G. Menestrina, CNR-ITC, Istituto di Biofisica – Sezione di Trento, Via Sommarive 18, 38050 Povo (TN), Italy. Fax: + 39 0461 810 628, Tel.: + 39 0461 314 256, E-mail: firstname.lastname@example.org
Ostreolysin is a 16-kDa cytolytic protein specifically expressed in primordia and fruiting bodies of the edible mushroom Pleurotus ostreatus. To understand its interaction with lipid membranes, we compared its effects on mammalian cells, on vesicles prepared with either pure lipids or total lipid extracts, and on dispersions of lysophospholipids or fatty acids. At nanomolar concentrations, the protein lysed human, bovine and sheep erythrocytes by a colloid-osmotic mechanism, compatible with the formation of pores of 4 nm diameter, and was cytotoxic to mammalian tumor cells. A search for lipid inhibitors of hemolysis revealed a strong effect of lysophospholipids and fatty acids, occurring below their critical micellar concentration. This effect was distinct from the capacity of ostreolysin to bind to and permeabilize lipid membranes. In fact, permeabilization of vesicles occurred only when they were prepared with lipids extracted from erythrocytes, and not with lipids extracted from P. ostreatus or pure lipid mixtures, even if lysophospholipids or fatty acids were included. Interaction with lipid vesicles, and their permeabilization, correlated with an increase in the intrinsic fluorescence and α-helical content of the protein, and with aggregation, which were not detected with lysophospholipids. It appears that either an unknown lipid acceptor or a specific lipid complex is required for binding, aggregation and pore formation. The inhibitory effect of lysophospholipids may reflect a regulatory role for these components on the physiological action of ostreolysin and related proteins during fruiting.
The oyster mushroom (or white-rot fungus) belongs to the genus Pleurotus which comprises a group of edible, ligninolytic fungi with medicinal, biotechnological, and environmental applications [1,2]. Despite its widespread and massive cultivation, a major lack of information remains on the cellular processes that lead to the initiation of fruiting body development, as is also true for other edible mushrooms. Several mushrooms have been examined for genes specifically expressed during formation of primordia and fruiting bodies . Recently, expressed sequence tags (ESTs) of P. ostreatus were compared within liquid-cultured mycelium and fruiting body to investigate changes in the genes expressed during fruiting . Among the 1069 ESTs identified in fruiting bodies, one set of unigene sequences, with a redundancy number of 29, was found to be differentially expressed. These sequences were highly homologous to the Aa-Pri1 gene expressed during primordia and fruiting body initiation by the mushroom Agrocybe aegerita[3,4]. Moreover, 13 of the ESTs, if translated, are identical with a 138-amino-acid protein (PriA) translated from P. ostreatus cDNA (EMBL/GenBank/DDBJ databases: Q8X1M9). The existence of the translation products was confirmed by isolation of the corresponding proteins, ostreolysin (TrEMBL db: P83467) and aegerolysin, specifically expressed in primordia and fruiting bodies of P. ostreatus and A. aegerita, respectively . These homologous, thermolabile proteins have a molecular mass of ≈ 16 kDa, a low isoelectric point, and hemolytic activity at nanomolar concentrations. Searches in the nucleotide and protein databases revealed that the sequence of the ostreolysin N-terminal 50 amino acids was 88% identical with the putative PriA protein of P. ostreatus and its translated ESTs. It was also homologous with the cDNA-derived amino-acid sequence of the putative Aa-Pri1 protein , with its isoform aegerolysin , with Asp-hemolysin from the mold Aspergillus fumigatus, with two Clostridium bifermentans hemolysin-like proteins expressed during sporulation , and with hypothetical proteins from Neurospora crassa (TrEMBL db: Q8WZT0) and Pseudomonas aeruginosa (TrEMBL db: Q9I710). It has been speculated that Aa-Pri1, and similar proteins, may have important roles in the initial phase of fungal fruiting, such as hyphae aggregation , or in apoptosis . Their exact biological role is, however, not yet clear.
In this work, we have undertaken a functional characterization of ostreolysin, with the aim of shedding some light on its physiological role(s), and possibly on the role of the whole group of homologous proteins, so far observed only in fungi and bacteria. As hemolytic activity was a common trait, we focused first on this aspect, in particular on the interaction with the lipid membrane. We have found that ostreolysin permeabilizes red blood cells and tumor cells, by forming pores in their plasma membrane. We were also able to reproduce pore formation in artificial membranes formed of erythrocyte total lipid extracts, suggesting the presence of a specific lipid acceptor(s). When the ability of lipids to inhibit ostreolysin-mediated hemolysis was investigated, we observed a strong and specific inhibition by a series of lysophospholipids, in particular lysophophatidylinositol and sphingosine-1-phosphate, and, to a lesser degree, by nonesterified fatty acids. However, the membrane acceptor for pore formation did not appear to be a lysophospholipid. Our studies rather suggest that ostreolysin, and related proteins, in addition to being hemolytic, may be modulated by lysophospholipids.
