(Received for publication, March 27, 1996, and in revised form, September 9, 1996)
Division of Immunology and Vaccine Development, National Institute of Cholera and Enteric Diseases, Calcutta 700010, India
Vibrio cholerae hemolysin is an
extracellular pore-forming monomeric protein with a native molecular
weight of about 60,000. In this study, we showed that the hemolysin
interacted with immobilized phospholipids and cholesterol and formed
oligomers in vesicles constituted from phospholipids alone with a
stoichiometry identical to those produced in rabbit erythrocyte
membrane. However, the hemolysin bound to glycoproteins with terminal
1-galactosyl residues and an association constant of 9.4 × 107 M
1 was estimated for the
hemolysin-asialofetuin complex by solid phase binding assay.
Oligomerization of the hemolysin in lipid bilayer converted the
sugar-binding monomer to a lectin with strong carbohydrate-dependent hemagglutinating activity
accompanied by inactivation of hemolytic activity and loss in ability
to interact with phospholipids. There was no evidence for erythrocyte
surface carbohydrates playing an essential role in interaction of the hemolysin with the cell. However, specific glycoproteins inhibited hemolysis of rabbit erythrocytes as well as interaction of the hemolysin with phospholipid. The results suggest (i) V. cholerae hemolysin is a monomer with distinct domains associated
with specific binding to carbohydrates and interaction with lipids,
(ii) the pore-forming property depends solely on the protein-lipid
interaction with no evidence for involvement of sugars, and (iii)
specific sugars can down-regulate the ability of the hemolysin to form pores in lipid bilayers.
Vibrio cholerae is a Gram-negative bacterium that produces cholera, a severe diarrheal disease of humans (1). In addition to cholera toxin (1), the critical enterotoxigenic factor of V. cholerae, the organisms generally express a protein that induces lysis of erythrocytes and other mammalian cells (2). The excreted form of this protein, the hemolysin, was purified and characterized as a monomer with a native molecular weight of 60,000 (3), cloned (4), and sequenced (5). The purified protein causes accumulation of fluid in a ligated rabbit ileal loop (6), suggesting that the protein is potentially capable of contributing to the enterotoxigenicity of V. cholerae (7).
Experiments on the mode of action of V. cholerae hemolysin
on the erythrocyte membrane indicated that the protein forms
transmembrane anion-selective oligomeric channels with a diameter of
0.9-1.0 nm, leading to colloid osmotic swelling of the cell and lysis (8, 9). A similar mechanism involving insertion of the monomeric cytolysin in the membrane lipid bilayer and lateral oligomerization by
diffusion in the bilayer plane to a channel-like assembly with a
hydrophobic outer and hydrophilic inner surface was earlier proposed
for pore-forming cytolysins in general (10) and demonstrated for a wide
range of cytolytic proteins such as Staphylococcus aureus
-toxin (11). Since events leading to membrane permeabilization are
reproduced in artificial lipid bilayers constituted from phospholipids alone (8, 12), nonlipid constituents of the membrane such as proteins
and carbohydrates do not appear to be critically involved in the
process. However, there is evidence which indicates that glycoconjugates do affect pore-forming activity of cytolysins. For
example, S. aureus
-toxin is inhibited by
GM1-ganglioside (13) and band 3 anion transport
glycoprotein of the erythrocyte membrane (14), and Aeromonas
hydrophila hemolysin is inactivated by erythrocyte membrane
glycopeptides (15). These observations suggest that specific
carbohydrates might regulate the interaction of the cytolysins with
membrane bilayers and their oligomerization behavior. However, the
possible existence of a carbohydrate-dependent regulatory
mechanism for a primarily lipid-mediated process has not been studied
or even recognized.
In this study, we have showed that V. cholerae hemolysin is
a lectin-like protein with a preference for the terminal
1-galactosyl moiety of glycoproteins. Our results provided evidence
that specific glycoconjugates can regulate the self-assembly of the
hemolysin in the erythrocyte membrane and artificial lipid
bilayers.
