Characterization of a high-affinity binding site for the pea albumin 1b entomotoxin in the weevil Sitophilus Fre ´ de ´ ric Gressent, Isabelle Rahioui and Yvan Rahbe ´ UMR 0203 INRA/INSA de Lyon, BF2I (Biologie Fonctionnelle, Insectes et Interactions), INSA Ba ˆ timent Louis Pasteur, Villeurbanne, France The toxicity of the pea albumin 1b (PA1b), a 37 amino-acid peptide extracted from pea seeds, for cereal weevils (Sito- philus oryzae, Sitophilus granarius and Sitophilus zeamais) was recently discovered. The mechanism of action of this new entomotoxin is still unknown and potentially involves a target protein in the insect tissues. This work describes the characterization of a high-affinity binding site for PA1b in a microsomal fraction of Sitophilus spp. extracts. Purified PA1b was labeled to a high specific radioactivity (c. 900 CiÆmmol )1 )using 125 I, and the iodinated ligand was found to be biologically active. Binding of this ligand to the microsomal fraction of S. oryzae extract was found to be saturable and reversible, with an affinity (K d )of2.6n M ,and a high maximal binding capacity (B max )of40pmolÆmg )1 of protein. A binding site displaying similar characteristics was detectable in the five susceptible weevils strains tested, as well as in the pea aphid or in the fruit fly. However, no binding activity was detectable in extracts from four S. oryzae strains previously shown to be resistant to the toxin through a recessive monogenic mechanism. Therefore, we suggest that this binding site might be involved in the mechanism of action of PA1b. Keywords: binding site; pea albumin 1b; Sitophilus;cystine- knot; leginsulin. The cereal weevils (Sitophilus oryzae, Sitophilus granarius and Sitophilus zeamais) are major pests of stored grains. At present, the use of chemical insecticides is the main answer to the damage caused by stored product pests, inducing ecotoxicity problems and the occurrence of resistance within insect populations. An alternative for controlling insects is physical methods, such as cooling or the use of nitrogen or carbon dioxide atmosphere. However, these methods are expensive and not always applicable. Plant protection also takes advantage of the genetic resources residing in crop plants to create resistant varieties, and of the use of biological toxins to build up transgenic plants tolerant to different insect species. The Bacillus thuringiensis toxins [1], proteins belonging to the lectin family [2], or enzymes inhibitors [3], are in use or being tested, but to date these molecules are not adapted to grain protection against weevils. Therefore, the recent finding of a plant peptide lethal for these insects has enlarged the possibilities for cereal grain protection [4]. This peptide was purified and sequenced from Pisum sativum seeds and, being previously known as a seed albumin [5], it was named PA1b for pea albumin 1b. PA1b is the result of the post-translational cleavage of the albumin proprotein PA1, also releasing a second peptide (PA1a). PA1b consists of 37 amino acids, with six cysteines involved in three disulfide bonds which confer the toxin its high stability. Biological activity is conserved when boiled, and the peptide is resistant to digestion by trypsin [6] or by bovine rumen fluid [7]. Although no primary sequence homologies could be found with other proteins in the available databases, it was recently established by NMR studies and molecular modeling that PA1b belongs to the cystine-knot family [8]. Characterized only recently as a structural family [9], cystine-knot peptides have now been identified from a variety of sources (plants, fungi, animal venoms, insects), and show very diverse biological activities (elicitor AVR9, trypsin inhibitor, antimicrobial, antiviral or antifungal activities, and many channel blockers [10]). A subfamily of these, the cyclic cystine-knot peptides, also named cyclo- tides, were investigated for their potential as new anti- biotics or anti-HIV properties. PA1b was the first entomotoxic cystine-knot peptide identified (peptidic sequence and disulfide bridges are shown in Fig. 1), and the interest of this structural family for insect targets was reinforced by identifying the cyclotide kalata B1 as a peptide active on larval growth of the Lepidoptera Helicoverpa punctigera [11]. Moreover, PA1b might belong to a multigenic family, as at least five isoforms of the peptide exist in a single pea genotype and it seems to be widespread in legumes [4,5]. Many cystine-knot peptides have inhibitory effects, against channels, proteases, or a-amylase [10]. However, to date, in vitro assaysusingPA1bfailedtorevealany Correspondence to F. Gressent, UMR 0203 INRA/INSA de Lyon, BF2I (Biologie Fonctionnelle, Insectes et Interactions), INSA Baˆ timent Louis Pasteur, 69621 Villeurbanne Cedex, France. Fax: + 33 4 72438534, Tel.:+ 33 4 72437982, E-mail: gressent@jouy.inra.fr Abbreviations: PA1b, pea albumin 1b; BBI, soybean Bowman–Birk inhibitor; E-64, trans-epoxysuccinyl- L -leucylamido(4-guanidino)- butane. (Received 21 February 2003, revised 24 March 2003, accepted 8 April 2003) Eur. J. Biochem. 