Comprehensive Proteomic Analysis of Saccharomyces cerevisiae Cell Walls

Fungal cells are surrounded by a cell wall, an essential organelle that enables cells to withstand the internal turgor pressure and provides protection against mechanical injury. Electron microscopy studies have revealed that the cell wall of the budding yeast Saccharomyces cerevisiae has a bilayered structure (1, 2). The inner part of the cell wall is electron transparent and consists mainly of a network of branched β-1,3-glucan molecules, held together by hydrogen bridges (3) and extended with covalently attached β-1,6-glucan and chitin molecules (4). The outer part of the wall is electron dense and mainly composed of mannoproteins that are covalently bound to the cell wall glycans. In related fungi such as the human pathogens Candida albicans and Candida glabrata, a similar overall cell wall structure exists (5, 6). The cell wall mannoproteins are thought and, in some cases, demonstrated to be involved in adhesion to host cells and inert surfaces, virulence, fungal morphogenesis, cell wall biogenesis, and possibly biofilm formation (7–12).

β-1,3-Glucan and chitin are individually synthesized by transmembrane protein complexes at the plasma membrane. Whether this is also the case for β-1,6-glucan is not clear. It is also unknown how β-1,3-glucan becomes branched, how linkages between different glycans are achieved, and how mannoproteins are attached to glucans. Recently, we have shown in C. albicans that several putative (trans)glycosidases are covalently linked to the glycan network (13). It is conceivable that these proteins are involved in branching and cross-linking of newly synthesized cell wall polymers or in cell wall remodeling in growing cells.

In terms of their linkage to the glycan lattice, two classes of covalently bound fungal cell wall proteins (CWPs)¹ can be distinguished: 1) glycosylphosphatidylinositol (GPI)-modified proteins, representing the major class of CWPs; and 2) a minor group of CWPs that can be liberated by treating cell walls with mild alkali (alkali-sensitive linkage (ASL)) CWPs. In addition, some proteins may be linked by disulfide bonds to other CWPs (14). Among the fungal GPI-modified proteins that have been experimentally confirmed to be covalently incorporated into the cell wall are flocculins and adhesins (15–18) as well as proteins that are classified as structural CWPs. “Structural” CWPs refer to CWPs with unknown but presumably nonenzymatic functions; usually, they are relatively small proteins with a high percentage of serine and threonine residues, indicating that they may be heavily O-glycosylated. Examples of such proteins are Cwp1p, Ssr1p, Tir1p, Tip1p, Ccw12p, and Sed1p of S. cerevisiae (19–22). In addition, several of the glycoside hydrolases that were identified in C. albicans are GPI-modified proteins (13). Originally, these enzymes were thought to be retained at the plasma membrane to play a role in cross-linking newly formed cell wall polymers synthesized by glycan synthases. Our data suggest that they might also be active while being linked to the cell wall. Furthermore, in C. albicans, two GPI-modified superoxide dismutases (13, 23) as well as the heme-binding protein Rbt5p (24) have been shown to be covalently incorporated into the cell wall.

In the class of ASL CWPs, the best characterized proteins are a family of proteins with conserved internal repeats (Pir proteins). S. cerevisiae Pir1p, Pir2p, and Pir4p and C. albicans Pir1p have been shown to be covalently bound to the cell wall matrix (13, 14). The Pir protein linkage to the cell wall β-1,3-glucan is devoid of interconnecting β-1,6-glucan molecules (25). Recently, experiments with truncated versions of S. cerevisiae Pir4p indicated that the internal repeat sequence is important for incorporation (26). However, the exact mechanism of cell wall attachment of Pir proteins is still unresolved. In C. albicans, the putative β-1,3-glucanase Scw1p was co-extracted with Pir1p (13), indicating that other proteins besides Pir proteins are linked through an alkali-labile linkage.

The first major aim of the work presented here is to determine whether fungi other than C. albicans also possess carbohydrate-active enzymes covalently linked to the cell wall. To this end, we have developed a mass spectrometric method to identify CWPs, unbiased with respect to the covalent linkages to the cell wall carbohydrates. Second, to obtain a better understanding of the function of covalently linked CWPs, we determined to what extent deletion mutants lacking CWP-encoding genes have altered cell wall properties. We show that cell walls of S. cerevisiae, like C. albicans, contain multiple covalently bound glycoside hydrolases and that many of the identified CWPs are required for normal cell wall formation. Finally, we show that besides Pir proteins, S. cerevisiae contains at least three other ASL CWPs.

