Bacillus thuringiensis Cry1Ac Toxin-Binding and Pore-Forming Activity in Brush Border Membrane Vesicles Prepared from Anterior and Posterior Midgut Regions of Lepidopteran Larvae

Abstract

It is generally accepted that Bacillus thuringiensis Cry toxins insert into the apical membrane of the larval midgut after binding to specific receptors, and there is evidence that the distribution of binding molecules along the midgut is not uniform. By use of the voltage-sensitive dye DiSC₃(5) and ¹²⁵I-labeled Cry1Ac, we have measured the effect of Cry1Ac in terms of permeabilization capacity and of binding parameters on brush border membrane vesicles (BBMV) prepared from the anterior and the posterior regions of the larval midgut from two insect species, Manduca sexta and Helicoverpa armigera. The permeabilizing activity was significantly higher with BBMV from the posterior region than with the one observed in the anterior region in both insect species. Instead, ¹²⁵I-Cry1Ac bound specifically to BBMV from the two midgut regions, with no significant differences in the binding parameters between the anterior and posterior regions within an insect species. N-acetylgalactosamine inhibition patterns on pore formation and binding differed between anterior and posterior midgut regions and between species, providing evidence of a multifaceted involvement of the sugar in the Cry1Ac mode of action. The analysis of binding and pore formation in different midgut regions could be an effective method to study differences in the mode of action of Cry1Ac toxin in different species.

The insecticidal activity of Bacillus thuringiensis proteins produced in parasporal crystals during sporulation (Cry proteins or Cry toxins) has been widely studied in lepidopteran insects, and there is relatively good evidence for the way the proteins act once ingested by a susceptible insect. There is general agreement in that the parasporal crystal dissolves after ingestion, and then the released protoxin is processed by gut proteases to an activated form; the active toxin crosses the peritrophic membrane and binds to specific receptors in the brush border membrane of midgut columnar cells and eventually leads to cell death (6, 13, 41, 43).

The events that take place after binding are not yet clear, although the permeabilization of the membrane induced by toxin insertion and pore formation has long been proposed (34) and there is much evidence which supports this view (6, 13, 41, 43). In a model for the mode of action of Cry1A toxins, cadherin and aminopeptidase N (APN) receptors have a pivotal role (7): Cry1A toxins are proposed to bind to cadherin first, and then, after proteolytical modification of the bound toxin, a homo-oligomer is formed, which is transferred to APN and then inserts into the membrane. In support of this model, a recent study has shown that Cry1A-modified toxins, which spontaneously form the oligomer, can bypass the step of binding to cadherin to produce the toxic effect (46). Membrane-bound alkaline phosphatase also seems to play a role in the mode of action of Cry1A toxins, and it has been shown to be a Cry1Ac binding molecule in Manduca sexta (39, 42) and Heliothis virescens (14, 30). A recent model has challenged the pore-forming model and proposes that only binding to cadherin is required to produce cell death; binding to APN or alkaline phosphatase is irrelevant in this model, and pore formation may be but a secondary effect of the toxic events triggered by toxin binding to cadherin (51, 52). It is important to note that evidence for the pore-forming model has been obtained mainly with in vitro experiments with artificial membranes or with brush border membrane vesicles (BBMV) (7, 19, 53), whereas the model of Zhang et al. has been developed from ex vivo observations with an insect ovarian cell line heterologously expressing the cadherin receptor.

Immunochemical localization in midgut tissue sections has been applied to study the distribution of Cry1A receptors along the larval midgut. Early work by Bravo et al. (8) showed that some Cry proteins concentrated, after ingestion, in the anterior region of the midgut, whereas some others concentrated in the posterior region. Chen et al. (12) have shown the differential distribution in the M. sexta midgut of APN, alkaline phosphatase, and cadherin, and the bound Cry1A proteins. Aimanova et al. (1) also found an uneven distribution of the cadherin protein along the midgut regions of M. sexta and H. virescens. One study has used BBMV to show differences in the permeabilizing activity of Cry1Ac in different regions of the midgut of M. sexta (11) in the presence and in the absence of N-acetylgalactosamine (GalNAc).

