Tripartite Hemolysin BL from Bacillus cereus
HEMOLYTIC ANALYSIS OF COMPONENT INTERACTIONS AND A MODEL FOR ITS CHARACTERISTIC PARADOXICAL ZONE PHENOMENON*

(Received for publication, June 25, 1996, and in revised form, October 11, 1996)

Douglas J. Beecher Dagger and Amy C. L. Wong

From the Food Research Institute, Department of Food Microbiology and Toxicology, University of Wisconsin, Madison, Wisconsin 53706

ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
FOOTNOTES
Acknowledgments
REFERENCES


ABSTRACT

Hemolysin BL (HBL) is a unique membrane-lytic toxin from Bacillus cereus composed of three distinct proteins, designated B, L1, and L2. HBL produces a paradoxical zone phenomenon in gel diffusion assays in sheep blood agar. Lysis does not begin immediately adjacent to the source of diffusion; rather, it begins several millimeters away. Cells near the source and at intersections of lysis zones remain intact longer. Here, we developed a spectrophotometric hemolysis assay system that measures the activities of the individual HBL components and used it to analyze the mechanisms of hemolysis and the paradoxical zone phenomenon. The B component was rate-limiting, and erythrocytes were slowly primed by B at an optimal concentration of about 1.3 nM to rapid lytic action by the combination of the L components (L1+2). All of the individual components bound to cells independently, and membrane-associated HBL components were neutralized by specific antibodies, suggesting that lysis was caused by formation of a membrane attack complex on the cell surface. Osmotic protection experiments indicate a colloid osmotic lysis mechanism. Concentrations of the B component above 1.3 nM caused inhibition of L1-mediated lysis, and L1 inhibited the priming reaction of B over a similar concentration range. From analyses of spectrophotometric and diffusion assays we constructed a basic model for the interactions between HBL components and for the paradoxical zone phenomenon in blood agar. In the latter, areas of slow lysis near diffusion sources are caused primarily by the accumulation of inhibitory levels of L1 reached before cells are primed by B.


INTRODUCTION

Bacillus cereus secretes hemolysin BL (HBL),1 a membrane-lytic system composed of three antigenically distinct proteins thought to contribute to diarrheal food poisoning (1) and necrotizing infections such as endophthalmitis (2). Separately, the HBL components are nontoxic, but when combined they exhibit a variety of toxic activities including hemolysis, cytotoxicity, vascular permeability, dermonecrosis, enterotoxicity, and ocular toxicity (1, 2, 3, 4). When HBL diffuses from a bacterial colony or a well in blood agar, it produces an unusual discontinuous hemolysis pattern (5), called a paradoxical zone phenomenon (6). Lysis begins away from the well or colony, followed by slow lysis nearer the source (5). A similar zone phenomenon is seen in spectrophotometric assays in which there is decreased hemolysis at HBL concentrations above a maximally effective dose (4).

HBL is the first known three-component bacterial toxin. The components are designated B, L1, and L2. Their respective sizes are 37.8, 38.5, and 43.2 kDa, and their pI values are 5.34, 5.33, and 5.33 (4), as determined electrophoretically. The genes for all three components have been cloned, and the sequences have been determined (7, 8). The amino acid sequences predicted from the nucleotide sequences are in complete agreement with published N-terminal amino acid sequences determined chemically (4). All of the proteins appear unique and exhibit no significant similarity with other known proteins. The three genes are immediately adjacent to one another, apparently in an operon (8). Methods for purification of relatively large quantities of pure HBL components were developed only recently (4); therefore, little is known about the mode of action of HBL at the molecular level.

Here we present a novel hemolysis assay system that has provided significant insight into the mechanism and interactions of HBL components. These assays are based on a fundamental difference between the activities of the B and L components and allow analysis of the activities of the individual HBL components. When sheep erythrocytes were treated with B they did not lyse, but they slowly became primed to lysis by the combination of the two L components (L1+2). The discrete proportion of cells primed at a given time could be used as a measure of the activity of B. The priming reaction, which took less than 1 h at 42 °C could be nearly stopped by reducing the temperature to 22 °C or below, creating a quasistable population of B-primed red blood cells (RBC-B'). The addition of L1+2 to RBC-B' at 22 °C caused rapid lysis of cells that was over 3000 times faster than the priming rate at that temperature. Consequently, numerous samples of the L components could be analyzed against an RBC-B' preparation with little change in the primed cells.

All three HBL components bound to erythrocytes independently of the others. Complete neutralization of membrane-bound L components by specific antibodies suggest that HBL causes hemolysis via the interaction of at least the L components in a "membrane attack complex" and not by enzymatic degradation of the membrane. Osmotic protection experiments indicated that lysis was by a colloid osmotic mechanism through transmembrane pores.

Analysis also indicated that concentrations of B above 1.3 nM caused inhibition of the lysis of RBC-B' by L1 in the presence of L2. Conversely, excess L1 inhibited the priming reaction of B. The data suggest that B and L1 each inhibit the other above a threshold concentration relative to the concentration of a membrane component. The L2 component was required for lysis, but it exhibited little or no interference with the other components.

