Bacillus thuringiensis Cry1Ab Mutants Affecting Oligomer Formation Are Non-toxic to Manduca sexta Larvae*

Pore-forming toxins are biological weapons produced by a variety of living organisms, particularly bacteria but also by insects, reptiles, and invertebrates. These proteins affect the cell membrane of their target, disrupting permeability and leading eventually to cell death. The pore-forming toxins typically transform from soluble, monomeric proteins to oligomers that form transmembrane channels. The Cry toxins produced by Bacillus thuringiensis are widely used as insecticides. These proteins have been recognized as pore-forming toxins, and their primary action is to lyse midgut epithelial cells in their target insect. To exert their toxic effect, a prepore oligomeric intermediate is formed leading finally to membrane-inserted oligomeric pores. To understand the role of Cry oligomeric pre-pore formation in the insecticidal activity we isolated point mutations that affected toxin oligomerization but not their binding with the cadherin-like, Bt-R1 receptor. We show the helix α-3 in domain I contains sequences that could form coiled-coil structures important for oligomerization. Some single point mutants in this helix bound Bt-R1 receptors with similar affinity as the wild-type toxin, but were affected in oligomerization and were severally impaired in pore formation and toxicity against Manduca sexta larvae. These data indicate the pre-pore oligomer and the toxin pore formation play a major role in the intoxication process of Cry1Ab toxin in insect larvae.

Bacillus thuringiensis (Bt) 2 is a Gram-positive bacterium that produces insecticidal Cry toxins. Cry proteins are widely used for insect control in agriculture and forestry and against mosquitoes, due to their high specificity and safety for humans and for the environment (1,2).
Although Bt Cry toxins are widely used as insecticides their mode of action is still not completely understood. These Cry toxins are pore-forming toxins that induce cell death by forming ionic pores following insertion into the membrane, causing osmotic lysis of larvae midgut cells (1)(2)(3). However, recently an alternative model proposed that these toxins activate a signal pathway through Bt-R 1 receptor interaction, which results in insect cell death without the participation of lytic pores into the membrane (4). It is important to note that this alternative model was proposed based on the effect of Cry1Ab toxin to cultured Trichoplusia ni H5 insect cells expressing the Manduca sexta toxin receptor, Bt-R 1 .
Nevertheless, in both models, receptor interaction with exposed regions in domains II and III of Cry1A toxins (1-4) is a key step that determines insect toxicity. In the case of Cry1A toxins, two receptors have been characterized in several lepidopteran species: cadherin-like proteins, known as Bt-R receptors (5) (Bt-R 1 in the case of M. sexta), and glycosylphosphatidylinositol-anchored proteins, as aminopeptidase-N or alkaline phosphatase (6,7).
In the pore-forming model, it is proposed that both receptors are important and participate in a sequential manner (3,8,9). After proteolytic activation of Cry1A protoxin by insect midgut proteases, the activated toxin binds to a Bt-R receptor, and this interaction facilitates additional cleavage of the N-terminal end of the toxin, resulting in the formation of a pre-pore oligomer (10). This oligomeric structure has been observed in several Cry toxin members of the three-domain family (11)(12)(13)(14)(15)(16)(17). The prepore binds to a second receptor, a glycosylphosphatidylinositol-anchored aminopeptidase, due to its higher affinity to this receptor, facilitating insertion of the oligomeric toxin into membrane lipid rafts resulting in pore formation (3,18). However, a glycosylphosphatidylinositol-anchored alkaline phosphatase protein has been also recognized as a Cry1Ac receptor in Heliothis virescens and M. sexta (7,8). This receptor could also participate in facilitating pore membrane insertion potentially explaining why Cry1Ac toxin mutations that were severely affected in aminopeptidase binding were still active against M. sexta larvae (19,20). In this model, the pre-pore oligomer is responsible for initiating cell death.
