INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The gram-positive, endospore-forming bacterium Bacillus thuringiensis produces parasporal crystalline inclusions that contain polypeptides (-endotoxins) that are toxic to a variety of insect species (16). Upon ingestion by an insect larva, these inclusions are solubilized in the alkaline environment of the midgut and are activated by midgut proteases (23, 40). The activated toxins then pass through the peritrophic matrix and subsequently bind to a highly specific receptor(s) on the midgut brush border membrane (14, 15, 41). Perhaps following a conformational change and/or oligomerization, the toxin induces the formation of a lytic pore in the midgut epithelial membrane that results in cell lysis, cessation of feeding, and death of the larva (27).

The ~130-kDa Cry1 -endotoxins undergo extensive proteolysis at both their C and N termini to produce a mature toxic moiety that has a molecular mass of approximately 60 kDa. This relatively protease-resistant toxic core is derived from the N-terminal half of the protoxin by removal of 500 to 600 amino acid residues from the C terminus and the first 27 to 29 N-terminal residues (2, 4, 7, 17, 30, 31, 36, 43). Activated Cry1Ac is generated by cleavages at R28 and K623 in the protoxin (4). It has been postulated that the C-terminal region that is lost through activation directs assembly of the crystal and facilitates efficient solubilization at alkaline pH values (42). It is interesting to note that Cry1Ac activation proceeds through seven specific cleavages, starting from the C terminus and perhaps involving a sequence of conformational changes, to remove the C-terminal half of the protoxin (8). The resulting 10- to 35-kDa protoxin-derived fragments are themselves rapidly proteolysed into peptides, and they apparently play no further role in toxicity. Whether this mechanism of activation is true for all Cry1 -endotoxins has not been elucidated to date.

Correct activation of a B. thuringiensis -endotoxin is likely to be a prerequisite for toxicity, and insufficient processing or overdigestion of a toxin may render it inactive. The battery of midgut proteases that an insect possesses is therefore likely to be a major determinant of toxin potency. The midgut lumina of lepidopteran insect larvae have been shown to contain a variety of alkaline proteases, mainly members of the serine protease class, that exhibit predominantly trypsinlike and chymotrypsinlike protease activities (9, 19, 20, 28). Such midgut proteases are likely to be responsible for -endotoxin activation.

A number of reports have suggested that -endotoxin proteolysis is a major determinant of toxicity. It has been demonstrated that a strain of Plodia interpunctella (Indian meal moth) resistant to the -endotoxins of B. thuringiensis subsp. entomocidus HD-198 exhibited a lower protoxin activation rate than susceptible insects exhibited due to a decrease in the total proteolytic activity of the gut extract (32, 33). Ingaki et al. (18) found the reverse to be true for Spodoptera litura processing of B. thuringiensis subsp. kurstaki HD-1. They reported that complete degradation of the toxin by proteases derived from the nonsusceptible organism S. litura was the likely cause of the lack of potency. Similarly, Keller et al. (20) suggested that reduced sensitivity of fifth-instar larvae of Spodoptera littoralis to Cry1C could be attributed to increased degradation of the toxin in the less susceptible larvae. Ogiwara et al. (31) compared the sites of proteolytic cleavage of the -endotoxins of B. thuringiensis subsp. kurstaki HD-1 and HD-73. They showed that the sites of N-terminal proteolysis depended on the insect that the proteases were derived from and that the difference may have accounted for the differences in potency. Bai et al. (3) showed that the protease levels in regurgitated gut juice from Pieris brassicae were higher than those found in gut juice from Mamestra brassicae and S. littoralis, and they suggested that this may account for the lack of susceptibility of the latter two insects to B. thuringiensis subsp. thuringiensis. They also showed that P. brassicae had a higher proportion of trypsinlike and chymotrypsinlike proteases than the other two insect species. Haider et al. (13) clearly demonstrated that differential proteolysis could determine the specificity of a toxin. Activation of the Cry1Ab protoxin from B. thuringiensis subsp. aizawai with lepidopteran gut enzymes yielded a 55-kDa protein that was toxic only to Lepidoptera. Further treatment of this 55-kDa polypeptide, or intact protoxin, with dipteran gut proteases resulted in production of a 53-kDa Diptera-specific toxic core.

