A Genomic Approach for the Identification and Classification of Genes Involved in Cell Wall Formation and its Regulation in Saccharomyces Cerevisiae

Using a hierarchical approach, 620 non-essential single-gene yeast deletants generated by EUROFAN I were systematically screened for cell-wall-related phenotypes. By analyzing for altered sensitivity to the presence of Calcofluor white or SDS in the growth medium, altered sensitivity to sonication, or abnormal morphology, 145 (23%) mutants showing at least one cell wall-related phenotype were selected. These were screened further to identify genes potentially involved in either the biosynthesis, remodeling or coupling of cell wall macromolecules or genes involved in the overall regulation of cell wall construction and to eliminate those genes with a more general, pleiotropic effect. Ninety percent of the mutants selected from the primary tests showed additional cell wall-related phenotypes. When extrapolated to the entire yeast genome, these data indicate that over 1200 genes may directly or indirectly affect cell wall formation and its regulation. Twenty-one mutants with altered levels of β1,3-glucan synthase activity and five Calcofluor white-resistant mutants with altered levels of chitin synthase activities were found, indicating that the corresponding genes affect β1,3-glucan or chitin synthesis. By selecting for increased levels of specific cell wall components in the growth medium, we identified 13 genes that are possibly implicated in different steps of cell wall assembly. Furthermore, 14 mutants showed a constitutive activation of the cell wall integrity pathway, suggesting that they participate in the modulation of the pathway either directly acting as signaling components or by triggering the Slt2-dependent compensatory mechanism. In conclusion, our screening approach represents a comprehensive functional analysis on a genomic scale of gene products involved in various aspects of fungal cell wall formation.


Introduction
The cell wall of S. cerevisiae is an essential organelle whose rigid structure determines cell shape, enables cells to withstand internal turgor pressure and protects cells against environmental stresses. The proteins (CWPs) covalently linked to glucans, namely GPI-CWPs (containing a glycosylphosphatidylinositol-derived structure) and Pir-CWPs (for Proteins with Internal Repeats), have been identified. The composition and structure of the yeast cell wall may vary considerably both during cell cycle progression (Klis, 1994;Cid et al., 1995), or when exposed to changing environmental conditions, e.g. heat stress (Jung and Levin, 1999) or the presence of pheromones (Klis, 1994;Cid et al., 1995). When yeast cells are grown in the presence of cell wall or cell membrane perturbing agents (Calcofluor white, SDS), cell wall construction is also adapted (Ketela et al., 1999). Studies with mutants lacking FKS1 or GAS1, involved in synthesis (Mazur et al., 1995) and remodeling of b1,3-glucan chains (Popolo and Vai, 1999;Mouyna et al., 2000), respectively, showed that in these mutants a compensatory mechanism is induced (Popolo et al., 1997;Dallies et al., 1998;Ram et al., 1998). As a result, chitin levels in the cell wall rise and more cell wall mannoproteins become bound, through b1,6glucan, to chitin (Kapteyn et al., 1997). In addition, the cell walls contain more Cwp1 and more Pir proteins, and transcription of FKS2, which encodes a catalytic subunit of the b1,3-glucan synthase complex, is upregulated Ram et al., 1998).
Partly due to its potential as a selective target for antifungal drugs, a great effort has been made in recent years to characterize genes involved in the construction and maintenance of the yeast cell wall. This has resulted in the cloning and characterization of genes involved in different aspects of cell wall biosynthesis, such as b1,3-glucan synthesis (FKS1, FKS2, Mazur et al., 1995), processing of b1,3-glucan chains (Popolo and Vai, 1999;Rodriguez-Peñ a et al., 2000), b1,6-glucan synthesis (Shahinian and Bussey, 2000) and chitin synthesis (CHS1 to CHS6, reviewed in Bulawa, 1993, andCHS7, Trilla et al., 1999). However, much less is known about genes that encode proteins involved in coupling reactions between cell wall macromolecules. In addition to genes directly involved in cell wall construction, a number of genes that regulate biosynthesis of the cell wall through modulation of the protein kinase C-directed cell wall integrity pathway have been characterized. This pathway, essential for maintaining a stable cell wall, is activated in response to a variety of stress conditions that trigger a compensatory mechanism De Nobel et al., 2000). It has recently been shown that most of the genes controlled by this pathway encode known or putative cell wall proteins and enzymes involved in cell wall biogenesis (Jung and Levin, 1999).
A genomic approach to further extend our knowledge of the cell biology of S. cerevisiae was undertaken by a EUROpean Functional Analysis Network for the yeast genome (EUROFAN). In this program, strains deleted in genes of unknown function generated by EUROFAN I were systematically screened using hierarchical approaches (EUROFAN II). As members of EUROFAN, we have analyzed the collection of deletants to improve our understanding of cell wall biogenesis of S. cerevisiae. For this purpose, in a first round of screens, we selected for mutants that appear to have altered cell walls by analyzing the sensitivity of cells challenged with the cell wall and cell membrane perturbing compounds Calcofluor white and SDS, and a sonication procedure. Additionally, cells were studied by light microscopy for having morphological defects. The mutants thus selected were further analyzed for additional cell wall phenotypes in a hierarchical manner using screens with increasing specificity and discriminatory power. These screens enabled us to identify candidate genes coding for structural cell wall proteins, proteins involved either in the biosynthesis, processing or assembly of the different cell wall macromolecules, and those involved in the regulation of cell wall formation. At the same time, these screens helped us to validate the assays chosen for the primary selection of potential cell wall-related mutants. Our hierarchical large-scale screening approach therefore allows the efficient identification of genes involved in different aspects of cell wall formation and its regulation. We propose that a similar approach will be effective for other fungi as well.