In other pore-forming toxins mutations affecting toxin oligomerization correlated with a severe disruption in toxicity, suggesting that oligomerization is an essential step (21)(22)(23)(24)(25). To determine the role of oligomer formation for Cry1Ab toxin action, we isolated and characterized single point mutations that affected toxin oligomerization. We show that domain I helix ␣-3 contains sequences that potentially form coiled-coil structures important for oligomerization. The ␣-helical coiledcoils, also called leucine zippers, constitute an important protein-folding motif (26). These coiled-coil structures are formed in the interaction between two to five ␣-helices and are involved in oligomerization of several proteins, forming specific oligomers with high thermodynamic stability (27). Sitedirected mutagenesis in helix ␣-3 results in Cry1Ab mutants, which still bind Bt-R 1 receptor as the wild-type toxin but are affected in their oligomerization, resulting in a complete loss of pore formation activity and non-toxic against M. sexta larvae. These results indicate that the pre-pore oligomer is an obligate intermediate in the intoxication process of Cry1Ab toxin.

MATERIALS AND METHODS
Construction of Coiled-coil Mutants-Residues localized at positions a and d of the coiled-coil structures predicted for helices ␣-3 and ␣-6 were selected for mutagenesis. Substitutions L100E, L103E, Y107E, and Y110E were produced by sitedirected mutagenesis (QuikChange, Stratagene, La Jolla, CA) using the pHT315Ab plasmid harboring the cry1Ab gene, and substitutions R93F, R99E, Y193D, L196D, I200D, and Y203D were obtained by overlap-extension PCR as described (28) using a ClaI fragment of cry1Ab gene cloned into pBluescript-SK (Stratagene, CA) as template. Automated DNA sequencing at facilities of the Instituto de Biotecnología-Universidad Nacional Autónoma de México verified the single point mutations. The mutated ClaI fragments were obtained by restriction and subcloned into pHT315Ab⌬ClaI. Acrystalliferous B. thuringiensis strain 407 was transformed with the recombinant plasmids containing the mutated cry1Ab genes as reported (29) and selected in Luria broth at 30°C supplemented with 10 g ml Ϫ1 erythromycin.
Cry1Ab Toxin Purification-B. thuringiensis transformant strains were grown for 3 days at 30°C in nutrient broth sporulation medium supplemented with 10 g ml Ϫ1 erythromycin until complete sporulation. Crystal inclusions were observed under phase-contrast microscopy and purified by sucrose gradients (30). Purified crystals were solubilized in 50 mM Na 2 CO 3 , 0.2% ␤-mercaptoethanol, pH 10.5. The monomeric toxins were obtained by trypsin activation in a mass ratio of 1:20 (1 h at 37°C). Phenylmethylsulfonyl fluoride (1 mM final concentration) was added to stop proteolysis. The oligomeric Cry1Ab structure was produced as described previously (10,11) or by incubation of 0.2 g of protoxin with scFv73 antibody (mass ratio 1:2) and 5% midgut juice from M. sexta larvae, in 100 l of solubilization buffer for 1 h at 37°C in the presence of phosphatidylcholine-small unilamellar vesicles (PC-SUVs) prepared as reported (11) at 12 M final concentration. The membrane fraction was separated by ultracentrifugation (30 min at 430,000 ϫ g), and the pellet was suspended in 20 l of 10 mM CHES, 150 mM KCl, pH 9. Finally, the Cry1Ab toxin inserted into the liposomes was visualized by Western blot. A control without PC-SUV was included, and under this condition no protein or pellet was recovered after centrifugation.
Western Blot of Cry1Ab Toxin-Wild-type and Cry1Ab mutant proteins incorporated into PC-SUV were boiled 5 min in Laemmli sample loading buffer, separated in SDS-PAGE, and electrotransferred onto nitrocellulose membrane. The proteins were detected in Western blots (10,11) using polyclonal anti-Cry1Ab (1/30,000, 1 h) and a secondary antibody coupled with horseradish peroxidase (HRP, Sigma) (1/5,000, 1 h) followed by luminol (ECL, Amersham Biosciences) as described by the manufacturers.
scFv73 Antibody Purification-Antibody scFv73 was purified from Escherichia coli cells by a nickel-agarose column as described previously (10).