V. cholerae strain V2 used for the isolation of the hemolysin was a clinical non-01 isolate, a generous gift from R. Sakazaki, Tokyo, Japan. Brain heart infusion (dehydrated medium) was from Difco. Glycoproteins, enzymes, lectins, and sugars were from Sigma. Chromatographic matrices, gel electrophoresis chemicals, and standard protein molecular weight markers were from Pharmacia, Uppsala, Sweden.
Protein PurificationV. cholerae non-01 V2 was cultured in brain heart infusion broth at 37 °C with shaking for 18 h. Bacteria were removed by centrifugation at 20,000 × g for 15 min at 4 °C, and the culture supernatant was concentrated 50-fold by filtration through a PM-10 membrane (Amicon). Lipid-protein vesicles in the ultrafiltrate were removed by sedimentation at 100,000 × g for 2 h. The supernatant was fractionated by 50% saturation with (NH4)2SO4, and the crude hemolysin was subjected to hydrophobic interaction chromatography on phenyl-Sepharose CL-4B column (38 × 1.6 cm) equilibrated with 25 mM Tris-HCl buffer containing 1 mM EDTA, pH 7.6 (buffer A). The column was washed with 3 bed volumes of buffer A, and the adsorbed proteins were eluted with a linear 0-60% ethylene glycol gradient. Fractions with hemolytic activity eluted as a sharp peak at 30% ethylene glycol concentration and were pooled, concentrated by ultrafiltration, and applied on a Bio-Gel P-100 column (39 × 1.6 cm) equilibrated with 100 mM NaCl in buffer A. The elution volume of the hemolysin corresponded to a molecular weight of 60,000 as determined by calibrating the column with a mixture of bovine serum albumin, ovalbumin, chymotrypsinogen, and lysozyme. Column eluates were monitored for protein concentration by absorbance at 280 nm and assayed for hemolytic activity. Homogeneity of the preparation was examined by polyacrylamide gel electrophoresis (PAGE)1 under native and denaturing conditions.
The oligomeric form of the hemolysin was isolated by incubating the purified hemolysin monomer (1 mg) with 5 ml of a suspension of phosphatidylcholine (PC)-cholesterol (1:1 by weight) vesicles showing a turbidity of 2.0 at 600 nm for 2 h at 25 °C. The hemolysin that was not incorporated into lipid vesicles and remained free in aqueous solution was removed by centrifugation at 10,000 × g for 10 min. The hemolysin oligomer entrapped in liposome was dispersed in 1% sodium deoxycholate in buffer A and subjected to size exclusion chromatography on Sephacryl S-300 (39 × 1.6 cm). The hemolysin oligomer eluted in the void volume and was dialyzed against phosphate-buffered saline (PBS, pH 7.2) for removal of the detergent.
Assay of Hemolytic and Hemagglutinating ActivityThe hemolyic activity was assayed by monitoring spectrophotometrically the release of hemoglobin at 541 nm or the decrease in turbidity of the erythrocyte suspension at 650 nm (16). The hemagglutinating activity was monitored visually.
Enzyme Treatment of ErythrocytesRabbit erythrocytes were desialylated by incubating the cell suspension (10% v/v) in 20 mM sodium phosphate, 45 mM sodium acetate, 120 mM NaCl buffer, pH 6.0, with Clostridium perfringens neuraminidase (Sigma; 1 unit/ml) for 60 min at 37 °C. Desialylated erythrocytes were washed and incubated in the same buffer with galactose oxidase (Sigma; 5 units/ml) for 2 h at 37 °C (17). Desialylation of erythrocytes and subsequent degradation of cell surface galactose residues were confirmed by expected changes in agglutinability of cells with peanut agglutinin (18).
Preparation of LiposomeA mixture (1:1 by weight) of PC and cholesterol was evaporated to dryness under reduced pressure, dispersed in PBS, and sonicated for 10 min at 23 kHz (Ultrasonic Cell Disruptor, Microson, Secfroid). Multilamellar vesicles were centrifuged off at 12,000 × g for 10 min at 4 °C.
