M. J. Adang, Department of Entomology, University of Georgia, Athens, GA 30602–2603, USA. Fax: + 1 706 542 2279, Tel.: + 1 706 542 2436, E-mail: email@example.com
We reported previously a direct correlation between reduced soybean agglutinin binding to 63- and 68-kDa midgut glycoproteins and resistance to Cry1Ac toxin from Bacillus thuringiensis in the tobacco budworm (Heliothis virescens). In the present work we describe the identification of the 68-kDa glycoprotein as a membrane-bound form of alkaline phosphatase we term HvALP. Lectin blot analysis of HvALP revealed the existence of N-linked oligosaccharides containing terminal N-acetylgalactosamine required for [125I]Cry1Ac binding in ligand blots. Based on immunoblotting and alkaline phosphatase activity detection, reduced soybean agglutinin binding to HvALP from Cry1Ac resistant larvae of the H. virescens YHD2 strain was attributable to reduced amounts of HvALP in resistant larvae. Quantification of specific alkaline phosphatase activity in brush border membrane proteins from susceptible (YDK and F1 generation from backcrosses) and YHD2 H. virescens larvae confirmed the observation of reduced HvALP levels. We propose HvALP as a Cry1Ac binding protein that is present at reduced levels in brush border membrane vesicles from YHD2 larvae.
Specific binding to insect midgut receptors is a key step in the mode of action of insecticidal Cry toxins from the bacterium Bacillus thuringiensis (Bt). Despite exceptions , in most cases Cry toxin specificity and potency correlate with the extent of toxin binding to midgut brush border membrane receptors in vitro[2,3]. Effective toxin binding to receptors results in toxin insertion and oligomerization on the midgut cell membrane, leading to pore formation and cell death by osmotic shock .
In brush border membrane vesicles (BBMV) from Heliothis virescens (tobacco budworm) larvae, three groups of binding sites (A, B, and C) for Cry1A toxins were proposed based on their toxin binding specificities [5,6]. The A binding sites, which bind Cry1Aa, Cry1Ab, Cry1Ac, Cry1Fa and Cry1Ja toxins, include the cadherin-like protein HevCaLP (J. L. Jurat-Fuentes, L. Gahan, F. Gould, D. Heckel and M. Adang, unpublished observation) and a 170-kDa N-aminopeptidase (APN) [5,7–9]. Currently, there is evidence that both HevCaLP  (J. L. Jurat-Fuentes, L. Gahan, F. Gould, D. Heckel and M. Adang, unpublished observation); and the 170-kDa APN [8,10] function as Cry1A toxin receptors. In the B binding site group, a 130-kDa protein has been shown to recognize both Cry1Ab and Cry1Ac. The C binding site group includes Cry1Ac toxin-binding proteins smaller than 100-kDa in size . We reported previously a correlation between altered glycosylation of 63- and 68-kDa glycoproteins that are part of the C binding site group and resistance to Cry1Ac in the H. virescens YHD2 strain .
Cry1 toxin-binding proteins of 60- to 80-kDa in size have been described in toxin overlays of BBMV proteins from H. virescens, Manduca sexta, and Plodia interpunctella. In 2D proteomic analysis of M. sexta BBMV proteins, McNall and Adang  reported Cry1Ac binding to a 65-kDa form of alkaline phosphatase (ALP, EC 220.127.116.11). Membrane-bound ALP from Bombyx mori and M. sexta are attached to the brush border cell membrane by a glycosylphosphatidylinositol (GPI) anchor [13–15]. Specific interactions between Cry1Ac and ALPs under native conditions resulting in inhibition of phosphatase activity have been reported for M. sexta and H. virescens. However, the potential role for alkaline phosphatases in Cry1Ac intoxication has not been addressed directly.