From these observations we propose a model for the paradoxical zone phenomenon seen in gel diffusion assays. Hemolysis is slow near the HBL source because excess (and therefore inhibitory) concentrations of B and L1 accumulate before cells can be primed by B. The initial location of lysis occurs at a point at which the diffusion rate has decreased enough so that the rate of priming by B exceeds the rate at which excess components accumulate.


EXPERIMENTAL PROCEDURES

Erythrocytes, HBL Components, and Antibodies

Except for the osmotic protection experiments described below, all protein and red blood cell (RBC) preparations were in Tris-buffered saline (TBS), pH 7.4, containing 50 mM Tris-Cl and 150 mM NaCl. Pure HBL components were purified as described previously (4). The concentrations of the components in stock solutions were determined by absorbance at 280 nm through a 1-cm light path using extinction coefficients of 1.32, 1.85, and 0.83 for B, L1, and L2, respectively (1). Sheep RBCs were washed in TBS by centrifugation until the supernatant was colorless and then held on ice until use the same day. Neutralization experiments were carried out with IgG fractions of specific polyclonal antisera described previously (1). The IgG fractions were purified with a fast protein liquid chromatography system, using a Protein G-Superose column as per the instructions of the manufacturer (Pharmacia Biotech Inc.). When tested in double immunodiffusion and dot blot assays, the titers of the antibodies to components B and L2 were roughly equal and were 2-4 times higher than those of the antibodies to L1. The antibodies were highly specific for their respective antigens and exhibited no cross-reactivity to the other HBL components in double immunodiffusion assays or on Western blots.

Osmotic Protection Experiments

Osmotic protection experiments were performed essentially as described by Moayeri and Welch (9). All samples had final compositions of 2.5% RBC (v/v) and 0.75 nM concentrations of each HBL component in 25 mM Tris-HCl, pH 7.4, containing isotonic concentrations of NaCl (154 mM) or carbohydrates (270 mM each) of different sizes. Carbohydrates used were ribose, fucose, glucose, maltose, and maltotriose, with respective molecular weights of 150.1, 164.2, 180.2, 342.3, and 504.4. Assuming that the dimensions of these carbohydrates are the same as those reported (9) for carbohydrates of identical molecular weights, their molecular diameters are 0.62 nm (ribose), 0.72 nm (glucose), 0.92 nm (maltose), and 1.14 nm (maltotriose). Hemolysis was initiated by the addition of 0.1 ml of 25% RBCs to 0.9 ml of preparations containing HBL and the respective carbohydrates or controls. The samples were held at 37 °C for 75 min and then centrifuged. Hemolysis was determined by the absorbance of the supernatants at 540 nm. Percentage of hemolysis was calculated by comparing the absorbances of the samples with positive (100% lysis by saponin) and negative controls containing NaCl and no HBL (6). All samples were prepared in triplicate.

Spectrophotometric Determination of the Activities of the B and L Components of HBL

Fig. 1 depicts the main steps and reactions in the assay of the HBL components. Standard RBC suspensions, consisting of 1.3 × 107 cells ml-1 were prepared by diluting washed cells into TBS such that the optical density (OD) at 630 nm through a 1-cm path length was 0.7. Below this value, the OD630 was directly proportional to the number of intact cells present. Naive standardized RBCs were primed to the lytic action of the L components by treatment with the B component. Cells in the B-primed state are referred to as RBC-B'. The priming reaction (reaction 1 in Fig. 1) was conducted for desired times or until all RBCs were primed (100% RBC-B'), at 42 °C with 1.3 nM B, unless otherwise indicated. After priming, the cells were cooled to 22 °C, and lysis was initiated in 1-ml cuvettes by the addition of 7 µl of the L component stock (100 times the desired final concentration) to 693 µl of RBC-B'. Lysis (reaction 2 in Fig. 1) was measured continually by the decrease of the OD630 over time. The activity of the B component was measured as the proportion of cells primed (%RBC-B'; Fig. 1) at a given time. This proportion was evident as a distinct terminal plateau in the lysis curve, and proportion of primed cells was estimated as illustrated in Fig. 1B. For some samples with low hemolytic activity, the terminal plateau was approached gradually over a long assay period. In these cases the plateau was measured at the point at which the OD630 did not change for at least 12 s.


Fig. 1. Spectrophotometric determination of the activities of the B and L components of HBL. A, a representation of steps comprising the spectrophotometric hemolysis assay for individual HBL components. In reaction 1 standardized RBC (1.3 × 107 RBC ml-1) are treated with 1.3 nM B at 37 or 42 °C. During this treatment, RBC do not lyse; rather, they slowly become primed to the lytic action of the combined L components. The priming reaction is "stopped" by cooling the cells to 22 °C or below, and primed cells (RBC-B') are distributed to cuvettes. The addition of the combination of the two L components (L1+2) produces reaction 2, the rapid lysis of RBC-B'. §, the absorbance of primed RBC is generally lower than the starting OD value due to slight dilution with sample and a small amount of spontaneous lysis, but OD does not differ from that of untreated control cells. B, the graph is a time trace of the lysis of B-primed cells upon the addition of L1+2, illustrating reaction 2 from panel A. Cells underwent priming (reaction 1) at 37 °C for 60, 100, and 145 min as indicated above each curve. After cooling, lysis was initiated by the simultaneous addition of 2 nM of each of the two L components at t = 0. The decrease over time in OD630 in each curve is due to hemolysis of RBC-B' by L1+2. The difference between the starting OD value and the plateau on each curve reflects the discrete proportion of erythrocytes that are primed (%RBC-B') at a given time. The values for the maximum hemolysis velocity (Delta OD630 min-1) are shown beside each curve. Velocity varies directly with the concentration of L1+2 for a given RBC-B' preparation and is used as a measure of L activity. C, priming reaction of B at 42, 37, and 22 °C. Cells underwent reaction 1 at 42, 37, or 22 °C for the times indicated on the x axis. %RBC-B' was calculated from the terminal plateaus of lysis curves in reaction 2 of the HBL assay. *, values corresponding to the lysis curves shown in B.
[View Larger Version of this Image (32K GIF file)]