EXPERIMENTAL PROCEDURES

Strains and Cell Culture—Wild-type strain FY833 (MATa his3Δ300 ura3–52 leu2Δ1 lys2Δ202 trp1Δ63) was grown in YPD (1% (w/v) yeast extract, 2% (w/v) bactopeptone, and 2% (w/v) glucose) and harvested at A₆₀₀ = 2. Phenotypic analyses were performed with the BY4741 strain (MATa his3Δ1 ura3Δ0 leu2Δ0 met15Δ0), and mutant derivatives thereof were obtained from Euroscarf (www.rz.uni-frankfurt.de/FB/fb16/mikro/euroscarf). Mutant strains are single-gene deletants in which the genes of interest are completely deleted and replaced with the Geneticin resistance-encoding KanMX4 module.

Cell Wall Isolation—The detailed procedure for cell wall isolation has been described in De Groot et al. (13). Briefly, cells were harvested by centrifugation, washed with cold H₂O, and then washed with 10 mm Tris-HCl, pH 7.5. Cells were then resuspended in 10 mm Tris-HCl, pH 7.5, and fully disintegrated with 0.25–0.50-mm glass beads (Emergo BV, Landsmeer, The Netherlands) in the presence of a protease inhibitor mixture (Sigma) using a Bio-Savant Fast Prep 120 machine (Qbiogene, Carlsbad, CA). To remove noncovalently linked proteins and intracellular contaminants, isolated cell walls were washed extensively with 1 m NaCl and extracted twice with 2% SDS, 100 mm Na-EDTA, 40 mm β-mercaptoethanol, and 50 mm Tris-HCl, pH 7.8, for 5 min at 100 °C. SDS-extracted walls were washed three times with water, aliquoted, freeze-dried, and stored at -20 °C until use.

Protein Extraction and Fractionation—GPI-modified CWPs were released by incubating the cell walls in undiluted HF-pyridine (Sigma-Aldrich) at 0 °C for 3 h. After quenching the reaction by diluting the reaction mixture with an equal amount of ice-cold H₂O, HF-pyridine was removed by dialysis overnight against H₂O. Pir proteins were released by incubating cell walls with 30 mm NaOH at 4 °C for 17 h. The reaction was stopped by adding neutralizing amounts of acetic acid, followed by dialysis of the released proteins against H₂O or 20 mm bis-Tris, pH 6.0. Fractionation of extracted mild alkali-sensitive CWPs was performed by anion-exchange chromatography using a MonoQ HR 5/5 column (Amersham Biosciences), essentially as described by De Groot et al. (13). Eluted protein fractions were dialyzed against H₂O, freeze-dried, and subjected to electrophoresis in the presence of SDS using linear 2.6–20% gradient polyacrylamide gels. Proteins were visualized by staining with Coomassie Brilliant Blue R-250.

Sample Preparation for Mass Spectrometric Analysis—Freeze-dried cell walls (4 mg) were resuspended in a solution containing 100 mm NH₄HCO₃ and 10 mm dithiothreitol and incubated for 1 h at 56 °C. After centrifugation (5 min at 3000 rpm), the pellet was S-alkylated in a solution containing 100 mm NH₄HCO₃ and 55 mm iodoacetamide for 45 min at room temperature in the dark. Cell walls were then washed three times with 50 mm NH₄HCO₃ and dried under vacuum. Reduction with dithiothreitol and S-alkylation with iodoacetamide of released CWPs in unfractionated protein pools (13) and excised protein bands (27) were done as described previously. For proteolytic cleavage, cell walls, unfractionated protein pools, and excised protein bands in 50 mm NH₄HCO₃ were incubated overnight at 37 °C with sequencing grade trypsin (Roche Applied Science) or at 25 °C with endoprotease Glu-C (Sigma), using a CWP/enzyme ratio of 50:1. For proteolytic digestion of cell walls and unfractionated protein extracts, we assumed that protein accounts for ∼2% (w/w) of the cell wall dry weight. Digested samples were centrifuged, and the supernatants containing the solubilized peptides were analyzed by nanoscale high pressure liquid chromatography electrospray ionization quadrupole time-of-flight tandem mass spectrometry (LC/MS/MS). Remaining pellets of cell walls were washed three times with H₂O, freeze-dried, and stored for immunoblot analysis.