Materials and methods
Proteins. Ostreolysin was purified from the fruiting bodies of freshly collected mushrooms as described previously . The protein stock solution was desalted, concentrated by ultrafiltration, and kept in aliquots at −20 °C. Before use, the protein was diluted in 140 mm NaCl/20 mm Tris/HCl buffer, pH 8.0 (vesicle buffer) unless otherwise stated. Nontoxic phospholipase A2, i.e. ammodytin I2 (880 U·mg−1), was a gift from Dr Igor Križaj, J. Stefan Institute, Ljubljana, Slovenia. Porcine trypsin in Hanks balanced salt solution, Bacillus sp. and Serratia marcescens proteases, Saccharomyces cerevisiae proteinase A, and Clostridium perfringens neuraminidase were all supplied by Sigma.
Cells. Bovine, sheep, or human erythrocytes were centrifuged from freshly collected citrated blood and washed twice with an excess of 0.9% saline and once with vesicle buffer. Transformed cell lines, HT 1080 from human fibrosarcoma and MCF 7 from human breast adenocarcinoma, were obtained from the Istituto Zooprofilattico Sperimentale della Lombardia e dell'Emilia, Brescia, Italy.
Lipids. A series of natural and synthetic lipids and derivatives were used. Egg phosphatidylcholine (PtdCho), 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (PamOleGroPCho), cholesterol, cardiolipin (PtdGroPtd), egg phosphatidic acid (PtdOH), egg phosphatidylethanolamine (PtdEtn), egg phosphatidylglycerol (PtdGro), liver phosphatidylinositol (PtdIns), 1,2-palmitoyl-oleoyl-sn-glycero-3-ethylphosphocholine (PamOleGroEthyl-PCho), egg sphingomyelin, and brain phosphatidylserine (PtdSer) were all obtained from Avanti Polar Lipids. Bovine brain gangliosides and cerebrosides, ceramides, l-α-lysophosphatidylinositol (lyso-PtdIns), egg yolk l-α-lysophosphatidylethanolamine (lyso-PtdEtn), egg yolk l-α-lysophosphatidylcholine (lyso-PtdCho), l-α-lysophosphatidylcholine, palmitoyl (lyso-PtdCho 16:0), l-α-lysophosphatidic acid, oleoyl (lyso-PtdOH), sphingosine 1-phosphate (Sph1P), myristic, palmitic and stearic acid, dilauroyl, dimyristoyl, dipentadecanoyl, dipalmitoyl and distearoyl phosphatidylcholine, and human low-density lipoprotein (LDL) were all from Sigma. Phosphatidylinositol 3-phosphate (PtdIns3P) was obtained from Matreya. All the lipids were dissolved in chloroform, or other organic solvents, in accordance with manufacturers' instructions.
Membranes of sheep erythrocytes were prepared by hypo-osmotic lysis in vesicle buffer diluted with distilled water (1 : 4), followed by 5 min centrifugation at 4 °C and 18 500 g. The supernatant was discarded, and the pellet was resuspended in vesicle buffer. The washing procedure was repeated 5 times to remove all cytosolic proteins. Total lipids were then extracted from pelleted membranes, essentially as described by Bligh & Dyer . The membranes were combined with 3 mL chloroform/methanol (1 : 2, v/v) and vortex-mixed for 30 s. Chloroform and water, 1 mL each, were then added, vortex-mixed again and gently centrifuged to separate solvent phases. The chloroform phase was removed, dried under argon, and used as total sheep erythrocyte lipids (SELs). To exclude the presence of small hydrophobic peptides co-extracted with the lipid phase, SEL extracts were dissolved in chloroform (50 µg·mL−1), applied to a standard TLC silica plate, and run with chloroform/methanol/acetic acid/acetone/water (35 : 25 : 4 : 14 : 2, v/v). The plates were then sprayed with the ninhydrin reagent and heated in an oven at 100 °C. Red–violet spots were observed only corresponding to the amine-containing lipids PtdEtn and PtdSer (as confirmed using the appropriate phospholipid standards). No other spots, not even at the origin or the front of the plate, were detected.
Total lipids from fruiting bodies of P. ostreatus, or fresh sheep brain, were obtained by Folch extraction : 10 g tissue and 10 mL distilled water were homogenized on ice, followed by centrifugation (26 300 g, 30 min, 4 °C), and extraction of the sediment with 100 mL chloroform/methanol (1 : 1, v/v). The extract obtained was separated from the sediment by centrifugation (110 g, 10 min, 25 °C). The sediment was re-extracted sequentially with chloroform/methanol (1 : 1, v/v), chloroform/methanol (1 : 2, v/v), and chloroform/methanol/water (60 : 30 : 4.5, v/v), 20 mL each. Respective supernatants were combined, dried by rotary evaporation, and kept at −20 °C under argon.
All the extraction procedures were performed in duplicate. To one half of all the extraction mixtures, 0.05% (w/v) butylhydroxytoluene (BHT; Sigma) was added as antioxidant; the other half was kept without antioxidant.