Size-exclusion Chromatography-60 M SUV containing the activated Cry1Ab toxins (mutant or wild-type) were solubilized 1 h at 25°C with 100 l of 10% (v/v) Triton X-100 and clarified by centrifugation at 152,000 ϫ g for 60 min at 4°C. Detergent extracts were then applied to Superdex 200 HR 10/30 (Amersham Biosciences, Uppsala, Sweden) fast-protein liquid chromatography size exclusion with a mobile phase of phosphatebuffered saline/1 mM EDTA plus 0.1% Triton X-100 (31). Fraction samples were analyzed in Western blot using polyclonal anti-Cry1Ab antibody as above.
Bioassays-Different doses of pure crystals (from 0.1 to 2000 ng/cm 2 ) were applied onto the diet surface of 24-well polystyrene plates (Cell Wells, Corning, NY), one first instar M. sexta larvae per well, using 24 larvae per toxin concentration in three repetitions (10). The plates were incubated at 28°C with 65% Ϯ 5% relative humidity and a light-dark photoperiod of 16:8 h. Mortality was recorded after 7 days, and the 50% lethal concentration (LC 50 ) was estimated by using Probit analysis (Polo-PC LeOra Software).
Preparation of Brush-border Membrane Vesicles-M. sexta eggs were reared on artificial diet. BBMVs from fourth instar M. sexta larvae were prepared as reported before (32).
Toxin Binding Assay and Binding Competition-BBMV protein (10 g) were incubated in 100 l of binding buffer (phosphate-buffered saline, 0.1% bovine serum albumin, w/v, 0.1% Tween 20, v/v, pH 7.6) with 5 nM biotinylated wild-type Cry1Ab toxin (RPN28, Amersham Biosciences) in the absence or presence of different -fold excesses (50 -1,000) of unlabeled toxins for 1 h. Unbound toxin was washed twice by centrifugation (10 min at 14,000 ϫ g). The resulting membrane pellet was boiled for 5 min in Laemmli sample loading buffer, loaded in 10% SDS-PAGE, and transferred to Hybond-ECL nitrocellulose membranes (Amersham Biosciences). The biotinylated toxin bound to the vesicles was visualized by incubating with streptavidin-HRP conjugate (1:4,000 dilution) for 1 h and developed with luminol as described by the manufacturers. Scanning of the 60-kDa signal was performed to quantify binding.
Cloning and Expression of Truncated Bt-R 1 Fragment-Total RNA from M. sexta midgut tissue was prepared using the acid guanidinium-phenol-chloroform method (33). cDNA was generated with a First Strand cDNA synthesis kit for reverser transcription-PCR (AMV, Roche Applied Science), using TCT TAA GCT TGC TGA AGG TGT CCG CGA TAA AG primer. A PCR reaction was performed with Vent-Polymerase (New England Biolabs, Beverly, MA) and primers TAC AGAATTCC ATG ATC GAC TTC CTC ACG GGT CAA ATT TCC and TCT TAAGCTTG CTG AAG GTG TCC GCG ATA AAG, that include EcoRI and HindIII restriction sites (underlined), respectively. The PCR product (1960 bp) was purified with the QIAquick PCR purification kit (Qiagen, Valencia, CA), digested with EcoRI and HindIII (New England Biolabs), and ligated into the previously digested vector pET22b (Novagen, EMD Biosciences, Inc.). DNA constructs were electroporated into E. coli BL21 cells. Protein expression of the His-tagged protein and purification on a nickel affinity column were performed as reported (34). This partial Bt-R 1 clone expresses a protein fragment corresponding to residues 810 -1480 of Bt-R 1 receptor and includes the toxin-binding regions from cadherin repeats CR7 to CR12.
Quantitative Determination of Binding Affinities by Competition ELISA-The Bt-R 1 fragment (10 nM) was incubated with increasing concentrations of Cry1Ab or mutant toxins (from 0.1 nM to 1 M) in 100 l of binding buffer for 1 h at 25°C. The incubation mixtures were transferred to a 96-well ELISA plate previously coated with 2.5 g of Cry1Ab toxin, to determine the amount of unbound Bt-R 1 fragment. The ELISA plates were washed three times and detected with anti-HisTag antibody coupled to HRP (Qiagen) and detected by HRP enzymatic activity. The concentration of Cry1Ab or mutant toxins at which the half-maximal ELISA signal is detected corresponds to the apparent dissociation constant K D (35)(36)(37).
