Hydropathic Complementarity Determines Interaction of Epitope 869HITDTNNK876 in Manduca sexta Bt-R1 Receptor with Loop 2 of Domain II ofBacillus thuringiensis Cry1A Toxins*

In susceptible insects, Cry toxin specificity correlates with receptor recognition. In previous work, we characterized an scFv antibody (scFv73) that inhibits binding of Cry1A toxins to cadherin-like receptor. The CDR3 region of scFv73 shared homology with an 8-amino acid epitope (869HITDTNNK876) of the Manduca sexta cadherin-like receptor Bt-R1 (Gomez, I., Oltean, D. I., Gill, S. S., Bravo, A., and Soberón, M. (2001) J. Biol. Chem. 276, 28906–28912). In this work, we show that the previous sequence of scFv73 CDR3 region was obtained from the noncoding DNA strand. However, most importantly, both scFv73 CDR3 amino acid sequences of the coding and noncoding DNA strands have similar binding capabilities to Cry1Ab toxin as Bt-R1 869HITDTNNK876 epitope, as demonstrated by the competition of scFv73 with binding to Cry1Ab with synthetic peptides with amino acid sequences corresponding to these regions. Using synthetic peptides corresponding to three exposed loop regions of domain II of Cry1Aa and Cry1Ab toxins, we found that loop 2 synthetic peptide competed with binding of scFv73 to Cry1A toxins in Western blot experiments. Also, loop 2 mutations that affect toxicity of Cry1Ab toxin are affected in scFv73 binding. Toxin overlay assays of Cry1A toxins to M. sexta brush border membrane proteins showed that loop 2 synthetic peptides competed with binding of Cry1A toxins to cadherin-like Bt-R1 receptor. These experiments identified loop 2 in domain II of as the cognate binding partner of Bt-R1 869HITDTNNK876. Finally, 10 amino acids from β-6-loop 2 region of Cry1Ab toxin (363SSTLYRRPFNI373) showed hydropathic pattern complementarity to a 10-amino acid region of Bt-R1(865NITIHITDTNN875), suggesting that binding of Cry1A toxins to Bt-R1 is determined by hydropathic complementarity and that the binding epitope of Bt-R1 may be larger than the one identified by amino acid sequence similarity to scFv73.

In this work, we show that the previous sequence of scFv73 CDR3 region was obtained from the noncoding DNA strand. However, most importantly, both scFv73 CDR3 amino acid sequences of the coding and noncoding DNA strands have similar binding capabilities to Cry1Ab toxin as Bt-R 1 869 HITDTNNK 876 epitope, as demonstrated by the competition of scFv73 with binding to Cry1Ab with synthetic peptides with amino acid sequences corresponding to these regions. Using synthetic peptides corresponding to three exposed loop regions of domain II of Cry1Aa and Cry1Ab toxins, we found that loop 2 synthetic peptide competed with binding of scFv73 to Cry1A toxins in Western blot experiments. Also, loop 2 mutations that affect toxicity of Cry1Ab toxin are affected in scFv73 binding. Toxin overlay assays of Cry1A toxins to M. sexta brush border membrane proteins showed that loop 2 synthetic peptides competed with binding of Cry1A toxins to cadherin-like Bt-R 1 receptor. These experiments identified loop 2 in domain II of as the cognate binding partner of Bt-R 1 869 HITDTNNK 876 . Finally, 10 amino acids from ␤-6loop 2 region of Cry1Ab toxin ( 363 SSTLYRRPFNI 373 ) showed hydropathic pattern complementarity to a 10amino acid region of Bt-R 1 ( 865 NITIHITDTNN 875 ), suggesting that binding of Cry1A toxins to Bt-R 1 is determined by hydropathic complementarity and that the binding epitope of Bt-R 1 may be larger than the one identified by amino acid sequence similarity to scFv73.
