Binding of a pleurotolysin ortholog from Pleurotus eryngii to sphingomyelin and cholesterol-rich membrane domains

Abstract

A mixture of sphingomyelin (SM) and cholesterol (Chol) exhibits a characteristic lipid raft domain of the cell membranes that provides a platform to which various signal molecules as well as virus and bacterial proteins are recruited. Several proteins capable of specifically binding either SM or Chol have been reported. However, proteins that selectively bind to SM/Chol mixtures are less well characterized. In our screening for proteins specifically binding to SM/Chol liposomes, we identified a novel ortholog of Pleurotus ostreatus, pleurotolysin (Ply)A, from the extract of edible mushroom Pleurotus eryngii, named PlyA2. Enhanced green fluorescent protein (EGFP)-conjugated PlyA2 bound to SM/Chol but not to phosphatidylcholine/Chol liposomes. Cell surface labeling of PlyA2-EGFP was abolished after sphingomyelinase as well as methyl-β-cyclodextrin treatment, removing SM and Chol, respectively, indicating that PlyA2-EGFP specifically binds cell surface SM/Chol rafts. Tryptophan to alanine point mutation of PlyA2 revealed the importance of C-terminal tryptophan residues for SM/Chol binding. Our results indicate that PlyA2-EGFP is a novel protein probe to label SM/Chol lipid domains both in cell and model membranes.

Sphingomyelin (SM) is a major sphingolipid in mammalian cells. Together with cholesterol (Chol), SM forms specific lipid microdomains in both model (1) and biomembranes (2). These domains are designated as lipid rafts (3–5). Presently, rafts are defined as dynamic yet ordered sterol and sphingolipid-enriched nanoscale lipid assemblies. Lipid rafts provide a unique environment for specific proteins by stabilizing their metastable resting state and supporting protein activation via specific lipid-lipid, protein-lipid, and protein-protein interactions (6). It has been proposed that lipid rafts play a critical role in a number of cellular events, such as signal transduction and membrane transport (7, 8). Rafts have also been suggested to be the point of cellular entry of a wide range of viruses, bacteria, and toxins, as well as a site of viral assembly and formation of both prions and Alzheimer amyloid (9–12).

Yet, the concept of lipid rafts has not reached a consensus in terms of their composition, size, or even actual functional existence in living cells (13, 14). Thus, reliable lipid probes are urgently needed in order to define the membrane lipid organization in situ. A plethora of cytolytic proteins have been shown to interact with raft-like domains (15). Several nontoxic mutants of these toxins that specifically bind either SM (16–18) or Chol (19) have been employed to study the distribution, dynamics, and function of SM/Chol domains. Because both sphingolipids and Chol are required to assemble lipid rafts (6), a protein that binds to the SM/Chol complex would be a very useful tool to study lipid microdomain organization in cell membranes.

Two mushroom proteins belonging to the aegerolysin protein family (20) are reported to bind SM/Chol membranes, namely ostreolysin (Oly) (21) and pleurotolysin (Ply)A (22). Oly is an ∼15 kDa acidic protein from the edible mushroom Pleurotus ostreatus (oyster mushroom) (21). The binding of Oly to SM/Chol and 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC)/Chol membranes has been reported (23). Electron paramagnetic resonance and Fourier-transformed infrared spectroscopy revealed the importance of highly ordered sterol-enriched membrane domains for the binding of Oly (24). When the Chol level is above 30 mol%, Oly exhibits cytotoxic effects by permeabilizing membranes composed of SM (23). PlyA, from P. ostreatus, binds to SM membranes containing ≥30 mol% Chol (22). PlyA requires formation of a complex with PlyB, an ∼59 kDa protein, to exhibit cytotoxic effects (22). However, detailed characterization of PlyA has not yet been performed. Recently, erylysin (Ery), an ortholog of Oly, was isolated from Pleurotus eryngii (25). Similar to Ply, Ery is a two-component hemolysin composed of EryA and EryB. To date, the binding specificity of Ery has not been studied.

In this study, we identified a novel ortholog of PlyA, termed PlyA2, from the extracts of P. eryngii, capable of binding specifically to SM/Chol liposomes. Subsequently, we constructed an enhanced green fluorescent protein (EGFP)-PlyA2 chimera and characterized its lipid specificity in a model and cellular membranes. Our results demonstrate that PlyA2-EGFP is a novel probe to study SM/Chol-rich lipid rafts.

