & Warner, 1999).

The Mpk1p MAP kinase cascade begins with the Pkc1p substrate, Bck1p (Costigan et al., 1992; Lee & Levin, 1992). BCK1 encodes a protein of the MAPKK kinase family. In response to activation by Pkc1p, Bck1p phosphorylates a pair of redundant MAPK kinases Mkk1p/Mkk2p (Irie et al., 1993). In turn, these MAPK kinases dually phosphorylate the MAP kinase Mpk1p/Slt2p on tyrosine and threonine residues (Lee et al., 1993). Phosphorylation of Mpk1p has been demonstrated to occur in response to high temperature, exposure to exogenous mating pheromone, or hypo-osmotic shock (Martin et al., 1993; Buehrer & Errede, 1997; Zarzov et al., 1996). Loss of PKC1 function, or that of any of the components of the MAP kinase cascade under its control, results in a cell lysis defect. This defect is attributable, at least in part, to a defect in the transcription of a variety of genes involved in cell wall biosynthesis (Igual et al., 1996). Regulation of the PKC1MPK1 cell integrity pathway is dependent on the Rho1p GTPase. GTP-bound Rho1p has been shown to physically associate with Pkc1p, resulting in its activation (Nonaka et al., 1995; Kamada et al., 1996). Essential for viability in S. cerevisiae, Rho1p localizes to sites of polarized growth and is implicated in cytoskeletal organization (Evangelista et al., 1997). In addition to its role in the cell integrity pathway, Rho1p in its GTP bound form activates the 1,3-β-glucan synthase complex through Fks1p (Qadota et al., 1996). Several regulators of Rho1 activity have been identified: Bem2p (Peterson et al., 1994) and Sac7p (Schmidt et al., 1997) are GTPase activating proteins, while Rom1p and Rom2p act as GDPGTP exchange factors for Rho1p (Ozaki et al., 1996).

In a body of work, links have been made between the cell surface and the PKC1MPK1 cell integrity pathway (Gray et al., 1997; Verna et al., 1997; Jacoby et al., 1998; Ketela et al., 1999; Rajavel et al., 1999; Philip & Levin, 2001; Sekiya-Kawasaki et al., 2002). Acting as sensors for the cell wall, a number of type I membrane proteins respond to vegetative and stress-related growth requirements by activating the PKC1MPK1 cell integrity pathway. Two families of cell-surface sensors have been described. The first, the WSC family of four related genes, appears to be necessary for vegetative growth. The archetype member Wsc1p functions upstream of the cell integrity pathway, with wsc1Δ mutants showing defective PKC pathway phenotypes. Like Wsc1p, both Mid2p and its functional homologue, Mtl1p, appear to be upstream activators of the cell integrity pathway (Ketela et al., 1999). Both Mid2p and Mtl1p show structural similarity to the WSC family, but show little amino acid sequence identity. All contain a single membrane-spanning domain, an N-terminal signal sequence, an extracellular domain rich in serine/threonine residues, and a short cytoplasmic tail. Identified as a gene necessary for survival upon exposure to mating pheromone (Ono et al., 1994), Mid2p is a type I integral plasma membrane protein. Null mutants of mid2 show normal vegetative growth, although some strains have a temperature-sensitive lysis defect (Rajavel et al., 1999). In addition to the mating-induced death phenotype, mid2Δ mutants are sensitive to caffeine, and resistant to the chitin-binding dye calcofluor white, while overexpression of MID2 leads to an increase in chitin levels and calcofluor white hypersensitivity (Ketela et al., 1999). Both Wsc1p and Mid2p are required for the activation of the cell integrity pathway (Gray et al., 1997; Ketela et al., 1999; Rajavel et al., 1999), and physically interact with Rom2p and increase GEF activity toward Rho1p (Philip & Levin, 2001). Despite these common functions, recent work indicates that Wsc1p is involved in Rho1p-dependent activation of the glucan synthase component, Fks1p, while Mid2p is not (Sekiya-Kawasaki et al., 2002).

To extend this idea, we performed a screen for genes synthetically interacting with MID2, and found that in the absence of Mid2p, components activated by Wsc1p and Rom2p are essential. To examine the signalling role of Mid2p, we performed a two-hybrid screen with the essential cytoplasmic tail of Mid2p, and identified ZEO1 (YOL109w). We show that ZEO1 encodes a 12 kDa protein that primarily localizes to the cell periphery; that zeo1Δ mutants are resistant to calcofluor white; and that Zeo1p is necessary for calcofluor white hypersensitivity on high-copy expression of MID2. We also provide genetic and biochemical evidence linking MID2 and ZEO1 with the PKC1MPK1 cell integrity pathway.

