Tripartite haemolysin BL: isolation and characterization of two distinct homologous sets of components from a single Bacillus cereus isolate

Douglas J. Beecher1 and Amy C. L. Wong1

Food Research Institute, Department of Food Microbiology and Toxicology, University of Wisconsin-Madison, 1925 Willow Drive, Madison WI 53706, USA1

Author for correspondence: Douglas J. Beecher. Tel: +1 608 263 6939 Fax: +1 608 263 1114. e-mail: dbeecher{at}facstaff.wisc.edu


   ABSTRACT
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Haemolysin BL (HBL), a three-component enterotoxic/necrotizing/vascular permeability toxin, is a likely virulence factor of Bacillus cereus diarrhoeal food poisoning and necrotic infections. This paper describes the isolation of two distinct homologous sets of all three HBL components from a single B. cereus isolate, MGBC 145. The proteins of one set (designated HBL, consisting of B, L1 and L2), were about 87–100% identical in N-terminal amino acid sequences to their respective prototype components from strain F837/76, and the proteins of the homologous set (HBLa, consisting of Ba, L1a and L2a) were all about 62–65% identical. Only the latter homologues differed immunochemically and physicochemically from the prototypes. HBL and HBLa exhibited similar haemolytic and vascular permeability potencies, and the homologues could be interchanged freely. There were no notable differences in activity between the L component homologues. However, components B and Ba were significantly different. Both were secreted as monomers, but unlike B, Ba was isolated as a relatively inactive complex that could be reactivated with urea. When Ba was substituted for B in gel-diffusion assays the distinct discontinuous haemolysis pattern typical of the presence of B did not occur. In suspension assays, excess B inhibited the haemolysis of B-primed cells by L1 (as previously described), but not that of Ba-primed cells. Excess Ba had the opposite effect and enhanced lysis of Ba-primed cells, but not that of B-primed cells. These differences reveal details about how the toxin components interact on target cell membranes. The authors’ observations indicate that HBL represents a new family of multicomponent toxins that was generated by a process of gene and operon duplication that occurred either intracellularly or by horizontal transfer, and raise the possibility of the existence of other related toxins in the genetically diverse B. cereus taxonomic group.

Keywords: Bacillus cereus, hemolysin BL, enterotoxin, gene duplication, discontinuous hemolysis

Abbreviations: BCA, bicinchoninic acid; HA, hydroxyapatite; HBL, haemolysin BL; TBS, Tris-buffered saline


   INTRODUCTION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Bacillus cereus is a ubiquitous soil-borne bacterium that, under appropriate circumstances, causes diarrhoeal and emetic food poisoning and a variety of typically necrotic non-gastrointestinal infections (Drobniewski, 1993 ; Kramer & Gilbert, 1989 ; Turnbull, 1986 ). Among the many potential virulence factors of B. cereus is haemolysin BL (HBL), a unique and potent three-component pore-forming toxin (Beecher & Macmillan, 1991 ; Beecher & Wong, 1994c ; Beecher & Wong, 1997 ). The toxic activities so far identified for HBL include haemolysis, vascular permeability and necrosis in rabbit skin (Beecher & Macmillan, 1991 ; Beecher & Wong, 1994c ), fluid accumulation in rabbit ileal loops (Beecher et al., 1995b ), toxicity to a number of transformed cell lines (Beecher, 1990 ), in vitro degradation of explanted rabbit retinal tissue, and in vivo ocular necrosis and inflammation in rabbits (Beecher et al., 1995a ). HBL is secreted by 45% of over 200 B. cereus, B. thuringiensis and B. mycoides isolates tested in our laboratory (Beecher & Wong, 1994b ; Schoeni & Wong, 1999 ). A distinctive feature of HBL is an unusual discontinuous haemolysis pattern produced in blood agar (Beecher & Macmillan, 1991 ; Beecher & Wong, 1997 ). When HBL diffuses from a bacterial colony or a well in blood agar, lysis begins away from the well or colony, followed by slow lysis nearer the source.

In B. cereus strain F837/76, the three HBL components are encoded in an operon by tandemly arranged genes in the order hblC, hblD and hblA, which respectively encode components designated L2, L1 and B (Heinrichs et al., 1993 ; Ryan et al., 1997 ). Immediately downstream from hblA is an open reading frame, hblB, which is about 85% identical to hblA in the first 158 predicted amino acids. The hblB gene product has not yet been isolated.

Many isolates in our Bacillus collection produce more than one antigen reactive to antibodies against individual HBL components (Schoeni & Wong, 1999 ). For example, on Western blots of denaturing electrophoresis gels B. cereus MGBC 145 produces two bands reactive with antibodies to HBL components B and L1, and one band reactive with antibodies to component L2. This strain is used in studies of experimental endophthalmitis in our laboratory and in others (Beecher et al., 1995a ; Callegan et al., 1999a , b ). Because the production of multiple toxin homologues might confound efforts to characterize the molecular basis of B. cereus virulence in intraocular infections, we decided to characterize the antigens responsible for producing multiple bands on Western blots. We isolated and characterized two distinct homologous variants of each HBL component from a single B. cereus strain, demonstrating that the variant antigens are distinct gene products. The activities of the variants were essentially the same as the prototype components, the homologues were interchangeable, and the requirement for three components to produce activity was maintained. A homologue of one component possessed several important differences in the details of its activity that elucidate some of its interactions with other components on erythrocyte membranes.