Bacillus thuringiensis (Bt) 1 is an aerobic, spore-forming bacteria that produces crystalline inclusions during the sporulation phase (1,2). These inclusions, which are toxic to larvae of several insects orders as well as to other invertebrates, are composed of proteins known as Cry toxins (1,2).
The inclusions are solubilized within the lepidopteran gut lumen to its highly alkaline pH and reducing conditions. Cry proteins are produced as protoxins that are activated by midgut proteases to release the toxin fragment (3). It is generally accepted that Cry toxins exert their pathological effect by forming lytic pores in the membrane of insect midgut epithelial cells (1,2).
The three-dimensional structures of Cry3A (coleopteran-specific), Cry1Aa (lepidopteran-specific) trypsin-activated toxins, and of Cry2A (dipteran-specific) protoxin have been resolved by x-ray diffraction crystallography (4 -6). The three proteins share many similar features and are comprised of three domains. In particular, Cry1Aa and Cry3A structures are more similar and have the following characteristics. The N-terminal domain I, a seven-␣-helix bundle in which helix ␣-5 is encircled by the other helices, is the pore-forming domain. Domain II consists of three anti-parallel ␤-sheets with exposed loop regions, and domain III is a ␤-sandwich (4,5). Domains II and III are involved in receptor binding (for reviews, see Refs. 1, 2, and 7).
The activated Cry toxin binds specifically to its receptors located on the midgut epithelium. In susceptible insects, Cry toxin specificity correlates with receptor recognition (8,9). The identification of epitopes involved in Cry toxin-receptor interactions could provide insights into the mechanism of insect specificity and the mode of action of these toxins. Also, this knowledge could offer tools for improving the specificity and toxicity of Cry toxins. Two Cry1A toxin receptors from various lepidopteran insects have been identified as aminopeptidase N (APN) and cadherin-like proteins (Bt-R 1 , Bt-R 175 ) (10 -16). In Lymantria dispar, besides APN and cadherin-like receptors, a high molecular weight anionic protein (Bt-R 270 ) that binds Cry1A toxins with high affinity was identified (17). However, an on-going debate is whether these are functional receptors leading to toxin pore formation activity. In contrast to APN, expression of the cadherin-like protein from Bombyx mori, Bt-R 175 , on the surface of Sf9 insect cells made these cells sensitive to Cry1Aa toxin (13). Also, a Heliothis virescens population resistant to Cry1Ac toxin contained a mutation in a cadherinlike coding gene (18). Moreover, in previous work we identified a scFv antibody (scFv73) that inhibited binding of Cry1A toxins to cadherin-like receptors, but not to APN, and reduced the toxicity of Cry1Ab to Manduca sexta larvae. Interestingly the CDR3 region of scFv73 shared homology with an eight-amino acid epitope of M. sexta cadherin-like receptor, Bt-R 1 , involved in Cry1A interaction (19). Evidence was obtained that showed Cry1A toxin binding to this cadherin epitope facilitates proteolytic cleavage of helix ␣-1 in domain I and formation of a tetramer oligomer pre-pore that is insertion-competent (20). Overall, these results suggest that binding to cadherin-like receptor is an important step in the mode of action Cry1A toxins.
Site-directed mutagenesis studies of Cry1A toxins revealed that loop ␣-8, located in the junction of domains I and II, and loop 2 and loop 3 regions of domain II are involved in receptor recognition and toxicity (21)(22)(23)(24)(25)(26). Interestingly enough, Cry1Aa and Cry1Ab toxins have different loop 2 and loop 3 amino acid sequences despite the fact that they interact with the same M. sexta receptors, APN and Bt-R 1 (10 -12, 24, 27, 28).