Immunoblot Analysis—Freeze-dried undigested cell walls and proteolytically treated cell wall residues were incubated with recombinant endo-β-1,6-glucanase (ProZyme, San Leandro, CA) to release GPI-modified proteins and with cold NaOH to release Pir proteins, as previously described (28). CWPs were separated by electrophoresis using linear 2.6–20% polyacrylamide gels and electrophoretically transferred onto Immobilon polyvinylidene difluoride membranes (Millipore). Membranes were probed with polyclonal antisera directed against S. cerevisiae GPI-modified CWPs Cwp1p and Ssr1p and the Pir protein Pir2p/Hsp150p, respectively. Sources of antisera and detailed immunoblotting procedures can be found in Kapteyn et al. (28).

Mass Spectrometric Analysis—Proteolytic digests derived from 40 μg of freeze-dried cell walls were fractionated on a 150 mm × 75-μm inner diameter reversed phase capillary column (PepMap C18; LC Packings, Amsterdam, The Netherlands). Sample introduction and mobile phase delivery at 300 nl/min were performed using a Ultimate nano-LC system (Dionex, Sunnyvale, CA) equipped with a 10-μl injection loop. Mobile phase A was water + 0.1% formic acid, and mobile phase B was acetonitrile + 0.1% formic acid. For the separation of peptides, a linear gradient of 5–95% B over 30 min was employed. Eluting peptides were directly electrosprayed into a Micromass quadrupole time-of-flight mass spectrometer (Waters, Manchester, UK). The most abundant ions from the survey spectrum, ranging from m/z 500 to 3500, were automatically selected for collision-induced fragmentation using Masslynx software. Fragmentation was conducted with argon as collision gas at a pressure of 4 × 10^-5 bars measured on the quadrupole pressure gauge. Resulting tandem mass spectrometry spectra were processed with the MaxEnt3 algorithm embedded in Masslynx Proteinlynx software to generate peak lists. Each LC/MS/MS run was repeated at least twice, thereby excluding abundant ions from previous runs.

Data Base Searching and Protein Identification—Tandem mass spectrometry peak lists were used to search the S. cerevisiae proteome from the Stanford Saccharomyces Genome Database (www.yeastgenome.org/) using the MASCOT search engine version 2.0. To identify N-terminal peptides, signal peptidase cleavage sites within GPI-modified and ASL CWPs were predicted using the SignalP 3.0 server (www.cbs.dtu.dk/services/SignalP/), and sequences of the matured proteins were added to the proteome file. The MASCOT searching parameters were as follows: allowing up to one missed cleavage, fixed carbamidomethyl modification, a peptide tolerance of 2.0 Da, and a tandem mass spectrometry tolerance of 0.8 Da. Probability-based MASCOT scores were used to evaluate protein identifications. Only matches with p < 0.05 for random occurrence were considered to be significant. Unassigned tandem mass spectrometry spectra were analyzed with Masslynx Pepseq software to identify peptides with potential posttranslational modifications.

Phenotypic Cell Wall Assays—To test the sensitivity of different mutants to the cell wall-perturbing agent Calcofluor White and to heat stress, cells were pre-grown overnight in YPD. From these cultures, 10-fold serial dilutions were prepared, and 10¹ to 10⁵ cells were spotted onto YPD plates or YPD plates containing 50 μg/ml Calcofluor White. Growth was monitored after 2 and 3 days at 30 °C or 39 °C. For the β-1,3-glucanase sensitivity assay, cells from a fresh overnight culture were inoculated in YPD at a starting A₆₀₀ of 0.1 and cultured at 30 °C to the early logarithmic phase (A₆₀₀ = 0.5–1.0). Cells were collected by centrifugation and gently resuspended in 50 mm Tris-HCl, 40 mm β-mercaptoethanol, pH 7.4, at A₆₀₀ = 1. Cells were incubated at room temperature for 1 h, after which 60 units/ml β-1,3-glucanase (Quantazyme; Quantum Biotechnologies Inc., Laval, Quebec, Canada) was added (t₀). The effect of β-1,3-glucanase treatment was followed by measuring the decrease of A₆₀₀ in time and expressed as the percentage of the A₆₀₀ at t₀.