The main goals of the present study were to identify the 68-kDa glycoprotein and characterize its oligosaccharide residues as a first step to investigate the specific alteration of this glycoprotein in Cry1Ac-resistant YHD2 larvae. Based on reported molecular sizes of insect alkaline phosphatases, and their interaction with Cry1 toxins, we hypothesized the 68-kDa glycoprotein to be a form of alkaline phosphatase. Immunoblotting and enzymatic activity experiments identified the 68-kDa protein as a GPI-anchored form of alkaline phosphatase we term HvALP (for H. virescens alkaline phosphatase). Ligand blots and glycosidase digestion demonstrated that an N-linked oligosaccharide containing a terminal N-acetylgalactosamine (GalNAc) residue on HvALP was necessary for Cry1Ac binding. Immunoblotting and specific alkaline phosphatase activity of BBMV proteins from susceptible and resistant larvae provided evidence that decreased HvALP levels were produced in YHD2 larvae. Our results provide evidence that HvALP is involved in Cry1Ac toxicity to H. virescens larvae.
Materials and methods
Insect strains and brush border membrane vesicle (BBMV) preparation
H. virescens laboratory strains YDK and YHD2 have been described previously . YDK is the unselected susceptible control colony for the Cry1Ac-selected YHD2 strain, which developed 10 000-fold resistance to Cry1Ac when compared to susceptible YDK larvae . After continuous selection with Cry1Ac, levels of resistance increased to 73 000-fold . Fifth instar larvae from each strain were dissected and midguts frozen and kept at −80 °C until used to prepare BBMV.
BBMV were isolated by the differential centrifugation method of Wolfersberger et al. . BBMV proteins were quantified by the method of Bradford , using BSA as standard, and kept at −80 °C until used. APN activity using leucine-p-nitroanilide as the substrate was used as a marker for brush border enzyme enrichment in the BBMV preparations. APN activities were enriched six- to eight-fold in the BBMV preparations compared to initial midgut homogenates.
Cry1Ac toxin purification and labeling
B. thuringiensis strain HD-73 obtained from the Bacillus Genetic Stock Center (Colombus, OH, USA) was used to produce Cry1Ac. Mutated Cry1Ac QNR(509–511)→AAA(509–511) was expressed in Escherichia coli MV 1190 kindly provided by D. Dean (Ohio State University, OH, USA), and purified as described elsewhere . This Cry1Ac mutant toxin lacks the GalNAc binding properties of the wild-type toxin . Cry1Ac crystalline inclusions were solubilized, activated and purified as described previously . Purified toxin samples (verified by 10% reducing SDS/PAGE) were pooled, the protein concentration determined as for BBMV proteins and stored at −80 °C until used.
Purified Cry1Ac (1 µg) was radiolabeled with 0.5 mCi of [125I]Na by the chloramine T method . Specific activities of labeled samples were 3–8 mCi·mg−1, as determined using the bindability method of Schumacher et al. . Labeled toxins were kept at 4 °C and used within 10 days.
Quantification of alkaline phosphatase and aminopeptidase activities
Specific alkaline phosphatase (ALP) and N-aminopeptidase (APN) enzymatic activities of BBMV proteins were measured using p-nitrophenyl phosphate disodium (pNPP) and leucine-p-nitroanilide (Sigma, St. Louis, MO, USA) as substrates, respectively. BBMV proteins (1 µg) were mixed with ALP buffer (100 mm Tris/HCl, pH 9.5, 100 mm NaCl, 5 mm MgCl2) or NaCl/Pi buffer (10 mm Na2HPO4, pH 7.5, 135 mm NaCl, 2 mm KCl) containing 1.25 mmpNPP or 0.8 mm leucine-p-nitroanilide, respectively. Enzymatic activities were monitored as changes in the A405-value for 5 min at room temperature (ALP) or at 37 °C (APN) in a microplate reader (Molecular Devices). One enzymatic unit was defined as the amount of enzyme that would hydrolyze 1.0 µmole of substrate to chromogenic product per min at the specific reaction pH and temperature. Data shown are the mean specific activities from at least four independent BBMV batches from each H. virescens strain measured in at least three independent experiments.