Strains, plasmids and media.
The S. cerevisiae strains used in this study are listed in Table 1 and oligonucleotide sequences are listed in Table 2. Removal of the MID2 cytoplasmic domain was accomplished by introducing BglII sites at residues 871 and 1177 via site-directed mutagenesis (Kunkel et al., 1987). A 306 bp fragment was removed and the resulting MID2 construct was religated. The two-hybrid bait plasmid (pEG202MID2_CYT) was created by generating an EcoRI restriction site through the modification of TGTATC (residues 744749 in the MID2 ORF) to GAATCC via site-directed mutagenesis on MID2 contained in pBSII KS(-) with primer 1. A BamHI restriction site was inserted directly downstream of the stop codon using primer 2. A 330 bp EcoRIBamHI fragment containing the cytoplasmic domain of MID2 was cloned into corresponding sites in pEG202 to generate an in-frame LexAMID2_CYT fusion. Two-hybrid bait constructs were confirmed by DNA sequencing (Applied Biosystems).

Table 1. Description of strains used in this study

Table 2. Oligonucleotide primers used in this study

Specific internal fragments of the MID2 cytoplasmic domain were amplified from BY4741 genomic DNA with Expand DNA polymerase (Boehringer Mannheim) using the primer pairs 3 and 4, 5 and 6, and 7 and 8. PCR fragments were digested with EcoRI and BamHI and subcloned into pEG202. Fidelity of the amplified fragments was confirmed by DNA sequencing.

ZEO1 was amplified from BY4741 genomic DNA using Expand DNA polymerase and the primer pair 9 and 10. A 2·5 kb fragment was excised with SacII and XhoI and cloned into the corresponding sites on pRS316 and pRS426. Fidelity of the PCR was confirmed by DNA sequencing.

Zeo1pHA was created by insertion of a single copy of the haemagglutinin (HA) epitope (YPTDVPDYA) directly upstream of the stop codon between residues 338 and 339 via site-directed mutagenesis (Kunkel et al., 1987) on single-stranded ZEO1 contained on pRS316. Correct insertion of the HA epitope was confirmed by DNA sequencing. A 2·5 kb SstIIXhoI fragment containing ZEO1HA was excised and cloned into pRS426. Zeo1pHA was shown to be functional by the observation that pRS316ZEO1HA could fully complement ZEO1 in a zeo1Δ rom2Δ mutant.

Synthetic lethal mutant screen.
Systematic genetic array analysis (SGA) was used to identify genes essential in a mid2Δ background, as described by Tong et al. (2001). Briefly, HAB976 (Table 1) was obtained in four steps. First, the mid2Δ : : KanMX4 from YD5241 (Table 1) was switched to mid2Δ : : NatMX4 by PCR-based transformation. Second, the Nat^R transformants were mated to Y3084 (Table 1) and the MATa/α diploids were transferred onto sporulation medium. MATα meiotic progeny were then selected on synthetic medium lacking leucine and arginine but containing canavanine. The mating type was confirmed by PCR, according to Huxley et al. (1990). Third, cells were replica plated onto medium containing nourseothricin to select for the deletion mutants. Fourth, cells were replica plated onto medium lacking lysine to identify lys2Δ derivatives. From three SGA screens, 194 potential positives were identified; 11 were confirmed by tetrad analysis.

Localization of Zeo1p.
A ZEO1GFP fusion was generated by inserting a BamHI and NotI site over the start codon via site-directed mutagenesis (Kunkel et al., 1987) with primer 11, using single-stranded ZEO1 contained on pRS316 as template. Clones positive for incorporation of the restriction sites were confirmed by DNA sequencing. Generation of an in-frame GFPZEO1 fusion (GFP F64L S65T; kindly provided by U. Stochaj) was accomplished via a GFP BamHBamHI and ZEO1 BamHIBamHI ligation. Correct orientation of the GFP construct was verified with a NotI diagnostic digestion. Functionality of the GFPZEO1 construct was demonstrated as described for ZEO1HA. Localization of GFPZEO1 was accomplished by examination of live mid-exponential-phase cells carrying pRS316 GFPZEO1.

Calcofluor white tests.
To test strains for sensitivity to calcofluor white, mid-exponential-phase cells were diluted and spotted either onto YEPD agar plates with the indicated amount of calcofluor white, or onto selective media buffered with 10 g MES l^-1 and adjusted to pH 6·2. Plates were incubated in the dark.