In this study, we investigated the role of proteolysis in determining the potency of Cry1Ac towards P. brassicae (large white butterfly) and M. brassicae (cabbage moth) larvae. Höfte and Whiteley (16) demonstrated that Cry1Ac was highly active towards P. brassicae, whereas it was relatively inactive against M. brassicae. Here, we demonstrate that differences in the way that Cry1Ac is proteolysed by proteases from the two insects may contribute to the large difference in toxin potency observed.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Strains and plasmids. Escherichia coli TG1 (35) was used for all standard molecular biology techniques. Plasmid pMSV.Cry1Ac (37) was used for expression of Cry1Ac -endotoxin crystals in the acrystalliferous strain B. thuringiensis subsp. israelensis IPS-78/11 (5).

Site-directed mutagenesis. Site-directed mutagenesis was performed as described by Smedley and Ellar (37) by using the Altered-Sites in vitro mutagenesis system (Promega) and plasmid pMSV.Cry1Ac as a template. Mutagenic oligonucleotides were synthesized by the Protein and Nucleic Acid Chemistry Facility, Department of Biochemistry, University of Cambridge, Cambridge, United Kingdom. Mutant plasmids were transformed into B. thuringiensis subsp. israelensis IPS-78/11 by electroporation (5).

Purification of -endotoxin inclusions. -Endotoxin inclusions were purified from sporulated cultures of B. thuringiensis subsp. israelensis IPS-78/11 containing the desired plasmid as described previously (39). The yield of purified crystals was determined by the method described by Lowry et al. (26) by using bovine serum albumin fraction V (BSA) as a standard.

Toxicity assays. P. brassicae larvae were reared from eggs that were kindly provided by Horticulture Research International, Wellesbourne, Warwickshire, United Kingdom. M. brassicae larvae were reared from eggs that were obtained from T. Carty, Institute for Virology and Environmental Microbiology, National Environment Research Council, Oxford, United Kingdom. A toxin crystal suspension or solubilized toxin was incorporated into molten artificial diet (50°C) at a range of concentrations. After the diet had set and dried, 20 third-instar larvae per concentration or 40 neonate larvae per concentration were placed on the diet, and mortality was determined after 6 days of incubation at 25°C under a light cycle consisting of 16 h of light and 8 h of darkness. The concentrations of toxin at which 50% of the larvae were killed (LC₅₀s) were determined by Probit analysis (11). To determine Cry1Ac toxicity, bioassays were repeated three times, and an average LC₅₀, with associated 95% confidence intervals, was calculated. In cases where mortality was low, the LC₅₀ could not be determined.

SDS-PAGE and immunoblotting. Proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) by using a modification of the method of Laemmli and Favre (22), as described by Thomas and Ellar (39). Proteins were transferred from the gel to nitrocellulose with a Bio-Rad semidry blot apparatus by using a blot buffer containing 39 mM glycine, 48 mM Tris (pH 9.0), 0.0375% (wt/vol) SDS, and 10% (vol/vol) methanol. Toxin was detected by the method described by Knowles et al. (21) or with an ECL Plus chemiluminescence detection kit (Amersham).

In vitro toxin processing. To prepare gut juice, fifth-instar larvae were chilled on ice for 10 min before they were dissected. The peritrophic membrane containing the food bolus was isolated and centrifuged at 30,000 × g at 4°C for 20 min. The supernatant was removed and recentrifuged for 20 min at the same speed, and the resulting supernatant was removed and filtered through a 0.22-µm-pore-size membrane. Cry1Ac protoxin inclusions were solubilized in 50 mM Na₂CO₃-10 mM dithiothreitol (DTT)(pH 10.0) for 30 min at 37°C. A Bio-Rad protein assay (Bio-Rad) was performed with the solubilized toxin, and the concentration was subsequently adjusted to 1 mg/ml. Gut juice was added to the solubilized toxin at a concentration of 5% (vol/vol), and a time course at 37°C was followed. Samples (50 µl) were removed at 10 min and 2, 8, and 24 h and separated into soluble and insoluble fractions by centrifugation at 30,000 × g at 4°C for 20 min; then each pellet was washed twice with ice-cold phosphate-buffered saline (PBS). The extent of toxin proteolysis was determined by SDS-13% PAGE as described above. Protease cleavage sites were determined by N-terminal sequencing of the -endotoxin polypeptides following SDS-13% PAGE and transfer of proteins onto polyvinylidene difluoride membranes. This was done by the Protein and Nucleic Acid Chemistry Facility, Department of Biochemistry, University of Cambridge, Cambridge, United Kingdom.

