Glycophorin as a receptor for Escherichia coli alpha-hemolysin in erythrocytes.

Escherichia coli alpha-hemolysin (HlyA) can lyse both red blood cells (RBC) and liposomes. However, the cells are lysed at HlyA concentrations 1-2 orders of magnitude lower than liposomes (large unilamellar vesicles). Treatment of RBC with trypsin, but not with chymotrypsin, reduces the sensitivity of RBC toward HlyA to the level of the liposomes. Since glycophorin, one of the main proteins in the RBC surface, can be hydrolyzed by trypsin much more readily than by chymotrypsin, the possibility was tested of a specific binding of HlyA to glycophorin. With this purpose, a number of experiments were performed. (a) HlyA was preincubated with purified glycophorin, after which it was found to be inactive against both RBC and liposomes. (b) Treatment of RBC with an anti-glycophorin antibody protected the cells against HlyA lysis. (c) Immobilized HlyA was able to bind glycophorin present in a detergent lysate of RBC ghosts. (d) Incorporation of glycophorin into pure phosphatidylcholine liposomes increased notoriously the sensitivity of the vesicles toward HlyA. (e) Treatment of the glycophorin-containing liposomes with trypsin reverted the vesicles to their original low sensitivity. The above results are interpreted in terms of glycophorin acting as a receptor for HlyA in RBC. The binding constant of HlyA for glycophorin was estimated, in RBC at sublytic HlyA concentrations, to be 1.5 x 10(-9) m.

Several pathogenic Gram-negative bacteria belonging to the genera Escherichia, Pasteurella, Bordetella, Morganella, or Actinobacillus produce and secrete 102-177-kDa toxic polypeptides that may become inserted into eukaryotic plasma membranes, leading to cell function impairment and eventually cell lysis (1). These toxins have been designated as the "RTX toxin family" on the basis of a common structural motif, namely a nonapeptide repeat, in the C-terminal part of the molecules. ␣-Hemolysin (HlyA) 1 from Escherichia coli is the prototype of this toxin family. Their members must undergo a transformation from a water-soluble to an insoluble, membrane-bound form to exert their effect on the target cell (for reviews, see Refs. [2][3][4][5]. Despite their similarities, members of the RTX family differ in cell target specificity. Whether the origin of this difference lies in the existence of different receptors is just beginning to be clarified. Results obtained with model phospholipid membranes for many of these toxins (6 -9) show that their lytic effects can be mimicked in the absence of a specific receptor. However, Lally et al. (10) showed that an integrin on the surface of human cells could be implicated in the recognition by Actinobacillus and Escherichia hemolysins. More recently, the CD18 component of ␤ 2 -integrin has been identified as a bovinespecific, leukocyte-specific receptor for Pasteurella hemolytica leukotoxin, a highly specific RTX toxin (11).
HlyA from E. coli belongs to the class of wide range host cell specificity toxins, and it has been reported to act on a variety of cell types from many species such as red blood cells, chicken embryo fibroblasts, rabbit granulocytes, mouse fibroblasts, lymphocytes, and macrophages (12). The cell type classically used for studies on the mechanism of action of HlyA has been the erythrocyte, but to date no receptor for the toxin in red blood cells has been reported. In fact, the binding studies by Eberspä cher et al. (13) using rabbit erythrocytes failed to show saturation, which was interpreted as HlyA binding nonspecifically lipids in the bilayer, although these authors did not exclude the possibility that their results might also be due to lack of saturation of receptors. The existence of a receptor has been hypothesized by Ludwig et al. (14) to explain some of their results. According to these authors, calcium was required for the lytic action of HlyA on erythrocytes but not on asolectin membranes. They concluded that the asolectin membrane did not contain the receptor to which the repeat domain of HlyA binds only in the presence of calcium in the erythrocyte membrane, but they did not demonstrate the existence of such a receptor. More recently, Bauer and Welch (15) described the saturable binding of hemolysin to sheep red blood cells with a maximum of 1000 -2000 molecules/cell, thus challenging the concept of nonspecific binding to the cell surface by RTX toxins.
In the present work, we have explored the putative role of glycophorin, an erythrocyte integral membrane protein, as a HlyA receptor. Our experiments have included intact and protease-treated erythrocytes and included liposomes (large unilamellar vesicles) with or without reconstituted glycophorin. We have found that hemolysin binds to glycophorin in erythrocytes, that this binding is abolished by trypsinization of the membranes, and that glycophorin purified from erythrocyte ghosts and reconstituted in liposomes increases significantly their sensitivity to the HlyA action. We conclude that glycophorin acts as a receptor for the toxin in this cell type.