Stabilization of RBC-B' and Comparison of Lytic Samples

With the standard concentrations of B and RBC, the priming rate at 42 °C was about 4.7% RBC-B' min-1, and at 22 °C the rate was only about 0.03-0.04% RBC-B' min-1 (2 to 3% h-1) (Fig. 1C), and priming was not detectable at 0 °C. Practically, this means that the priming reaction can be nearly stopped at 22 °C. Since the L components cause rapid lysis of RBC-B' at 22 °C, multiple samples tested sequentially on temperature-stabilized RBC-B' can be compared under the assumption that the cells are identically primed.

Gel Diffusion Assay

The gel diffusion assay was performed, and the blood agar gel was prepared as described previously (5). Each well (3-mm diameter) received 4 pmol (about 150 ng) of the appropriate HBL components. The gel was photographed after 8 h at 24 °C.


RESULTS

Osmotic Protection from Lysis by HBL

It is generally thought that pore-forming hemolysins cause lysis by a colloid osmotic mechanism. When pores form in the membrane, external solutes that are smaller than the pores enter the cell, creating an osmotic gradient. This in turn produces a net flow of water into the cell, which swells and lyses from excess osmotic pressure. Osmotic protection experiments are the standard methods for detecting colloid osmotic lysis. Theoretically, if pores of defined size are produced, and the diameters of the external solutes are larger than the diameters of the pores, those solutes cannot enter the cell. This prevents the production of an osmotic gradient, thereby protecting cells from lysis (9).

Hemolysis due to HBL decreased in the presence of isotonic concentrations of carbohydrates of increasing size as shown in Fig. 2. These results indicate that HBL lyses cells by a colloid-osmotic mechanism and suggest that, under these conditions, functional pores of <= 1.2 nm in diameter were formed. This pore size should only be considered a rough estimate, because calculated values vary greatly with assay conditions (9).


Fig. 2. Osmotic protection of sheep erythrocytes from HBL by carbohydrates of increasing size. Sheep erythrocytes were suspended in isosmotic concentrations of NaCl or carbohydrates of increasing sizes as indicated and treated with 0.75 nM of each HBL component. The percentage of cells lysed was determined after 75 min at 37 °C. Additional experimental details are provided under "Experimental Procedures."
[View Larger Version of this Image (20K GIF file)]


The Slow Priming Reaction of the B Component Versus the Rapid Lytic Reaction of L1+2

RBC did not lyse when treated with isolated B, but became primed to the lytic action of the combined L components. Fig. 1B depicts three lysis curves generated by the simultaneous addition of both L components (L1+2) to RBC that had been treated at 37 °C for the times indicated above each curve. The difference between the starting OD and that of the terminal plateau following the lysis phase in each curve, measures the discrete proportion of cells primed by B to lytic action by L1+2 (%RBC-B'). This value at 42, 37, and 22 °C (by 1.3 nM B) is plotted versus time in Fig. 1C. Note that the time scale is in hours for the priming reaction (Fig. 1C) and in seconds for the lysis reaction of primed cells by L1+2 (Fig. 1B). The priming reaction of B at 47, 37, and 22 °C was, respectively, 1, 2, and 3 orders of magnitude lower than the lysis velocity of 100% RBC-B' caused by comparable concentrations of L1+2 at 22 °C. This slow priming rate of B corresponds to the slow hemolysis reaction that occurs in the presence all three HBL components (4), which required more than 1.5 h for complete lysis at 37 °C. This shows that the priming reaction of the B component is the rate-limiting step of HBL hemolysis.

Independent Binding of L Components to RBC and Neutralization of Membrane-bound L Components

Naive RBC were treated at 22 °C for 20 min with 5, 10, or 50 nM L1 or 5 nM L2. The cells were washed twice by centrifugation with cold TBS such that the free L concentration was reduced by a factor of at least 5 × 105. Washed cells were then diluted to an OD630 of 0.7 with TBS. The washed L-treated RBC and naive RBC, were then primed at 42 °C with B for 1 h, as described above. Lysis was initiated by the addition of the appropriate L component (5 nM) as indicated in Fig. 3.