MATERIALS AND METHODS

Lipids

The following were purchased from Avanti Polar Lipids (Alabaster, AL): SM (brain, porcine); L-α-phosphatidylcholine (PC) (egg, chicken); 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC); DPPC; 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4-yl) (N-NBD-PE); 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE); L-α-phosphatidylserine (PS) (brain, porcine); lactosyl ceramide (LacCer) (Galβ1-4Glcβ1-1Cer); and galactosyl ceramide (GalCer) (Galβ1-1Cer). From Sigma (St. Louis, MO): Chol; ergosterol; β-sitosterol; GM1 [Galβ1-3GalNAcβ1-4(Neu5Acα2-3)Galβ1-4Glcβ1-1Cer; brain, bovine]; sphingosine (SPH); and D-ribo-phytosphingosine (Phyto) (yeast). From Wako Pure Chemical Industries (Osaka, Japan): GM2 [GalNAcβ1-4(Neu5Acα2-3)Galβ1-4Glcβ1-1Cer; brain, bovine]. From Matreya (Pleasant Gap, PA): GM3 (Neu5Acα2-3Galβ1-4Glcβ1-1Cer; buttermilk, bovine). From Research Biochemicals International (Natick, MA): GD1a [Neu5Acα2-3Galβ1-3GalNAcβ1-4(Neu5Acα2-3)Galβ1-4Glcβ1-1Cer]. From Enzo Life Sciences (Farmingdale, NY): sphingosylphosphorylcholine (SPC).

Other materials

Sphingomyelinase (SMase) from Staphylococcus aureus and methyl-β-cyclodextrin (MβCD) were from Sigma. Anti-bis(monoacylglycero)phosphate/lysobisphosphatidic acid antibody was prepared as described (26). Anti-EEA1, anti-GM130 antibodies were from BD Bioscience (Franklin Lakes, NJ). Anti-CD59 was from Chemicon International, anti-GM3 from Seikagaku Corporation (Tokyo, Japan), and transferrin-Alexa Fluor 546 conjugate was from Molecular Probes. Two peptides of PlyA2 consisting of residues 45–60 and 99–115 were synthesized, and rabbit polyclonal antibodies against the cocktail of these peptides were prepared by Medical and Biological Laboratories Co., Ltd. (Japan).

Preparation of P. eryngii extract

P. eryngii (50 g) was homogenized in 50 ml of 50 mM Tris-HCl buffer (pH 8.5) containing protease inhibitor. The homogenate was centrifuged at 11,600 g for 30 min at 4°C. The supernatant was centrifuged further in a Beckmann Optima ultracentrifuge at 58,863 g for 30 min at 4°C. The final supernatant was aliquoted and kept at −80°C.

Liposome sedimentation assay

Multilamellar vesicles (MLVs) were prepared by hydrating lipid film with PBS (pH 7.5) and vortex mixing. P. eryngii extract or purified-proteins were incubated with MLVs in PBS for 30 min at 37°C. The mixture was centrifuged at 21,600 g for 10 min at room temperature. The liposomes were washed twice with PBS and were subjected to SDS-PAGE followed by Coomassie brilliant blue (CBB) staining.

Determination of amino acid sequence of PlyA2

After blotting on polyvinylidene difluoride membranes, protein bands were stained with CBB. The protein bands were excised and subjected to Edman degradation using a Procise cLC or HT protein sequencing system (Applied Biosystems).

Molecular cloning of the cDNA encoding PlyA2

A coding sequence for PlyA2 was obtained from P. eryngii cDNA by PCR amplification.

Construction, expression, and purification of recombinant proteins

Fragments of PlyA2 were cloned into the expression vector pET28 containing Hexa-histidine (His) fragment (Merck KGaA, Germany). The coding sequences of PlyA2 tryptophan mutants were optimized, and the DNA fragments were synthesized by GenScript through OptimumGene^TM codon optimization analysis. The resulting fragments were cloned into pET-28 vectors. To construct the vectors for purification of EGFP-fused proteins, gene encoding EGFP was first cloned into pET-28 vectors, and then PlyA2 was recloned into the N terminal or C terminal of EGFP in pET-28-EGFP vectors, respectively. The resulting plasmids were transformed into Escherichia coli strain BL-21(DE3). Bacterial culture was grown at 30°C in lysogeny broth medium with 100 μg/ml kanamycin until an optical density (OD 600 nm) of 0.5 was reached. Expression of EGFP-protein was induced by the addition of 0.1 mM isopropyl-β-D-thiogalactopyranoside overnight and the cells were harvested by centrifugation at 5,000 g for 10 min at 4°C. Cell lysates were prepared by resuspending the pellet in binding buffer consisting of 20 mM sodium phosphate buffer (pH 7.4) containing 500 mM NaCl, 20 mM imidazole, and protease inhibitor cocktail set I (Calbiochem), followed by sonication (Misonix). Insoluble protein and cell debris were removed by centrifugation at 9,000 g at 4°C for 20 min. For purification of EGFP-proteins, a Ni-NTA (Qiagen) column was used according to the supplier's protocol. The resulting fractions were concentrated, followed by exchanging buffer (PBS, pH 7.5) using an Amicon Ultra-15 Ultracel®-30K centrifugal unit (Millipore). The protein concentration was determined using the Bradford assay.