Materials
Egg phosphatidylcholine was obtained from Lipid Products (South Nutfield, United Kingdom). Neuraminidase (from Clostridium perfrin-* This work was supported by Direccion General de Investigacion Cientifica y Tecnica Grant PB96-0171, Basque Government Grant EX99/05, and University of the Basque Country Grant G03/98. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ Recipient of a fellowship from the Basque government. § To whom correspondence should be addressed. Tel.: 34-94-6012542; Fax: 34-94-4648850; E-mail: gbzoseth@lg.ehu.es. 1 The abbreviations used are: HlyA, ␣-hemolysin; PC, phosphatidylcholine; LUV, large unilamellar vesicle(s); RBC, red blood cell(s); PE, phosphatidylethanolamine. gens), trypsin (from bovine pancreas), ␣Ϫchymotrypsin (from bovine pancreas), and glycosidase F (from Flavobacterium meningosepticum) were all obtained from Sigma. Horse erythrocytes were purchased from Microlab (Madrid, Spain). 8-aminonaphthalene-1,2,3-trisulfonic acid and p-xylene bis(pyridinium bromide) were from Molecular Probes, Inc. (Eugene, OR). Purified human glycophorin and an antibody against human glycophorin A were purchased from Sigma.

Purification of Glycophorin
Glycophorin from horse erythrocyte ghosts was purified according to a modification of the method described by Hamaguchi and Cleve (16). Lyophilized ghosts were suspended in 10 mM Tris (pH 7.4) and extracted with 9 volumes of chloroform/methanol (2:1). After removal of the chloroform/methanol, 9 volumes of ethanol were added to the collected water phases, resulting in a clear solution. Subsequently, 0.1 volume of a 0.1 M Tris, 10 mM EDTA (pH 7.4) solution was added, which resulted in precipitation of the glycoprotein. After centrifugation (16,500 ϫ g, 30 min, 4°C) the protein was redissolved in distilled water and dialyzed against 10 mM Tris-HCl (pH 9.0) and subsequently against 10 mM Tris-HCl (pH 7.4) and distilled water, prior to lyophilization. Final purification was achieved by extraction with a 50% phenol solution in water at low protein concentration and at room temperature, followed by centrifugation (65,000 ϫ g, 1 h, 0°C).

Incorporation of Glycophorin in Lipid Bilayers
Glycophorin was reconstituted in lipid vesicles by the method of MacDonald and MacDonald (17). Lyophilized glycophorin was dissolved in 1 mM Tris-HCl, pH 7.4, and 225 volumes of chloroform/methanol (2:1) were added. Egg phosphatidylcholine (PC) dissolved in the same solvent was evaporated to dryness together with the glycoprotein to form a dry lipid-protein film. Then the film was placed in a vacuum pump for at least 1 h to evaporate traces of solvent. The film was then hydrated with 12.5 mM 8-aminonaphthalene-1,2,3-trisulfonic acid, 45 mM p-xylene bis(pyridinium bromide), 20 mM Tris-HCl, 70 mM NaCl, (pH 7.4). To remove multilamellar structures containing little no or protein, the suspension was centrifuged at 10,000 ϫ g for 10 min at 4°C, and the pellet was discarded. Finally, a Sepharose 2B-300 column (1 ϫ 15 cm; Sigma) was used to remove glycophorin and dye not incorporated into the vesicles.