Chitin assay.
Total cellular chitin was measured as described by Bulawa et al. (1986), and outlined by Ketela et al. (1999). In brief, washed cells (approx. 25 mg wet cells) were resuspended in 500 µl 6 % (v/v) potassium hydroxide and incubated at 80 °C for 90 min. After cooling at room temperature, 50 µl glacial acetic acid was added. Insoluble material was washed twice with water and resuspended in 250 µl 50 mM sodium phosphate (pH 6·3); 2 mg Streptomyces griseus chitinase (Sigma) was added, and incubated at 25 °C for 2 h. Tubes were centrifuged at 15 000 g for 5 min; 250 µl of the supernatant was transferred to a fresh tube and 1 mg Helix pomatia β-glucuronidase (Sigma) was added. Tubes were incubated at 37 °C for 2 h and then assayed for N-acetylglucosamine content.

Western blot and solubilization tests.
For solubilization tests, total cell extracts were prepared from mid-exponential-phase cells grown in selective medium, by vigorous vortexing in lysis buffer [50 mM Tris/HCl (pH 7·5), 1 mM EDTA, 5 % (v/v) glycerol] in the presence of glass beads and protease inhibitors (Complete protease inhibitor cocktail; Boehringer Mannheim). The resulting slurry was clarified by centrifugation at 3500 g for 10 min at 4 °C to remove unbroken cells and cell walls. The resulting supernatant was divided and one aliquot subjected to centrifugation at 50 000 g at 4 °C for 30 min. The supernatant was withdrawn, and the pellet resuspended in an equal volume of lysis buffer. Samples were resolved by SDS-PAGE and then subjected to Western blotting. Immunodetection of Zeo1pHA was achieved by using anti-HA monoclonal antibody HA11 (Babco) at 1 : 1000 dilution and horseradish-peroxidase-conjugated secondary antibody (Amersham Life Sciences) at a 1 : 2000 dilution. Bands were visualized by enhanced chemiluminescence (Amersham Life Sciences).

Measurement of Mpk1p phosphorylation.
The detection of Mpk1p/Slt2p phosphorylation was carried out as described by Martin et al. (2000). In brief, cells were grown overnight in selective medium at room temperature. For heat-shocked cells, overnight cultures were diluted and grown for 3 h at 37 °C. A 25 ml sample of mid-exponential-phase cells was collected in an equal volume of ice-cold water and centrifuged at 4 °C for 4 min at 1623 g. The cell pellet was briefly washed in 1 ml ice-cold water and transferred to a pre-chilled 1·5 ml collection tube. Cells were repelleted for 20 s at 15 000 g. The supernatant was immediately removed and the pellet resuspended in 300 µl ice-cold lysis buffer [50 mM Tris/HCl pH 7·5, 10 % (v/v) glycerol, 1 % (v/v) Triton X-100, 0·1 % (w/v) SDS, 150 mM NaCl, 50 mM NaF, 1 mM sodium orthovanadate, 50 mM β-glycerol phosphate, 5 mM sodium pyrophosphate], in the presence of glass beads and protease inhibitors (Complete protease inhibitor cocktail; Boehringer Mannheim). Equal concentrations of protein were fractionated by 8 % (v/v) SDS-PAGE. Membranes were probed with phospho-p44/42 MAPK (Thr²⁰²/Tyr²⁰⁴) antibody (New England Biolabs) at 1 : 500 dilution to detect dually phosphorylated Mpk1p. Control of sample loading was monitored using anti-HA, as described above.

Synthetic interaction analysis of MID2
To broadly assess Mid2p function, we made a systematic genetic analysis (Tong et al., 2001) on a mid2 deletion mutant, globally examining the synthetic interactions of MID2 with mutants in all nonessential genes. Previous work (Ketela et al., 1999) identified synthetic lethal interactions in two genes, WSC1 and FKS1. Under the vegetative growth conditions tested we found these, and an additional set of nine genes which when individually deleted led to a synthetic growth defect in a mid2 deletion background (Table 3). Seven of these genes are involved in aspects of the cell wall synthesis or in the PKC1-dependent cell integrity pathway response. Of these, FKS1, CCW12, VAN1 and FPS1 encode a 1,3-β-glucan synthase component, a structural cell wall component, a mannosyltransferase and a glycerol transporter required for normal response to osmotic stress, respectively (Douglas et al., 1994; Luyten et al., 1995; Mrsa et al., 1997; Jungmann et al., 1999). Individual deletion of these genes leads to cell surface stress. Mid2p is required for survival of these mutants, probably by activating the PKC1MPK1 cell integrity pathway (Ketela et al., 1999; Philip & Levin, 2001). Concordant with this explanation are the known synthetic lethal interactions between components of the PKC pathway with FKS1 and FPS1 (Garret-Engele et al., 1995; Tamas et al., 1999) and transcription of CCW12 being Mpk1p-dependent (Baetz et al., 2001). Furthermore, three genes (WSC1/SLG1, ROM2 and SMI1/KNR4) that act to maintain normal activity of Fks1p also show synthetic interactions with MID2. WSC1 and ROM2 encode upstream activators of the cell integrity pathway. Wsc1p is a membrane sensor and Rom2p a GDPGTP exchange factor for Rho1p. WSC1 and ROM2 signal through Rho1p which, in turn, both activates the PKC pathway and stimulates 1,3-β-glucan synthesis by interacting directly with Fks1p (Qadota et al., 1996; Sekiya-Kawasaki et al., 2002). SMI1 is a positive regulator of glucan synthase activity, and shows genetic interactions with components of the PKC pathway (Martin-Yken et al., 2002)