270, 2429–2435 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03611.x enzyme inhibitory effect [4]. Therefore, the aim of the present work was to investigate the mode of action of the toxin, as part of a wider study to determine whether PA1b could be a valuable tool for insect control. An important point is to understand the molecular mechanisms of the sensitivity to the toxin, since resistant strains do exist naturally: the screening of up to 90 Sitophilus spp. strains for the susceptibility to PA1b has revealed that three strains, all belonging to the S. oryzae species, contained individuals able to feed on pea seeds [12] and resistant to the purified toxin. Other pest insects (the pea aphid Acyrthosiphon pisum and the flour moth Ephestia kuehniella) were found to be susceptible to the toxin, but the fruit fly Drosophila melano- gaster was shown to be resistant [4]. Within the S. oryzae species, a genetic analysis of the resistance demonstrated that this trait was driven by a single recessive autosomal gene [12]. This result suggested that a single weevil gene product could be responsible for the susceptibility or resistance towards the toxin. On this basis, the mechanism of action of the peptide was likely to involve a discrete molecular target at the insect side. Therefore our work focused on the characterization of a PA1b binding site in Sitophilus spp. extracts, using a 125 I labeled 3741 Da isoform of the toxin (peptide sequence in Fig. 1 [4]). Given the biological features presented here, this binding site could potentially be regarded as the molecular target of the PA1b toxicity in weevils. Experimental procedures Biological material Cereal weevils (S. oryzae, zeamais and granarius, Coleoptera, Curculionidae) were reared on wheat seeds at 27.5 °C 70% relative humidity. Nine strains were used, differing in their genetic ability to thrive on pea seeds and resist the toxic activity of pea albumin PA1b: five susceptible strains ÔS. oryzae WAA42, Benin, and Bouriz, S. zeamais LS, and S. granarius BrayardÕ and four fully resistant S. oryzae strains ÔISOR3, Mex1, China and GVÕ harboring a recessive pea-resistance allele [12]. The fruit fly D. mel- anogaster (Diptera)larvaewereprovidedbyR.Allemand (UMR 5558 CNRS-UCBL, Lyon, France). Tests of toxicity The toxicity assays for PA1b or modified PA1b were performed on the S. oryzae susceptible strain WAA42 and on its partially isogenic resistant strain ISOR3 (three backcross of a resistant ÔChinaÕ genotype over the suscepti- ble WAA42 genotype), according to the protocol described by Delobel et al. [4]. Briefly, in all described experiments, 30 insects were fed with the toxin to be assayed (100 lgÆg )1 of food), and mortality was then monitored daily up to 14 days after the beginning of the treatment. Isolation of membranes All the operations were performed at 4 °C. Insects (0.5–1 g) were ground with a mortar and pestle in liquid nitrogen. The resulting powder was resuspended in 10–20 mL of extraction buffer (20 m M Tris/HCl pH 8; 0.25 M sucrose; 2m M MgCl 2 ;10l M E-64), and extracted six times for 5 s using an UltraTurrax blender. The slurry was centrifuged for 10 min at 3000 g. The supernatant was further centri- fuged to give a second pellet at 10 000 g (10 min) and a third one at 45 000 g for 45 min (microsomal fraction). The resulting fractions, sedimenting at 3000, 10 000 or 45 000 g, were resuspended in binding buffer (20 m M Tris buffer pH 7.0 containing 10 l M E-64) with 30% glycerol and stored at )80 °C until use. Purification of the toxin One batch of purified toxin isoform, showing a molecular mass of 3741 Da by mass spectrometry, and a pea albumin extract (named SRA1) were provided by J. Gueguen and E. Ferrasson (Laboratoire de Biochimie et Technologie des Prote ´ ines, Nantes, France). From SRA1, a mixture of PA1b isoforms was obtained by a 45-min incubation at )20 °Cin acetone/water (80 : 20) followed by a 10-min centrifugation at 12 000 g. The supernatant was dried under