Other reagents. Poly(ethylene glycol)s were from Pharmacia or Fluka. Triton X-100 was from Merck. Eagle's minimum essential medium, Dulbecco's phosphate buffered saline, Mes, 3-(4,5-dimethylthiazolyl-2)-2,5-diphenyl tetrazolium bromide (MTT), and calcein were obtained from Sigma, and fetal bovine serum was from Euroclone (Wetnerby, UK).
Hemolytic activity was measured by a turbidimetric method as described previously . Typically, 25 µL ostreolysin solution in vesicle buffer was added to 175 µL either sheep, human or bovine erythrocyte suspension with an apparent D650 of 0.1. The decrease in D650 was recorded for 30 min using a kinetic microplate reader (Molecular Devices) to define the time necessary for 50% hemolysis (t0.5), and the maximal rate of hemolysis, i.e. the maximal slope of the hemolysis kinetics. All the experiments were performed at 25 °C. HC0.5 (µg·mL−1) was defined as the hemolysin concentration causing 50% of lysis in 2 min. If not otherwise stated, sheep erythrocytes were used.
For osmotic protection, glucose, sucrose, raffinose, stachiose, and a series of poly(ethylene glycol)s with molecular masses ranging from 900 to 6000 Da were used, as previously described . Sheep erythrocytes were mixed with 30 mm of an osmotic protectant in vesicle buffer, then 0.5 µg·mL−1 ostreolysin was added and the time course of hemolysis was followed for up to 60 min in the kinetic microplate reader. The size of the pores produced by ostreolysin was estimated as described by Renkin .
Cytotoxic activity of ostreolysin was assayed using HT 1080 (human fibrosarcoma cells) and MCF 7 (human breast adenocarcinoma cells). After thawing, cells were grown for a week in Eagle's minimum essential medium, supplemented with 10% fetal bovine serum, 2 mm l-glutamine, 0.15 mg·mL−1 gentamycin, and 1 mm sodium pyruvate in the case of MCF 7 cells. The cell lines were grown as monolayers in 75-cm2 tissue culture flasks, in a humidified CO2 incubator (5% CO2, 37 °C). When cells reached 80% confluence, they were washed using Dulbecco's phosphate buffered saline with 1 mm EDTA, and then trypsinized with 1 mL porcine trypsin (2.5 mg·mL−1 in Hanks balanced salt solution). Thereafter, 2 mL Eagle's minimum essential medium was added to block tryptic activity, and the cells were washed three times with Eagle's minimum essential medium without fetal bovine serum, by 5 min centrifugation at 200 g. Finally, they were resuspended in 1 mL of the same medium, counted in a Burker chamber, and plated at a similar cell density. Various dilutions of ostreolysin in culture medium were then added for 2 h, followed by 10% fetal bovine serum and a further 22 h of incubation (5% CO2, 37 °C). When this complete, cell viability was checked by a standard MTT test, as described .
Preparation of vesicles
Lipid films were formed by removing the organic solvents from a lipid solution in a rounded flask with rotary evaporation and final vacuum drying. Lipids, at a final concentration of 1–10 mg·mL−1, were swollen in vesicle buffer and vortex-mixed vigorously to obtain multilamellar vesicles. Small unilamellar vesicles (SUVs), or micelles of lysophospholipids and fatty acids, were prepared by 30 min pulsed sonication with a Vibracell ultrasonic disintegrator (Sonics and Materials), using output scale 4 and 50% duty cycle (room temperature). For large unilamellar vesicles (LUVs), the multilamellar vesicle suspension was subjected to eight cycles of freeze–thawing, and pressure-extruded through 0.1-µm polycarbonate filters (Millipore).
Permeabilization of lipid membranes was studied on LUVs loaded with fluorescent calcein. These were prepared essentially as above, except that calcein (at the self-quenching concentration of 80 mm) was included in the vesicle buffer. Extravesicular calcein was removed by gel filtration on a Sephadex G-50 (medium) column. Dimensions and homogeneity of the vesicles were routinely estimated by dynamic light scattering using a Malvern Zeta-Sizer 3 apparatus (Malvern, UK) as described previously .
Inhibition of ostreolysin-induced hemolysis
Binding of ostreolysin to lipids, human LDL and sheep erythrocyte membranes was estimated by measuring the residual hemolytic activity of unbound lysin. Typically, 75 µL micelles, SUVs, LUVs, LDL, or erythrocyte membranes, all at various concentrations in vesicle buffer, were pipetted into a multiwell plate. Then 25 µL ostreolysin (4 µg·mL−1) was added to each well, and the plate was incubated for 30 min at 37 °C to allow ostreolysin binding. Hemolysis was started by adding 100 µL erythrocyte suspension in vesicle buffer and recorded for 30 min. The lysing mixture had an initial D650 of 0.1.