In regard to the receptor binding epitopes, in B. mori APN, a region of 63 residues involved in Cry1Aa binding was identified. This site was specific for Cry1Aa toxin since it was not involved in Cry1Ac binding (29). In cadherin-like receptors, a region located between residues 1245 and 1391 of Bt-R 175 was identified as a binding region important for Cry1Aa interaction (13). We identified an 8-amino acid epitope in Bt-R 1 ( 869 HIT-DTNNK 876 ) and 2 amino acid epitopes in B. mori Bt-R 175 ( 873 IIDTNNK 880 and 1296 LDETTN 1301 ) involved in binding of Cry1A toxins (19). To study the mechanism of receptor interaction it is important to identify the Cry1A cognate binding partner for Bt-R 1 869 HITDTNNK 876 epitope. The molecular basis of protein-protein interactions remains largely unknown, although accumulating evidence indicates that proteins can interact through amino acid sequences displaying inverse hydropathic profiles leading to the concept of hydropathic complementarity (30). One model for interacting complementary structures postulates secondary structures, ␤-strands and ␣-helices, in which the hydrophilic surfaces are oriented toward the aqueous phases, whereas hydrophobic surfaces face each other (31), and another model suggests complementary surface contour (32). This concept has been crucial for the molecular recognition theory that proposes that peptides whose sequences are obtained from the noncoding DNA strands likely bind the amino acid sequence of the coding strand since they have inverted hydropathic patterns (30,33). Among more than 40 examples (for review, see Ref. 34) hydropathic complementarity has been successfully applied to produce biologically active synthetic analogs of receptor binding sites (33,35,36) and ligands (37) and to map binding epitopes of natural ligands with their receptors (38,39). Also, hydropathic complementarity has been shown to determine several peptide-antibody interactions (40). However, there are few examples of naturally occurring peptides or proteins whose similarity in binding properties is correlated with similar hydropathic patterns.
In this work, we demonstrate that the scFv73 CDR3 region interacts with the same epitope in Cry1Ab toxin as the 869 HITDTNNK 876 epitope of Bt-R 1 . Also, we identified domain II loop 2 of Cry1A toxins as the cognate binding partner of the Bt-R 1 869 HITDTNNK 876 epitope. Finally, hydropathic profiles analyses of the loop 2 regions in the toxins and of the cadherin-like receptor Bt-R 1 binding epitope revealed that hydropathic complementarity could account for the interaction of Cry1A toxins with their cadherin-like receptors and that the binding epitope in Bt-R 1 is likely to include residues 865 NITIHITDTNN 875 .
Purification of Cry1A Toxins-The Cry1Aa and Cry1Ab crystals produced by Bt strains were isolated, trypsin-activated, and purified by Q-Sepharose as described (19,43,44). Mutant F371A was expressed in E. coli and purified as previously described (45).
Purification and Characterization of scFv73-scFv73 antibody was purified from E. coli cells to homogeneity by a nickel-agarose column as described (19).
Western Blotting of Cry1Ab with scFv73-Trypsin-activated Cry1Aa or Cry1Ab toxin was separated in 9% SDS-PAGE, transferred onto a nitrocellulose membrane polyvinylidene difluoride, and blocked with skim milk (5%). The membranes were then incubated in 200 nM scFv73 antibody followed by anti-c-Myc antibody (Sigma) (1:5000 dilution) and then a secondary goat anti-mouse antibody conjugated with peroxidase (Sigma) (1:5000 dilution). Blots were visualized using luminol (ECL, Amersham Biosciences). Amino acid sequences of synthetic peptides used for competition experiments are shown in Table I. Quantification of competition with synthetic peptides was determined by scanning the optical density of bands in blots. Fig. 2 blots were replicated at least twice, and representative results are shown. Data points of Fig. 3 are the means of three replicates, and error deviations are shown.
Preparation of Brush Border Membrane Vesicles (BBMV)-M. sexta were reared on an artificial diet from eggs kindly supplied by Dr. Jorge Ibarra (Centro de Investigación y de Estudios Avanzados, Irapuato, Mexico). BBMV from fifth instar M. sexta larvae were prepared as reported (19,46).
