Multiple Functional Domains of the Yeast l,3-β-Glucan Synthase Subunit Fks1p Revealed by Quantitative Phenotypic Analysis of Temperature-Sensitive Mutants
Hiroki Okada, Mitsuhiro Abe, Masayo Asakawa-Minemura, Aiko Hirata, Hiroshi Qadota, Kana Morishita, Shinsuke Ohnuki, Satoru Nogami, Yoshikazu Ohya

Abstract

The main filamentous structural component of the cell wall of the yeast Saccharomyces cerevisiae is 1,3-β-glucan, which is synthesized by a plasma membrane-localized enzyme called 1,3-β-glucan synthase (GS). Here we analyzed the quantitative cell morphology and biochemical properties of 10 different temperature-sensitive mutants of FKS1, a putative catalytic subunit of GS. To untangle their pleiotropic phenotypes, the mutants were classified into three functional groups. In the first group, mutants fail to synthesize 1,3-β-glucan at the proper subcellular location, although GS activity is normal in vitro. In the second group, mutants have normal 1,3-β-glucan content but are defective in polarized growth and endocytosis. In the third group, mutations in the putative catalytic domain of Fks1p result in a loss of the catalytic activity of GS. The differences among the three groups suggest that Fks1p consists of multiple domains that are required for cell wall construction and cellular morphogenesis.

THE cell wall is a rigid extracellular structure in plants and fungi that defines cell morphology. In the cell wall of the yeast Saccharomyces cerevisiae, the major structural components are 1,3-β-glucan, 1,6-β-glucan, chitin, and mannoproteins (Cabib et al. 2001). 1,3-β-glucan, which is the most abundant component of the cell wall, is essential for the rigidity of yeast cells.

The enzyme responsible for synthesis of 1,3-β-glucan is 1,3-β-glucan synthase (GS). Yeast GS is composed of at least two subunits: a putative catalytic subunit and a regulatory subunit. The catalytic subunit is Fks1p, a membrane-localized protein (Douglas et al. 1994). Fks1p and its alternative protein Fks2p share 88% similarity, including the region of the putative catalytic domain (Mazur et al. 1995; Kelly et al. 1996). The simultaneous deletion of both genes results in a lethal phenotype, indicating that yeast GS is essential (Inoue et al. 1995). Fks1p is localized to the plasma membrane at the site of cell wall remodeling (Utsugi et al. 2002). The regulatory subunit is Rho1p, a Rho-type small GTPase, which is essential for vegetative growth (Drgonova et al. 1996; Mazur and Baginsky 1996; Qadota et al. 1996). Rho1p acts as a molecular switch and its GTP-bound form specifically activates GS.

In addition to its function in 1,3-β-glucan synthesis, Fks1p has been suggested to have other functions, since deletion of FKS1 results in pleiotropic defects not only in alteration of cell wall structures such as 1,6-β-glucan (Dijkgraaf et al. 2002), mannoproteins (Ram et al. 1995), and chitin (Dallies et al. 1998; Garcia-Rodriguez et al. 2000), but also in endocytosis, which is unrelated to cell wall biogenesis (deHart et al. 2003).

Because GS and its subunits have multiple functions and their absence causes pleiotropic defects, the physiological functions of GS in cell morphogenesis have remained unclear. Conditional-lethal mutations have been induced in the regulatory subunit of GS and characterized to elucidate the function of Rho1p (Yamochi et al. 1994; Ozaki et al. 1996; Qadota et al. 1996; Drgonova et al. 1999; Saka et al. 2001; Roh et al. 2002). It has been revealed that the different rho1 mutations cause distinctive effects on cellular growth, likely because Rho1p binds to and activates some effectors including Fks1p and Fks2p (Mazur and Baginsky 1996; Inoue et al. 1999). In contrast, few conditional-lethal mutations in the catalytic subunit of GS have so far been investigated (Dijkgraaf et al. 2002; Suzuki et al. 2004).

To untangle the complex effects of loss of gene functions, quantitative and high-resolution phenotypic analysis is a powerful approach. Yeast cell morphology is one of the most useful phenotypes for such analysis because it reflects various cellular events, such as cell cycle progression, establishment of cell polarity, and regulation of cell size (Ni and Snyder 2001; Giaever et al. 2002; Jorgensen et al. 2007). Previously, we developed an image processing program named CalMorph that automatically processes digital images of each yeast cell into quantitative data (Ohtani et al. 2004). Cells were photographed with simultaneous staining of mannoproteins (as a cell wall component marker), actin patches, and nuclear DNA for quantification of cell morphology. By using this system, we analyzed 4718 yeast nonessential gene deletion mutants and showed that approximately half of them have altered cell morphology (Ohya et al. 2005). Moreover, it has been reported that hierarchical cluster analysis based on similarities in cell morphology is a useful method for elucidating functional correlations of gene products (Ohnuki et al. 2007).

