Analysis of β-1,3-Glucan Assembly in Saccharomyces cerevisiae Using a Synthetic Interaction Network and Altered Sensitivity to Caspofungin
Guillaume Lesage, Anne-Marie Sdicu, Patrice Ménard, Jesse Shapiro, Shamiza Hussein, Howard Bussey


Large-scale screening of genetic and chemical-genetic interactions was used to examine the assembly and regulation of β-1,3-glucan in Saccharomyces cerevisiae. Using the set of deletion mutants in ∼4600 nonessential genes, we scored synthetic interactions with genes encoding subunits of the β-1,3-glucan synthase (FKS1, FKS2), the glucan synthesis regulator (SMI1/KNR4), and a β-1,3-glucanosyltransferase (GAS1). In the resulting network, FKS1, FKS2, GAS1, and SMI1 are connected to 135 genes in 195 interactions, with 26 of these genes also interacting with CHS3 encoding chitin synthase III. A network core of 51 genes is multiply connected with 112 interactions. Thirty-two of these core genes are known to be involved in cell wall assembly and polarized growth, and 8 genes of unknown function are candidates for involvement in these processes. In parallel, we screened the yeast deletion mutant collection for altered sensitivity to the glucan synthase inhibitor, caspofungin. Deletions in 52 genes led to caspofungin hypersensitivity and those in 39 genes to resistance. Integration of the glucan interaction network with the caspofungin data indicates an overlapping set of genes involved in FKS2 regulation, compensatory chitin synthesis, protein mannosylation, and the PKC1-dependent cell integrity pathway.

THE cell wall is a major organelle that surrounds cells, is responsible for cell shape and osmotic stability, and acts as a filter for large molecules. The cell wall is composed mainly of β-1,3 and β-1,6-glucans, mannoproteins, and chitin, with the relative proportions of these constituents varying with growth conditions and the cellular developmental program. β-1,3-Glucan is the principal cell wall component, to which the other components are crosslinked (Smitset al. 1999; Kliset al. 2002). Synthesis of β-1,3-glucan occurs at the plasma membrane. Glucan synthase is thought to contain a catalytic subunit, encoded by the two homologous genes FKS1 and FKS2/GSC2 (Mazuret al. 1995), and a regulatory subunit, the small GTPase Rho1p (Drgonovaet al. 1996; Mazur and Baginsky 1996; Qadotaet al. 1996). FKS1 and FKS2 encode a pair of integral membrane proteins with 16 predicted transmembrane domains that share 88% identity. Deletion of FKS1 leads to a decrease in β-glucan and an increase in chitin and mannoprotein levels in the cell wall. The deletion of FKS2 causes no obvious cell wall defect, but a fks1Δ fks2Δ double mutant is inviable (Mazuret al. 1995). The yeast genome contains a third gene, FKS3, whose product is 72% identical to Fks1p and Fks2p. The role of FKS3 remains unknown, but a fks3Δ mutant has no apparent cell wall defects or genetic interactions with FKS1 or FKS2 (Dijkgraafet al. 2002). In addition to the Rho1p regulatory subunit, other proteins are required for normal levels of β-1,3-glucan. The SMI1/KNR4 gene was cloned by complementation of a Hansenula mrakii K9 killer toxin (a glucan synthase inhibitor) resistant mutant (Honget al. 1994). The smi1Δ mutant has a highly permeable cell wall and shows both decreased glucan synthase activity and cell wall β-1,3-glucan content (Honget al. 1994; Martinet al. 1999). Genetic and biochemical evidence suggests that Smi1p acts in the PKC1-SLT2 signaling cascade by modulating the kinase activity of Slt2p (Martin-Yken et al. 2002, 2003).

Cell wall composition changes during growth, budding, mating, and sporulation, and these dynamic processes require remodeling of the crosslinking of β-1,3- and β-1,6-glucans to themselves and to other cell wall components. Gas1p, a GPI-anchored protein localized to the extracellular face of the plasma membrane, has β-1,3-glucanosyltransferase activity and is involved in this remodeling (Mouynaet al. 2000). A null gas1 mutant releases β-glucosylated proteins into the medium and shows increased chitin and mannoprotein levels (Ramet al. 1998). Such increased levels of cell wall components can compensate for a defect in a specific polymer: for instance, a decrease in β-1,3-glucan is buffered by an increase in chitin made by chitin synthase III (Valdiviesoet al. 2000; Garcia-Rodriguezet al. 2000b; Carottiet al. 2002). CHS3 encodes chitin synthase III, and Chs3p is responsible for synthesis of the chitin in a ring at the bud neck, in the lateral wall, and in response to external stress (Roncero 2002).