Cry1Ac has the peculiarity, not shared by other Cry1A proteins, of binding to APN through GalNAc residues through a lectin-like pocket in its domain III (10, 32, 33, 35). Cry1Ac binding to H. virescens membrane-bound alkaline phosphatase has also been shown to be dependent on the presence of an N-linked oligosaccharide containing a terminal GalNAc residue (30). Because of other similarities with APNs (both are glycosylphosphatidylinositol-anchored membrane proteins), it seems likely that membrane-bound alkaline phosphatases in other insect species are also GalNAc-mediated Cry1Ac binding molecules. Cry1Ac also binds to Bombyx mori peritrophic membrane more efficiently than Cry1Aa does, and its binding is greatly reduced by preincubation with GalNAc (24). By making use of this property of Cry1Ac, one can dissect the binding of this protein by using GalNAc as an inhibitor and thus discriminate binding to GalNAc-bearing molecules (such as APN and alkaline phosphatase) from binding to other molecules in the brush border (such as cadherin) (11, 15) or the peritrophic membrane (24).

Binding analyses of Cry toxins to lepidopteran BBMV have been carried out since 1988 (27) and have showed the importance of midgut receptors in the mode of action of Cry toxins. However, to our knowledge, binding studies have never been performed with BBMV prepared from different regions of the midgut. In the present work, we wanted to see whether the differences in permeabilization observed by Carroll et al. (11) in regions of M. sexta midgut would be also reflected at the level of toxin binding to BBMV. Furthermore, we have included a second insect species, also susceptible to Cry1Ac (2, 3), to determine how general the results obtained with M. sexta are.

MATERIALS AND METHODS

Cry1Ac toxin preparation.

Cry1Ac was obtained from the recombinant B. thuringiensis EG11070 strain (Ecogen, Inc., Langhorne, PA) and was trypsin-activated and chromatographically purified as described previously (15). The protein concentration was determined by densitometry after sodium dodecyl sulfate-polyacrylamide gel electrophoresis and staining with Coomassie blue.

Cry1Ac protein aliquots used for BBMV permeability assays (reported in fluorescence measurements) were dialyzed overnight at 4°C against 20 mM Tris-HCl, 150 mM CsCl at pH 8.6, with regenerated cellulose membranes (molecular mass cutoff of 7,000 Da; Pierce, Rockford, IL) in order to remove the Na⁺ present in the elution buffer used for chromatography purification.

Preparation of BBMV.

Last-instar larvae from laboratory colonies of M. sexta and H. armigera were excised, and the midguts were pulled out. Then the midgut was divided into three equidistant parts, and the central region was discarded (the reason being that if Cry1Ac receptors are distributed nonhomogeneously along the midgut, forming a gradient, the anterior and posterior regions would be the ones showing maximum differences). Tissue from the anterior and posterior regions was washed separately in MET buffer (250 mM mannitol, 17 mM Tris-HCl, and 5 mM EGTA, pH 7.5), blotted on filter paper, and frozen in liquid nitrogen in batches of about 5 g of tissue. Samples were stored in liquid nitrogen until used.

BBMV for binding analyses and for enzyme assays were prepared following the Mg²⁺ differential precipitation method (49). BBMV were resuspended in one-half MET and stored at −80°C until required. For pore formation assays, BBMV were prepared according to the Ca²⁺ precipitation method (17). The final BBMV pellet was resuspended in a small volume of resuspension buffer (300 mM mannitol, 1 mM KCl, 10 mM HEPES-Tris, pH 7.2) to obtain a concentrated preparation of BBMV (final membrane protein concentration of 5 to 10 mg/ml). BBMV to be used in pore formation experiments were used within 2 h after preparation.

The BBMV protein concentration was determined by the Bradford method (5) using bovine seroalbumine as standard.

BBMV permeability assays.

BBMV permeability to K⁺ was monitored by recording the fluorescence quenching of the voltage-sensitive cyanine dye 3,3′-dipropylthiodicarbocyanine iodide, DiSC₃(5). Vesicle membranes have a basal permeability to K⁺; thus, in the presence of a K⁺ inwardly directed electrochemical gradient, there is an influx of cations which leads to a release of the dye and thus to an increase in fluorescence. In the presence of pore-forming agents, there is a significantly higher influx and, thus, a significantly higher increase in fluorescence (44).