In this study, to understand the physiological functions of GS, we isolated 10 different temperature-sensitive (ts) fks1 alleles in an fks2Δ background and characterized the phenotypes associated with the mutations. Quantitative and high-resolution phenotypic analysis of the fks1 mutants revealed that they could be statistically classified into three groups. Our results suggested that Fks1p consists of multiple domains required for cellular morphogenesis.

MATERIALS AND METHODS

Media and strains:

Yeast growth, tetrad analysis, and mating-type determination were performed as described previously (Kaiser et al. 1994). Yeast transformation was carried out using the lithium acetate method (Ito et al. 1983). Yeast cells were grown in rich media (YPD) [1% Bacto yeast extract (Difco, Detroit), 2% polypeptone (Wako Chemicals, Osaka, Japan), and 2% glucose (Wako Chemicals)] or in synthetic growth media (SD) [0.67% yeast nitrogen base without amino acids (Difco) and 2% glucose] supplemented appropriately. For −Ura or −Trp selection, 0.5% casamino acid (Difco) was added to SD. To induce the expression of GAL1 promoter-driven FKS1, 2% galactose and 0.1% sucrose were used as carbon sources instead of glucose (SGS). Standard procedures were used for all DNA manipulations and Escherichia coli transformation (Sambrook et al. 1989). The yeast strains used are listed in Table 1. All strains are isogenic derivatives of YPH499, YPH500, or YPH501 (Sikorski and Hieter 1989). YOC792 and YOC793 were constructed by disruption of both FKS1 and FKS2 in YPH499 and YPH500 carrying a plasmid pYO916 (GAL1 promoter-FKS1, URA3 marker), respectively.

View this table:
TABLE 1

Yeast strains used in this study

Plasmids:

The plasmids used in this study are listed in Table 2. pYO973 was generated by cloning the SphI-NheI fragment of YEp-GSC1 into pSN314. pYO975 was generated by replacing the MscI-SalI region of pYO973 with an MscI-SalI linker. pYO976 was generated by replacing the SalI-SacI region of pYO973 with an SalI-SacI linker. pYO977 was generated by replacing the SacI-AatII region of pYO973 with a SacI-AatII linker. pYO978 was generated by replacing the AatII-NarI region of pYO973 with an AatII-NarI linker. pYO979 was generated by deleting the BglII-BglII region of pYO973. pYO980 was generated by blunting the KpnI site of pRS314. pYO981 was generated by cloning the SphI-NheI fragment of pYO1751 (Inoue et al. 1995), containing the genomic FKS1 gene, into pYO980. pYO982 was generated by deleting the KpnI-KpnI region of pYO981. Construction of plasmids pYO2163–pYO2173 is described below.

View this table:
TABLE 2

Plasmids used in this study

Isolation of ts fks1 mutants:

To isolate ts fks1 mutants, the random mutations were introduced across the entire FKS1 region by the error-prone PCR method (Cadwell and Joyce 1992). The procedure involves amplification of the FKS1 region under mutagenic PCR conditions. The mutagenic PCR cycle is as follows: (94° for 20 sec, 68° for 4 min) × 30 cycles, (92° for 1 min, 60° for 1 min, 72° for 2 min) × 30 cycles, and (75° for 10 min) × 1 cycle. The amplified PCR fragment and KpnI-digested vector (YOC982) were used for cotransformation of YOC792 (fks1Δ fks2Δ and carrying the pYO916 plasmid). About 8600 transformants were obtained on glucose plates (−Trp, −Ura). Next, 1000 transformants were chosen, streaked on YPD plates, and incubated at 23° and 37° to screen for ts mutants. The candidates were selected on a plate containing 5-fluoroorotic acid (5-FOA) to eliminate the pYO916 plasmid. After rescue of the mutagenized plasmids, these plasmids were used for retransformation of YOC792 to confirm the phenotype. Finally, 17 alleles were isolated.