RESULTS

CWPs Are Efficiently Digested by Proteases While Being Linked to the Glucan Network—Previously, we have identified covalently bound CWPs of C. albicans by releasing specific classes of CWPs using biochemical methods, followed by proteolytic digestion of the liberated protein pools and tandem mass spectrometry (13). Although it led to the identification of 14 CWPs in C. albicans, the use of this method is limited to the identification of CWPs that can be liberated from cell walls with established methods. To abolish this limitation, we now aimed to identify CWPs without a prior protein solubilization step. Thus, generation of proteolytic fragments of CWPs for protein sequencing, using endoproteinases, was performed while the CWPs were still bound to the cell wall lattice. A major concern about this method was whether the CWPs would be fully accessible to the proteinases used. To investigate this, isolated SDS-treated cell walls were incubated with the endoproteinases trypsin or Glu-C, followed by immunoblot analysis of the proteins remaining on the cell wall lattice. Digestion of isolated cell walls with endo-β-1,6-glucanase resulted in release of the major class of covalently bound CWPs in S. cerevisiae, the GPI-modified CWPs, as was demonstrated with antisera raised against the abundant GPI-modified CWPs Cwp1p (Fig. 1) and Ssr1p (data not shown). However, in cell walls that were treated with trypsin or Glu-C prior to digestion with endo-β-1,6-glucanase, we did not detect these proteins or smaller fragments thereof, indicating that GPI-modified CWPs were efficiently digested despite their connection to the cell wall. Similarly, using antibodies against the Pir protein Pir2p/Hsp150p, we tested release of mild alkali-extractable proteins. In contrast to undigested samples, alkali extracts of cell walls digested with trypsin or Glu-C walls did not react with anti-Pir2p serum (Fig. 1), indicating that Pir proteins were efficiently digested when using isolated walls in suspension as substrate as well.

Identification of 19 Covalently Bound CWPs—For the identification of covalently linked CWPs of S. cerevisiae, wild-type cells were grown to mid-log phase. SDS-treated walls, devoid of noncovalently associated proteins, were directly incubated with trypsin and Glu-C to obtain peptide fragments that were separated and sequenced by LC/MS/MS. For each LC/MS/MS run, the complete set of peptide tandem mass spectra was submitted to MASCOT for protein sequence data base searching. The high confidence limit settings (p < 0.05) that were used in the analysis of the peptide data, together with the identification of multiple peptides for most of the proteins, allowed for the unambiguous identification of 18 CWPs from log-phase S. cerevisiae cells using this technique (Table I). Details of the mass spectrometric analysis using MASCOT can be found in Supplementary Table I. The amount of sequenced peptides per identified protein ranged from 1 to 13, with an average of 5. Of the 18 proteins, 10 were identified in both tryptic and Glu-C extracts, 7 were found only in tryptic extracts, and 1 was detected only in Glu-C extracts.

Among the identified CWPs are 12 predicted GPI-modified proteins (29, 30). At least five of these are putative carbohydrate-active enzymes (31) possibly involved in modifying the cell wall glycan network during growth. Gas1p, Gas3p, and Gas5p are classified in glycoside hydrolase family 72 and are thought to be directly involved in β-1,3-glucan remodeling or cross-linking of β-1,3-glucan and β-1,6-glucan chains (32). Crh1p and Utr2p belong to glycoside hydrolase family 16 and seem to have a role, directly or indirectly, in linking chitin to the β-1,3-glucan network (33). For Ecm33p, a member of a family of four GPI-modified proteins (the Sps2p family), little functional information is available. However, recently it has been reported that the proteins of this family contain a receptor L-domain for ligand binding similar to the mammalian type 1 insulin-like growth factor receptor and the insulin receptor (34). Furthermore, similar to GAS1, deletion of ECM33 is known to result in a strong hypersensitivity to the cell wall perturbant Calcofluor White, in large swollen cells, in an increased amount of β-1,6-glucosylated proteins secreted to the growth medium, and in increased levels of activated Slt2p, the mitogen-activated protein kinase of the cell wall integrity pathway (34, 35). This suggests that this protein has a crucial role in cell wall biogenesis and is required to ensure proper cell wall integrity.

One of the identified GPI-modified proteins is the phospholipase Plb2p. Finding Plb2p in the cell wall was rather surprising because phospholipases are generally known to be located and active at the plasma membrane. On the other hand, Plb2p lacks adjacent basic residues in the region immediately upstream of the GPI modification site. Such a dibasic motif is often present in proteins that are predominantly localized at the plasma membrane, whereas this is usually not the case for proteins destined to be cell wall-localized (29, 36). Consistent with this, Plb1p and Plb3p do have a dibasic motif and are not detected in cell walls using mass spectrometry.