View this table:

Yeast strains used

To uncover the network of genes involved in β-1,3-glucan biology, we made two studies, using the collection of yeast mutants singly disrupted in each gene. As part of a larger study of the yeast genetic network (Tonget al. 2004), we looked for mutations leading to a growth defect when combined with a FKS1, FKS2, GAS1, or a SMI1 deletion. We found that deletion of 135 genes impaired growth of fks1Δ, fks2Δ, gas1Δ,or smi1Δ mutants and we analyze these interactions here. As a complementary approach, we looked for single deletions leading to an altered sensitivity to the β-1,3-glucan synthase inhibitor, caspofungin. Caspofungin is an echinocandinlike antifungal lipopeptide that inhibits β-1,3-glucan synthesis in vitro by affecting the FKS1 and FKS2 gene products in Saccharomyces cerevisiae (for a review see Deresinski and Stevens 2003; Letscher-Bru and Herbrecht 2003). Mutant alleles in FKS1 and FKS2 that lead to echinocandin resistance in S. cerevisiae have been identified (Douglaset al. 1994; Mazuret al. 1995). We globally tested for altered sensitivity of ∼4600 haploid deletion mutants and the ∼1100 heterozygous diploids in essential genes and found 91 genes with such a phenotype.


Strains, media, and drugs: All strains used (Table 1) are available from the deletion project consortium (Winzeleret al. 1999). Haploid deletion mutants were previously arrayed on 16 768-format plates using a colony picker (Tonget al. 2001). Three plates with 1058 diploid strains heterozygous for essential genes were also arrayed. Arrays were propagated at 30° on standard YEPD (10 g/liter yeast extract, 20 g/liter bacto-peptone, 20 g/liter glucose) or YEPD supplemented with 200 μg/ml G-418 (Invitrogen, Carlsbad, CA). Caspofungin acetate (Cancidas; Merck, Whitehouse Station, NJ) was a gift from Elitra Canada (Montreal). Nourseothricin (clon-NAT) was purchased from Werner Bioagent (Jena, Germany).

SGA analysis: The synthetic genetic array (SGA) analysis procedure is fully described elsewhere (Tong et al. 2001, 2004) and is briefly summarized below.

Construction of query strains: Query strains HAB1122, HAB1123, HAB1124, HAB1125, HAB1126, and HAB1127 (Table 1) were obtained in four steps. First, the KanMX4 from BY4741-derived strains (Table 1) was switched to NatMX4 by PCR-based transformation. Second, the nourseothricin-resistant transformants derived from chs3 or other mutants were mated to Y3656 or Y3084 (Table 1), respectively, and the MATa/α diploids were transferred to 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 clonNAT to select for the deletion mutants. Fourth, cells were replica plated onto medium lacking lysine to identify lys2Δ derivatives.

SGA screens: A given query strain was pinned onto a fresh YEPD plate at a density of 768/plate, and then the deletion mutant array was pinned on top of the query cells. The resulting diploids were selected on medium containing G418 and clonNAT. Arrays were then pinned onto sporulation medium. After a 5-day incubation at 22° spores were pinned onto haploid selection medium to select for growth of MATa spore progeny. This step was performed twice. Then, meiotic progeny carrying the deletion mutation derived from the deletion mutant array parental strain were selected on medium containing G418. Finally, double mutants were selected on haploid-selection medium containing G418 and clonNAT for 2 days. Colony size was then scored by visual plate inspection. Each screen was done in triplicate, and putative interactions scored multiple times (∼1800 interactions for the six SGA screens) were subjected to confirmation tests.

Confirmation of synthetic interactions: Spores were germinated for 2 days at 30° in liquid haploid selection medium. The MATa progeny were diluted in sterile water and plated out on medium that selects for the query-gene mutation [clon-NAT], the deletion mutant array mutation [G418], or both the query-gene and deletion mutant array mutations [clonNAT/G418], and then incubated at 30° for ∼2 days. Colony growth under the three conditions was compared and the double mutants were scored as synthetic sick (SS), synthetic lethal (SL), or no interaction (No). Tetrad analysis was used to test synthetic interactions in 42 cases. In all, 248 interactions were positive.

Accuracy of the procedure: Since screens were done in triplicate followed by a confirmation procedure, we expect our data set to be largely devoid of false positives. However, some interactions may have been missed (false negatives). A search in the literature and databases indicated that 10, 1, 1, 11, 5, and 8 synthetic-lethal interactions were reported for FKS1, FKS2, FKS3, GAS1, SMI1, and CHS3, respectively. Of these 36 interactions, 6 engaged essential genes and thus were not seen with our procedure. Of the 30 remaining “observable” interactions, 22 were also found in our screen. On the basis of this, we estimate the rate of false negatives to be ∼30%, which is consistent with an estimate made on a larger SGA data set (Tonget al. 2004). Some true interactions would be missed if they involve one of the ∼500 genes whose deletion led to a systematic defect in our assay and were excluded from analysis (Tonget al. 2001). For example, 25 of the 45 nonessential genes whose deletion leads to caspofungin hypersensitivity fall into this group.