Analytical ProceduresProtein was estimated by a modified Folin-Ciocalteu method (19) using bovine serum albumin as a standard. SDS-PAGE was performed (20) in a 12.5% or 4-20% linear polyacrylamide gradient gel. Desialylation of glycoproteins was performed by incubation in 50 mM sodium acetate, 10 mM CaCl2 buffer, pH 5.5, with C. perfringens neuraminidase (1 unit/5 mg) for 1 h at 37 °C and monitored by estimation of sialic acid by the thiobarbituric acid method (21). Periodate oxidation of asialofetuin was carried out by incubating the glycoprotein (10 mg) in 25 mM sodium acetate buffer, pH 4.5, with 25 mM sodium metaperiodate at 4 °C for 22 h in the dark followed by decomposition of the excess periodate by ethylene glycol (22).
Anti-hemolysin AntiserumAntiserum to the purified hemolysin was raised by injecting male albino rabbits intramuscularly with 100 µg of the protein emulsified with Freund's complete adjuvant (Difco) followed by three booster injections of 100 µg of protein alone at an interval of 10 days. Animals were bled 4 days after the last injection.
Solid-phase Binding AssayBinding of the purified hemolysin to immobilized asialofetuin and lipids was quantitated by enzyme-linked immunosorbent assay (ELISA) described previously (23). Wells of polystyrene microtiter plates (Corning) were coated in triplicate with glycoprotein (1 µg/100 µl of PBS) or lipids (1 µg/100 µl of methanol) by overnight incubation at 4 °C. They were washed thrice with PBS, 0.05% Tween 20, and free sites were blocked with 200 µl of defatted milk powder (3% in PBS), washed, and incubated with various concentrations of the hemolysin for 2 h at 25 °C. Wells were decanted and washed as above. The hemolysin bound by the immobilized glycoprotein or lipids was estimated by sequential incubation with rabbit anti-hemolysin antiserum and horseradish peroxidase-conjugated goat anti-rabbit IgG. Nonspecific interaction was minimized by including in the incubation mixture defatted milk powder (1%). Color was developed with o-phenylenediamine and monitored at 492 nm. Wells coated with defatted milk powder served as blanks.
Purified V. cholerae hemolysin induced 50%
hemolysis of a 1% suspension of rabbit erythrocytes at a concentration
of 33 ng/ml. As expected for a process dependent on self-aggregation of
monomers (10), the equilibrium value of the percent hemolysis showed sigmoidal dependence on the hemolysin concentration and was sensitive to the amount of protein within a rather narrow concentration range
from 30 to 40 ng/ml (Fig. 1A). The time
course of hemolysis of rabbit erythrocytes was relatively more
sensitive to the concentration of the hemolysin, with a lag phase that
increased sharply with decreasing hemolysin concentration (Fig.
1B). V. cholerae hemolysin showed erythrocyte
species preference, although not very pronounced, with rabbit
erythrocytes being 6- and 8-fold more sensitive than human and sheep
cells, respectively.
As reported by earlier workers (9), the hemolysin inserted itself in the rabbit erythrocyte membrane and formed oligomers that migrated in SDS-PAGE with an apparent molecular weight of 240,000 when incubated with SDS at 60 °C (Fig. 1C). Insertion of the hemolysin in the membrane and subsequent oligomerization to an SDS-stable tetramer was also observed in PC-cholesterol (Fig. 1D, lane 6) as well as PC vesicles (Fig. 1D, lane 5). Since vesicles constituted from PC alone could effectively mimic the role of erythrocyte membrane in inducing oligomerization of the hemolysin, the process did not require participation of nonlipid constituents of the biomembrane. The interaction of hemolysin with immobilized cholesterol as determined by an ELISA-based assay (Fig. 1E) was significantly stronger than that with PC or phosphatidylethanolamine (PE). However, the extent of difference did not suggest a marked discrimination of polar head groups of the lipids by the protein.