   METHODS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Bacterial isolates.
B. cereus strain MGBC 145 was used as the source of HBL homologues. This strain was isolated from the infected eye of a patient with post-traumatic endophthalmitis (Beecher et al., 1995a ; Callegan et al., 1999a , b ). B. cereus strain F837/76, an isolate from an infected prostate, was used as the source of prototype HBL components in the present study. This isolate was originally chosen as the source of HBL because it is exceptionally toxigenic (Beecher & Macmillan, 1990 ).

Production of crude haemolysin BL.
The B. cereus isolates were grown at 32 °C for 5 h in 3 litres of casein/glucose/yeast extract broth with shaking at 200 r.p.m. The culture supernatants were each concentrated by ammonium sulfate precipitation to 50 ml. The concentrates were used as sources of crude HBL. The culture medium, growth conditions and concentration procedure are described in detail elsewhere (Beecher & Wong, 1994c ).

Chromatography.
Details (columns, flow rates, elution gradients, etc.) of purification steps involving anion-exchange (Whatman DE-52 and Pharmacia Mono-Q) and hydroxyapatite (HA) chromatography were as previously described (Beecher & Wong, 1994c ) with the following exceptions. The pH 5·9 Bistris anion-exchange equilibration buffer was 20 mM instead of 25 mM, and the HA was Macro-Prep ceramic HA (20 µm beads, Bio-Rad) instead of BioGel HTP. Salt concentrations of fractions were estimated assuming that the gradients were linear and began eluting one column volume after they were started. Column volumes were estimated empirically by changes in the apparent absorbance of eluate after abrupt changes in salt concentration.

Three additional chromatographic steps were used. First, rather than dialysis, buffer exchange of the concentrated culture supernatant was by gel-filtration chromatography on a 2·6x78 cm column of medium-grade Sephadex G-25 (Pharmacia) equilibrated with 20 mM Bistris/HCl, pH 5·9, and eluted at a flow rate of 7·5 ml min-1. Second, for final isolation of the L1a component (see below for nomenclature), fractions from the pH 5·9 Mono-Q column were dialysed against 25 mM Tris/HCl, pH 7·4, and applied to the Mono-Q column equilibrated with the same buffer. Proteins were eluted with a linear gradient of NaCl from 0 to 0·5 M over 50 ml at a flow rate of 1 ml min-1. Third, for isolation of the L2a the nonbound DE-52 fraction containing L2a was applied to a 1 ml (0·5x5 cm) HA column (as compared to the 1·6x8 cm column usually used) equilibrated with 10 mM NaCl and eluted at 1 ml min-1 with a linear gradient of sodium phosphate from 0 to 0·3 M over 30 ml and collected into 1 ml fractions.

Protein determinations.
Concentrations of purified proteins were routinely determined by measuring A280 using previously estimated absorption coefficients (1 mg ml-1) of 1·3, 1·8 and 0·8 for B, L1 and L2 (from strain F837/76) respectively. Initial experiments suggested that the absorption coefficients of B and Ba may differ significantly. Thus, when homologue doses were to be compared, concentrations were measured using a bicinchoninic acid assay (Pierce Chemical Co.) in which components from strain F837/76 were used as concentration standards for their respective homologues. The BCA assay measures primarily peptide bonds rather than aromatic amino acid residues, thus decreasing protein-to-protein variation.

Assay for HBL components and sphingomyelinase.
HBL components and sphingomyelinase were identified in chromatography fractions immunochemically with dot blots. The different B and L1 homologues were distinguished from one another by their different apparent molecular masses on Western blots. The L2 homologues were identical in size, but could be distinguished by their widely different chromatographic behaviour. The antibodies and procedures used for the immunoassays were described previously (Beecher et al., 1995a ).

Electrophoresis and protein blotting.
PAGE was performed using precast ReadyGels (Bio-Rad) composed of polyacrylamide with a Tris/glycine buffer system, using Mini Protean II dual-slab electrophoresis cells (Bio-Rad). SDS gels (Laemmli, 1970 ) were composed of 12% polyacrylamide, and native gels (Davis, 1964 ; Ornstein, 1964 ) were composed of 4–15% acrylamide gradients. The minigels were stained with Coomassie brilliant blue R-250 (0·1%, w/v) or silver (Nesterenko et al., 1994 ), or were used for protein blotting. The Pharmacia LMW Marker Kit was used for size standards (phosphorylase b, bovine serum albumin, ovalbumin, carbonic anhydrase and lactalbumin).

Proteins resolved in SDS and native PAGE gels were transferred to PVDF membranes (Millipore) in a Mini Trans-Blot electrophoretic transfer cell (Bio-Rad) at 100 V, constant voltage, for 1 h or 1·25 h, respectively. Isoelectric points were determined on a PhastSystem (Pharmacia) using Pharmacia IEF standards and Phastgel IEF gels with pH ranges of 3–9 and 4–6·5. All gels were run at 2000 V, 2 mA, 3·5 W for 75 V h prior to loading samples and at 200 V, 2 mA, 3·5 W for 15 V h during sample loading. Proteins were focused at 2000 V, 5 mA, 3·5 W for 510 V h (pH 3–9 gels) or 410 V h (pH 4–6·5 gels).

N-terminal sequence analysis.
Purified proteins were prepared for N-terminal sequence analysis as described by Matsudaira (1990) . Electroblotted proteins were sent to the Protein/Nucleic Acid Shared Facility at the Cancer Center of the Medical College of Wisconsin for N-terminal sequence analysis by Edman degradation.