Ligand Blot Assay-Ten micrograms of purified Bt-R 1 fragment was separated in 10% SDS-PAGE and electrotransferred to nitrocellulose membrane. The membrane was blocked with bovine serum albumin and incubated for 2 h with 10 nM biotinylated Cry1Ab wild-type or mutant toxins. Unbound toxins were removed by washing 3ϫ with 20 ml of phosphate-buffered saline buffer. Bound toxins were detected by incubation with streptavidin-HRP conjugate (1:5000) for 1 h and visualized with luminol (ECL, Amersham Biosciences).
Pore-forming Activity of Cry1Ab Toxins-Black lipid bilayers were made as reported (38) with egg-derived PC (Avanti Polar Lipids). Bilayer capacitance values were between 250 and 300 picofarads. Buffers 300 mM KCl, 10 mM CHES, pH 9, and 10 mM KCl, 10 mM CHES, pH 9, were added to the cis and trans compartments, respectively. Once a bilayer was formed, 2.4 M SUV containing the activated Cry1Ab toxins (mutant or wildtype) was added to the cis compartment; the trans compartment was held at virtual ground. Single-channel currents were recorded with a patch clamp amplifier (3900A, Dagan Corp., Minneapolis, MN) as described (39). Currents were filtered at 200 or 500 Hz, digitalized on-line at 1 or 2 kHz, and analyzed using a Digidata 1200 interface and Axotape and pClamp software (Axon Instruments, Foster City, CA).

Regions of Domain I Involved in
Oligomerization-Previous studies indicated that domain I of Cry1Ab toxin was involved in oligomerization (40). Because domain I is a seven ␣-helix-bundle we hypothesized that coiled-coil structures of certain ␣-helices might be involved in oligomer formation. To analyze this, we determined the possibility of coiled-coil structure formation by using the program COILS that compares a sequence to data base of known coiled-coils and derives a similarity score, then by comparing this score to the distribution of scores in globular and coiled-coil proteins, this program calculates the probability that the sequence will adopt a coiled-coil conformation (41). A value of 1 is the highest value indicating that the amino acid sequence has a high probability to form a coiled-coil structure. The amino acid sequence of coiled-coil structures is characterized by heptads of residues, (abcdefg) n , with a unique pattern of internal a and d positions occupied mostly by apolar residues forming and hydrophobic core and positions g and e occupied by charged residues. Analyses of protein sequences of some Cry toxins revealed that helix ␣-3 gave the highest score (0.4 for Cry1Ab, 0.97 for Cry2Aa, 0.99 for Cry3A, and 0.90 for Cry3Ba). Although the primary structure of helix ␣-3 is not conserved among Cry toxins, it is surprising that positions a and d of the predicted coiled-coil structure are highly conserved among members of the three-domain Cry toxin family (Fig. 1A). Alignment of helix ␣-3 revealed that the corresponding residues at positions d (Trp-117, Tyr-110, Leu-103, and Ala-96) are present in all Cry1, Cry3, Cry7, Cry8, and Cry9 toxins, whereas Cry4, Cry2, and Cry10 toxins show some conservative substitutions. The same analysis showed that residues corresponding to positions a (Arg-93, Leu-100, Tyr-107, and Phe-114) are also conserved among members of the threedomain Cry toxin family, the first residue being a positive charged residue (Arg or Lys), whereas the three following residues in these positions correspond to hydrophobic amino acids (Tyr, Phe, Val, or Leu). The second highest score for Cry toxins was helix ␣-6 (0.1 for Cry1Ab), but it is surprising that in some Cry toxins as Cry1Ca, Cry1Fa, Cry4Ba, Cry5Ba, Cry7Aa Cry8Aa, Cry13Aa, Cry14Aa, Cry20Aa, Cry21Aa, and Cry30Aa the probability score to form coiled-coils for helix ␣-6 ranged up to 0.99.