Caspofungin sensitivity/resistance screening procedures: Robotic procedures: Mutants were pinned onto YEPD plates with or without caspofungin. Final caspofungin concentrations were 500 ng/ml (from a 1-mg/ml stock in 1% DMSO) for sensitivity testing and 5 μg/ml (from a 10-mg/ml stock in 1% DMSO) for resistance screening. Growth was scored after overnight incubation at 25°. Strains showing significant growth defects on a 0.5-μg/ml caspofungin plate (mutants in 157 nonessential genes and 103 essential genes) or growing on a 5-μg/ml caspofungin plate (mutants in 116 nonessential genes) were individually confirmed by the spotting assay described below.

Confirmation and scoring procedure: Due to high cell density (768 colonies/plate) and the pinning geometry of the plate during the robotic screening, the caspofungin concentrations used during the screening were higher than those used in the confirmation test. Cells were grown in liquid YEPD to log-phase, diluted to OD600 0.5, serial diluted 10-fold four times, and 2.5 μl was spotted onto YEPD ± caspofungin. Since haploids are less sensitive to the drug than diploids, confirmation of the hypersensitivity phenotype was performed at 100 and 200 ng/ml caspofungin for the former and at 150 ng/ml caspofungin for the latter. Phenotypes were scored after overnight incubation at 25°, by checking growth of mutants in the presence or absence of the drug and comparison to growth of the wild-type strain. Sensitive haploid mutants that failed to grow at 100 ng/ml, or at 200 ng/ml, or grew very slowly at 200 ng/ml were scored as - - -, - -, or -, respectively. Sensitive diploid strains were scored - - - if they did not grow in the presence of drug and - - when they grew poorly. Confirmation of the resistance phenotype was performed at 400 ng/ml.

Probability of overlap between networks: The probability P that Nhit genes are found in two data sets composed of N1 and N2 interacting genes was estimated using the formula P=P(N1,Nhit)P(N,Nhit)P(N2,Nhit)P(N,Nhit) (Parsonset al. 2004), where N is the total number of interactions tested (i.e., the number of strains in the array), and P(N, M) = N!/(M!(NM)!).


Synthetic interactions with mutants in β-1,3-glucan assembly

To identify genes buffering defects in β-1,3-glucan synthesis, we searched for genes required for viability or optimal growth in fks1Δ, fks2Δ, fks3Δ, gas1Δ,or smi1Δ backgrounds. Haploid deletion mutants in 4598 genes were arrayed and crossed with strains individually deleted for FKS1, FKS2, FKS3, GAS1,or SMI1. The resulting diploids were selected, sporulated, and haploid double mutants scored for growth. Double mutants showing a growth defect were scored as candidate interactants. Random spore analysis was used to confirm a synthetic interaction. In all, 76, 71, 48, and 1 genes were found to interact with FKS1, SMI1, GAS1, and FKS2, respectively. No synthetic interactions were found with the fks3 null strain. We found 77 synthetic lethal interactions and 118 double mutant combinations leading to growth defects. The 135 genes involved in these interactions are grouped in eight categories according to their cellular function (Table 2) and are displayed as a network of 195 interactions depicted as edges linking two nodes (Figure 1).

Synthesis and regulation of the cell wall: In addition to FKS1, FKS2, GAS1, and SMI1, 27 genes group here. Some are involved in the synthesis of cell wall components such as chitin (BNI4, CHS3, CHS4, CHS5, CHS6, and CHS7), β-1,6-glucan (KRE1), or protein glycosylation (CWH41, MNN10, MNN11, ROT2, and VAN1). This indicates that when β-1,3-glucan synthesis is impaired, correct synthesis of other cell wall constituents is required for normal cell growth or viability.

Components of the PKC1-SLT2 cell integrity pathway, such as sensors (MID2 and SLG1/WCS1), regulators (BEM2 and ROM2), kinases (BCK1 and SLT2), and transcription factors (RLM1, SSD1, and SWI4), occur in the interaction network. This mitogen-activated protein (MAP) kinase cascade orchestrates morphological change by regulating cell wall assembly in response to stress and low osmolarity (Heinischet al. 1999). Moreover, the ability to respond to low osmolarity with the glycerol channel Fps1p is essential in the fks1 and smi1 mutants. Thus, appropriate osmosensing and a functional cell integrity pathway buffer mutants defective in β-1,3-glucan synthesis.

Finally, five poorly characterized genes (DFG16/ECM41, ECM7, ECM21, IMG1, and RIM20) are candidates for involvement in cell wall assembly, with their mutants having altered sensitivity to environmental stresses or cell surface perturbing agents.

Polarity and secretory pathway function: Since cell polarity, vesicular transport, endocytosis, and membrane biogenesis are needed for coordinating cell wall assembly during yeast growth (Pruyne and Bretscher 2000a,b), the 29 genes involved in these cellular functions are grouped in a single category. Genes involved in cell polarity and showing interaction in the network encode regulators of the Cdc42p GTPase (BEM1, BEM4), scaffold proteins regulating the directionality of actin polymerization from the bud tip (BNI1, SPA2), regulators of septin assembly at the bud neck (CLA4, ELM1), and factors with a role in cell morphology and budding (BUD19, HBT1). Stages of secretion found among interacting genes involved in vesicular transport are ER to Golgi (BRE5, EMP24, and UBP3), intra-Golgi (KRE11), Golgi to bud neck (CSR2 and SBE2), and vacuole assembly (CCZ1, VPS61, VPS63, and VPS67). Genes required for correct endocytosis are found in the glucan network, encoding regulators of cortical actin patch assembly (EDE1, RVS167, SHE4, SLA1, and YLR338W) or regulators of vesicle trafficking from endosomes to the Golgi (RIC1, RGP1, and YPT6). A further three genes are involved in membrane biogenesis (FAT1, FEN1/ELO2/ GNS1, and OPI3).