Hemolysin Oligomer, a Nonlytic Lectin with Hemagglutinating ActivityThe hemolysin oligomer formed in PC-cholesterol vesicles was isolated by solubilization of the protein-lipid complex in 1% sodium deoxycholate followed by gel filtration on Sephacryl S-300 in the presence of the detergent. The oligomer, unlike the monomer that was stable indefinitely at 4 °C, tended to precipitate in aqueous solution in storage and showed polydispersity during size exclusion chromatography on Sepharose CL-4B (not shown), indicating nonstoichiometric aggregation of the oligomers. The oligomer had no hemolytic activity toward rabbit erythrocytes at a concentration as high as 1 mg/ml. Absence of hemolytic activity of the protein corresponded with complete loss of ability to interact with immobilized PC, PE, or cholesterol (Fig. 1E). Interestingly, the hemolysin oligomer agglutinated rabbit erythrocytes strongly at a concentration of 10 ng/ml and human or sheep erythrocytes at about a 200-fold higher concentration.
Although monosaccharides such as mannose, glucose, galactose,
L-fucose, N-acetylglucosamine, and
N-acetylgalactosamine were noninhibitory at a
concentration of 100 mM, fetuin and calf thyroglobulin, glycoproteins with multiple galactose-containing asparagine, and serine/threonine linked sugar units (25) inhibited the hemagglutinating activity of the oligomer (Table I). Ovalbumin or bovine
pancreatic DNase, each of which contains a single oligomannosidic unit
devoid of galactose (25) did not inhibit the oligomer (Table I).
Inhibitory activities of fetuin and thyroglobulin increased 10-20-fold
by exposure of the penultimate 1-galactosyl moiety by desialylation (Table I). Treatment of asialofetuin with periodate, which selectively degrades the terminal galactose residue (26) led to complete loss of
inhibitory activity (Table I). These observations demonstrated that
agglutination of erythrocytes by the hemolysin oligomer involved binding to carbohydrate receptors, with the protein displaying a
distinct preference for terminal
1-galactosyl residue of complex glycoconjugates.
|
Oligomerization of the hemolysin to a hemagglutinating
lectin suggested that the monomer, although incapable of agglutinating erythrocytes by cross-linking apposing cells apparently due to lack of
multivalency, might itself be endowed with carbohydrate-binding activity. The hemolysin bound to asialofetuin with an association constant of 9.4 × 107 M1 as
determined by the Scatchard plot (27) of the ratio of the bound to the
free hemolysin against bound hemolysin (Fig.
2A) but not to ovalbumin or periodate-treated
asialofetuin, which lack terminal
1-galactosyl moiety (not shown).
Further, binding of the hemolysin to asialofetuin was partially
reversed by preincubation of the protein with galactose and
N-acetylgalactosamine but not by mannose (Fig.
2B). Pretreatment of immobilized asialofetuin with peanut
agglutinin, a lectin specific for terminal
1-galactosyl moiety of
glycoconjugates (18), led to a significant decrease in binding to the
hemolysin (Fig. 2C). Concanavalin A (28) did not affect the
binding of asialofetuin to the hemolysin under similar conditions.
Identification of the hemolysin as a lectin-like protein specific for
glycoconjugates with terminal 1-galactosyl residues suggested that
it might be capable of binding to erythrocyte surface carbohydrate
receptors. In order to see whether this sugar-specific interaction was
a prerequisite for induction of hemolysis the potential carbohydrate
receptors for the hemolysin were blocked by the hemolysin oligomer as
well as the peanut agglutinin. Although the time course of hemolysis
was somewhat affected, as indicated by a marked shortening of the
prelytic period, the equilibrium value of percent hemolysis was not
affected (Fig. 3A) suggesting that binding of
the hemolysin to carbohydrate receptors was not essential for insertion
of the protein in the membrane bilayer and subsequent oligomerization
to a channel. Neuraminidase-treated rabbit erythrocytes that possessed
an increased surface density of glycoconjugates with terminal rather
than penultimate
1-galactosyl residue were significantly less
sensitive to the hemolysin than untreated cells in kinetics as well as
equilibrium value of percent lysis (Fig. 3B). The
sensitivity of neuraminidase-treated erythrocytes could be restored to
that of the untreated cells by selective degradation of
1-galactosyl
moiety (Fig. 3B), demonstrating that interaction of the
hemolysin with its cell surface carbohydrate ligand was functionally
unproductive. It seems, therefore, that cell surface carbohydrates
contributed to the resistance of erythrocytes to the pore-forming
action of the hemolysin.