Haemolysis assays: gel diffusion.
For all haemolysis assays (diffusion and turbidimetric), defibrinated sheep blood (Crane Laboratories) was washed in 50 mM Tris/HCl, 150 mM NaCl, pH 7·4 (TBS) prior to use. Gel-diffusion assays for haemolysin BL were as described previously (Beecher, 1990 ; Beecher & Wong, 1994c ) except that 0·3 U ml-1 of B. cereus phosphatidylcholine-preferring phospholipase C (Boehrenger-Mannheim) was added to the gels. This enzyme enhances and preserves the discontinuous haemolysis pattern of HBL by an as-yet unknown mechanism (Beecher & Wong, 1996 ). All gel-diffusion assays were at 22–25 °C.

Haemolysis assays: turbidimetric.
Haemolysis was also measured with turbidimetric assays (Beecher & Wong, 1997 ) modified to a microtitre plate format. Washed sheep erythrocytes were diluted into TBS such that 200 µl in a well of a 96-well microtitre plate had an OD620 of 0·7. At this OD620 there were approximately 6·1x107 sheep erythrocytes ml-1. Cell suspensions were warmed to the assay temperature prior to addition to the assay plate. Lysis was measured as the decrease in OD620 over time using a temperature-controlled SpectraFluor spectrophotometer/microplate reader (Tecan). Haemolysis rates were calculated as the maximum rate of change during the assay period in OD620 min-1x1000 ({Delta}mOD620 min-1) using DeltaSoft 3, a microplate analysis software package for the Macintosh (Biometallics Inc.).

Lysis of naive erythrocytes.
Assay samples consisted either of individual HBL components, combinations of components, or control samples. Samples were either first mixed and diluted, then transferred to empty microtitre wells for the assay, or diluted and then added separately to the empty assay wells. Premixed samples were added to assay wells in volumes from 5 to 20 µl (usually 10 µl), but always in the same volume for a given assay. Samples added separately were generally in 5 µl. Sample volumes were all made identical with TBS when necessary. Immediately prior to the start of the assay, 200 µl of naive erythrocytes (cells that have not been treated with any HBL component) warmed to the assay temperature were added to wells containing the samples and lysis was monitored as described above.

Priming of erythrocytes by B components.
The priming activity of the HBL B component was described in detail by Beecher & Wong (1997) . Erythrocytes treated with the B component do not lyse, but slowly become sensitized to the rapid lytic action of the combined L1 and L2 components. The priming reaction nearly stops at or below 25 °C, whereas the L components rapidly (1–5 min) lyse primed cells, but not unprimed cells. Here, B or Ba was added to erythrocytes at 42 °C, then cooled to 25 °C for the lysis reaction. The extent of priming was quantitated by adding 5 nM L1+2 to B-treated erythrocytes at 25 °C and measuring the percentage of cells that lysed in less than 5 min.

Quantitation of L component activity.
To determine the activity of the L components the rate of the rapid lysis of B-primed erythrocytes was measured (Beecher & Wong, 1997 ). Sheep erythrocytes were treated with B homologues at 42 °C until 100% of the cells were primed. Primed cells were then added to wells containing the desired amount of L components and lysis was measured as described above and expressed as {Delta}mOD620 min-1.

Urea treatments.
Some samples of component Ba were treated with 8 M urea to dissociate complexes and then tested for haemolysis and priming activity. When these samples were to be compared with non-urea-treated components, the same final concentration of urea was added to the erythrocytes separately from the components. During the priming reaction, if present, the urea concentration was 0·11 M. When 25 nM excess B or Ba was tested, if urea was present, the final concentration was 0·4 M. None of the urea concentrations used had an effect on the lysis rate of L1+2 components or the priming rate of B. Urea at 4 M in wells of gel-diffusion assays had no effect on activity of binary combinations of components (i.e. there was no lysis), and had the effects on ternary combinations shown in Results.

Vascular permeability assay.
Vascular permeability activity was assayed as described by Glatz et al. (1974) . Briefly, 50 µl samples were injected intradermally in the shaved backs of rabbits, and after 3 h, Evans blue dye was injected into the marginal ear vein. Samples possessing vascular permeability activity produced areas of oedema and bluing in proportion to the concentration of the toxin. High toxin concentrations also produced areas of necrosis. Female New Zealand White rabbits of 3–4 kg (Covance Research Products) were used for all assays. All activities are reported as means of duplicate or triplicate samples.

Nomenclature.
Prototype HBL components isolated from strain F837/76 are referred to as B(F) L1(F) and L2(F) or else the strain from which they were purified is stated. Among the proteins isolated from strain MGBC 145, the homologues exhibiting the closest relationship to the prototypes from F837/76 (i.e. 87–100% identity in the N-terminal sequences) are designated B, L1 and L2. The more distantly related homologues (i.e. 62–65% identity in the N-terminal sequences) are designated with a subscript ‘a’, i.e. Ba, L1a and L2a. Component Ba was isolated as a complex. Under the assumption that the complex is a dimer, Ba in complex form is sometimes referred to as 2Ba and the urea-treated monomer as Ba. Combinations of the L components are similarly indicated by L1+2(F), L1+2 and L1a+2a. Other combinations are specifically identified, e.g. B(F)+L1+L2a. Combinations, such as HBL or L1+2, were composed of equimolar concentrations of each component unless otherwise specified.


   RESULTS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Isolation of HBL homologues
The following description applies to the isolation of components from B. cereus MGBC 145. For comparisons with the components from MGBC 145, the prototype HBL components were also isolated from B. cereus F837/76 (data not shown).