Vesicle Sizing by Quasielastic Light Scattering
The size of the reconstituted vesicles and of the control protein-free liposomes was characterized by quasielastic light scattering using a Malvern Zeta-Sizer spectrometer.

Determination of Vesicle-bound Glycophorin
To determine the recovery of glycophorin in the reconstituted vesicles, a modification of the resorcinol-sialic acid assay was used according to van Zoelen et al. (18). The standard curve was constructed with sialic acid and the same amount of lipid present in the assay vesicle aliquots. Sample aliquots were diluted to a standard volume of 0.5 ml with distilled water. 0.5 ml of resorcinol reagent, prepared as described by van Zoelen et al. (18), was added to each test tube. The tubes were placed in a boiling water bath for 20 min. The samples were allowed to cool to room temperature, and 1.25 ml of n-butyl acetate/1-butanol (85:15, v/v) were added with vortexing. The samples were then centrifuged for 10 min in a clinical centrifuge. The upper organic phase was transferred to glass cuvettes, and absorbance was read at 580 nm.

Trypsin Treatment of Proteoliposomes
Glycophorin-containing vesicles were incubated for 24 h at room temperature with continuous stirring in the presence of 1 mg/ml trypsin. After incubation, the liposomes were pelleted by centrifugation (125,000 ϫ g, 60 min, 4°C) and resuspended in fresh 20 mM Tris-HCl, 150 mM NaCl (pH 7.4) buffer. The sialic acid content of the treated liposomes was measured as described above.

Enzymic Treatments of Horse Erythrocytes
Erythrocytes were treated with trypsin (1 mg/ml), chymotrypsin (1 mg/ml), or neuraminidase (2.2 units/ml) at 37°C for 1 h. After incubation, the cells were washed several times to remove the enzymes. The sialic acid content of the cells was tested after the treatment in each case.

Hemolysis Assay
A standard red blood cell suspension was used, obtained by diluting the erythrocytes with saline so that 37.5 l of the mixture in 3 ml of distilled water gave an absorbance of 0.6 at 412 nm. Equal volumes of the standard suspension of washed horse erythrocytes were added to serial 2-fold dilutions of hemolysin in buffer (150 mM NaCl, 10 mM CaCl 2 , 20 mM Tris-HCl, pH 7.0). The mixtures were incubated at 37°C for 45 min and then left at room temperature for a few hours, so that erythrocyte sedimentation occurred. The absorbance of the supernatants, appropriately diluted with distilled water, was read at 412 nm. The blank (zero hemolysis) consisted of a mixture of equal volumes of buffer and erythrocytes.

Hemolysin Binding to Erythrocytes
For the standard binding assay, 1 ϫ 10 9 cells/ml were incubated with purified toxin for 15 min at 37°C and then collected by centrifugation at 16,000 ϫ g for 15 min at room temperature and washed with 5 mM phosphate buffer (pH 8.0). The pellet was then resuspended with 2ϫ sample buffer for electrophoresis and heated for 10 min at 100°C. Cell-associated toxins were quantified in immunoblots by comparing densitometric scans of bound toxin signals with calibrated signals of varying amounts of purified toxin electrophoresed and blotted under identical conditions. The binding data were analyzed using an equation described by Gutfreund (21) that relates the fractional saturation Y (equivalent to bound HlyA/maximal bound HlyA) with the total concentration of toxin as follows.
The equation holds for the case of a toxin binding to a single class of sites. In this case, the slope provides an estimate of the dissociation constant K d of the toxin-receptor complex, while the intersection with the x axis gives , the concentration of the binding sites.