vacuum, and more than 95% of the resulting powder consisted of a mixture of PA1b isoforms (as checked by HPLC analysis). PA1b or modified PA1b were purified or analyzed by reverse phase C 18 HPLC column eluted at 1 mLÆmin )1 with a gradient of water; trifluoroacetic acid 0.1%/acetonitrile; trifluoroacetic acid 0.1% (80 : 20 for 2 min, then to 40 : 60 in 20 min). PA1b peptide isoforms were detected by their absorbance at 210 nm (DAD 440, KONTRON, France), quantified by the measurement of the peak area and referred to known quantities of pure peptides used as standards. Labeling of the toxin The labeling was carried out in 37 lL of 400 m M Tris/HCl buffer pH 7.5, 60% methanol with 1.66 lg of the purified 3741 Da PA1b and 1 mCi of carrier free Na 125 I(Amer- sham). Chloramine-T (1.2 lg) in 3 lLof50m M Tris/HCl buffer pH 7.5, 60% methanol was added at t ¼ 0, 1, 2, and 3 min. After 10 min of incubation the reaction was stopped by adding 10 lgoftyrosinein140lLofwater.The radiolabeled peptide was separated from non-incorporated iodine by reverse-phase HPLC. In these conditions, incor- poration of iodine ranged from 40 to 50%, and the specific radioactivity of the 125 I-labelled PA1b ligand on labeling day ranged from 890 to 1120 CiÆmmol )1 . The labeled toxin was stored in 60% methanol at )20 °C and used within a month after labeling. For iodination of the peptide using nonlabeled iodine, the protocol was as described above, except that 200 lgofthe 3741 Da isoform, 0.3 mg of nonlabeled NaI and four times 50 lg of chloramine-T were used in a 100-lL volume. Reduction and alkylation of the toxin Reduction of the 3741 Da toxin (2 mgÆmL )1 ) was carried out in Tris/HCl buffer (pH 8, 0.1 M ) in the presence of Fig. 1. Peptide sequence [4] and disulfide bridges [8] of the 3741 Da PA1b isoform. 2430 F. Gressent et al. (Eur. J. Biochem. 270) Ó FEBS 2003 20 m M dithiothreitol. The mixture was boiled for 10 min, allowing a complete reduction of the peptide. After cooling at room temperature, one volume of 0.2 M iodoacetamide was added and the mixture was incubated in the dark for 1 h. Alkylation of the cysteine residues was quantitative, and the alkylated peptide was then purified by reverse-phase HPLC. Binding assays Microsomal proteins (1–3 lg) were incubated in the pres- ence of 0.4 n M 125 I-labelled PA1b in binding buffer supple- mented with 0.2% CHAPS in a total volume of 204 lL. The nonspecific binding component was determined in the presence of 1 l M nonlabeled mixture of the PA1b isoforms. Incubations were performed for 2 h at room temperature in 96-well microtiter plates, then assays were transferred onto MultiScreen filter plates containing GF/B filters and vacuum drained using the MultiScreen system (Millipore, USA). Filters were washed with 200 lL of cold washing buffer (10 m M Tris/HCl buffer pH 7, methanol 60%) and transferred on gamma counter tubes. The radioactivity was counted on a gamma counter (Riastar, Packard Instrument, USA), and each point was the mean of triplicates. Binding data were analyzed using the RADLIG 4 software (BIO- SOFT, Cambridge, UK), and plots were drawn using the ORIGIN 5 software (Microcal, USA). Protein determination Protein content was measured by the bicinchoninic acid procedure developed by PIERCE (Rockford, USA) with bovine serum albumin as reference. Results The iodinated toxin displays a high specific radioactivity and is biologically active The 3741 Da isoform of PA1b contains a single tyrosine residue in position 31. Assays of iodination were performed using IodoBeads and chloramine-T. The yield of 125 I incorporated on the toxin was rather low (less than 5%) with IodoBeads, while a 40–50% incorporation was achieved using chloramine-T. The need for a high specific radioactivity for binding studies, and the fact that native PA1b and labeled PA1b cannot be separated by HPLC, led us to use stoichiometric amount of toxin and iodine. Under these conditions, the specific radioactivity of the 125 I-labelled PA1b, calculated by the ratio of the radioactivity measured by gamma counting and the amount of peptide evaluated by absorbance at 210 nm during HPLC analysis, was about 942 CiÆmmol )1 . This value is similar to the theoretical value of 890 CiÆmmol )1 for a 40% 125 I incorporation (40% of the 2220 CiÆmmol )1125 I initial activity). We next analyzed by HPLC the stability of the radio- ligand. When stored at )20 °C in 60% methanol, the 125 I-labelled PA1b showed a moderate but significant radiolysis, since about 25% of the toxin was degraded after one month, and less than 20% of radioligand remained intact after 6 months. Then, the radioligand was used only during the month following its labeling. The biological activity of the iodinated peptide was verified. For this purpose, a nonradioactive iodinated ligand was synthesized as described in the Experimental procedures section. The mass spectrometry analysis of the product revealed that more than 95% of the peptide incorporated two 127 I atoms. Toxicity assays on S. oryzae showed that the mortality on susceptible strains WAA42 was similar to that of the control (100 lg of native 3741 Da peptide) for comparable toxin amounts, and that the resistant strain ISOR3 was not affected (data not shown). 