DE-52 anion-exchange chromatography.
The desalted fraction from the G-25 column was applied directly to the DE-52 column and eluted with a linear gradient of NaCl as described in Methods and by Beecher & Wong (1994c) . The elution characteristics of the B, L1 and L2 components from MGBC 145 were essentially identical to those from F837/76, while elution of the ‘a’ components varied to different extents. The following list indicates what fractions were pooled to isolate each component and the calculated concentration range of NaCl comprised by the fractions. L2a, nonbound fraction (isocratic elution); L1a and L2, fractions 23–26 (0·10–0·13 mM NaCl); L1, (some L2) fractions 27–32 (0·13–0·17 mM NaCl); B, fractions 33–38 (0·17–0·21 mM NaCl); Ba, fractions 39–46 (0·21–0·26 mM NaCl).

HA chromatography.
Except for L2a, the fraction pools listed above were each applied to the 1·6x8 cm HA column and eluted with identical gradients of phosphate buffer. L2a was applied to a smaller HA column as described in Methods. Upon elution from HA, the proteins are nearly pure, so that comparison of their A280 elution profiles becomes informative. Fig. 1(a) compares the elution profiles of each component from HA relative to the phosphate eluent concentration. Only the elution profiles of L2 and L2a were exceptionally different from one another. At this point, L2a was essentially pure, and no further separations were performed. Also of interest, L1a and sphingomyelinase behaved identically on HA.



View larger version (22K):
[in this window]
[in a new window]
 
Fig. 1. Chromatographic elution profiles of HBL and HBLa components. (a) Elution profiles from HA columns. The top four profiles are from the 16 ml HA column and the bottom profile is from the 1 ml HA column (details in Methods). (b) Elution profiles from Mono-Q (anion-exchange) columns. All Mono-Q columns were run at pH 5·9 except for one depicted in the inset, which was run at pH 7·4. Proteins were eluted with linear salt gradients and flow rates as described in Methods. Labels adjacent to peaks identify the major proteins in those peaks. The L1a/S peaks contained both L1a and sphingomyelinase (S). Inset, separation of L1a from sphingomyelinase on the Mono-Q column at pH 7·4. The L1a and S peaks are labelled and eluted at 110 and 140 mM, respectively. The first peak contained no proteins larger than 10 kDa. The x- and y-axes do not apply to this profile, which is included only to illustrate the effectiveness of separation at pH 7·4.

 
Mono-Q anion-exchange chromatography.
Except for L2a, the fractions containing the respective HBL component homologues were pooled and applied to a Mono-Q column. Fig. 1(b) compares the Mono-Q elution profiles for each component. All of the homologous component pairs behaved differently on this column (essentially reflecting their behaviour on DE-52). Note that both peaks that eluted from the column depicted in the top panel in Fig. 1(b) represent the B component. The later peak represents intact B, and the early peak represents B with its first two N-terminal residues (Ser-Glu) cleaved by endogenous protease (unpublished observations). Also note that, as on HA, L1a and sphingomyelinase behaved identically on the Mono-Q column at pH 5·9, but could be easily separated at pH 7·4 (inset, Fig. 1b). Under these conditions, the L1a peak occurred at 110 mM NaCl, and the sphingomyelinase peak was at 140 mM NaCl.

Physicochemical comparison of components
Fig. 2 shows the purified HBL homologues from strains F837/76 and MGBC 145 on SDS-PAGE. Table 1 compares their N-terminal sequences and their estimated molecular masses and isoelectric points. Also shown for comparison is the predicted N-terminal sequence from the hblB open reading frame immediately downstream from the B gene (hblA).



View larger version (83K):
[in this window]
[in a new window]
 
Fig. 2. SDS-PAGE gel of purified HBL and HBLa components. Lanes 3–5, 7–9 and 11–13 contain 1 µg of each purified component as labelled at the top of the gel. Components with a dot above them were isolated from B. cereus F837/76. All other components were from B. cereus MGBC 145. Lanes 1, 2, 6, 10, 14 and 15 contained low-molecular-mass protein standards (s). The gel was run and stained as described in Methods.

 

View this table:
[in this window]
[in a new window]
 
Table 1. N-terminal sequence alignments and characteristics of HBL and HBLa components

 
Between aligned residues, B, L1 and L2 were 92, 100 and 87% identical to their respective prototype components from F837/76. Ba, L1a and L2a were respectively 60, 65 and 63% identical to B, L1 and L2. Interestingly, L2a was more similar to L2(F) over 36 residues (67% identical) than to L2 (63% identical). Ba was 67% identical to the hblB predicted amino acid sequence from F837/76. It is doubtful that these identities reflect the overall identities between the respective sequences; for example, B(F) and the predicted hblB amino acid sequence are approximately 85% identical over 158 residues, but only 60% identical over 43 aligned N-terminal residues.

Isolation of Ba as a complex
We recently found, in experiments unrelated to this study, that treatment of the B component from strain F837/76 at 65 °C converts it to stable but relatively inactive complexes, primarily dimers (analysis to be presented elsewhere). The complexes can be identified as slowly migrating bands on native PAGE gels and can be dissociated to the monomer with 8 M urea, which also restores activity. The native PAGE gel in Fig. 3(a) shows that, in our preparation, Ba was a complex that can be dissociated with urea. Unlike B, which forms mixtures of increasingly larger complexes over time, Ba was present primarily as a single complexed species that migrated at the same rate as the initial complex formed by B. The other proteins purified in this study were all monomers, because their migration distances on native PAGE were not affected by 6–8 M urea (not shown).