Table 3. Synthetic interactions with MID2

Our results are consistent with cell wall assembly being regulated by two distinct networks involving Rho1p (Sekiya-Kawasaki et al., 2002). One involves signalling from Mid2p through Rho1p to Pkc1p and the PKC1MPK1 pathway, with the second involving specific signalling from Wsc1p and Rom2p through Rho1p to activate Fks1p, as well as to activate the PKC1MPK1 pathway. A prediction from this idea is that the effects of a MID2 deletion should be confined to the PKC1MPK1 pathway. To test this we explicitly examined for synthetic effects of a mid2 deletion with mutants in bck1 and mpk1, downstream components of the PKC1MPK1 pathway. Neither bck1mid2 nor mpk1mid2 double mutants showed synthetic effects. Thus, absence of the Wsc1p/Rom2p/Fks1p network is required for synthetic lethality with a mid2 deletion. Of the remaining genes interacting with MID2, SAC6 and YLR338w (which overlaps with VRP1) are involved in polarity and cytoskeleton function, ILM1 is required for mitochondrial inheritance, and YLR111W is of unknown function.

The cytoplasmic tail of Mid2p is essential for function
As Mid2p shows signalling distinct from Wsc1p, we examined how this signalling specificity was mediated. To act as a stress sensor this plasma-membrane-spanning protein must be able both to detect stress and to relay this information intracellularly. The large highly O-mannosylated extracellular domain presumably involved in signal detection is essential for function (Ketela et al., 1999; and see Fig. 1). To test if the cytoplasmic tail of Mid2p is required for this intracellular relay, we constructed a Mid2p mutant with a deletion of the C-terminal cytoplasmic domain (amino acid residues 252376) from MID2, generating MID2_ΔCYT under the control of the MID2 promoter. Although this mutant localized normally to the cell periphery (not shown), it was unable to confer sensitivity to calcofluor white when expressed at multicopy levels (Fig. 1). Philip & Levin (2001) demonstrated that a deletion of this domain also fails to complement the mating-induced death phenotype of mid2Δ mutants. These results indicate that the cytoplasmic domain is required for a fully functional Mid2p.

(30K): Fig. 1. The cytoplasmic and extracellular domains of Mid2p are essential for function. Mid-exponential-phase BY4742 cells (wild-type) containing vector only, multicopy MID2, multicopy MID2ΔS/T, or multicopy MID2ΔCYT, were diluted to a concentration of 3x106 cells ml-1; 5 µl of this suspension and three subsequent 10-fold serial dilutions were each spotted onto medium containing the indicated concentration of calcofluor white.

Zeo1p interacts with the cytoplasmic domain of Mid2p in a two-hybrid screen
To explore the role of the cytoplasmic domain in Mid2p function, a two-hybrid screen was made to identify physically interacting proteins. The sequence encoding the cytoplasmic domain of MID2 (amino acid residues 250376) was subcloned in-frame to the LexA DNA-binding domain of a two hybrid vector (pEG202-NLS; C. Boone). Sequence analysis showed that all clones were full-length and in-frame with the Gal4 activating domain of the pACTII bait plasmid. One clone containing the gene ZEO1 (YOL109w) was identified several times in the screen and studied further (Fig. 2a). To map the Zeo1p-binding domain of the Mid2p tail, a variety of LexAMID2_ΔCYT fusions were made and tested for interaction with pACTIIZEO1 (Fig. 2c). A shortened domain between amino acid residues 250343 was sufficient for the interaction, while a fusion between residues 250313 showed a weakened interaction. The polypeptide directly adjacent to the transmembrane domain is required for Mid2p interaction with Zeo1p, since cytoplasmic domain fusions that do not contain this region (amino acid residues 283343 and 314376; 3 and 4 respectively in Fig. 2b) showed no interaction.