Suitable amounts of M. sexta BBMV and H. armigera BBMV were preincubated for 30 and 45 min, respectively, at 4°C with Cry1Ac toxin (final concentration of 40 μg/mg of BBMV proteins) or just with the toxin buffer (20 mM Tris-HCl at pH 8.6, 150 mM CsCl) in controls. After preincubation, the BBMV suspensions were transferred (at a final concentration of 22 μg BBMV protein/ml) into polyacryl cuvettes containing the same buffer used for BBMV resuspension to which 6 μM DiSC₃(5) (Molecular Probes, Società Italiana Chimici, Italy) had been added. Extravesicular increments of K⁺ concentrations were made by three successive additions of 2 M KCl to give final concentrations of 20, 40, and 60 mM KCl, respectively. Since intra- and extravesicular buffers had the same ionic composition, pH, and osmolarity and the apical membranes of lepidopteran columnar cells are almost impermeable to Cl⁻, the membrane potential variations depended exclusively on the potassium movements across the membranes. Each experiment was tested three times or more with at least two independent BBMV preparations. DiSC₃(5) fluorescence was measured in a Varian (Cary Eclipse) spectrofluorometer equipped with a thermostatized (25°C) four-cuvette holder, using excitation and emission wavelengths of 645 and 665 nm, respectively.

For the GalNAc inhibition assays, 400 mM GalNAc was added to the preincubation reaction mixture and then the assays proceeded as indicated above.

Enzyme assays.

Alkaline phosphatase activity in BBMV was determined spectrophotometrically by using 4-nitrophenyl phosphate in diethanolamine buffer (Sigma, St. Louis, MO). The increase in absorbance at 405 nm was continuously read for 6 min at room temperature. The amount of BBMV to obtain an absorbance reading in the lineal range depended on the midgut region and the insect species. Nine and 15 μg of proteins were used with BBMV from the anterior regions, whereas 0.3 and 1.5 μg were used with BBMV from the posterior regions of M. sexta and H. armigera, respectively. Enzyme activities were calculated using an extinction coefficient of 18,450 M⁻¹ cm⁻¹ for 4-nitrophenol.

Leucine aminopeptidase activity was determined by using 4 mM l-leucyl-p-nitroanilide with 0.5 to 8 μg BBMV proteins in 37.5 mM Tris-HCl, pH 7.5. Absorbance was measured at 410 nm during 3 min at room temperature. Enzyme activities were calculated using an extinction coefficient of 9,900 M⁻¹ cm⁻¹ for 4-nitroaniline.

Cry1Ac labeling.

Radiolabeling of Cry1Ac was performed using the chloramine-T method of incubating 20 μg of protein with 0.25 mCi of Na¹²⁵I (GE Healthcare, Barcelona, Spain) (47). The specific radioactivity obtained was 1.42 mCi/mg of Cry1Ac.

Binding assays.

Binding assays were carried out in a final volume of 0.1 ml in binding buffer (phosphate-buffered saline-0.1% bovine serum albumin). BBMV (4 or 2 μg of membrane protein from H. armigera or M. sexta, respectively) were incubated for 1 h at room temperature with 320 pM ¹²⁵I-Cry1Ac. Competition experiments were carried out by adding increasing concentrations of unlabeled Cry1Ac and binding parameters (dissociation constant [K_D] and concentration of binding sites [R_t]) were estimated with the LIGAND computer program (40). These binding experiments were replicated at least twice.

For the GalNAc inhibition assays, ¹²⁵I-Cry1Ac was preincubated with 25 mM GalNAc for 45 min at room temperature prior to the start of the binding assay with the addition of 10 μg BBMV (15). Nonspecific binding was determined by adding a 1,000-fold excess of unlabeled Cry1Ac. Specific binding was estimated by subtracting the nonspecific binding from the total binding. These assays were replicated at least four times.

RESULTS

Cry1Ac permeabilization of BBMV from different midgut regions.