Fig. 3. Independent binding to RBC of the L1 and L2 components. Standardized naive RBC were pretreated with either L1 (top) or L2 (bottom) at the concentrations indicated beside each curve. The L-treated cells were washed and then primed with B for 1 h. Lysis of primed cells was initiated at t = 0 by the addition of 5 nM L2 to the L1-treated cells (top) or 5 nM L1 to the L2-treated cells (bottom). The control samples were naive RBCs that were not pretreated with L1 or L2 but were primed with B for 1 h. Each L component was added to a separate control sample at t = 0, and then lysis was initiated after 300 s by the addition of the complementary L component.
[View Larger Version of this Image (31K GIF file)]


Cells that were pretreated with either L component and then primed with B lysed upon the addition of the complementary L component. Control RBC-B' without pretreatment with an L component did not lyse upon the addition of either L component, but lysed rapidly upon the subsequent addition of the complementary L component. Pretreatment with L1 was less effective than with L2, possibly due to a lower affinity of L1 for RBC.

Neutralization of Membrane-associated HBL Components

In the experiment depicted in Fig. 3, specific neutralizing antibodies to L1 and L2 were added at t = 0 to RBC-B'-L1 (5 nM L1) and RBC-B'-L2 (5 nM L2), respectively. Each antibody completely inhibited lysis by the complementary L component for at least 1 h.

When specific antibodies to the B component were mixed with B prior to the addition to naive RBC, the priming reaction was completely inhibited. However, when those antibodies were added after the priming reaction (i.e. added to 100% RBC-B') there was only partial inhibition of lysis of by L1+2. The simultaneous addition of L1+2 and the B-specific antibodies to RBC-B' resulted in about 30% inhibition of lysis velocity (see point at t = 0 in Fig. 4). Pretreatment of RBC-B' with the antibodies increased inhibition of lysis in a time-dependent manner (Fig. 4). Inhibition also increased with increasing antibody dose (data not shown). As a control, the L1+2 preparation was treated with the anti-B antibody prior to the addition to RBC-B'. There was no significant effect on inhibition of lysis although the concentration of antibody exposed to L1+2 was 100 times that exposed to RBC-B'. This indicates that the inhibitory effect of the antibody was due to interaction with B on the surface of RBC-B' and not due to an interaction with the L1+2 preparation. These results demonstrate a detectable difference between B in the unprimed versus the primed state.


Fig. 4. Inhibition of L-induced lysis of RBC-B' by antibodies specific for component B. Hemolysis of 100% RBC-B' was initiated by the addition of L1+2 in the presence or absence of specific antibodies to the B component of HBL. The antibodies were preincubated with either the RBC-B' cells or with the L1+2 preparation to distinguish whether inhibition was specific for B or due to cross-reaction with L1+2. The x axis indicates the time that the RBC-B' or L1+2 was preincubated with the antibodies. At t = 0 on the x axis the antibodies, L1+2, and RBC-B' were all mixed simultaneously. Circles, RBC-B' were incubated with antibodies before the addition of L1+2 for the times indicated on the x axis. Squares, L1+2 was incubated with antibodies before the addition to RBC-B' for the times indicated on the x axis. The inhibitory activity of the antibodies is expressed on the y axis as the percentage by which the maximum lysis velocity was decreased compared with simultaneously run control samples containing no antibody (% inhibition).
[View Larger Version of this Image (21K GIF file)]


Inhibition of Hemolysis by Excess Component B

We previously found that, when equal concentrations of all three HBL components were simultaneously varied, hemolysis exhibited a paradoxical zone phenomenon (4). Hemolysis was higher within a certain low concentration range than at higher concentrations. The effect of the concentration of the B component on the priming rate of B and the lysis velocity of L1+2 is illustrated in Fig. 5A. The extent of priming and the L-induced lysis velocity were both maximal around 1.3 nM B. When the time course of priming at 37 °C was followed, the maximum priming rate at the inflection point of the sigmoidal priming curve was 1.15% min-1 at 1.3 nM B and 0.61% min-1 at 4.5 nM B, supporting the idea that there is an antagonistic effect of excess B on itself (not shown). Above 4.5 nM B, the priming rate increased slightly, but lysis velocity was further inhibited.


Fig. 5. Inhibition of L1-induced lysis by excess component B. A, effect of the concentration of B on the priming reaction of B and the lysis velocity of L1+2. Erythrocytes were primed at 37 °C for 60 min with the B concentrations indicated on the x axis. Lysis was initiated by the addition of 5 nM L1+2. The percentage of primed cells (%RBC-B') and lysis velocity were measured as described under "Experimental Procedures." B, inhibition of hemolysis of RBC-B' by Bi. Bi (5 nM) was added to primed erythrocytes prior to, simultaneously with, or following L1+2 (2 nM). The labels for each curve indicate the time (min) at which Bi was added relative to t = 0 on the x axis. Inset, plot of percentage of inhibition of velocity versus the time of the addition of Bi. C, inhibition of L1 and not L2 by Bi. Lysis curves A and B were generated by the addition of L components to 100% RBC-B' at 0 s and 180 s (closed arrow) in the order indicated beside each curve. When excess B was added at 0 s, inhibition of both A and B occurred. However, when Bi was added at 180 s, inhibition occurred for B and not for A. bullet , the percentage of inhibition of the lysis velocity of curve A (L1 at 0 s; L2 at 180 s) that occurred when Bi was added between 0 and 180 s. For each symbol, the abscissa corresponds to the time at which Bi was added, and the ordinate corresponds to the percentage of inhibition for that sample. Open arrow, the point at which RBC-B' were completely lysed by the addition at t = 0 s of the combination of both L components (2 nM).
[View Larger Version of this Image (22K GIF file)]