To assess the effect of partial enzymatic hydrolysis, either pure phospholipids or LDL were treated with phospholipase A2 (ammodytin I2). In these experiments, LDL (1 mg protein·mL−1), or pure sonicated PtdCho, PtdOH, PtdEtn, PtdIns, PtdSer, and PtdGro (1 µmol lipid), were incubated for 5 min with 1 U ammodytin I2 in vesicle buffer supplemented with 2 mm CaCl2. Thereafter, 25 µL ostreolysin (4 µg·mL−1) was added, and the hemolytic assay was performed as described above. The appearance of lipid hydrolytic products of ammodytin I2, i.e. lysophospholipids and fatty acids, was confirmed by standard TLC and electrospray ionization mass spectroscopy. Ammodytin I2 alone did not affect ostreolysin-induced hemolysis, and was not hemolytic by itself, or in combination with LUVs or LDL.
Permeabilization of LUVs
Vesicle permeabilization were assayed in a fluorescence microplate reader (SLT Fluostar, Männedorf, Switzerland) with excitation and emission set at 494 nm and 520 nm, respectively. Ostreolysin at various concentrations in vesicle buffer (100 µL) was dispensed into a multiwell microplate, followed by an appropriate amount of calcein-loaded LUVs. The release of calcein was then recorded as described [15,16]. For pH values ranging from 6.5 to 9.5, we used the Tris/HCl vesicle buffer supplemented with 1 mm EDTA, and for values between 4.0 and 6.0 we used 140 mm NaCl/20 mm Mes/1 mm EDTA. Inhibition of calcein release by lysophospholipids was studied by either preincubating ostreolysin with various amounts of sonicated lyso-PtdIns for 20 min before adding the LUVs or preincubating LUVs and then adding ostreolysin.
Steady-state intrinsic fluorescence of ostreolysin, either alone or in combination with lipids, was measured at 25 °C in a Fluoromax spectrofluorimeter (Spex, Edison, NJ, USA) equipped with a thermostatically controlled cell holder and a magnetic stirrer. Excitation and emission slits were set at 5 nm. Samples were excited at 295 nm. Fluorescence emission spectra of 320 nm ostreolysin were taken over the range 300–450 nm. Intrinsic tryptophan fluorescence signals were corrected for the dilution factor, and the background was subtracted using the appropriate blanks. To monitor the kinetics of ostreolysin interaction with lipids (SUVs composed of SELs or sonicated lysophospholipids), tryptophan emission intensity was recorded at 339 nm. All the fluorescence measurements were taken in 50 mm Tris/HCl buffer, pH 8.0.
Proteins or proteolipid complexes were dissolved in electrophoresis buffer containing SDS without reducing agents. The samples were analyzed by SDS/PAGE using an 8–25% gradient polyacrylamide gel (Phast System; Pharmacia); gels were double stained, first with Coomassie Blue and then, after destaining, with silver nitrate.
FTIR spectroscopy was used to assess the secondary structure of ostreolysin in solution or adsorbed to the lipid phase by analysis of the amide I′ band as described . Ostreolysin was incubated for 30 min with LUVs composed of pure SELs (protein/lipid, 1 : 11.3, w/w) or LUVs composed of PamOleGroPCho/lyso-PtdIns (9 : 1, w/w) (protein/lipid, 1 : 1000, mol/mol), all in 10 mm Hepes buffer, pH 8.0. The mixtures were centrifuged, together with controls, in an Optima TL ultracentrifuge (Beckman). A fixed-angle rotor (TLA-100.2) was used at 400 000 g, for 1.5 h at 5 °C. After centrifugation, the supernatant and the pellet (resuspended in a starting volume of 10 mm Hepes buffer, pH 8.0) were checked for residual hemolytic activity and analysed by SDS/PAGE. Finally, they were deposited on germanium crystals and gently dried by nitrogen flushing. Spectra were collected, in an ATR geometry, on a FTS 185 spectrometer (Bio-Rad), with MCT detector, first on hydrated, and then on deuterated films, with or without a polariser set at either 0 ° or 90 ° (with respect to the plane of reflections).
In the case of lipid-bound protein, the spectrum of the protein was obtained by subtracting the contribution of the lipid alone, with a weight that minimized the band remaining at 1738 cm−1 (stretching of the carbonyl groups in the phospholipids). This was also necessary in view of the fact that SELs contain lipids comprising the ceramide moiety, e.g. sphingomyelin and gangliosides, that contribute a signal in the amide I′ region. This amounted to ≈ 40% of the total. The protein spectrum was then subtracted from the original lipid–protein one to provide the lipid-alone contribution.
Secondary structures were obtained from analysis of the amide I′ band. The original spectrum was deconvoluted to obtain the component frequencies, which were assigned as follows: bands in the regions 1696–1680 cm−1 and 1670–1660 cm−1, β-turn; band at 1672 ± 2 cm−1, antiparallel β-sheet; band at 1651 ± 3 cm−1, α-helix; band at 1640 ± 2 cm−1, random coil; bands in the region 1638–1616 cm−1, β-sheet (parallel plus antiparallel). These were used to curve fit the original spectrum, and the relative areas were taken as the proportion of the related structure present. Additional bands around 1610 cm−1 and 1600 cm−1, derive from side chain contributions, and were excluded from the total .