125 I-Labelled PA1b binds specifically to a proteinaceous component of a S. oryzae extract The results presented in Table 1 show that the three subfractions obtained by differential centrifugation of a crude extract of S. oryzae strain WAA42 (susceptible strain) were able to bind specifically the 125 I-labelled PA1b ligand. The specific binding activity was the highest in the 45 000 g fraction (microsomal fraction), which presented an enrich- ment of threefold on a protein basis. However, in terms of total activities, the binding was found with similar values in the 10 000 and 45 000 g fractions. The nonspecific binding component was about 20% of the total binding activity in all three fractions. The specific binding value increases with the increase of microsomal proteins, until it reaches a plateau for 25–30% of the radioligand bound (about 10–15 lg of microsomal proteins, Fig. 2). The microsomal Table 1. Binding activities of 1.1 n M 125 I-labelled PA1b to particulate fractions prepared by differential centrifugation of S. oryzae WAA42 extracts (5 lg of proteins). Particulate fraction 3000 g 10 000 g 45 000 g Total protein (mg) 18.1 12.4 6.9 Total binding activity (c.p.m. · 10 6 ) 24.5 45.2 49.9 Non-specific binding activity (% of total binding) 22.3 18.9 18.3 Specific binding activity (c.p.m.Ælg protein )1 ) 1355 3643 7243 Fig. 2. Dose dependence of 125 I-labelled PA1b specific binding to microsomal proteins from S. oryzae extract. Different amounts (0.8– 30 lg) of proteins of the microsomal fraction of S. oryzae extract were incubated in the presence of 0.4 n M 125 I-labelled PA1b. Ó FEBS 2003 A binding site for the PA1b entomotoxin (Eur. J. Biochem. 270) 2431 fraction of S. oryzae WAA42 was then used for further characterization, with 1–3 lg of protein loaded per assay, in order to work in the linear part of the curve. The proteinaceous origin of the binding activity was determined by its sensitivity to proteinase K and to heat- denaturation. The results reported in Table 2 show that the binding activity decreased strongly in the presence of proteinase K, which in control experiments displayed no activity towards the peptide ligand itself in the conditions of the binding experiment (data not shown). Adding an antiprotease cocktail (leupeptin, phenylmethanesulfonyl fluoride and E-64), or heat-denaturing of the proteinase K before the incubation with microsomal fraction proteins, totally prevented the loss of binding activity. Furthermore, the microsomal fraction proteins were incubated 15 min in binding buffer at temperatures ranging from 20 to 70 °C before performing the binding experiment. The results show that the binding activity began to decrease when proteins were preincubated at 40 °C and was practically undetect- able at 60 °C (respectively 95 and 7% as compared to the control, data not shown). The binding of the S. oryzae microsomal fraction to 125 I-labelled PA1b is reversible, saturable, and displays a high affinity Binding of the 125 I-labelled PA1b to the microsomal fraction proceeded in a time-dependent manner and an apparent equilibrium was reached after a 90-min incubation at room temperature. Addition of excess unlabeled ligand after equilibrium had been reached led to the loss of specifically bound radioligand, and this displacement was complete within 60 min (Fig. 3). The rate constants for association (k on ) was 2.03 · 10 5 M )1 Æs )1 , and for dissociation (k off )was 8.77 · 10 )4 s )1 . The calculated equilibrium dissociation constant (K d ¼ k off /k on ) was then of about 4.3 n M . A saturation experiment, using a freshly synthesized 125 I- labelled PA1b, was performed with labeled ligand ranging from 0.35 to 3.2 n M . The saturation plot showed that the specific binding was saturable, and the deduced Scatchard plot revealed a K d of 2.6 n M and a maximal binding capacity (B max ) of 36 pmol per mg of