Immunoblotting Analysis
Samples from SDS-polyacrylamide gels were transferred to nitrocellulose by the method of Towbin et al. (22). Blots were blocked with 10% skim milk in TBST buffer (10 mM Tris-HCl, 150 mM NaCl, 0.05% Tween 20 (w/v), pH 7.5) for 1 h at room temperature. They were then incubated with a solution containing a polyclonal rabbit anti-hemolysin (or antiglycophorin) antibody (1:1000) in 5% skim milk/TBST overnight at 4°C, washed with TBST buffer, and finally reacted with peroxidase-conjugated anti-rabbit Ig antibody (Sigma) (1:2000) in TBST buffer with 5% skim milk for 1 h at room temperature.
Immunoblots were developed by a femtogram sensitivity chemiluminescent method (Super Signal West femto maximum sensitivity substrate; Pierce).

Binding of HlyA to Glycophorin
HlyA Binding to a Hi-Trap Affinity Column-An N-hydroxysuccinimide-activated Hi-Trap column (1 ml) from Amersham Pharmacia Biotech was used to bind purified hemolysin. The column was first washed three times (3 ϫ 2 ml) with cold 1 mM HCl to remove isopropyl alcohol. Then the purified protein solution (500 g/ml) prepared in the coupling buffer (0.2 M NaHCO 3 , 0.5 M NaCl, 6 M urea, pH 8.3) was added and left overnight at 4°C. Note that HlyA is stable in 6 M urea (23). The deactivation of any excess active groups that had not coupled to the ligand protein and the washing out of the nonspecifically bound ligands were performed following exactly the procedure described by the supplier.
Binding Assay-Deoxycholate extracts of horse erythrocyte ghosts (1 mg/ml) were injected into the derivatized affinity column and left incubating for 1 h at 25°C. After this incubation, the column was washed with 4 volumes of 20 mM NaHCO 3 , 150 mM NaCl, pH 8.3, buffer to eliminate unbound proteins. Elution of the bound protein was performed by lowering the pH to 3.0 (0.1 M glycine, 150 mM NaCl, pH 3.0, buffer). The eluted fractions were concentrated, and pH was adjusted to neutrality. To visualize the protein, SDS-polyacrylamide gel electrophoresis and immunoblotting were performed as described previously using anti-glycophorin antibodies and ECL reagent (Western blotting detection reagent) from Amersham Pharmacia Biotech.

Antibody Preparation
Hemolysin (or glycophorin) was boiled for 3 min in electrophoresis buffer and subjected to preparative scale SDS-polyacrylamide gel electrophoresis. A section of the gel containing the protein band (ϳ500 g of protein) was treated as indicated in Rivas et al. (24) and injected into New Zealand White rabbits to raise hemolysin (or glycophorin)-specific antiserum, which jointly with horseradish peroxidase-conjugated antirabbit immunoglobulin G was used to visualize hemolysin (or glycophorin) by immunoblotting.

Protection against Hemolysis by Anti-glycophorin Antibodies
A polyclonal anti-glycophorin serum obtained as described above was cleaned with a Hi-Trap Protein G column (1 ml; Amersham Pharmacia Biotech) equilibrated with phosphate buffer (20 mM phosphate, pH 7.0). 0.5-ml serum samples were applied, and the retained antibodies were then eluted with 0.1 M glycine, pH 2.7, buffer. The pH of the collected fractions (14 ml) was raised to pH 7.0.
RBC standardized at A 412 ϭ 0.6 were incubated for 1 h at room temperature with the anti-glycophorin antibodies (dilution 1:2.5). After incubation, the treated erythrocytes were washed three times with 20 mM Tris, 150 mM NaCl, pH 7.0, buffer, and then a typical hemolysis assay was performed adding the toxin.
In a parallel assay, the RBC were treated with preimmune serum in the same way as described above.

