|Home | About | Journals | Submit | Contact Us | Français|
The killer toxin from Pichia membranifaciens CYC 1106, a yeast isolated from fermenting olive brines, binds primarily to the (1→6)-β-d-glucan of the cell wall of a sensitive yeast (Candida boidinii IGC 3430). The (1→6)-β-d-glucan was purified from cell walls of C. boidinii by alkali and hot-acetic acid extraction, a procedure which solubilizes glucans. The major fraction of receptor activity remained with the alkali-insoluble (1→6)-β- and (1→3)-β-d-glucans. The chemical (gas-liquid chromatography) and structural (periodate oxidation, infrared spectroscopy, and 1H nuclear magnetic resonance) analyses of the fractions obtained showed that (1→6)-β-d-glucan was a receptor. Adsorption of most of the killer toxin to the (1→6)-β-d-glucan was complete within 2 min. Killer toxin adsorption to the linear (1→6)-β-d-glucan, pustulan, and a glucan from Penicillium allahabadense was observed. Other polysaccharides with different linkages failed to bind the killer toxin. The specificity of the killer toxin for its primary receptor provides an effective means to purify the killer toxin, which may have industrial applications for fermentations in which salt is present as an adjunct, such as olive brines. This toxin shows its maximum killer activity in the presence of NaCl. This report is the first to identify the (1→6)-β-d-glucan as a receptor for this novel toxin.
Cell walls determine the shape of fungal cells and are essential for their integrity. They consist mainly of carbohydrates, some free and some linked to protein. The main components of the yeast cell wall are a (1→3)-β-d-glucan (50%) that also contains some (1→6)-β-linked branches (5%) (11) and a mannoprotein, most of which is carbohydrate. A (1→6)-β-d-glucan, also containing some (1→3)-β-linked branches (14%), is a relatively minor constituent (15%), and chitin (0.6 to 9%) is present at an even lower level (18). The latter is concentrated in the bud-scar region (18). The role of cell surface polysaccharides as receptors for proteins in many cell events is widely accepted, but the mechanism of their action is poorly understood. As well as acting as receptors for bacteria, viruses, and toxins, surface polysaccharides may be involved in cell interactions, such as cell associations, distribution, and turnover (15).
Killer yeasts act on sensitive yeasts by liberating killer toxins that are either proteins or glycoproteins. The K1 killer toxin of Saccharomyces cerevisiae acts in two steps (20). First, the toxin is adsorbed by a glucan of the cell wall. Then, the toxin is bound to a receptor in the cell membrane, damaging the membrane and releasing K+, ATP, and other metabolites and destroying the pH gradient of the membrane (8). Binding of toxin to the wall receptor appears to be a necessary prelude to cell killing (1, 4). Previous workers (1, 5) defined a specific cell wall receptor for killer toxin by measuring binding of toxin to sensitive cells and to resistant mutants with defective receptors. The receptor, probably a polysaccharide or glycoprotein, was solubilized from yeast cell walls by an endo-(1→3)-β-d-glucanase action and is heat and pronase resistant but periodate sensitive (1).
Various primary receptors for other killer toxins have been reported. (1→3)-β-d-Glucans and (1→6)-β-d-glucans probably act as cell wall receptors of Hansenula mrakii LKB 169 (17). The (1→6)-β-d-glucans are primary receptors for S. cerevisiae K1 and K2 killer toxins (15) and Hanseniaspora uvarum killer toxin (32). Mannoproteins are receptors for KT28 of S. cerevisiae (35, 36) and Zygosaccharomyces bailii killer toxin (33), and chitin is a receptor for Kluyveromyces lactis killer toxin (41). Thus, any of the principal components of the cell wall could be the primary receptor for a killer toxin.
The phenomenon of killer activity in yeast was originally observed with Saccharomyces (23) and was later found in other genera (19, 27). Recently, interest in the development of bacteriocins as food preservatives (14) and in the use of the killer factors for industrial applications has increased (7, 13, 29, 30). However, the role that killer activity may have as a mechanism of antagonism among yeasts in natural environments is not clear, and the conditions governing their behavior in various niches are mostly unknown. In spite of this lack of knowledge, the use of killer toxins to control yeast populations during fermentations has been postulated for beer and wine (7).