Liposome flotation assay

MLVs containing 0.05% N-NBD-PE were incubated with proteins in PBS at room temperature for 30 min. One-half milliliter of this suspension was mixed with 1 ml of 2.1 M sucrose in PBS, loaded at the bottom of an ultracentrifuge tube (MLS50; Beckman Coulter), and overlaid with 1.5 ml of 1.25 M sucrose and 2.25 ml of 0.8 M sucrose. The gradient was centrifuged for 1 h at 35,608 g at 4°C using a Beckman Coulter Optima^TM ultracentrifuge. Top fraction (200 μl) was collected and subjected to SDS-PAGE and Western blotting. Fluorescence intensity of N-NBD-PE was measured (λex, 483 nm; λem, 533 nm) to monitor the position and concentration of liposomes in the gradient.

Cell culture

HeLa cells were grown at 37°C in Dulbecco's modified Eagle's medium (DMEM) (Invitrogen, Carlsbad, CA) supplemented with 10% FBS (18)

Cell surface labeling after SMase or MβCD treatment

HeLa cells were treated with 1.25 IU/ml SMase or 5 mM MβCD in DMEM for 30 min at 37°C. Cells were then fixed with 4% paraformaldehyde, treated with 50 mM NH₄Cl and 0.1% BSA, and labeled with 1 μg/ml PlyA2-EGFP.

Effect of preincubation with liposomes on PlyA2-EGFP labeling

PlyA2-EGFP was preincubated with 1 mM (total lipid) brain SM/Chol (1:1) and POPC/Chol (1:1) MLVs in PBS for 30 min at 37°C. Then, the mixture was incubated with fixed HeLa cells for 30 min at room temperature. After washing, the specimen was observed under a confocal microscope.

Colocalization of PlyA2-EGFP and lipid raft markers

HeLa cells were fixed with 4% paraformaldehyde, blocked with 0.1% BSA, and doubly labeled with 1 μg/ml PlyA2-EGFP and anti-CD59 or anti-GM3. In the case of transferrin receptor staining, prior to fixation, the cells were washed with ice-cold DMEM and incubated with transferrin-Alexa Fluor 546 conjugate (250 μg/ml) on ice for 30 min. Cells were then labeled with 1 μg/ml PlyA2-EGFP.

Confocal microscopy

Confocal images were obtained on a Zeiss 510 or Zeiss 700 confocal microscope equipped with a C-Apochromat 63XW Korr (1.2 NA) objective.

Intracellular labeling of HeLa cells with PlyA2-EGFP and organelle markers

HeLa cells were fixed with 4% paraformaldehyde, permeabilized by freezing in liquid nitrogen then thawing, blocked with 0.1% BSA, and labeled with 1 μg/ml PlyA2-EGFP and various antibodies.

Measurement of hemolysis

The hemolytic activity of PlyA2-EGFP was measured as described (27). Sheep erythrocyte suspensions (1 ml, containing 3 × 10⁷ cells) were incubated with various concentrations of PlyA2-EGFP at 37°C for 30 min and then centrifuged at 500 g for 5 min to precipitate the erythrocytes. Aliquots of the supernatants were taken, and the optical densities at 415 nm were measured to determine the percentage of hemoglobin released from the erythrocytes. Total hemoglobin contents were determined by measuring hemoglobin released after freezing and thawing of the erythrocyte preparations.

Expression of PlyA2 in cytoplasm

The PlyA2 fragment was cloned into pAcGFP-Hyg-N1 vector (Clontech) used for expression. Transient transfection was performed in DMEM containing FBS by using Lipofectamine LTX reagent (Invitrogen) according to the instructions provided by manufacturer.