RESULTS
When the lytic activity of HlyA was tested under similar conditions for both erythrocytes and liposomes (large unilamellar vesicles), in suspensions containing equal phospholipid concentrations, the dose-response curves showed a much higher sensitivity (over 1 order of magnitude) for the red blood cells ( Fig. 1 and Table I). The sensitivity of LUV varies with lipid composition, but the one used in the experiment shown in Fig.  1 provides LUV with highest sensitivity toward HlyA. Specifically, liposomes made of total red blood cell lipids or of a lipid mixture representing the cells outer lipid leaflet were even less sensitive (6). One reason that could explain the superior sen-sitivity of the cells would be the presence of a surface receptor. To test this possibility, erythrocytes were treated with trypsin under controlled conditions. An assay of HlyA-induced lysis on trypsinized erythrocytes revealed that their sensitivity had decreased to the level of liposomes ( Fig. 2 and Table I). This was a strong indication in favor of a protein receptor for HlyA on the erythrocyte surface.
A further clue was provided by the fact that red blood cell sensitivity to HlyA was not decreased when cells were treated with chymotrypsin (Table I). Glycophorin, a major integral membrane protein on the red blood cell surface, is known to be sensitive to trypsin digestion (25, 26) but resistant to chymotrypsin (27,28). Since glycophorin contains 60 weight % carbohydrate, including ϳ30 sialic acid residues per protein molecule (29), hemolysin was assayed on red blood cells incubated with either neuraminidase or glycosidase F. Under our conditions, neuraminidase released virtually all of the protein neuraminic acid. However, these carbohydrate-cleaving enzymes did not modify significantly the sensitivity of red blood cells toward HlyA (Table I), indicating that the integrity of the sugar moiety in the putative receptor was not essential for the lytic action of the toxin.
In a further series of experiments, HlyA was incubated with purified glycophorin and then added to either liposomes (Fig.  3A) or red blood cells (Fig. 3B). In both cases, binding of free glycophorin to HlyA decreased, and eventually abolished, the toxin lytic effect. The dose-dependent character of this effect is shown in Fig. 3B for the inhibition of hemolysis. The high glycophorin/HlyA molar proportion required for completely abolishing hemolysis was probably the consequence of an intrinsic tendency to aggregation by glycophorin (30). A qualitatively similar inhibition was observed when HlyA was preincubated with varying concentrations of a commercial preparation of human glycophorin (not shown).
The identity of the receptor with glycophorin was confirmed by an experiment in which horse erythrocytes were treated with an anti-glycophorin polyclonal antibody and then with HlyA. The results (Fig. 4) showed that, in the range of low toxin concentrations at which nonspecific binding to the bilayer does not occur (see Fig. 2), the antibody provided protection against lysis. Higher HlyA concentrations presumably displaced the antibody, since they both competed for the same receptor, and lysis was restored. Alternatively, the higher HlyA concentrations could be using the membrane lipid as target. RBC treatment with a preimmune serum did not provide any protection against HlyA challenge (Fig. 4). Very similar results were  observed when human RBC were incubated first with an antihuman glycophorin A antibody and then with hemolysin (not shown).
The specific binding of HlyA to glycophorin was demonstrated by an experiment in which red blood cell membranes were solubilized by 1% deoxycholate. The solubilized extract was then passed through a Hi-Trap affinity column containing bound HlyA. After repeated washing with buffer to remove unbound material, the column was eluted with a low pH buffer. The eluted fractions were concentrated, brought to neutral pH, and subjected to SDS-polyacrylamide gel electrophoresis and immunoblotting. The results are shown in Fig. 5. Glycophorin was retained by the HlyA affinity column (lane 3), and when the deoxycholate extract was incubated with an excess HlyA prior to its passage through the column, no binding was detected (lane 4).