Toxin Overlay Assays-Toxins were biotinylated using biotinyl-Nhydroxysuccinimide ester (Amersham Biosciences) according to the manufacturer's instructions. Protein blot analysis of BBMV preparations was performed as described previously (19,44). To determine the ability of peptides to compete with the Cry1A toxin, different concentrations of the peptides were incubated with biotinylated Cry1A toxins in washing buffer (0.1% Tween 20, 0.2% bovine albumin in phosphatebuffered saline) for 1 h at room temperature before adding the mixture to nitrocellulose membranes. Fig. 4 blots were replicated at least twice, and representative results are shown.
Biosensor Analysis of scFv73 Affinities to Cry1A-All surface plasmon resonance (SPR) measurements was performed using a Biacore X and CM5 sensor chips (Biacore) as described (19). HBS-P buffer (10 mM HEPES, pH 7.4, 0.15 M NaCl, 3 mM EDTA, 0.05% surfactant P20) was used throughout the analyses. Briefly, the 30-kDa ligand, scFv73 in 20 mM ammonium acetate pH 5 buffer, was immobilized on flow cell 2, giving densities of less than 150 relative units. The experimental and control surfaces were then blocked with a 5-min injection of 1 M ethanolamine. Cry1Ab wild type and the F371A mutant (65 kDa, apparently homogenous based on SDS-PAGE) were injected over both flow cells at a flow rate of 30 l/min. Association and dissociation were monitored for 120 and 180 s, respectively, before the surfaces were regenerated (47). The data obtained from triplicate injections of each toxin concentration (15-750 nM) over both surfaces were corrected by double referencing (47) and fitted using global analysis software available within Biaevaluation 3.1 (Biacore). A variety of controls were done as previously described (19).
Docking Construction-Using coordinates for the insecticidal toxin Cry1Aa (PDB code 1CIY) as template, an initial three-dimensional model for Cry1Ab was constructed using module homology from Insight II (Biosym/MSI). Subsequently, energy of the model was minimized using the CNS program (48). A fragment of Bt-R 1 (residues 864 -880) was used to perform a Fasta search using the PDB data base. The resulting highest score fragment (residues 153-169 from ADP ribosyl cyclase, PDB code 1LBE) was used as template to structure the Bt-R 1 fragment using Insight II. Hydrophobic docking was performed using the GRAMM program by applying the recommended parameters (49). The first 100 Bt-R 1 fragment positions were explored selecting those that bind close to loop 2 of domain II of Cry1Ab model. The positions of selected results were adjusted by performing an annealing procedure (CNS), fixing the position of Cry1Ab atoms, and adjusting the position of Bt-R 1 fragment atoms.
Hydropathic Pattern Determination-Hydropathic profiles based on the Kyte-Doolitle algorithm were calculated using Hypscan software kindly provided by Dr. J. E. Blalock (University of Alabama, Birmingham, AL).

scFv73 CDR3 and Bt-R 1 Epitope 869 HITDTNNK 876 Bind the Same Region in Cry1Ab
Toxin-To generate a strategy to mutate CDR3 of scFv73, both DNA strands of the CDR3 region were sequenced since the previous sequence was short and corresponded to only one DNA strand. The new sequence revealed that the previous CDR3 amino acid sequence (RITQTTNR) was obtained from the noncoding DNA strand (5Ј to 3Ј) (Fig. 1). The amino acid sequence of the CDR3 coding strand (TVGSLSNS) shares no identity with the sequence of the binding epitope in Bt-R 1 869 HITDTNNK 876 (named here BtR 1 -Cry1A). Previously, we demonstrated that synthetic peptides corresponding to the epitope BtR 1 -Cry1A or to the amino acid sequence of the noncoding scFv73 CDR3 peptide (CDR non-cod, Table I) competed with binding of Cry1Aa and Cry1Ab toxins to Bt-R 1 (19). To determine whether scFv73 binds to the same regions in Cry1Ab toxin as epitope BtR 1 -Cry1A, competition assays of scFv73 binding to Cry1Ab with synthetic peptides corresponding to scFv73 CDR3 (coding and noncoding) and to the BtR 1 -Cry1A epitope were performed. Fig. 2A shows a Western blot of Cry1Ab toxin detected with scFv73, indicating that peptides homologous to scFv73 coding or noncoding CDR3 and BtR 1 -Cry1A epitope competed with binding of scFv73 to Cry1Ab. A non-relevant peptide (scramble of scFv73 CDR3 sequence) did not compete with binding of scFv73 to Cry1Ab ( Fig. 2A). This result shows that the coding and noncoding CDR3 amino acid sequences bind to the same epitope in Cry1Ab toxin as the BtR 1 -Cry1A binding epitope.