Four of the identified GPI-modified proteins, Cwp1p, Tip1p, Tir1p, and Ssr1p, are relatively small proteins that have a high content of serine and threonine residues and are heavily O-glycosylated (Saccharomyces Genome Database, www.yeastgenome.org/). These proteins therefore do not seem to have enzymatic functions, but they may be important to determine the cell surface properties of yeast cells. The last predicted GPI-modified protein is Pry3p (pathogen related in yeast), an unknown protein that has similarity with the plant PR-1 class of pathogen-related proteins.

Three CWPs that were identified in log-phase cells belong to the small family of Pir proteins (14). Discriminating among the four different members of this family using protein sequencing is hampered by the fact that they contain conserved repetitive sequences and produce only a few unique peptide sequences from the C-terminal region (Fig. 2). For instance, the identified tryptic peptide IGSIVANR with a mass of 828.48 Da is present in all four Pir proteins. Obtained sequences of additional discriminating peptides did unambiguously identify Pir1p, Pir2p, and Pir4p in log-phase yeast cells; however, no peptide uniquely specifying Pir3p was found. It is probable that this may be explained by the lack of short tryptic peptides in the C-terminal region of Pir3p because expression levels of PIR3 are comparable with those of the other PIR genes in log-phase cells (37). Because transcript profiling studies indicated that PIR3 expression is up-regulated during stationary phase (38), we grew wild-type cells to stationary phase and analyzed a tryptic digest of isolated cell walls for the presence of Pir3p. In this case, we were able to determine the sequence of a peptide with a mass of 2270.05 Da corresponding to the C terminus of Pir3p, which demonstrated that Pir3p is also incorporated in the cell wall, at least during stationary phase. Apart from identifying Pir3p, analysis of stationary phase cells confirmed most of the protein identifications from log-phase cells but did not reveal other new identifications (results not shown).

Two of the remaining non-GPI-modified proteins, Scw4p and Scw10p, belong to the Bgl2p family of β-1,3-glucanases/β-1,3-glucanosyl transferases (glycoside hydrolase family 17). Interestingly, the orthologous protein Scw1p of C. albicans was recently found to be present in protein extracts of cell walls that had also been pre-treated with reducing agents (13). Covalent incorporation of this family of proteins is therefore not unique for a single organism and may be more common in fungi. The last protein we identified in log-phase cells was Ybr162cp/Tos1p. For Tos1p (target of SBF), we obtained three peptide sequences. In previous studies, Tos1p has been detected in laminarinase-treated cell wall extracts, suggesting a tight association of this protein with the cell wall (39). This is consistent with our data showing that Tos1p is covalently bound to the cell wall network.

Covalently Bound CWPs of S. cerevisiae Are Either GPI-modified or Attached in an Alkali-labile Manner—To understand how the three proteins that are neither GPI-modified nor Pir proteins are covalently linked to the wall and to confirm the linkage type of the other identified CWPs, the two known types of CWPs were separately released from isolated cell walls before trypsin digestion. GPI-modified CWPs can be specifically released with HF-pyridine (13). HF-pyridine cleaves the phosphodiester bonds through which GPI-modified CWPs are linked to β-1,6-glucan chains. Extracted protein pools were then proteolytically digested with trypsin and analyzed by LC/MS/MS, which generated amino acid sequences of 34 different peptides (Supplementary Table I) originating from nine different proteins. These nine proteins corresponded to predicted GPI-modified proteins that were also identified in trypsin extracts with the direct cell wall digestion method. This result confirms GPI modification for most of the predicted GPI-modified proteins identified above and demonstrates the specificity of the HF-pyridine extraction toward GPI-modified proteins.

Pir CWPs and Scw1p of C. albicans are linked to β-1,3-glucan through a linkage that is destroyed by treatment with cold 30 mm NaOH (13, 14). NaOH extraction of cell walls from log-phase yeast cells yielded 17 proteolytic peptide sequences originating from seven proteins. These were the six non-GPI modified proteins (Pir1p, Pir2p, Pir4p, Scw4p, Scw10p, and Tos1p), which were identified in log-phase cells using direct digestion of cell walls, plus Cwp1p (Table I). The presence of the latter in this extract confirms earlier observations that, in addition to GPI-dependent incorporation, Cwp1p can alternatively be linked to the cell wall in a mild alkali-sensitive manner (28). Extraction by mild alkali of Scw4p and Scw10p is consistent with the detection of Scw1p in alkali extracts of C. albicans cell walls (13). Identification of Tos1p in the same extract indicates that other proteins besides Pir proteins and Bgl2P family members can be incorporated in the cell wall via an alkali-labile linkage.