The inhibitory effect of excess B on L1+2 was observed even when B was added to previously primed erythrocytes (Fig. 5B). Erythrocytes were completely primed (100% RBC-B') and stabilized, and then excess B (an inhibitory concentration, Bi) was added prior to, simultaneously with, or following L1+2. The addition of Bi simultaneously with L1+2 resulted in nearly 60% inhibition of the lysis velocity. Inhibition did not increase if B was preincubated with RBC-B', and inhibition decreased drastically when B was added after L1+2. This indicates that under these conditions, B did not alter the cell surface, but it did interfere either with the binding or an intermediate step of one or both L components, and B did not interfere with lysis once binding or an intermediate step occurred.

The sequential addition of L1, L2, and Bi to RBC-B' in various permutations indicated that Bi inhibits L1 and not L2 (Fig. 5C). When either L component was added to RBC-B', the cells did not lyse until the second L was added (Fig. 5C, curves A and B). When Bi was added at t = 0, lysis was inhibited regardless of the order of addition of the L components. But when Bi was added with an L component at t = 180 s, inhibition only occurred when L1 was added at that time (curve B), suggesting that B interferes with L1 and not L2. With respect to curve A (L1 at 0 s, L2 at 180 s), Bi inhibited lysis when added at 0 s but not when added at 180 s. The kinetics of the loss of inhibitory activity of Bi is graphically illustrated by the closed circle scatter plot in Fig. 5C. When L1 was added first (at t = 0), less inhibition occurred the later Bi was added until no inhibition occurred at 180 s.

Inhibition of B by Excess L1

RBC were primed by B in the presence of increasing concentrations of either L1 or L2, and the extent of priming was measured by the addition of the complementary L component (Fig. 6A). L1 drastically inhibited the priming reaction above ~1.5 nM. The inhibition dose response was very similar to the inhibition of L1 by B in Fig. 5A. The priming reaction of B could be inhibited by adding an excess concentration of L1 at any time before 100% RBC-B' occurred (Fig. 6A, inset). The L2 component did not significantly inhibit the priming reaction in the concentration range tested, just as Bi did not inhibit L2.


Fig. 6. Inhibition of component B by excess L1. A, effect of the concentrations of L1 and L2 on the priming reaction of B. RBCs were primed at 42 °C with 1.3 mM B in the presence of either L1 or L2 at the concentrations indicated on the x axis. The extent of priming was measured after 1 h by the addition of the complementary L component (5 nM) to lyse susceptible cells as described under "Experimental Procedures." Inset, inhibition of the B component by the addition of L1 (5 nM) at any time during priming. RBCs were primed with 1.3 nM B at 37 °C. Excess L1 was either not added (circles) or added at 5 (squares) or 15 min (triangles) as indicated by the arrows. The extent of priming (%RBC-B') was determined at the times indicated on the x axis by the addition of L2 to RBC-B'. B, inhibition of hemolysis in a gel diffusion assay by excess L1. Lack of lysis was observed within the intersections of HBL hemolysis zones. The diffusion assay was performed, and the blood agar gel was prepared as described previously (10). Each well (3-mm diameter) received 4 pmol (~150 ng) of the respective HBL components. Wells labeled H received all three HBL components. The remaining wells received components B, L1, and L2 as indicated. The gel was photographed after 8 h at 24 °C.
[View Larger Version of this Image (46K GIF file)]


An implication of the above observations is that, in blood agar, the inhibition of lysis near the HBL diffusion source and at the intersections of lysis zones (10) is a function of the mutual inhibitory activity between B and L1. To test this idea, the separated components were added to wells adjacent to wells containing all three components (Fig. 6B). Consistent with the above observations, the L1 component inhibited lysis in the adjacent zone, and the L2 component had no apparent effect on the adjacent pattern. Lysis was enhanced adjacent to the B component. This superficially contradicts the observation that excess B inhibits lysis by L1, but it is consistent with the overall kinetics of the system, as discussed below.


DISCUSSION

Among the greater than 300 described bacterial toxins, there are several that require two different, individually soluble, proteins for toxic activity (11) but none that require three components. These binary toxins include staphylococcal gamma -hemolysin and the related leukocidins (12, 13); anthrax toxin (14); Clostridium botulinum C2 and the related toxins of Clostridium perfringens, Clostridium spiroforme, and Clostridium difficile (15); and Bacillus sphaericus binary toxin (16). For all of these (except possibly the B. sphaericus toxin), the components exhibit a compulsory cell-binding order. Also the protective antigen of anthrax toxin, which acts as a docking protein, must be cleaved by a cellular protease before the complementary proteins will bind. The observations presented here allowed us to address related questions about the mechanism of HBL and helped answer three basic questions. Does HBL cause lysis enzymatically or by formation of a membrane attack complex? Do the HBL components act randomly or in a compulsory order? What is the basis for the paradoxical zone phenomenon exhibited by HBL?