scFv73 Binds to Loop 2 of Domain II of Cry1A Toxins-Previously we demonstrated that scFv73 binds to domain II of Cry1Ab toxin (19). Site-directed mutagenesis of Cry1A domain II sequences demonstrated that the exposed loops of this domain are involved in receptor interaction (21)(22)(23)(24)(25)(26). To map the Cry1A cognate binding epitope of BtR 1 -Cry1A, synthetic peptides corresponding to the three loops of Cry1A toxins were synthesized and used to compete with binding of scFv73 to Cry1A toxins in Western blots. Fig. 2, B and C, shows that loop 2 peptides of Cry1Aa and Cry1Ab toxins competed with binding of scFv73 to their corresponding Cry1A toxin in contrast to peptides corresponding to loop 1 and loop 3. Moreover, loop 2 peptide of Cry1Ab competed the binding of scFv73 to Cry1Aa although in a lesser extent (Fig. 2C).

Cry1Ab Loop 2 Amino Acids Involved in Receptor Binding
and Toxicity Affect scFv73 Binding-The biological significance of the binding of Cry 1A loop 2 to scFv73 was addressed by the use of synthetic peptide representing known mutants of Cry1Ab toxin to determine whether mutant loop 2 peptides binding to scFv73 correlated with the characteristic of known mutants in terms of binding and toxicity. Cry1Ab loop 2 F371A and double mutant R368E/R9E showed decreased M. sexta toxicity and binding to BBMV (22,26). Synthetic peptides containing F371A and R368E/R9E double mutations were synthesize to determine their ability to compete with scFv73 binding to Cry1Ab toxin. Optical density of Cry1Ab toxin bands in Western blots was scanned to quantify the degree of binding in the presence of different concentrations of peptides used as competitors. Synthetic loop 2 peptides containing the R368E/ R9E or F371A mutations did not compete with scFv73 binding to Cry1Ab toxin in contrast to the wild-type loop 2 synthetic peptide, L2Ab (Fig. 3). To determine whether scFv73 binds the Cry1Ab F371A mutant, Western blots of F371A mutant detected with scFv73 were performed that showed that scFv73 binds Cry1Ab F371A mutant (data not shown). However, lower concentrations of the wild-type loop 2 peptide are required to compete with binding of scFv73 to F371A toxin than are needed to compete with scFv73 binding to wild-type Cry1Ab (Fig. 3). These data suggest that scFv73 binds F371A mutant with lower affinity than Cry1Ab toxin.
The binding affinity of Cry1A toxins to scFv73 was in the range of 20 -50 nM (19). To determine the effect of the loop 2 mutation F371A in the interaction with scFv73, we performed real time binding kinetics by SPR. Fig. 4 show sensograms of Cry1Ab and F371A mutant binding to scFv73. Cry1Ab F371A has a lower affinity compared with wild-type Cry1Ab toxin, 114 versus 55.8 nM (Table II). The apparent lower affinity, 114 nM, was due to differences in the association as well as the dissociation rates (Fig. 4). This result shows that F371A moderately affects the interaction of Cry1Ab with scFv73.