View larger version (95K):
[in this window]
[in a new window]
 
Fig. 3. Production of Ba as a monomer and isolation as a dimer. Lanes 1–5, native PAGE of components B and Ba in complex and monomer forms. Each lane was loaded with 0·25 µg protein and the gel was stained with silver. Lane 1 (B65), component B treated at 65 °C, 10 min, to form complexes. Lane 2 (B), untreated component B. Lane 3 (Ba), untreated component Ba. Lane 4, blank. Lane 5 (BaU), component Ba added to gel in the presence of 6 M urea. Note: 8 M urea eliminates upper band (not shown). Lanes 6 and 7, Western blot probed with polyclonal antiserum to B. Lane 6 (F), 20 µl culture supernatant from B. cereus F837/76 (HBL prototype strain). Lane 7 (M), 20 µl culture supernatant from B. cereus MGBC 145 (HBLa-producing strain).

 
Fresh culture supernatants were run on a native PAGE gel and analysed by Western blotting to determine whether Ba is secreted as a dimer. Fig. 3(b) compares supernatants from strains MGBC 145 and F837/76 (which produces only B). No complex was detected in either supernatant. Both supernatants did exhibit one slow-moving band, which did not correspond to the complexes. The identities of these bands are unknown.

Haemolysis by HBL and HBLa components: similarities
Activity of component combinations.
In turbidimetric haemolysis assays, sheep erythrocytes were treated with all possible binary and ternary combinations of the prototype-like and ‘a’ components from MGBC 145. None of the binary combinations was lytic. Among the 20 possible three-component combinations, all eight of those possessing B, L1 and L2 homologues were lytic. None of the 12 combinations that possessed redundant homologues (e.g. B+L1+L1a, B+Ba+L1, etc.) was lytic. Thus, the three-component requirement for HBL activity was maintained, and homologous components could be interchanged freely.

Priming activity of B and Ba.
During the priming reaction, which occurs independently of the L components, naive erythrocytes gradually become sensitive to rapid lytic action by the L components (Beecher & Wong, 1997 ). This slow priming occurred here for both B and Ba from MGBC 145. B at 5 nM completely primed erythrocytes at 40 °C in 1 h, whereas 10 nM Ba monomer (urea-treated) required 1·25 h to prime all of the cells. Untreated Ba complex also primed erythrocytes. However, its activity has not yet been characterized thoroughly because we had only limited quantities, and the monomer appears to be the natural form (Fig. 3b).

Rapid lytic activities of all L components.
We tested all combinations of L and La components for their abilities to lyse B- and Ba-primed cells. For all combinations possessing L1 and L2 homologues (e.g. L1+2, L1a+2, L1a+2a, etc.) lysis by 0·5–10 nM of the L components was rapid and complete (data not shown). L1a appeared to be about half as potent as L1. However, minor concentration inaccuracies cannot yet be ruled out. There were no apparent differences in activities between L2 and L2a.

Haemolysis by HBL and HBLa components: differences
Haemolysis in blood agar.
When B was present the discontinuous haemolysis pattern typical of HBL occurred for all L component combinations (i.e. L1+2, L1+2a, L1a+2 and L1a+2a). This suggests that the L1 and L1a components interact in similar ways with the B component to produce the pattern (L2 neither inhibits B or L1, nor is it inhibited by them and thus does not contribute to the discontinuous pattern: Beecher & Wong, 1997 ). However, when Ba replaced B, the discontinuous pattern did not occur for any L combination, i.e. lysis was continuous from the well edge to the edge of the lysis zone.

Fig. 4 shows the effect of varying the concentration of L1 on the haemolysis pattern in the presence of B or Ba (L1a produced the same effects in blood agar; not shown). The discontinuous pattern occurred at all L1 concentrations in the presence of B, but lysis occurred more quickly near the well at lower L1 concentrations, which appear nearly continuous in the figure. A discontinuous pattern was not evident at any time around the wells containing Ba.



View larger version (79K):
[in this window]
[in a new window]
 
Fig. 4. Effect of urea and L1 concentration on the haemolysis pattern in sheep blood produced by B and Ba. Wells in rows received 100 ng of either B or Ba in the presence or absence of 8 M urea as indicated on the left. Wells in columns received L1 at the concentrations indicated at the bottom. All wells contained 100 ng L2. All four horizontal strips were on the same blood agar gel. The image was recorded after 6 h at 25 °C.

 
Fig. 4 also indicates that 8 M urea significantly enhanced the activity of Ba, but quantitation of the enhancement was not possible because of the inhibitory effects of excess L1. The relative potencies of B and Ba in blood agar were estimated by placing serial dilutions of each in wells surrounding a central well containing 100 ng L1+2. In this method B primes cells surrounding the well without the inhibitory effect of excess L1 (Beecher & Wong, 1997 ). The lowest amounts of B, Ba and Ba+8 M urea exhibiting lytic activity were 0·3, 4 and 1·25 ng per well respectively.

Inhibition of L1 components by B and Ba.
Just as excess L1 inhibits the priming reaction of B to produce discontinuous haemolysis on blood agar, it also inhibits priming in erythrocyte suspensions (Beecher & Wong, 1997 ). Likewise, the addition of excess B to previously primed erythrocytes inhibits the lysis rate of L1 (Beecher & Wong, 1997 ). We examined the abilities of excess B and Ba components to inhibit the lysis of B- and Ba-primed cells by L1 (preliminary experiments indicated that there were no discernible differences between L1 and L1a components; therefore only results using L1 are shown). The results are shown in Table 2.