GPI-modified CWPs are attached to β-1,6-glucan via a GPI remnant at their C terminus. In these proteins, functional domains are generally found in the N-terminal regions (Table I). In contrast, Pir proteins have a conserved four-cysteine domain in their C-terminal regions, and in Scw4P and Scw10P functional domains are in the C-terminal regions as well. This may indicate that there is a relation between protein organization and the manner of cell wall incorporation.

Identification of ASL CWPs using Large Scale Fractionation—To identify possible additional, less abundant, ASL CWPs and to test the sensitivity of the direct cell wall digestion method, we undertook a large scale fractionation approach. Starting with a >100-fold increased amount of starting material from an independent culture, cell walls were treated with mild alkali, followed by anion-exchange chromatography and SDS-PAGE analysis of the separated protein fractions. Protein bands visualized with Coomassie Brilliant Blue R-250 were excised from gel and analyzed by LC/MS/MS, resulting in the identification of seven proteins (Fig. 3). These proteins were the seven proteins that were already identified in the small scale unfractionated mild alkali extract and were also detected with the direct cell wall digestion method. The increased amount of individual peptides subjected to LC/MS/MS with the large scale method only allowed us to increase the sequence coverage of these proteins (Supplementary Table I). These results prove the sensitivity and reproducibility of the small scale methods to identify CWPs and emphasize the specificity of the mild alkali extraction to select for a subclass of covalently bound CWPs.

Not all ASL CWPs Contain Pir-specific Tandem Repeats— Mutagenesis studies with Pir4p of S. cerevisiae indicated that covalent attachment of Pir proteins to the cell wall matrix is governed by the presence of internal, Pir-specific, tandem repeats (26) conforming to the consensus sequence Q[IV]XDG-Q[IVP]Q. Interestingly, GPI-modified CWP Cwp1p, which is also present in NaOH extracts, contains one perfect copy of such a repeat (Fig. 4), whereas CaScw1p contains a glutamine-rich region that is partly similar. However, Pir-specific repeats are absent in S. cerevisiae Scw4p, Scw10p, and Tos1p. This prompted us to perform pairwise protein alignments and BLAST analyses to identify other common features of the alkali-extracted CWPs of S. cerevisiae and C. albicans. The only obvious similarity seems to be the presence of adjacent basic residues (KR and KK), known to be potential substrates for proprotein cleavage by the maturating enzyme Kex2p, in the N-terminal regions of all these proteins except Cwp1p (Fig. 4). Moreover, in the proteins that do not contain Pir-specific internal repeats, the Kex2p site is preceded by at least three additional positively charged residues (including at least two histidines). It is possible that a strongly positively charged Kex2p region may be an additional factor contributing to covalent incorporation of CWPs in a mild alkali-sensitive manner.

Phenotypic Analysis of Mutant Strains Lacking CWP-encoding Genes—To investigate the relevance of the identified CWPs for attaining normal cell wall structure and integrity, BY4741-derived single gene knock-out strains carrying deletions in the corresponding genes were subjected to cell wall-related phenotypic tests (Fig. 5). First, we tested sensitivity to Calcofluor White (CFW), a drug that interferes with cell wall biosynthesis by binding to chitin. Mutants with aberrant cell wall structures are often hypersensitive to the presence of CFW in the growth medium (40). As described in earlier reports and consistent with the observation that the levels of chitin in cell walls of GAS1 and ECM33 deletion mutants are drastically increased, these mutants are highly sensitive to the addition of CFW to the growth medium (34, 41, 42). Those deleted in CRH1, SCW4, SSR1, PIR2, PLB2, PRY3, and TOS1 showed only slight hypersensitivity to CFW, whereas the sensitivity to CFW of the remaining mutants was comparable with that of the wild type (data not shown). Second, we tested sensitivity to sustained heat stress. Again, mutants deleted in GAS1 and ECM33 showed the most severe phenotypes and were unable to grow at 39 °C. The tos1Δ mutant showed a slightly decreased ability to grow at this high temperature, whereas the others showed wild-type behavior. Third, sensitivity to Quantazyme, a recombinant β-1,3-glucanase that hydrolyzes β-1,3-glucan, was tested. Eight mutants showed increased resistance to Quantazyme in comparison to wild type. Of these mutants, gas1Δ, tos1Δ, and scw4Δ were most resistant; pry3Δ, tir1Δ, and ecm33Δ showed intermediate resistance; and pir2Δ and plb2Δ showed slightly increased resistance. Interestingly, the Quantazyme resistance of tos1Δ is not accompanied by a dramatically increased sensitivity to CFW, as was observed for gas1Δ and ecm33Δ. This suggests that chitin incorporation in tos1Δ is not increased but that β-1,3-glucan molecules are less accessible to the β-1,3-glucanase (for instance, by alterations in the outer protein layer) or more resistant to the enzyme by increased branching. Taken together, the results of these phenotypic tests underline the notion that the identified covalently bound CWPs, including the newly identified Pry3p, Tos1p, Plb2p, and Ecm33p, are important to the cell for normal cell wall construction.