Subcloning of ts fks1 alleles:

To determine the mutation(s) that is responsible for the temperature sensitivity, the fks1 ts alleles were subcloned into plasmids pYO975–pYO979 to produce fks1 alleles that were mutated only in the restricted regions. The resultant plasmids were transformed into the YOC792 strain, and the transformants were selected on SGS−Trp plates at 23°. Transformants were patched onto duplicate YPD plates, and the plates were incubated at 23° and 37°. The region responsible for the temperature sensitivity of the original fks1 allele was determined by checking the temperature sensitivity of the subclones.

Out of the 17 original fks1 alleles, 10 showed temperature sensitivity after subcloning. Plasmids carrying the ts mutations were digested with SphI and NheI, and the resultant fragment was cloned into plasmid pYO969 to produce plasmids pYO2163–pYO2173. These plasmids were digested with SacII and transformed into YOC792 to integrate the fks1 subcloned allele into the ADE3 locus. The transformants were selected on SGS−Trp + 1/3Ade plates and white colonies were selected for appropriate integrants. The colonies were patched onto the plate containing 5-FOA to eliminate the original pYO916 plasmid. The resulting strains (YOC1001, -1002, and -1071–1090) were used for further analysis.

Quantification of cellular morphology:

To reduce the possibility that cell morphology was changed by unexpected mutation(s), we quantified the cell morphology of homoallelic diploids of 10 fks1 mutants (YOC4318–4328). Yeast strains were cultured at 25° in YPD media until the early log phase and shifted for 4 hr at 35.5°. After incubation, cells were fixed and specifically stained for three components: fluorescein isothiocyanate concanavalin A (FITC-Con A) (Sigma, St. Louis), rhodamine–phalloidin (Invitrogen, Carlsbad, CA), and 4′,6-diamidino-2-phenylindole (Wako Chemicals) to stain the mannoprotein, actin, and nucleus, respectively. Cells were observed and images were captured using AxioImager M1 with a 100× Plan-Apochromat objective lens (Carl Zeiss, Oberkochen, Germany) equipped with a CoolSNAP HQ cooled-CCD camera (Roper Scientific) and AxioVision software (Carl Zeiss). The images were analyzed by the image processing software CalMorph version 1.3 as described previously (Ohya et al. 2005).

Data preprocessing:

To generate a uniform distribution of each parameter value, we transformed the parameter values to Ri statistics (the sum of the ranks in the ith sample) of the Kruskal–Wallis test (Kruskal and Wallis 1952) and modified them to reflect the direction of morphological change. All the observations were ranked together with the lowest values first. Note that the same values were ranked in average order. Then, we summed the rank order of samples to Ri and computed Ri′, which is a modified Ri statistic defined by the functionMath

where ni is the number of observations in the ith sample and N is the number of observations in all samples combined (Σni).

Parameter reduction:

Because some of the 501 parameters that are outputs of CalMorph (detailed in SCMD, http://scmd.gi.k.u-tokyo.ac.jp/datamine/) are not independent (Ohya et al. 2005), we reduced parameters by eliminating redundancy. The redundancy among parameters was estimated by Spearman's rank correlation coefficient (R) between an arbitrary 2 of 501 parameter values obtained by quantification of cellular morphology of the wild-type (WT) strain (n = 20). Two hundred twenty-one of the 501 parameters were regarded to be correlated with at least one other parameter at R > 0.9 (P < 0.01 with Bonferroni correction). The 221 parameters were classified into 75 groups where each group consisted of the parameters whose correlation coefficient to each other is >0.9. Then, the representative parameter of each group was defined as a parameter whose sum of the R values was maximum among parameters of each group. Finally, we combined the representative 75 parameters with the resting 280 of 501 parameters into 355 parameters that were used in this study (supporting information, Table S1).

Statistical analysis:

All statistical analyses of the quantified morphological data were performed using R ver. 2.7.0 (http://www.r-project.org/).

Measurement of GS activity of the membrane fraction:

The membrane fraction was prepared as described previously (Abe et al. 2001). In brief, log-phase cultures were resuspended in 1 mm EDTA and 500 mm NaCl containing 1 mm phenylmethylsulfonyl fluoride and lysed with glass beads. The crude lysate was centrifuged at 1500 × g for 5 min so that cell debris and unbroken cells were separated. After centrifugation at 100,000 × g for 30 min, the pellet was suspended in a buffer containing 50 mm Tris–HCl (pH 7.5), 1 mm EDTA, and 33% glycerol, and the suspension was taken as the membrane fraction. GS activity was measured according to the procedure described previously (Inoue et al. 1995).