Loop 2 Peptides Compete Binding of Cry1Ab Toxin to Bt-R 1 -We showed scFv73 and the BtR 1 -Cry1A epitope bind to the same Cry1Ab epitope ( Fig. 2A) and that this region corresponds   (Fig. 2B). To corroborate these data, we performed toxin overlay assays to determine whether Cry1Aa and Cry1Ab loop 2 peptides could compete with the interaction of Cry1A toxins to Bt-R 1 . Fig. 5 shows the Cry1Aa and Cry1Ab toxins bind APN (120 kDa) and Bt-R 1 (210 kDa) of M. sexta BBMV as previously reported (10 -12, 19). Competition with synthetic peptides corresponding to loop regions of these toxins showed that synthetic peptides corresponding to loop 2 of Cry1Aa and Cry1Ab competed with binding of their corresponding toxin to Bt-R 1 and not to APN. Interestingly, whereas the loop 3 peptide of Cry1Aa toxin also competed with binding of Cry1Aa toxin to Bt-R 1 , the Cry1Ab toxin loop 3 peptide did not compete. Neither loop 1 peptide had an effect on binding of the corresponding toxin to Bt-R 1 (Fig. 5).
Inverse Hydropathy Determines Interaction of Cry1A Loop 2 Sequences with Bt-R 1 -As shown above, BtR 1 -Cry1A epitope binds to Cry1Aa and Cry1Ab toxins at loop 2 despite the fact that these loop 2 epitopes share low amino acid identity. To determine whether the interaction of loop 2 regions with BtR 1 -Cry1A epitope involves inverted hydropathic patterns, the hydropathic profiles of loop 2 Cry1A epitopes were determined as well as those of Bt-R 1 -Cry1A and the CDR3 regions. The patterns were compared for similarity or complementarity between them using the computer program Hypscan as indicated under "Materials and Methods." Fig. 6A shows that the hydropathic profile of the scFv73 coding and noncoding CDR3 amino acid sequences and that of BtR 1 -Cry1A are very similar. Fig. 6, B and C, shows Cry1Aa and Cry1Ab loop 2 regions have inverted hydropathic profiles relatives to the Bt-R 1 epitope, with Cry1Ab loop 2 more complementary to the BtR 1 -Cry1A epitope than the Cry1Aa loop 2 region. Regions of Cry1A toxins that showed inverse hydropathic patterns to the BtR 1 -Cry1A epitope included five residues of ␤-6 and five residues of loop 2 in Cry1Ab 363 SSTLYRRPFNI 373 , whereas Bt-R 1 included residues 865 NITIHITDTNN 875 . The amino acid sequence identity between the Cry1Aa and Cry1Ab regions that have inverted hydropathic complementarity to Bt-R 1 is 60%, where 5 of 6 identical residues are from the ␤-6 structure and one from the loop 2 region. DISCUSSION Understanding the molecular basis of Cry toxin specificity will help in the rational design of improved toxin formulations useful in insect pest management. The identification of epitopes involved in Cry toxin-receptor interaction could pro- k on is the association rate constant; k off is the dissociation rate constant and K d is the apparent affinity (k on /k off ). Values are representative of triplicate injections that were analyzed globally. vide insights into the mechanism of insect specificity and the mode of action of these toxins. In this work we identified loop 2 of domain II of Cry1A toxins as the cognate binding epitope of the M. sexta receptor Bt-R 1 869 HITDTNNK 876 . This finding highlights the importance of BtR 1 -Cry1A binding epitope since extensive mutagenesis of loop 2 of Cry1A toxins has shown that this loop region is important for receptor binding and toxicity (22,23). Besides loop 2, loop ␣-8 and loop 3 of Cry1A toxins are also important for receptor interaction and toxicity (21, 24 -26). The Bt-R 1 epitopes involved in binding loop ␣-8 and loop 3 regions still remains to be identified. In this work we analyzed the role of the three exposed loop regions of domain II on receptor interaction. The role of loop ␣-8 on receptor Bt-R 1 interaction still remains to be analyzed since mutagenesis of this region has been shown to have an important role on receptor interaction (21). Loop 2 peptides of Cry1Aa and Cry1Ab toxins competed with binding of both toxins to Bt-R 1 in toxin overlay assays (Fig. 5). However, only the peptide corresponding to loop 3 of Cry1Aa but not that of Cry1Ab competed with binding of the corresponding Cry1A toxin to Bt-R 1 (Fig. 5). Cry1Aa loop 3 shares no amino acid sequence identity with Cry1Ab or Cry1Ac loop 3 regions (24). Therefore, our results suggest that loops 2 and 3 of Cry1Ab and Cry1Aa toxins contribute differentially to the binding of these toxins to Bt-R 1 , with Cry1Aa loop 3 more important in the interaction with Bt-R 1 receptor than Cry1Ab loop 3. However, we cannot rule out the possibility that Cry1Ab loop 3 has structural constraints for receptor interaction, therefore explaining the lack of competition by the Cry1Ab loop 3 peptide in Cry1Ab binding to Bt-R 1 (Fig. 5).