In previous work (22, 25, 26), members of this group found that Pichia membranifaciens, the dominant species of yeast isolated from spontaneously fermenting olive brines, had killer characteristics. We also found that sodium chloride, in concentrations similar to those found in the brines, enhanced the apparent toxicity of some of the strains and broadened the killer spectrum (22). We concluded that salt may confer an additional advantage on killer strains and could affect the development of spontaneous fermentations and the properties of the product.
Candida boidinii IGC 3430 can be isolated from olive brines in the first stage of the fermentation process and is sensitive to P. membranifaciens killer toxin only in the presence of 0.1 to 1 M sodium chloride. C. boidinii is of industrial interest because of the pernicious effects (lipolytic activity and lactic acid assimilation) it has on the fermentation of olive brines (25). Thus, the P. membranifaciens killer toxin might be commercially useful in controlling C. boidinii, and other spoilage yeasts, in fermentations containing moderate to high levels of sodium chloride.
The objectives of this study were to isolate and characterize cell wall fractions from the sensitive yeast C. boidinii IGC 3430 to determine the nature of the receptor for the killer toxin from P. membranifaciens CYC 1106. This study is the first to identify the primary receptor, (1→6)-β-d-glucan, for this killer toxin, a killer toxin with some properties of industrial interest.
The killer strain, P. membranifaciens CYC 1106 (Complutense Yeast Collection, Biology Faculty, Complutense University, Madrid, Spain), was isolated from olive brines and identified in the Gulbenkian Institute of Science. The sensitive strain was C. boidinii IGC 3430 (Portuguese Yeast Culture Collection, Biotechnology Unit, Faculty of Sciences and Technology, New University of Lisbon, Caparica, Portugal). The K1 strain of S. cerevisiae was provided by H. Bussey (Department of Biology, McGill University, Montreal, Quebec, Canada). These strains were maintained on agar slants containing 0.5 g of yeast extract (Difco, Detroit, Mich.), 1 g of proteose peptone no. 3 (Difco), 2 g of glucose, 2 g of agar, and water to make 100 ml.
The yeast growth medium was YNB–D–Brij-58 (yeast nitrogen base–dextrose–Brij-58): 1% (wt/vol) glucose and 0.67% (wt/vol) yeast nitrogen base (Difco). This medium was buffered with 0.2 M sodium citrate-phosphate buffer (pH 4.0) supplemented with 0.01% (wt/vol) Brij-58 (polyoxyethylene 20 cetyl ether) (Serva, Heidelberg, Germany). This detergent was employed as an adjunct because the addition of nonionic detergents enhanced killer toxin production significantly (data not shown). Killer activity was determined in YMA-MB: 1% (wt/vol) glucose, 0.3% (wt/vol) yeast extract (Difco), 0.3% (wt/vol) malt extract (Difco), 0.5% (wt/vol) proteose peptone no. 3 (Difco), 30 mg of methylene blue/liter, 6% (wt/vol) NaCl, and 2% (wt/vol) agar (22, 25). Incubation was at 20°C; killer factor is rapidly inactivated at temperatures above 25°C.
Killer strains were cultivated for 3 days at 20°C in 2-liter Erlenmeyer flasks with 1 liter of YNB–D–Brij-58. Cultures were incubated in a rotary bed shaker (150 rpm). After centrifugation (4,000 × g, 10 min, 4°C) the supernatant was adjusted to a final glycerol concentration of 15% (vol/vol) and concentrated to a volume of 75 ml by tangential ultrafiltration with a 10-kDa-cutoff membrane (Minisette membrane cassette, omega type [Filtron Technology Corporation, Northborough, Mass.]). These partially purified concentrated supernatants were used as the killer toxin concentrate.
We assayed killer toxin with a diffusion test (42), using 6-mm-diameter antibiotic assay AA Whatman paper disks on YMA-MB seeded with the sensitive strain. The diameter of the inhibition zone was used as a measure of the yeast killer activity, and killer toxin activity was expressed in arbitrary units (AU) (3). Under the experimental conditions used, a linear relationship was observed between the logarithm of the protein concentration in the solution tested and the diameter of the inhibition zone. One AU is defined as the amount of protein resulting in an inhibition zone with a 1-mm diameter.