Incorporation of [14C]glucose into 1,3-β-glucan:

As described previously (Abe et al. 2003), cells were grown to early log phase at 25° and were then cultured at either the permissive or the restrictive temperature for 2 hr. The cultured cells were diluted to OD600 of 0.5 with 1 ml of 0.5% glucose media containing 10 μCi of [14C]glucose and labeled for 2 hr. The labeled cells were harvested and extracted with 1 N NaOH at 80° for 30 min. The insoluble pellets were resuspended in 10 mm Tris–HCl, pH 7.5, containing 5 mg/ml Zymolyase 100T (Seikagaku, Tokyo) and incubated at 37° for 20 hr. After digestion, the zymolyase-resistant material was removed by centrifugation (15,000 × g for 20 min) and the supernatant was filtered through an Amicon Centricon YM-10 membrane (molecular weight cutoff is 10,000; Millipore, Bedford, MA). The flow-through fraction was dried by a vacuum evaporator and applied to Unifilter-GF/C (Packard Instrument Co.). The differences in the incorporation rates in strains were normalized by ΔOD600, measured before and after the labeling period. Since a significant value of ΔOD600 was required for the normalization, we preferred to use the mild condition (34°) as the restrictive temperature.

Morphological observations:

For cell wall staining of yeast cells, early log-phase cells were harvested by centrifugation. For aniline blue staining, the cells were washed twice with PBS and mildly sonicated for 10–20 sec. The washed cells were incubated in 0.05% aniline blue (Wako Chemicals) for 5 min and observed by fluorescence microscopy with an Olympus (Tokyo) U-MNV DM455 filter set (excitation wavelength, 400–410 nm; emission wavelength, 455 nm). For calcofluor white staining, the cells were washed twice in distilled water (DW) and mildly sonicated. The washed cells were incubated in 1 mg/ml calcofluor white (Sigma) for 5 min, washed twice with DW, and then observed. For FITC-Con A staining, the cells were washed twice with P buffer (10 mm sodium phosphate, pH 7.2, 150 mm NaCl). The washed cells were incubated in 0.1 mg/ml FITC-Con A in P buffer for 10 min, washed twice with P buffer, and then observed.

Cells were viewed on an Olympus BX-FLA microscope or a Leitz DMR microscope (Leica, Wetzlar, Germany). Images were captured using a CoolSNAP HQ CCD camera (Nippon Roper, Tokyo) and Metamorph Imaging software (Universal Imaging). All images were processed for publication using Adobe Photoshop software.

Immunoelectron microscopic analysis:

Thin sections of yeast cells were prepared by the freeze-substituted fixation method as described previously (Abe et al. 2003). For 1,3-β-glucan immunolabeling, a mouse monoclonal antibody against 1,3-β-glucan (Biosupplies, Parkville, Australia) and a secondary antibody conjugated with 10-nm gold particles were used. The labeled thin sections were viewed under an H7600 electron microscope (Hitachi, Tokyo) at 100 kV.

RESULTS

Isolation of 10 fks1 ts mutants:

Yeast strains harboring fks1 ts mutations were generated by random mutagenesis of the FKS1 gene (see materials and methods). Although all the mutants grew normally at 23°, they failed to grow above 35.5° (Figure S1). Subcloning and sequence analysis revealed that each mutant possessed from one to five mutations conferring temperature sensitivity (Table 3).

View this table:
TABLE 3

Mutation points of the ts fks1 mutants

Cluster analysis of the cell morphology of the fks1 mutants:

To classify the fks1 mutants, we applied hierarchical cluster analysis based on cell morphology of mutants and the FKS1 WT strain. Homoallelic diploids were incubated at 35.5°, stained with fluorescent dyes, and photographed (see materials and methods). In each sample, values for 501 morphological parameters were obtained by the CalMorph image-processing program (Figure S2). The experiments were replicated 5 times for mutants and 20 times for the WT. For using cluster analysis, we averaged the replicated data for each parameter value of each strain to obtain a standardized value (Ohnuki et al. 2007, see materials and methods). By hierarchical cluster analysis of the 501 parameters, we identified three classes with an approximately unbiased probability value (AU P-value: calculated using a multiscale bootstrap technique) (Suzuki and Shimodaira 2006) >0.95 (Figure 1A). Class I contains fks1-1082, fks1-1132, and WT; class II contains fks1-1163, fks1-1014, and fks1-1104; and class III contains fks1-1114, fks1-1144, fks1-1154, and fks1-1125. Only fks1-1093 did not belong to any of these classes. Interestingly, three of four class III mutants (fks1-1114, fks1-1144, and fks1-1154) possessed mutations in the putative catalytic domain (aa 829–973) (Kelly et al. 1996).