View this table:
[in this window]
[in a new window]
 
Table 2. Effect of excess B, Ba or 2Ba on erythrocyte lysis by components L1+2

 
Sheep erythrocytes were primed at 42 °C with 5 nM B or 10 nM Ba. The primed erythrocytes were lysed with 5 nM L1+2 in the presence or absence of 25 nM B or Ba. The Ba was added either in 8 M urea to test the monomer, or only in buffer to test the complex (2Ba). Final urea concentrations in samples and controls are described in Methods.

There were striking differences between the effects of excess B and excess Ba on the lysis rate produced by L1. As expected, excess B inhibited the lysis of B-primed cells. However, excess Ba actually enhanced haemolysis of Ba-primed cells. Neither excess B nor excess Ba significantly affected the lysis rate of cells primed by the other protein. Complexed Ba (2Ba in Table 2) did not significantly affect lysis of cells primed by either B or Ba.

Most proteins unfold in 8 M urea, and it was possible that the enhanced activity of urea-treated Ba was due to the unfolding and refolding of Ba rather than dissolution of the complex per se. When B, which was already in monomer form, was first dissolved in 8 M urea, its inhibition of lysis differed by less than 10 % compared to untreated B. This suggests that urea had little effect on the monomer.

Vascular permeability activity of HBL and HBLa
The ability of the HBL and HBLa components to cause vascular permeability and necrosis in rabbit skin was tested using combinations that appeared likely to be active based on their haemolytic characteristics. Samples containing 0·5–1·5 µg of single components and binary combinations produced no reaction. There were no clear differences between samples containing 0·5 µg of each component of HBL from F837/76, or HBL and HBLa from MGBC 145, which produced respectively 20±1·6, 21±1·2 and 20±2·0 mm of bluing and 3·7±2·5, 4·4±3·7 and 4·0±1·2 mm of necrosis (means±SD, n=3). Similar reactions were obtained when Ba was mixed with L1+2 and when B(F) was mixed with L1a+2a. When B and Ba concentrations were varied with constant L1+2, necrosis occurred with 0·5 µg of B and Ba, but not with 0·25 and 0·125 µg. The only difference appeared in the bluing response between 0·125 µg of B and 0·125 µg of Ba, which produced 23±0·5 and 9±2·6 mm zones respectively.

Ba was not treated with urea prior to vascular permeability testing to avoid any complications that high concentrations of urea might produce. This component will be further characterized when conditions are developed to produce significant concentrations of Ba monomer in physiological salt concentrations.


   DISCUSSION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
We have noted that numerous B. cereus strains produce multiple bands on Western blots probed with antibodies specific for components of HBL. The presence of multiple toxin homlogues might complicate attempts to understand virulence mechanisms. Therefore, we tried to determine whether the different bands seen on Western blots represent distinct gene products. Using antibodies to track the chromatographic behaviour of the different forms of the HBL components, we isolated two distinct homologues of each component from the medically important B. cereus isolate MGBC 145. Both sets of homologues, HBL and HBLa, were haemolytic, dermonecrotic, and caused vascular permeability in rabbit skin with similar potencies. Overall, there were no discernible differences in activity between the L homologues, but there were significant differences in activity between B and Ba.

The homologous components differed from one another in size, net charge and chromatographic behaviour. L1 and L1a behaved most similarly, and L1a coeluted with sphingomyelinase from HA and from anion-exchangers at pH 5·9. However, it could be separated from sphingomyelinase at pH 7·4. L2 and L2a were essentially identical in size, but because they differed in pI by a full pH unit, they were readily separated by ion-exchange chromatography, and the exceptionally late elution of L2a from HA made it easy to isolate in as few as two chromatographic steps.

Ba presented a peculiar case. It was isolated as a complex, which may explain its incongruous chromatographic elution. Its apparent pI was 6·0 (Table 1), and one would expect that it would elute earlier from anion-exchange columns than B (pI 5·3). Instead, it eluted later, at 1·3 times the elution volume of B. The heat-generated complexes of B (upper bands, Fig. 3a, lane B65) behave in the same way (data to be published elsewhere). It appears, however, that Ba is naturally produced as a monomer, which is how it appeared in fresh B. cereus culture supernatant (Fig. 3b). We are investigating the factors that result in complex formation during purification.

All of the ‘a’ components were 62–65% identical to their respective homologues over 30–43 residues. Similar identity (60%) in strain F837/76 between the N-termini of B and the predicted hblB sequence rises to approximately 85% over the first 158 residues (Heinrichs et al., 1993 ), suggesting that the HBLa components may prove to be more closely related to HBL components than indicated by the N-termini. The high sequence similarity among the ‘a’ homologues suggests that the three genes may have been either simultaneously duplicated or simultaneously transferred horizontally, and this would have been facilitated by their tandem arrangement in an operon (Ryan et al., 1997 ). We have also determined that the four hbl genes in strain F837/76 all arose via duplication of a single gene (Beecher, 1997 ). Note that HBL has been mapped to a portion of the B. cereus chromosome that exhibits exceptional variability compared with other regions, and that this variable region is sometimes located on large extrachromosomal DNA fragments that appear to be stable, but might also prove to be large mobile plasmids (Carlson et al., 1996a , b ; Carlson & Kolstø, 1994 ). This raises fundamental questions about the extent and role of gene duplication and horizontal transfer in the evolution of B. cereus.