CDR3 of scFv73 exhibits similar binding properties as Bt-R 1 since scFv73 binding to Cry1Ab toxin is affected by loop 2 mutations that also affect Cry1Ab receptor binding (Fig. 3). In addition, by using scFv73 as a surrogate for Bt-R 1 , we found that binding of Cry1A toxins to this epitope facilitates proteolytic cleavage of helix ␣-1 of domain I and the formation of a tetramer oligomer pre-pore structure, showing that scFv73 has functional activity similar to that of the natural Bt-R 1 receptor (20).
Prior SPR binding analysis showed the binding of the Cry1Ab loop 2 F371A mutant to APN is barely affected (50), and this mutant has a 2-fold lower affinity for Bt-R 1 as judged by binding to scFv73 (Table II, Fig. 4). However, the irreversible binding and toxicity is affected (22,23). As mentioned above, binding to Bt-R 1 facilitates proteolytic cleavage of helix ␣-1 and the formation of a pre-pore structure that is membrane insertion-competent, probably by inducing a conformational change that makes helix ␣-1 accessible for proteolytic degradation (20). Therefore, it is possible that binding of Cry1Ab F371A mutant to Bt-R 1 does not promote the conformational change that facilitates helix ␣-1 degradation, thus affecting the formation of the pre-pore structure. This hypothesis could explain the effect of the F371A mutation on irreversible toxin binding since it is postulated that formation of the pre-pore is a prerequisite for membrane insertion (20). Analysis of pre-pore formation by the Cry1Ab F371A mutant induced by scFv73 binding will enable us to determine whether this step is affected (work in progress).
Competition of Cry1Aa and Cry1Ab toxins to APN and Bt-R 1 in toxin overlay assays in the presence of loop 2 synthetic peptides showed that loop 2 is important for interaction of the Cry1A toxin to Bt-R 1 but not for APN (Fig. 5). These data support the fact that Bt-R 1 has an important role in insect toxicity. However, we cannot rule out the possibility that the APN receptor has a role in Cry1Ab toxicity since several loop 2 point mutations affect both APN binding and toxicity (26,50).
The fact that some Cry1A loop 2 mutations affect APN binding suggests that a similar binding epitope as BtR 1 -Cry1A could be present in APN. However, the loop 2 peptides did not compete with Cry1Ab binding to APN. These data are in apparent contradiction. An important methodological difference is that SPR binding studies (26,50) determined binding of native proteins in contrast to toxin overlay assays, which determine binding to denatured proteins after SDS-PAGE electrophoresis. Thus, structural differences could account for the lack of competition of loop 2 peptide on Cry1A toxins binding to APN observed in this work.