Figure 1.—

Cluster analysis of the fks1 mutants based on similarities in cell morphology. (A) Three classes detected by cluster analysis. Dissimilarity indicates a positive angle (0°–180°) between the vectors of the 501 parameter values. Blue values indicate the AU P-value (calculated with the multiscale bootstrap technique) in the dendrogram. Orange rectangles indicate robustly clustered mutant classes with an AU P-value >0.95. Red and green boxes indicate modified Ri′ statistics reflecting morphological differences between cells (see materials and methods). Positive and negative values of the modified Ri′ statistic are depicted in red and green, respectively. Class indicates a mutant group that shows similar morphological features. (B) Coherently changed parameters in each class. The red and green boxes indicate significantly increased and decreased parameters at a significance level of 0.01, respectively (Ohnuki et al. 2007).

In vitro GS activity of the fks1 mutants:

The GS activity in a membrane fraction of the class III (fks1-1114, fks1-1144, fks1-1154, and fks1-1125) cells was virtually reduced after cultivation at 23° and 37° (Figure 2). Kinetic analysis of GS activity in these four mutants revealed a decreased Vmax value at 37° (Table 4). Immunoblotting analysis with an anti-Fks1p antibody revealed that the Fks1p levels of these four mutants were normal (data not shown), suggesting that none of the mutations affect Fks1p stability. These results suggested that the four class III mutants had lost the catalytic activity of GS. In contrast to these mutants, the other mutants did not show a clear decrease in the GS activity or the Vmax value, suggesting that these mutations affect activities of GS other than in vitro glucan synthesis.

Figure 2.—

In vitro GS activity of membrane fractions isolated from fks1 cells. Membrane fractions were isolated from cells cultured at 23°, and GS activity was assayed at 23° (left). Membrane fractions were isolated from the cells cultured at 23° and shifted to 37° for 2 hr, and GS activity was assayed at 30° (right). The data represent the means and standard deviations of at least four experiments.

View this table:
TABLE 4

Km and Vmax values of GS activity

Correlation between in vitro GS activity and cell morphology:

To clarify the relationship between mutation sites and GS activity, the GS activity of all mutants at the restrictive temperature was plotted along with the Fks1p protein sequence (Figure 3). In the diagram, mutants in the same class appeared near one another. Mutations of class I, which showed GS activity equivalent to WT, were located near amino acid 300. Although the class II and III mutations overlapped in location, they displayed difference in GS activities. Taken together, these data indicated that GS activity and mutation site location influence cell morphology.

Figure 3.—

Scatter plot of in vitro GS activity against fks1 mutation sites. Small circles indicate mutation sites of fks1 mutants. Circles enclosed by dotted lines indicate the classes detected and shown in Figure 1. The red arrow indicates the GS activity value of Fks1p (WT). The top parts and the bottom parts of white boxes indicate predicted exoplasmic and cytoplasmic regions, respectively. Black bars indicate predicted transmembrane regions. The small gray box indicates the putative catalytic domain.

To find cell morphological traits that are affected by GS activity, we scanned correlations between GS activity and the morphological parameters of all strains. To avoid redundancy among all 501 parameters, we reduced parameters according to the method described previously (Ohya et al. 2005) (Table S1, see materials and methods). Then, 355 parameters were selected for further analysis. By linear regression analysis between values of GS activity at 37° and 355 parameters, 6 parameters were detected with a squared correlation coefficient (R2) > 0.718 (P < 0.001) (Table 5); the false discovery rate (FDR) estimated that <1 such parameter would have been detected by chance (Storey et al. 2004). Among the 6 parameters, 3 parameters were related to actin localization. For instance, a scatter plot of GS activity values vs. the values of morphological parameter A114 (proportion of cells with delocalized actin patches in unbudded cells, Figure 4) indicates a negative correlation between GS activity and actin localization. To confirm this result, we quantified the cell morphology of WT cells following echinocandin B (GS inhibitor) treatment with concentrations of 0, 1, 2, 3, and 4 mg/liter (n = 5) (S. Ohnuki, S. Oka, S. Nogami and Y. Ohya, unpublished results). The parameter A114 was detected to show a significant dose-dependent increase at P-value <0.01 by a Jonckheere–Terpstra trend test. Another actin-related parameter, A7-1_A1B (proportion of actin region in budded and mononucleic cells), also showed a significant dose-dependent increase. These results indicated that the GS activity influences actin localization.