The sensitive strain was grown on YMB (YMA-MB without salt, methylene blue, and agar) and harvested at late stationary phase (30 h). Cell walls from C. boidinii IGC 3430 were prepared by mechanical disruption (10). After being washed, cell walls were freeze-dried and stored in a desiccator. Glucans were extracted as described by Manners et al. (24) (fractions S-1, S-2, P-1, and P-2). Mannoproteins were extracted and partially purified from cell walls by Cetavlon (cetyltrimethylammonium bromide) fractionation (35). We purified chitin by the method of Fleet (11).
Binding of killer toxin was estimated from the amount of killer toxin remaining in solution after incubation. Approximately 2,400 AU of killer factor ml−1 was added to a suspension of 20 mg of the cell wall fractions ml−1. Samples were stirred gently and then centrifuged (10,000 × g, 30 s). The amount of killer activity remaining in solution after incubation was measured by the diffusion test method.
Polysaccharides with (1→4)-α as the main glucosidic linkage (amylose, amylopectin, and polygalacturonic acid); pullulan, a (1→4)-α- and (1→6)-α-polysaccharide; xylan and chitin, (1→4)-β-polysaccharides; laminarin, a (1→3)-β-polysaccharide; liquenan, a (1→3)-β- and (1→4)-β-polysaccharide; and (1→6)-β-polysaccharides (pustulan and a glucan obtained from Penicillium allahabadense) were used for toxin binding. These polysaccharides (15 mg each) were added to 1 ml of killer toxin (150 AU ml−1). Mixtures were shaken gently and then centrifuged (10,000 × g, 30 s). The amount of killer activity remaining in solution was assayed. Amylose and amylopectin were obtained from Sigma Chemical Co. (St. Louis, Mo.). Polygalacturonic acid, xylan, pullulan, pustulan, chitin, laminarin, and liquenan were gifts from C. Vázquez and M. J. Martínez.
The P-1 fraction (20 mg) of the cell wall was incubated in the presence of 1 ml of the killer toxin concentrate. Test suspensions were stirred at 20°C. At intervals, 80-μl samples were centrifuged (10,000 × g, 30 s), and 40 μl was used for determination of the killer activity.
The P-1 and P-2 fractions and pustulan (2 mg each) were suspended in 200 μl of 100 mM NaIO4 in 10 mM sodium citrate-phosphate buffer (pH 4.0). As a control, cell wall fractions were suspended in 10 mM sodium citrate-phosphate buffer (pH 4.0). These mixtures were incubated in the dark for 2 h and then were centrifuged (10,000 × g, 1 min) and washed (10 mM sodium citrate-phosphate buffer [pH 4.0]). The pellets were suspended in the killer toxin concentrate (2,400 AU ml−1) and incubated at 20°C for 1 h. The pellets were then centrifuged (9,000 × g, 0°C, 30 min), and the amount of killer activity remaining in the supernatant after centrifugation was measured.
Polysaccharide fractions (5 mg) were hydrolyzed with 2 M H2SO4 in sealed evacuated tubes in an oven at 100°C for 5 h. After neutralization with a saturated BaCO3 solution and centrifugation (5,000 × g, 5 min), the sugars were identified and quantified by gas-liquid chromatography of the corresponding alditol acetates (21) on 3% SP-2340 on 100/120 Supelcoport (Supelco, Inc., Bellefonte, Pa.). A 2-m by 2-mm glass column was used at 200 to 230°C: 3 min at 200°C, a temperature rise of 10°C min−1 to 230°C, and 8 min at 230°C. The N2 flow rate was 30 ml min−1. A hydrogen flame ionization detector with a sensitivity of 10−10, at a sample size of 3 μl, was used in a Perkin-Elmer-Sigma (Wellesley, Mass.) 3 and 10 gas chromatograph. Inositol was used as the internal standard. Peaks were identified on the basis of sample coincidence with the relative retention times of standards.