Hemolysins are thought to lyse cells by two general mechanisms (17). First, the cell surface may be enzymatically altered to an extent that causes membrane degeneration. This usually involves phospholipases, and there are numerous examples of hemolysis that require two enzymes acting in a compulsory order (18). In some systems a phospholipase acts on the cell, which is subsequently lysed by a nonphospholipase enzyme or nonenzymatic protein. In the second hemolysis mechanism, lysin molecules insert into the membrane and form functional transmembrane pores (19), or "leaky patches" (20). Such lesions are generally composed of oligomers of one protein or multiple proteins. Complement is an example of a lytic system that produces lesions composed of more than one different protein. Complement lesions are called membrane attack complexes (MACs) to account for their multicomponent composition (21). Hemolysins may also exhibit detergent-like activity and "solubilize" membranes, but this mechanism may not occur for hemolytic proteins (17).

The experiment shown in Fig. 3 demonstrates that both L components bound to erythrocyte membranes independently of any other HBL component. After priming of L1- or L2-treated and washed cells by B, the addition of antibodies specific to either membrane-bound L component completely neutralized hemolysis by the complementary L component. This indicates that the L components remained associated with the membranes after washing and that each needed to be present for a competent lytic lesion to be formed. It also strongly suggests that a competent lesion is only formed when L1 and L2 physically interact on the cell surface, which, by definition, describes a MAC. If the L components were simple enzymes that caused lysis by independently damaging membranes, then antibodies to an L component would not reverse the membrane damage and protect cells from lysis by the other HBL components. It appears, therefore, that HBL ultimately causes lysis by forming a lesion analogous to a complement MAC that has a minimum structure consisting of an L1-L2 heterodimer. Osmotic protection of erythrocytes treated with HBL (Fig. 2) suggests that the putative MAC results in formation of transmembrane pores but does not indicate whether the pores are protein-lined or simply perturbations of the phospholipid bilayer surrounding the HBL components.

Antibodies specific for the B component also interfered with the ability of the L components to lyse B-primed cells (Fig. 4), suggesting that L components interact with B to form an integral three-component heteromeric membrane lesion. However, although the anti-B antibodies completely neutralized the autonomous priming reaction of B, they could only partially inhibit lysis once cells were primed. If HBL lysis does involve a three-component MAC, there are several possibilities as to what might occur during the priming reaction that decreases the ability of the anti-B antibodies to neutralize lysis. During priming, the B component may insert into the membrane, possibly forming oligomers. This would allow antibodies to bind only to exposed portions of B and not to portions sequestered within the membrane. Any conformational changes experienced by B might also decrease the avidity of antibodies prepared to its native conformation. It is also possible that B is altered by a membrane component, such as occurs in the proteolytic processing of the protective antigen component of anthrax toxin (14). Antibodies prepared to native B may bind poorly to the processed version.

The inability of anti-B antibodies to completely neutralize lysis of RBC-B' by L1+2 might also be diagnostic of an enzymatic mechanism for B. A situation can be envisioned in which B binds to a membrane and converts a substrate to a product that is subsequently recognized by the L components, leading to lysis. If the product was to accumulate in a pool in the immediate vicinity of the membrane-bound B, partial inhibition might occur if B-associated antibody sterically prevented access of L components to the product. Such a mechanism seems rather unlikely, but it cannot be ruled out with present observations.

The next question addressed by our data is whether the HBL components act in an ordered or random manner. It is clear from Figs. 1 and 3 that all three HBL components bind to erythrocytes independently (i.e. randomly). However, as pointed out by Ponder in 1948 (6), hemolysis is a relatively complex multistep process. Pore assembly models for staphylococcal alpha -toxin and for streptolysin O provide examples of such a multistep process (22). alpha -Toxin monomers bind to membranes, diffuse laterally into a prepore complex on the membrane surface, and subsequently insert and form pores. SLO also binds as monomers and then undergoes a rate-limiting nucleation process involving formation of dimers on the membrane surface, followed by dimer insertion, and then rapid accumulation of monomers around the dimer nucleus to form the functional lesion. The complexity of these processes demonstrates that simple association of toxin components with erythrocytes does not provide a picture of the overall hemolysis mechanism.

Our evidence suggests that HBL components form pores after they form a complex. The L1 and L2 components rapidly contribute to the complex in a manner that does not appear to depend on the sequence of their addition. However, the B component clearly must undergo an autonomous, rate-limiting reaction before a functional lesion can be formed with the L components. Therefore, although all of the components bind independently to the cell, final lesion formation appears to follow a specific order with respect to the B and L components. It is possible that the appearance of an ordered reaction is an artifact of the kinetics of the individual components. All of the components might undergo analogous independent priming reactions, with the reaction of B being extremely slow relative to the reactions of the L components. However, our data suggest that the priming process followed by B is fundamentally different from those followed by the L components, because antibodies could completely protect cells when reacted with either membrane-associated L component, but not when reacted with B-primed cells.