Figure 4.—

Scatter plot of parameter A114 (proportion of cells with delocalized actin patches in unbudded cells) against in vitro GS activity. Small circles indicate values of different fks1 strains. Circles enclosed by dotted lines indicate the classes detected in Figure 1. The squared correlation coefficient (R2) of a linear regression model was 0.858 (P < 0.001).

View this table:
TABLE 5

Morphological parameters correlated with GS activity

Morphological alteration in the fks1 mutants:

To test for other functional defects in each mutant, we characterized the cell morphology of each class as shown in Figure 1A. We searched for morphological parameters showing similar values within the same class according to the method described previously (Ohnuki et al. 2007) (P < 0.001, Figure 1B). The numbers of such parameters were 96 in class I, 84 in class II, and 109 in class III; the FDR estimated that <1 such parameter of each class would have been detected by chance. For example, class I mutants shared a smooth elliptical cell shape, a high population of budded cells, and localized actin cytoskeleton; class II mutants shared a large depolarized cell shape, delocalized actin cytoskeleton in bud, a high population of cells with large buds, actin delocalization in bud, and irregular nuclear migration; class III mutants shared a small round cell shape, a high population of cells with small buds, and delocalized actin cytoskeleton (Figure S3). These differences in representative features may indicate the existence of distinct functional defects among the classes of fks1 mutants. Then, we chose three mutants from each class as class-representative mutants (fks1-1082 from class I, fks1-1163 from class II, and fks1-1154 from class III) for further analyses.

The class II and III mutants shared delocalized actin phenotype; we further observed the distribution of Spa2p, which localizes at the incipient bud site and bud tip and regulates polarized actin cytoskeleton (Ayscough et al. 1997). In the class II mutants, the Spa2p localization was lost (data not shown), suggesting that the class II mutants likely lose cell polarity leading to delocalized actin cytoskeleton.

In vivo 1,3-β-ghican synthesis in fks1 mutants:

To investigate in vivo 1,3-β-glucan synthesis, [14C]glucose incorporation was measured at 25° or after being shifted to 34° for 2 hr (Figure 5A). At 25°, [14C]glucose incorporation into 1,3-β-glucan was highly decreased in class III fks1-1154 cells. In contrast, at 34° 1,3-β-glucan synthesis was slightly reduced in fks1-1082 and fks1-1163 cells and significantly reduced in fks1-1154 cells. These results suggest that in vivo 1,3-β-glucan synthesis was mildly reduced in mutants of both classes I and II.

Figure 5.—

In vivo 1,3-β-glucan synthesis in fks1 mutants. (A) Incorporation of [14C]glucose into 1,3-β-glucan of fks1 mutants. Cells were incubated at 25° (solid bars) or shifted to 34° for 2 hr (open bars), labeled with [14C]glucose for 2 hr, and incorporation of glucose into 1,3-β-glucan was measured. Data represent the means and standard deviations from at least three experiments. (B) Cell wall staining in fks1 mutants. WT, fks1-1082, fks1-1163, and fks1-1154 were incubated at 37° for 4 hr. 1,3-β-glucan, mannoproteins, and chitin were stained with aniline blue, FITC-Con A, and calcofluor white, respectively. All cells were stained without fixation.

To monitor the localization of synthesized 1,3-β-glucan, we stained cells with aniline blue (Figure 5B). The WT cells exhibited uniform staining of 1,3-β-glucan across the entire cell surface. At the restrictive temperature, the class I fks1-1082 cells appeared to lose staining specifically in the small bud. Although the class II fks1-1163 cells had abnormal morphology, the cells exhibited the normal staining pattern across the entire cell surface. The class III fks1-1154 cells displayed a tiny bud-like projection and lost aniline blue staining at this projection. These results suggested that the glucan in class I and III mutants is not synthesized at the proper locations.

Alteration of cell wall components in fks1 mutants:

To monitor the alteration of cell wall components, we stained mannoproteins and chitin, using FITC-Con A and calcofluor white, respectively (Figure 5B). Staining with FITC-Con A showed that signals were distributed uniformly on the cell surface at a higher level than WT, suggesting that the mutants have increased mannoproteins. Staining with calcofluor white showed that strong chitin-staining signal appeared in the class II fks1-1163 and the class III fks1-1154 cells but not in the class I fks1-1082 cells. These results suggested that cell wall structure in the fks1 mutants was altered.

Alteration was also observed in the ultrastructure of the cell wall in the fks1 mutants by electron microscopy (Figure 6). In the class II fks1-1163 and class III fks1-1154 cells, the thickness of the cell wall was increased, and an abnormal layer was observed other than the 1,3-β-glucan layer.