(39K): Fig. 2. (a) The cytoplasmic domain of Mid2p interacts with Zeo1p in a two-hybrid system. W303 containing pEG202MID2CYT (aa250376), with either vector alone (pACTII) or pACTIIZEO1, were streaked onto medium with or without adenine. (b) Mid2p interaction with Zeo1p is dependent on discrete portions of the cytoplasmic domain. W303 containing pEG202MID2CYT (aa250-376), pEG202MID2CYT (aa250313), pEG202MID2CYT (aa313376), pEG202MID2CYT (aa283343), or pEG202MID2CYT (aa250343) and pACTIIZEO1 were streaked on medium with or without adenine and incubated at 30 °C for 3 days.

Characterization and subcellular localization of Zeo1p
To examine the physical characteristics of Zeo1p, a functional HA-tagged protein, Zeo1pHA, was generated which migrates at its predicted molecular mass of approximately 12 kDa (Fig. 3b). To test membrane association of Zeo1p, partially purified extracts from cells expressing Zeo1pHA were fractionated into supernatant (soluble) and pellet (membrane-associated) portions by centrifugation. Zeo1pHA was found in both the high-speed (HSS) (50 000 g) spin supernatant, as well as the pellet fraction (Fig. 3b). To test whether membrane association is peripheral or integral, partially purified, membrane-containing cell extracts were treated with urea, sodium chloride and sodium carbonate to disrupt peripheral or proteinprotein associations, or with Triton X-100 to disrupt integral association. Zeo1pHA was solubilized in all treatments, indicating that the insolubility is due to peripheral membrane and/or Zeo1pmembrane protein interactions (Fig. 3c).

(88K): Fig. 3. Cellular localization studies on Zeo1p. (a) GFP fluorescence analysis was performed on BY4742 (wild-type) cells expressing pRS316ZEO1GFP. (b) Total cell extracts were centrifuged at 50 000 g and the resulting pellet (P) and soluble (S) fractions were analysed by Western blotting. (c) Immunoblot analysis of cell fractions from BY4742 containing pRS316ZEO1HA to demonstrate membrane association. P, pellet; S, supernatant.

Direct immunofluorescence microscopy was performed to establish the subcellular location of Zeo1p. A functional Zeo1pGFP fusion protein was constructed by inserting GFP immediately downstream of the ZEO1 start codon. Examination of the fluorescing cells maintaining centromeric Zeo1pGFP revealed Zeo1pGFP distribution to be mainly confined to the cell periphery (Fig. 3a). Since Zeo1p is a soluble protein with peripheral membrane association, it appears that this association is with the plasma membrane. Given the two-hybrid interaction with Mid2p, this localization could be, at least in part, due to association with Mid2p. However, examination of both mid2Δ and mtl1Δ cells maintaining centromeric Zeo1pGFP showed the same peripheral pattern of staining (not shown).

A zeo1Δ mutant is resistant to calcofluor white, and MID2-induced hypersensitivity to calcofluor white is Zeo1p-dependent
Chitin is vital to maintain cell wall integrity. Calcofluor white is a fluorescent dye that intercalates with nascent chitin chains, preventing microfibril assembly (Elorza et al., 1983) and inhibiting growth. Previously, we demonstrated that mid2Δ cells are resistant to calcofluor white and that MID2 is required for synthesis of supplemental chitin under conditions of cell stress (Ketela et al., 1999). If Zeo1p provides a structural link in Mid2p-mediated signalling, then zeo1Δ cells should also display a calcofluor white resistance phenotype. The results in Fig. 4(a) show that at a calcofluor white concentration of 15 µg ml^-1 on rich medium, zeo1Δ cells are more resistant to calcofluor white than wild-type cells. Cells harbouring MID2 on a 2 µm plasmid show enhanced sensitivity to calcofluor white, and we tested if ZEO1 was involved in this hypersensitivity. Fig. 4(b) shows that zeo1Δ mutants overexpressing MID2 were less sensitive to calcofluor white than a corresponding wild-type. This attenuation of hypersensitivity to calcofluor white was partial, suggesting the involvement of additional proteins in this process. Resistance to calcofluor white is often associated with defects in chitin synthesis, and this is the case for mid2Δ-associated calcofluor white resistance and MID2-induced hypersensitivity (Ketela et al., 1999). Overexpression of MID2 causes hypersensitivity to calcofluor white by increasing total cell wall chitin levels (Ketela et al., 1999). Since zeo1Δ mutants are able to attenuate MID2-mediated hypersensitivity to calcofluor white, we tested if this was through a reduction in the MID2-associated increase in chitin. However, we found that this increase in chitin was independent of ZEO1.