View this table:

Genes showing synthetic interaction with FKS1, FKS2, GAS1, or SMI1

Transport of cell wall assembly components to sites of cell wall expansion requires cell polarity control, forward transport through the secretory pathway, and endocytosis-mediated recycling. In mutants defective in β-1,3-glucan synthesis, such polarized transport is essential to bring compensatory components to the cell wall.

Transcription regulation and stress response: Genes encoding transcription factors (CRZ1, HAP1, and IXR1), transcription factor regulators (RSC2 and SNF1), a subunit of RNA polymerase I (RPA14), and RNA processing factors (DBP7, GRS1, LSM1, LSM6, PSP2, and TOP1) were found. An additional set of nine genes putatively involved in stress responses also falls into this group (AAD4, HIT1, IMP2/YIL154C, LYS2, QRI5, SIS2, YDJ1, YJR046W, and YMR073C). The regulated expression of genes compensating for defects in β-1,3-glucan synthesis may depend upon these gene products.

Figure 1.

—Network of synthetic interactions with CHS3, FKS1, GSC2/FKS2, GAS1, and SMI1. Genes engaged in interactions are represented as nodes. Nodes are colored according to functional categories, assigned on the basis of information from the literature.

Ion homeostasis and signal transduction: Excluded here are genes involved in the cell integrity pathway discussed above. The five remaining genes are involved in the downregulation of the high-osmolarity glycerol response (HOG) pathway (PTC1), calcium signaling (CCH1, CNB1, and SPF1/COD1), and phosphate transport (PHO86). Ion homeostasis, by acting through signaling cascades, may contribute to the onset of processes essential for viability of β-1,3-glucan mutants.

Ubiquitin-regulated protein degradation: Constituents of the 26S proteasome (PRE9 and RPN10) and ubiquitin-conjugating enzymes (PEX4 and UBC4) were found. Ubiquitination regulates a number of processes required in β-1,3-glucan mutants, such as endocytosis or cell cycle progression.

Cell cycle: Regulation of cell cycle progression by cyclin action (PHO85) or destruction (DOC1 and YNL171C) is crucial in coordinating cell wall synthesis and cell growth and buffers β-1,3-glucan mutant defects.

Other genes and poorly characterized genes: Of the other genes interacting with FKS1, GAS1,or SMI1 (Table 2), 17 have a known function not discussed above, and 24 are poorly characterized or of unknown function.

Screen for genetic interactions with CHS3

As chitin synthesis can compensate for mutational defects in β-1,3-glucan synthesis, CHS3 and other genes required for Chs3p function show genetic interactions with FKS1, SMI1, and GAS1. To further investigate this compensation process, we reasoned that genes involved in balancing chitin and β-1,3-glucan synthesis should be required for the normal growth of mutants with defects in chitin synthase or β-1,3-glucan synthase. Thus, we searched among the genes required in the absence of CHS3 for those that are also required in the absence of FKS1, SMI1,or GAS1. An SGA analysis was performed with the chs3 null strain, and 53 gene deletions affected growth, with 26 of these also found in the glucan network (Figure 1). The remaining genes are listed in Tong et al. (2004). These 26 overlapping genes fall mainly into two categories: secretory pathway polarization (12 genes) and synthesis and cell wall regulation (8 genes). Thus, the proper localization of cell wall synthesis components buffers both glucan and Chs3p-dependent chitin synthesis.

Screens for altered sensitivity to caspofungin

To broaden our view of β-1,3-glucan biology, we searched for genes whose deletion led to altered sensitivity to caspofungin, a glucan synthase inhibitor. As caspofungin is thought to inhibit both Fks1p and Fks2p, such an analysis should give insights distinct from our interaction approach that examines the buffering effects of genes on mutants individually deleted for the FKS1 or FKS2 target genes. A screen for growth in the presence of caspofungin was made with 4598 haploid strains deleted for nonessential genes and 1058 strains heterozygous for essential genes. As the wild-type diploid had a higher sensitivity to caspofungin than the wild-type haploid (Figure 2), screens were performed at concentrations specific for these two cell types. The search for hypersensitive mutants was performed at a subinhibitory caspofungin concentration, while for screening resistant mutants a drug concentration that inhibited growth of the wild type was used (Figure 2). Strains were first grown on YEPD and then replicated onto YEPD with or without caspofungin. The hits were then confirmed by a spotting assay (see materials and methods and Figure 2). We found 45 haploid deletion mutants to be hypersensitive to caspofungin. Of these, 23 were also tested for haplo-insufficiency, with 16 (69%) showing a haplo-insufficient sensitivity phenotype as heterozygous diploids (Table 3 and supplementary Table 1 at In addition, among the ∼1100 heterozygous null mutants in essential genes, 7 were caspofungin hypersensitive (Table 3). Finally, a screen for haploid deletion mutants able to grow at high caspofungin concentration gave mutants in 39 genes with caspofungin resistance (Table 4 and supplementary Table 2 at

Figure 2.