Calcofluor White and SDS spot assays
The spot assay used to screen for hypersensitivity or resistance to Calcofluor White was adapted from the assay described by Ram et al. (1994). Calcofluor white plates were prepared by adding Calcofluor white (from a 1% (w/v) stock solution) to sterile YEPD medium at 70uC to a final concentration of 50 mg/ml. On these plates, 3 ml drops were spotted that contained y10 5 , 10 4 , 10 3 , 10 2 and 10 cells/ml, respectively, and after two days at 30uC growth was scored.
Similar to the Calcofluor white assay, mutants that are sensitive to SDS were identified by adding SDS (400 mg/ml) for TRP1 strains and (12 mg/ml) for trp1 auxotrophs) to the YEPD medium.

Sonication assay
The sonication protocol was adapted from Ruiz et al. (1999). Cells were grown overnight in 5 ml YED cultures. One ml of each cell suspension was sonicated for 30 seconds on ice at a wave amplitude of 2 microns in a Vibra Cell (Sonic & Materials, Connecticut, USA). The percentage of sonicationsensitive cells was determined by the addition of 300 ml 0.005% propidium iodide (PI) to 100 ml of untreated (control) and treated cells, and measuring the number of PI positive cells using a FACStar Plus flow cytometer (Becton Dickinson, San Jose, CA). The relative sonication-sensitivity or resistance of each mutant was determined by calculating the ratio of the percentage of damaged cells of the mutant and that of the isogenic wild type due to sonication. Mutants with a ratioi1.5 were considered sonication-sensitive, while those with a ratio j0.5 were defined as resistant.

Calcofluor White staining and fluorescence microscopy
The protocol was adapted from Pringle (1991) for screening large numbers of strains. Cells from an overnight culture (28uC in YED), were inoculated 1 : 10 in fresh YED medium. To enhance the detection of cell wall-related phenotypes, the cells were incubated for 5 h at 37uC. From each culture a 200 ml sample was taken for centrifugation (1 min) and the supernatant removed. Cells were resuspended in 50 ml of a 10 mg/ml solution of Calcofluor white (Fluorescent Brightener 28, Sigma) and observed using an Olympus IMT-2 fluorescence microscope provided with an Olympus BH2-RFL-T3 lamp, an appropriate set of filters (UV to blue for excitation and blue to green for emission) and a 100r immersion objective.

Immunoblot analysis of Slt2 activation
Slt2 activation was determined basically as described by Martín et al. (2000). Briefly, cells were grown to mid-log phase, diluted to A 600 =0.2 and grown for one generation at 24uC or 39uC. After cell breakage, cell extracts were separated from glass beads and cell debris by centrifugation and collected in new tubes. Protein concentrations (A 280 ) were determined and 80 mg protein samples were fractionated by SDS-PAGE and analysed by Western blotting. Membranes were probed with anti-phospho-p44/42 MAPK (Thr202/Tyr204) antibodies (New England Biolabs) to detect dually phosphorylated Slt2, and then stripped and reprobed with anti-GST-Slt2 antibodies (Martín et al., 1993) in order to monitor the total amount of Slt2 in each lane.
(Sigma) and YED+12 mM caffeine+0.5 M sorbitol (Merck) plates. Growth was scored after two days at 28uC for YED plates and after three days for the caffeine-containing ones. Strains that showed sensitivity to caffeine were further studied by spotting serial tenfold dilutions (y10 4 , 10 3 , 10 2 and 10 cells).
Calcium chloride and lithium chloride spot assays Cells suspensions as described above for the caffeine sensitivity assay were spotted on YED and YED+0.2 M CaCl 2 plates. Growth was scored after two days at 28uC. Sensitivity to 0.2 M LiCl was evaluated by an equivalent procedure in order to discern strains that are calcium-specific from those that are sensitive to high ionic conditions. Strains that showed sensitivity to calcium and not to lithium were further studied by spotting serial tenfold dilutions (y10 4 , 10 3 , 10 2 and 10 cells).

Zymolyase assay
Sensitivity to Zymolyase was determined based on the method described by Lussier et al. (1997). Cells from strains grown overnight in YEPD were washed in water and resuspended in 10 mM Tris-HCl pH 7.4, at a concentration of 10 7 cells/ml. Zymolyase-20T (Seikagaku Corporation, Tokyo, Japan) was added at a concentration of 5 mg/ml and the cell density at the start of the incubation was measured as A 600 . Incubations were performed for 4 h at 37uC and A 600 was read at one-hour intervals to make sure that the decrease in cell density was in the linear range. Mutants were considered sensitive to Zymolyase when the A 600 after a 4 h incubation was 70% or less compared to that of the FY1679a strain and resistant when it was 130% or higher.

Killer toxin assays
Sensitivities to killer toxins K1, K28 and HM-1 were determined using a seeded plate assay (Boone et al., 1990). Killer toxin-producing strains were grown overnight on YEPD plates at 30uC and cells from these plates were resuspended in water to obtain concentrated cell suspensions. Strains to be tested were grown overnight in 5 ml YEPD, washed with sterile water and resuspended at a cell density of A 600 =0.5. Seeded plate medium consisted of 20 ml YEPD-agar supplemented with 50 mM sodium citrate buffer (pH 3.7-3.8) and 30 mg/ml methylene blue. This medium was inoculated with 100 ml cell suspension of strains to be tested and the plates were allowed to dry completely before spotting 3 ml of the killer toxin producing strains. For each deletion strain to be tested two plates were prepared, one of which was incubated at 22uC and another at 30uC for two to three days. Resistance/ sensitivity was determined by comparing the growth inhibition halos of mutants with the halo of the haploid reference strain FY1679a.

Cyclosporin sensitivity assay
Cells were grown overnight at 28uC in YED medium and 200 ml cell suspensions containing 10 3 cells were deposited in wells of 96-well microtiter plates. For each mutant three conditions were used; YED medium, YED+50 mg/ml Cyclosporin A (Sandoz Pharma AG, Basel) and YED+100 mg/ml Cyclosporin A. Plates were incubated one day at 28uC and optical density was measured to determine the level of growth.

