Killer toxins secreted by yeasts are proteinaceous toxins with diverse activities (Gulubev, 1998). These toxins are mostly specific for certain strains of yeasts, and are usually not effective against other organisms. Some killer toxins have been extensively studied (Bussey, 1991; Tipper and Schmitt, 1991) and the receptor molecules on the cell wall (Hutchins and Bussey, 1983; Schmitt and Radler, 1987) and the target molecules on the plasma membrane (Komiyama et al., 1998) have been identified. However, the most important processes, such as the molecular mechanism of internalization of the killer toxins from the cell surface and the mechanism of self-protection, termed ‘immunity’, remain unknown.
One of these killer toxins, SMKT, is produced by the halotolerant yeast Pichia farinosa KK1. This toxin kills yeasts of several genera in the presence of a high concentration of NaCl or sorbitol (Suzuki and Nikkuni, 1989) and is therefore named ‘salt-mediated killer toxin’ (SMKT). SMKT consists of two distinct subunits, α (63 amino acid residues) and β (77 amino acid residues) that are processed from a preprotoxin encoded by the chromosomal gene SMK1 (Suzuki and Nikkuni, 1994).
The wild-type strain of Saccharomyces cerevisiae is sensitive to SMKT. However, mutant strains that lack the SPF1 gene encoding the P-type ATPase are highly resistant to both exogenous SMKT (Suzuki and Shimma, 1999) and endogenously expressed SMK1 product (Suzuki et al., 2000). SMKT interacts with the cell surface of spf1 disruptants that have defects in the glycosylation process, suggesting that modified cell wall components sequester SMKT away from the target and confer resistance to exogenous SMKT (Suzuki and Shimma, 1999).
The two subunits of SMKT are folded together in a single ellipsoidal domain in the crystal structure (Kashiwagi et al., 1997) and they interact tightly with each other below pH 5. However, the subunits dissociate above pH 5, resulting in aggregation of the α subunit and concomitant loss of the toxic activity. Circular dichroism (CD) analysis has shown that the soluble β subunit loses its secondary structure after dissociation from the α subunit (Suzuki et al., 1997).
With regard to the crystal structure, the topology of SMKT is identical to that of the fungal killer toxin KP4 produced by Ustilago maydis, despite of the lack of sequence similarity between KP4 and SMKT (Kashiwagi et al., 1997). KP4 is likely to act as a Ca2+ channel inhibitor (Gu et al., 1995). On the other hand, SMKT is similar to K1 toxin of S.cerevisiae, i.e. it has a subunit structure consisting of a hydrophobic α subunit and a hydrophilic β subunit. A recent study reported that the K1 toxin activates the K+ channel Tok1p (Ahmed et al., 1999), although it is unclear how the K1 toxin with an optimum pH in the range of pH 4.6–4.8 interacts with the channel in the buffer at pH 7. Further studies are required to determine whether Tok1p is the true target of the toxin or whether de novo formation of ion channels by the toxin directly perturbs the cell membrane (Martinac et al., 1990).
As a simple model of biomembranes, liposomes are a useful system to elucidate interactions between biological materials and membranes at the molecular level. The use of dark-field microscopy allows us to visualize the intact morphology and dynamic behaviour of individual liposomes in suspension and allows obtaining real-time images (Magariyama et al., 1995; Saitoh et al., 1998). In the present study, we examined the effect of SMKT on the permeability of lipid bilayers to investigate the direct interaction of SMKT with membranes. We then directly observed the effects of SMKT on liposomes and yeast cells under a dark-field microscope. Our results showed that SMKT interacts only with the sensitive cells in a peripheral manner, as demonstrated in sodium carbonate extraction experiments. The mode of action of SMKT will be discussed, taking into account its interaction with the cell membranes.
Materials and methods
Phosphatidylcholine (PC), phosphatidylserine (PS), phosphatidylglycerol (PG) and cholesterol were purchased from Avanti Polar Lipids (Birmingham, AL) and used directly without further purification. Mastoparan X was purchased from Bachem (Bubendorf, Switzerland). All other chemicals were obtained locally and were of a high analytical grade.