—Assay for altered sensitivity to caspofungin. Dilutions of exponentially growing wild-type haploid or diploid (WT or WT/WT, respectively) and mutant strains were spotted onto YEPD plates containing the indicated caspofungin concentration.

Genes involved in multidrug sensitivity: Recently, a set of yeast mutants that are hypersensitive to a range of inhibitory compounds has been identified (Parsonset al. 2004). A number of these mutants also show hypersensitivity to caspofungin (supplementary Table 1). These genes are involved in a wide range of cellular functions: assembly of the vacuolar H+-ATPase (PPA1/ VMA16, TFP3/VMA11, VMA2, VMA4, VMA5, VMA7, VMA10, VMA13, VMA22, and VPH2), late endosomal trafficking (SNF7 and STP22), ergosterol synthesis (ERG6), transcription (CCR4 and SPT20), nuclear migration (SPC72), glycogen turnover (GPH1), and signal transduction (SLT2).

Although no global compendium of multidrug-resistant yeast mutants is currently available, a literature search revealed that mutations in 12 additional genes confer resistance to a number of drugs as well as caspofungin (supplementary Table 2). These genes are involved in lipid biosynthesis (CSG2, FEN1, MCT1, SUR1, and SUR4), ER-to-Golgi trafficking (ERV14), and signal transduction (CKA2 and CWH43) or are of unknown function. As mutants in the genes in supplementary Tables 1 and 2 show altered sensitivity to a diverse set of bioactive compounds, their altered sensitivity to caspofungin is likely nonspecific.

Genes specifically involved in caspofungin toxicity: These are grouped in five categories (Tables 3 and 4).

Synthesis and regulation of the cell wall: Deletion of FKS1 leads to hypersensitivity, while deletion of FKS2 leads to caspofungin resistance relative to a wild type. Mutants deleted for genes required for chitin synthase III-dependent chitin deposition (CHS37) and protein mannosylation (MNN10) are also hypersensitive to the drug.

View this table:

Genes whose deletion confers hypersensitivity to caspofungin

Deletion of SLG1 or TUS1, two genes acting upstream of the cell integrity pathway, confers caspofungin resistance. In contrast, deletion of genes acting in the downstream part of the pathway such as SMI1 or SIT4 leads to caspofungin hypersensitivity.

Cytoskeleton and vesicular transport: Two essential genes (CCT2 and CCT5) have heterozygous diploids that are hypersensitive. In addition, the deletion of SLA1, RCY1, components of the endocytic pathway, or the vacuolar protein-sorting gene VPS66 leads to a moderate increase in caspofungin sensitivity. Among genes whose deletion decreases susceptibility to caspofungin are DNF2 and LEM3, both involved in membrane trafficking, and SEC66 and VID24, both involved in protein trafficking.

Signal transduction and stress: Deletion of any of the five genes in this group increases caspofungin resistance. These genes encode components of signaling cascades (MRK1 and SSK2) or factors activated by or mediating sensitivity to various stresses (MCA1, PGU1, and SMF2).

View this table:

Genes whose haploid deletion mutants show enhanced resistance to caspofungin

Transcription and protein synthesis: A set of 11 nonessential and 3 essential genes in this group shows increased caspofungin sensitivity when deleted. They are involved in all stages from transcription to translation: histone acetylation (SPT10), chromatin remodeling (ARG82 and SNF2), transcription regulation (SWI6), RNA-polymerase I and II transcription (RPB3, RRN6, and THP1), RNA processing and transport (CDC40, NPL3, and YHR085W), and regulation of translation (ARD1, RSM27, and ZUO1). The absence of the Crz1p transcription factor or the ribosomal protein L18B leads to increased resistance to caspofungin.

Genes of other and unknown function: Deletion in 6 and 13 other genes was found to confer hypersensitivity and resistance, respectively (Tables 3 and 4).


A network of genetic interactions with FKS1, FKS2,GAS1, and SMI1

By compiling the synthetic genetic interactions of a set of mutants with defects in β-1,3-glucan synthesis, we have generated a network of 135 genes involved in 195 interactions.

