N-terminal Activation Is an Essential Early Step in the Mechanism of Action of the Bacillus thuringiensis Cry1Ac Insecticidal Toxin*

A variant form of the Bacillus thuringiensis Cry1Ac toxin that is not cleaved at the N terminus during proteolytic activation with trypsin was found to be incapable of forming pores in Manduca sexta brush border membrane vesicles in vitro and had reduced insecticidal activity in vivo . Binding studies indicated an altered binding pattern of the mutant toxin in that bound toxin could not be fully displaced by a high molar excess of fully trypsin-activated toxin. These results suggest that proteolytic removal of the N-terminal peptide of Cry1Ac is an important step in toxin activation. The entomocidal Cry1 toxins of Bacillus thuringiensis are synthesized as inactive protoxins of around 130 kDa within the bacterial cytoplasm (1). Upon ingestion by a susceptible insect the protoxin is activated through the proteolytic removal of an N-terminal peptide of 25–30 amino acids and approximately half of the remaining protein from the C terminus. The activated toxin then binds to specific sites on the brush border membrane of the midgut epithelium before inserting into the membrane and forming a pore. B. thuringiensis toxins pro-duced in transgenic plants have often been expressed as pro-teins truncated at the C terminus because this has resulted in higher expression levels (2). These truncated forms are regu-larly described as the active form of the


A variant form of the Bacillus thuringiensis Cry1Ac toxin that is not cleaved at the N terminus during proteolytic activation with trypsin was found to be incapable of forming pores in Manduca sexta brush border membrane vesicles in vitro and had reduced insecticidal activity in vivo. Binding studies indicated an altered binding pattern of the mutant toxin in that bound toxin could not be fully displaced by a high molar excess of fully trypsin-activated toxin. These results suggest that proteolytic removal of the N-terminal peptide of Cry1Ac is an important step in toxin activation.
The entomocidal Cry1 toxins of Bacillus thuringiensis are synthesized as inactive protoxins of around 130 kDa within the bacterial cytoplasm (1). Upon ingestion by a susceptible insect the protoxin is activated through the proteolytic removal of an N-terminal peptide of 25-30 amino acids and approximately half of the remaining protein from the C terminus. The activated toxin then binds to specific sites on the brush border membrane of the midgut epithelium before inserting into the membrane and forming a pore. B. thuringiensis toxins produced in transgenic plants have often been expressed as proteins truncated at the C terminus because this has resulted in higher expression levels (2). These truncated forms are regularly described as the active form of the toxin despite having an intact N terminus. The role of the C-terminal extension to the active toxin is believed to be in the formation of crystalline inclusion bodies within the bacterium and is dispensable for toxicity (3). Little is known about the role of the N-terminal peptide and whether its removal is important in the mechanism of action of the toxin. It has previously been observed that if an engineered Cry1Ab or Cry1Ca toxin lacking the N-terminal peptide is expressed in Escherichia coli, growth of the E. coli culture is severely affected (4 -6). It was speculated that the N-terminally truncated toxin was free to interact with the cell membrane of the E. coli host cell and that the first 28 amino acids prevented the Cry toxin from inserting into the membrane. When expressed in B. thuringiensis an N-terminally truncated Cry1Ca toxin was expressed at a much lower level than the full-length toxin, and the formation of crystals was repressed. Although this might have indicated a role of the N terminus in promoting crystallization, the low levels of expression and lack of crystals might also have been an effect of an inappropriate interaction between the toxin and the host cell (6). In this article we compare the toxicity, binding, and pore-forming abilities of activated Cry1Ac with a partially activated recombinant form of Cry1Ac in which the N terminus remains intact after treatment with trypsin.

EXPERIMENTAL PROCEDURES
Preparation of the Toxin Samples-Site-directed mutagenesis was used to remove the N-terminal tryptic cleavage site of Cry1Ac using methods described previously (7). Two amino acid substitutions were made, R28T and I29M. Both mutant and wild-type toxins were expressed in E. coli JM109, and crystalline material was purified from sonicated cell cultures by sucrose density centrifugation. The protoxin was solubilized in 50 mM Na 2 CO 3 , pH 11.4 (37°C, 1 h) and then activated by incubation with trypsin (1 mg/ml; 37°C, 1 h) in the same buffer. The trypsin-treated toxin was purified by Mono Q FPLC 1 and stored in buffer A: 100 mM methylglucamine chloride, 10 mM 2-N-cyclohexylaminoethanesulfonic acid, pH 9.5 ( Fig. 1).