The lipid to protein ratio (L/P) in the pellet, was calculated from the following algorithm [18,19]:
where nres is the number of residues of the protein (assumed to be 140), A90° is the absorption with the 90 ° polarizer, and S are order parameters calculated from the ratio of the parallel and perpendicular absorption bands. SL is for the lipid chains, derived from the symmetric and asymmetric CH2 stretching (bands centered at 2850 and 2920 cm−1, respectively), using θ (the angle between the direction of the dipole moment change and that of the long axis of the molecule) set at 90 °. Samide I′ is the order parameter for the amide I′ band (between 1600 and 1700 cm−1, with θ = 0 °). The integrals were calculated from the corrected spectra, that with suffix L from the lipid alone, and that with suffix amide I′ from the protein alone. The order parameter for the α-helix, Sα, was obtained using the Lorentzian components at 1650 ± 3 cm−1 with θ = 30 °[20–22].
Hemolytic and cytotoxic activity of ostreolysin
Ostreolysin was able to lyse sheep, bovine or human erythrocytes, all with an HC0.5 of about 1 µg·mL−1 (or 64 nm). The time course of hemolysis was characterized by an initial lag phase followed by a relatively fast lysis, both dependent on protein concentration (Fig. 1A). Even when delayed, hemolysis always ran to completion (not shown). The maximal rate of hemolysis, but not the 1/t0.5 values, exhibited saturation with ostreolysin concentration (Fig. 1B).
Osmotic protectants larger than 1.500 kDa markedly decreased the rate of hemolysis, and complete protection was observed with molecules of 6000 kDa or more (Fig. 1C). The inner diameter of the ostreolysin-induced pore was estimated to be about 4 nm, by fitting the experimental data to the Renkin equation .
Exposure of HT 1080 (fibrosarcoma) and MCF 7 (mammalian tumor) cells to ostreolysin showed cytotoxicity with an effective concentration producing a 50% effect of 10 µg·mL−1 (or 640 nm, Fig. 2). Direct microscope observation of cell morphology confirmed that ostreolysin had a similar effect on both cell lines, producing swelling, blebbing and degranulation. The activity had already peaked after 2 h of incubation.
Inhibition of ostreolysin-induced hemolysis
Hemolytic activity of ostreolysin could be inhibited by preincubation with either washed erythrocyte membranes or LUVs of SELs, or from mushroom fruiting bodies or sheep brain. Notably, the extent of inhibition was strongly dependent on the presence or absence of an antioxidant (BHT) during the storage of the extracted lipids. Without BHT, the LUV concentration that decreased 1/t0.5 by a factor of two was 3 µg·mL−1 for SELs (Fig. 3), and about 6 µg·mL−1 and 120 µg·mL−1 for mushroom and brain lipid extracts (Table 1). When lipid extracts were stored with BHT, instead, a similar inhibition was obtained only at ≈ 50 times higher lipid concentrations.
Table 1. Inhibition of ostreolysin-induced hemolysis by natural and synthetic lipids, treated or not with phospholipase A2 (PLA2, ammodytin I2), oxidizing conditions (200 µm CuSO4 and 2 mm H2O2), or an antioxidant scavenger (BHT). Ostreolysin (0.5 µg·mL−1) was preincubated for 30 min at 37 °C with various amounts of lipid. Sheep erythrocytes were added and the remaining hemolytic activity was measured. The numbers reported are the concentration that caused 50% inhibition of ostreolysin-induced hemolysis (in µg·mL−1). NI, No inhibition (up to 750 µg·mL−1); –, not determined.
Dilution factor in vesicle buffer.
The same result was obtained with the polysaccharide chitin.
The same values were observed when eythrocyte membrane extracts were treated with Bacillus sp. or S. marcescens proteases, endonuclease, trypsin, and neuraminidase.
Ostreolysin inhibition was explained by permanent binding to SEL LUVs. In fact, after incubation with an excess of SEL LUVs, and ultracentrifugation of the mixture, almost all the protein was found in the sediment, as demonstrated by SDS/PAGE (Fig. 4A) and FTIR spectroscopy (Fig. 7). In the mean time, its hemolytic activity was completely abolished. As controls, LUVs alone also sedimented, whereby ostreolysin alone remained, fully active, in the supernatant. Similar experiments indicated that ostreolysin did not cosediment in a tight proteolipid complex with LUVs composed of PamOleGroPCho/lyso-PtdIns (9 : 1, w/w). By FTIR spectroscopic analysis, the amount of cosedimented protein was less than 15% of that observed with SEL LUVs, too little to give a band in SDS/PAGE (Fig. 4B, lane 5). Furthermore, in the case of PamOleGroPCho/lyso-PtdIns LUVs, both the sediment and the supernatant were hemolytic, suggesting that the small amount of sedimented protein was probably that entrapped between the LUVs. Together these results suggest the presence of ostreolysin acceptor molecule(s) in the erythrocyte membrane, the lipid nature of which was confirmed by the fact that interaction was not decreased by membrane treatment with proteases (Bacillus sp. or S. marcescens proteases, proteinase A, trypsin) or neuraminidase. In addition, the hypothetical presence of nonlipid components, such as short hydrophobic peptides, in SELs was directly excluded by TLC analysis. Notably, ostreolysin formed aggregates of around 34, 64, and 100 kDa (probably dimers, tetramers, and hexamers) when bound to SEL LUVs, but much less in the absence of lipids (Fig. 4), further evidence for a specific interaction.