Homology modeling

The amino acid sequence was submitted to the homology modeling and threading server I-TASSER with or without specified restraints (28–30). The protonation state of the resulting top rated model was predicted utilizing the H++ server (31). Subsequently, the models were parameterized, solvated, and equilibrated for 10 ns at 310 K in the presence of 0.15 mM KCl utilizing the NAMD software package (32) and the CHARMM 36 force field (33). The bonded interactions were calculated every time-step, full range electrostatic interactions every other time-step, short-range nonbonded interaction cutoff was set to 12 Å and a smoothing function employed at 10 Å, and pair-lists were recalculated every 10 time-steps with a pair-list distance of 14 Å. The temperature and pressure (1 bar) were maintained utilizing Langevin dynamics and Langevin piston method, respectively. After reevaluation and adjustment of the protonation state by H++ server, the models were equilibrated for an additional 10 ns and subjected to a 10 ns production run. Molecular structures were visualized by visual molecular dynamics (34).

RESULTS

Two mushroom-derived aegerolysins, Ply and Oly, have been reported to bind SM/Chol membranes (22, 23). In order to find a new SM/Chol binding protein, we screened for proteins that bind liposomes composed of SM and Chol (1:1) using extracts of P. eryngii. The mushroom extract was incubated with SM/Chol (1:1) liposomes for 30 min at 37°C, centrifuged, and the pellet and supernatant were subjected to SDS-PAGE (Fig. 1A). The liposome-bound pellet gave a low molecular weight single band (Fig. 1A, lane 4, arrow). The bound protein was extracted from the gel and subjected to amino acid sequencing as described in Materials and Methods. The N-terminal amino acid sequence of the protein was determined as AYAQXVIIII. Then, using an in-house P. eryngii transcriptome database, the protein was identified to be a 138 amino acid polypeptide which exhibits two amino acid substitution (K22V and T100A) from the PlyA from P. ostreatus (35) (Fig. 1B, triangles). This was an identical protein to the protein that we recently identified in the transcriptome of P. eryngii as a homolog of PlyA (36). We designated the gene as Pe. Pleurotolysin A (Pe. PlyA) (Accession number AB777517). Here we name the protein PlyA2. PlyA2 belongs to the aegerolysin family and exhibits a high homology to other family members, EryA and ostreolysin (Oly) (80% identity). EryA and Oly share a high degree of sequence identity (98%) (25).

Fig. 1.

Isolation and identification of SM/Chol binding protein PlyA2 from P. eryngii. A: Screening of SM/Chol binding protein from P. eryngii. Lane 1, molecular weight (MW) marker. Lane 2, mushroom extract. Lane 3, supernatant fraction after incubation of P. eryngii extract with SM/Chol (1:1) MLVs for 30 min at 37°C followed by sedimentation. Lane 4, pellet fraction after incubation. B: Alignment of amino acid sequence of PlyA2 with PlyA, EryA, and Oly. Protein names are given on the left. Sequence numbers refer to PlyA2 and are shown above. Triangles indicate dissimilar amino acid residues between PlyA and PlyA2. Arrows indicate conserved tryptophan residues. Single letters, amino acids; black background, identical; gray background, conserved substitution.

Next, we prepared recombinant PlyA2 which was conjugated to EGFP on the N terminus (EGFP-PlyA2) as well as the C terminus (PlyA2-EGFP). The new constructs were mixed with SM/Chol (1:1), 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC)/Chol (1:1), and SM/POPC (1:1) liposomes to characterize their binding specificity. In all cases EGFP-PlyA2 was exclusively recovered in the supernatant (Fig. 2A). This strongly suggests that EGFP fusion to the N terminus of PlyA2 completely abolished the lipid binding capability. In contrast, PlyA2-EGFP was detected in the pellet fraction upon incubation with SM/Chol (1:1) liposomes (Fig. 2B). A considerable amount of the protein still remained in the supernatant, suggesting that the protein is weakly associated to the liposomes. Nevertheless, complete recovery of PlyA2-EGFP in the supernatant of POPC/Chol and SM/POPC liposomes strongly indicates that the presence of both SM and Chol are required for binding.

Fig. 2.