Multilamellar vesicles (MLVs), calcein-entrapped large unilamellar vesicles (LUVs) for dye-release experiments, and small unilamellar vesicles (SUVs) were prepared as described previously (Machida et al., 2000), with the following modification. After evaporation of the solvent, dried lipid film was hydrated with 10 mM phosphate buffer (pH 4.5).
SMKT was prepared as described previously (Suzuki et al., 1997; Suzuki and Nikkuni, 1994).
Strains and media
Saccharomyces cerevisiae H2071 (MATa, his3-Δ1, leu2-3, 112, trp1-289, ura3-52) (Iida et al., 1994) and CS202A (spf1::LEU2 in H2071 background) (Suzuki and Shimma, 1999) were used as a sensitive strain and a resistant strain, respectively. YPD containing 1 M sorbitol was used for culturing yeast strains before SMKT treatment. The killer assay medium contained 2.1% YM broth (Difco), 1.5 M sorbitol and 25 mM citrate–phosphate buffer (pH 3.5).
Leakage of calcein from the vesicles was monitored fluorometrically using Millipore Cytofluor 2300/2350 (Millipore, Bedford, MA)(excitation at 490 nm and emission at 530 nm), as described previously (Machida et al., 2000).
Liposomes were examined with a dark-field microscope (Zeiss Axiophot, Germany) at 25°C. In order to observe the effect of SMKT on liposomes, 2 µl diluted MLV solution was placed on a cover slip (18×18 mm), which was then mounted on a glass slide. SMKT was introduced between the glass slide and the cover slip by placing a small droplet of 10 µM SMKT solution on the slide at the edge of the cover slip.
Yeast cells in the logarithmic phase of growth (1×107/ml) were incubated in the killer assay medium in the presence or absence of 0.1 µM SMKT for 1 h at 30°C. Cells were also treated with 3.7% formaldehyde or heated at 65°C for 10 min in the presence or absence of 1% SDS. Those cells were directly observed under a dark-field microscope (Olympus BX50, Japan) equipped with a digital camera (Fujix HC2500, Japan) or under a phase-contrast microscope (Leica DMLB) with a digital camera (Leica DC200).
Sodium carbonate extraction
Spheroplasts were prepared as described previously (Kawamoto et al., 1990) with the following modification. After 2-mercaptoethanol treatment, cells were washed with Solution B containing 0.1 M potassium-phosphate buffer (pH 7.5) and 1.2 M sorbitol and were suspended in Solution B containing 50 µg/ml Zymolyase 100T (Seikagaku Co., Japan). Spheroplasts prepared from logarithmically growing cells (2×108) and untreated cells (2×108) were incubated for 5 min in the killer assay medium containing 1.2 M sorbitol in the presence or absence of 1 µM SMKT. After washing nine times with 1 ml ice-cold citrate–phosphate buffer (pH 3.5) containing 1.2 M sorbitol (CPB/sorb), the cells and spheroplasts were finally washed with 200 µl CPB/sorb, followed by extraction with 200 µl 0.1 M sodium carbonate (SC) (pH 11) containing 1.2 M sorbitol. After 10 min incubation on ice, the SC extracts were separated by centrifugation at 1000×g for 1 min. After washing once with SC, SC-treated cells were broken with glass beads, as described previously (Suzuki, 1999). SMKT-treated and washed spheroplasts were also broken with glass beads. Lysates of the SC-treated cells and the spheroplasts were centrifuged at 100 000×g for 1 h to separate the supernatants (S100) and precipitates (P100). The P100 fractions and the unbroken cell debris were suspended in 50 µl SDS sample buffer (2% SDS, 50 mM Tris–HCl, pH 6.8, 10% glycerol, 2.5% β-mercaptoethanol, 0.1% bromophenol blue). Proteins in the final wash, the SC extracts and the S100 fractions were precipitated in the presence of 12.5% trichloroacetic acid, washed with acetone, and dissolved in 50 µl SDS sample buffer. In the case of SC-treated spheroplasts, the precipitates after SC extraction were suspended in 50 µl SDS sample buffer.