We then assayed a series of pure lipids, or lipid mixtures, for ostreolysin inhibition (Table 1). None of the fully acylated lipids of varying length and degree of saturation, nor cholesterol were inhibitory, unless supplemented with a certain proportion of SELs, as shown for the case of PamOleGroPCho (Fig. 3). Ostreolysin-induced hemolysis was, however, markedly inhibited by pure sonicated lysophospholipids at concentrations at which these compounds were not themselves hemolytic (Fig. 5A). The most effective inhibitor was lyso-PtdIns, causing 50% reduction of 1/t0.5 at 0.2 µg·mL−1 (≈ 10−7m), whereas egg lyso-PtdCho, lyso-PtdCho 16:0, Sph1P, lyso-PtdEtn, and lyso-PtdOH induced the same effect at 0.7, 0.7, 1.1, 2.3, and 50 µg·mL−1 (or 10−4m), respectively. This suggests that their activity was neither dependent on the charge of the polar group (the two negatively charged, lyso-PtdIns and lyso-PtdOH, were the most and the least effective, respectively), nor on the fatty acid composition (lyso-PtdCho 16:0 and egg lyso-PtdCho had the same effect). Instead, it may depend, at least in part, on the chemical nature of the polar head, because lyso-PtdCho and Sph1P (with the same choline head group) had similar effects. All lysophospholipids showed a similar sigmoidal dose-dependent inhibition, except for Sph1P, which had a less steep dependence (Fig. 5A).
The interaction of ostreolysin with lysophospholipids was further analysed by preparing PamOleGroPCho LUV containing 10% of different lysophospholipids (Fig. 5B). Apart from LUVs containing Sph1P, the inhibitory ability of these mixtures was dramatically decreased with respect to pure lysophospholipids, and a similar level of inhibition was obtained only with concentrations at least 100-fold higher (corresponding to a 10-fold higher amount of the lysophospholipid present). The order of inhibitory activity was Sph1P > lyso-PtdIns > lyso-PtdCho > lyso-PtdEtn. When the amount of lyso-PtdIns included in PamOleGroPCho LUVs was varied (Fig. 5C), it was again apparent that the loss of inhibitory activity was larger than the corresponding decrease in lyso-PtdIns concentration.
The inhibition of ostreolysin by lysophospholipids was further confirmed by enzymatic hydrolysis of pure phospholipids. We found that even a partial hydrolysis of pure PtdCho, PtdOH, PtdEtn, PtdIns, PtdSer, and PtdGro by ammodytin I2, a phospholipase A2, markedly inhibited ostreolysin-induced hemolysis (Table 1). Furthermore, although intact LDL was not inhibitory, it became so after 10 min hydrolysis with ammodytin I2 (Table 1).
As the binding of single-chained lysophospholipids could be promoted by their fatty acid moiety, we also assayed myristic, palmitic and stearic acid (sonicated for 30 min) for inhibition. We found that all of them could inhibit ostreolysin-induced hemolysis, but only with a molar efficiency ≈ 10-fold lower than lysophospholipids. The inhibition was virtually independent of the fatty acid chain length (Table 1).
In agreement with the absence of inhibition, we found that calcein-loaded LUVs composed of pure phospholipids or sphingolipids and cholesterol, in various combinations, could not be permeabilized by ostreolysin. Even LUVs containing up to 10% of lyso-PtdIns, or made of total lipids from P. ostreatus (with or without BHT), were insensitive, despite being inhibitory to various extents. Only LUVs containing SEL extracts could be permeabilized. The extent of calcein release was dependent on the lysin dose and the pH of the bathing solution (Fig. 6). It was optimal in the pH range 8.0–9.0, where 0.5 µg·mL−1 ostreolysin produced 50% calcein release from SEL LUVs. In contrast with inhibition, SEL LUV permeabilization was not affected by the presence of BHT during lipid storage. The release was, however, abolished by the presence of sonicated lysophospholipids at sublytic concentrations.
FTIR spectra were recorded for ostreolysin alone, or cosedimented with either SEL LUVs or PamOleGroPCho:lyso-PtdIns (9 : 1, w/w) LUVs. The secondary structure of ostreolysin was estimated by FTIR spectroscopy, analysing the amide I′ band. Spectra were first deconvoluted to find a suitable set of single Lorentzian components, the sum of which was then used to curve-fit the original spectra. The single Lorentzian bands were attributed to four secondary structures (Fig. 7, Table 2). The resulting curves suggested that ostreolysin was composed of ≈ 50%β-structure (comprising 15%β-turn and 35%β-sheet), plus 20%α-helix and 30% random coil.