Ligand, lectin and immunoblots of BBMV proteins
BBMV proteins (15 or 2 µg) were separated by SDS/PAGE 8%, and gels were either stained or electrotransferred to poly(vinylidene difluoride) Q membrane filters (Millipore). After overnight transfer, filters were blocked for 1 h at room temperature with NaCl/Pi buffer containing 0.1% Tween-20 (PBST) and 3% BSA.
For immunoblots, blocked filters were probed with a 1 : 25 000 dilution of polyclonal serum against the membrane-bound form of alkaline phosphatase (mALP) from B. mori (kindly provided by M. Itoh, Kyoto Institute of Technology, Kyoto, Japan) for 1 h. After washing with PBST containing 0.1% BSA, blots were probed with anti-rabbit serum (Sigma) conjugated to horseradish peroxidase (HRP) or alkaline phosphatase. Filters were developed using enhanced chemiluminescence (ECL; Amersham BioSciences) for peroxidase conjugates, or Nitro Blue tetrazolium and 5-bromo-4-chloroindol-2yl phosphate for alkaline phosphatase conjugates. No endogenous alkaline phosphatase activity was detected with Nitro Blue tetrazolium/5-bromo-4-chloroindol-2yl in blots of BBMV proteins when samples were boiled before electrophoresis. Periodate oxidation treatment of blots prior to immunoblotting did not alter antigenicity of BBMV proteins (data not shown); evidence that the serum used recognized protein and not sugar epitopes.
For lectin blots, blocked filters containing separated BBMV proteins were incubated with lectins from Canavalia ensiformis (ConA, at 0.05 µg·mL−1), Artocarpus integrifolia (Jac, at 0.5 µg·mL−1), Glycine max[soybean agglutinin (SBA), at 1 µg·mL−1], Ricinus communis (RCA-I, at 5 µg·mL−1), Dolichus biflorus (DBA, at 5 µg·mL−1), Sophora japonica (SJA, at 5 µg·mL−1), Wistaria floribunda (WFL, at 1 µg·mL−1), Helix pomatia (HPL, at 1 µg·mL−1), or Griffonia simplicifolia (GSL-I, at 5 µg·mL−1) for 1 h in blocking buffer (PBST plus 3% BSA). Con A, Jac, SBA, and HPL were purchased from Sigma; RCA-I, SJA, WFL, and GSL-I were from Vector laboratories (Burlingame, CA, USA).
Lectins conjugated to HRP were visualized by ECL. Blots of biotinylated lectins were probed with streptavidin–HRP conjugate (Vector) and then visualized as HRP-conjugated lectins. As controls for nonspecific lectin binding, lectins were incubated with specific hapten sugars (Table 1) for 30 min at room temperature before probing BBMV blots. This treatment eliminated or greatly decreased lectin binding to BBMV proteins on filters (see below, and data not shown).
Table 1. Sugar specificities of lectins (based on ) used in blots and respective hapten sugars used for lectin specificity controls. Several lectins were selected according to their specificity of binding to Gal, N-acetylgalactosamine (GalNAc), N-acetylglucosamine (GlcNAc), Man or Glc.
Canavalis ensiformis (ConA)
Artocarpus integrifolia (Jac)
0.8 m Gal
Glycine max (SBA)
0.2 m GalNAc
Ricinus communis (RCA-I)
0.2 m Gal
Dolichus biflorus (DBA)
0.2 m GalNAc
Sophora japonica (SJA)
0.2 m Gal
Wistaria floribunda (WFL)
0.2 m GalNAc
Helix pomatia (HPL)
0.2 m GalNAc
Griffonia simplicifolia (GSL)
0.2 m Gal
For SBA binding competition, filters were blocked as above, and then 12 µg·mL−1 of Cry1Ac or the Cry1Ac mutant protein QNR(509–511)→AAA(509–511) was added to the blocking buffer along with SBA lectin (1 µg·mL−1). After 1 h incubation and washing, filters were developed as described for lectin blots.
Ligand blots were performed as described previously . [125I]Cry1Ac (1 nm) was used to probe blotted BBMV proteins in blocking buffer for 1 h at room temperature. After washing, filters were exposed to photographic film at −80 °C for 24 h.