Aliquots (10 µl) of samples were analysed by Tricine SDS–PAGE, followed by immunoblotting as described previously, with the following modification (Suzuki, 1999). Protein A-purified rabbit antiserum against the α subunit, that against the β subunit and horseradish peroxidase-linked anti-rabbit immunoglobulin (Bio-Rad Laboratories) were used at dilutions of 1:5000, 1:1000 and 1:3000, respectively. Anti-yeast 3-phosphoglycerate kinase, mouse monoclonal 22C5-D8 (Molecular Probes, Eugene, OR) and horseradish peroxidase-linked anti-mouse immunoglobulin (Bio-Rad Laboratories, Richmond, CA) were used at dilutions of 1:10 000 and 1:3000, respectively. For detection, the ECL Plus Western Blotting Detection Reagents (Amersham-Pharmacia) were used.
Intact cells and spheroplasts were incubated in the presence or absence of SMKT in the killer assay medium containing 1.2 M sorbitol at 30°C for 1 h and diluted in Solution B. In the case of intact cells, cells were appropriately diluted and directly spread on YPD plates. For measurement of the regeneration rate of spheroplasts, 0.1 ml of the appropriate dilution of spheroplasts were mixed with 5 ml YPD containing 1.2 M sorbitol and 3% agarose warmed at 50°C and spread on YPD plates. After 3–5 days incubation, developed colonies were counted.
Results and discussion
Analysis of killer toxins always involves difficulties associated with assessing the activity in the physiological state, because most killer toxins, including killer toxins of S. cerevisiae (Pfeiffer and Radler, 1984) and SMKT (Suzuki and Nikkuni, 1989), have an optimum pH in the acidic range. Considering the habitat of yeasts, as yeasts prefer to grow under acidic conditions, in fruit juice, the finding that killer toxins show optimum activity at acidic pH seems reasonable. The physiological conditions for mammalian cells are not always the same as those for yeasts. Therefore, one should be careful when interpreting the results obtained under conditions that favour the loss of toxin activity. Liposomes prepared under acidic conditions are a powerful tool to overcome these problems. In this study, SUVs, MLVs and calcein-entrapped LUVs were prepared using a buffer at pH 4.5. Such a system allowed us to analyse the interaction between the pH-sensitive killer toxin and the cell membranes.
Figure 1 shows the time course of calcein leakage from LUVs of PC/PG (7:3) in the presence of SMKT. Mastoparan-X, a tetradecapeptide toxin from hornet venom, was used as a representative membrane-destabilizing peptide. In the absence of SMKT, no leakage was observed. However, addition of SMKT or mastoparan-X caused rapid release of calcein, although the effect of SMKT was less than half that of mastoparan-X. Leakage increased with increasing SMKT concentration. In order to examine the membrane specificity of SMKT, several types of calcein-entrapped LUVs were prepared by changing the ratio of acidic phospholipids and cholesterol. Maximum leakage from LUVs of PC/PS (1:1), PC/PS (10:1), PC/PS/Chl (10:1:1), and PC/PG/Chl (7:3:1) in the presence of 5 µM SMKT was 14%, 33%, 28% and 34%, respectively. No significant difference in calcein leakage was observed in these LUVs, suggesting that SMKT non-specifically destabilized the LUVs.
Spherical liposomes yield a ring-like image because the density of lipids increases at the circumference when a sphere is projected in one plane and where materials with a high refractive index such as lipids distribute in high density yield a bright image by dark-field microscopy (Saitoh et al., 1998).