Papulacandin B and Echinocandin B assays
Sensitivities to the glucan synthase inhibitors Papulacandin B (Ciba-Geigy, Basel, Switzerland) and Echinocandin B (Eli Lilly, Indianapolis, USA) were determined by spotting serial dilutions of cell suspensions, as described for the Calcofluor white spot-assay, on YEPD plates containing the inhibitors. The concentrations used were 1 mg/ml Papulacandin B and 0.2 mg/ml Echinocandin B, respectively.
In vitro b1,3-Glucan synthase and chitin synthase activities For determination of b1,3-glucan synthase and chitin synthase activities, cells grown to midlogarithmic-phase were homogenized and the membrane fraction was isolated by centrifugation at 48.000rg for 30 min. ChsI, ChsII, and ChsIII activity measurements on the membrane fractions were performed as described by Choi and Cabib (1994), and b1,3-glucan synthase activity was determined as described by Ishiguro et al. (1997) Functional analysis of cell wall mutants 127 either with 150 mM GTP or without GTP in the assay mixture. Glucan and chitin synthase activities were defined as the amount of glucose and N-acetylglucosamine incorporated into glucan and chitin, repectively. Enzyme activities of deletion mutants per milligram of protein were compared to the wild type strain FY1679a and were considered as being increased if >120% and decreased if <80% compared to wild-type.

Immunoblot analysis of culture supernatant
Yeast strains were grown in SC to mid-log phase (A 600 y2). Cells and culture media were carefully separated by centrifugation (2r10 min at 1,600rg). Culture supernatants of the different strains were diluted to a concentration equivalent to A 600 =1.6 with 10 mM Tris-HCl, pH 6.8, and from these solutions, two-fold serial dilutions were prepared. These dilutions were spotted onto Immobilon-NC (Millipore) using a Bio-Dot apparatus (Biorad). For Western analysis using b1,3-glucan antibodies, Immobilon-P (Millipore) instead of Immobilon-NC was preferred. Spots were allowed to dry overnight. Western immunoblot analyses were performed according to Klis et al. (1998) using monoclonal antibodies against b1,3-glucan (Biosupplies Australia) and polyclonal antisera raised against b1,6-glucan-BSA (Kapteyn et al., 1995), Ssr1 (Moukadiri et al., 1997), Cwp1 (Shimoi et al., 1995), and Pir2/Hsp150 (Russo et al., 1992). Western blots were visualized with ECL Western blotting detection reagents (Amersham) according to the manufacturer's instructions. Signal intensities of spots were determined by densitometric scanning of spots in the linear range of the X-ray films. A mutant was considered to have increased levels of a certain cell wall component in the medium when the amount was i1.7r the wild type level. For SDS-PAGE analysis of medium proteins, proteins were precipitated overnight in 80% cold ethanol and washed twice with 80% acetone. Precipitated proteins corresponding to the equivalent of 200 ml culture supernatant of cells grown to A 600 =0 2 were separated by electrophoresis using linear 2.2-20% polyacrylamide gels and electrophoretically transferred onto Immobilon polyvinylidene (PVDF) membranes (Montijn et al., 1994). Immunoblot analyses were performed as described above.

General approach
Deletion mutants generated by EUROFAN I (Oliver, 1996) were systematically screened for cell wall-related phenotypes. Efficient analysis of the whole collection of EUROFAN I mutants was achieved by using a hierarchical screening approach (Figure 1). This method employed rapid screening of all mutants with easy-to-perform assays in order to select potential cell wall-related genes. To elucidate which aspect of cell wall formation was affected, the mutants that scored positively were analyzed further using tests of increasing specificity. The detailed results of our screens are compiled in an EUROFAN database that is accessible at: http:// www.mips.biochem.mpg.de/proj/eurofan/eurofan_2/n7/ index.html

Primary screens for the identification of cell wall-related genes
For primary identification of potential cell wallrelated genes, the 620 EUROFAN mutants, each deleted in an individual ORF, were analyzed using phenotypic assays which are indicative of mutations leading to a defective cell wall. Our primary tests scored for altered sensitivity to Calcofluor White, SDS, sonication, and for abnormal morphologies. Calcofluor White is a fluorescent agent that binds to chitin and interferes with the polymerization of this cell wall component (Elorza et al., 1983;Roncero and Durá n, 1985). For this reason, it has been widely used for chitin staining (Pringle et al., 1989) but also for the identification of cell wall-related mutants after random mutagenesis (Roncero et al., 1988;Ram et al., 1994;Lussier et al., 1997). SDS is a detergent that affects membrane stability and indirectly also cell wall construction. It can thus be used to reveal cell wall defects that result in increased accessibility of SDS to the plasma membrane (Shimizu et al., 1994;Igual et al., 1996;Bickle et al., 1998). The use of sonication is also a powerful tool for the identification of cell wall alterations; in addition, a procedure applicable at large scale has been developed (Ruiz et al., 1999). Finally, the cell wall is responsible for maintaining cell shape and, therefore, morphological abnormalities may be indicative of alterations in cell wall dynamics during morphogenesis. Microscopic observation of cells stained with Calcofluor white allowed the identification of mutants showing different chitin deposition patterns and other alterations related to morphogenesis, such as abnormal septation, hyperpolarized growth or abnormal budding patterns. Examples of mutants displaying a variety of morphogenetic defects are shown in Figure 2.
In total, among the 620 mutants that were analyzed, 145 (23%) mutants showed phenotypic abnormalities in at least one of the screens used ( Table 1). The number of mutants selected by each primary screen ranged from 51 mutants (8.2%) showing abnormal morphology to 90 mutants (14.5%) with altered sensitivity to Calcofluor white. The sets of mutants obtained with the single screens showed limited overlap. For example, 35% and 39% of the sonication-sensitive mutants showed altered sensitivity to SDS and Calcofluor, respectively, and indeed, only six of the 145 mutants selected by the primary screens were positive in all four primary screens. This indicates that our primary screens are largely independent from each other. More extensive overlap was found only in one case: 63% of the sonication-sensitive mutants had an abnormal shape, supporting the idea that mutants affected in morphogenetic processes have defects at particular sites in the cell wall which might lead to loss of cell integrity after sonication (Cid et al., 1998;Ruiz et al., 1999).