To detect HvALP in the filters used for lectin or ligand blotting, after development, filters were washed in PBST plus 0.1% BSA overnight. Blocking and HvALP immunodetection were performed as described above. To avoid interference with lectin or toxin detection, bound mALP antisera was detected by anti-rabbit sera conjugated to alkaline phosphatase.
Digestion of BBMV proteins with peptide-N-glycosidase F
Release of N-linked oligosaccharides from BBMV proteins was achieved by digestion of blotted BBMV proteins with peptide-N-glycosidase F (PNG-F). BBMV proteins (15 µg) were separated by 8% SDS/PAGE and transferred to poly(vinylidene difluoride) Q filters as above. Filters were incubated in 5 mL of NaCl/Pi buffer (pH 7.4) containing 0.1% SDS, 0.5% Triton-X-100 and 30 U of PNG-F (Boehringer-Mannheim) for 17 h at 37 °C. After treatment, filters were blocked and probed as for SBA lectin blots or [125I]Cry1Ac ligand blots. Controls, which had no PNG-F in the incubation buffer, showed no differences in lectin or toxin binding when compared to SBA and [125I]Cry1Ac blots (data not shown).
Detection of GPI anchors
The presence of glycosylphosphatidylinositol (GPI) anchors in BBMV proteins was detected following the method described by Luo et al. . Briefly, after phosphatidylinositol-specific phospholipase C (PIPLC) digestion of BBMV blots, cleaved GPI anchors were detected by immunological detection of the exposed cross-reacting determinant (CRD) epitope contained in the residue of the GPI anchor by probing with anti-CRD sera (kindly provided by K. Mensa-Wilmot, University of Georgia, Athens, GA, USA). Blots were probed with anti-rabbit–HRP conjugate (Sigma) before developing with enhanced chemiluminescence as above. In controls, which had no PIPLC in the blocking buffer, no proteins were detected (data not shown).
Detection of alkaline phosphatase activity in SDS/PAGE gels and blots
To detect alkaline phosphatase activity in BBMV, proteins (15 or 2 µg) solubilized in sample buffer  were not heat-denatured before gel loading. After 8% SDS/PAGE and transfer to poly(vinylidene difluoride) Q, filters were washed with ALP buffer for 15 min at room temperature. After addition of 330 µg·mL−1 of Nitro Blue tetrazolium and 165 µg·mL−1 of 5-bromo-4-chloroindol-2yl to the ALP buffer, alkaline phosphatase activity was visualized by the formation of a purple-red precipitate. Reactions were stopped by incubation of filters in 50 mL of NaCl/Pi, pH 7.5 containing 200 µL of 500 mm EDTA pH 8.0.
Identification of the 68-kDa BBMV glycoprotein as alkaline phosphatase
To test the hypothesis that the 68-kDa protein with altered glycosylation in the Cry1Ac-resistant YHD2 larvae was a form of ALP, we used sera developed against the mALP from B. mori to detect homologs of this protein in BBMV from H. virescens. Although no protein amount differences were detected in Coomassie blue stained gels (Fig. 1A), the 68-kDa protein had reduced SBA binding in BBMV from YHD2 larvae (Fig. 1B). This protein was recognized by sera against mALP (Fig. 1C) and displayed ALP activity in blots of BBMV proteins (Fig. 1D), demonstrating that this protein is a form of alkaline phosphatase. PIPLC digestion was used to determine whether the 68-kDa protein was GPI anchored to BBMV in H. virescens. As shown in Fig. 1E, after PIPLC digestion, anti-CRD sera recognized the 68-kDa protein in H. virescens BBMV, suggesting that this protein is GPI-anchored to the brush border membrane. Based on these results, we named the 68-kDa GPI-anchored glycoprotein as HvALP for H. virescens alkaline phosphatase.