Preparation of Brush Border Membrane Vesicles (BBMVs)-Manduca sexta larvae were reared on artificial diet (Bio-Serv) at 23°C with a photoperiod of 16 h of light and 8 h of darkness. BBMVs from fifth instar larvae were prepared and analyzed as reported previously (8) except that neomycin sulfate (2.4 g/ml) was included in the buffer (300 mM mannitol, 20 mM 2-mercaptoethanol, 5 mM EGTA, 1 mM EDTA, 0.1 mM phenylmethanesulfonyl fluoride, 150 g/ml pepstatin A, 100 g/ml leupeptin, 1 g/ml soybean trypsin inhibitor, 10 mM HEPES-HCl, pH 8.0). The vesicles were dialyzed overnight against 400 volumes of 150 mM KCl, 10 mM HEPES-HCl, pH 7.5 (Sigma) and sonicated for six periods of 30 s each at 25°C (BRANSON 1200 sonic bath, Danbury, CT) in the same solution. BBMV enrichment was estimated according to the alkaline phosphatase (AP) and cytochrome-c oxidase activity relative to the initial homogenate (9.5-fold increase in AP/mg of protein). Vesicle orientation was determined from the AP activity in the presence and absence of 0.02% Triton X-100.
Fluorescence Measurements-Pore formation activity was assayed by monitoring changes in membrane potential with the fluorescent, positively charged dye 3,3Ј-dipropylthiodicarbocyanine (Dis-C 3 -(5), Molecular Probes, Eugene, OR) as described previously (9). Fluorescence was recorded at 620/670 nm in a Hansatech system (Norfolk, England). Hyperpolarization causes dye internalization into the BBMVs and a decrease in fluorescence; depolarization causes the opposite effect. BBMVs (10 g) previously loaded with 150 mM KCl were suspended in 900 l of 150 mM N-methyl-D-glucamine chloride, 10 mM HEPES-HCl, pH 8 buffer. After equilibration of the dye (2 min), 50 nM toxin was added. Changes in membrane potential were monitored by successive additions of KCl to the BBMV suspension. Analyses of the slope (m) of ⌬F (%) versus K ϩ equilibrium potential (E K ϩ) (mV) curve are reported in this work. E K ϩ was calculated with the Nernst equation.
Binding Assay-Binding analysis and homologous competition of wild-type and mutant toxins to M. sexta BBMVs were performed in a semiquantitative binding assay. Toxin was biotinylated by using biotinyl-N-hydroxysuccinimide ester (RPN28, Amersham Biosciences) according to the manufacturer's instructions. Binding was performed in 100 l of binding buffer (PBS, 0.1% (w/v) bovine serum albumin, 0.1% (v/v) Tween 20, pH 7.6). 20 g of BBMV protein were incubated with 10 * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ To whom correspondence should be addressed. Tel.: 44-1273-678917; Fax: 44-1273-678433; E-mail: n.crickmore@sussex.ac.uk. nM biotinylated toxin in the absence or presence of different (50 -1000)fold excesses of unlabeled, activated Cry1Ac toxin for 1 h. The unbound toxin was removed by centrifugation for 10 min at 14,000 ϫ g. The BBMVs were suspended in 100 l of binding buffer and washed twice with the same buffer. Finally the BBMVs were suspended in 20 l of PBS, pH 7.6, and an equal volume of 2ϫ Laemmli sample loading buffer (0.125 M Tris-HCl, pH 6.8, 4% SDS, 20% glycerol, 10% 2-mercaptoethanol, 0.01% bromphenol blue) was added. Samples were boiled for 3 min, electrophoresed in SDS-polyacrylamide gels, and electrotransferred to nitrocellulose membranes. The biotinylated protein that remained bound to the vesicles was visualized by incubating with streptavidin-peroxidase conjugate (1:4000 dilution) for 1 h followed by luminol as described by the manufacturers.
Bioassays-Bioassays were conducted by means of the diet-surface contaminating procedure. A constant volume of the sample to be tested (35 l in PBS) was applied on the diet surface contained in 24-well plates (Cell Wells, Corning Glass Works, Corning, NY). Two neonate M. sexta larvae were placed in each well, and 24 larvae were assayed for each protein concentration. The plates were incubated under the following conditions: 28°C, 65 Ϯ 5% relative humidity, and a 16:8 h, light:dark photoperiod. Mortality was recorded after 7 days. The effec-tive dose estimates (50% lethal concentration of the toxins, LC 50 ) were calculated using PROBIT analysis.