Specific screens for the identification of proteins involved in cell wall biosynthetic reactions
In a second round of screens, the 145 mutants selected by primary analysis were further analyzed using screens of higher specificity in order to discriminate between mutants related to different aspects of cell wall formation and mutants that present a more general phenotype ( Figure 1). We have not screened for mannosylation and phosphorylation defects of cell wall proteins. For this we refer to the results of the Eurofan II Secretion and Protein Trafficking Node at MIPS http:// www.mips.biochem.mpg.de/proj/eurofan/eurofan_2/n5/ index.html.
To identify genes encoding proteins belonging to cell wall biosynthetic enzyme complexes or proteins involved in the modulation of the synthetic activities, we analyzed the 145 selected mutants for their sensitivities to the b1,3-glucan synthase inhibitors Papulacandin B and Echinocandin B, to the killer toxins K1, K28, and HM-1, which bind to specific cell wall components, to Zymolyase, an enzyme cocktail comprising b1,3-glucanase and protease activities, and to the calcineurin inhibitor Cyclosporin A. Papulacandin B and Echinocandin B act either by hindering some components of the b1,3glucan synthase complex or by inhibiting the incorporation of the glucans into the extracellular matrix (Font de Mora et al., 1993;Ram et al., 1994). Interestingly, although resistance or sensitivity to both inhibitors are usually correlated, we found a group of deletants that are sensitive to one of these drugs and behave as wild type for the other, pointing out that the mechanism of action of both compounds is different (Table 2). Increased resistance of mutant cells to the killer toxin K1 has often been found to correlate with low levels of b1,6-glucan, the receptor molecule for this toxin, in the wall (Shahinian and Bussey, 2000). Similarly, low levels of mannan are expected to correlate with increased resistance to killer toxin K28 (Schmitt and Radler, 1987) and low levels of b1,3-glucan with increased resistance to killer toxin HM-1   (131) nd, not determined. Six hundred and twenty deletion mutants were screened. Deletants that scored positively in one or more of the primary screens (145 strains), were further analyzed ( Figure 1). The number of mutants that also scored positively in one or more secondary tests are presented between parentheses. On average, 90% of the positives in the primary screens scored also positively in one or more of the secondary screens. The data for each mutant can be found at MIPS: http://www.mips. biochem.mpg.de/proj/eurofan/eurofan_2/n7/index.html. 130 P. W. J. de Groot et al. (Kasahara et al., 1994). Altered Zymolyase sensitivity may reflect changes in the b1,3-glucan layer (Ovalle et al., 1998) or changes in the external mannoprotein layer that result in altered permeability to cell wall degrading enzymes (De Nobel et al., 1991). Increased sensitivity to K1, K28, HM-1 and Zymolyase was found for 29, 50, 76 and 42 mutants, respectively whereas increased resistance to K1, HM-1 and Zymolyase was found for 17, 8 and 2 mutants respectively. For the K28 toxin, we only scored for increased sensitivity because the growth inhibition halo of the wild type strain was rather small. Of the K1-hypersensitive mutants, all except one were also hypersensitive to the HM-1 toxin, indicating that the amount of b1,6-glucan that will be incorporated in the cell wall strongly depends on the amount of b1,3-glucan present in the wall. This was expected since the b1,6-glucan in the wall is covalently linked to the b1,3-glucan network. Twenty-nine mutants were hypersensitive to both Zymolyase and killer toxin HM-1, indicating that these mutants at least seem to have alterations in their b1,3-glucan layer. It is well known that Cyclosporin A functions as a calcineurin inhibitor that negatively affects FKS2 expression (Zhao et al., 1998), and sensitivity to this drug is therefore related to glucan synthesis. FKS2 codes for one of the subunits of the glucan synthase complex whose expression is increased in the absence of Fks1 function (Mazur et al., 1995;Ram et al., 1998). Sensitivity of fks1D mutants to Cyclosporin A is therefore due to its effect on FKS2 expression. We detected only one mutant, yjl029cD, that was hypersensitive to this drug. This mutant also showed diminished glucan synthase activity (see below and Table 2), supporting the idea that the activity of Yjl029c is directly related to glucan biogenesis.
The tests described above indicated that among the 145 selected mutants, a group of mutants may