Characterization of the glycan moiety of HvALP by lectin blotting
To investigate the oligosaccharides present on HvALP from Cry1Ac susceptible larvae, we performed lectin blotting using selected lectins (Table 1) and BBMV proteins from YDK larvae. After lectin blotting, HvALP on blots was detected by sera against B. mori mALP to confirm lectin binding to HvALP. As shown in Fig. 2, HvALP was recognized by lectins from Canavalia ensiformis (ConA), Glycine max (SBA), and Wistaria floribunda (WFL). The different pattern of BBMV proteins being recognized by both SBA and WFL (both bind terminal GalNAc) was probably due to the existence of terminal GalNAc in linkages poorly recognized by one of the lectins. Conversely, no binding to HvALP was detected using lectins from Artocarpus integrifolia (Jac), Ricinus communis (RCA), Dolichus biflorus (DBA), or Helix pomatia (HPL). Although proteins of similar size to HvALP were bound by Griffonia simplicifolia (GSL) and Sophora japonica (SJA) lectins, immunodetection of HvALP in these filters demonstrated that the detected lectin binding proteins were not HvALP. To further test the existence of terminal GalNAc on N-linked oligosaccharides on HvALP, we performed digestion of blotted BBMV proteins with PNG-F, which releases N-linked oligosaccharides as N-glycosides from polypeptide chains. Digestion of BBMV proteins with PNG-F eliminated binding of SBA to HvALP (Fig. 2), supporting the hypothesis that this protein has N-linked oligosaccharides with terminal GalNAc residues. Binding of SBA to other BBMV proteins was also decreased after PNG-F digestion, suggesting the presence of GalNAc or galactose on N-linked oligosaccharides in these proteins.
Importance of ALP glycosylation for Cry1Ac binding
To test the hypothesis that Cry1Ac toxin bound to the terminal GalNAc residue on HvALP, we competed SBA binding to HvALP with Cry1Ac. We did not perform the reciprocal competition assay due to the 106-fold lower affinity of SBA for GalNAc (Kd = 0.3 mm); when compared to Cry1Ac affinity for its binding sites (Kd = 1.1 nm). When comparing SBA binding to BBMV with Cry1Ac competition blots (Fig. 3A), Cry1Ac prevented SBA binding to HvALP as well as to other BBMV proteins, indicative of toxin binding to terminal GalNAc residues on these proteins. Binding of SBA to the 170-kDa APN was almost unaffected by the presence of Cry1Ac. As a control for toxin binding not due to GalNAc recognition, we competed SBA binding with a Cry1Ac mutant, QNR(509–511)→ AAA(509–511), which lacks GalNAc binding . SBA binding to HvALP was unchanged by QNR(509–511)→AAA(509–511), demonstrating that Cry1Ac bound to terminal GalNAc on HvALP.
To provide further support for the hypothesis of Cry1Ac binding to GalNAc on HvALP, we performed ligand blots with [125I]Cry1Ac. Cry1Ac bound to several BBMV proteins, including HvALP (Fig. 3B). When N-linked oligosaccharides were released from HvALP by PNG-F digestion, Cry1Ac did not bind to this protein, demonstrating that toxin binding was dependent on the presence of N-linked oligosaccharides on HvALP. Binding to other Cry1Ac binding proteins was also decreased greatly by PNG-F digestion, indicating the importance of N-linked protein glycosylation for Cry1Ac binding on blots.
Reduced HvALP correlates with resistance to Cry1Ac
To investigate the possibility that reduced SBA binding to HvALP from YHD2 larvae (Fig. 1B) was a result of decreased HvALP protein levels, we compared HvALP from YHD2, YDK, and larvae from the F1 generation of backcrosses between YDK and YHD2 adults, using immunodetection and alkaline phosphatase activity blots. Two different types of F1 larvae, according to the sex of the susceptible parent, were used to determine the potential existence of sex linkage. As shown in Fig. 4B, sera against the membrane-bound form of alkaline phosphatase from B. mori recognized HvALP in BBMV from YDK, YHD2 and F1 larvae. No differences in intensity of recognition were observed between HvALP from YDK and F1 vesicles, while recognition of HvALP in YHD2 was clearly reduced. To confirm reduction in HvALP antigen in BBMV from YHD2, we increased the protein load by three-, five- and tenfold to compare to YDK and F1 vesicles. Increased BBMV protein concentrations as observed in the stained gel (Fig. 4A), resulted in augmented HvALP recognition (lanes 3, 4 and 5 in Fig. 4B), clearly suggesting a reduction in HvALP protein levels in BBMV from YHD2 larvae. Visual comparison of the lanes with increasing YHD2 protein loads and the YDK and F1 lanes in the blots (Fig. 4B) suggested a three- to fivefold reduction in HvALP antigen levels in BBMV from YHD2 larvae when compared to YDK or F1 vesicle proteins.