Figure 6.—

Immunoelectron microscopic observation of fks1 cells. (a) WT, (b) fks1-1163, and (c) fks1-1154 were shifted from 25° to 37° for 4 hr, prepared for electron microscopy by freeze-substituted fixation methods, and immunolabeled using an anti-l,3-β-glucan antibody. Bar, 1 μm.

Endocytosis defect in the fks1 mutants:

Since actin plays roles in endocytosis (Robertson et al. 2009), mutants defective in actin localization such as the class II and III mutants may have defective endocytosis. As expected, the class II fks1-1163 and class III fks1-1154 cells showed reduced levels of Lucifer yellow uptake in the vacuole, indicating that endocytosis is defective in these mutants (Figure 7). In contrast, in WT and class I fks1-1082 cells, Lucifer yellow was accumulated in the vacuole as a result of endocytosis. Defective endocytosis is known to induce a longer lifetime of Sla1-GFP (Kaksonen et al. 2003). The Sla1-GFP lifetime in fks1 mutants in classes II and III but not in class I was prolonged, suggesting defects in endocytosis (data not shown).

Figure 7.—

Endocytosis in fks1 mutants. (a) WT, (b) fks1-1082, (c) fks1-1163, and (d) fks1-1154 were incubated at 37° for 4 hr in YPD media and further incubated in YPD containing Lucifer yellow at 37° for 30 min. Cells were washed and observed with fluorescence microscopy.

Intragenic complementation between fks1 mutations:

Intragenic complementation is a genetic phenomenon in which diploid strains homozygous for the parental alleles do not grow at the restrictive temperature, but diploids bearing different recessive alleles (heteroallelic diploids) are able to grow. Intragenic complementation is often observed in mutants of multifunctional proteins (Ohya and Botstein 1994; Saka et al. 2001; Yahara et al. 2001). At 34°, homoallelic diploids of the fks1 ts alleles grew weakly (class I: fks1-1082/fks1-1082) or did not grow (class II, fks1-1163/fks1-1163; and class III, fks1-1154/fks1-1154) (Table 6). Among heteroallelic diploids, the pairs fks1-1082/fks1-1163 and fks1-1163/fks1-1154 exhibited weak intragenic complementation. A different heteroallelic pair (fks1-1082/fks1-1154) did not show intragenic complementation. Thus, intragenic complementation divides fks1 mutants into at least two classes.

View this table:
TABLE 6

Intragenic complementation

DISCUSSION

Fks1p, a putative catalytic subunit of GS, was functionally dissected on the basis of quantitative and high-dimensional phenotyping of ts mutants. By hierarchical cluster analysis of 501 morphological traits, we classified mutants into three functional groups. Extensive analyses of the mutants in each group revealed multiple functions of Fks1p that are required for cell wall construction, organization of a cell polarity, and endocytosis (Figure 8). The N-terminal domain or the site of the class I mutations is responsible for cellular glucan synthesis. The N-terminal part of the central cytoplasmic region, to which class II mutations are localized, functions in polarized growth and endocytosis. The class III mutations localize to the putative catalytic domain at the cytoplasmic region, which is required for in vitro GS activity.

Figure 8.—

Functional map of Fks1p. A black curved line, circles on the line, and a curved dotted line indicate Fks1p, fks1 mutation sites, and the putative catalytic domain, respectively. The gray hatched box indicates the plasma membrane. The class I mutations induce defects in in vivo glucan syntheses. The class II mutations induce defects in polarized cell wall remodeling and in endocytosis. The class III mutations induce defects in glucan synthesis and in endocytosis.

GS activity in class III mutants:

Class III mutants exhibited reduced GS activity with small Vmax values. Since three of four class III mutations (fks1-1114, fks1-1144, and fks1-1154) are localized to the putative catalytic domain (aa 829–973) (Kelly et al. 1996), the mutations likely affect the polymerization reaction, leading to reduced GS activity. The remaining class III mutant (fks1-1125) has mutations near the membrane-spanning region of the C terminus. Since this mutant exhibited an increased Km value for GS activity at 23°, this region may be required for high affinity with the substrate, UDP-glucose, although a consensus binding motif for UDP-glucose has not been identified in Fks1p.

Linear regression analysis and experiment of GS inhibitor treatment indicated correlation between GS activity and actin localization. Several reports suggest that cell wall construction and actin organization are related. First, coordinated cell wall construction requires an actin cytoskeleton (Kopecka and Gabriel 1995). Second, absence of bgs4+, an FKS1 homolog of the yeast S. pombe, results in dispersed actin patches (Cortes et al. 2005), and because the cell wall stress induces actin depolarization (Delley and Hall 1999), it is possible that the cell wall damage due to defective 1,3-β-glucan synthesis causes actin delocalization.