To determine the role of each ␣-helix in oligomer formation, we assayed for Cry1Ab oligomer formation in the presence of synthetic peptides with the amino acid sequences corresponding to the different helices of domain I. It is important to mention that the structure of each of these peptides was analyzed by CD spectroscopy revealing that they adopt predominantly ␣-helical structure (data not shown), similar to what has been observed for synthetic peptides of Cry3Aa toxin (42,43) and consistent with their structure in the intact Cry1Ab molecule. In this assay, Cry1Ab protoxin was activated in the presence of antibody scFv73 that mimics a cadherin-binding epitope (10,18). Fig. 2 shows that the synthetic peptide corresponding to helix ␣-3 inhibited formation of the 250-kDa oligomer. In contrast the synthetic peptide corresponding to helix ␣-6 did not affect oligomer formation. Neither of the other ␣-helices of domain I inhibited oligomer formation (data not shown).
Construction of Cry1Ab Mutants in Helices ␣-3 and ␣-6-To determine the role of helix ␣-3 in oligomer formation, we performed site-directed mutagenesis of residues in positions a (R93F, L100E, Y107E) and d (L103E and Y110E) of the predicted coiled-coil structure. Also, the R99E mutant, in position g of the predicted coiled-coil, was constructed, because this Cry1Aa mutation previously showed a loss of toxicity and pore formation (44), effects that could be explained by impairment in oligomer formation. We also mutagenized helix ␣-6 residues in positions a (Y193D and I200D) and d (L196D and Y203D) of the predicted coiled-coil structure. Fig. 1B shows the location of the mutated residues in the monomeric Cry1A toxin. Residues in position a (illustrated as space-filling model in red) and d (illustrated as space-filling model in blue) of the coiled-coil sequence of helices ␣-3 and ␣-6 face the interior of domain I, whereas residue Arg-99 (space-filling model in green) faces the surface of the protein.
Some of the mutant toxins produced small bipyramidal crystals that were highly susceptible to degradation with trypsin (R93F, L103E, Y110E, Y193D, and L196D), suggesting that these residues could be important in maintaining the monomeric toxin structure. Thus, these mutants were not further analyzed. The rest of the protein mutants (R99E, L100E, Y107E, I200D, and Y203D) produce crystal inclusions similar to the wild-type toxin. After trypsin digestion these mutants gave 60-kDa fragments as the wild-type toxin, which were recognized by several monoclonal and polyclonal Cry1Ab-specific antibodies (data not shown), suggesting that these mutations did not cause major structural disturbance. Table 1 shows the toxicity of the Cry1Ab mutants to M. sexta larvae. The mutants R99E and Y107E located in helix ␣-3 were severally affected in toxicity with LC 50 Ͼ 2000 ng/cm 2 . In contrast, the L100E mutant and the two mutants in helix ␣-6, showed slightly higher or similar toxicity to M. sexta larvae as the wild-type Cry1Ab toxin.