We predicted that reduced HvALP amounts in BBMV from YHD2 larvae would result in reduced alkaline phosphatase activity. Alkaline phosphatase activity in blots of BBMV proteins from YDK and F1 larvae was similar, and higher than activity in YHD2 vesicles (Fig. 4C). In agreement with reduced protein levels observed in Fig. 4B, specific alkaline phosphatase activity in suspensions of BBMV from YHD2 insects was reduced three- to fourfold when compared to YDK or F1 vesicles (Table 2). N-aminopeptidasespecific activity was used as control, with no significant differences found between BBMV from YDK, YHD2 or F1 larvae. These results were evidence for reduced amounts of HvALP in BBMV from YHD2 larvae resulting in reduced alkaline phosphatase activity and correlating with resistance to Cry1Ac and reduced Cry1Ac toxin binding.
Table 2. Specific alkaline phosphatase (ALP) and N-aminopeptidase (APN) activities of BBMV suspensions from YDK, YHD2 and F1 larvae. Specific activity of BBMV suspensions is expressed in units per milligram of BBMV protein (U·mg−1). One enzymatic unit was defined as the amount of enzyme that would hydrolyze 1.0 µmole of substrate to chromogenic product per min at the specific reaction pH and temperature. SD; standard deviation of the mean based on at least six independent measurements.
ALP activity (U·mg−1 ± SD)
APN activity (U·mg−1 ± SD)
223 ± 91
2192 ± 427
77 ± 37
2364 ± 290
YDK♀ × YHD2♂
375 ± 12
3156 ± 62
YHD2♀ × YDK♂
292 ± 12
2921 ± 275
In the Cry1Ac-resistant H. virescens strain YHD2, knockout of the cadherin-like protein HevCaLP  resulted in reduction of Cry1Aa but not Cry1Ab or Cry1Ac binding  (J. L. Jurat-Fuentes, L. Gahan, F. Gould, D. Heckel and M. Adang, unpublished results). The patterns of Cry1Ac binding molecules in BBMV from YDK and YHD2 larvae, including the 170-kDa APN, were identical . To explain decreased Cry1Ac toxin binding after continuous selection of YHD2 larvae with Cry1Ac, we hypothesized a key role for two BBMV glycoproteins of 63- and 68-kDa in Cry1Ac binding and toxicity .
In this study we identified the 68-kDa glycoprotein as a membrane-bound form of alkaline phosphatase we term HvALP (H. virescens alkaline phosphatase). As observed in other insect alkaline phosphatases, HvALP was GPI-anchored to the cell membrane. In insect larvae, alkaline phosphatases have been localized along the midgut, in Malpighian tubules, and in embryos . Serum used to detect HvALP was developed originally against the mALP from B. mori, which was localized to the brush border of columnar cells along the middle and posterior midgut . As GPI anchored proteins, alkaline phosphatases are located preferentially in lipid rafts . Zhuang et al.  reported isolation of lipid rafts from H. virescens midgut epithelium containing a GPI-anchored protein of 66-kDa. Based on molecular size, the GPI anchor, and localization in rafts, we believe HvALP and the 66-kDa protein reported by Zhuang et al.  are equivalent. Alkaline phosphatases have been reported previously to interact with Cry1Ac toxin in ligand blots of BBMV from M. sexta[13,16]. Moreover, direct inhibition of alkaline phosphatase activity by Cry1Ac has been reported in H. virescens and M. sexta. Together with our current results, these observations are evidence of a direct interaction between Cry1Ac and membrane-bound forms of alkaline phosphatase.