Table 2. FTIR spectroscopic determination of the secondary structure of ostreolysin with and without lipids. Values are mean ± SD. β1, Antiparallel β-sheet; β2, parallel and antiparallel β-sheet; t, β-turn; α, α-helix; r, random coil; βtot, total β-structure (i.e. β1 + β2 + t).
% Secondary structure
6 ± 1
28 ± 3
14 ± 2
19 ± 2
33 ± 3
Ostreolysin + SELs
3 ± 1
20 ± 2
11 ± 2
31 ± 3
35 ± 3
A significant association of ostreolysin with SEL LUV pellets was observed. The lipid/protein molar ratio estimated using Eqn (1) was ≈ 300, corresponding to a w/w ratio of 12 : 1. When compared with the precentrifugation ratio of 11 : 1, this suggested that more than 90% of the protein was associated. With PtdCho:lyso-PtdIns (9 : 1, w/w) LUVs instead, the estimated lipid/protein ratio was ≈ 1900, indicating a much weaker association, if any.
When ostreolysin bound to SEL LUVs, we observed an increase in its α-helical structure from 20% to 30% (Table 2). This occurred mainly at the expense of β-structures. Such an increase may suggest rearrangement of a portion of the protein with insertion of a newly formed α-helix into the lipid matrix. Interestingly, from the 0 ° and 90 ° polarized spectra of the inserted protein, it was possible to calculate the dichroic ratio of the helix and its orientation around the perpendicular to the plane of the membrane . We obtained an average angle of 45 °. However, considering that the average orientation shown by the lipid chains in the same spectra was between 42 ° and 44 °, we could recalculate the relative orientation of the α-helix with respect to the lipid chains , obtaining an angle of 20–22 °. This suggested an α-helix orientation nearly perpendicular to the plane of the membrane.
Intrinsic tryptophan fluorescence was finally used to explore changes in the local environment of ostreolysin. As in the case of permeabilization, only SEL vesicles affected the fluorescence of ostreolysin. The intensity increased and the emission maximum shifted from 339 to 333 nm (Fig. 8A), suggesting that at least some of the tryptophan residues of the protein are transferred into a more hydrophobic environment. The time course of fluorescence increase was rather fast, as shown in Fig. 8B. Fluorescence intensity was maximal at a lipid/protein ratio (w/w) above 16.3, corresponding to an approximate molar ratio of 400. No changes were detected on addition of pure lyso-PtdIns (Fig. 8C).
Our study provides direct evidence that ostreolysin has at least two different ways of interacting with lipids. First, it can permeabilize cell membranes and artificial lipid bilayers of specific composition, and secondly, it is modulated by lysophospholipids. In fact, the latter class of physiologically very important lipid derivatives efficiently inhibits the most adverse effect of this protein, i.e. cell lysis.
Ostreolysin is equally lytic to human, bovine and sheep erythrocytes, and only slightly less potent on some human tumor cell lines (Figs 1A and 2). There are several pieces of evidence that hemolysis is of the colloid-osmotic type, caused by the formation of ostreolysin pores in the lipidic portion of the cell membrane. Hemolysis could be prevented by the presence of osmotically active solutes large enough to exceed the pore size (Fig. 1C), as previously reported for other pore-forming proteins [10,24]. The Renkin estimate of the ostreolysin inner pore diameter was ≈ 4 nm, which is similar to that of flammutoxin, a 31-kDa cytolytic protein from the edible mushroom Flammulina velutipes. In addition, ostreolysin was able to release the fluorescent marker calcein (diameter ≈ 1.1 nm) from LUVs comprised of total erythrocyte lipids (Fig. 6). The pore may result from aggregation of several protein molecules, as we have observed the occurrence of SDS-resistant ostreolysin aggregates of two, four, and six monomers on SEL LUVs. Similar aggregates appeared only very faintly in the absence of lipids (Fig. 4A). The rather long lag phase preceding fast hemolysis (Fig. 1A) may also indicate that the formation of a functional pore requires the growth of ostreolysin aggregates on, or within, the erythrocyte membrane. The observation that maximal hemolysis rate was saturated at high ostreolysin concentrations, whereas 1/t0.5 was not (Fig. 1B), confirmed that lysin binding and aggregation (which are likely to affect 1/t0.5) are slower processes than diffusion of solutes through the opened pores (limiting the maximal rate). Asp-hemolysin, a similar protein, has also been reported to form large aggregates on erythrocytes, which could be visualized by electron microscopy .