Midgut Juice Isolation and in Vitro Processing of Protoxins-Fifth instar M. sexta larvae were chilled for 10 min on ice, and midgut tissue was dissected. Midgut juice was separated from solid material by centrifugation (5,000 rpm for 10 min) and filtered through 0.22-m filters. The midgut juice was stored at Ϫ70°C. 2 g of protoxin in 30 l of 50 mM Na 2 CO 3 /NaHCO 3 , pH 11.0 were incubated at 37°C for 2 h in the presence of different amounts of M. sexta midgut juice (1:10, 1:100, and 1:1000, midgut juice:protoxin, w/w). Proteolysis was stopped by adding phenylmethanesulfonyl fluoride (1 mM, final concentration), and the samples were briefly centrifuged to remove insoluble material. 5-l samples were subjected to SDS-PAGE.

RESULTS AND DISCUSSION
In vitro pore formation assays performed with either fully or partially activated Cry1Ac toxin clearly demonstrate that the partially activated toxin is incapable of forming pores in M. sexta BBMVs since no membrane polarization was observed upon addition of toxin (Fig. 2). To test whether this could have been due to reduced binding of the partially activated toxin, binding assays were performed (Fig. 3). The results show that initial binding of labeled, fully activated Cry1Ac toxin to M. sexta BBMVs is competitive and largely abolished in the presence of an excess of unlabeled toxin. In contrast, although the partially activated toxin binds to the BBMVs a significant proportion remains bound even in the presence of the large excess of unlabeled fully activated toxin. When the assay was repeated in the absence of BBMVs no labeled toxin was observed in the insoluble fraction.
Bioassays were performed with protoxin forms of the wildtype and mutant toxins. The wild-type protoxin was found to have an LC 50 of 0.1 (0.08 -0.15) ng/cm 2 , whereas the mutant protoxin had an activity of 2.5 (1.8 -3.9) ng/cm 2 . Given that no appreciable in vitro pore formation activity could be observed with the mutant toxin it was interesting to observe that the mutant protoxin demonstrated in vivo toxicity, albeit ϳ25-fold lower than the wild-type toxin. A possible explanation for this is that proteinases other than trypsin within the M. sexta gut were fully activating the mutant toxin. This was tested by incubating wild-type and mutant toxins with varying concentrations of M. sexta gut extract and monitoring activation by SDS-PAGE. The results are shown in Fig. 4. At the highest concentration of gut extract used both protoxins appear fully activated, suggesting that alternative enzymes do indeed exist. At higher dilutions of gut extract it can be seen that while the wild-type protoxin was fully activated, the mutant protoxin was only partially activated at the N terminus. N-terminal sequencing has confirmed that the higher molecular weight band has the same N terminus as the protoxin. The reduced toxicity of the mutant protoxin could thus be explained by the reduced rate of N-terminal activation. The fact that mutant toxin, activated by endogenous proteinases, has insecticidal activity eliminates the possibility that the lack of pore-forming ability described earlier could have been due to a misfolded and thus inactive protein.
It has previously been suggested that the presence of the N-terminal peptide might prevent binding to non-target membranes (4), suggesting that the pore-forming potential of domain I is only realized after proteolytic removal of this peptide. In the homologous Cry1Ab protoxin a similar mutation that prevented N-terminal processing also showed reduced toxicity to M. sexta larvae (4). Indirect evidence for a possible role of the N-terminal fragment in modulating binding of the toxin has come from the solution of the structure of the Cry2Aa toxin (10). Unlike previously reported structures for the Cry3Aa and Cry1Aa toxins, the Cry2Aa toxin had not been proteolytically activated and therefore contained an uncleaved N-terminal peptide. The structure revealed that the N-terminal region masks a region of the toxin believed to be involved in the interaction between the toxin and the brush border membrane of the target insect. The N terminus of the Cry1 toxins might similarly prevent an interaction between the toxin and the target membrane, an important initial step in pore formation. The altered binding properties of the mutant toxin could be due to an additional binding activity between the N-terminal peptide and the BBMVs following binding of the toxin to its receptor. This additional binding affinity would be strong enough to prevent full displacement of the toxin by an excess of competitor.
The conclusion to be drawn from this work is that N-terminal cleavage of a Cry1 protoxin is required before the toxin can properly be considered active. The reduced activity of the mutant toxin in vivo suggests that it might be possible to engineer a toxin that would only be fully active in insects sharing a specific combination of receptor and gut proteinase.