Functional analysis of cell wall mutants 131
be affected in the synthesis of b1,3 glucan, b1,6 glucan or mannan. To obtain more direct evidence that some of the proteins are involved in the activity of the b1,3 glucan synthase complex, we determined the in vitro activity of this enzyme in 50 mutants selected on the basis of results obtained in the killer, Zymolyase or Papulacandin B/Echinocandin B assays ( Table 2). 21 of these mutants have significantly altered levels of b1,3-glucan synthase activity, either lowered or increased, compared to the wild type strain, indicating that in these mutants the deletion involved a protein that is required for normal b1,3-glucan synthase activity. For each mutant the relative glucan synthase activities in the absence and presence of GTP correlated (not shown), indicating that in none of the deletants was the activation of b1,3-glucan synthase by GTP affected. 13 of the 14 mutants with reduced glucan synthase activity were hypersensitive to the HM-1 killer toxin and 11 of them also showed hypersensitivity towards K28 toxin. This suggests that the increased killer sensitivity in these mutants may be due to their lower levels of b1,3-glucan and/or probably altered levels of mannoproteins in the cell wall. In seven mutants, an increase of b1,3-glucan synthase activity was detected suggesting that perhaps proteins involved in negative regulation of b1,3-glucan synthesis are affected. 14 of the 21 mutants with altered b1,3-glucan synthase activity are deleted in currently identified genes. Among these is a group of genes that is responsible for a wide range of functions in the cell (such as NCL1, MED2, SEH1, ELP2, CTK2 or PTK2) suggesting that the observed effects on b1,3-glucan synthase activity may be indirect. Others perform cellular functions that are more directly related to cell wall construction, morphogenesis or secretion (VAM6, VPS53, ARP5, INP52, TLG2), but defects in b1,3glucan synthase have not been previously reported for those mutants.
In the Calcofluor white spot-assay, 21 mutants appeared to be less sensitive than the wild type strain suggesting a possible role in chitin synthesis (Roncero et al., 1988). We examined the sensitivity of these mutants to Calcofluor white in concentrations up to 500 mg/ml. Five of the mutants, for which resistance to Calcofluor white had not been reported before, could tolerate this concentration and these were therefore assayed for ChsI, ChsII and ChsIII activities (Figure 3). Chitin synthase III activity is responsible for more than 90% of the chitin in the cell wall. ChsIII is downregulated in yjl046wD, which lacks a putative protein with similarity to Caenorhabditis elegans lipoate-protein ligase, and in yor275cD, which lacks a putative protein with SH3 domain-binding motifs and has similarity with the pH signal transduction pathway gene palA of Aspergillus nidulans. In both mutants in vitro ChsI activity is also affected. Upregulation of chitin synthase III activity was found in the mutant ynl294D, which lacks an ORF containing six putative transmembrane domains and also showed a decrease in ChsI activity. Surprisingly, only this mutant showed increased sensitivity to Zymolyase, a phenotype previously observed in mutants lacking ChsIII activity (Bulawa, 1993). Thus, a clear relation between Zymolyase sensitivity, Calcofluor white resistance and chitin synthase activities could not be detected in this group of mutants. Mutants ynl058cD and ynl233wD (bni4D) had normal levels of ChsI and ChsIII activities, but in both mutants increased levels of ChsII activity were found. However, the antifungal effect of Calcofluor not only depends on its binding to cell wall chitin but also on the presence of a functional HOG pathway (Garcia-Rodriguez et al., 2000). In other words, Calcofluor resistance may also be caused by the inability to respond to Calcofluor treatment.

Secondary screens to identify potential assembly enzymes
The various components of the cell wall are synthesized individually and are to a large extent processed and covalently coupled to each other at the cell surface ( Figure 4A). To identify enzymes involved in processing and assembly reactions, we focused on potential cell wall-related proteins that are known to be located at the cell surface (Yeast Proteome Database: http://www.proteome.com, Costanzo et al., 2000), proteins that show significant identity with known cell surface proteins (BLAST, http://www.ncbi.nlm.nih.gov/BLAST) and proteins with sequence features that are predictive of cell surface proteins (N-terminal signal sequence and/or transmembrane domains or a C-terminal GPI-anchor in combination with absence of retention signals) according to PSORT II (http://psort. nibb.ac.jp) analysis (Figure 1). 56 of the initially selected mutants were defective in proteins that met these criteria. To further select for mutants affected in the final cell wall construction steps, we anticipated that such mutants would secrete increased levels of cell wall components in the growth medium. Analysis of culture supernatants was performed by dot-blot immunoanalysis using antibodies directed against b1,3-glucan, b1,6-glucan (Kapteyn et al., 1995), the GPI-CWPs Cwp1 (Shimoi et al., 1995) and Ssr1 (Moukadiri et al., 1997), and the Pir-CWP Pir2 (Russo et al., 1992). An example of such an analysis is presented in Figure 4B, showing mutants yjl062wD (gpi7D/ las21D) and ydl231cD with significantly increased levels and ybr078wD (ecm33D) with a slightly decreased level of Ssr1, respectively, compared to the wild type. We found that for the wild type strain the levels of b1,3-glucan, b1,6-glucan and Cwp1 are (4) In wild type cells, a minor part of the chitin in the cell wall is bound to non-reducing ends of b1,6-glucan. (5) Pir-CWPs are linked to the b1,3-glucan chains without interconnecting b1,6-glucan but the precise linkage of Pir-CWPs to the cell wall network is unknown. (B) Dot-blot immunoanalysis of culture supernatants of putative cell wallrelated mutants using anti-Ssr1 antibodies. Two-fold serial dilutions of culture supernatants were spotted; the first spots correspond to the equivalent of 400 ml culture supernatants of cells grown to A 600 =2. (C) SDS-PAGE immunoblot analysis of precipitated proteins of culture supernatants of putative cell wall-related mutants with b-1,6-glucan antiserum, anti-Cwp1 antiserum, anti-Ssr1-antiserum and anti-Pir2 antiserum. Strains were grown to A 600 =2 in SC medium and precipitated proteins of 200 ml culture supernatant were loaded per lane. 1, wild type strain FY1679a; 2, ybr078wD (ecm33D); 3, yjl062wD (gpi7D/las21D); 4, ydl231cD Functional analysis of cell wall mutants 133 rather low and just above the detection limit of our immunoanalysis. We therefore focused this analysis on the identification of mutant strains that showed significant increases, rather than decreases, of cell wall components in the culture supernatant. Table 3 shows the results of this analysis for ORFs whose deletion caused an increase in the release of cell wall components to the medium suggesting a role for them in cell wall construction. Six mutants showed increased signal intensities upon analysis of culture supernatants with anti-Cwp1 and anti-Ssr1 antibodies, four of them also showing an increase upon analysis with anti-Pir2 antiserum. Some correlation between increased secretion of Cwp1 and Ssr1 in cell wall mutants was expected since these proteins both belong to the group of GPI-CWPs. On the other hand, induced expression of Cwp1, and not Ssr1, is known to be a phenotypic trait of mutants having a compromised cell wall structure . Pir2 is linked to b1,3-glucan independent of GPI-structures or b1,6-glucan ( Figure 4A). Elevated release of Pir2 and GPI-proteins into the growth medium suggests that these mutants might have a defect in linking cell wall components to b1,3glucan caused by a defect in either an intermolecular coupling step or remodeling of b1,3-glucan. Processing of linear b1,3-glucan is believed to create a branched molecule with acceptor sites for b1,6glucan, chitin and Pir-proteins (Smits et al., 1999). Because GPI-proteins are linked to the b1,3-glucan network via b1,6-glucan, mutants defective in the branching of b1,3-glucan are expected to also have increased levels of b1,6-glucan in the growth medium. One of the mutants displaying increased release of the three analyzed CWPs, ydl231cD, deleted in an ORF specifying an unidentified putative membrane protein, indeed shows this phenotype. Alternatively, a defect in the incorporation of GPI-proteins might induce increased production of Pir proteins to compensate for lower levels of GPI-proteins in the wall. Inefficient incorporation of this Pir2 in the wall might explain the increased levels of Pir2 in the medium. This phenotype is seen in mutants yjl062wD, ybr183wD (ypc1D) and ynl080cD. We found two genes in whose absence the culture medium became specifically enriched in Pir-cell wall proteins, suggesting Determined by Western analysis using anti-b1,3-glucan (3G, only on dot-blots); anti-b1,6-glucan (6G), anti-Cwp1, anti-Ssr1 and anti-Pir2 antibodies. +, increased level of cell wall component in the culture solution of mutant relative to FY1679a. 4 From YPD or MIPS, or 5 predicted from present data.