The process of destruction of liposomes in the presence of SMKT was directly observed under a dark-field microscope. When the buffer was added to the liposome suspension from the edge of the cover slip, a flow of the liposomes was observed. In Figure 2A, a huge MLV is shown at the centre, while running small MLVs appear as light tracks. Under these conditions, no morphological changes were observed. Addition of 5 µM bovine serum albumin to the liposome suspension had no effect (data not shown). On the other hand, when SMKT was added to the liposome suspension, a rapid morphological change was noted. The outer lamella of the huge MLV (Figure 2A) separated off and was divided into small vesicles very rapidly, resulting in aggregation of the small vesicles (Figure 2B). Because SMKT was added from the edge of the cover slip and diffused from there into the liposome suspension, the concentration of SMKT was not uniform. Immediate destruction of liposomes was observed in the area near the edge, whereas a relatively slow change in the morphology of the liposomes was observed in the area 5 mm from the edge. These observations are in agreement with theresult that the calcein leakage is dependent on the concentration of SMKT (Figure 1). After destruction of the liposomes, no Brownian movement of the liposomes was observed and the liposomes did not flow even when buffer was added, suggesting that these vesicles were stuck to the glass surface.
To exclude the possibility that the interaction between SMKT and the membrane occurs only invitro and does not relate to the killing mechanism, we investigated the effect of SMKT on membranes of yeast cells in vivo. The sensitive wild-type cells and resistant spf1 disruptant cells were treated with SMKT and examined by dark-field microscopy (Figure 3A). In the absence of SMKT, each yeast cell appeared as a ring-like structure with a clear-cut outline, indicating that the yeast cells retained a spherical shape. However, SMKT-treated sensitive cells appeared hazy as a turbid sphere. The outline and inside of the cells could not be distinguished, suggesting that the spherical membrane structure was destroyed by SMKT. The number of cells with a turbid image increased with increasing concentrations of SMKT and prolonged incubation time. Although maximum leakage from LUV was less than 5% in the presence of 0.1 µM SMKT, approximately 60–70% of the sensitive cells showed a turbid image in the presence of 0.1 µM SMKT after 1 h incubation at 30°C. On the other hand, spf1 disruptants displayed a ring-like image and cells with a turbid image were not observed even in the presence of 1 µM SMKT. These results indicate that there are some sensitivity determinants in the sensitive cells.
To show that the dark-field image of SMKT-treated cells was due to a direct effect of SMKT, the sensitive cells were treated under several conditions and examined by dark-field microscopy (Figure 3B). Cells treated at 65°C for 10 min showed a turbid image similar to SMKT-treated cells. However, cells treated at 65°C for 10 min in the presence of 1% SDS displayed a ring-like image. By this treatment, membrane structures were destroyed and organelles were not visible by phase-contrast microscopy (data not shown). On the other hand, when cells were treated with 3.7% formaldehyde, a standard fixative for observing cell structure, they displayed a ring-like image in dark-field microscopy. These treatments resulted in 100% cell death. These results suggest that the turbid image is due to denaturation and aggregation of cell components and is not necessarily due to cell death. Solubilization of such aggregates by addition of SDS resulted in a ring-like image, although this image may represent an intact cell wall. It is possible that aggregation of membranes and other cellular components may disrupt the spherical structure of the plasma membrane, resulting in eventual cell death.
Interaction of SMKT with the sensitive cells
To investigate how SMKT interacts with yeast cells, we determined the conditions that could effectively strip SMKT from the cells. Experiments were performed using the wild-type strain and the spf1 disruptant strain. Cells were treated with 1 µM SMKT for 5 min, washed with CPB/sorb until unbound SMKT was not detected in the washing solution. Then cells were treated with 0.1 M SC, which is typically used to extract proteins that are associated with membranes in a peripheral manner (Fujiki et al., 1982). We also analysed SC-extracted SMKT, SMKT binding to cells, and 3-phosphoglycerate kinase (PGK) as a marker of the cytoplasmic proteins (Figure 4A).
Both the α and β subunits were extracted by SC-treatment in the case of wild-type cells. Since PGK was not detected in SC-treated fractions, the results indicated that SC-treatment does not cause cell lysis. This result suggests that SMKT is associated with the cell surface in a peripheral manner. The α subunit was detected in the P100 fraction of the lysates of SC-treated wild-type cells but not in the S100 fraction. When SC-treatment of cells was repeated, more than 90% of SMKT was extracted on the first SC-treatment (data not shown). In this experiment, SC-treated cells were washed with SC again and then broken with glass beads to obtain cell lysates. Therefore, the α subunit in the P100 fraction is not likely to be a contamination from peripherally interacting SMKT and is likely to interact with membranes in a manner different from the SC-extractable α subunit.