In addition to HBL and HBLa, we have found (data not shown) that strain MGBC 145 expresses another homologous three-component toxin designated nonhaemolytic enterotoxin (NHE) by Lund & Granum (1996) . Data from several studies suggest that, among strains that express either HBL or NHE, 80% of them express both (Buchanan & Schultz, 1994 ; Day et al., 1994 ; Rusul & Yaacob, 1995 ). These data are based on detection of the L2 protein of HBL, or the NHEA protein, by two commercial kits (Beecher & Wong, 1994a ).

Our observations highlight practical problems both for the diagnosis of toxigenic strains and for the identification of B. cereus virulence factors, particularly since there may be additional unidentified HBL homologues. Since the discontinuous haemolysis pattern typical of HBL appears to be produced only in the presence of the prototype B component, discontinuous lysis around colonies or culture supernatants (Beecher & Wong, 1994b ) now appears to be of limited diagnostic value. For maximum value, other methods should detect all relevant toxin components or component combinations. However, not all of these components and combinations are completely defined. In attempts to identify virulence factors, possible contributions from multiple, potentially interacting, multicomponent toxins must be kept in mind.

Although numerous physical phenomena have been documented for HBL, little is known about how the components interact at the molecular level beyond the fact that the they bind to cells independently and must physically interact to form pores. It is also clear that there is a change in the physical state of B between its initial association with the membrane (Bm) and its primed state (B'). This change may involve conformational changes, oligomerization, membrane insertion, or all three. The experiment represented in Table 2 provides some insight into such additional interactions as discussed below.

We previously found that the lysis of B-primed erythrocytes by L1+2 was inhibited by excess B added simultaneously with the L components, and that it is L1 that is inhibited, not L2 (Beecher & Wong, 1997 ). Table 2 shows that excess B and Ba have different effects, and that the specific effect depends on which was used to prime the cells. A significant effect occurred only when the protein added in excess was the same as that used to prime cells. This indicates that the excess Bm must physically interact with B' (likewise for excess and Ba'), and that B and Ba are sufficiently unrelated that they do not interact with one another.

The excess B may inhibit lysis simply by obstructing L1 from contact with B', or by producing inefficient lytic complexes composed of Bm, B' and L1. Enhancement of lysis by Ba was unexpected, and the mechanism by which this occurs is unclear. It might occur either if enhances the incorporation of L1 into the lesion, or if –Ba'–L1 complexes are more efficient than Ba'–L1 complexes.

There is a structural basis for suggesting that L1 binds more efficiently to –Ba' complexes. B and L1 from strain F837/76 appear to be distant homologues, and the most significant structural difference between them is that B is missing 40 residues of a 60-residue apparent transmembrane segment present in L1 (Beecher, 1997 ). The observed size difference of approximately 4·4 kDa between B and Ba (Table 1) equates with a difference of approximately 40 residues. In essence, Ba may act as a hybrid between B and L1 by virtue of retaining the full transmembrane segment that is characteristic of L1.

While our observations emphasize practical difficulties for diagnosis and identification of virulence factors, they also provide promising opportunities for structure–function analysis of the HBL family of toxins. The structural and functional differences between B and Ba are significant and may provide insight into the mechanism of HBL toxicity as well as an explanation for the unique haemolytic phenomena of this toxin.


   ACKNOWLEDGEMENTS
 
This work was supported by grant 96-35201-3765 from the National Research Initiative Competitive Grants Program of the US Department of Agriculture, and by the College of Agricultural and Life Sciences, University of Wisconsin-Madison.


   REFERENCES
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Beecher, D. J. (1990). Isolation and characterization of hemolysin BL: a novel three-component diarrheagenic hemolysin from Bacillus cereus. PhD thesis, The State University of New Jersey, Rutgers.

Beecher, D. J. (1997). Hemolysin BL: structure, function, and diversity of the dermonecrotic enterotoxin from Bacillus cereus. In First International Workshop on The Molecular Biology of Bacillus cereus, Bacillus anthracis, and Bacillus thuringiensis, p. 38. Edited by L. Andrup. Copenhagen: National Institute of Occupational Health. ISBN 87-7534-573-0.

Beecher, D. J. & Macmillan, J. D. (1990). A novel bicomponent hemolysin from Bacillus cereus.Infect Immun 58, 2220-2227.[Medline]

Beecher, D. J. & Macmillan, J. D. (1991). Characterization of the components of hemolysin BL from Bacillus cereus.Infect Immun 59, 1778-1784.[Medline]

Beecher, D. J. & Wong, A. C. L. (1994a). Identification and analysis of the antigens detected by two commercial Bacillus cereus diarrheal enterotoxin immunoassay kits.Appl Environ Microbiol 60, 4614-4616.[Abstract]

Beecher, D. J. & Wong, A. C. L. (1994b). Identification of hemolysin BL-producing Bacillus cereus isolates by a discontinuous hemolytic pattern in blood agar.Appl Environ Microbiol 60, 1646-1651.[Abstract]

Beecher, D. J. & Wong, A. C. L. (1994c). Improved purification and characterization of hemolysin BL: a hemolytic dermonecrotic vascular permeability factor from Bacillus cereus.Infect Immun 62, 980-986.[Abstract]

Beecher, D. J. & Wong, A. C. L. (1996). Interactions affecting hemolysis between enterotoxic hemolysin BL and two phospholipases C from Bacillus cereus. In 96th General Meeting of the American Society for Microbiology, abstract no. B-24. New Orleans: ASM Press.