A set of genes compensates for defects in glucan synthesis: Many (51/135, 37%) of the genes in the glucan network are connected to more than one query gene. This core set of genes is engaged in 112/195 (57%) interactions and 10 genes interact with FKS1, GAS1, and SMI1. FKS1 shares 43/76 interactions with GAS1 or SMI1, GAS1 shares 24/48 interactions with FKS1 or SMI1, while SMI1 shares 45/71 interactions with FKS1 or GAS1. The majority of multiply connected genes (62%) occupy two categories: synthesis and regulation of the cell wall (18 genes) and polarization and secretory pathway function (14 genes). This reflects the underlying coordination of polarized growth and cell wall assembly in the mitotic cell cycle. The regulation and orchestration of these processes depend on the integrity of the actin cytoskeleton, the cell polarity machinery, and a functional PKC1-SLT2 pathway. Other categories represented are ion homeostasis and signal transduction, cell cycle, and ubiquitin-related protein degradation. Importantly our glucan network core identifies genes of unknown function that appear central to the buffering of glucan defects and that are likely new components of the pathways discussed above. These include ILM1, MMS22, RIM21, YGL046W, YGL110C, YML117W, YOL003C, and YPL041C.

Cell integrity pathway: Eight of the 10 genes interacting with FKS1, SMI1, and GAS1 are involved in regulation of cell wall assembly through the cell integrity pathway. These genes encode components (BEM2, ROM2, and SLT2) or downstream targets (RLM1 and SWI4) of the PKC1-SLT2 pathway. In addition, CNB1 and PHO85 act in concert with this pathway under stress conditions (Zhaoet al. 1998; Huanget al. 2002). Furthermore, overactivation of the HOG signaling pathway by deletion of the protein phosphatase PTC1, as well as deletion of the glycerol channel FPS1, is deleterious to fks1 and smi1 mutants. These findings support the view that the HOG and PKC1-SLT2 pathways play opposing roles in regulating cell wall synthesis (Reynoldset al. 1998).

Chitin compensation: CHS3 and CHS5 both interact synthetically with FKS1, SMI1, and GAS1, as Chs3p-dependent chitin synthesis compensates for stress generated by defects in glucan synthesis (Garcia-Rodriguezet al. 2000b; Valdiviesoet al. 2000; Carottiet al. 2002). This chitin stress response is regulated, at least in part, by components of the PKC1-SLT2 pathway (Mazzoniet al. 1993; Igualet al. 1996; Valdivia and Schekman 2003). In addition, a group of genes (BNI1, ELM1, RVS167, SLA1, SPA2, and VPS67) required for normal polarized growth and morphogenesis interact with both FKS1 and SMI1, suggesting that compensatory chitin synthesis at the bud neck is essential in both fks1 and smi1 mutants. These gene products may also participate in the targeting of other cell wall synthesis components.

Gene-specific interactions: FKS1 interacts specifically with FKS2 and the double mutant is synthetically lethal. Since deletion of FKS1 triggers expression of FKS2, a fraction of FKS1-specific interactions involve genes required for FKS2 expression or for Fks2p function. For example, Crz1p and Snf1p are both positively involved in FKS2 induction (Stathopoulos and Cyert 1997; Zhaoet al. 1998). In this category are genes required for or induced during stress responses (IMP2/YIL154C, IXR1, HIT1, QRI5, YDJ1, YJR046W, and YMR073C). Many of these genes are important for the oxidative stress response, a process known to involve calcium signaling through calcineurin (Serranoet al. 2002), and thus, could influence FKS2 expression. FKS1 also interacts with genes involved in transcription, RNA processing, and translation, again suggesting that their buffering of Fks1p loss is through an altered Fks2p level. Finally, genes involved in endocytosis (EDE1 and SHE4) and cell polarity (CSR2 and HBT1) that interact with FKS1 may be required for cellular targeting of Fks2p.

The SMI1 deletion is buffered by genes acting in different areas of cell wall assembly, such as crosslinking glucan fibrils (GAS1), efficient β-1,6-glucan synthesis (KRE1), or crosslinks between β-1,3-glucan and Van1p- and Mnn10p-dependent protein mannosylation. In addition, SMI1 shows interactions with genes required for chromosome segregation (ARP1, CTF4, and CTF8) and polarity establishment (CLA4 and BEM1), two processes requiring bud neck integrity. This suggests that a SMI1 deletion results in defective bud neck assembly or function.

Survival of a gas1 null mutant appears to require the correct synthesis and assembly of cell wall β-1,6-glucan. We found a set of gas1-interacting mutants in genes affecting this process or resistance to K1 killer toxin, which requires β-1,6-glucan as a receptor (BUD19, CWH41, IMG1, KRE1, KRE11, NBP2, PHO85, RGP1, RIC1, ROT2, RSC2, SMI1, VPS61, VPS63, and YPT6; see Pageet al. 2003). A number of potential cell wall regulating genes also interact with GAS1 and are candidates for involvement in β-1,6-glucan biology. These include ECM7 (Lussieret al. 1997; Giaeveret al. 2002), YAL053W (Lagorceet al. 2003), and SSD1, a regulator of cell wall composition (Kaeberlein and Guarente 2002).