134
P. W. J. de Groot et al.
that these genes encode proteins that have a role in coupling Pir proteins to the b1,3-glucan network ( Figure 4A). Yol092w is a putative transmembrane protein that belongs to a family of three members, whereas Ynl294c has no homolog in the yeast genome. In the absence of Ybr255w, which is predicted to be associated with the membrane of the endoplasmic reticulum, the levels of b1,6-glucan in the culture supernatant are increased. However the deletant does not become resistant to K1 toxin suggesting that the b1,6-glucan levels of the wall have not decreased. Possibly, the synthesis of b1,6glucan is upregulated in this mutant. Surprisingly, the increased level of b1,6-glucan in the culture supernatant of this mutant is not accompanied by a significant increase in the GPI-cell wall proteins Cwp1 and Ssr1. This was confirmed using SDS-PAGE analysis of precipitated proteins (see below) and can not be ascribed to the presence of free b1,6glucan in the growth medium. Likewise, mutants ecm33D and ynl159cD also have increased levels of protein-bound b1,6-glucan but no increase of Cwp1 and Ssr1, in their culture supernatants. Possibly, the bulk of b1,6-glucan in the culture supernatant of these mutants is bound to GPI-proteins other than Cwp1 and Ssr1.
Mutants showing higher levels of cell wall components in the growth medium were further analyzed by SDS-PAGE and immunodetection of proteins precipitated from the culture supernatant. For the b1,6-glucan, Cwp1, Ssr1 and Pir2 levels, this analysis generally confirmed our results obtained with the dot-blot analysis. This is exemplified in Figure 4C, showing immunoblots of three mutants, ybr078wD, yjl062wD and ydl231cD, that appear to have increased levels of different cell wall components in the growth medium compared to the wild type, as determined by dot-blot analysis (Table 3). Mutant yjl062wD releases high amounts of GPI-proteins into the culture medium whereas the amount of b1,6-glucan is similar to that of the wild type. This suggests that in this mutant coupling of GPI-proteins to b1,6-glucan is affected, which is consistent with recent studies that showed that Gpi7/Las21, like Mcd4, is involved in the addition of ethanolaminephosphate onto the core structure of GPI anchors (Flury et al., 2000). Furthermore, a significant increase of Pir2 in the culture supernatant of yjl062wD was found which might be explained by increased expression of Pir-proteins as a compensation for the low amount of GPI-proteins in the cell wall. In some of the mutants, analysis with anti-Pir2 antibodies revealed that in addition to Pir2 (t220 kDa) a weak signal at about 40 kDa was detected. This protein may correspond to Pir4/Cis3 which immunoreacted with polyclonal anti-Pir2 antibodies . For ydl231cD, an additional protein band of about 60 kDa is observed, that is postulated to be a processed form of the Pir2 protein .
None of the mutants showing immunoreactivity in the dot-blots with the anti-b1,3-glucan antibodies (Table 3) gave rise to clear signals by immunoanalysis of precipitated medium proteins (not shown). This indicates that b1,3-glucan molecules detected in the medium are not linked to protein.

Secondary screens to identify potential cell wall regulatory proteins
To identify proteins directly affecting the cell wall integrity pathway and proteins that are otherwise involved in regulation of cell wall formation, the 145 initially selected mutants were analyzed for sensitivity to calcium, sensitivity to caffeine, osmotic remediability of caffeine hypersensitivity using sorbitol, and Slt2/Mpk1 MAP kinase activation (Figure 1). Intracellular calcium levels are thought to influence cell morphogenesis and cell wall biosynthesis in several ways. For instance, calcium is likely to be involved in the regulation of actin-dependent morphogenesis via Cdc24 (Miyamoto et al., 1991), in polarized secretion via calmodulin (Peters and Mayer, 1998) and in regulation of glucan synthesis via calcineurin (Zhao et al., 1998). We therefore speculated that many of the mutants might be hypersensitive to the presence of calcium in the growth medium. The results of the calcium sensitivity test show that 24% of the mutants analyzed were indeed sensitive to calcium (but not pleiotropically hypersensitive to high salinity). Conversely, most of the calcium-hypersensitive mutants (92%) had additional phenotypes related to defects in cell wall formation. There does not seem to be a strong correlation between altered morphology and calcium hypersensitivity since only 22% of the morphological mutants were hypersensitive to calcium. However, a high proportion of mutants with decreased glucan synthase activity were also more sensitive to calcium (Table 2).