Binding specificity of EGFP-tagged PlyA2 to various liposomes. N-terminally EGFP-tagged PlyA2 (EGFP-PlyA2) (A) or C-terminally EGFP-tagged PlyA2 (PlyA2-EGFP) (B) were incubated with various liposomes followed by centrifugation as described above. The supernatant (S) and pellet (P) fractions were subjected to SDS-PAGE followed by CBB staining. SM/Chol, brain sphingomyelin/cholesterol (1:1) liposomes; POPC/Chol, 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine/cholesterol (1:1) liposomes; SM/POPC, brain sphingomyelin/1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (1:1) liposomes; DPPC/Chol, 1,2-dipalmitoyl-sn-glycero-3-phosphocholine/cholesterol (1:1) liposomes; PS/Chol, brain phosphatidylserine/cholesterol (1:1) liposomes; GM1/Chol, Galβ1-3GalNAcβ1-4(Neu5Acα2-3)Galβ1-4Glcβ1-1Cer/cholesterol (1:1) liposomes; GM2/Chol, GalNAcβ1-4(Neu5Acα2-3)Galβ1-4Glcβ1-1Cer/cholesterol (1:1) liposomes; GM3/Chol, Neu5Acα2-3Galβ1-4Glcβ1-1Cer/cholesterol (1:1) liposomes; GD1a/Chol, Neu5Acα2-3Galβ1-3GalNAcβ1-4(Neu5Acα2-3)Galβ1-4Glcβ1-1Cer/cholesterol (1:1) liposomes; LacCer/Chol, Galβ1-4Glcβ1-1Cer/cholesterol (1:1) liposomes; GalCer/Chol, Galβ1-1Cer/cholesterol (1:1) liposomes; SPC/Chol, sphingosylphosphorylcholine/cholesterol (1:1) liposomes; SPH/Chol, sphingosine/cholesterol (1:1) liposomes; Phyto/Chol, D-ribo-phytosphingosine/cholesterol (1:1) liposomes.

The SM and Chol mixture forms a characteristic liquid-ordered membrane (37). A liquid-ordered phase is also observed in the mixture of Chol and PC with saturated fatty acid (37). Thus, in Fig. 2B, we examined the binding of PlyA2-EGFP to the mixture of Chol and DPPC. The detected negligible binding to DPPC/Chol liposomes further supports the notion that PlyA2-EGFP specifically recognizes SM/Chol assemblies and does not simply associate with membranes in the liquid-ordered phase. Additionally, PlyA2-EGFP did not exhibit any binding affinity towards mixtures of Chol and other phospholipids such as PS, or glycosphingolipids such as GM1, GM2, GM3, GD1a, LacCer, and GalCer, or degradation products of sphingolipids and intermediates of sphingolipid biosynthesis such as SPC, SPH, and Phyto (Fig. 2B).

The influence of sterol structure on PlyA2-EGFP binding specificity was characterized by incubating the protein with equimolar mixtures of SM and various sterols (Fig. 3). PlyA2-EGFP exhibited a similarly high binding affinity to SM/Chol as well as to SM/β-sitosterol and SM/7-dehydrocholesterol, but only a weak binding affinity toward SM/ergosterol. In contrast, SM mixtures containing lanosterol, 6-ketocholestanol, and epicholesterol completely abolished the binding of PlyA2-EGFP. This indicates the importance of the stereochemical configuration of the 3-hydroxyl group on SM/sterol domain organization and PlyA2-EGFP binding specificity, while modifications of the sterol tail exhibited only little influence. In contrast, the presence of bulky groups on the A- and B-ring of the sterol backbone completely abolished PlyA2-EGFP binding.

Fig. 3.

Binding specificity of PlyA2-EGFP to SM vesicles containing various sterols. A: PlyA2-EGFP was incubated with SM (lane 1), SM/Chol (1:1) (lane 2), SM/β-sitosterol (1:1) (lane 3), SM/lanosterol (1:1) (lane 4), SM/7-dehydrocholesterol (1:1) (lane 5), SM/ergosterol (1:1) (lane 6), SM/6-ketocholestanol (1:1) (lane 7), and SM/epicholesterol (1:1) (lane 8) for 30 min at 37°C. The liposome-bound protein and free protein were separated by a sucrose gradient centrifugation. The top fractions were subjected to SDS-PAGE followed by Western blotting using anti-His antibodies. B: Chemical structure of sterols used in the experiment. Shaded regions indicate the structural difference with Chol.

While our in vitro liposome results indicate that PlyA2-EGFP specifically binds to SM/Chol membranes, we cannot exclude the possibility that the protein also binds other cellular components, such as membrane proteins or glycoconjugates. Therefore, we examined the effect of PlyA2-EGFP preincubation with liposomes on cell surface staining (Fig. 4). Preincubation with SM/Chol liposomes abolished cell labeling, whereas in the case of POPC/Chol liposome preincubation, cell labeling was not affected. Additionally, treatment of cells with SMase or MβCD, to remove cell surface SM or Chol, respectively (18, 38), abolished PlyA2-EGFP binding (Fig. 4). This strongly indicates that PlyA2-EGFP selectively binds to a SM/Chol complex on the cell surface, rendering it a suitable probe to label SM/Chol-rich membrane domains.

Fig. 4.