microsomal protein (Fig. 4). The Hill number was 0.98, suggesting the presence of a single class of binding site. Moreover, a competition experiment using the homologous nonradioactive 127 I 2 - labelled PA1b as competitor confirmed the results obtained by the saturation experiment on the 45 000 g fraction (K i ¼ 2.8 n M , B max ¼ 39 pmol per mg of protein). A single PA1b binding site is detectable in membrane fractions of the S. oryzae extract, and is specific for the native peptide Competition experiments were performed on the three subfractions of S. oryzae extracts using the 3741 Da PA1b as the competitor. The results showed that the native peptide displayed the same affinity as the iodinated toxin on the 45 000 g fraction (K i ¼ 2.2 n M ; B max ¼ 40 pmolÆmg of protein). Similar affinities were found on the 3000 and 10 000 g fractions with lower B max (a single class of binding site displaying K i of 2 and 3.5 n M ,andB max of 7.4 and 15.6 pmol per mg of protein, respectively, data not shown). Fig. 4. Saturation plot and deduced Scatchard plot (inset) of the binding of 125 I-labelled PA1b to the microsomal fraction of S. oryzae extract. Experiments were performed using 1 lg of proteins and increasing concentrations of 125 I-labelled PA1b (0.35–3.2 n M ). Table 2. Effect of proteinase K on the 125 I-labelled PA1b binding activity on the microsomal fraction of S. oryzae WAA42 extract. Treatment a Binding activity (%) Control b 100 Proteinase K (0.2 lg) 24 Proteinase K (2 lg) 7 Denaturated proteinase K (0.2 lg) c 107 Proteinase K (0.2 lg) + antiproteases d 104 a Microsomal fraction aliquots (2 lg of proteins) were incubated with effectors for 10 min at 37 °C before measuring the binding activity using 0.4 n M of 125 I-labelled PA1b. b Incubation in binding buffer without antiproteases only. c Protease was denaturated by boiling for 10 min. d Phenylmethylsulfonyl fluoride (1 m M ),4l M leupeptin, and 10 l M E-64. Fig. 3. Association and dissociation kinetics of 125 I-labelled PA1b to the microsomal fraction of S. oryzae extract. Binding experiments were performed using 2 lgofproteinsand0.4n M 125 I-labelled PA1b. Solid squares (j)andcircles(d) indicate total binding and non-specific binding during association, respectively, whereas triangles (.) indicate total binding during dissociation. The dissociation process was initi- ated by the addition of 1 l M nonlabeled 3741 Da PA1b (arrow). 2432 F. Gressent et al. (Eur. J. Biochem. 270) Ó FEBS 2003 Next, the selectivity of the binding site was examined by competition experiments as well: the results showed that neither a 7-kDa peptide (a Bowman–Birk serine protease inhibitor), nor the bovine insulin used as competitors were able to compete with the 125 I-labelled PA1b. Moreover, when the 3741 Da PA1b was reduced and the cysteine residues alkylated, the modified peptide completely lost its binding activity. Biological tests performed on S. oryzae showed that the alkylated peptide was not toxic for the susceptible or resistant weevil strains (data not shown). The binding activity correlates with the susceptibility or resistance of the Sitophilus spp. strains The specific binding activity of the microsomal fraction of different weevil strains, of the pea aphid A. pisum and of the fruit fly D. melanogaster was determined. Figure 5 shows that the five susceptible weevil strains tested (S. oryzae WAA42, Be ´ nin and Bouriz, S. granarius Brayard and S. zeamais LS) displayed a high specific binding activity, between 1700 and 2200 c.p.m. per lgofprotein.By contrast, no binding activity could be detected when incubating the 125 I-labelled PA1b with the microsomal fraction proteins of extracts of the four strains of S. oryzae resistant to the toxin (China, Mex1, ISOR3 and GV), even when increasing 25-fold the amount of microsomal proteins. The pea aphid (susceptible to the toxin) was able to bind the 125 I-labelled PA1b, to a lesser extent than the susceptible weevil strains. Surprisingly, the microsomal fraction of an extract of the resistant insect D. melanogaster also displayed a strong binding capacity. The determination of the characteristics of the binding site of these different strains or species is presented in Table 3. The binding activities of extracts from the five weevil strains of the three species showed very similar affinities (K i between2and9n M ) and maximal binding capacities (B max between 33 and 40 pmol per mg of protein). The D. melanogaster, but principally the A. pisum micro- somal fractions displayed lower affinities (K i ¼ 16 and 58 n M , respectively), but similar B max (40 pmol per mg protein) compared to the values obtained on weevil extracts. The radioligand is stable in the presence of the microsomal fraction of Sitophilus extract In order to determine the fate of the labeled toxin during the binding test, the 125 I-labelled PA1b ligand was incubated for 5hwith5lg of proteins of the microsomal fraction from S. oryzae WAA42 (susceptible strain) or from S. oryzae ISOR3 (resistant strain) in the conditions used for binding experiments. Following incubation, the radioactivity was dissociated from the binding site and recovered at more than 98% by a 60% methanol extraction. The HPLC analysis of this fraction revealed a single radioactive peak at a retention time corresponding to PA1b (data not shown). Thus, the radioligand was not degraded during incubation with extracts from either the susceptible or the resistant strain. Discussion The recent finding of the insecticidal properties of the PA1b peptide(s) [4] has opened new possibilities for cereal grain protection against weevils. However, the occurrence of naturally resistant strains of S. oryzae is an important problem, and could limit the usefulness of this peptide for plant protection. Although the mechanism of action of the toxin is still totally unknown, genetical analysis of the resistance within the S. oryzae species, implicating a single recessive gene [12], suggested that the toxicity might involve a specific receptor on the insect side. On this basis, the search for a PA1b binding protein may be the first step towards the cloning of a potential receptor in susceptible insects. In this work, we characterized the binding of PA1b to a proteinaceous component of a particulate fraction of S. oryzae extracts. The binding was saturable and reversible, and the binding site exhibits a high affinity for the native 3741 Da toxin (K d ¼ 2.4 n M ), an affinity compatible with a role in the toxicity mechanism. Furthermore, the binding site showed an unusually high maximal binding capacity (B max ¼ 40 pmol per mg of proteins, assuming that the Table 3. Affinities (K i ) and maximal binding capacity (B max )ofPA1bon different species and strains of insects measured by competitive inhibition of the 125 I-labelled PA1b binding. Insects species and strains K i ± SEM (n M ) B max (pmolÆmg protein) S. oryzae WAA42 2.6 ± 0.3 40 S. oryzae Be ´ nin 6 ± 1.8 39 S. oryzae Bouriz 4.5 ± 0.72 33 S. granarius Brayard 6.3 ± 1.6 40 S. zeamais LS 9 ± 1.6 34 A. pisum 58 ± 10 40 D. melanogaster 16 ± 3.3 40 Fig. 5. Specific binding of the 125 I-labelled PA1b ligand on the micro- somal fraction extracts from different insect species or strains. Bars indicate specific binding of the 125 I-labelled PA1b (0.4 n M )(±SEM) to 2 lg proteins from the microsomal fraction of S. oryzae strain WAA42 (1), Be ´ nin (2), Bouriz (3), GV (4), Mex1 (6), ISOR3 (8) and China (10), S. zeamais LS (12), S. granarius Brayard (13), A. pisum (14) and D. melanogaster (15). Bars labeled 5, 7, 9 and 11 correspond to specific binding displayed by 50 lg of microsomal proteins extracted from the resistant S. oryzae strains GV, Mex1, ISOR3 and China, respectively. Ó FEBS 2003 A binding site for the PA1b entomotoxin (Eur. J. Biochem. 270) 2433 binding implicated only one toxin molecule per binding site). PA1b seems to bind to a single protein, as only one binding site is detectable in the microsomal fraction, and the binding activity in the 3000 and 10 000 g fractions is probably due to the same protein as the binding in the three subfractions displayed the same characteristics. In terms of ligand specificity, we presently do not have available site directed mutants of the toxin, and a single isoform was purified from pea seeds to homogeneity, thus restricting the possibilities to identify the PA1b amino acids involved in ligand recognition. However, the fact that the iodinated ligand is still biologically active and displays an affinity comparable with the affinity of the native peptide indicates that the tyrosine residue is probably not implicated in ligand binding. On the contrary, the reduced, alkylated peptide looses its binding capacity, as its activity on weevils, suggesting that the tertiary structure of PA1b is critical. This is not surprising, as this is a common feature of many biologically active sulfur-rich peptides. This result, together with the absence of recognition of a BBI peptide or the bovine insulin, confirms that the binding site is probably highly specific for the PA1b tertiary structure. It was suggested that PA1b