Functional analysis of cell wall mutants 135
Caffeine, besides being a well-known inhibitor of cAMP phosphodiesterase, has also been found to stimulate dual phosphorylation of Slt2, the MAP kinase component of the cell wall integrity signal transduction pathway . Mutants involved in the cell wall integrity pathway are more sensitive to caffeine, displaying a lytic phenotype in the presence of this compound that can be prevented by osmotic stabilization (Martín et al., 1996). 71 of the 145 analyzed mutants appeared caffeine-sensitive but only 17 of them were rescued by the addition of sorbitol suggesting that these osmotically-remedial mutants have cell wall defects that are due to a failure in the regulation of the cell wall integrity by this pathway. Eleven of the seventeen osmotically stabilized mutants had altered morphologies (Table 4), which is consistent with the notion that the cell wall integrity pathway is involved in coordinating cell wall synthesis with the cell cycle and morphogenetic events (Igual et al., 1996;Gray et al., 1997).
More direct information on activation of the Pkc1-Slt2-pathway was obtained by measuring the levels of the activated form of the MAP kinase Slt2. Mutations in genes involved in positive modulation of the pathway are expected to fail in activating the pathway whereas the absence of negative modulators should result in activation of the pathway even in the absence of stimulating signals. On the other hand, this pathway is activated in response to environmental conditions that jeopardize cell wall stability in order to induce a cell wall compensation mechanism (Jung and Levin, 1999;De Nobel et al., 2000). Constitutive activation of Slt2 has been also seen in mutants affected in cell wall functions, such as kre9D, gas1D and fks1D, in response to the cell wall defects displayed by these mutants (De Nobel et al., 2000). The 145 initially selected mutants were analyzed by Western analysis using antibodies raised against the dually phosphorylated region (Thr202/Tyr204) of p44/42 MAP kinases, which have been shown to recognize the active form of the yeast Slt2 MAP kinase . Compared to wild-type, fourteen mutants showed elevated levels of Slt2 phosphorylation at 24uC, indicative of proteins that directly regulate the activity of the pathway or proteins whose absence triggers a constitutive activation of the cell wall integrity pathway as a consequence of a weakened cell wall (Table 5). Interestingly, among these mutants are mutants that are deleted in ORFs specifying putative transcription factors, typical signaling molecules (a GTP-binding protein and a protein kinase), chaperones and proteins involved in cell wall maintenance. In wild-type cells that are incubated at 39uC, an increased amount of the cellular Slt2 becomes phosphorylated . After 2 h of growth at 39uC, five of the 14 cell wall-related mutants with elevated Slt2 phosporylation at 24uC showed increased Slt2 activation compared to the wild type grown at high temperature (Table 5). Two of the mutants displaying this phenotype are lacking in putative members of the DNAJ molecular chaperones. This class of proteins, usually induced by heat shock and other environmental stresses, play a role in protecting cells from these adverse conditions. Increased activation of Slt2 at 24uC but not at 39uC, as observed for the other nine mutants, suggests that these proteins might be involved in maintaining Slt2 activity at 24uC at a low basal level. An example of the MAP kinase phosphorylation assay is presented in Figure 5, including the yjl056cD (zap1D) strain, which shows increased activation at 24uC, and the ygl128cD strain, in which hyperactivation of the pathway occurs both under inducing (39uC) and non-inducing conditions (24uC).
Mutants showing decreased Slt2 phosphorylation at 39uC were not detected, indicating that none of the mutations studied is indispenable for heat-shock induced activation of the Pkc1-Slt2 pathway.