Effects of various treatments on the labeling of HeLa cells with PlyA2-EGFP. Control, PlyA2-EGFP only; SM/Chol and POPC/Chol, PlyA2-EGFP preincubated with indicated liposome mixtures; SMase and MβCD, cells treated with SMase and MβCD, respectively prior to fixation and PlyA2-EGFP labeling. Scale bar, 20 μm.

We then compared the cell surface distribution of PlyA2-EGFP with established lipid raft markers (Fig. 5). Figure 5 indicates that there was partial colocalization between PlyA2-EGFP and raft-associated glycosylphosphatidylinositol-anchored protein CD59 (arrows) (39, 40). Based on dot-by-dot comparison, about 45% of CD59-positive dots were colocalized with PlyA2-EGFP. In contrast, only 5% of nonraft protein, transferrin receptor visualized with plasma membrane bound Alexa-labeled transferin (39), was colocalized (images are shown in Fig. 5). In contrast to the protein raft marker, ganglioside GM3 was not significantly colocalized with SM/Chol domains labeled by PlyA2-EGFP (Fig. 5). GM3 antibody stained about 30% of our HeLa cell population. The different localization of raft lipids is consistent with our previous observation that lysenin did not colocalize with GM1-binding cholera toxin in Jurkat cells (16).

Fig. 5.

Colocalization of PlyA2-EGFP with lipid raft markers. Cells were fixed and doubly labeled with PlyA2-EGFP and anti-CD59 or anti-GM3. Alternatively, cells were labeled with transferrin followed by fixation and PlyA2-EGFP labeling. Scale bar, 10 μm.

In order to visualize the intracellular distribution of SM/Chol (Fig. 6) domains, HeLa cells were fixed and permeabilized by freezing and thawing to avoid lipid loss prior to treatment with PlyA2-EGFP. Double labeling of PlyA2-EGFP with an antibody specific to bis(monoacylglycero)phosphate/lysobisphosphatidic acid, the late endosome marker (26, 41), indicates the enrichment of SM/Chol domains in the internal membranes of late endosomes. On the contrary, no significant colocalization of PlyA2-EGFP with EEA1 and GM130, organelle markers for early endosomes and cis-Golgi, respectively, was observed.

Fig. 6.

Intracellular distribution of PlyA2-EGFP labeling in HeLa cells. Cells were fixed and permeabilized by freezing and thawing. Cells were then doubly labeled with PlyA2-EGFP and various organelle markers. Scale bar, 20 μm.

No hemolytic activity of PlyA2-EGFP against sheep red blood cells was detected up to a concentration of 10 μg protein/ml (Fig. 7A), akin to its homolog PlyA. Staining of the plasma membrane of living HeLa cells with PlyA2-EGFP revealed no cytotoxic effect (Fig. 7B). Similarly, expression of PlyA2-AcGFP in HeLa cells did not exhibit any obvious cytotoxic effect (Fig. 7C). Interestingly, PlyA2-AcGFP did not label the cytoplasmic leaflet of the plasma membrane, suggesting an asymmetric distribution of suitable SM/Chol-rich domains between the leaflets of the plasma membrane.

Fig. 7.

PlyA2-EGFP labels living cells. A: Hemolysis of sheep red blood cells was measured as described in Materials and Methods. B: Living HeLa cells were labeled with exogenously added PlyA2-EGFP. C: PlyA2-AcGFP was expressed in the cytoplasm of HeLa cells as described in Materials and Methods. Scale bar, 20 μm.

Previously we demonstrated the importance of conserved tryptophan residues for the SM binding specificity of lysenin family proteins (42). PlyA2 contains six tryptophan residues at positions 6, 28, 92, 96, 103, and 112 (Fig. 1B), and these tryptophan residues are well conserved among PlyA, PlyA2, Oly, and EryA. Mutations were introduced in PlyA2 by replacing individual tryptophan with alanine. The resulting mutants were designated as W6A, W28A, W92A, W96A, W103A, and W112A. These proteins were expressed in E. coli as C-terminus His-tag conjugates. Two proteins, W92A and W103A, could not be expressed in E. coli. The remaining mutants were subjected to flotation assays to evaluate their binding affinity to SM/Chol liposomes (Fig. 8A). The binding of the W28A and W112A mutants to SM/Chol liposomes was considerably reduced and in the case of the W96A mutant, completely abolished. In contrast, the W6A mutant retained its binding affinity and specificity. Similarly, only the W6A mutant exhibited a comparable staining pattern to that of wild-type PlyA2 in fixed HeLa cells, while the other clones failed to stain the cells (Fig. 8B).

Fig. 8.