could have a low sequence similarity with the BBI peptide [5], and Watanabe et al. [13] showed that the soybean homologue of PA1b (named leginsulin in this paper) is able to compete with bovine insulin for binding to a soybean globulin exhibiting functional simi- larity to the rat insulin receptor. In our case, neither bovine insulin nor BBI were able to displace PA1b from its binding site, suggesting that the potential binding capacity of PA1b to an insulin binding protein, or a putative similarity with BBI, were not implicated in the toxicity mechanism in weevils. Within the Sitophilus genus, the presence of the binding activity fully correlates with the susceptibility to the toxin. All susceptible strains of the three species display similar affinities, whereas no binding activity could be detected in the four S. oryzae resistant strains, including ISOR3, a three generation backcross isoline to WAA42 [12]. This result strongly suggests that the high-affinity binding site is the molecular target of the peptide in Sitophilus,and that it may play a major role in the toxicity process. As the labeled ligand was not degraded during the binding experiment, the absence of binding activity on the resistant strains demonstrates that the resistance mechan- ism within the Sitophilus genus involves either a modifi- cation or the absence of the molecular target. Indeed, the absence of binding in resistant strains could be due to the absence of the target protein, or to a mutation within the site of the binding to PA1b. Actually, target mutation resistance is widely used by insects to resist a wide range of toxins, and was reported for different insecticide targets, as for the GABA receptor, the sodium channels or the acetylcholinesterases [14]. The presence and density of the binding site in all insect species tested, and among three insect orders (e.g. Coleoptera, Diptera and Hemiptera) suggests that this protein is widely represented and conserved in insects, and that PA1b could share a similar mechanism of action in all susceptible species. However, the presence of the binding site in the resistant species D. melanogaster,with characteristics roughly similar to those of the Sitophilus susceptible strains, suggests that a different mechanism of insensitivity exists in this species. This mechanism seems not to implicate the binding site, and could take place either upstream or downstream of the perception step. We could for example speculate that certain insects are able to enzymatically degrade the peptide in a nonactive form, which is one reported way of resistance of insects against Bt toxins [15], and proteinase inhibitors [16], or that a protein in a putative signal transduction cascade may differ in Drosophila. Also, the possibility that the PA1b target may be very common in insects will require further biological assays to test side-effects of a PA1b-based biocontrol strategy, in particular on useful insects (e.g. ladybird beetle, trichogramma, bee). In conclusion, this work allowed the first in vitro biochemical characterization of a high-affinity binding site for the product of a plant resistance gene in insects, with binding characteristics fully correlated to the genetical analysis [12], and thus falling in the gene for gene paradigm. The occurrence of the binding site in different susceptible insect species indicated that this binding site is probably the molecular target of the toxin, and that resistance within the S. oryzae species potentially involves a variation in this protein. However, the role of the binding protein in insects, and therefore the overall molecular mechanism of toxicity is still unknown. To date, the cystine-knot peptides comprised mainly ion channel blockers and enzyme inhibitors. How- ever, ion channel effectors were only found in animal venoms, with the exception of a unique sulfur-rich plant peptide (the CSab-type c-zeathionin) which has been found recently to be a Na + channel blocker [17]. However, all cystine-knot plant peptides with known targets at the present time are enzyme inhibitors (including for example trypsin, a-amylase and carboxypeptidase) [18]. PA1b is of plant source, but all inhibitory assays conducted until now has failed to detect any effect on enzymatic activities, including papain, trypsin and chymotrypsin-like activities [4]. Although PA1b could have some low similarity with BBI peptides, the absence of consensus residues in the active site results in probable lack of any antiprotease activity [5]. Then, the purification of the binding site, and