Any interaction between SMKT and spf1 disruptants was detected. Neither the α nor the β subunit was extracted by SC-treatment from spf1 disruptants and neither was detected in the cell lysates, suggesting that SMKT may be washed out during the washing process in the case of resistant cells. We have previously indicated that the interaction of SMKT with the surface of spf1 disruptants was observed only when cells were fixed with formaldehyde just after SMKT treatment (Suzuki and Shimma, 1999). Detection by indirect immunofluorescence microscopy was performed without zymolyase treatment. Under the conditions employed, no intracellular SMKT was detected by mean of fluorescent-labelled IgG because of the barrier function of the cell wall. Therefore, it is reasonable to conclude that SMKT that interacts with the membrane under the cell wall was not detected in the case of wild-type cells and that only SMKT superficially interacting with the spf1 disruptants could be detected.
To test whether the cell walls are involved in such interaction, the SC-extraction from SMKT-treated spheroplasts was simultaneously performed (Figure 4B). Both the α and β subunits were extracted by SC from the wild-type spheroplasts as was observed in the wild-type cells. After SC-treatment, very small amount of cell debris was recovered by centrifugation at 1000×g and PGK was extracted, indicating that SC-treatment destroyed the spheroplasts in spite of the presence of 1.2 M sorbitol. To exclude the possibility that SC-extracted SMKT from the spheroplasts was derived from the inside of the spheroplasts, SMKT-treated spheroplasts were broken with glass beads and the lysates were separated into the S100 fractions and the P100 fractions. Neither the α nor β subunit was detected in the S100 fractions, suggesting that SMKT was associated with spheroplasts of sensitive strain and was extracted by SC, as was observed in wild-type cells.
A small amount of the α subunit was detected in the samples from spheroplasts of the spf1 strain. This result does not contradict our previous finding that spf1 disruptants acquire resistance by trapping SMKT on the cell surface (Suzuki and Shimma, 1999). In the absence of the resistant cell walls, SMKT may interact to some extent with the cell membranes of spf1 disruptants. However, more SMKT was extracted by SC in the case of wild-type spheroplasts, suggesting that SMKT preferably interacts with the membranes of sensitive cells.
To determine the role of the cell wall in the killing activity of SMKT, we next examined the killing effects of SMKT on intact cells and spheroplasts. In the case of intact cells, the viability of the wild-type cells was 3.6% after treatment with 5 µM SMKT for 1 h, whereas that of spf1 disruptants was 80%. The regeneration rate of the wild-type spheroplasts was 28% in the absence of SMKT and was 1.4% after treatment of 5 µM SMKT for 1 h, suggesting that wild-type spheroplasts are sensitive to SMKT. On the other hand, the regeneration rate of spf1 spheroplasts was in the range of 2–3% in the presence and absence of SMKT, while less than 5% of unspheroplasted cells remained in the spheroplasts. These results suggest that Spf1p is also involved in the regeneration of spheroplasts. However, it is difficult to conclude that spf1 spheroplasts are resistant to SMKT when the regeneration rate is less than 5%, because it is impossible to exclude that the regenerated colonies were derived from residual intact cells.
As a simple model of biomembranes, we used liposomes to show that SMKT non-specifically destroys membranes. However, interaction of SMKT with yeast cells was more complicated. Using SC treatment, we obtained for the first time a direct evidence for SMKT interaction with the membrane of sensitive cells. Since SMKT interacts with the membrane of wild-type cells rather than that of spf1 disruptants, a specific molecule that was synthesized in the presence of Spf1p may involved in the interaction with SMKT. Isolation of the target molecule of SMKT would elucidate not only the killing mechanism but also the function of Spf1p.
We thank Dr Y. Magariyama for his support concerning the dark-field microscopy, Drs K. Yoda, Y. Noda and F. Tsuchiya for helpful discussion, and Dr Y. Arata for encouragement. This study was supported in part by a grant-in-aid (Pioneer program) from the Ministry of Agriculture, Forestry and Fisheries, Japan.