As reported for other insect alkaline phosphatases , HvALP was glycosylated . Binding of ConA to HvALP was evidence for the presence of N-linked oligosaccharide structures, as this lectin recognizes the trimannosidic core characteristic of N-linked glycans . Binding of both SBA and WFL suggested the presence of either GalNAc or galactose at the nonreducing end of the oligosaccharide. Absence of RCA-I binding to HvALP suggested lack of terminal galactose, confirming that SBA and WFL were binding to a terminal GalNAc residue. Terminal GalNAc in glycoproteins is usually part of an O-linked glycan . Interestingly, none of the lectins with high specificity for O-linked oligosaccharide structures (Jac, DBA, HPL, SJA) bound HvALP, indicating that terminal GalNAc bound by SBA and WFL was part of a complex or hybrid type N-linked oligosaccharide.
Even though N-linked oligosaccharides with complex type cores are rare in insects , mALP from B. mori was found to possess oligosaccharides of the biantennary complex type . Terminal GalNAc has been proposed as binding site for Shiga-like and heat-labile toxins from E. coli[36,37]. Additionally, the role of GalNAc as binding epitope for Cry1Ac toxin has been studied extensively [38–41]. Lack of DBA and HPL binding is evidence that the terminal GalNAc on HvALP is not in a GalNAcα1→3 linkage. Considering that terminal GalNAc in other α-linkages has not been reported to occur on N-linked oligosaccharides, and both SBA and WFL bind α- as well as β-linked GalNAc, terminal GalNAc on HvALP is probably β-linked. Terminal βGalNAc has been reported in N-linked oligosaccharides of proteins synthesized by the parasite Dirofilaria immitis and in microvillar glycoproteins of 68-kDa in size from Anopheles stephensi midguts [42,43]. Even though both terminal GalNAcβ1→3 and GalNAcβ1→4 can be found in biological samples, only terminal GalNAcβ1→4 has been described to occur on glycoproteins. Lepidopteran insect cell lines express a β1→4-GalNAc transferase that functions in the synthesis of complex-type carbohydrate chains . N-linked oligosaccharides containing terminal GalNAcβ1→4 have been reported in hemocyanin from the pond snail Lymnaea stagnalis, bovine milk , antigenic glycoproteins from Schistosoma mansoni, and bee venom . Terminal GalNAcβ1→4Gal has been proposed as adherence receptor for Streptococcus pneumoniae and E. coli infection in humans [49,50].
Binding of Cry1Ac to proteins of 68-kDa in size in ligand blots of H. virescens BBMV has been reported previously [1,5,11]. Our ligand blotting and competition results are evidence for Cry1Ac binding to the terminal GalNAc residue on HvALP. An interesting possibility is that terminal GalNAcβ1→4 may serve as a general recognition epitope for Cry1Ac toxin on alternative toxin receptors. Zhuang et al.  proposed a potential role for GPI anchored proteins such as HvALP in toxin action after observing a correlation between partition of Cry toxin to lipid rafts, toxin aggregation, and pore formation. Although speculative, Cry1Ac may bind to GalNAcβ1→4 on HvALP to initiate toxin oligomerization and pore formation, due to putative HvALP localization in lipid rafts. Similarly, the aerolysin enterotoxin from the bacterium Aeromonas hydrophila binds to βGlcNAc on the GPI anchor of alkaline phosphatase before insertion on target cell membranes [51,52]. In support of the terminal GalNAcβ1→4 as a Cry toxin binding epitope, mutations in a predicted UDP-GalNAc:GlcNAc β1,4-N-acetylgalactosaminyltransferase resulted in resistance to Cry5B and Cry14A Bt toxins in Caenorhabditis elegans. Further analysis of purified oligosaccharides from HvALP as well as other putative toxin receptors would be necessary to obtain more conclusive and detailed linkage information on oligosaccharides with terminal GalNAc.