Oxidation of a polysaccharide and quantitative determination of the periodate consumed and the formic acid generated can provide information on the nature and proportion of the glycosidic linkages in the polysaccharide (12, 16). Periodate oxidation was performed according to the method of Aspinall and Ferrier (2). Twenty-five milligrams of fractions P-1 and P-2 (with pustulan and glucose as controls) were suspended in 25 ml of distilled water and mixed with 25 ml of 30 mM NaIO4. Aliquots of the oxidation mixture were taken daily, and absorbance (223 nm) was measured until constant values were obtained (7 days). Periodate consumption was determined by measuring the decrease of absorbance of the reaction mixture, and the formic acid produced was titrated with standard 0.02 N NaOH to a methyl red end point.
Infrared (IR) spectra were obtained in KBr disks (300 mg of KBr with 1 to 2 mg of the corresponding polysaccharide) on a Perkin-Elmer 1420 IR spectrophotometer.
1H spectra were recorded at 35°C with a Varian (Palo Alto, Calif.) XL 300 spectrophotometer operating at 300 MHz. The polysaccharides (30 mg) were dissolved in 0.8 ml of D2O.
P. membranifaciens CYC 1106 showed killer activity against C. boidinii and S. cerevisiae only in the presence of NaCl (Table (Table1).1). P. membranifaciens and C. boidinii were resistant to the K1 killer toxin of S. cerevisiae, indicating that the P. membranifaciens killer toxin and K1 are different toxins. C. boidinii was killer toxin negative.
The yield of cell walls was 21% of the initial dry weight. The walls were subjected to extraction for glucan, and we recovered more than 80% of the initial wall material (Table (Table2).2). The remaining 20% was lost in washing. Fraction S-1 (16%) reacted with Fehling's solution to produce a dense flocculent precipitate, suggesting that it was a mannan (31). The mannoprotein fraction of C. boidinii as assayed by Cetavlon precipitation accounted for only 10% of the cell wall, a smaller value than that obtained in the Fehling's precipitation (S-1 fraction). The chitin content (2%) was similar to that obtained for other yeasts (9, 39).
The (1→6)-β-d-glucans, pustulan, and the polysaccharide obtained from P. allahabadense (34) were effective in toxin binding. None of the polysaccharides tested composed of (1→4)-β-, (1→4)-α-, (1→3)-β-, (1→6)-α-, or some mixture of these linkages bound the killer toxin. These results were consistent with the presence of a (1→6)-β-d-glucan toxin receptor.
The P-1 and P-2 cell wall fractions from C. boidinii IGC 3430 are the only ones that bind killer toxin and presumably contain the toxin receptor. These alkali-insoluble cell wall fractions contain mostly (1→3)-β-d-glucan and (1→6)-β-d-glucan. After two successive hot-acid extractions, the receptor was associated with the acid-soluble fraction (P-1), which retained 91% of the toxin-binding activity. The P-2 fraction retained 29%, indicating that although material with receptor activity was obtained by acid extraction, some receptor activity was not solubilized by this procedure.
Sodium metaperiodate severely reduced killer toxin binding to the P-1 and P-2 fractions, suggesting that (1→3)-β-d-glucans or other periodate-resistant polysaccharides were not involved in toxin binding.
Killer toxin from P. membranifaciens CYC 1106 was very quickly adsorbed to the P-1 fraction of the cell wall (760 AU min−1 under the conditions tested), with 75% bound within the first 2 min and little or no additional binding for at least the next 3 min.
The cell wall fractions P-1 and P-2 were composed primarily of glucose (95 to 98%) and mannose (2 to 5%). The mannose in these fractions may be due to contamination with cell wall mannans. Pustulan, a (1→6)-β-d-glucan from Umbilicaria pustulata, was used as the control. Pustulan is primarily composed of glucose (98%) but also contains some arabinose. The P-2 fraction was more resistant to digestion with H2SO4 (yield of digestion, 55%) than P-1 (74%).