the cloning of the corresponding gene in susceptible and resistant strains will probably help to answer these questions, and will provide us with the molecular tools to monitor the outburst of resistant populations, and eventually to engineer a pea- related toxin into a peptide active on both resistant and susceptible weevils. Acknowledgements We thank Francesca Cardinale for carefully reviewing this manuscript. We thank J. Gueguen and E. Ferrasson for providing us the 3741 Da toxin isoform and the PA1b isoform mixture. References 1. Schnepf, E., Crickmore, N., Van Rie, J., Lereclus, D., Baum, J., Feitelson, J., Zeigler, D.R. & Dean, D.H. (1998) Bacillus thur- ingiensis and its pesticidal crystal proteins. Microbiol. Mol. Biol. Rev. 62, 775–806. 2. Peumans, W.J. & Vandamme, E.J.M. (1995) Lectins as plant defense proteins. Plant Physiol. 109, 347–352. 2434 F. Gressent et al. (Eur. J. Biochem. 270) Ó FEBS 2003 3. Gatehouse, A.M.R. & Gatehouse, J.A. (1998) Identifying proteins with insecticidal activity: use of encoding genes to produce insect- resistant transgenic crops. Pestic. Sci. 52, 165–175. 4. Delobel, B., Grenier, A., Gueguen, J., Ferrasson, E. & Mbailao, M. (1998) Utilisation D’un Polypeptide De ´ rive ´ D’une Albumine PA1b de Le ´ gumineuse Comme Insecticide. French Patent 98 05877. 5. Higgins, T.J.V., Chandler, P.M., Randall, P.J., Spencer, D., Beach,L.R.,Blagrove,R.J.,Kortt,A.A.&Inglis,A.S. (1986) Gene structure, protein structure and regulation of the synthesis of a sulfur-rich protein in pea seeds. J. Biol. Chem. 261, 11124–11130. 6. Hancock, K.R., Ealing, P.M. & White, D.W. (1994) Identification of sulfur-rich proteins which resist rumen degradation and are hydrolysed rapidly by intestinal proteases. Br.J.Nutr.72, 855–863. 7. Spencer, D., Higgins, T.J.V., Freer, M., Dove, H. & Coombe, J.B. (1988) Monitoring the fate of dietary proteins in rumen fluid using gel electrophoresis. Br.J.Nutr.60, 241–247. 8. Jouvensal, L., Louis, S., Quillien, L., Gue ´ guen,J.,Rahbe ´ ,Y.& Vovelle, F. (2002) PA1b, an insecticidal peptide extracted from the seed of the pea Pisum sativum:1H)2D-NMR study and molecular modelling. In Proceedings of the XXth International Conference on Magnetic Resonance in Biological Systems, p. 232. Toronto, Canada. 9. McDonald, N.Q. & Hendrickson, W.A. (1993) A structural superfamily of growth factors containing a cystine-knot motif. Cell 73, 421–424. 10. Craik, D.J., Daly, N.L. & Waine, C. (2001) The cystine knot motif in toxins and implications for drug design. Toxicon 39, 43–60. 11. Jennings,C.,West,J.,Waine,C.,Craik,D.&Anderson,M. (2001) Biosynthesis and insecticidal properties of plant cyclotides: the cyclic knotted proteins from Oldenlandia affinis. Proc. Natl Acad. Sci. USA 98, 10614–10619. 12. Grenier, A.M., Mbauguinam, M. & Delobel, B. (1997) Genetical analysis of the ability of the rice weevil Sitophilus oryzae (Coleoptera, Curculionidae) to breed on split peas. Heredity 79, 15–23. 13. Watanabe, Y., Barbashov, S.F., Komatsu, S., Hemmings, A.M., Miyagi, M., Tsunasawa, S. & Hirano, H. (1994) A peptide that stimulates phosphorylation of the plant insulin-binding protein. Eur. J. Biochem. 224, 167–172. 14. Hemingway, J. & Ranson, H. (2000) Insecticide resistance in insect vectors of human disease. Annu. Rev. Entomol. 45, 371–391. 15.Keller,M.,Sneh,B.,Strizhov,N.,Prudovsky,E.,Regev,A., Koncz, C., Schell, J. & Zilberstein, A. (1996) Digestion of d-endotoxins by gut proteases may explain reduced sensitivity of advanced instar larvae of Spodoptera littoralis to CryIC. Insect Biochem. Mol. Biol. 26, 365–373. 16. Girard, C., Le Me ´ tayer, M., Bonade ´ -Bottino, M., Pham-Dele ` gue, M.H. & Jouanin, L. (1998) High level of resistance to proteinase inhibitors may be conferred by proteolytic cleavage in beetle larvae. Insect Biochem. Mol. Biol. 28, 229–237. 17. Kushmerick,C.,deSousaCastro,M.,SantosCruz,J.,Bloch,C.Jr & Beirao, P.S.L. (1998) Functional and structural features of c-zeathionins, a new class of sodium channel blockers. FEBS Lett. 440, 302–306. 18. Norton, R.S. & Pallaghy, P.K. (1998) The cystine-knot struc- ture of ion channel toxins and related polypeptides. Toxicon 11, 1573–1583. Ó FEBS 2003 A binding site for the PA1b entomotoxin (Eur. J. Biochem. 270) 2435 . a seed albumin [5], it was named PA1b for pea albumin 1b. PA1b is the result of the post-translational cleavage of the albumin proprotein PA1, also releasing. Characterization of a high-affinity binding site for the pea albumin 1b entomotoxin in the weevil Sitophilus Fre ´ de ´ ric Gressent, Isabelle Rahioui and