Purified glycophorin was then reconstituted into pure PC bilayers as described under "Experimental Procedures." The resulting vesicles were about 400 nm in diameter, after quasielastic light scattering measurements, and the lipid/protein molar ratio was ϳ1100:1. As a control, pure PC vesicles were prepared by extrusion through 400-nm pore filters. The results of lysis assays performed with these liposomes are shown in Fig. 6. Incorporation of glycophorin into PC bilayers rendered the vesicles much more sensitive to HlyA, the difference being of 1 order of magnitude, similar to what was found for red blood cells in comparison with PC/PE/cholesterol liposomes (see Table I for the absolute D 50 values). Note that PC bilayers are intrinsically more resistant to HlyA than those composed of PC/PE/cholesterol (6). Pure PC was used for glycophorin reconstitution to obtain more homogeneous lipoprotein preparations. As a further control, liposomes containing reconstituted glycophorin were treated with trypsin as indicated under "Experimental Procedures." When treated with HlyA, trypsinized proteoliposomes became much less sensitive to the toxin, as indicated in Table I.
Precise determination of binding constants for glycophorin and HlyA (or for any RTX toxin receptor) is prevented by the fact that, since the toxin can also bind and lyse pure lipid bilayers, the classical experiments in which an excess of unlabeled ligand is added to displace nonspecifically bound labeled ligand are not feasible (10). A good estimate, however, could be obtained by measuring binding at low toxin concentration, under conditions where presumably nonreceptor-mediated binding to the lipid bilayer is negligible. This was achieved by  incubating red blood cells with increasing amounts of HlyA, washing as indicated under "Experimental Procedures," and separating the proteins by electrophoresis. The toxin was then quantified using a high sensitivity chemiluminescent method, by comparison with known standards treated in exactly the same way. A representative experiment is shown in Fig. 7. Binding appeared to increase hyperbolically with total toxin concentration. A derived plot according to Gutfreund (21) provided a straight line for the range of toxin concentrations under study. This plot was no longer linear for HlyA concentrations above 10 nM, probably because then nonreceptor-mediated binding became significant. The apparent binding parameters obtained using this procedure were as follows: K s ϭ 1.5 Ϯ 0.8 ϫ 10 Ϫ9 M and number of receptors per cell 3584 Ϯ 284 (average values Ϯ S.D.; n ϭ 3). DISCUSSION The nature and even the existence of a receptor for E. coli HlyA in eukaryotic cells have been the object of some debate. The findings of high lytic activities of the toxin on pure lipid vesicles and bilayers (6,8) indicated that no protein receptor was essential for the lytic effects of HlyA to be shown. More recent data by Lally et al. (10) and Li et al. (11) have demonstrated that HlyA and other RTX toxins recognize a specific protein, a ␤ 2 -integrin, on the surface of human target cells. In the present paper, we have shown that, in the presence of glycophorin, lipid bilayers become more sensitive to HlyA by over an order of magnitude and that HlyA binds with high affinity to red blood cells containing glycophorin. Thus, glycophorin appears to act as a receptor for HlyA, notwithstanding the fact that the toxin can also bind the membrane lipid matrix in the absence of receptors. Several isoforms and multiple variants of glycophorin are known (reviewed in Refs. 31 and 32), but no attempts have been made by us at this stage to analyze separately their HlyA binding abilities.
Other bacterial toxins share with HlyA its capacity to bind membranes with or without a receptor, in the latter case, of course, at higher concentrations. A particularly good paral- FIG. 5. Binding of HlyA to glycophorin. RBC ghost membranes were solubilized with deoxycholate and passed through an HlyA affinity column (see "Experimental Procedures"). Lane 1, purified glycophorin, directly processed through electrophoresis and blotting. Lane 2, whole deoxycholate extract. Lane 3, low pH eluent containing protein specifically bound to HlyA. Lane 4, as in lane 3, except that the deoxycholate extract had been incubated with soluble hemolysin (100 g/ml) prior to passage through the column. In all cases, a polyclonal rabbit anti-glycophorin antibody was used to visualize the protein as described under "Experimental Procedures."  lelism between rather different toxins exists between HlyA and aerolysin, a channel-forming, non-RTX toxin from Aeromonas spp. Aerolysin will lyse liposomes, at concentrations 2-3 orders of magnitude higher than RBC, and human glycophorin works as a receptor for this toxin (33,34). Recently, Vibrio cholerae E1 Tor hemolysin, also a non-RTX toxin, has been shown to recognize glycophorin B of human erythrocytes (35).