As can be seen in Fig. 1 and 2, the intensity of fluorescence increases in the buffer control samples upon the addition of KCl. This is because BBMV have an intrinsic permeability to K⁺, and this increase is taken as the negative control of these experiments. As a positive control, we used valinomycin, a K⁺ ionophore. Preliminary dose-response experiments indicated that 23 μM valinomycin gave a maximal increase in fluorescence intensity, corresponding to the free influx of K⁺ (data not shown). When BBMV were preincubated with Cry1Ac, an increase of fluorescence intensity was observed, significantly different from that of the negative control, which indicates that pores are being formed (Fig. 1).

FIG. 1.

Effect of Cry1Ac toxin on K⁺ permeability of BBMV from M. sexta anterior (A) or posterior (B) midgut regions and from H. armigera anterior (C) or posterior (D) midgut regions. BBMV, preloaded with 300 mM mannitol, 1 mM KCl, 10 mM HEPES-Tris at pH 7.2, were preincubated at 4°C with the buffer in which the toxin was dissolved (control and valinomycin samples) or with Cry1Ac (40 μg/mg of BBMV proteins). After preincubation, BBMV were diluted in a cuvette with the resuspension buffer supplemented with 6 μM DiSC₃(5). The ionophore valinomycin was added to the cuvette simultaneously to the BBMV addition. KCl was added at the times indicated by the arrows to obtain extravesicular final concentrations of 20, 40, and 60 mM. Each trace represents the mean ± SE of at least three independent experiments. AU, arbitrary units of fluorescence.

FIG. 2.

Effect of GalNAc on Cry1Ac-induced K⁺ permeabilization of BBMV from M. sexta anterior (A) or posterior (B) midgut regions and from H. armigera anterior (C) or posterior (D) midgut regions. BBMV, preloaded with 300 mM mannitol, 1 mM KCl, 10 mM HEPES-Tris at pH 7.2, were preincubated at 4°C with the buffer in which the toxin was dissolved (control and valinomycin samples) or with Cry1Ac (40 μg/mg of BBMV proteins) in the presence of 400 mM GalNAc. Other details are as described in the legend for Fig. 1.

In both insect species, 40 μg Cry1Ac/mg BBMV protein caused maximal permeabilization of BBMV from the posterior region, comparable to that caused by valinomycin (Fig. 1B and D). This result is in contrast to its effect on the anterior region, for which the permeabilizing capacity was lower than that of valinomycin, though significantly different from that of the negative control (Fig. 1A and C).

GalNAc inhibition of Cry1Ac-induced permeabilization of BBMV.

When GalNAc was added to the preincubation reaction mixture, K⁺ permeabilization by Cry1Ac in the posterior region of both insect species was completely inhibited, giving values as the buffer control (Fig. 2B and D). The same complete inhibition was observed with BBMV from the anterior region of H. armigera (Fig. 2C); however, with BBMV from the anterior region of M. sexta, permeabilization was hardly inhibited by GalNAc (Fig. 2A), resembling the values obtained with Cry1Ac in the absence of GalNAc (Fig. 1A).

Alkaline phosphatase and leucine aminopeptidase activities in different midgut regions.

The activities of these two brush border membrane marker enzymes, which have been identified as midgut receptors for Cry1A toxins, were measured in the anterior and posterior regions of H. armigera and M. sexta. As shown in Table 1, both activities are significantly higher in the posterior region than in the anterior region in both species. A comparison between species showed that the specific activity of both enzymes in the posterior region was higher in M. sexta than in H. armigera, whereas the reversed situation applied to the anterior region. A comparison within a species between midgut regions showed that differences in specific enzyme activity were much smaller in H. armigera than in M. sexta (3-fold versus 100-fold, respectively, for both enzyme activities).

TABLE 1.

Alkaline phosphatase and leucine aminopeptidase activities in different midgut regions of H. armigera and M. sexta

Insect species	Sp acta
	Alkaline phosphatase		Leucine aminopeptidase
	Anterior	Posterior	Anterior	Posterior
H. armigera	2.5 ± 0.4	8.9 ± 0.9	0.4 ± 0.1	1.1 ± 0.1
M. sexta	0.4 ± 0.1	43 ± 6	0.13 ± 0.02	14.4 ± 0.5

^aExpressed in μmol/min per mg of protein. Results are means ± standard deviations of at least four replicates.