(43K): Fig. 4. zeo1Δ confers resistance to calcofluor white and is required for MID2-induced hypersensitivity. (a) Mid-exponential-phase cells were diluted to a concentration of 3x106 cells ml-1; 5 µl of this suspension and three subsequent tenfold serial dilutions were spotted onto YEPD plates containing the indicated concentration of calcofluor white and incubated at 30 °C for 3 days. (b) BY4742 cells (WT) containing pRS425 (vector alone) or pRS425MID2, and zeo1Δ cells containing pRS425MID2 or pRS425 (vector alone) were spotted in serial dilution on the indicated medium.

Zeo1p affects basal activity of Mpk1p in a Mid2p-dependent manner
Mid2p is required for activation of the Mpk1 MAPK in response to stress (Ketela et al., 1999; Rajavel et al., 1999). To determine if ZEO1 is needed for this activation, we measured heat-induced activation of Mpk1p in a zeo1Δ strain. While heat-induced phosphorylation of Mpk1p was not reduced in a zeo1Δ strain (not shown), the basal level of Mpk1p phosphorylation was elevated in a zeo1Δ strain relative to wild-type at 22 °C (Fig. 5a). This result suggests a role for Zeo1p in the negative regulation of the cell integrity pathway. Given the relationship between Mid2p and Zeo1p, we asked if the zeo1Δ increase in basal activation of Mpk1p required Mid2p. As shown in Fig. 5(b), a mid2Δzeo1Δ mutant had basal Mpk1p activity, indicating the need for Mid2p in this activation.

(78K): Fig. 5. zeo1Δ mutants show a Mid2p-dependent increase in the basal level of phosphorylation of Mpk1p. (a) Lanes are loaded with equal concentrations of total cellular extracts from wild-type, zeo1Δ and mpk1Δ. (b) Total cellular extracts from wild-type (WT), mid2Δ, zeo1Δ and mid2Δzeo1Δ, all carrying pFL44Mpk1HA. Mpk1p phosphorylation was detected by anti-phospho-p44/42 MAPK (Thr202/Tyr204) (New England Biolabs). Loading control for Mpk1HA is demonstrated by anti-HA antibody HA11.

Mid2p and Zeo1p signal to the cell integrity pathway via the Rho1p GTPase
Mid2p acts on the Rho1p GTPase to activate the PKC1MPK1 cell integrity pathway (Ketela et al., 1999; Sekiya-Kawasaki et al., 2002). To further explore roles for MID2 and ZEO1 in this pathway we investigated the interactions between MID2, ZEO1 and components of the Rho1p GTPase switch. Sac7p has been demonstrated to be a GTPase activating protein for Rho1p (Schmidt et al., 1997). Strains lacking SAC7 have a variety of strain-specific phenotypes, and when grown at 18 °C, sac7Δ mutants grow slowly and have abnormal actin assembly (Schmidt et al., 1997). A sac7Δmid2Δ mutant was constructed and at 22 °C all cells exhibited normal growth. However, at 18 °C, while sac7Δ cells had reduced viability, sac7Δmid2Δ double mutants grew to wild-type levels (Fig. 6a). Restoration of low-temperature growth in sac7Δ cells can also be achieved by deletion of ROM2 (Schmidt et al., 1997). Conversely, overexpression of MID2 in a sac7Δ strain leads to inviability (Fig. 6b). In contrast, ZEO1 appears to play an opposing role to MID2 in a genetic interaction with Sac7p (Fig. 6c), where a zeo1 mutant exacerbates the slow growth of a sac7 mutant at 37 °C. High-copy expression of ZEO1 does not affect growth of a sac7Δ mutant, nor does it suppress the 37 °C growth phenotype (data not shown). Since SAC7 is required to facilitate the removal of GTP from Rho1p and in effect render the protein inactive, Rho1p is in a constitutively active GTP-bound state in a sac7Δ strain, and contributes to cell inviability (Schmidt et al., 1997). Taken together our observations suggest that Mid2p and Zeo1p have opposing actions on Sac7p activity in vivo.

(66K): Fig. 6. MID2 and ZEO1 genetically interact with SAC7. (a) Serial dilutions of wild-type (WT) BY4742, mid2Δ, sac7Δ and mid2Δsac7Δ mid-exponential phase cells were spotted onto YPD and incubated at room temperature and 18 °C for 35 days. (b) Wild-type (WT) BY4742 cells and sac7Δ cells containing either pYES2 vector or pYES2MID2 were streaked onto selective media containing either glucose or galactose and incubated at 30 °C for 23 days. (c) Serial dilutions of wild-type (WT) BY4742, zeo1Δ, sac7Δ and zeo1Δsac7Δ mid-exponential-phase cells were spotted onto YPD medium and incubated at 22 °C or 37 °C for 23 days.