Although hemolysis did not require binding to cell surface carbohydrate
receptors, specific glycoproteins inhibited the hemolytic activity of
V. cholerae hemolysin. Asialofetuin and asialothyroglobulin, but not ovalbumin or periodate-treated asialofetuin, caused a concentration-dependent decrease in the rate of lysis (Fig.
3C), indicating that inhibition of hemolysis involved
specific binding of the hemolysin to 1-galactosyl moiety of
glycoproteins. Since the hemolysin preincubated with asialofetuin
migrated in Sephacryl S-300 as a lytically active monomer, the
inhibition by glycoproteins did not arise from sugar-induced
oligomerization of the hemolysin to lytically inactive aggregates.
The results seemed to imply that carbohydrates inhibited a process that
did not require binding of the protein to carbohydrate receptors. That
this was indeed so was clearly indicated by the inhibition of
interaction of the hemolysin with immobilized PC by asialofetuin but
not by periodate-treated asialofetuin (Fig. 3D). This result
implied that glycoproteins with terminal 1-galactosyl moiety
inactivated the hemolysin by inducing transition of the protein to a
form incapable of interacting with lipid bilayers rather than by
competing with cell surface carbohydrate receptors.
In this communication, we showed that V. cholerae
hemolysin is a lectin-like protein that binds to glycoproteins with
terminal 1-galactosyl moiety. As observed with other pore-forming
cytolysins (10), the hemolytic activity of the protein depends on its
interaction with membrane lipids that lead to formation of oligomeric
pores by lateral collision in the bilayer plane. Specific glycoproteins inhibit V. cholerae hemolysin by inducing conversion of the
protein to a form incapable of inserting itself in lipid bilayers
rather than by competing with erythrocyte surface carbohydrate
receptors. Inhibition of hemolysis by glycoproteins is, therefore,
mechanistically different from that of well characterized
lectin-induced cellular events like agglutination of erythrocytes and
mitogenesis of lymphocytes (29). While the biological function of a
lectin is usually a sequel to its ability to interact specifically with
cell surface sugars, the sugar-binding property of V. cholerae hemolysin appears to be involved in regulating a function
not dependent on interaction of the protein with sugars. A regulatory
rather than functional role for carbohydrates might also resolve the
apparent inconsistency in the finding that, although S. aureus
-toxin was inactivated by the erythrocyte membrane band
3 glycoprotein (14), the sensitivity of erythrocytes to lysis was not
correlated with the amount of the band 3 glycoprotein in the membrane
(30). A mechanistically similar regulation of the
carbohydrate-independent interaction of concanavalin A with pure PC
vesicles by methyl
-mannoside was reported earlier (31) and was
attributed to a sugar-induced conformational change in the protein (32,
33).
The hemolysin oligomer that forms transmembrane diffusion channels in artificial lipid bilayers (8) and erythrocyte membrane (9) aggregated in water to a polydisperse mixture and adsorbed to phenyl-Sepharose matrix with an affinity that could be neutralized at an ethylene glycol concentration of 80%2 compared to 30% required for the desorption of the hemolysin monomer under similar conditions. Apparently, the presence of a lipid bilayer-spanning apolar domain in the hemolysin oligomer (10) contributed to the increased global hydrophobicity. However, the oligomer, in contrast to the monomer, failed to insert itself in the erythrocyte membrane or to bind to immobilized phospholipids (Fig. 1E). These observations would suggest that a relatively small domain of the hemolysin monomer was involved in mediating the interaction of protein with lipids.