Cry1Ac binding to different midgut regions.

Specific binding of ¹²⁵I-Cry1Ac was found in both the anterior and the posterior regions of the midgut in the two insect species tested, as can be seen from the competition curves in Fig. 3. No apparent differences were observed between the competition curves within the same insect species, which is corroborated by the lack of significant differences (P > 0.05) in the binding parameters between the anterior and the posterior regions (Table 2). Both insects bound Cry1Ac with similar affinities (they had similar K_D values), whereas M. sexta showed higher R_t values than H. armigera did.

FIG. 3.

Binding of ¹²⁵I-Cry1Ac as a function of increasing concentrations of unlabeled Cry1Ac in H. armigera (A) and M. sexta (B). BBMV were prepared from the anterior (•) or the posterior region (○) of the midgut. Each data point represents the mean of two replicates, and the error bars show standard deviations.

TABLE 2.

K_Ds and R_ts for ¹²⁵I-Cry1Ac binding to BBMV from different midgut regions from H. armigera and M. sexta^a

Binding parameter	Value for insect species
	H. armigera		M. sexta
	Anterior	Posterior	Anterior	Posterior
KD (nM)	3.0 ± 1.3	3.4 ± 0.3	5.0 ± 0.6	3.9 ± 1.3
Rt (pmol/mg)	5.1 ± 1.3	7.9 ± 6.3	47.5 ± 3.5	66.3 ± 23

^aResults are shown as means ± standard errors. K_D and R_t values were calculated from at least two experiments.

GalNAc inhibition of ¹²⁵I-Cry1Ac binding.

Figure 4 shows the specific binding of ¹²⁵I-Cry1Ac to BBMV in the presence and in the absence of GalNAc represented as a percentage of the total radioactivity added. Considering the specific binding for each sample in the absence of GalNAc as 100%, GalNAc inhibition accounted for 11.6% and 68% of the specific binding in the anterior and posterior regions, respectively, of the H. armigera midgut. In M. sexta, the specific binding inhibition was 29% for the anterior region and 52% in the posterior region. GalNAc inhibition was statistically significant (P < 0.05) in all samples except for the anterior midgut region of H. armigera.

FIG. 4.

Specific binding of ¹²⁵I-Cry1Ac to BBMV from the anterior and posterior regions of the midgut from H. armigera and M. sexta in the presence (white bars) or in the absence (gray bars) of 25 mM GalNAc. Each point represents the mean of at least four determinations, and the error bars show standard deviations. Asterisks show the values which were significantly different from those for the control without GalNac (P < 0.05), for which the percentage of inhibition is given.

DISCUSSION

It is known that in the presence of a transmembrane electrical potential, the fluorescence quenching of the voltage-sensitive dye DiSC₃(5) can be used to monitor the diffusional K⁺ movements across the vesicular membrane in insect midgut BBMV (9, 21, 37, 38). In the present study, we measured the ability of Cry1Ac to induce the formation of pores with K⁺ channel properties by recording the fluorescence signal variations of the dye DiSC₃(5) with increasing inwardly directed K⁺ gradients.

The results showed that the preincubation with Cry1Ac in both species affected the permeability in BBMV from the posterior region, leading to the complete dissipation of the K⁺ electrochemical gradient during the permeability assay. On the contrary, the effect of the toxin was lower in the BBMV from the anterior region than in BBMV from the posterior region. Furthermore, this permeabilizing activity was completely inhibited by GalNAc in the posterior regions of both insects and in the anterior region of H. armigera, showing that GalNAc-bearing molecules (APN, alkaline phosphatase, or other) are involved in the permeabilization process in these regions. However, BBMV from the anterior region from M. sexta did not show any detectable inhibition of permeabilizing activity by GalNAc. Our results with M. sexta regarding pore formation and its inhibition by GalNAc are in good agreement with the results from the kinetics experiments performed by Carroll et al. (11), who used a more indirect method (light scattering) to measure permeabilization activity of Cry1Ac. However, in contrast to our results, these authors failed to obtain significant differences between both midgut regions when preincubating the toxin with BBMV. The main reason for this discrepancy may be the presence of an alkaline pH inside the vesicles in the experiments by Caroll et al. (pH 9 was present in their study, in contrast to pH 7.2 in our study), which may have reasonably affected the integrity of the vesicles before starting the permeability experiment (18, 31). Moreover, the long preincubation time (1 h) at room temperature could further affect the quality of the vesicles used. These authors proposed that there must be a different mechanism in M. sexta that leads to pore formation in the anterior region different from the one that takes place in the posterior region. It is worth noting that this conclusion cannot be extended to H. armigera, which showed complete inhibition of both midgut parts, at least under the experimental conditions tested.