Qualitative binding assay. The procedure used for the qualitative binding assay was essentially that of Knowles et al. (21). Cry1Ac was activated with 1% (vol/vol) purified gut juice (see above) for 1 h at 37°C, and this was followed by centrifugation at 30,000 × g at 4°C for 20 min to remove insoluble material. One hundred micrograms of brush border membrane vesicles (BBMV) (44) was incubated with 10 µg of activated toxin in 100 µl (final volume) of PBS containing 1 mg of BSA per ml for 60 min at room temperature without shaking. The mixture was centrifuged at 30,000 × g at 4°C for 20 min, and the resultant BBMV pellet was washed twice with cold PBS containing 1 mg of BSA per ml. Samples of the supernatant containing unbound toxin and the pellet containing toxin bound to membrane were separated by SDS-13% PAGE, followed by immunoblotting as described above.

In vivo toxin processing. Newly molted fifth-instar larvae of M. brassicae and P. brassicae were starved for 12 h. A 10-mg/ml Cry1Ac toxin crystal suspension (or 10 mg of BSA per ml as a control) was made up in a 25 mM sucrose solution containing approximately 10 mg of cabbage powder per ml. Fifteen microliters of the toxin suspension was fed to the larvae by placing a drop in front of an advancing larva with a micropipette. After the toxin suspension had been completely imbibed, the larvae were incubated at 22°C for either 30 min or 3 h; 10 larvae were used for each treatment. After this the larvae were dissected and separated into three fractions (midgut, insoluble fraction of the food bolus, and soluble fraction of the food bolus) as follows (all subsequent fractionation steps were performed in a cold room at 4°C). To prepare the midgut fraction, midguts were isolated from toxin-treated and BSA-fed larvae and then washed by vortexing them in 500 µl of ice-cold PBS containing an inhibitor cocktail consisting of 10 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 10 µM leupeptin, 10 µM antipain, 20 µg of (4-amidophenyl)-methanesulfonyl fluoride per ml, and 135 µM L-1-chloro-3-(4-tosylamide)-7-amino-2-heptanone. The midguts were then homogenized in 100 µl of fresh PBS containing the inhibitor cocktail (PBS/inhibitors) and subsequently washed three times by repeated resuspension of the pellet in PBS/inhibitors and centrifugation at 15,000 × g at 4°C for 15 min in a minifuge. The final pellet was resuspended in 100 µl of PBS/inhibitors, and protein was precipitated with 25% trichloroacetic acid (TCA) at 20°C for 2 h, which was followed by centrifugation at 30,000 × g at 4°C for 15 min. The pellet was washed twice with ice-cold acetone, and then the pellet was air dried. The precipitate was then boiled for 5 min in 80 µl of 1× SDS-PAGE loading buffer and separated by SDS-PAGE and immunoblotting performed as described above. To prepare the insoluble fraction of the food bolus, after dissection peritrophic membranes containing the food bolus were removed from the gut, and a 10× PBS/inhibitors solution was added to give a 1× concentration. The sample was then centrifuged at 15,000 × g at 4°C for 15 min in a minifuge, and the supernatant was removed (see below). The pellet was then homogenized in 100 µl of ice-cold PBS/inhibitors and subsequently washed three times by repeated centrifugation at 15,000 × g at 4°C for 15 min in a minicentrifuge and resuspension in PBS/inhibitors. The final pellet was then TCA precipitated and processed in the same way that the midgut membrane sample was processed (see above). To prepare the soluble fraction of the food bolus, after centrifugation of the peritrophic membranes containing the food bolus as described above, the supernatant was removed and centrifuged again at 30,000 × g at 4°C for 15 min in a minifuge. The supernatant was removed, TCA precipitated, and processed in the same way that the midgut membrane fraction was processed (see above).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Toxicity of Cry1Ac against P. brassicae and M. brassicae larvae. Höfte and Whiteley (16) reported the activities of a number of -endotoxins towards P. brassicae and M. brassicae, which demonstrated that Cry1Ac was the most potent toxin against P. brassicae whereas it was relatively inactive against M. brassicae. However, the LC₅₀s for the two insect species were expressed in different units, making direct comparisons difficult. It was therefore essential to obtain bioassay data produced under standardized conditions. Toxicity assays were performed with purified Cry1Ac -endotoxin inclusions and third-instar P. brassicae and M. brassicae larvae (LC₅₀s are shown in Table 1).