Tryptophan mutants fail to bind SM/Chol-rich membranes. A: Recombinant PlyA2 or tryptophan mutants were incubated with SM/Chol or PC/Chol liposomes for 30 min at 37°C. Proteins bound to liposomes were recovered by a sucrose gradient centrifugation and detected by Western blotting using anti-His antibodies. B: Fixed HeLa cells were incubated with PlyA2 or tryptophan mutants, followed by treatment with anti-His antibodies and the secondary antibodies conjugated with Alexa 488. Scale bar, 20 μm.

Due to the absence of sufficiently homologous protein templates, the PlyA2 model was generated via an iterative process based on protein threading, amino acid charge prediction, and molecular dynamic simulation as described in Materials and Methods. The first model (model 1) was generated by unrestrained submission of the PlyA2 amino acid sequence to I-TASSER (supplementary Figs. I, II) and was based on chain A of 3VSF (exo-β-1,3-galactanase from Clostridium thermocellum). Model 1 achieved a confidence score (C-score) of −2.05 on a scale of −5 to 2 of low to high confidence respectively, where a C-score of about −1.5 and higher is considered to indicate a correct fold of the model (43). Interestingly, the second ranked template structure, 3LIM_A (FraC from sea anemone Actinia fragacea), is a member of the actinoporin family of pore forming toxins. Actinoporins, such as equinatoxin II (Eqt2) [Protein Data Bank (PDB) code 1IAZ, 1O72] and sticholysin II (Stl2) (PDB code 1GWZ), are known to bind to SM-rich membranes (44). Sequence alignment revealed an ∼15% sequence identity and up to 65% sequence similarity between PlyA2 and prominent members of the actinoporin family (Fig. 9). The extended gap region corresponds to the complete α2-helix of the actinoporin templates. Still, a high similarity of the folding state as well as the binding mechanism between members of the aegerolysin and actinoporin families has been proposed, based on circular dichroism (CD) and fluorescence spectra (45). Resubmission of PlyA2 sequence aligned to FraC produced the second PlyA2 model (model 2) with a C-score of −1.76 and a TM-score of 0.5 (Fig. 10). In line with CD spectral data of Oly (45), both models exhibit a high degree of β-structure. In contrast, only model 2 matched the α-helix content of the CD data, despite omission of the α2-helix of the template structure. During the initial equilibration period of the molecular dynamic simulation, only the low quality region after the extended gap of model 2 was significantly altered. Residue 112 moved nearly 2 nm from its initial position, until the favorable aryl/aryl interaction with F91 stabilized its position for the remaining simulation. The location of the relatively close C terminus was not altered due to a stabilizing salt bridge between K25 and D132. In both models, W92 and W103 exhibit an edge-to-face orientation, indicating their importance to stabilize aryl/aryl cross-strand interactions (46). This is in good agreement with the unsuccessful expression of W92A and W103A, possibly due to a failure of the mutants to acquire the correct folding state. W96 is located at the loop between two β-strands, namely β5-β6 (model 1) and β4-β5 (model 2), resembling W112 in FraC. Mutation studies of W112 of FraC revealed its importance during initial membrane binding due to primarily hydrophobic interaction (47). Mutation of W112 of FraC to alanine resulted in a complete loss of binding, similar to our W96A mutant. Abolishment of stabilizing W28-F90 aryl/aryl cross-strand interactions could be the cause for reduced binding affinity of the W28A mutant, according to model 2. In contrast, in model 1, the reduced binding affinity after W28 mutation is difficult to rationalize. Superimposition of the β-sheet core of both models with Stl2 (PDB code 1O72) in the presence of choline phosphate (CP) revealed the close proximity of W112 in model 2 to the superimposed CP binding site of Stl2. In contrast, loss of binding of the W112A mutant cannot be explained in terms of model 1. In summary, model 2 appears to be in better agreement with experimental results compared with model 1.

Fig. 9.

Alignment of amino acid sequence of PlyA2 with prominent members of the actinoporin family. Protein names are given on the left. Sequence numbers are indicated above and refer to Eqt2. Choline binding site residues of the actinoporin family are indicated by arrows. Single letters, amino acids; –, alignment gap; black background, identical; gray background, conserved substitution.

Fig. 10.

Cartoon representation of model 2. Position of displayed residues as indicated. N, N terminus; C, C terminus.

DISCUSSION

In this study, we screened for SM/Chol binding proteins from the edible mushroom P. eryngii and identified a new member of the aegerolysin family, namely PlyA2. EGFP chimeras of aegerolysin family proteins promise to be useful lipid probes. Yet, conjugation of EGFP to small proteins of only 15–17 kDa may alter their activity and selectivity. Indeed, in our hands, only PlyA2-EGFP retained its SM/Chol binding activity, among several aegerolysin proteins tested.