Membranes made of the total lipid extract from fruit bodies of P. ostreatus were not susceptible to permeabilization, but were nonetheless strong inhibitors of ostreolysin-induced hemolysis. One explanation could be that ostreolysin may bind to these vesicles, but not permeabilize them. Another possibility is that they contain a diffusible component that can be transferred from the vesicle to the protein and inactivate it. The observation that lysophospholipids, either alone or in combination with other lipids (Fig. 5), can inhibit ostreolysin hemolytic and permeabilizing activity, but not cosediment it (Fig. 4), supports the latter explanation. Concentrations of various lysophospholipids necessary for 50% inhibition of hemolysis were always below their critical micellar concentrations, which is ≈ 70 µm for lyso-PtdIns  and 1.3 mm for lyso-PtdOH (Avanti Polar Lipids, web page). This suggests that lysophospholipids may inhibit ostreolysin pore-forming activity in their monomeric, rather than micellar, form.
It is known that biological membranes , and also LDL , contain various amounts of lysophospholipids that could be diffusively exchanged between membranes  and that lysophospholipid content may be increased by oxidative processes  or the action of phospholipase A2. Accordingly, we were able to modify both noninhibitory vesicles and normal LDL to become inhibitors of ostreolysin hemolysis by 2 h of oxidation with 2 mm H2O2 in the presence of 200 µm CuSO4, as reported also for Asp-hemolysin and LDL [31,32]. Moreover, the inhibitory activity, but not the permeabilization, was clearly higher if the vesicles were prepared from lipids stored without an antioxidant scavenger (Table 1). As oxidative degradation of phospholipids results in a variety of products, in addition to lysophospholipids , we also employed partial digestion of LUVs and LDL with ammodytin I2, a phospholipase A2, to prove that emerging lysophospholipids were in fact responsible for the observed inhibition (Table 1).
Fluorescence and FTIR spectroscopy confirmed that the interaction of ostreolysin with SEL membranes and lysophospholipids (or lysophospholipid-containing membranes) was different. Whereas addition of SEL LUVs enhanced the protein intrinsic fluorescence and blue-shifted the wavelength of emission maximum, lysophospholipids did not (Fig. 8). Similarly, the FTIR spectra revealed structural changes in ostreolysin on binding to SEL SUVs, but little or no binding with LUVs containing 10% lyso-PtdIns. The collective results (Figs 7 and 8, and Table 2) suggest that binding to the lipid bilayer and pore formation induced changes in ostreolysin conformation concomitant with transfer of certain tryptophan residues to a more hydrophobic environment, which were not seen on interaction with lysophospholipids. Such changes may correspond to the movement of the tryptophan residues into the lipid phase, or into a hydrophobic pocket created by aggregation of the protein molecules. According to the FTIR spectroscopic results, the transferred protein portion was conceivably an α-helix, inserted almost perpendicular to the plane of the membrane.
In conclusion, we have observed two different interaction modes of ostreolysin with lipids. One occurs on mammalian cell membranes, or on model membranes containing total erythrocyte lipids, and consists of tight and stable binding, followed by permeabilization via the formation of 4 nm pores. This goes together with tryptophan fluorescence changes indicative of protein insertion into the lipid film, and an increase in α-helix at the expense of β-structure. Such effects are independent of the presence of antioxidants. The other is the effect of lysophospholipids, and some related components such as fatty acids, which induce loss of the pore-forming ability, without permanent association with membranes containing these components, nor a change in the intrinsic fluorescence. This effect also occurs in the presence of membranes or proteins containing these components (e.g. mushroom membranes or LDL), probably via exchange of lysophospholipid molecules through the solution. It is enhanced by conditions that favor the production of lysophospholipids, e.g. the absence of scavengers, direct oxidation or phospholipase A2 attack.
The mode of action of ostreolysin is probably also common to other members of this protein family, so far found only in fungi and bacteria. For example, the homologous Asp-hemolysin of A. fumigatus was reported to be a lyso-PtdCho-binding protein  and to enhance infection of mice by this organism . However, the edible mushroom P. ostreatus is not a pathogenic species and it therefore seems unlikely that ostreolysin and similar proteins, such as Aa-Pri1, are toxins. Specific expression of ostreolysin and aegerolysin during initiation of fungal fruiting [4,5], expression of a hemolysin-like protein of B. bifermentans at the stage of sporulation , and our present observations may rather suggest that these proteins are involved in cell signaling. Lysophospholipids, in particular lyso-PtdIns and Sph1P, are in fact well-known signaling molecules which participate in a variety of cell processes  including differentiation . It is tempting to speculate that the level of lysophospholipids present in the organism may regulate the membrane-binding and pore-forming ability of these proteins, thus controlling their participation in cell differentitation in an as yet undiscovered way.
The Slovenian authors were supported by a grant from the Ministry of Science, Education and Sport of the Republic of Slovenia. The Italian authors were supported by a grant from Consiglio Nazionale delle ricerche (CNR), Istituto Trentino di Cultura (ITC) and, in part, also by a grant from Provincia Autonoma di Trento (PAT) Fondo Progetti (Project AgriBio). K. S. was the recipient of a NATO fellowship (band 219.33) during her stay in Trento. We thank Dr Gabriella Viero for technical help and Dr Graziano Guella for some TLC and electrospray ionization mass spectroscopy analyses.