Mutants in Helix ␣-3 Bind to the Bt-R 1 Receptor-To determine if the non-toxic Cry1Ab mutants had altered receptor binding, trypsin-activated proteins were labeled with biotin, and a qualitative binding assay to M. sexta BBMV was performed (Fig. 3A). All mutants were able to bind to BBMV with similar affinity, because heterologous binding competition of biotinylated-Cry1Ab toxin with unlabeled mutant proteins revealed that all mutants competed with the binding of Cry1Ab as the unlabeled wild-type Cry1Ab (Fig. 3B). To characterize the binding of Cry1Ab mutants to Bt-R 1 receptor, ligand blots of biotinylated proteins against a Bt-R 1 protein fragment that contains cadherin repeats CR7 to CR12 was performed. This cadherin region contains all the Cry1A binding sites characterized so far (45,46). Fig. 3C shows that all mutants bound to the Bt-R 1 protein fragment (Fig. 3C). Finally, the binding affinity of the mutants to the Bt-R 1 protein fragment was quantitatively determined by ELISA binding competition assays. Table 1 shows that all Cry1Ab mutants bound the Bt-R 1 fragment with a similar dissociation constant as wild-type Cry1Ab. This ELISA procedure permits the determination of true association rate constants in solution, and several reports have shown good   agreement in the determination of K D values by ELISA binding competition assays and those obtained by conventional methods (immunoprecipitation of the radiolabeled antigen, fluorescence transfer, or surface plasmon resonance) (35)(36)(37)45). Toxicity Directly Correlates with in Vitro Oligomer Formation and Pore Formation-The Cry1Ab mutants were proteolytically activated in the presence of antibody scFv73 to determine if the mutations affected the formation of pre-pore oligomeric structure. Fig. 4A shows that the two protein mutants that were severely affected in toxicity, R99E and Y107E located in helix ␣-3, did not form the 250-kDa oligomeric structure. In contrast, L100E, I200D, and Y203D mutants produced the 250-kDa oligomer. To determine if the R99E and Y107E mutants could form unstable oligomers, the oligomeric structures were revealed after SDS-PAGE under less rigorous conditions avoiding boiling of the sample before electrophoresis. Fig. 4B shows that the R99E did form an oligomeric structure, but this structure was extremely sensitive to heat denaturation in contrast with the wild-type Cry1Ab oligomer that is quite stable and resisted all temperatures, including boiling (Fig. 4B). Similarly, the Y107E oligomer was found to be heat-labile (data not shown).
We also analyzed oligomer formation by size-exclusion chromatography. Protein samples activated in the presence of scFv73 antibody and SUV were delipidated with Triton X-100 and loaded in a Superdex 200 HR column (10,11,31). Previous studies with Cry1Ab showed that oligomeric structure eluted in fractions 9 -11 of the column, whereas monomeric structure eluted in fractions 16 -18 (10, 11). Fig. 4C shows that oligomer structure of the wild-type toxin eluted in fractions 10 -12, whereas with mutants Y107E and R99E no oligomer was observed in these fractions, only the monomeric structure was observed in fractions 17 and 18 of the column. These data suggest that oligomer structures of the mutants Y107E and R99E were unstable and highly sensitive to detergent delipidation.
Black lipid bilayers were used to analyze the pore formation activity of the mutant and Cry1Ab protein samples that were proteolytically activated in the presence of scFv73 and liposomes. Current amplitude induced by toxin samples incorporated in the same concentration of SUV was analyzed. Control sample containing only SUV gave a negative response. These experiments were performed in a KCl gradient 300/10 mM in the cis/trans compartments to facilitate liposomes insertion in the bilayer. The observed responses with the wild-type Cry1Ab and mutants I200D, Y203D, and L100E, showed stable channels with high open probability, and these responses were very similar to the previously characterized pore formation induced by pure Cry1Ab pre-pore oligomer (11). Mutants I200D and Y203D, located in helix ␣-6, showed similar I/V curves as with the Cry1Ab toxin (Fig. 5). In contrast, the observed response of mutant L100E located in helix ␣-3 showed a higher current that suggests a more efficient insertion of the toxin in liposomes during the activation process (Fig. 5). Finally, mutants R99E and Y107E located in helix ␣-3 were severely affected in permeability, showing no change in the current or a low response when liposomes containing these proteins were added to the lipid bilayer, suggesting an important role of these residues in gating and permeability.