In vitro toxin processing. Studies of Cry1Ac activation with gut juice derived from P. brassicae and M. brassicae were performed. Soluble and insoluble fractions were isolated and separated by SDS-13% PAGE, which was followed by staining with Coomassie brilliant blue (Fig. 1). It should be noted that on SDS-PAGE gels with the molecular weight markers used, activated Cry1Ac appeared to migrate at a molecular weight lower than that estimated from its amino acid sequence. This anomaly may have been due to aberrant migration of the activated toxin and/or the 47.5-kDa protein marker. For this reason all estimates of the molecular weights of all toxin products were based on data for the soluble activated form of Cry1Ac (bands 1 and 5) having a molecular mass of 60 kDa.

View larger version (60K): [in this window] [in a new window]   FIG. 1.   Proteolytic activation of Cry1Ac with M. brassicae and P. brassicae gut juice. Solubilized -endotoxin inclusions were activated with 5% (vol/vol) gut juice from M. brassicae (lanes 2 to 9) or P. brassicae (lanes 10 to 17). Ten micrograms of toxin was loaded onto a SDS-13% PAGE gel, and electrophoresis was followed by Coomassie brilliant blue staining. Lanes 2, 4, 6, 8, 10, 12, 14, and 16, insoluble toxin fraction; lanes 3, 5, 7, 9, 11, 13, 15, and 17, soluble toxin fraction. Lane 1 contained molecular weight markers. Toxin was activated for 10 min (lanes 2, 3, 10, and 11), 2 h (lanes 4, 5, 12, and 13), 8 h (lanes 6, 7, 14, and 15), or 24 h (lanes 8, 9, 16, and 17).

Qualitative binding of Cry1Ac to BBMV from M. brassicae and P. brassicae. A qualitative binding assay based on the method described by Knowles et al. (21) was used to investigate the interactions of Cry1Ac with BBMV prepared from M. brassicae and P. brassicae (44). As a control, the assay was also performed in the absence of BBMV. Soluble material containing unbound toxin and insoluble material containing toxin bound to BBMV were analyzed by SDS-13% PAGE, followed by immunoblotting (Fig. 2).

View larger version (52K): [in this window] [in a new window]   FIG. 2.   Binding of Cry1Ac to P. brassicae and M. brassicae BBMV. Qualitative binding was performed with activated toxin and BBMV from P. brassicae (A) and M. brassicae (B). Twenty microliters of either supernatant or pellet was analyzed by SDS-13% PAGE followed by immunoblotting with anti-Cry1Ac polyclonal antisera. Lane 1, molecular weight markers; lanes 2 and 4, soluble material; lanes 3 and 5, insoluble material. Lanes 2 and 3 contained Cry1Ac incubated in the absence of BBMV, and lanes 4 and 5 contained Cry1Ac incubated with BBMV.

Role of membrane-associated and midgut lumen proteases in Cry1Ac binding. As described above, the forms of Cry1Ac bound to BBMV from P. brassicae and M. brassicae differed. The differences in proteolysis may contribute to the ability of the toxin to irreversibly associate with the membrane and consequently form lytic pores. To determine whether membrane-associated or gut juice-derived proteases were responsible for the specific proteolysis of membrane-bound toxin, qualitative binding was performed by using Cry1Ac activated with either 1% (vol/vol) P. brassicae gut juice or 1% (vol/vol) M. brassicae gut juice incubated with BBMV from both insects (Fig. 3).