Accumulating evidence indicates that proteins can interact through amino acid sequences displaying inverted hydropathic profiles, implying that amino acid sequences that share low amino acid sequence identity can interact with the same epitopes if they share a similar hydropathic profile (30). Analysis of hydropathic patterns of Cry1A loop 2 regions and that of BtR 1 -Cry1A epitope showed that the interaction of these regions is determined by inverse hydropathic patterns (Fig. 6). Hydropathic profiles of amino acid sequences of scFv73 CDR3 and BtR 1 -Cry1A epitope were similar (Fig. 6A) and inverted to a region of Cry1A that corresponds to the end of ␤-6 and loop 2 (Fig. 6, B and C). If we analyze the hydropathic pattern of the amino acid region adjacent to the Bt-R 1 -Cry1A epitope, we find an inverse pattern with Cry1Ab region including five residues of ␤-6 and five residues of loop 2 in both toxins. This analysis predicts that mutations of Cry1A ␤-6 residues 363 SSTLY 367 would affect receptor binding and toxicity (work in progress). The Bt-R 1 included residues 865 NITIHITDTNN 875 , suggesting that the binding epitope could be larger than the one previously mapped based on amino acid sequence identity with scFv73 CDR3 (19). Using a computer-simulated docking method, we analyzed the binding of a peptide corresponding to Bt-R 1 864 GNITIHITDTNNKVPQAE 880 to Cry1Ab. We observed that FIG. 7. Molecular surface model of Cry1Ab toxin. Exposed surfaces of helix ␣-8 (residues 275-293 (blue); Arg-281 (deep blue)), loop 1 (residues 310 -314 (violet)), loop 2 (368 -375, red), residues R-368 -R-369 and Phe-371 (orange)), ␤-6 (360 -367 (green)), and loop 3 (residues 438 -443 (yellow)). The lowest potential energy docking of Bt-R 1 864 -880 peptide to Cry1Ab ␤-6-loop 2 region (rank 70) is shown. the binding of this Bt-R 1 amino acid sequence to ␤-6-loop 2 region was favorable in steric and energy-pairing analyses among several other favorable binding possibilities obtained (Fig. 7). This result provides support to this region of Bt-R 1 interacting with ␤-6-loop 2 of Cry1A toxins. The inverted hydropathic pattern of Cry1Ab loop 2 sequences with that of Bt-R 1 -Cry1A was more complementary than that of loop 2 of Cry1Aa (Fig. 6, B and C); these data correlates with a slightly lower affinity of Cry1Aa to Bt-R 1 compared with the binding affinities of Cry1Ab and Cry1Ac toxins (51). The finding that inverse hydropathic patterns could determine the interaction of loop 2 of Cry1A toxins with Bt-R 1 could be used in the rational design of more toxic Cry1A mutant proteins by optimizing the profile of ␤-6 loop 2 to that of the epitope in the receptor. Mutants with optimized inverted hydropathic patterns are predicted to have improved affinity toward its receptor as has been shown for other examples of interacting proteins (34).
In this study we show that amino acid sequences obtained from the coding and noncoding DNA strands of scFv73 CDR3 bind to loop 2 of Cry1A toxins. The molecular recognition theory predicts that peptides obtained from noncoding DNA strands (anti-peptides) bind to amino acid sequence epitopes of the coding strand (30,52). In the case of scFv73 CDR3, the amino acid sequence of the noncoding strand has similar binding properties as the coding DNA strand CDR3 amino acid sequence ( Fig. 2A). There are other examples of anti-peptides that have similar binding properties as the coded amino acid sequence in the literature (35,51). The analysis of hydropathic profiles of the coding and noncoding scFv73 CDR3 amino acid sequences revealed that they share a similar hydropathic pattern (Fig. 6A), explaining their ability to interact with the same epitope in the toxin even though they share no sequence identity. The noncoding amino acid sequence of scFv73 CDR3 (RITQTTNR) shares amino acid sequence identity with Bt-R 1 869 HITDTNNK 876 epitope (19). Isolation of peptides that mimic natural ligands by phage display have been successful in mapping protein epitopes by searching amino acid sequence similarities (53). Nevertheless, as was pointed out before (39), the most reliable way to map binding epitopes is to search hydropathic profile similarities rather than amino acid sequences similarities. We propose that amino acid sequence similarities could also be searched considering both amino acid sequences of the coding and noncoding DNA strands of the interacting epitopes.