Beecher, D. J. & Wong, A. C. L. (1997). Tripartite hemolysin BL from Bacillus cereus: hemolytic analysis of component interactions and a model for its characteristic paradoxical zone phenomenon.J Biol Chem 272, 233-239.[Abstract/Free Full Text]

Beecher, D. J., Pulido, J. S., Barney, N. P. & Wong, A. C. L. (1995a). Extracellular virulence factors in Bacillus cereus endophthalmitis: methods and implication of involvement of hemolysin BL.Infect Immun 63, 632-639.[Abstract]

Beecher, D. J., Schoeni, J. L. & Wong, A. C. L. (1995b). Enterotoxic activity of hemolysin BL from Bacillus cereus.Infect Immun 63, 4423-4428.[Abstract]

Buchanan, R. L. & Schultz, F. J. (1994). Comparison of the Tecra VIA kit, Oxoid BCET-RPLA and CHO cell culture assay for the detection of Bacillus cereus diarrhoeal enterotoxin.Lett Appl Microbiol 19, 353-356.[Medline]

Callegan, M. C., Booth, M. C., Jett, B. D. & Gilmore, M. S. (1999a). Pathogenesis of gram-positive bacterial endophthalmitis.Infect Immun 67, 3348-3356.[Abstract/Free Full Text]

Callegan, M. C., Jett, B. D., Hancock, L. E. & Gilmore, M. S. (1999b). Role of hemolysin BL in the pathogenesis of extraintestinal Bacillus cereus infection assessed in an endophthalmitis model.Infect Immun 67, 3357-3366.[Abstract/Free Full Text]

Carlson, C. R. & Kolstø, A. B. (1994). A small (2·4 Mb) Bacillus cereus chromosome corresponds to a conserved region of a larger (5·3 Mb) Bacillus cereus chromosome.Mol Microbiol 13, 161-169.[Medline]

Carlson, C. R., Johansen, T. & Kolstø, A. B. (1996a). The chromosome map of Bacillus thuringiensis subsp. canadensis HD224 is highly similar to that of the Bacillus cereus type strain ATCC 14579.FEMS Microbiol Lett 141, 163-167.[Medline]

Carlson, C. R., Johansen, T., Lecadet, M.-M. & Kolstø, A.-B. (1996b). Genomic organization of the entomopathogenic bacterium Bacillus thuringiensis subsp. berliner 1715.Microbiology 142, 1625-1634.

Davis, B. J. (1964). Disc electrophoresis. II. Method and application to human serum proteins.Ann NY Acad Sci 121, 405-427.

Day, T. L., Tatani, S. R., Notermans, S. & Bennett, R. W. (1994). A comparison of ELISA and RPLA for detection of Bacillus cereus diarrhoeal enterotoxin.J Appl Bacteriol 77, 9-13.[Medline]

Drobniewski, F. A. (1993). Bacillus cereus and related species.Clin Microbiol Rev 6, 324-338.[Abstract]

Glatz, B. A., Spira, W. M. & Goepfert, J. M. (1974). Alteration of vascular permeability in rabbits by culture filtrates of Bacillus cereus and related species.Infect Immun 10, 299-303.[Medline]

Heinrichs, J. H., Beecher, D. J., Macmillan, J. D. & Zilinskas, B. A. (1993). Molecular cloning and characterization of the gene encoding the B component of hemolysin BL from Bacillus cereus.J Bacteriol 175, 6760-6766.[Abstract]

Kramer, J. M. & Gilbert, R. J. (1989). Bacillus cereus and other Bacillus species. In Foodborne Bacterial Pathogens, pp. 21-70. Edited by M. P. Doyle. New York & Basel: Marcel Dekker.

Laemmli, U. K. (1970). Cleavage of structural proteins during the assembly of the head of bacteriophage T4.Nature 227, 680-685.[Medline]

Lund, T. & Granum, P. E. (1996). Characterisation of a non-haemolytic enterotoxin complex from Bacillus cereus isolated after a foodborne outbreak.FEMS Microbiol Lett 141, 151-156.[Medline]

Matsudaira, P. (1990). Limited N-terminal sequence analysis.Methods Enzymol 182, 602-612.[Medline]

Nesterenko, M. V., Tilley, M. & Upton, S. J. (1994). A simple modification of Blum’s silver stain method allows for 30 minute detection of proteins in polyacrylamide gels.J Biochem Biophys Methods 28, 239-242.[Medline]

Ornstein, L. (1964). Disk electrophoresis. I. Background and theory.Ann NY Acad Sci 121, 321-349.

Rusul, G. & Yaacob, N. H. (1995). Prevalence of Bacillus cereus in selected foods and detection of enterotoxin using TECRA-VIA and BECT-RPLA.Int J Food Microbiol 25, 131-139.[Medline]

Ryan, P. A., Macmillan, J. D. & Zilinskas, B. A. (1997). Molecular cloning and characterization of the genes encoding the L1 and L2 components of hemolysin BL from Bacillus cereus.J Bacteriol 179, 2551-2556.[Abstract]

Schoeni, J. L. & Wong, A. C. L. (1999). Heterogeneity observed in the components of hemolysin BL, an enterotoxin produced by Bacillus cereus.Int J Food Microbiol 53, 159-167.[Medline]

Turnbull, P. C. B. (1986). Bacillus cereus toxins. In Pharmacology of Bacterial Toxins, pp. 397-448. Edited by F. Dorner & J. Drews. Oxford: Pergamon Press.

Received 24 August 1999; revised 7 December 1999; accepted 16 March 2000.