It was previously reported that some mutation in the anthrax protective antigen (PA), that result in non-toxic proteins unable to translocate the lethal factor, and edema factor are also defective in pore formation and produce defective oligomeric structures that lost SDS-resistance (47). These PA mutations were characterized as dominant negative, because when mixed FIGURE 4. Analysis of oligomer formation of Cry1Ab and mutant toxins. A, the oligomeric structure of Cry1Ab toxins was produced by incubation of soluble protoxin with scFv73 antibody (which mimics Bt-R 1 receptor), 5% midgut juice from M. sexta larvae, and small unilamellar vesicles (PC-SUV). Membrane pellets were recovered by centrifugation, and protein samples were boiled 3 min, loaded in the SDS-PAGE 1% SDS, transferred to nitrocellulose membrane, and detected by Western blot using polyclonal anti-Cry1Ab antibody. B, analysis of heat sensitivity of the oligomeric structure of Cry1Ab toxin and mutant R99E. Oligomeric structures of Cry1Ab and mutants R99E and Y107E were incubated 5 min at different temperatures, loaded in the SDS-PAGE 0.1% SDS, transferred to nitrocellulose membrane, and detected by Western blot. C, oligomeric structures of Cry1Ab and mutant R99E were delipidated with Triton X-100 and loaded in Superdex 200 HR column. Fractions of the size-exclusion chromatography were boiled 3 min, loaded in the SDS-PAGE 1% SDS, transferred to nitrocellulose membrane, and detected by Western blot using polyclonal anti-Cry1Ab antibody. with wild-type toxin (tested in 1:8 mixtures wt/mutant) the toxicity of wt toxin was inhibited (47). We analyzed if mutant R99E was able to inhibit toxicity of wild-type Cry1Ab. Mixtures of 1:10 (Cry1Ab: R99E) showed a measurable inhibitory activity (up to 60% inhibition of Cry1Ab toxicity) similarly to the dominant negative mutants of PA.

DISCUSSION
Bt Cry proteins belong to a class of bacterial toxins known as pore-forming toxins that are secreted as water-soluble proteins and undergo conformational changes to form oligomers that insert into cell membranes of their target (48).
Previous studies involving partial unfolding of the oligomeric Cry1Ab structure in the presence of urea and limited proteolysis suggested that the ␣-helical domain I of Cry1Ab toxin is involved in monomer-monomer interaction of the pre-pore (40). Because ␣-helical coiled-coils structures are involved in the oligomerization of several proteins (27), we decided to look for coiled-coils in Cry toxins. We found that helix ␣-3, of domain I of Cry toxins, has the highest probability to form coiled-coil structures. The internal hydrophobic positions a and d of coiled-coils are important for protein-protein interactions with a matching ␣-helix, and mutagenesis of these positions affects the stability of this interaction (49). The presence of charged residues in these positions correlates with proper alignment, orientation, and selectivity of coiled-coils and contributes considerably to their stability (27). In addition, a salt bridge within the ␣-helix that forms the coiled-coil stabilizes this structure and is important in triggering the leucine zipper (50). Helix ␣-3 in Cry1A toxins contains an intrahelical salt bridge (51). These data in addition to the high conservation of positions a and d of helix ␣-3 in the three-domain Cry family suggest that this helix is highly stable and could play an important role in toxin oligomerization.
Previously, helix ␣-3 of a closely related toxin (Cry1Aa) was subjected to mutagenesis of charged residues, but none of the mutated residues corresponded to positions a or d of the predicted coiled-coil structure, and the authors did not analyze the capacity of these mutants to form the oligomeric structure nor the interaction of these mutants with the Bt-R 1 receptor (44). However, the authors concluded that most of the helix ␣-3 mutants were affected to some extent in their rate of pore formation (44). In the same work, substitutions in residue Arg-99 resulted in a complete loss of pore activity (44). All these data could be explained if the isolated mutants in helix ␣-3 were unable to form stable oligomers that are important for making pore channels. In other reports the Arg-93 and Ala-92 of Cry1Ac toxin (located at the beginning of helix ␣-3) were changed to different amino acids, only the conservative change R93K was fully active, some other changes of Arg-93 (Gly, Ala, and Ser) showed 100-to 1000-fold reduced toxicity, whereas all other changes resulted in complete loss of toxicity (52,53). The authors concluded that a positive charge in this position could participate in the formation of specific binding domain or that the conformation of this domain could indirectly disrupt the specificity domain of the toxin (52). The mutations A92E and A92D were also severely affected in toxicity and correlated with loss of pore formation (52,53). It was proposed that the irre-versible insertion of these mutants into the BBMV was affected (53). Residue Arg-93 is located in position a of the predicted coiled-coil structure of helix ␣-3 (Fig. 1), and as stated above this is a highly conserved amino acid within the Cry toxin family. Unfortunately, the authors neither analyzed oligomer formation nor pore activity of the Arg-93 mutants. We isolated a mutant R93F in this work, but due to the small crystal inclusion produced by this mutant we did not analyzed the pore formation or the oligomer formation.