DISCUSSION

This study describes the identification of 19 covalently bound CWPs of S. cerevisiae using a rapid, sensitive, and versatile method based on mass spectrometry. In contrast to our earlier approach, which involved solubilization of CWPs prior to proteolytic digestion, in the current set-up, cell walls are directly subjected to proteolytic enzymes. This has two major advantages. First, the new method is faster, which increases the throughput while maintaining a sensitivity comparable to large scale approaches. Second, the new approach allows for the identification of proteins irrespective of the nature of their covalent linkages to the cell wall lattice. In other words, we can make a full inventory of covalently bound CWPs, even when knowledge about protein-glycan linkages is lacking. Our method will therefore also be applicable to other fungi, provided that they contain CWPs that are not shielded from cleavage by endoproteinases. Conceivably, this may lead to the discovery of novel types of connections between glycoproteins and cell wall matrices.

Several lines of evidence indicate that all proteins identified in this study are genuine covalently linked CWPs. First, all peptides identified in this study (Supplementary Table I) originated from secretory proteins. Our cell wall preparations are therefore free from any intracellular contaminants. Second, all identified proteins have GPI signatures or can be solubilized from cell walls by 30 mm NaOH, which specifically releases covalently bound CWPs (13, 14, 25). Third, phenotypic tests showed that mutants lacking genes coding for many of the identified CWPs have strong cell wall-related phenotypes, indicating that their cell wall integrity is affected by a change in cell wall composition. Fourth, the cell wall identification of all proteins reported in our study is validated by independent studies that show covalent linkage to cell wall glycans (7 proteins) or cell surface association (the other 12 proteins) (Table I). In most cases, the evidence is obtained from fractionation studies, in which glucanase-extractable or biotin-labeled proteins are analyzed by N-terminal sequencing (14, 19, 20, 43), by immunoblot analysis (21, 39, 44, 45), or by measuring the activity of a reporter protein that is placed in front of C-terminal GPI anchor sequences (46, 47). Only in a few cases (Cwp1p, Crh1p, and Crh2p) has cell surface localization been demonstrated in vivo using fluorescence microscopy (33, 48, 49). Localization studies of CWPs using (green fluorescent protein) fusion constructs are hampered by the fact that these proteins undergo various processing steps and other posttranslational modifications. This may easily lead to mislocalization and misinterpretation of results. For instance, in GPI-modified proteins, the C-terminal part encodes a hydrophobic GPI-modified signal peptide, which is removed in the endoplasmic reticulum by a transamidase complex. C-terminally attached green fluorescent protein fusion constructs, as reported by Huh et al. (50), whose data are accessible at Saccharomyces Genome Database (www.yeastgenome.org/), will therefore show the localization of the detached GPI signal peptide, instead of the mature GPI-modified protein. The same approach also failed to show cell surface localization of the ASL CWPs Scw4p, Scw10p, Pir1p, Pir2p, and Pir4p, all of which are well-known, biochemically confirmed cell surface proteins (14, 43, 51).

In S. cerevisiae, 60–70 proteins are predicted to be GPI-modified by in silico analysis (29, 30). Nevertheless, we strongly believe that the 19 proteins identified in this study, including 12 predicted GPI-modified proteins, represent almost all CWPs that are covalently bound to cell wall glycans of S. cerevisiae under the conditions tested. Approximately 40 of the predicted GPI-modified proteins are suggested to remain predominantly attached to the plasma membrane rather than becoming covalently linked to the cell wall (7). Also, many GPI-modified proteins are differentially regulated. For instance, perturbations of the yeast cell wall (37, 52, 53) or hypoxic growth conditions (54) cause S. cerevisiae cells to use a very different set of CWPs. Consistent with our results, in an extensive search for CWPs, only 13 different CWPs were detected on SDS gels in exponential-phase cells of S. cerevisiae (14, 19, 43). Recently, analysis of the cell wall proteome of C. albicans, using LC/MS/MS, also resulted in the identification of 12 GPI-modified CWPs from log-phase cells (13). However, we cannot rule out that our analysis may have missed proteins if their proteolytic fragments fall out of the range detectable by LC/MS/MS, especially when they are glycosylated. Notably, most of the identified peptides derive from predicted functional domains (55) or from other regions with a relatively low frequency of serines and threonines.