View larger version (59K): [in this window] [in a new window]   FIG. 3.   Role of membrane-associated and midgut lumen proteases in Cry1Ac binding. Cry1Ac activated with either P. brassicae or M. brassicae gut juice was mixed with BBMV from both insects, and qualitative binding assays were performed. Lane 1, prestained molecular weight markers; lanes 2, 4, 6, and 8, Cry1Ac remaining in the supernatant; lanes 3, 5, 7, and 9, Cry1Ac present in the pellet fraction. The following combinations of activated toxin and BBMV were used: lanes 2 and 3, M. brassicae-activated Cry1Ac and M. brassicae BBMV; lanes 4 and 5, P. brassicae-activated Cry1Ac and P. brassicae BBMV; lanes 6 and 7, P. brassicae-activated Cry1Ac and M. brassicae BBMV; and lanes 8 and 9, M. brassicae-activated Cry1Ac and P. brassicae BBMV.

In vivo toxin processing. Most of the studies on toxin activation described previously were carried out in vitro by using purified gut juice preparations. Below we describe a novel experiment that was designed to monitor toxin activation and fate in vivo. The aim of the experiment was to determine whether the cleavage events that were observed in vitro (see above) also occur in the guts of the two insects. Three gut fractions were isolated from toxin-treated and BSA-fed insects and were analyzed by SDS-13% PAGE, followed by immunoblotting: a midgut membrane fraction, an insoluble fraction of the gut including the peritrophic membrane, and a soluble fraction of the gut (Fig. 4).

View larger version (58K): [in this window] [in a new window]   FIG. 4.   In vivo processing of Cry1Ac in P. brassicae and M. brassicae larvae. Immunoblotting of in vivo-activated Cry1Ac was performed. Lanes 1 to 9, samples derived from M. brassicae; lanes 10 to 18, samples derived from P. brassicae. Lanes 1, 4, 7, 10, 13, and 16 contained fractions isolated from insects fed BSA (control); lanes 2, 5, 8, 11, 14, and 17 contained fractions isolated from insects that were fed toxin and incubated for 30 min; and lanes 3, 6, 9, 12, 15, and 18 contained fractions isolated from insects that were fed toxin and incubated for 3 h. Lanes 1 to 3 contained the M. brassicae midgut membrane fraction; lanes 4 to 6 contained the M. brassicae insoluble fraction from the insect gut; lanes 7 to 9 contained the M. brassicae soluble fraction from the insect gut; lanes 10 to 12 contained the P. brassicae midgut membrane fraction; lanes 13 to 15 contained the P. brassicae insoluble fraction from the insect gut; lanes 16 to 18 contained the P. brassicae soluble fraction from the insect gut; and lane 19 contained molecular weight markers.

View larger version (59K): [in this window] [in a new window]   FIG. 5.   ECL Plus detection of membrane-associated Cry1Ac. The membrane fraction from the in vivo processing experiment was detected by using an ECL Plus kit (Amersham) in combination with anti-Cry1Ac polyclonal antisera. Lane 1, molecular weight markers; lane 2, in vitro-activated Cry1Ac (positive control); lane 3, M. brassicae midgut membrane isolated from BSA-fed insects (control); lane 4, P. brassicae midgut membrane isolated from BSA-fed insects; lane 5, M. brassicae midgut membrane isolated from Cry1Ac-fed insects after 3 h of incubation; lane 6, P. brassicae midgut membrane isolated from Cry1Ac-fed insects after 3 h of incubation. The arrow indicates the position of the membrane-associated toxin.

Construction and analysis of Arg423 mutant toxins. In an attempt to prevent proteolytic breakdown of the toxin and to determine whether cleavage at Arg423 contributed to the low potency of Cry1Ac towards M. brassicae, site-directed mutagenesis was used to replace Arg423 with Ala, Gln, or Ser. Oligonucleotides were designed that would enable these substitutions and at the same time remove a BglI restriction endonuclease site from the pMSV.Cry1Ac plasmid (Table 3).