General approach
We have developed a hierarchical approach to identify and classify genes involved in cell wall Figure 5. Anti-phospho-Slt2 immunoblot analysis of 80 mg protein extracts obtained from the mutant strains yjl056cD and ygl128cD, and the isogenic wild type strain FY1679a and the mutant strain FYDK (slt2D) as a control. Cells were cultured to mid-log phase at 24uC and, where indicated, shifted to 39uC for 2 h prior to collection. Anti-Slt2 immunoblot analysis was performed on the same membrane to verify that similar amounts of Slt2 were present in every lane (data not shown) dynamics. In the present work, the complete EUROFAN set of 620 mutants bearing deletions in non-essential genes was screened using a few simple tests that are indicative of cell wall defects. We used the following primary screens, namely altered sensitivity to Calcofluor white, SDS, and sonication, and altered morphology. Initially, we also investigated whether additional cell wall mutants could be identified by screening for cell lysis at 37uC. However, analysis of a subset of the EUROFAN mutants for this phenotype (data not shown) indicated that, although being selective for mutants with altered cell integrity, this test did not result in the identification of additional mutants and studying cell lysis at 37uC as a primary test was therefore not continued. Mutants selected with our primary screens (145 or 23%) were analyzed further with tests of increasing specificity to detect genes involved in specific aspects of cell wall formation or in regulatory signaling pathways. These secondary screens revealed that 91% of the mutants identified by the Calcofluor white assay, 98% of the mutants identified by the SDS test and 86% of the mutants identified by sonication have additional cell wallrelated phenotypes. In accordance with the close relationship existing between the formation of a functional cell wall and morphogenesis, additional phenotypes were found for 94% of the morphological mutants. Thus, all the primary screens used in this work are indeed strongly selective for mutants with an altered cell wall. Except for the mentioned correlation between the mutants selected by the sonication assay and those identified by microscopic observation, our primary screens showed limited overlap indicating that they were largely independent from each other.
Altogether, for 131 (90%) of the mutants identified in the first round, additional cell wall-related phenotypes were found, indicating that the number of potentially false positives was acceptable and that our screens were efficient. Extrapolated for the entire yeast genome containing more than 6000 genes, our data suggest that at least 1200 genes directly or indirectly affect the cell wall. However, because this study focussed on non-essential gene deletants only, the actual number of genes affecting cell wall biosynthesis is likely to be even higher. Using constructs in which genes of interest are cloned behind a doxycycline-repressible promoter, we are currently applying the same approach to study the involvement of essential genes in cell wall biosynthesis (not shown). Furthermore, as is inherent to this type of approach, our screens did not identify all cell wall-related genes. It is known that, due to redundancy, deletion of some individual genes encoding cell wall proteins do not cause clear phenotypes (Mrsa et al., 1997;Rodriguez-Peñ a et al., 2000;Lussier et al., 1997). In fact, some known cell wall-related genes in the EUROFAN collection (GAS4, CRH1 and PST1) not selected by our screens, belong to gene families. Finally, our screening results show that genes performing a wide range of cellular functions, ranging from transcription, translation, metabolism, or having a mitochondrial function are linked to cell wall metabolism (Table 6). This points to the existence of regulatory associations between apparently unrelated cellular processes. Genome-wide screens such as the one presented here will be very powerful tools to uncover such associations.
Cell wall synthesis and cell wall assembly enzymes 26 mutants showed altered b1,3-glucan or chitin synthetic activities indicating that the corresponding proteins may be involved in the synthesis of these cell wall polymers. They could be directly modulating the activity of the synthase complexes or affecting transport of components of the synthetic machinery. For example, two of the mutants displaying a diminished glucan synthase activity (Table 2) are defective in proteins involved in the regulation of protein trafficking to the membrane: Vps53 (Yjl029c) whose role in protein sorting at the late Golgi compartment has been recently described (Conibear and Stevens, 2000) and the syntaxin Tlg2 (Yol018c). A related member of the syntaxin family, Tlg1, has been reported to be required for the correct localization of chitin synthase III by mediating trafficking of this enzyme to polarized growth sites (Holthuis et al., 1998). The cyclosporin hypersensitivity displayed by the vps53D mutant is consistent with the presumed role of the corresponding protein in glucan biogenesis. Furthermore, a strong correlation with calcium hypersensitivity was observed among the mutants with decreased b1,3-glucan synthase activity, reinforcing the idea of the involvement of calcium in the regulation of the synthesis of this polymer.
In addition, a role in the regulation of the expression of biosynthetic enzymes can not be 138 P. W. J. de Groot et al.
ruled out for proteins whose absence leads to altered synthesis of cell wall polysaccharides. Transcription of CHS3 and FKS1, the genes encoding chitin synthase III, and a catalytic subunit of the b1,3-glucan synthase complex, respectively, was found to be regulated by the cell integrity pathway (Igual et al., 1996;Jung and Levin, 1999). Moreover, an increase in the chitin content of the cell wall has been proposed to be part of a set of reactions induced in response to cell wall perturbation (Popolo et al., 1997;Ram et al., 1998). Therefore, altered biosynthesis of cell wall polymers in some of these mutants may be a consequence of compensation mechanisms triggered by a defective cell wall. For example, deletion of YNL058c, an ORF whose transcription is also regulated by this pathway (Jung and Levin, 1999), led to increased chitin synthase II activity. Additionally, mutant yor275D, displaying a significant reduction both in chitin synthase I and III, is defective in a protein that has similarity to Bro1, which has been reported to interact with components of the cell integrity pathway (Nickas and Yaffe, 1996). Our knowledge of enzymes involved in cell wall assembly is still limited. To identify candidate enzymes involved in cell wall construction steps, we inferred that mutants lacking such enzymes release increased amounts of cell wall components to the growth media. Examples of genes whose deletion causes such phenotypes are GAS1, which is involved in processing of short linear b1,3-glucan chains (Mouyna et al., 2000), and MCD4 which is involved in transfer of ethanolaminephosphate onto the core structure of GPI anchors (Gaynor et al., 1999). Mutants lacking these genes secrete high amounts of cell wall proteins in the culture media Gaynor et al., 1999). In agreement with this, we found high amounts of the GPI-proteins Ssr1 and Cwp1 in the culture supernantant of the mutant deleted in YJL062w (GPI7/ LAS21) which belongs to the MCD4 gene family (Flury et al., 2000). Immunoanalysis of culture Table 6. Genes identified in the primary screening. Classification according to MIPS and YPD databases supernatants revealed, among others, candidate genes involved in coupling of Pir-proteins to the b1,3-glucan network (YNL294c and YOL092w), in b1,3-glucan remodeling (YBR078w, YDL231c and YNL159c) and in GPI anchor biosynthesis (YJL062w, YNL080c). This method therefore proved to be an effective tool to identify proteins involved in assembly of cell wall components.

Regulation of cell wall synthesis
The activation of the cell integrity pathway has been also studied in the selected 145 mutants. Two classes of mutants with altered activation of this pathway were expected. First, mutants lacking components or modulators of the regulatory signaling pathway and, second, mutants which are able to trigger the compensatory mechanism mediated by this pathway because they are defective in proteins playing an important role in cell wall formation. In agreement with this, typical signaling proteins and proteins known to be involved in cell wall biogenesis and maintenance together with other proteins of unknown function were identified by this screen. Most of the mutants displaying a constitutive activation of the cell integrity pathway were also more sensitive to caffeine, although this phenotype was osmotically remediable for only two of them. In total, 17 mutants showed caffeine hypersensitivity which could be stabilized by sorbitol. Among them, 15 mutants did not show an altered pattern of MAP kinase Slt2 activation, suggesting that they have alterations in the cell wall that do not constitutively activate the cell integrity pathway.

Concluding remarks
The use of the hierarchical approach described here has the advantage that a large collection of mutants can be screened rapidly using a few tests that are indicative of putative cell wall defects. A total of 65 genes with no previously experimentally-proved or predicted function have been identified among the EUROFAN collection as having cell wall-related phenotypes (Table 6). Based on previous phenotypic analysis of mutants or by database analysis of protein sequences, a number of proteins among the EUROFAN collection had been suggested to be related to cell wall biogenesis or maintenance. Nine of these, among which Ecm33, Rot2 and Gpi7, were also identified by our screening procedure. Additionally, nine genes were selected that had previously been related to morphogenesis (Table 6). In conclusion, the systematic phenotypic screening of deletion mutants has proved to be a powerful tool in selecting a wide range of mutants with cell wall-related phenotypes of S. cerevisiae and should also be applicable to related fungi having similar cell wall organizations, such as Candida albicans (Kapteyn et al., 2000)