The enzymatic assays indicated much higher APN activity in the posterior region than in the anterior region, in agreement with previous reports on APN activity in M. sexta (11, 48) and with the histochemical determination of APN1 distribution in the M. sexta midgut (12). In addition, we also found higher alkaline phosphatase activity in the posterior region in both species, which is in disagreement with the reported midgut distribution of membrane-bound alkaline phosphatase in M. sexta as assessed by immunolocalization (12). A possible explanation for this discrepancy may be the occurrence of more than one type of membrane-bound alkaline phosphatase in M. sexta. Our results for Cry1Ac-induced BBMV permeabilization and its GalNAc inhibition in the posterior parts of both insects would be perfectly explained by the accumulation of APN and/or alkaline phosphatase in the posterior region, although the involvement of additional molecules cannot be excluded. The GalNAc complete inhibition of the Cry1Ac permeabilizing activity in the anterior region of H. armigera could also be due to the presence of APN and/or alkaline phosphatase in this region. However, the lack of GalNAc inhibition in the anterior region of M. sexta midgut may reflect the occurrence of isozymes other than those found in the posterior part.

Chen et al. (12), by the use of specific antibodies, determined the distribution of three Cry1A-binding proteins in the M. sexta midgut of second-instar larvae. The researchers showed that APN1 was concentrated in the posterior region, whereas the cadherin (Bt-R1) and membrane-bound alkaline phosphatase were more or less evenly distributed along the midgut. They also showed that in vitro binding to tissue sections of Cry1Ac, but not Cry1Aa or Cry1Ab, took place preferentially in the posterior region. The combination of our results and the observations of Chen et al. suggest that APN1 may be the key receptor involved in GalNAc-dependent pore formation, but they do not support (or reject) the involvement of alkaline phosphatase. Immunodetection of these receptors in B. mori has shown a similar distribution of APN along the midgut; however, in contrast to the situation in M. sexta, cadherin was found to be more abundant in the posterior region of B. mori (23). It seems that the distribution of these receptors varies depending on the species studied, in agreement with our observations of H. armigera.

The pore-forming activity of Cry1Ac in the anterior region of M. sexta, which is not inhibited by GalNAc, must be independent of APN1. Therefore, in addition to cadherin as the main Cry1Ac-binding protein, other molecules different from APN1 must be involved in Cry1Ac insertion and pore formation in the anterior region.

It was known that Cry1Ac binds with high affinity to binding sites in BBMV prepared from whole midguts from M. sexta (28, 47) and H. armigera (2, 15, 25). Our results using ¹²⁵I-Cry1Ac have shown that specific binding occurs in the two midgut regions from both insect species (Fig. 3). What was not so obvious was that ¹²⁵I-Cry1Ac binding was so quantitatively similar in both regions, i.e., we did not find significant differences in the binding parameters between the anterior and posterior regions within an insect species (Table 2). These results indicate that whatever the region which most contributes to the BBMV preparation, this should not affect binding parameters, at least in the two insect species tested. Given that APN is concentrated in the posterior region (determined not only by APN1 immunolocalization but also by enzyme activity), other Cry1Ac-binding molecules must compensate for this in the anterior region, with either their concentrations or their affinities for this toxin. Aimanova et al. (1), working with first-instar larvae, have shown that cadherin was detected only in the anterior region of the H. virescens midgut and that although it could be detected in M. sexta in both regions, it was concentrated in the anterior region in this insect.