View larger version (72K): [in this window] [in a new window]   FIG. 6.   Proteolytic activation of the R423Q and R423S mutants with M. brassicae gut juice. Solubilized R423Q (A) or R423S (B) mutant -endotoxin inclusions were activated with 5% (vol/vol) gut juice from M. brassicae for 24 h and separated into insoluble and soluble fractions. Lane 1, molecular weight markers; lane 2, insoluble toxin; lane 3, soluble toxin.

Toxicity of the R423Q and R423S mutants. Due to differences in the purities of the wild-type Cry1Ac and mutant crystal preparations, toxin crystals were solubilized in 50 mM Na₂CO₃-10 mM DTT (pH 10) for 1 h before they were incorporated into the artificial diet. Due to the low potency of Cry1Ac towards M. brassicae larvae, a single concentration, 50 µg of toxin/g of diet, was used to challenge neonate M. brassicae larvae (Table 4).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Previously, toxicity data for P. brassicae and M. brassicae have been generated by using different methods (i.e., diet incorporation and leaf dip assays), and this has made comparisons of toxin potencies for these two insect species difficult. Using B. thuringiensis IPS-78/11 purified Cry1Ac toxin inclusions, we performed three independent bioassays and combined the resulting data to obtain mean LC₅₀ with 95% confidence limits. These data demonstrated that Cry1Ac was at least 2,000 times more toxic towards P. brassicae larvae than towards M. brassicae larvae. This difference provides a useful model with which to investigate the factors that contribute to toxin potency and insect resistance.

By using fifth-instar larval gut juice, similarities and differences in Cry1Ac activation were identified for the two insect species. As determined by using proteases from both insects, the N terminus of the ~60-kDa, activated, soluble form of Cry1Ac occurred at Ile29, a putative trypsinlike cleavage site. This finding is consistent with the findings of Bietlot et al. (4) and suggests that differences in the soluble form of activated Cry1Ac are unlikely to account for differences in toxin potency. In addition to differences in the cleavage positions within domain I of Cry1Ac producing insoluble forms of the toxin, M. brassicae was also shown to process Cry1Ac at Arg423, a trypsinlike cleavage site not recognized by P. brassicae proteases. Arg423 is located towards the C-terminal end of domain II between putative -sheets 9 and 10, before the exposed loop 3. The importance of the loop regions in toxin binding to insect midgut membranes and in determining activity has been demonstrated (34, 38, 45). Domain III has also been shown to be important in toxin binding to putative midgut receptors (6, 10, 24). If cleavage at Arg423 occurs in vivo, then it is highly likely that removal of loop 3 of domain II and all of domain III would destroy the structural integrity of the toxin molecule and consequently result in an inactive form. Interestingly, both insects generated a ~60-kDa insoluble form of Cry1Ac with the same N terminus as the soluble toxin (Ile29). It is possible that C-terminal processing induces the toxin to adopt a more open conformation that ultimately produces an insoluble product that comigrates with the soluble form of the toxin during SDS-PAGE. Although providing no information about processing at the C terminus of the toxin, which may also affect potency, these data highlighted clear differences in the pattern of Cry1Ac proteolysis in the two insect species. Although the two insects seem to possess similar arrays of protease classes, the specificities of the proteases themselves seem to differ. The different processing of Cry1Ac in the two insects may contribute to the observed differences in toxin potency, and in the case of M. brassicae, degradation of Cry1Ac may act as a specific resistance mechanism. A similar pattern of toxin proteolysis was observed in vivo, implying that the conditions used in in vitro experiments were representative of the gut environment. It is interesting to note that large proportions of toxin were present in the insoluble fractions of the guts of both insects.

Milne et al. (29) reported that a protein complex present in the midgut of Choristoneura fumiferana (spruce budworm) could inactivate Cry1Aa by precipitation followed by proteolysis, thus accounting for the low potency of this toxin towards this insect. It is possible that a similar resistance mechanism exists in M. brassicae. However, it would be paradoxical if P. brassicae had evolved such a process since this insect is highly susceptible to Cry1Ac. It is unlikely that the insoluble form of the toxin retains activity, since the aggregate would be unable to pass through the peritrophic matrix and subsequently bind to a receptor on the underlying midgut epithelium.