Genetic interactions with FKS2 and FKS3: In contrast with the many interactions found for FKS1, FKS2 interacted only with FKS1, and no interactions were found with FKS3. Differential expression of these genes likely underlies their interaction patterns. FKS1 is expressed during vegetative growth on glucose, a growth condition where the FKS2 transcript is largely undetectable. FKS2 is, however, induced under specific conditions such as starvation, stress, and in stationary phase (Mazuret al. 1995). Little is known about FKS3 function; its expression is regulated by Ste12p upon pheromone exposure (Zeitlingeret al. 2003), and the fks3 null mutant shows a slight sporulation defect (Deutschbaueret al. 2002).

Functional links between glucan and chitin synthesis: Defective β-1,3-glucan assembly is compensated for by an increased synthesis of chitin. In our synthetic analysis of Chs3p-dependent chitin synthesis, we found that, as with glucan mutants, this is largely buffered by genes involved in the regulation of cell wall assembly and secretory pathway polarization. Indeed almost half of the genes interacting with CHS3 are found in the glucan network, highlighting their common function in buffering the cell wall from adversity. A significant overlap of CHS3 interactants with FKS1 (16 genes, P = 5 × 10–62) and SMI1 (17 genes, P = 1 × 10–66) interactants was found, with 11 genes interacting with FKS1, SMI1, and CHS3 (BNI1, BRE5, DOC1, ILM1, PRE9, RVS167, SLA1, SLT2, SWI4, VPS67, and YNL171C), 5 genes interacting with CHS3 and FKS1 (EDE1, HBT1, RPL20B, SHE4, and YLR338W), and 6 genes interacting with CHS3 and SMI1 (ASC1, BCK1, CLA4, GAS1, MNN10, and VAN1). This further indicates that proper localization of cell wall building components, through polarization of the secretory apparatus, is essential in achieving a balance of chitin and glucan levels.

Genes involved in caspofungin sensitivity

A synthetic-lethal analysis reveals pairwise interactions among genes. Application of this approach to the FKS gene family is complicated by the need to compare more complex combinations of mutants. In this situation, a drug inhibiting a protein family offers a powerful alternative “chemogenomics” strategy. As caspofungin differentially inhibits both Fks1p and Fks2p, targets that are singly dispensable but together are essential, the basis for phenotypes of deletion mutants with altered sensitivity to this drug is likely to be complex. In general, deletion of genes required to maintain Fks1p and Fks2p activity would lead to lower glucan synthase activity levels and hypersensitivity to caspofungin. In addition, the absence of genes whose products buffer cells from loss of glucan synthesis would be more vulnerable to such loss, and thus caspofungin hypersensitive. In this case, the mutant-caspofungin interaction can be viewed as being “synthetic” (Parsonset al. 2004). As the complete loss of the Fks1p and Fks2p targets is lethal, resistance of this kind cannot occur, but is possible with mutant alleles (Douglaset al. 1994; Mazuret al. 1995) or if targets are overproduced (Rineet al. 1983). Altered sensitivity to caspofungin can also arise through detoxification by vacuole enzymes or mutant defects that affect membrane permeability and hence accessibility of the drug to its targets.

Caspofungin toxicity and regulation of glucan synthesis: The two FKS targets show different levels of sensitivity to this drug class, with Fks2p being more sensitive to echinocandin and aerothricin than Fks1p (Mazuret al. 1995; Kondohet al. 2002). Our work accords with these findings, with fks1 and fks2 mutants being caspofungin hypersensitive and resistant, respectively. Consistent with this, the crz1 mutant, known to be defective in FKS2 induction (Stathopoulos and Cyert 1997), is more resistant to caspofungin than the wild type. Mutants in regulatory components affecting glucan synthase activity show a complex set of responses. The slg1/ wsc1 mutant defective in Rho1p-dependent activation of Fks1p/Fks2p (Mazur and Baginsky 1996; Sekiya-Kawasakiet al. 2002) is caspofungin resistant, as is a tus1 mutant that is also implicated in Rho1p signaling. In contrast, the ability to activate the Rho1p-dependent PKC1-SLT2 pathway buffers cells against caspofungin, as slt2 and smi1 mutants are hypersensitive. However, deletion of SSK2, leading to defective activation of the HOG pathway, as well as that of HLR1, a multicopy suppressor of osmosensitivity of a ste11ssk2ssk22 triple mutant (Alonso-Mongeet al. 2001), confers resistance to caspofungin. These findings, together with results on calcofluor white sensitivity (Garcia-Rodriguezet al. 2000a) highlight the crosstalk between the cell integrity and the HOG pathways and the importance of coordinating these opposing signaling pathways for cell wall assembly.

Figure 3.

—Overlap of the FKS1/FKS2/SMI1 genetic interaction network and the caspofungin chemical-genetic network. (A) Venn diagram summarizing the number of genes leading to altered caspofungin sensitivity when deleted and/or showing synthetic interaction with FKS1, FKS2, or SMI1. The numbers of genes not classified as multidrug sensitive are indicated in small font. (B) Network of chemical genetic interactions with caspofungin and the genetic interactions with FKS1, FKS2, and SMI1.