GalNAc inhibition of Cry1Ac binding has also shown a different pattern in M. sexta and in H. armigera (Fig. 4). In the former, this sugar inhibited binding in the anterior and posterior regions, although relatively more inhibition occurred in the posterior region. In contrast, in H. armigera, binding was only significantly inhibited in the posterior region. The inhibition found in both insects in the posterior region is in agreement with the differential distribution of APN. The binding component not inhibited by GalNAc indicates the contribution of other types of molecules in the binding of Cry1Ac. The inhibition of Cry1Ac binding in the anterior region of M. sexta seems to indicate that this region possesses GalNAc-dependent Cry1Ac-binding molecules which do not take part in the pore formation process since GalNAc did not inhibit pore formation in this region. In H. armigera, the Cry1Ac binding inhibition in the anterior region is not statistically significant. However, given that GalNAc inhibits pore formation in that region, it follows that GalNAc must be inhibiting binding to some receptor molecules. Since the reduction in binding is so small (Fig. 4), the GalNAc-bearing molecules must represent a minor part of the binding molecules in that region.

Taken together, the results suggest that all Cry1Ac permeabilizing activity in BBMV from the posterior region of the midgut of the two insect species studied is due to Cry1Ac binding to GalNAc residues, probably from APN and alkaline phosphatase, and the same applies to BBMV from the anterior region in H. armigera. However, another type of GalNAc-independent binding takes place in the anterior region of M. sexta, which also leads to membrane permeabilization. This is the reason why Carroll et al. (11) proposed that Cry1Ac is acting on the anterior region of M. sexta midgut by a different mechanism from the posterior region. The lack of differences between the two midgut regions in terms of specific binding of Cry1Ac in the two species, together with the different GalNAc binding-inhibition patterns, indicates that there is a distinct distribution of Cry1Ac-binding proteins, some GalNAc dependent and some not, and in turn, some of the GalNAc-dependent Cry1Ac-binding proteins do not seem to be involved in membrane permeabilization (those of the anterior region of M. sexta midgut).

It is important to note that despite the GalNAc-mediated binding and pore-forming activity of Cry1Ac observed in vitro, two independent groups obtained evidence that Cry1Ac binding through GalNAc residues does not have a major effect on in vivo toxicity (10, 29). They showed that Cry1Ac mutants, unable to bind to GalNAc residues and to purified APN from M. sexta, had impaired pore-forming capacity and reduced binding affinity for M. sexta BBMV; however, their toxicity to this insect remained practically unaffected. The binding of Cry1Ac to midgut proteins through GalNAc residues may mask other toxin-receptor interactions, in binding and permeabilization analysis, with more relevance to the in vivo toxicity. There is an increasing body of evidence showing that not all binding to BBMV is effective and that there are binding molecules in the brush border that do not contribute to the toxicity of Cry toxins. For example, there are some B. thuringiensis-resistant strains with high levels of resistance to Cry1Aa, Cry1Ab, and Cry1Ac toxins, but not to other B. thuringiensis proteins, which have lost binding to Cry1Aa or Cry1Ab, but not to Cry1Ac (4, 20, 22, 26, 36, 45, 50). This type of highly selective resistance is typical of alterations in the binding site and we expect an alteration in a shared receptor to be the cause of resistance (16). These cases have been explained by researchers proposing either different binding molecules or different epitopes/binding sites, one shared by all Cry1A toxins but also others that only Cry1Ac would bind. These unique Cry1Ac binding sites may very likely be the GalNAc residues. It is therefore important to be able to distinguish between the Cry1Ac binding which is dependent on GalNAc residues (which seems to be futile for in vivo toxicity) and the GalNAc-independent one (which may be effective).

The approach used in this work, although indirect, has succeeded in revealing the complex distribution of Cry1Ac binding molecules along the larval midgut regions of two lepidopteran species. In addition to the interest of the work per se, the differential distribution of GalNAc-dependent and -independent Cry1Ac-binding molecules, and of GalNAc-dependent and -independent Cry1Ac-mediated pore formation, can have practical applications. One could use different regions of the midgut to prepare BBMV enriched with different types of binding molecules and then use GalNAc to dissect further binding and/or pore-forming activity. Depending on the insect species studied, the analysis of binding and pore formation in different midgut regions could be an effective method to study differences in susceptibility to Cry1Ac toxin in different species and also to study the molecular basis of Cry1Ac resistance in some resistant insect strains minimizing interference from futile binding.