The positive regulation of Rho1p is provided by the GDPGTP exchange factors Rom1p and Rom2p (Ozaki et al., 1996). Although cells lacking ROM1 do not display obvious phenotypes, rom2Δ mutants grow slowly, are prone to lysis at high temperature (Ozaki et al., 1996), and display abnormal mating projection morphology (Manning et al., 1997), suggesting that Rom2p is the major Rho1p GEF in yeast. Since Mid2p and Zeo1p have opposing roles in their effect on sac7Δ cells, and since the mid2rom2 double mutant is synthetically lethal (Table 3), we investigated the interaction of Zeo1p with Rom2p. Results of tetrad analyses generating zeo1Δrom2Δ and mid2Δrom2Δ double mutants are shown in Fig. 7(a), with the zeo1Δrom2Δ mutant largely suppressing the slow growth of a rom2Δ single mutant while the mid2Δrom2Δ mutant is inviable. These findings are consistent with Mid2p and Zeo1p playing reciprocal roles in signalling to Rho1p. As shown in Fig. 7(b), MID2 expression from a high-copy vector is able to suppress the temperature-sensitive phenotype of a rom2Δ mutant, implying that MID2 can signal independently of Rom2p.

(49K): Fig. 7. MID2 and ZEO1 genetically interact with ROM2. (a) Tetratype tetrad from HAB970 (mid2Δ/MID2 rom2Δ/ROM2) and HAB974 (zeo1Δ/ZEO1 rom2Δ/ROM2) heterozygous diploids grown on YEPD at 30 °C for 23 days. (b) Tenfold serial dilutions of wild-type (WT) BY4742 and rom2Δ strains carrying vector only (pRS426) or pRS426MID2 grown on YEPD and incubated at the indicated temperatures for 23 days.

Our analysis of the synthetic interactions of Mid2p extends recent work (Sekiya-Kawasaki et al., 2002) and indicates that this sensor acts via Rho1p on the PKC1MPK1 cell integrity pathway. The synthetic interactions seen with MID2 and WSC1, ROM2, FKS1, SMI1 and FPS1 show that this network upstream of the cell integrity pathway acts to activate 1,3-β-glucan synthesis through Rho1pFks1p in a way that is Mid2p-independent. Further, the synthetic interactions seen with MID2 and the 1,6-α-mannosyltransferase encoding VAN1 and the cell wall mannoprotein encoding CCW12 suggest that Rho1p may also activate mannoprotein synthesis in a Mid2p-independent fashion. Thus, Mid2p and Wsc1p show distinct signalling patterns, with Wsc1p acting through Rom2p to effect its action (Philip & Levin, 2001). While two-hybrid evidence links Mid2p and Rom2p, our genetic evidence and that of Sekiya-Kawasaki et al. (2002) show that much of Mid2p action is Rom2p-independent. To address the nature of Mid2p signalling we looked for proteins interacting with the cytoplasmic tail of Mid2p; we identified Zeo1p in a two-hybrid screen, and showed that Zeo1p acts to negatively regulate the cell integrity pathway in a Mid2p-dependent fashion.

Zeo1p is peripherally associated with the plasma membrane
Zeo1p is present in extracts in both soluble and insoluble forms and localizes largely to the cell periphery. Although it lacks a transmembrane domain, Zeo1p localizes to the cell surface, presumably through proteinprotein interactions or interactions with the inner surface of the plasma membrane. As Zeo1p localization is not dependent on the presence of either Mid2p or Mtl1p alone, it may bind to both of these plasma membrane proteins, and possibly to others. Null mutants of ZEO1 resemble mid2Δ mutants in being resistant to calcofluor white, with the mid2Δzeo1Δ double mutant being no more resistant to calcofluor white than either single mutant, consistent with their operating in a single pathway. The ZEO1 gene is required for the high-copy MID2-induced hypersensitivity to calcofluor white, providing further evidence of a biological link between Zeo1p and Mid2p. Since high-copy expression of ZEO1 does not induce hypersensitivity to calcofluor white, it appears that Mid2p requires Zeo1p, but that Zeo1p itself cannot transduce this signal. As the suppression of MID2-induced hypersensitivity to calcofluor white by zeo1Δ is only partial, there may be other proteins involved in transducing this signal.