According to our measurements, the apparent number of receptors per cell is ϳ3500. This figure is similar to the one proposed by Bauer and Welch (15) (ϳ2000) but lower than the figure given by Eberspä cher et al. (13) (up to 50,000). The reason for the discrepancy is probably found in the procedures used. In our case, the binding experiments have been performed, by design, at sublytic toxin concentrations, and the same is true of the measurements by Bauer and Welch (15). Eberspä cher et al. (13), on the other hand, measured binding under lytic conditions. Thus, our figure represents probably the maximum number of receptors that can be occupied without cell lysis taking place under our conditions. In turn, this shows that a (large) number of HlyA molecules may bind the cell without producing lysis. The pathway from receptor binding to cell lesion is still to be explored, but the results by Moayeri and Welch (36), showing a different conformation for hemolysin when bound in a nonlytic (at 2°C) or a lytic conformation, are interesting in this context.
Both glycophorin and ␤ 2 -integrin are transmembrane, glycosylated proteins, that form homo-or heterodimers (27,37,38). However, they do not seem to have any significant degree of sequence homology. 2 Nevertheless, it should be noted that they are both cytoskeleton-linked proteins. (See in particular Ref. 32 for the association between glycophorin A and skeleton proteins.) This may be related to early ultrastructural studies showing that HlyA-treated RBC lose their discoidal shape, giving rise to spherocytes, with 10 -20 small projections spaced over the surface of the RBC (40). The latter authors found that the microscopic observations were consistent with an effect on the spectrin-actin cytoskeleton. More recently, receptor clustering, or receptor segregation, has been postulated as a cytoskeleton-related form of signaling (41). The possibility that the sublytic effects of HlyA (i.e. leukotriene generation, phosphatidylinositol hydrolysis, cytokine release, or superoxide generation (42)(43)(44)(45)(46)) could be related to receptor segregation in the cell surface, via cytoskeleton-anchored receptors, is worth considering.
Our study has not addressed the issue of the HlyA domain responsible for cell binding. This is an additional question that remains open. Ludwig et al. (14) suggested that the nonapeptide repeat in the protein would be responsible for receptor binding. Similarly, a mutant Pasteurella leukotoxin devoid of the amphipathic helices but conserving the acylated region as well as the repeat domain can bind and aggregate cells, but it does not cause lysis (47). While these data would be compatible with the idea of glycophorin as a receptor, Bauer and Welch (15) have indicated that several regions in the protein may be involved in cell binding. The latter concept can also explain the simultaneous presence of high and low affinity binding, respectively, to receptor protein and bilayer lipids. The polar nonapeptide repeat region would bind the receptor, and the nonpolar parts could interact directly with the lipid bilayer. One such nonpolar region is formed by the two acyl residues in active HlyA (linked covalently to Lys 564 and Lys 690 ) (48). The fatty acyl chains appear to be essential for insertion of HlyA into the lipid bilayer in a lytic conformation (49). Doubly lipidmodified protein motifs exhibit excellent membrane-anchoring features (50). It is interesting in this context that several Pseudomonas proteins are driven to eukaryotic plasma targets by fatty acylation (51).
A further point for consideration is the fact that RBC from different species present different susceptibilities to HlyA. 3 Since there is interspecific variability in glycoprotein structure (30,39), different affinities for HlyA of the various glycophorins can be at the origin of that phenomenon. Binding measurements with cells from different species will be required to test this hypothesis.