Scw4p, Scw10p, and Tos1p neither contain consensus sequences for GPI modification nor belong to the Pir protein family. Similar to Scw1p of C. albicans (13), these proteins were co-extracted with Pir proteins by treatment with NaOH, and they are resistant to treatment with hot SDS/β-mercaptoethanol and dithiothreitol. Scw4p and Scw10p have been described as proteins that can be released from cell walls by extraction with reducing agents (43). Apparently, a fraction of these CWPs do not contain alkali-labile linkages to the glucan network but bind to the cell wall in a manner that is sensitive to reduction of disulfide bonds. Scw4p, Scw10p, and Tos1p do not contain Pir-specific internal repeats, which are required to incorporate Pir proteins in an alkali-labile manner (26). A common feature of alkali-extracted proteins is the presence of a Kex2p substrate site (KK/KR) in the N-terminal region, and in Scw4p, Scw10p, Tos1p, and CaScw1p this site is preceded by at least three additional positively charged amino acids. We hypothesize that this may contribute to covalent incorporation of CWPs in a mild alkali-sensitive manner. However, this would imply that such alkali-extractable proteins are not processed by Kex2p because Kex2p will remove the positively charged regions from the mature proteins. Strikingly, Kex2p processing of Pir proteins has only been shown for secreted proteins (14), and unprocessed Pir1p of C. albicans is incorporated into the cell wall (13).

Whether a GPI-modified protein localizes to the cell wall or to the plasma membrane is primarily determined by the amino acids immediately upstream of the site of GPI anchor addition. Adjacent basic residues in this region promote plasma membrane retention rather than cell wall incorporation (29, 30, 56). Recently, it has been demonstrated that long regions rich in serine and threonine residues can override this dibasic signal and redirect a GPI plasma membrane protein to the cell wall (57). Consistent with this, of the 12 identified GPI-modified CWPs, only Gas1p and Ecm33p have a dibasic motif, and in both cases this is preceded by a stretch of consecutive serine residues. Gas1p has been shown to be localized in the plasma membrane (58). The fact that we detect Gas1p and Ecm33p in the cell wall may also be explained by their high expression levels or by the high sensitivity of our method (13).

The identified GPI-modified CWPs Gas1p, Gas3p, Gas5p, Crh1p, and Crh2p specify putative (trans)glycosidases able to modify the carbohydrates of the cell wall. Furthermore, Ecm33p is also crucial to attain a normal cell wall structure. Orthologues of these three GPI-modified protein families have previously been isolated from plasma membranes in Aspergillus fumigatus (59). This raises an important question: at which location (plasma membrane or cell wall) are these proteins mainly functional? Interestingly, a mutant lacking ECM33 in S. cerevisiae was only partially complemented by the homologous (58% identity) protein Pst1p (34). As suggested by these authors, these proteins possibly have different functions; however, the improper complementation may also be caused by different localization of the proteins because Pst1p has never been shown to be covalently incorporated into the cell wall. Apart from five GPI-modified CWPs, Scw4p and Scw10p also specify (trans)glycosidases. Because these proteins are not GPI-modified, they enter the cell surface as soluble proteins and are not bound to the plasma membrane. Interestingly, members from all three glycoside hydrolase families were also shown to be covalently bound to the cell wall in the pathogenic yeast C. albicans (Table I), and orthologues are present in the genomes of many other ascomycetic fungi with similar cell wall structures. Taken together, these observations indicate that the activity of the proteins is not restricted to the plasma membrane, but they may be actively involved in cell wall biosynthesis while being covalently bound to the cell wall. Alternatively, these enzymes may play a role in biofilm formation. Biofilms usually consist of a polymorphic population of cells embedded in a glycan matrix and are an important factor in pathogenesis of Candida spp. (60, 61). Biofilm formation has also been demonstrated in the budding yeast S. cerevisiae (62). Actually, natural environments of fungi are often solid surfaces that they can colonize and to which they irreversibly adhere. Future studies will aim to obtain a better understanding of the function of these proteins, to analyze their occurrence in cell walls of other fungi, and to biochemically decipher the alkali-labile linkage(s).