Previous work has reported that zeo1Δ mutants grow slowly on galactose, and that high-copy ZEO1 expression leads to Zeocin resistance (MIPS code: HRB113). Zeocin is a copper-chelated glycopeptide antibiotic of the bleomycin family. Although bleomycin antibiotics perturb the plasma membrane of cells, their activity is believed to be primarily due to their ability to bind and degrade DNA. The ble gene isolated from the bacterium Streptoalloteichus hindustanus encodes a 14 kDa protein that has the ability to bind Zeocin and inhibit its DNA strand cleavage activity (Drocourt et al., 1990; Calmels et al., 1991). Zeo1p has no similarity to the ble gene product and it is unclear how ZEO1 confers resistance to Zeocin in Saccharomyces cerevisiae. A number of genomic expression studies show ZEO1 transcriptional activity to be frequently altered under cell stress (Gasch et al., 2000; Posas et al., 2000). For example, ZEO1 transcription is increased fourfold in HOG1-deleted cells exposed to 0·4 M NaCl for 10 min (Posas et al., 2000). These experiments implicate ZEO1 in the response of S. cerevisiae to cellular stress.

Zeo1p affects Mpk1p phosphorylation in a Mid2p-dependent manner
Phosphorylation of Mpk1p in response to stress is partially dependent on Mid2p (Ketela et al., 1999; Rajavel et al., 1999). Although zeo1Δ mutants fail to show a reduction in Mpk1p phosphorylation on heat stress, they do show a constitutive activation of Mpk1p at 22 °C. Thus, Zeo1p may act as a negative regulator for this signalling pathway, with its absence leading to an elevated basal Mpk1p phosphorylation that is Mid2p-dependent. Since the influence of Zeo1p on the calcofluor white phenotype of Mid2p is only partial, other proteins may also modulate these Mid2p responses.

Mid2p and Zeo1p signal the cell integrity pathway in a ROM2-independent manner
In our work several lines of evidence indicate that Mid2pZeo1p signalling to the cell integrity pathway can be independent of Rom2p. The pattern of synthetic interactions of MID2 with WSC1 and its network with ROM2 suggests a non-overlapping function for Mid2p, a finding strengthened by the lack of synthetic lethality between MID2 with BCK1 or MPK1, and indicating that Mid2p acts directly through Rho1p to this cell integrity pathway. In a second line of evidence, the death of mid2Δ cells on exposure to mating pheromone appears to be due to insufficient activation of Rho1p, since this phenotype is largely suppressed by over expression of Rho1p (Ketela et al., 1999). Therefore, if activation of Rho1p requires Rom2p, then rom2Δ mutants should also be sensitive to mating pheromone. Although rom2Δ mutants do exhibit morphological defects in response to mating pheromone, they remain viable (Manning et al., 1997). Further, overexpression of ROM2 is unable to suppress the pheromone-induced death phenotype of a mid2Δ mutant. Finally, high-copy expression of MID2 is able to suppress the high-temperature growth phenotype of a rom2Δ mutant. Since this phenotype is due to a reduced Rho1p GEF activity, Mid2p can thus signal to Rho1p in a Rom2p-independent manner.

Mid2p and Zeo1p appear to play opposing roles in the regulation of Rho1p. Rom2p has been shown to have a major role in activating Rho1p (Ozaki et al., 1996). Indeed, like MID2, RHO1 is capable of suppressing the temperature sensitivity of a rom2 mutant. Further, Rom2p (Ozaki et al., 1996) and Mid2p (Phillip & Levin, 2001) stimulate Rho1p-specific GDPGTP exchange activity. Since the slow growth of rom2Δ mutants is at the level of Rho1p, the synthetic lethality of rom2Δmid2Δ mutants and the suppression of rom2Δ growth defects in a rom2Δzeo1Δ double mutant suggest that Zeo1p and Mid2p play reciprocal roles in stimulating Rho1p. The idea that Mid2p and Zeo1p signal at the Rho1p level is further supported by genetic interactions of Mid2p with the Rho1p GTP-activating protein, Sac7p, where mid2 mutants can rescue the slow growth phenotype of sac7 mutants at 18 °C and overexpression of MID2 is lethal in a sac7Δ background. Genetic interactions between MID2, ZEO1 and components of the Rho1p GDPGTP exchange factor and GTP-activating protein suggest that Mid2p acts positively, while Zeo1p acts as a negative regulator of the pathway. For the Mid2p response to calcofluor white, Zeo1p is a critical component of this signalling pathway. Indeed, Philip & Levin (2001) demonstrated that in response to calcofluor white, GTP loading of Rho1p is increased, suggesting that the Zeo1p control of the Mid2p response to calcofluor white is at the level of the Rho1p GTPase.

Supported by a Discovery Grant from the Natural Sciences and Engineering Research Council of Canada and by Genome Canada and Genome Quebec.