The IR spectrum of the cell wall fractions of C. boidinii (Fig. (Fig.1)1) is characteristic of a glucan, having the β-configuration, an absorption band at 890 cm−1, and a lack of absorption at 850 cm−1 (34). The lack of an absorption band at 1,750 cm−1 shows the absence of the carbonyl group of uronic acids or esterified organic acids (e.g., P. allahabadense glucan).
The 85% conversion of hexoses to formic acid (Table (Table3)3) in the oxidation mixture is evidence that the P-1 fraction was a (1→6)-glucan. The P-2 fraction consumed two-thirds less periodate than P-1, suggesting that the P-2 fraction contains (1→3)-linkages. P-1 was primarily composed of (1→6)- linkages (85%), while P-2 was a glucan with both (1→6)- linkages (24%) and (1→3)- linkages (76%) (Table (Table3).3).
Evidence from the competition studies with pure polysaccharides and enzymatic and chemical degradation of cell wall fractions and from resistant mutants (kre mutants) suggests that (1→6)-β-d-glucan is the K1 killer toxin receptor in the cell wall of S. cerevisiae (1, 5, 15), although other cell wall components may be the primary receptors for killer toxins from other strains or yeast species (17, 33, 41). We tested polysaccharides of known composition as receptors of the killer toxin. Toxin binding was specific for the (1→6)-β- linkage, since none of the polysaccharides tested composed of (1→3)-β-, (1→4)-β-, (1→4)-α-, or (1→6)-α- linkages showed killer toxin binding.
The results of cell wall fractionation studies of the receptor are less easily interpreted, but they are consistent with (1→6)-β-d-glucan involvement. Yeast mannans may be covalently linked to protein. Mannoproteins are antigenic determinants, and they may function as receptors or provide support for other receptor systems (11, 35, 37, 38). Purified mannoprotein fractions did not bind P. membranifaciens killer toxin, showing that mannoproteins are not involved as primary receptors of this killer toxin. Some killer toxins (e.g., K. lactis toxin) have chitin as the primary receptor (41). The chitin content in the walls of C. boidinii IGC 3430 was low (2%), and the receptor for the killer toxin was not related to chitin, since it did not bind killer toxin. Alkali insolubility, together with the periodate degradation and binding tests, was consistent with a (1→6)-β-d-glucan component. These results are similar to those obtained with S. cerevisiae by Hutchins and Bussey (15). The P. membranifaciens killer toxin is different from some S. cerevisiae killer toxins (e.g., KT28) and similar to others (e.g., K1, K2, and K3). The similarities include a (1→6)-β-d-glucan receptor and rapid adsorption of killer toxin to the P-1 fraction. The amount of (1→6)-β-d-glucans present was directly related to the amount of killer toxin bound. This relationship was similar regardless of the origin of the polysaccharide (yeast cell wall or lichen).
We think that the (1→6)-β-d-glucans in the cell wall of C. boidinii IGC 3430 facilitate the primary contact of the killer toxin from P. membranifaciens CYC 1106. The use of more specific carbohydrate-degrading enzymes and more chemical and structural analysis of cell walls from toxin-sensitive strains are needed to further elucidate the nature of this site.
This toxin shows similarities to the K1 killer toxin of S. cerevisiae, but the K1 toxin becomes less stable with increasing ionic strength (28, 40). Both P. membranifaciens and C. boidinii are resistant to the K1 killer toxin (Table (Table1);1); therefore, P. membranifaciens and K1 killer toxins probably are distinct toxins with different structures or activities.
In the search for food biopreservatives or therapeutic agents against yeast and fungi, studies of mycocins (killer toxins) have been common (6, 30). To determine their effectiveness in food and therapeutic systems, it is necessary to obtain relatively large quantities of these proteins in a pure and concentrated form. We think that the receptor from C. boidinii described here might provide an effective means to purify, via a receptor-mediated affinity chromatographic technique, the P. membranifaciens CYC 1106 killer toxin and possibly other toxins with similar biological activities. This killer toxin could be used for application in some industrial processes, such as with olive brines, soy sauce production, or other fermenting processes in which high levels of sodium chloride are present.
This work was supported by the EU project AIR-CT93-0830.
We thank I. Spencer-Martins for help at various stages of this work.