Cellular glucan synthesis in class I mutants:

Although class I mutants did not show clear reduction of in vitro GS activity, they have reduced levels of in vivo glucan incorporation and weak glucan-staining signal specifically in the small bud at the restrictive temperature. These results imply that the mutants have a defect in the synthesis of cellular glucan. Failure of intragenic complementation between class I and III mutants also supports this idea because it can be interpreted that defective cellular glucan synthesis in class I mutants is not complemented by glucan-synthesis-defective class III mutants. A possible explanation for the reduced in vivo GS activity is the defect in localization of Fks1p. However, this is unlikely because Fks1p localization in an FKS2 background was not altered in class I or III mutants (data not shown). According to hydropathy analysis, class I mutation sites are located in the cytoplasmic region (Douglas et al. 1994; Kelly et al. 1996), leading to another possible speculation that the region interacts with a factor(s), which is necessary for 1,3-β-glucan synthesis. However, immunoprecipitation of GS with an anti-Fks1p antibody showed none of the mutants altered the level of interaction with Rho1p (data not shown). Thus, the class I mutants might have a defect in the interaction of Fks1p with a factor(s) other than Rho1p.

Between class I and III mutants, intragenic complementation was not detected and we discussed the possible reasons above. However, we could not rule out the alternative possibility. The difference between class I and III mutants might be caused by the difference of threshold of cellular response for the level of GS activity, other than the difference of functional domains. This idea would explain the result that the class I mutants exhibited normal-cell-like morphology. Further study is necessary to test this possibility.

Polarized growth and endocytosis in class II mutants:

The class II mutants were found to be defective in cell polarity and endocytosis. It should be noted that bgs1+ (another FKS1 homolog) is also implicated in polarized growth in fission yeast (Ishiguro et al. 1997; Cortes et al. 2002). Thus, in both the budding and the fission yeasts, polarized growth is somehow involved in the GS function. The class II mutants may be defective in the interaction with a factor(s) involved in cell polarity leading to abnormal cell wall structure and loss of cell polarity. Loss of cell polarity explains why the mutants exhibit a large and round morphology. However, at the moment, we could not exclude the possibility that the primary defect of the mutants is in endocytosis. It has been reported that mutants defective in endocytosis exhibit an abnormal cell wall and cell polarity in the mother cell (Pruyne and Bretscher 2000).

Interestingly, the endocytosis defect was observed in Δfks1 but not in Δfks2 cells irrespective of their high similarity (deHart et al. 2003). In addition, existence of Fks2p suppresses temperature sensitivity of both class I fks1-1082 and class III fks1-1154 cells, but not class II fks1-1163 cells (data not shown), suggesting that the function in endocytosis is specific to Fks1p.

The class II and III mutants showed increased chitin levels, consistent with the previous reports showing that cell-wall-deficient mutants such as Δgas1 or Δfks1 show increased chitin synthesis (Garcia-Rodriguez et al. 2000; Carotti et al. 2002). Interestingly, these fks1 ts mutants also showed defects in endocytosis. It was previously reported that blockade of endocytosis stops internalization of Chs3p, the chitin synthase subunit, leading to a significant increase in chitin synthesis (Reyes et al. 2007). Therefore it is possible that loss of Fks1p function induces chitin accumulation due to defective endocytosis, to repair cell wall damage caused by 1,3-β-glucan shortage.

In summary, the present study has demonstrated that the putative catalytic subunit of GS, Fks1p is composed of multiple functional domains. This study has also demonstrated that Fks1p is involved in many cell morphological processes including actin regulation and polarized cell growth. Although further study is necessary to elucidate physiological functions of Fks1p, the present method based on cell morphology is a useful approach to untangle the pleiotropic functions of a protein.

Acknowledgments

We are grateful to Osamu Kondoh and Yukako Kondoh for technical advice on measurement of GS activity. We are also grateful to current and former members of the Laboratory of Signal Transduction, especially to Takahiko Utsugi and Mariko Sekiya-Kawasaki for technical support and discussion at the initial stage of this work. We also thank the community of R developers for their work. This work was supported by grants from the Ministry of Education, Science and Sports and Culture of Japan.

Footnotes

  • Received September 23, 2009.
  • Accepted January 20, 2010.

References

View Abstract