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

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 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.

Fig. 1 depicts the main steps and reactions in the assay of the HBL components. Standard RBC suspensions, consisting of 1.3 × 10⁷ cells ml¹ 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.

1

%RBC-B

1

%RBC-B

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 min¹, and at 22 °C the rate was only about 0.03-0.04% RBC-B min¹ (2 to 3% h¹) (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.

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

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).

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.

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.

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.

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.

Inhibition of L-induced lysis of RBC-B by antibodies specific for 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% min¹ at 1.3 nM B and 0.61% min¹ 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.

%RBC-B

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, B_i) was added prior to, simultaneously with, or following L₁₊₂. The addition of B_i 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 B_i to RBC-B in various permutations indicated that B_i 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 B_i was added at t = 0, lysis was inhibited regardless of the order of addition of the L components. But when B_i 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), B_i inhibited lysis when added at 0 s but not when added at 180 s. The kinetics of the loss of inhibitory activity of B_i 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 B_i was added until no inhibition occurred at 180 s.

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 B_i did not inhibit L₂.

%RBC-B

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 -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 -toxin and for streptolysin O provide examples of such a multistep process (22). -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 B_i was added after cells were completely primed (100% RBC-B). The inhibition of lysis did not increase if RBC-B were pretreated with B_i (Fig. 5B, inset), indicating that inhibition was not due to an enzymatic action of B_i on the cells. Inhibition decreased, however, if B_i was added after L₁ (Fig. 5, B and C), indicating that B_i 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.