In addition, it was suggested that helix ␣-5, located in the central position of domain I, was involved in Cry1Ac toxin oligomerization, because several point mutations in this helix disrupted oligomerization (54). The helix ␣-5 mutants affected in toxin oligomerization resulted in loss of toxicity against M. sexta with the exception of mutant H168R, which showed high insecticidal activity but could not form the 200-kDa oligomeric structure observed with wild-type Cry1Ac in SDS-PAGE (54). However, these authors also showed that this highly active mutant is able to form an oligomeric structure of Ͼ200 kDa that was also observed with the wild-type Cry1Ac toxin, probably a 250-kDa structure. Unfortunately, the pore-forming activities of the reported Cry1Ac mutants in helix ␣-5 were not analyzed, and it will be important to analyze their functionality to understand their role in toxin activity (work in progress).
As mentioned above, Zhang et al. (4) proposed that binding of monomeric Cry1A toxin with Bt-R 1 receptor is sufficient to induce insect death by a signal transduction mechanism (4). In this work, we identified single-point mutants in helix ␣-3 that severely affect in vitro pore formation and toxicity without affecting their ability to bind the receptor Bt-R 1 . It was previously demonstrated that the Cry1Ab oligomeric structure is highly stable, even after boiling and urea denaturation (11,40), and that purified Cry1Ab oligomer structure can be resolved as a SDS-resistant band in SDS-PAGE (11). Analysis of SDS resistance of the oligomeric structure produced by R99E and Y107E mutant toxins showed that these mutants produce a defective oligomeric structure, which is highly sensitive to temperature and impaired in pore formation and toxicity against M. sexta larvae. These results are similar to the dominant negative mutants isolated in another toxin, the anthrax protective antigen PA (47). These PA mutants produce unstable heptamers that lost SDS resistance, lost pore formation activity, and were non-toxic. In this case, the mutant proteins can be used as antitoxins that co-assemble with wild-type PA toxin producing defective heptameric structures that lost toxicity, indicating the importance of having stable protein structures to perform their function in vivo (47). Mutant R99E showed a similar inhibitory effect when mixed with the wild-type Cry1Ab, suggesting that helix ␣-3 plays an important role in oligomerization. Exhaustive mutagenesis of helix ␣-3 will help to determine the residues involved in pre-pore contacts and in dominant negative phenotype (work in progress).
The helix ␣-3 mutants bound a Bt-R 1 protein fragment with similar high nanomolar affinity as has been reported for Cry1Ab binding to the full-length Bt-R 1 receptor (55). This suggests that binding to Bt-R 1 receptor is not sufficient by itself to cause insect death and supports that oligomer and pore formation are two steps necessary to kill the larvae. In addition, we

Cry/Ab Oligomer Intermediate Is Required for Insect Toxicity
recently reported the characterization of scFvM22 antibody that competed with the binding of Cry1Ab oligomer to aminopeptidase-N receptor and inhibited the toxic effects of Cry1Ab, suggesting that the interaction of Cry1Ab oligomeric pre-pore with aminopeptidase-N is also important for toxicity (56).
We cannot, however, exclude the possibility that intracellular responses could have a role and synergize the effect of the toxin in insect larvae. With other pore-forming toxins, such as the ␣-toxin from Staphylococcus aureus and aerolysin produced by Aeromonas hydrophila, cell death is triggered by two mechanisms, pore formation and apoptosis, depending on the cell type and on the dose of toxin (57,58). Overexpression of antiapoptotic protein could block aerolysin-induced apoptosis, although this effect is overcome if higher toxin concentrations are used, where cells die quickly and apoptotic pathway is not triggered (58). In the case of aerolysin it was demonstrated that apoptosis is not directly triggered by binding of the toxin to its receptor, but rather it is caused by the production of a small number of channels in the membrane (58) indicating that the intracellular downstream effects are triggered by pore formation and membrane depolarization. The data presented in this work also indicate that pore formation induced by oligomeric toxin is required for Cry1A toxicity in vivo.