As we did not previously observe Cry1Aa or Cry1Ab binding to HvALP on ligand blots , we propose that HvALP is part of the C group of binding sites. According to the current toxin binding model, alteration of C binding sites would explain reduced Cry1Ac binding, as observed in BBMV from YHD2 insects . Our initial hypothesis, to explain reduced Cry1Ac and SBA binding to HvALP in YHD2 larvae, was based on possible alteration of protein glycosylation in resistant insects. Results from immunoblotting and alkaline phosphatase activity detection revealed instead that HvALP protein levels were decreased in BBMV from YHD2 larvae. Therefore, decreased SBA binding to HvALP from YHD2 vesicles was due to reduced protein levels and not to altered glycosylation. Due to limiting YHD2 materials, oligosaccharide analysis was only performed in BBMV from YDK larvae, hence potential alterations of HvALP glycosylation in YHD2 larvae cannot be excluded. BBMV from the F1 generation of reciprocal crosses recovered HvALP levels observed for the susceptible parents independently of the sex of the susceptible progenitor, demonstrating autosomal recessive transmission of this trait. Considering that F1 generation larvae bound Cry1Ac toxin and were only twofold resistant to Cry1Ac , our results are evidence for a direct correlation between decreased HvALP levels and increased resistance to Cry1Ac.
Electrophoretic variations of alkaline phosphatase between different strains or developmental stages have been reported for Drosophila melanogaster, Aedes aegypti, and B. mori[56,57], although the physiological consequences of these variations are not clearly understood. In the Tsunomata B. mori strain, reduced mALP activity correlated with undetectable levels of mALP antigen, while there were no alterations in gene copy or transcript size . These results suggested that electrophoretic mALP polymorphisms were due to post-transcriptional processes. The fact that Tsunomata larvae were viable and fertile under normal conditions suggests lack of dramatic fitness costs associated with reduced mALP levels. Interestingly, YHD2 larvae do not survive through pupation when grown in cotton or Bt cotton , suggesting dramatic fitness costs associated with resistance in this species. We believe these costs are the result of the existence of multiple resistance mechanisms in YHD2 larvae. The existence of such effects is crucial when designing approaches to delay evolution of resistance against Bt crops.
Insect alkaline phosphatases have been proposed to function in active absorption of metabolites and transport processes , although there is also evidence for participation in cell adhesion and differentiation . Interestingly, knockout of HevCaLP, another protein predicted to function in cell adhesion processes, results in Cry1 resistance in YHD2 larvae . According to these important functions, significant fitness costs associated with reduced ALP activity would be expected, although information from the Tsunomata B. mori strain may suggest the contrary.
The specific mechanism by which YHD2 larvae reduce HvALP expression needs further investigation. As stated above, information from B. mori mALP suggests that decreased HvALP activity may not be related to changes in gene copy number or transcription. An alternative hypothetical mechanism to reduce HvALP in midgut brush border membranes was proposed previously by Lu and Adang . According to this hypothesis, GPI-anchored proteins would be selectively solubilized by endogenous PIPLC digestion in Bt-resistant insects. Such treatment would result in elimination of potential Cry toxin binding sites such as aminopeptidases and alkaline phosphatases from the midgut epithelium. In support of this hypothesis, B. mori mALP is solubilized by midgut epithelium enzymes to form digestive fluid alkaline phosphatase (dALP), which is highly resistant to degradation by midgut proteases .
Our results demonstrate a direct correlation between decreased HvALP levels and resistance in H. virescens. HvALP may be a critical component in toxicity, or alternatively, the reduced HvALP levels observed in resistant larvae may indicate broader alterations in the brush border membrane. One possibility is that resistant larvae have altered membrane components such as lipid rafts that affect the amounts of HvALP localized to the brush border membrane. The specific role of HvALP in Cry1Ac intoxication needs further investigation. We believe HvALP has potential as a resistance marker, so that biochemical and DNA-based tests may be developed to detect emergence of resistance to Bt crops in field populations. These questions are currently being addressed in our laboratory.
The authors express their gratitude to Dr. Fred Gould (North Carolina State University, Raleigh, NC, USA) for providing the Heliothis materials used for this research.