The cause of the paradoxical zone phenomenon of HBL is the third major mechanistic question addressed by these data. Figs. 5 and 6 indicate that B and L1 are mutually inhibitory above a threshold concentration and that this antagonism is responsible for the paradoxical hemolysis behavior seen in gel diffusion and suspension assays.

In a dose-response curve for the priming reaction of B (Fig. 5A), the lysis velocity (reaction 2 in Fig. 1A) of L1+2 decreased with increasing concentrations of B even when most cells were primed. Fig. 5B shows that lysis velocity could be inhibited even if Bi was added after cells were completely primed (100% RBC-B'). The inhibition of lysis did not increase if RBC-B' were pretreated with Bi (Fig. 5B, inset), indicating that inhibition was not due to an enzymatic action of Bi on the cells. Inhibition decreased, however, if Bi was added after L1 (Fig. 5, B and C), indicating that Bi interferes with the incorporation of L1 into the lesion and not with the function of a preformed lesion.

The kinetics of the inhibition of L1 by B in Fig. 5C is noteworthy. Inhibition decreased linearly with time (r = 0.976). This suggests a first order process for the incorporation of L1 into lesions, which is in distinct contrast to the complex shape of hemolysis curves. This type of measurement was possible only because of the special situation in which B does not appear to interfere with lysis once L1 has been incorporated into a lesion.

Fig. 6 shows that, above a threshold concentration, L1 also inhibited the priming reaction of B. The dose-response of L1-mediated inhibition of the priming reaction was nearly identical to that for inhibition of L1 by excess B (Fig. 5A). In addition, excess L1 inhibited hemolysis caused by HBL placed in adjacent wells in blood agar (Fig. 6B).

The hemolysis data presented here provide an explanation for the paradoxical zone phenomenon of HBL and the lysis patterns seen in Fig. 6B. HBL hemolysis in blood agar is discontinuous and begins away from a well containing the components. However, diffusing molecules create a continuous concentration gradient, and the components must therefore reach a concentration that promotes hemolysis at every point between the well edge and the zone in which lysis begins; this is the paradox, and we propose the following model to explain it. A cell will not lyse unless it has been primed by B, and maximal priming occurs only at permissive concentrations of B and L1. When membrane-bound B or L1 exceeds a critical ratio versus a membrane constituent, the other component is inhibited from forming a functional lesion (or perhaps produces an inefficient lesion). Priming by B is slow and, since L1 can inhibit B at any time during the reaction (Fig. 6A, insert), priming is inhibited near a well when L1 diffuses rapidly enough to accumulate to inhibitory levels before cells are primed. Hemolysis only begins in a zone in which the B concentration is high enough to prime cells, but the diffusion rate of L1 is too low for it to reach an inhibitory concentration before the cells are primed. Inhibition at the intersections of zones is due to accumulation of excess L1 before cells are primed. It is unlikely that inhibition is caused by an excess ratio of B to L1 or vice versa, because a paradoxical zone phenomenon occurs in suspension assays simply by increasing the concentration of all three components at a constant ratio of each to the others (4).

The influence on hemolysis patterns of the inhibitory effect of L1 is seen in Fig. 6B. Just as lysis is inhibited at the intersections of HBL hemolysis zones, lysis was also inhibited at the intersection of an HBL hemolysis zone and an L1 diffusion zone. As expected from Fig. 6A, L2 had no appreciable effect on an adjacent hemolysis pattern. The enhanced lysis between wells containing either HBL or B only seems contradictory in light of the observation that excess B inhibits lysis by L1, but it is consistent with the mechanism of HBL. The B component primes cells independently of the other components (Fig. 1, B and C), and once primed, cells are rapidly lysed by any concentration of L1+2. The lysis within the B diffusion zone occurred simply because cells were primed by B in the absence of inhibitory concentrations of L1. The priming reaction of B is over 1000 times slower than the lytic action of L1 at room temperature. Therefore, even in the presence of excess B, the lysis rate of RBC-B' will not be noticeably slower than the rate at which cells are primed by B. On the other hand, any inhibition by L1 of the priming reaction of B should be profoundly evident. This seems to be born out in the lysis patterns in Fig. 6B.

Our hemolysis data provide some insights into the mechanism of HBL, which appears to involve novel modes of interactions between proteins and membranes. The three proteins that comprise HBL are collectively called a toxin because of an early simplifying assumption that they represent a single functional unit to the bacterium that produces them. This work supports that assumption. New findings show that the genes for all three components are arranged immediately adjacent to one another, probably in an operon (8), lending additional weight to the idea that the three proteins comprise a functional unit. HBL is unlike any previously described cytolytic toxin, from its complex composition to the unique hemolytic phenomena that it causes. Further study of this toxin promises to add a new perspective to perceptions about the mechanisms and utility of cytolytic toxins.