A possible explanation for this paradox is that if proteolysis occurs while the toxin is bound to a specific receptor, then instead of precipitating out of solution, as it does in vitro, it undergoes a conformational change, exposing hydrophobic residues that enable the toxin to oligomerize and/or insert into the hydrophobic membrane. The ability of a toxin to form a pore may therefore be dependent on the sites at which proteolytic nicking occurs, as well as the accessibility of the sites. Domain I of the Cry -endotoxins exists as a helical bundle with the hydrophobic faces of the amphipathic helices surrounding a hydrophobic central helix. Proteolytic nicking within domain I may impart greater flexibility to the molecule, perhaps allowing helix pairs from the bundle to act as initiators of membrane insertion (25), or alternatively may induce oligomerization of several nicked toxin molecules. Hence, exposure of hydrophobic segments of the toxin by selective proteolysis when the toxin is close to the membrane may favor irreversible association with the membrane, followed by oligomerization and pore formation.

Knowles et al. (21) showed that Cry1Ac was associated with BBMV from Manduca sexta, Heliothis virescens, and P. brassicae as a doublet. More recently, Aronson et al. (1) showed that binding of Cry1Ac and Cry1Ab to M. sexta and H. virescens BBMV induced oligomerization of toxin molecules. This suggests that toxin proteolysis at the membrane surface may play a role in oligomerization and subsequent toxin insertion into the membrane. However, Aronson et al. (1) did not determine whether the oligomerization of toxin molecules occurs in nonsusceptible insects, such as M. brassicae, and this possibility should be investigated.

In the qualitative binding experiments described in this paper, Cry1Ac was found to bind to BBMV from two insects. The forms of Cry1Ac associated with the midguts of the two insects differed. In M. brassicae, Cry1Ac was bound as a doublet with products having molecular masses of 60 and 58 to 59 kDa, and the higher-molecular-mass form accounted for approximately 90% of the bound toxin. In P. brassicae, the toxin was bound as a distinct doublet with products having molecular masses of 60 and ~56 kDa that were present in similar amounts. This implies that the two insects are likely to have different Cry1Ac-binding and/or postbinding mechanisms. The generation of a ~56-kDa form of Cry1Ac may be a prerequisite for activity, and differential processing of membrane-bound Cry1Ac may account for the large difference in toxin potency in these two insect species. The sites at which processing occurs with gut juice may also be recognized by membrane-associated proteases in order to generate the nicked forms of membrane-bound Cry1Ac. The similar sizes of the lower-molecular-weight membrane-associated forms of the toxin and the gut juice-activated nicked toxin products support this possibility. Using a chemiluminescence detection method, we found that Cry1Ac was present as a doublet in midguts from intoxicated P. brassicae larvae and as a single band in the M. brassicae membrane. These findings support the view that differential toxin proteolysis may play a role in membrane binding, and thus toxicity, in vivo.

Formation of the lower-molecular-weight bound products in P. brassicae and M. brassicae BBMV was independent of the source of the gut juice used to activate Cry1Ac. This suggests that proteases associated with the BBMV, not the proteases present in insect gut juice, are likely to be responsible for the proteolytic nicking of Cry1Ac in its bound form. Alternatively, it may suggest that the interaction of Cry1Ac with the P. brassicae receptor(s) and/or membrane may induce a specific conformational change that does not occur during binding to M. brassicae BBMV and exposes amino acids to an increased level of proteolytic attack.

Removal of Arg423, a putative trypsinlike cleavage site within Cry1Ac, prevented degradation of Cry1Ac to the ~40- and ~20-kDa fragments with M. brassicae gut juice. Bioassays demonstrated that the potency of the Cry1AcR423S and Cry1AcR423Q mutants was not dramatically increased. This suggests that although degradation of Cry1Ac to ~40- and 20-kDa products in the gut of M. brassicae may contribute to the low potency of the toxin towards this insect, it is unlikely to be the major determinant. It is probable that other factors, such as binding to a specific functional receptor(s) on the midgut brush border membrane, specific proteolysis within domain I to generate an active hydrophobic form of the toxin, and the propensity of the toxin to insert into the target membrane and induce the formation of a pore, play additional roles in the specificity and potency of Cry1Ac. Modification of the Cry1Ac toxin molecule at sites that control these properties may increase the potency even further.