Genes involved in processes compensating for the inhibition of a target are required for survival in presence of a drug. For example, chitin synthesis is upregulated by cell wall stress, and as with the genetic interaction data, deletion of CHS3 or the ancillary genes (CHS47) leads to caspofungin hypersensitivity. Deletion of components of the endocytic pathway (RCY1 or SLA1) also leads to a moderate increase in caspofungin sensitivity. Thus, transport of cell surface components likely buffers perturbed cell wall synthesis in caspofungin-treated cells, with defects in the proper recycling of these components resulting in increased drug sensitivity. For instance, cortical actin patches are important for dynamic Fks1p localization, with cell wall remodeling and SLA1 deletion resulting in mislocalization of Fks1p (Liet al. 2002; Utsugiet al. 2002). In this context, the dynamics of Fks1p and/or Chs3p localization in an rcy1 null mutant merit examination.

View this table:

Genes whose transcription is increased and whose deletion mutant shows altered growth in a fks1, smi1, or gas1 null background or upon caspofungin exposure

Comparison of the synthetic and chemical-genetic networks

Both the synthetic interaction and the caspofungin phenotype data sets should identify genes involved in buffering cells against defective β-1,3-glucan synthesis. However, each set has limitations, such as gene family issues with the synthetic interactions and the multidrug sensitivities and possible “off target” side effects of caspofungin. Integration of the 189 genes in the two networks (Figure 3A) shows a central overlapping core of 14 genes. Of these 14 genes with altered caspofungin sensitivity, 11 show synthetic interactions with FKS1 and 9 with SMI1 (Figure 3B). The overlap of these two sets is highly significant (P = 1 × 10–39 and 9 × 10–33 for FKS1 and SMI1, respectively) and is consistent with caspofungin acting at the level of Fks1p and Fks2p in inhibiting β-1,3-glucan synthesis.

Deletion mutants in 98 genes of the glucan network have wild-type sensitivity to caspofungin (Figure 3A), indicating that caspofungin treatment does not pheno-copy cell wall mutations. This may be because at the subMIC concentration used here, caspofungin does not fully inhibit its target. A prediction of this is that the viable synthetic double mutants should show enhanced caspofungin sensitivity when compared to the single mutants. Finally, a set of 77 genes whose deletion alters caspofungin sensitivity is absent from the glucan network (Figure 3A). Mutation in a fraction of these could affect growth of an fks1 fks2 double mutant but not that of the singly deleted mutants. In addition, some of these genes may actually show synthetic interactions, and be false negatives (see materials and methods), while others may buffer against off target side effects of caspofungin.

Genetic interactions, fitness under stress condition, and transcription

We have compared our data on synthetic interactions and altered caspofungin sensitivity with the relevant transcriptional profiling data (Table 5). Three genes found in our functional core (Figure 3B), CHS3, FKS2, and SLT2, show altered transcriptional profiles. These three core genes capture much of glucan buffering, namely the need for compensatory chitin, an alternative glucan synthase component, Fks2p, and a cell integrity signal transduction pathway. Other genes found in the functional network that are transcriptionally regulated are KRE11, YAL053W, and GPH1.

It is striking that most genes that genetically buffer fks1, smi1, and gas1 mutants or lead to altered caspofungin sensitivity show no changes in transcriptional levels. Thus, it appears that most genes functionally involved in responding to glucan defects do so in ways that are transcription independent. Presumably, the existing cellular location and activity of these gene products coupled with their normal levels of synthesis are sufficient to achieve an effective buffering.

Our work also emphasizes that many transcriptionally regulated genes have no apparent effect on fitness during perturbation of β-1,3-glucan synthesis, a situation seen previously for a range of conditions (Giaeveretal. 2002). This indicates that the yeast repertoire of transcriptional responses may be limited and stereotyped. Here, no β-1,3-glucan-specific response is invoked, but rather a more general response occurs, in which only a fraction of genes are functionally effective, but where the stereotypic response set has been evolutionarily selected as a “tool box” to cope with a more broadly based set of insults.

Issues of drug resistance

Drug resistance is a major clinical issue. Our work on resistance is confined to null mutants in nonessential genes and so is not comprehensive; for example, point mutants in FKS1 or FKS2 leading to resistance would not be seen. However, despite these limitations, we found 39 S. cerevisiae genes leading to decreased caspofungin sensitivity when deleted (see Table 3 and supplementary Table 1). In particular, deletion of several genes encoding putative membrane-associated proteins with unknown function led to enhanced resistance to caspofungin; they could encode either additional targets for caspofungin (YDR326C and YDR479C) or proteins mediating effects of the drug (YBR144C, YGR283C, YIL110W, and YPL056C). Mutations in the fungal pathogen orthologs of these 39 genes could lead to increased resistance.


We thank Charles Boone, Huiming Ding, and Ainslie Parsons for making unpublished results available, helpful discussions, and comments on the manuscript. This work was supported by Genome Canada and Genome Quebec and by a Discovery grant from the Natural Sciences and Engineering Research Council of Canada.


  • Communicating editor: M. Johnston

  • Received October 28, 2003.
  • Accepted January 14, 2004.


View Abstract