FOOTNOTES

*   This work was supported by the National Research Initiative Competitive Grants Program of the U.S. Department of Agriculture Grant 93-37201-9194 and by the College of Agriculture and Life Sciences, University of Wisconsin. The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Dagger    To whom correspondence should be addressed: Food Research Inst., Dept. of Food Microbiology and Toxicology, University of Wisconsin, 1925 Willow Dr., Madison, WI 53706. Tel.: 608-263-1177; Fax: 608-263-1114; E-mail: dbeecher{at}facstaff.wisc.edu.
1    The abbreviations used are: HBL, hemolysin BL; L1+2, a combination of HBL components L1 and L2; RBC, red blood cells; RBC-B', erythrocyte(s) primed by the B component of HBL to the lytic action of the combination of the two L components; Bi, inhibitory concentration of B; TBS, Tris-buffered saline; MAC, membrane attack complex.

Acknowledgments

We thank W. W. Cleland for helpful discussion and Jean L. Schoeni for purification of IgG fractions of antisera by Protein G chromatography.


REFERENCES

  1. Beecher, D. J., Schoeni, J. L., and Wong, A. C. L. (1995) Infect. Immun. 63, 4423-4428 [Abstract]
  2. Beecher, D. J., Pulido, J. S., Barney, N. P., and Wong, A. C. L. (1995) Infect. Immun. 63, 632-639 [Abstract]
  3. Beecher, D. J., and Macmillan, J. D. (1991) Infect. Immun. 59, 1778-1784 [Medline] [Order article via Infotrieve]
  4. Beecher, D. J., and Wong, A. C. L. (1994) Infect. Immun. 62, 980-986 [Abstract]
  5. Beecher, D. J., and Macmillan, J. D. (1990) Infect. Immun. 58, 2220-2227 [Medline] [Order article via Infotrieve]
  6. Ponder, E. (1948) Hemolysis and Related Phenomena, pp. 168-257, Grune and Stratton, New York
  7. Heinrichs, J. H., Beecher, D. J., Macmillan, J. D., and Zilinskas, B. A. (1993) J. Bacteriol. 175, 6760-6766 [Abstract]
  8. Ryan, P. A., Macmillan, J. D., and Zilinskas, B. A. (1996) 96th General Meeting of the American Society for Microbiology, New Orleans, May 19-23, p. 384, American Society for Microbiology, Washington, D. C.
  9. Moayeri, M., and Welch, R. A. (1994) Infect. Immun. 62, 4124-4134 [Abstract]
  10. Beecher, D. J., and Wong, A. C. L. (1994) Appl. Environ. Microbiol. 60, 1646-1651 [Abstract]
  11. Alouf, J. E. (1994) in Bacterial Protein Toxins, Zentralblatt für Bakteriologie (Freer, J., Aitken, R., Alouf, J. E., Boulnois, G., Falmagne, P., Fehrenbach, F., Montecucco, C., Piemont, Y., Rappuoli, R., Wadström, T., and Witholt, B., eds) Suppl. 24, pp. 3-28, Gustav Fischer, Stuttgart
  12. Prévost, G., Supersac, G., Colin, D. A., Couppié, P., Sire, S., Hensler, T., Petiau, P., Meunier, O., Cribier, B., König, W., and Piémont, Y. (1994) in Bacterial Protein Toxins, Zentralblatt für Bakteriologie (Freer, J., Aitken, R., Alouf, J. E., Boulnois, G., Falmagne, P., Fehrenbach, F., Montecucco, C., Piemont, Y., Rappuoli, R., Wadström, T., and Witholt, B., eds) Suppl. 24, pp. 284-293, Gustav Fischer, Stuttgart
  13. Prévost, G., Cribier, B., Couppié, P., Petiau, P., Supersac, G., FinckBarbançon, V., Monteil, H., and Piémont, Y. (1995) Infect. Immun. 63, 4121-4129 [Abstract]
  14. Leppla, S. H. (1991) in Sourcebook of Bacterial Protein Toxins (Alouf, J. E., and Freer, J. H., eds), pp. 277-302, Academic Press, London
  15. Rappuoli, R., and Pizza, M. (1991) in Sourcebook of Bacterial Protein Toxins (Alouf, J. E., and Freer, J. H., eds), pp. 1-21, Academic Press, London
  16. Baumann, P., Clark, M. A., Baumann, L., and Broadwell, A. W. (1991) Microbiol. Rev. 55, 425-436
  17. Linder, R., and Bernheimer, A. W. (1991) in Sourcebook of Bacterial Protein Toxins (Alouf, J. E., and Freer, J. H., eds), pp. 101-115, Academic Press, London
  18. Fehrenbach, F. J., and Jürgens, D. (1991) in Sourcebook of Bacterial Protein Toxins (Alouf, J. E., and Freer, J. H., eds), pp. 187-213, Academic Press, London
  19. Bhakdi, S., and Tranum-Jensen, J. (1991) Immunol. Today 12, 318-320 [Medline] [Order article via Infotrieve]
  20. Esser, A. F. (1991) Immunol. Today 12, 316-318 [Medline] [Order article via Infotrieve]
  21. Esser, A. F. (1994) Toxicology 87, 229-247 [Medline] [Order article via Infotrieve]
  22. Bhakdi, S., Bayley, H., Valeva, A., Walev, I., Walker, B., Weller, U., Kehoe, M., and Palmer, M. (1996) Arch. Microbiol. 165, 71-79 [CrossRef]

©1997 by The American Society for Biochemistry and Molecular Biology, Inc.