PlyA2-EGFP bound to SM/Chol as well as SM/7-dehydrocholesterol liposomes. In contrast, no binding affinity toward SM/ergosterol and DPPC/Chol liposomes was detected. Both 7-dehydrocholesterol and ergosterol are known to facilitate liquid-ordered sphingolipid/sterol domains (48). This indicates that PlyA2-EGFP binding is based on direct protein-lipid head-group interaction and is not due to the liquid-ordered state of the membrane. Thus, PlyA2-EGFP selectively binds liquid-ordered domains composed of SM and Chol. In contrast, Oly is reported to bind to SM/Chol and DPPC/Chol liposomes, indicating a preference for Chol-rich liquid-ordered phase (23). PlyA2-EFGP binding to SM/sterol liposomes is highly sensitive to alterations of the sterol A and B rings. In contrast, bulky modifications of the sterol hydrocarbon side-chain did not influence binding affinity, indicating PlyA2 binding is predominantly restricted to the head-group region of the SM/sterol assembly.

The importance of tryptophan residues for the activity of SM-binding toxins, such as Eqt2 (49, 50) and lysenin (42), is well established. Six tryptophan residues are conserved within the aegerolysin family members PlyA2, PlyA, Oly, and EryA. Our point mutants W28A, W96A, and W112A lost their binding affinity to SM/Chol liposomes and the cell surface. In contrast, the W6A mutant retained its binding specificity. Previously, it has been shown that the N-terminal 14 amino acid residues of PlyA are important for membrane attachment (35). Indeed, 10 out of these 14 amino acid residues are categorized as hydrophobic, and based on model 2, are clustered at the N-terminal half of the α-helix (supplementary Fig. III). In contrast, the C-terminal half of the α-helix is dominated by a cluster of polar amino acid residues, likely protruding from the membrane. This is in line with the observed loss of binding activity of EFGP-PlyA2 due to N-terminal conjugation.

The homology model 2 is not only in good agreement with our point mutation and EGFP-conjugation results, but also satisfies CD spectroscopic data. Sequence alignment of PlyA2 with members of the actinoporin family indicated that the choline binding site residues are similar and the central proline residue is conserved (Fig. 9, actinoporin binding site residues indicated with arrows). Nevertheless, tyrosines 133, 137, and 138, integral components of the actinoporin choline binding site located on the α2-helix, are absent. Optimized superimposition of the β-strand core of model 2 and Stl2 in the presence of CP suggests that the absence of these three tyrosine residues is compensated by residues F75, Y91, and W112 (Fig. 11).

Fig. 11.

Putative choline binding site of model 2. A: Model 2. B: Superposition of amino acid residues (as indicated) from Stl2 choline binding site. Color scheme of model 2: carbon (cyan), oxygen (red), and nitrogen (blue). Color scheme of Stl2: carbon (lime), oxygen (pink), nitrogen (iceblue), and phosphorous (purple).

To date, several proteins capable of binding specifically to SM or Chol have been reported, such as lysenin, Eqt2, and the nontoxic C-terminal fragment of perfringolysin O (domain 4) (51). Lysenin, a pore forming toxin from earthworm, binds to SM clusters (52). Nontoxic truncated lysenin revealed the heterogeneity of lipid rafts (16) and the role of SM-rich domains during cell division (18). In contrast, Eqt2 labeled selectively SM-rich domains in the cytoplasmic leaflet of the Golgi apparatus (17), indicating the presence of different pools of SM in cell membranes (53). Utilizing domain 4, a Chol-binding probe, and photoactivation localization microscopy, we demonstrated that SM-rich and Chol-rich domains do not always give a similar cell surface labeling pattern (38). Additionally, we recently identified a novel SM/Chol binding protein from the mushroom Grifola frondosa (Makino et al., in preparation). This highlights the importance of multiple probes capable of recognizing different assemblies of SM and Chol to understand the detailed organization of cellular membranes.

In conclusion, PlyA2-EGFP is a nontoxic lipid binding protein capable of selectively associating with SM/Chol-rich membrane domains. Cell surface labeling was achieved by exogenous addition, while expression in the cytoplasm did not result in labeling of internal membranes. Partial colocalization of PlyA2-EGFP and raft-resident protein indicates that PlyA2-EGFP labels a sub-population of lipid rafts. Consequently, PlyA2-EGFP is a complementary addition to the lipid probe toolkit to study lipid distribution and dynamics in situ.