We have not characterized biochemically or genetically the site(s) of the hemolysin molecule involved in interaction with lipids and glycoproteins. They appear to be associated with distinct domains of the protein for the following reasons. (i) Oligomerization of the monomer abolished lipid-binding activity without essentially affecting the sugar-binding property, and (ii) it is unlikely from considerations of steric complementarity as well as types of noncovalent forces involved in binding of protein to sugars and lipids for the same molecular domain of the protein to interact with these two classes of compounds. These considerations suggest that lipids bind to the active or functional site of the hemolysin and glycoproteins to the regulatory site.
Two classes of rabbit erythrocyte surface sites may likewise be
distinguished, pore formation or hemolysis sites represented by exposed
patches of phospholipid-cholesterol bilayer and agglutination sites
represented by sugar moieties of glycoproteins or glycolipids. While
the oligomeric form of the hemolysin binds to carbohydrate receptors
only leading to agglutination, the monomeric form of the hemolysin is
potentially capable of interacting with both these sites.
Glycoprotein-induced conversion of the hemolysin to a lytically
inactive form incapable of interacting with lipid bilayers suggested
that cell surface sugars might contribute to the insensitivity of
erythrocytes to the hemolysin by causing a decrease in the effective
concentration of the hemolysin available for productive interaction
with phospholipid cholesterol bilayer leading to pore formation. This
contention was corroborated by the inverse correlation between
susceptibility of rabbit erythrocytes to V. cholerae
hemolysin and the surface density of glycoconjugates specifically
recognized by the hemolysin (Fig. 3B). Similar
considerations may explain the decrease in the rate of insertion of
S. aureus -toxin in
phospholipid-GM1-ganglioside liposome with increasing glycolipid content (34). Although pore-formation process is facilitated
by a high concentration of the inserted monomer in the membrane bilayer
(Fig. 1B) (9, 10), kinetics of the overall process and
especially the length of the lag phase are regulated primarily by
membrane events that include diffusion of the monomer in the bilayer
plane and self-aggregation by lateral collision (10, 12). These events
are highly sensitive to parameters that affect the property of the
membrane bilayer matrix as a diffusional medium (12). It appears,
therefore, that the shortening of the prelytic period of hemolysis on
preincubation of erythrocytes with lectins (Fig. 3A) was due
to perturbation of the membrane bilayer induced by cross-linking of
glycoconjugate receptors by multivalent lectins (29), rather than due
to suppression of an abortive interaction of the hemolysin with
sugars.
Biological relevance of the sugar-binding property of V. cholerae hemolysin has not been considered in this communication. Recent reports suggest that the lectin-like function of a protein does
not necessarily determine its interaction with cell surface. For
example, discoidin I, a galactose-specific lectin of cellular slime
mold, binds to cells by a carbohydrate-independent mechanism similar to
those mediated by fibronectin (35). The sugar-binding function of the
protein is, however, involved in correct localization and exocytosis of
the protein in the presence of specific oligosaccharides (36). Export
of proteins by Gram-negative bacteria to the extracellular milieu
requires passage through lipid barriers of the outer and inner
membranes. Since V. cholerae hemolysin is excreted in a monomeric, lytically active form, it must have evolved a mechanism to
arrest premature self-aggregation to a lytically inactive form during
exposure to lipid bilayers of bacterial membranes. This is, perhaps,
true of pore-forming cytolysins in general. Recently, it has been found
by genetic studies that an oligosaccharide sequence of the
lipopolysaccharide mediates prevention of premature aggregation of
Escherichia coli -hemolysin (37), suggesting that
specific bacterial glycoconjugates might mimic the role of molecular
chaperones (38). Evidence from genetic studies also indicates that the membrane insertion and trimerization behavior of E. coli
porins (39) and the localization of IcsA, an outer membrane protein responsible for intracellular spread of Shigella flexneri
(40), are regulated, by an unknown mechanism, by the oligosaccharide moiety of lipopolysaccharide. The observations suggest that these outer
membrane proteins may have lectin-like functions that might serve, in
the presence of appropriate glycoconjugates, as information necessary
for targeting them to the correct destinations. Further study may
elucidate the biological significance of sugar-binding motif of a
protein that does not appear to be designed to mediate its cognitive
interaction with cell surface carbohydrates.